diff options
Diffstat (limited to 'net/core/dev.c')
| -rw-r--r-- | net/core/dev.c | 875 |
1 files changed, 557 insertions, 318 deletions
diff --git a/net/core/dev.c b/net/core/dev.c index 76e6438f4858..e1bb6d7856d9 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -77,7 +77,9 @@ #include <linux/hash.h> #include <linux/slab.h> #include <linux/sched.h> +#include <linux/sched/isolation.h> #include <linux/sched/mm.h> +#include <linux/smpboot.h> #include <linux/mutex.h> #include <linux/rwsem.h> #include <linux/string.h> @@ -153,41 +155,21 @@ #include <linux/prandom.h> #include <linux/once_lite.h> #include <net/netdev_rx_queue.h> +#include <net/page_pool/types.h> +#include <net/page_pool/helpers.h> +#include <net/rps.h> #include "dev.h" #include "net-sysfs.h" static DEFINE_SPINLOCK(ptype_lock); struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly; -struct list_head ptype_all __read_mostly; /* Taps */ static int netif_rx_internal(struct sk_buff *skb); static int call_netdevice_notifiers_extack(unsigned long val, struct net_device *dev, struct netlink_ext_ack *extack); -/* - * The @dev_base_head list is protected by @dev_base_lock and the rtnl - * semaphore. - * - * Pure readers hold dev_base_lock for reading, or rcu_read_lock() - * - * Writers must hold the rtnl semaphore while they loop through the - * dev_base_head list, and hold dev_base_lock for writing when they do the - * actual updates. This allows pure readers to access the list even - * while a writer is preparing to update it. - * - * To put it another way, dev_base_lock is held for writing only to - * protect against pure readers; the rtnl semaphore provides the - * protection against other writers. - * - * See, for example usages, register_netdevice() and - * unregister_netdevice(), which must be called with the rtnl - * semaphore held. - */ -DEFINE_RWLOCK(dev_base_lock); -EXPORT_SYMBOL(dev_base_lock); - static DEFINE_MUTEX(ifalias_mutex); /* protects napi_hash addition/deletion and napi_gen_id */ @@ -200,8 +182,9 @@ static DECLARE_RWSEM(devnet_rename_sem); static inline void dev_base_seq_inc(struct net *net) { - while (++net->dev_base_seq == 0) - ; + unsigned int val = net->dev_base_seq + 1; + + WRITE_ONCE(net->dev_base_seq, val ?: 1); } static inline struct hlist_head *dev_name_hash(struct net *net, const char *name) @@ -216,35 +199,60 @@ static inline struct hlist_head *dev_index_hash(struct net *net, int ifindex) return &net->dev_index_head[ifindex & (NETDEV_HASHENTRIES - 1)]; } -static inline void rps_lock_irqsave(struct softnet_data *sd, - unsigned long *flags) +#ifndef CONFIG_PREEMPT_RT + +static DEFINE_STATIC_KEY_FALSE(use_backlog_threads_key); + +static int __init setup_backlog_napi_threads(char *arg) +{ + static_branch_enable(&use_backlog_threads_key); + return 0; +} +early_param("thread_backlog_napi", setup_backlog_napi_threads); + +static bool use_backlog_threads(void) { - if (IS_ENABLED(CONFIG_RPS)) + return static_branch_unlikely(&use_backlog_threads_key); +} + +#else + +static bool use_backlog_threads(void) +{ + return true; +} + +#endif + +static inline void backlog_lock_irq_save(struct softnet_data *sd, + unsigned long *flags) +{ + if (IS_ENABLED(CONFIG_RPS) || use_backlog_threads()) spin_lock_irqsave(&sd->input_pkt_queue.lock, *flags); else if (!IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_save(*flags); } -static inline void rps_lock_irq_disable(struct softnet_data *sd) +static inline void backlog_lock_irq_disable(struct softnet_data *sd) { - if (IS_ENABLED(CONFIG_RPS)) + if (IS_ENABLED(CONFIG_RPS) || use_backlog_threads()) spin_lock_irq(&sd->input_pkt_queue.lock); else if (!IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_disable(); } -static inline void rps_unlock_irq_restore(struct softnet_data *sd, - unsigned long *flags) +static inline void backlog_unlock_irq_restore(struct softnet_data *sd, + unsigned long *flags) { - if (IS_ENABLED(CONFIG_RPS)) + if (IS_ENABLED(CONFIG_RPS) || use_backlog_threads()) spin_unlock_irqrestore(&sd->input_pkt_queue.lock, *flags); else if (!IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_restore(*flags); } -static inline void rps_unlock_irq_enable(struct softnet_data *sd) +static inline void backlog_unlock_irq_enable(struct softnet_data *sd) { - if (IS_ENABLED(CONFIG_RPS)) + if (IS_ENABLED(CONFIG_RPS) || use_backlog_threads()) spin_unlock_irq(&sd->input_pkt_queue.lock); else if (!IS_ENABLED(CONFIG_PREEMPT_RT)) local_irq_enable(); @@ -341,13 +349,22 @@ int netdev_name_node_alt_create(struct net_device *dev, const char *name) return 0; } -static void __netdev_name_node_alt_destroy(struct netdev_name_node *name_node) +static void netdev_name_node_alt_free(struct rcu_head *head) { - list_del(&name_node->list); + struct netdev_name_node *name_node = + container_of(head, struct netdev_name_node, rcu); + kfree(name_node->name); netdev_name_node_free(name_node); } +static void __netdev_name_node_alt_destroy(struct netdev_name_node *name_node) +{ + netdev_name_node_del(name_node); + list_del(&name_node->list); + call_rcu(&name_node->rcu, netdev_name_node_alt_free); +} + int netdev_name_node_alt_destroy(struct net_device *dev, const char *name) { struct netdev_name_node *name_node; @@ -362,10 +379,7 @@ int netdev_name_node_alt_destroy(struct net_device *dev, const char *name) if (name_node == dev->name_node || name_node->dev != dev) return -EINVAL; - netdev_name_node_del(name_node); - synchronize_rcu(); __netdev_name_node_alt_destroy(name_node); - return 0; } @@ -373,8 +387,10 @@ static void netdev_name_node_alt_flush(struct net_device *dev) { struct netdev_name_node *name_node, *tmp; - list_for_each_entry_safe(name_node, tmp, &dev->name_node->list, list) - __netdev_name_node_alt_destroy(name_node); + list_for_each_entry_safe(name_node, tmp, &dev->name_node->list, list) { + list_del(&name_node->list); + netdev_name_node_alt_free(&name_node->rcu); + } } /* Device list insertion */ @@ -385,12 +401,10 @@ static void list_netdevice(struct net_device *dev) ASSERT_RTNL(); - write_lock(&dev_base_lock); list_add_tail_rcu(&dev->dev_list, &net->dev_base_head); netdev_name_node_add(net, dev->name_node); hlist_add_head_rcu(&dev->index_hlist, dev_index_hash(net, dev->ifindex)); - write_unlock(&dev_base_lock); netdev_for_each_altname(dev, name_node) netdev_name_node_add(net, name_node); @@ -404,7 +418,7 @@ static void list_netdevice(struct net_device *dev) /* Device list removal * caller must respect a RCU grace period before freeing/reusing dev */ -static void unlist_netdevice(struct net_device *dev, bool lock) +static void unlist_netdevice(struct net_device *dev) { struct netdev_name_node *name_node; struct net *net = dev_net(dev); @@ -417,13 +431,9 @@ static void unlist_netdevice(struct net_device *dev, bool lock) netdev_name_node_del(name_node); /* Unlink dev from the device chain */ - if (lock) - write_lock(&dev_base_lock); list_del_rcu(&dev->dev_list); netdev_name_node_del(dev->name_node); hlist_del_rcu(&dev->index_hlist); - if (lock) - write_unlock(&dev_base_lock); dev_base_seq_inc(dev_net(dev)); } @@ -442,6 +452,12 @@ static RAW_NOTIFIER_HEAD(netdev_chain); DEFINE_PER_CPU_ALIGNED(struct softnet_data, softnet_data); EXPORT_PER_CPU_SYMBOL(softnet_data); +/* Page_pool has a lockless array/stack to alloc/recycle pages. + * PP consumers must pay attention to run APIs in the appropriate context + * (e.g. NAPI context). + */ +static DEFINE_PER_CPU(struct page_pool *, system_page_pool); + #ifdef CONFIG_LOCKDEP /* * register_netdevice() inits txq->_xmit_lock and sets lockdep class @@ -551,7 +567,7 @@ static inline void netdev_set_addr_lockdep_class(struct net_device *dev) static inline struct list_head *ptype_head(const struct packet_type *pt) { if (pt->type == htons(ETH_P_ALL)) - return pt->dev ? &pt->dev->ptype_all : &ptype_all; + return pt->dev ? &pt->dev->ptype_all : &net_hotdata.ptype_all; else return pt->dev ? &pt->dev->ptype_specific : &ptype_base[ntohs(pt->type) & PTYPE_HASH_MASK]; @@ -653,7 +669,7 @@ int dev_get_iflink(const struct net_device *dev) if (dev->netdev_ops && dev->netdev_ops->ndo_get_iflink) return dev->netdev_ops->ndo_get_iflink(dev); - return dev->ifindex; + return READ_ONCE(dev->ifindex); } EXPORT_SYMBOL(dev_get_iflink); @@ -738,9 +754,9 @@ EXPORT_SYMBOL_GPL(dev_fill_forward_path); * @net: the applicable net namespace * @name: name to find * - * Find an interface by name. Must be called under RTNL semaphore - * or @dev_base_lock. If the name is found a pointer to the device - * is returned. If the name is not found then %NULL is returned. The + * Find an interface by name. Must be called under RTNL semaphore. + * If the name is found a pointer to the device is returned. + * If the name is not found then %NULL is returned. The * reference counters are not incremented so the caller must be * careful with locks. */ @@ -821,8 +837,7 @@ EXPORT_SYMBOL(netdev_get_by_name); * Search for an interface by index. Returns %NULL if the device * is not found or a pointer to the device. The device has not * had its reference counter increased so the caller must be careful - * about locking. The caller must hold either the RTNL semaphore - * or @dev_base_lock. + * about locking. The caller must hold the RTNL semaphore. */ struct net_device *__dev_get_by_index(struct net *net, int ifindex) @@ -924,6 +939,18 @@ struct net_device *dev_get_by_napi_id(unsigned int napi_id) } EXPORT_SYMBOL(dev_get_by_napi_id); +static DEFINE_SEQLOCK(netdev_rename_lock); + +void netdev_copy_name(struct net_device *dev, char *name) +{ + unsigned int seq; + + do { + seq = read_seqbegin(&netdev_rename_lock); + strscpy(name, dev->name, IFNAMSIZ); + } while (read_seqretry(&netdev_rename_lock, seq)); +} + /** * netdev_get_name - get a netdevice name, knowing its ifindex. * @net: network namespace @@ -935,7 +962,6 @@ int netdev_get_name(struct net *net, char *name, int ifindex) struct net_device *dev; int ret; - down_read(&devnet_rename_sem); rcu_read_lock(); dev = dev_get_by_index_rcu(net, ifindex); @@ -944,12 +970,11 @@ int netdev_get_name(struct net *net, char *name, int ifindex) goto out; } - strcpy(name, dev->name); + netdev_copy_name(dev, name); ret = 0; out: rcu_read_unlock(); - up_read(&devnet_rename_sem); return ret; } @@ -1201,7 +1226,10 @@ int dev_change_name(struct net_device *dev, const char *newname) memcpy(oldname, dev->name, IFNAMSIZ); + write_seqlock(&netdev_rename_lock); err = dev_get_valid_name(net, dev, newname); + write_sequnlock(&netdev_rename_lock); + if (err < 0) { up_write(&devnet_rename_sem); return err; @@ -1212,13 +1240,13 @@ int dev_change_name(struct net_device *dev, const char *newname) dev->flags & IFF_UP ? " (while UP)" : ""); old_assign_type = dev->name_assign_type; - dev->name_assign_type = NET_NAME_RENAMED; + WRITE_ONCE(dev->name_assign_type, NET_NAME_RENAMED); rollback: ret = device_rename(&dev->dev, dev->name); if (ret) { memcpy(dev->name, oldname, IFNAMSIZ); - dev->name_assign_type = old_assign_type; + WRITE_ONCE(dev->name_assign_type, old_assign_type); up_write(&devnet_rename_sem); return ret; } @@ -1227,15 +1255,11 @@ rollback: netdev_adjacent_rename_links(dev, oldname); - write_lock(&dev_base_lock); netdev_name_node_del(dev->name_node); - write_unlock(&dev_base_lock); - synchronize_rcu(); + synchronize_net(); - write_lock(&dev_base_lock); netdev_name_node_add(net, dev->name_node); - write_unlock(&dev_base_lock); ret = call_netdevice_notifiers(NETDEV_CHANGENAME, dev); ret = notifier_to_errno(ret); @@ -1245,9 +1269,11 @@ rollback: if (err >= 0) { err = ret; down_write(&devnet_rename_sem); + write_seqlock(&netdev_rename_lock); memcpy(dev->name, oldname, IFNAMSIZ); + write_sequnlock(&netdev_rename_lock); memcpy(oldname, newname, IFNAMSIZ); - dev->name_assign_type = old_assign_type; + WRITE_ONCE(dev->name_assign_type, old_assign_type); old_assign_type = NET_NAME_RENAMED; goto rollback; } else { @@ -2073,6 +2099,11 @@ void net_dec_egress_queue(void) EXPORT_SYMBOL_GPL(net_dec_egress_queue); #endif +#ifdef CONFIG_NET_CLS_ACT +DEFINE_STATIC_KEY_FALSE(tcf_bypass_check_needed_key); +EXPORT_SYMBOL(tcf_bypass_check_needed_key); +#endif + DEFINE_STATIC_KEY_FALSE(netstamp_needed_key); EXPORT_SYMBOL(netstamp_needed_key); #ifdef CONFIG_JUMP_LABEL @@ -2242,7 +2273,8 @@ static inline bool skb_loop_sk(struct packet_type *ptype, struct sk_buff *skb) */ bool dev_nit_active(struct net_device *dev) { - return !list_empty(&ptype_all) || !list_empty(&dev->ptype_all); + return !list_empty(&net_hotdata.ptype_all) || + !list_empty(&dev->ptype_all); } EXPORT_SYMBOL_GPL(dev_nit_active); @@ -2253,15 +2285,14 @@ EXPORT_SYMBOL_GPL(dev_nit_active); void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev) { - struct packet_type *ptype; + struct list_head *ptype_list = &net_hotdata.ptype_all; + struct packet_type *ptype, *pt_prev = NULL; struct sk_buff *skb2 = NULL; - struct packet_type *pt_prev = NULL; - struct list_head *ptype_list = &ptype_all; rcu_read_lock(); again: list_for_each_entry_rcu(ptype, ptype_list, list) { - if (ptype->ignore_outgoing) + if (READ_ONCE(ptype->ignore_outgoing)) continue; /* Never send packets back to the socket @@ -2302,7 +2333,7 @@ again: pt_prev = ptype; } - if (ptype_list == &ptype_all) { + if (ptype_list == &net_hotdata.ptype_all) { ptype_list = &dev->ptype_all; goto again; } @@ -3791,6 +3822,10 @@ no_lock_out: return rc; } + if (unlikely(READ_ONCE(q->owner) == smp_processor_id())) { + kfree_skb_reason(skb, SKB_DROP_REASON_TC_RECLASSIFY_LOOP); + return NET_XMIT_DROP; + } /* * Heuristic to force contended enqueues to serialize on a * separate lock before trying to get qdisc main lock. @@ -3830,7 +3865,9 @@ no_lock_out: qdisc_run_end(q); rc = NET_XMIT_SUCCESS; } else { + WRITE_ONCE(q->owner, smp_processor_id()); rc = dev_qdisc_enqueue(skb, q, &to_free, txq); + WRITE_ONCE(q->owner, -1); if (qdisc_run_begin(q)) { if (unlikely(contended)) { spin_unlock(&q->busylock); @@ -3927,6 +3964,11 @@ static int tc_run(struct tcx_entry *entry, struct sk_buff *skb, if (!miniq) return ret; + if (static_branch_unlikely(&tcf_bypass_check_needed_key)) { + if (tcf_block_bypass_sw(miniq->block)) + return ret; + } + tc_skb_cb(skb)->mru = 0; tc_skb_cb(skb)->post_ct = false; tcf_set_drop_reason(skb, *drop_reason); @@ -4420,20 +4462,11 @@ EXPORT_SYMBOL(__dev_direct_xmit); /************************************************************************* * Receiver routines *************************************************************************/ +static DEFINE_PER_CPU(struct task_struct *, backlog_napi); -int netdev_max_backlog __read_mostly = 1000; -EXPORT_SYMBOL(netdev_max_backlog); - -int netdev_tstamp_prequeue __read_mostly = 1; -unsigned int sysctl_skb_defer_max __read_mostly = 64; -int netdev_budget __read_mostly = 300; -/* Must be at least 2 jiffes to guarantee 1 jiffy timeout */ -unsigned int __read_mostly netdev_budget_usecs = 2 * USEC_PER_SEC / HZ; int weight_p __read_mostly = 64; /* old backlog weight */ int dev_weight_rx_bias __read_mostly = 1; /* bias for backlog weight */ int dev_weight_tx_bias __read_mostly = 1; /* bias for output_queue quota */ -int dev_rx_weight __read_mostly = 64; -int dev_tx_weight __read_mostly = 64; /* Called with irq disabled */ static inline void ____napi_schedule(struct softnet_data *sd, @@ -4452,18 +4485,16 @@ static inline void ____napi_schedule(struct softnet_data *sd, */ thread = READ_ONCE(napi->thread); if (thread) { - /* Avoid doing set_bit() if the thread is in - * INTERRUPTIBLE state, cause napi_thread_wait() - * makes sure to proceed with napi polling - * if the thread is explicitly woken from here. - */ - if (READ_ONCE(thread->__state) != TASK_INTERRUPTIBLE) - set_bit(NAPI_STATE_SCHED_THREADED, &napi->state); + if (use_backlog_threads() && thread == raw_cpu_read(backlog_napi)) + goto use_local_napi; + + set_bit(NAPI_STATE_SCHED_THREADED, &napi->state); wake_up_process(thread); return; } } +use_local_napi: list_add_tail(&napi->poll_list, &sd->poll_list); WRITE_ONCE(napi->list_owner, smp_processor_id()); /* If not called from net_rx_action() @@ -4475,12 +4506,6 @@ static inline void ____napi_schedule(struct softnet_data *sd, #ifdef CONFIG_RPS -/* One global table that all flow-based protocols share. */ -struct rps_sock_flow_table __rcu *rps_sock_flow_table __read_mostly; -EXPORT_SYMBOL(rps_sock_flow_table); -u32 rps_cpu_mask __read_mostly; -EXPORT_SYMBOL(rps_cpu_mask); - struct static_key_false rps_needed __read_mostly; EXPORT_SYMBOL(rps_needed); struct static_key_false rfs_needed __read_mostly; @@ -4495,7 +4520,7 @@ set_rps_cpu(struct net_device *dev, struct sk_buff *skb, struct netdev_rx_queue *rxqueue; struct rps_dev_flow_table *flow_table; struct rps_dev_flow *old_rflow; - u32 flow_id; + u32 flow_id, head; u16 rxq_index; int rc; @@ -4518,16 +4543,16 @@ set_rps_cpu(struct net_device *dev, struct sk_buff *skb, goto out; old_rflow = rflow; rflow = &flow_table->flows[flow_id]; - rflow->filter = rc; - if (old_rflow->filter == rflow->filter) - old_rflow->filter = RPS_NO_FILTER; + WRITE_ONCE(rflow->filter, rc); + if (old_rflow->filter == rc) + WRITE_ONCE(old_rflow->filter, RPS_NO_FILTER); out: #endif - rflow->last_qtail = - per_cpu(softnet_data, next_cpu).input_queue_head; + head = READ_ONCE(per_cpu(softnet_data, next_cpu).input_queue_head); + rps_input_queue_tail_save(&rflow->last_qtail, head); } - rflow->cpu = next_cpu; + WRITE_ONCE(rflow->cpu, next_cpu); return rflow; } @@ -4572,7 +4597,7 @@ static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb, if (!hash) goto done; - sock_flow_table = rcu_dereference(rps_sock_flow_table); + sock_flow_table = rcu_dereference(net_hotdata.rps_sock_flow_table); if (flow_table && sock_flow_table) { struct rps_dev_flow *rflow; u32 next_cpu; @@ -4582,10 +4607,10 @@ static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb, * This READ_ONCE() pairs with WRITE_ONCE() from rps_record_sock_flow(). */ ident = READ_ONCE(sock_flow_table->ents[hash & sock_flow_table->mask]); - if ((ident ^ hash) & ~rps_cpu_mask) + if ((ident ^ hash) & ~net_hotdata.rps_cpu_mask) goto try_rps; - next_cpu = ident & rps_cpu_mask; + next_cpu = ident & net_hotdata.rps_cpu_mask; /* OK, now we know there is a match, * we can look at the local (per receive queue) flow table @@ -4606,7 +4631,7 @@ static int get_rps_cpu(struct net_device *dev, struct sk_buff *skb, */ if (unlikely(tcpu != next_cpu) && (tcpu >= nr_cpu_ids || !cpu_online(tcpu) || - ((int)(per_cpu(softnet_data, tcpu).input_queue_head - + ((int)(READ_ONCE(per_cpu(softnet_data, tcpu).input_queue_head) - rflow->last_qtail)) >= 0)) { tcpu = next_cpu; rflow = set_rps_cpu(dev, skb, rflow, next_cpu); @@ -4660,9 +4685,9 @@ bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, if (flow_table && flow_id <= flow_table->mask) { rflow = &flow_table->flows[flow_id]; cpu = READ_ONCE(rflow->cpu); - if (rflow->filter == filter_id && cpu < nr_cpu_ids && - ((int)(per_cpu(softnet_data, cpu).input_queue_head - - rflow->last_qtail) < + if (READ_ONCE(rflow->filter) == filter_id && cpu < nr_cpu_ids && + ((int)(READ_ONCE(per_cpu(softnet_data, cpu).input_queue_head) - + READ_ONCE(rflow->last_qtail)) < (int)(10 * flow_table->mask))) expire = false; } @@ -4709,6 +4734,11 @@ static void napi_schedule_rps(struct softnet_data *sd) #ifdef CONFIG_RPS if (sd != mysd) { + if (use_backlog_threads()) { + __napi_schedule_irqoff(&sd->backlog); + return; + } + sd->rps_ipi_next = mysd->rps_ipi_list; mysd->rps_ipi_list = sd; @@ -4723,6 +4753,23 @@ static void napi_schedule_rps(struct softnet_data *sd) __napi_schedule_irqoff(&mysd->backlog); } +void kick_defer_list_purge(struct softnet_data *sd, unsigned int cpu) +{ + unsigned long flags; + + if (use_backlog_threads()) { + backlog_lock_irq_save(sd, &flags); + + if (!__test_and_set_bit(NAPI_STATE_SCHED, &sd->backlog.state)) + __napi_schedule_irqoff(&sd->backlog); + + backlog_unlock_irq_restore(sd, &flags); + + } else if (!cmpxchg(&sd->defer_ipi_scheduled, 0, 1)) { + smp_call_function_single_async(cpu, &sd->defer_csd); + } +} + #ifdef CONFIG_NET_FLOW_LIMIT int netdev_flow_limit_table_len __read_mostly = (1 << 12); #endif @@ -4734,7 +4781,7 @@ static bool skb_flow_limit(struct sk_buff *skb, unsigned int qlen) struct softnet_data *sd; unsigned int old_flow, new_flow; - if (qlen < (READ_ONCE(netdev_max_backlog) >> 1)) + if (qlen < (READ_ONCE(net_hotdata.max_backlog) >> 1)) return false; sd = this_cpu_ptr(&softnet_data); @@ -4774,36 +4821,45 @@ static int enqueue_to_backlog(struct sk_buff *skb, int cpu, struct softnet_data *sd; unsigned long flags; unsigned int qlen; + int max_backlog; + u32 tail; - reason = SKB_DROP_REASON_NOT_SPECIFIED; + reason = SKB_DROP_REASON_DEV_READY; + if (!netif_running(skb->dev)) + goto bad_dev; + + reason = SKB_DROP_REASON_CPU_BACKLOG; sd = &per_cpu(softnet_data, cpu); - rps_lock_irqsave(sd, &flags); - if (!netif_running(skb->dev)) - goto drop; + qlen = skb_queue_len_lockless(&sd->input_pkt_queue); + max_backlog = READ_ONCE(net_hotdata.max_backlog); + if (unlikely(qlen > max_backlog)) + goto cpu_backlog_drop; + backlog_lock_irq_save(sd, &flags); qlen = skb_queue_len(&sd->input_pkt_queue); - if (qlen <= READ_ONCE(netdev_max_backlog) && !skb_flow_limit(skb, qlen)) { - if (qlen) { -enqueue: - __skb_queue_tail(&sd->input_pkt_queue, skb); - input_queue_tail_incr_save(sd, qtail); - rps_unlock_irq_restore(sd, &flags); - return NET_RX_SUCCESS; + if (qlen <= max_backlog && !skb_flow_limit(skb, qlen)) { + if (!qlen) { + /* Schedule NAPI for backlog device. We can use + * non atomic operation as we own the queue lock. + */ + if (!__test_and_set_bit(NAPI_STATE_SCHED, + &sd->backlog.state)) + napi_schedule_rps(sd); } + __skb_queue_tail(&sd->input_pkt_queue, skb); + tail = rps_input_queue_tail_incr(sd); + backlog_unlock_irq_restore(sd, &flags); - /* Schedule NAPI for backlog device - * We can use non atomic operation since we own the queue lock - */ - if (!__test_and_set_bit(NAPI_STATE_SCHED, &sd->backlog.state)) - napi_schedule_rps(sd); - goto enqueue; + /* save the tail outside of the critical section */ + rps_input_queue_tail_save(qtail, tail); + return NET_RX_SUCCESS; } - reason = SKB_DROP_REASON_CPU_BACKLOG; -drop: - sd->dropped++; - rps_unlock_irq_restore(sd, &flags); + backlog_unlock_irq_restore(sd, &flags); +cpu_backlog_drop: + atomic_inc(&sd->dropped); +bad_dev: dev_core_stats_rx_dropped_inc(skb->dev); kfree_skb_reason(skb, reason); return NET_RX_DROP; @@ -4858,6 +4914,12 @@ u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp, xdp_init_buff(xdp, frame_sz, &rxqueue->xdp_rxq); xdp_prepare_buff(xdp, hard_start, skb_headroom(skb) - mac_len, skb_headlen(skb) + mac_len, true); + if (skb_is_nonlinear(skb)) { + skb_shinfo(skb)->xdp_frags_size = skb->data_len; + xdp_buff_set_frags_flag(xdp); + } else { + xdp_buff_clear_frags_flag(xdp); + } orig_data_end = xdp->data_end; orig_data = xdp->data; @@ -4887,6 +4949,14 @@ u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp, skb->len += off; /* positive on grow, negative on shrink */ } + /* XDP frag metadata (e.g. nr_frags) are updated in eBPF helpers + * (e.g. bpf_xdp_adjust_tail), we need to update data_len here. + */ + if (xdp_buff_has_frags(xdp)) + skb->data_len = skb_shinfo(skb)->xdp_frags_size; + else + skb->data_len = 0; + /* check if XDP changed eth hdr such SKB needs update */ eth = (struct ethhdr *)xdp->data; if ((orig_eth_type != eth->h_proto) || @@ -4920,11 +4990,35 @@ u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp, return act; } -static u32 netif_receive_generic_xdp(struct sk_buff *skb, +static int +netif_skb_check_for_xdp(struct sk_buff **pskb, struct bpf_prog *prog) +{ + struct sk_buff *skb = *pskb; + int err, hroom, troom; + + if (!skb_cow_data_for_xdp(this_cpu_read(system_page_pool), pskb, prog)) + return 0; + + /* In case we have to go down the path and also linearize, + * then lets do the pskb_expand_head() work just once here. + */ + hroom = XDP_PACKET_HEADROOM - skb_headroom(skb); + troom = skb->tail + skb->data_len - skb->end; + err = pskb_expand_head(skb, + hroom > 0 ? ALIGN(hroom, NET_SKB_PAD) : 0, + troom > 0 ? troom + 128 : 0, GFP_ATOMIC); + if (err) + return err; + + return skb_linearize(skb); +} + +static u32 netif_receive_generic_xdp(struct sk_buff **pskb, struct xdp_buff *xdp, struct bpf_prog *xdp_prog) { - u32 act = XDP_DROP; + struct sk_buff *skb = *pskb; + u32 mac_len, act = XDP_DROP; /* Reinjected packets coming from act_mirred or similar should * not get XDP generic processing. @@ -4932,41 +5026,36 @@ static u32 netif_receive_generic_xdp(struct sk_buff *skb, if (skb_is_redirected(skb)) return XDP_PASS; - /* XDP packets must be linear and must have sufficient headroom - * of XDP_PACKET_HEADROOM bytes. This is the guarantee that also - * native XDP provides, thus we need to do it here as well. + /* XDP packets must have sufficient headroom of XDP_PACKET_HEADROOM + * bytes. This is the guarantee that also native XDP provides, + * thus we need to do it here as well. */ + mac_len = skb->data - skb_mac_header(skb); + __skb_push(skb, mac_len); + if (skb_cloned(skb) || skb_is_nonlinear(skb) || skb_headroom(skb) < XDP_PACKET_HEADROOM) { - int hroom = XDP_PACKET_HEADROOM - skb_headroom(skb); - int troom = skb->tail + skb->data_len - skb->end; - - /* In case we have to go down the path and also linearize, - * then lets do the pskb_expand_head() work just once here. - */ - if (pskb_expand_head(skb, - hroom > 0 ? ALIGN(hroom, NET_SKB_PAD) : 0, - troom > 0 ? troom + 128 : 0, GFP_ATOMIC)) - goto do_drop; - if (skb_linearize(skb)) + if (netif_skb_check_for_xdp(pskb, xdp_prog)) goto do_drop; } - act = bpf_prog_run_generic_xdp(skb, xdp, xdp_prog); + __skb_pull(*pskb, mac_len); + + act = bpf_prog_run_generic_xdp(*pskb, xdp, xdp_prog); switch (act) { case XDP_REDIRECT: case XDP_TX: case XDP_PASS: break; default: - bpf_warn_invalid_xdp_action(skb->dev, xdp_prog, act); + bpf_warn_invalid_xdp_action((*pskb)->dev, xdp_prog, act); fallthrough; case XDP_ABORTED: - trace_xdp_exception(skb->dev, xdp_prog, act); + trace_xdp_exception((*pskb)->dev, xdp_prog, act); fallthrough; case XDP_DROP: do_drop: - kfree_skb(skb); + kfree_skb(*pskb); break; } @@ -5004,24 +5093,24 @@ void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog) static DEFINE_STATIC_KEY_FALSE(generic_xdp_needed_key); -int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb) +int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff **pskb) { if (xdp_prog) { struct xdp_buff xdp; u32 act; int err; - act = netif_receive_generic_xdp(skb, &xdp, xdp_prog); + act = netif_receive_generic_xdp(pskb, &xdp, xdp_prog); if (act != XDP_PASS) { switch (act) { case XDP_REDIRECT: - err = xdp_do_generic_redirect(skb->dev, skb, + err = xdp_do_generic_redirect((*pskb)->dev, *pskb, &xdp, xdp_prog); if (err) goto out_redir; break; case XDP_TX: - generic_xdp_tx(skb, xdp_prog); + generic_xdp_tx(*pskb, xdp_prog); break; } return XDP_DROP; @@ -5029,7 +5118,7 @@ int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb) } return XDP_PASS; out_redir: - kfree_skb_reason(skb, SKB_DROP_REASON_XDP); + kfree_skb_reason(*pskb, SKB_DROP_REASON_XDP); return XDP_DROP; } EXPORT_SYMBOL_GPL(do_xdp_generic); @@ -5038,7 +5127,7 @@ static int netif_rx_internal(struct sk_buff *skb) { int ret; - net_timestamp_check(READ_ONCE(netdev_tstamp_prequeue), skb); + net_timestamp_check(READ_ONCE(net_hotdata.tstamp_prequeue), skb); trace_netif_rx(skb); @@ -5330,7 +5419,7 @@ static int __netif_receive_skb_core(struct sk_buff **pskb, bool pfmemalloc, int ret = NET_RX_DROP; __be16 type; - net_timestamp_check(!READ_ONCE(netdev_tstamp_prequeue), skb); + net_timestamp_check(!READ_ONCE(net_hotdata.tstamp_prequeue), skb); trace_netif_receive_skb(skb); @@ -5352,7 +5441,8 @@ another_round: int ret2; migrate_disable(); - ret2 = do_xdp_generic(rcu_dereference(skb->dev->xdp_prog), skb); + ret2 = do_xdp_generic(rcu_dereference(skb->dev->xdp_prog), + &skb); migrate_enable(); if (ret2 != XDP_PASS) { @@ -5373,7 +5463,7 @@ another_round: if (pfmemalloc) goto skip_taps; - list_for_each_entry_rcu(ptype, &ptype_all, list) { + list_for_each_entry_rcu(ptype, &net_hotdata.ptype_all, list) { if (pt_prev) ret = deliver_skb(skb, pt_prev, orig_dev); pt_prev = ptype; @@ -5713,7 +5803,7 @@ static int netif_receive_skb_internal(struct sk_buff *skb) { int ret; - net_timestamp_check(READ_ONCE(netdev_tstamp_prequeue), skb); + net_timestamp_check(READ_ONCE(net_hotdata.tstamp_prequeue), skb); if (skb_defer_rx_timestamp(skb)) return NET_RX_SUCCESS; @@ -5743,7 +5833,8 @@ void netif_receive_skb_list_internal(struct list_head *head) INIT_LIST_HEAD(&sublist); list_for_each_entry_safe(skb, next, head, list) { - net_timestamp_check(READ_ONCE(netdev_tstamp_prequeue), skb); + net_timestamp_check(READ_ONCE(net_hotdata.tstamp_prequeue), + skb); skb_list_del_init(skb); if (!skb_defer_rx_timestamp(skb)) list_add_tail(&skb->list, &sublist); @@ -5833,21 +5924,21 @@ static void flush_backlog(struct work_struct *work) local_bh_disable(); sd = this_cpu_ptr(&softnet_data); - rps_lock_irq_disable(sd); + backlog_lock_irq_disable(sd); skb_queue_walk_safe(&sd->input_pkt_queue, skb, tmp) { if (skb->dev->reg_state == NETREG_UNREGISTERING) { __skb_unlink(skb, &sd->input_pkt_queue); dev_kfree_skb_irq(skb); - input_queue_head_incr(sd); + rps_input_queue_head_incr(sd); } } - rps_unlock_irq_enable(sd); + backlog_unlock_irq_enable(sd); skb_queue_walk_safe(&sd->process_queue, skb, tmp) { if (skb->dev->reg_state == NETREG_UNREGISTERING) { __skb_unlink(skb, &sd->process_queue); kfree_skb(skb); - input_queue_head_incr(sd); + rps_input_queue_head_incr(sd); } } local_bh_enable(); @@ -5859,14 +5950,14 @@ static bool flush_required(int cpu) struct softnet_data *sd = &per_cpu(softnet_data, cpu); bool do_flush; - rps_lock_irq_disable(sd); + backlog_lock_irq_disable(sd); /* as insertion into process_queue happens with the rps lock held, * process_queue access may race only with dequeue */ do_flush = !skb_queue_empty(&sd->input_pkt_queue) || !skb_queue_empty_lockless(&sd->process_queue); - rps_unlock_irq_enable(sd); + backlog_unlock_irq_enable(sd); return do_flush; #endif @@ -5932,7 +6023,7 @@ static void net_rps_action_and_irq_enable(struct softnet_data *sd) #ifdef CONFIG_RPS struct softnet_data *remsd = sd->rps_ipi_list; - if (remsd) { + if (!use_backlog_threads() && remsd) { sd->rps_ipi_list = NULL; local_irq_enable(); @@ -5947,7 +6038,7 @@ static void net_rps_action_and_irq_enable(struct softnet_data *sd) static bool sd_has_rps_ipi_waiting(struct softnet_data *sd) { #ifdef CONFIG_RPS - return sd->rps_ipi_list != NULL; + return !use_backlog_threads() && sd->rps_ipi_list; #else return false; #endif @@ -5967,7 +6058,7 @@ static int process_backlog(struct napi_struct *napi, int quota) net_rps_action_and_irq_enable(sd); } - napi->weight = READ_ONCE(dev_rx_weight); + napi->weight = READ_ONCE(net_hotdata.dev_rx_weight); while (again) { struct sk_buff *skb; @@ -5975,13 +6066,14 @@ static int process_backlog(struct napi_struct *napi, int quota) rcu_read_lock(); __netif_receive_skb(skb); rcu_read_unlock(); - input_queue_head_incr(sd); - if (++work >= quota) + if (++work >= quota) { + rps_input_queue_head_add(sd, work); return work; + } } - rps_lock_irq_disable(sd); + backlog_lock_irq_disable(sd); if (skb_queue_empty(&sd->input_pkt_queue)) { /* * Inline a custom version of __napi_complete(). @@ -5991,15 +6083,17 @@ static int process_backlog(struct napi_struct *napi, int quota) * We can use a plain write instead of clear_bit(), * and we dont need an smp_mb() memory barrier. */ - napi->state = 0; + napi->state &= NAPIF_STATE_THREADED; again = false; } else { skb_queue_splice_tail_init(&sd->input_pkt_queue, &sd->process_queue); } - rps_unlock_irq_enable(sd); + backlog_unlock_irq_enable(sd); } + if (work) + rps_input_queue_head_add(sd, work); return work; } @@ -6156,6 +6250,27 @@ struct napi_struct *napi_by_id(unsigned int napi_id) return NULL; } +static void skb_defer_free_flush(struct softnet_data *sd) +{ + struct sk_buff *skb, *next; + + /* Paired with WRITE_ONCE() in skb_attempt_defer_free() */ + if (!READ_ONCE(sd->defer_list)) + return; + + spin_lock(&sd->defer_lock); + skb = sd->defer_list; + sd->defer_list = NULL; + sd->defer_count = 0; + spin_unlock(&sd->defer_lock); + + while (skb != NULL) { + next = skb->next; + napi_consume_skb(skb, 1); + skb = next; + } +} + #if defined(CONFIG_NET_RX_BUSY_POLL) static void __busy_poll_stop(struct napi_struct *napi, bool skip_schedule) @@ -6177,8 +6292,13 @@ static void __busy_poll_stop(struct napi_struct *napi, bool skip_schedule) clear_bit(NAPI_STATE_SCHED, &napi->state); } -static void busy_poll_stop(struct napi_struct *napi, void *have_poll_lock, bool prefer_busy_poll, - u16 budget) +enum { + NAPI_F_PREFER_BUSY_POLL = 1, + NAPI_F_END_ON_RESCHED = 2, +}; + +static void busy_poll_stop(struct napi_struct *napi, void *have_poll_lock, + unsigned flags, u16 budget) { bool skip_schedule = false; unsigned long timeout; @@ -6198,7 +6318,7 @@ static void busy_poll_stop(struct napi_struct *napi, void *have_poll_lock, bool local_bh_disable(); - if (prefer_busy_poll) { + if (flags & NAPI_F_PREFER_BUSY_POLL) { napi->defer_hard_irqs_count = READ_ONCE(napi->dev->napi_defer_hard_irqs); timeout = READ_ONCE(napi->dev->gro_flush_timeout); if (napi->defer_hard_irqs_count && timeout) { @@ -6222,23 +6342,23 @@ static void busy_poll_stop(struct napi_struct *napi, void *have_poll_lock, bool local_bh_enable(); } -void napi_busy_loop(unsigned int napi_id, - bool (*loop_end)(void *, unsigned long), - void *loop_end_arg, bool prefer_busy_poll, u16 budget) +static void __napi_busy_loop(unsigned int napi_id, + bool (*loop_end)(void *, unsigned long), + void *loop_end_arg, unsigned flags, u16 budget) { unsigned long start_time = loop_end ? busy_loop_current_time() : 0; int (*napi_poll)(struct napi_struct *napi, int budget); void *have_poll_lock = NULL; struct napi_struct *napi; + WARN_ON_ONCE(!rcu_read_lock_held()); + restart: napi_poll = NULL; - rcu_read_lock(); - napi = napi_by_id(napi_id); if (!napi) - goto out; + return; if (!IS_ENABLED(CONFIG_PREEMPT_RT)) preempt_disable(); @@ -6254,14 +6374,14 @@ restart: */ if (val & (NAPIF_STATE_DISABLE | NAPIF_STATE_SCHED | NAPIF_STATE_IN_BUSY_POLL)) { - if (prefer_busy_poll) + if (flags & NAPI_F_PREFER_BUSY_POLL) set_bit(NAPI_STATE_PREFER_BUSY_POLL, &napi->state); goto count; } if (cmpxchg(&napi->state, val, val | NAPIF_STATE_IN_BUSY_POLL | NAPIF_STATE_SCHED) != val) { - if (prefer_busy_poll) + if (flags & NAPI_F_PREFER_BUSY_POLL) set_bit(NAPI_STATE_PREFER_BUSY_POLL, &napi->state); goto count; } @@ -6275,18 +6395,22 @@ count: if (work > 0) __NET_ADD_STATS(dev_net(napi->dev), LINUX_MIB_BUSYPOLLRXPACKETS, work); + skb_defer_free_flush(this_cpu_ptr(&softnet_data)); local_bh_enable(); if (!loop_end || loop_end(loop_end_arg, start_time)) break; if (unlikely(need_resched())) { + if (flags & NAPI_F_END_ON_RESCHED) + break; if (napi_poll) - busy_poll_stop(napi, have_poll_lock, prefer_busy_poll, budget); + busy_poll_stop(napi, have_poll_lock, flags, budget); if (!IS_ENABLED(CONFIG_PREEMPT_RT)) preempt_enable(); rcu_read_unlock(); cond_resched(); + rcu_read_lock(); if (loop_end(loop_end_arg, start_time)) return; goto restart; @@ -6294,10 +6418,31 @@ count: cpu_relax(); } if (napi_poll) - busy_poll_stop(napi, have_poll_lock, prefer_busy_poll, budget); + busy_poll_stop(napi, have_poll_lock, flags, budget); if (!IS_ENABLED(CONFIG_PREEMPT_RT)) preempt_enable(); -out: +} + +void napi_busy_loop_rcu(unsigned int napi_id, + bool (*loop_end)(void *, unsigned long), + void *loop_end_arg, bool prefer_busy_poll, u16 budget) +{ + unsigned flags = NAPI_F_END_ON_RESCHED; + + if (prefer_busy_poll) + flags |= NAPI_F_PREFER_BUSY_POLL; + + __napi_busy_loop(napi_id, loop_end, loop_end_arg, flags, budget); +} + +void napi_busy_loop(unsigned int napi_id, + bool (*loop_end)(void *, unsigned long), + void *loop_end_arg, bool prefer_busy_poll, u16 budget) +{ + unsigned flags = prefer_busy_poll ? NAPI_F_PREFER_BUSY_POLL : 0; + + rcu_read_lock(); + __napi_busy_loop(napi_id, loop_end, loop_end_arg, flags, budget); rcu_read_unlock(); } EXPORT_SYMBOL(napi_busy_loop); @@ -6385,7 +6530,7 @@ int dev_set_threaded(struct net_device *dev, bool threaded) } } - dev->threaded = threaded; + WRITE_ONCE(dev->threaded, threaded); /* Make sure kthread is created before THREADED bit * is set. @@ -6476,7 +6621,7 @@ void netif_napi_add_weight(struct net_device *dev, struct napi_struct *napi, * threaded mode will not be enabled in napi_enable(). */ if (dev->threaded && napi_kthread_create(napi)) - dev->threaded = 0; + dev->threaded = false; netif_napi_set_irq(napi, -1); } EXPORT_SYMBOL(netif_napi_add_weight); @@ -6654,8 +6799,6 @@ static int napi_poll(struct napi_struct *n, struct list_head *repoll) static int napi_thread_wait(struct napi_struct *napi) { - bool woken = false; - set_current_state(TASK_INTERRUPTIBLE); while (!kthread_should_stop()) { @@ -6664,15 +6807,13 @@ static int napi_thread_wait(struct napi_struct *napi) * Testing SCHED bit is not enough because SCHED bit might be * set by some other busy poll thread or by napi_disable(). */ - if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state) || woken) { + if (test_bit(NAPI_STATE_SCHED_THREADED, &napi->state)) { WARN_ON(!list_empty(&napi->poll_list)); __set_current_state(TASK_RUNNING); return 0; } schedule(); - /* woken being true indicates this thread owns this napi. */ - woken = true; set_current_state(TASK_INTERRUPTIBLE); } __set_current_state(TASK_RUNNING); @@ -6680,61 +6821,48 @@ static int napi_thread_wait(struct napi_struct *napi) return -1; } -static void skb_defer_free_flush(struct softnet_data *sd) +static void napi_threaded_poll_loop(struct napi_struct *napi) { - struct sk_buff *skb, *next; + struct softnet_data *sd; + unsigned long last_qs = jiffies; - /* Paired with WRITE_ONCE() in skb_attempt_defer_free() */ - if (!READ_ONCE(sd->defer_list)) - return; + for (;;) { + bool repoll = false; + void *have; - spin_lock(&sd->defer_lock); - skb = sd->defer_list; - sd->defer_list = NULL; - sd->defer_count = 0; - spin_unlock(&sd->defer_lock); + local_bh_disable(); + sd = this_cpu_ptr(&softnet_data); + sd->in_napi_threaded_poll = true; - while (skb != NULL) { - next = skb->next; - napi_consume_skb(skb, 1); - skb = next; + have = netpoll_poll_lock(napi); + __napi_poll(napi, &repoll); + netpoll_poll_unlock(have); + + sd->in_napi_threaded_poll = false; + barrier(); + + if (sd_has_rps_ipi_waiting(sd)) { + local_irq_disable(); + net_rps_action_and_irq_enable(sd); + } + skb_defer_free_flush(sd); + local_bh_enable(); + + if (!repoll) + break; + + rcu_softirq_qs_periodic(last_qs); + cond_resched(); } } static int napi_threaded_poll(void *data) { struct napi_struct *napi = data; - struct softnet_data *sd; - void *have; - - while (!napi_thread_wait(napi)) { - for (;;) { - bool repoll = false; - - local_bh_disable(); - sd = this_cpu_ptr(&softnet_data); - sd->in_napi_threaded_poll = true; - - have = netpoll_poll_lock(napi); - __napi_poll(napi, &repoll); - netpoll_poll_unlock(have); - sd->in_napi_threaded_poll = false; - barrier(); - - if (sd_has_rps_ipi_waiting(sd)) { - local_irq_disable(); - net_rps_action_and_irq_enable(sd); - } - skb_defer_free_flush(sd); - local_bh_enable(); - - if (!repoll) - break; + while (!napi_thread_wait(napi)) + napi_threaded_poll_loop(napi); - cond_resched(); - } - } return 0; } @@ -6742,8 +6870,8 @@ static __latent_entropy void net_rx_action(struct softirq_action *h) { struct softnet_data *sd = this_cpu_ptr(&softnet_data); unsigned long time_limit = jiffies + - usecs_to_jiffies(READ_ONCE(netdev_budget_usecs)); - int budget = READ_ONCE(netdev_budget); + usecs_to_jiffies(READ_ONCE(net_hotdata.netdev_budget_usecs)); + int budget = READ_ONCE(net_hotdata.netdev_budget); LIST_HEAD(list); LIST_HEAD(repoll); @@ -8415,27 +8543,29 @@ static void dev_change_rx_flags(struct net_device *dev, int flags) static int __dev_set_promiscuity(struct net_device *dev, int inc, bool notify) { unsigned int old_flags = dev->flags; + unsigned int promiscuity, flags; kuid_t uid; kgid_t gid; ASSERT_RTNL(); - dev->flags |= IFF_PROMISC; - dev->promiscuity += inc; - if (dev->promiscuity == 0) { + promiscuity = dev->promiscuity + inc; + if (promiscuity == 0) { /* * Avoid overflow. * If inc causes overflow, untouch promisc and return error. */ - if (inc < 0) - dev->flags &= ~IFF_PROMISC; - else { - dev->promiscuity -= inc; + if (unlikely(inc > 0)) { netdev_warn(dev, "promiscuity touches roof, set promiscuity failed. promiscuity feature of device might be broken.\n"); return -EOVERFLOW; } + flags = old_flags & ~IFF_PROMISC; + } else { + flags = old_flags | IFF_PROMISC; } - if (dev->flags != old_flags) { + WRITE_ONCE(dev->promiscuity, promiscuity); + if (flags != old_flags) { + WRITE_ONCE(dev->flags, flags); netdev_info(dev, "%s promiscuous mode\n", dev->flags & IFF_PROMISC ? "entered" : "left"); if (audit_enabled) { @@ -8486,25 +8616,27 @@ EXPORT_SYMBOL(dev_set_promiscuity); static int __dev_set_allmulti(struct net_device *dev, int inc, bool notify) { unsigned int old_flags = dev->flags, old_gflags = dev->gflags; + unsigned int allmulti, flags; ASSERT_RTNL(); - dev->flags |= IFF_ALLMULTI; - dev->allmulti += inc; - if (dev->allmulti == 0) { + allmulti = dev->allmulti + inc; + if (allmulti == 0) { /* * Avoid overflow. * If inc causes overflow, untouch allmulti and return error. */ - if (inc < 0) - dev->flags &= ~IFF_ALLMULTI; - else { - dev->allmulti -= inc; + if (unlikely(inc > 0)) { netdev_warn(dev, "allmulti touches roof, set allmulti failed. allmulti feature of device might be broken.\n"); return -EOVERFLOW; } + flags = old_flags & ~IFF_ALLMULTI; + } else { + flags = old_flags | IFF_ALLMULTI; } - if (dev->flags ^ old_flags) { + WRITE_ONCE(dev->allmulti, allmulti); + if (flags != old_flags) { + WRITE_ONCE(dev->flags, flags); netdev_info(dev, "%s allmulticast mode\n", dev->flags & IFF_ALLMULTI ? "entered" : "left"); dev_change_rx_flags(dev, IFF_ALLMULTI); @@ -8586,12 +8718,12 @@ unsigned int dev_get_flags(const struct net_device *dev) { unsigned int flags; - flags = (dev->flags & ~(IFF_PROMISC | + flags = (READ_ONCE(dev->flags) & ~(IFF_PROMISC | IFF_ALLMULTI | IFF_RUNNING | IFF_LOWER_UP | IFF_DORMANT)) | - (dev->gflags & (IFF_PROMISC | + (READ_ONCE(dev->gflags) & (IFF_PROMISC | IFF_ALLMULTI)); if (netif_running(dev)) { @@ -8830,7 +8962,7 @@ int dev_change_tx_queue_len(struct net_device *dev, unsigned long new_len) return -ERANGE; if (new_len != orig_len) { - dev->tx_queue_len = new_len; + WRITE_ONCE(dev->tx_queue_len, new_len); res = call_netdevice_notifiers(NETDEV_CHANGE_TX_QUEUE_LEN, dev); res = notifier_to_errno(res); if (res) @@ -8844,7 +8976,7 @@ int dev_change_tx_queue_len(struct net_device *dev, unsigned long new_len) err_rollback: netdev_err(dev, "refused to change device tx_queue_len\n"); - dev->tx_queue_len = orig_len; + WRITE_ONCE(dev->tx_queue_len, orig_len); return res; } @@ -8914,7 +9046,7 @@ int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa, } EXPORT_SYMBOL(dev_set_mac_address); -static DECLARE_RWSEM(dev_addr_sem); +DECLARE_RWSEM(dev_addr_sem); int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa, struct netlink_ext_ack *extack) @@ -9090,7 +9222,7 @@ int dev_change_proto_down(struct net_device *dev, bool proto_down) netif_carrier_off(dev); else netif_carrier_on(dev); - dev->proto_down = proto_down; + WRITE_ONCE(dev->proto_down, proto_down); return 0; } @@ -9104,18 +9236,21 @@ int dev_change_proto_down(struct net_device *dev, bool proto_down) void dev_change_proto_down_reason(struct net_device *dev, unsigned long mask, u32 value) { + u32 proto_down_reason; int b; if (!mask) { - dev->proto_down_reason = value; + proto_down_reason = value; } else { + proto_down_reason = dev->proto_down_reason; for_each_set_bit(b, &mask, 32) { if (value & (1 << b)) - dev->proto_down_reason |= BIT(b); + proto_down_reason |= BIT(b); else - dev->proto_down_reason &= ~BIT(b); + proto_down_reason &= ~BIT(b); } } + WRITE_ONCE(dev->proto_down_reason, proto_down_reason); } struct bpf_xdp_link { @@ -9668,11 +9803,11 @@ static void dev_index_release(struct net *net, int ifindex) /* Delayed registration/unregisteration */ LIST_HEAD(net_todo_list); DECLARE_WAIT_QUEUE_HEAD(netdev_unregistering_wq); +atomic_t dev_unreg_count = ATOMIC_INIT(0); static void net_set_todo(struct net_device *dev) { list_add_tail(&dev->todo_list, &net_todo_list); - atomic_inc(&dev_net(dev)->dev_unreg_count); } static netdev_features_t netdev_sync_upper_features(struct net_device *lower, @@ -10237,9 +10372,9 @@ int register_netdevice(struct net_device *dev) goto err_ifindex_release; ret = netdev_register_kobject(dev); - write_lock(&dev_base_lock); - dev->reg_state = ret ? NETREG_UNREGISTERED : NETREG_REGISTERED; - write_unlock(&dev_base_lock); + + WRITE_ONCE(dev->reg_state, ret ? NETREG_UNREGISTERED : NETREG_REGISTERED); + if (ret) goto err_uninit_notify; @@ -10305,25 +10440,12 @@ err_free_name: } EXPORT_SYMBOL(register_netdevice); -/** - * init_dummy_netdev - init a dummy network device for NAPI - * @dev: device to init - * - * This takes a network device structure and initialize the minimum - * amount of fields so it can be used to schedule NAPI polls without - * registering a full blown interface. This is to be used by drivers - * that need to tie several hardware interfaces to a single NAPI - * poll scheduler due to HW limitations. +/* Initialize the core of a dummy net device. + * This is useful if you are calling this function after alloc_netdev(), + * since it does not memset the net_device fields. */ -int init_dummy_netdev(struct net_device *dev) +static void init_dummy_netdev_core(struct net_device *dev) { - /* Clear everything. Note we don't initialize spinlocks - * are they aren't supposed to be taken by any of the - * NAPI code and this dummy netdev is supposed to be - * only ever used for NAPI polls - */ - memset(dev, 0, sizeof(struct net_device)); - /* make sure we BUG if trying to hit standard * register/unregister code path */ @@ -10343,12 +10465,30 @@ int init_dummy_netdev(struct net_device *dev) * because users of this 'device' dont need to change * its refcount. */ +} - return 0; +/** + * init_dummy_netdev - init a dummy network device for NAPI + * @dev: device to init + * + * This takes a network device structure and initializes the minimum + * amount of fields so it can be used to schedule NAPI polls without + * registering a full blown interface. This is to be used by drivers + * that need to tie several hardware interfaces to a single NAPI + * poll scheduler due to HW limitations. + */ +void init_dummy_netdev(struct net_device *dev) +{ + /* Clear everything. Note we don't initialize spinlocks + * as they aren't supposed to be taken by any of the + * NAPI code and this dummy netdev is supposed to be + * only ever used for NAPI polls + */ + memset(dev, 0, sizeof(struct net_device)); + init_dummy_netdev_core(dev); } EXPORT_SYMBOL_GPL(init_dummy_netdev); - /** * register_netdev - register a network device * @dev: device to register @@ -10446,8 +10586,9 @@ static struct net_device *netdev_wait_allrefs_any(struct list_head *list) rebroadcast_time = jiffies; } + rcu_barrier(); + if (!wait) { - rcu_barrier(); wait = WAIT_REFS_MIN_MSECS; } else { msleep(wait); @@ -10499,6 +10640,7 @@ void netdev_run_todo(void) { struct net_device *dev, *tmp; struct list_head list; + int cnt; #ifdef CONFIG_LOCKDEP struct list_head unlink_list; @@ -10529,12 +10671,11 @@ void netdev_run_todo(void) continue; } - write_lock(&dev_base_lock); - dev->reg_state = NETREG_UNREGISTERED; - write_unlock(&dev_base_lock); + WRITE_ONCE(dev->reg_state, NETREG_UNREGISTERED); linkwatch_sync_dev(dev); } + cnt = 0; while (!list_empty(&list)) { dev = netdev_wait_allrefs_any(&list); list_del(&dev->todo_list); @@ -10552,12 +10693,13 @@ void netdev_run_todo(void) if (dev->needs_free_netdev) free_netdev(dev); - if (atomic_dec_and_test(&dev_net(dev)->dev_unreg_count)) - wake_up(&netdev_unregistering_wq); + cnt++; /* Free network device */ kobject_put(&dev->dev.kobj); } + if (cnt && atomic_sub_and_test(cnt, &dev_unreg_count)) + wake_up(&netdev_unregistering_wq); } /* Convert net_device_stats to rtnl_link_stats64. rtnl_link_stats64 has @@ -10634,6 +10776,8 @@ struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev, ops->ndo_get_stats64(dev, storage); } else if (ops->ndo_get_stats) { netdev_stats_to_stats64(storage, ops->ndo_get_stats(dev)); + } else if (dev->pcpu_stat_type == NETDEV_PCPU_STAT_TSTATS) { + dev_get_tstats64(dev, storage); } else { netdev_stats_to_stats64(storage, &dev->stats); } @@ -10942,13 +11086,14 @@ void free_netdev(struct net_device *dev) dev->xdp_bulkq = NULL; /* Compatibility with error handling in drivers */ - if (dev->reg_state == NETREG_UNINITIALIZED) { + if (dev->reg_state == NETREG_UNINITIALIZED || + dev->reg_state == NETREG_DUMMY) { netdev_freemem(dev); return; } BUG_ON(dev->reg_state != NETREG_UNREGISTERED); - dev->reg_state = NETREG_RELEASED; + WRITE_ONCE(dev->reg_state, NETREG_RELEASED); /* will free via device release */ put_device(&dev->dev); @@ -10956,6 +11101,19 @@ void free_netdev(struct net_device *dev) EXPORT_SYMBOL(free_netdev); /** + * alloc_netdev_dummy - Allocate and initialize a dummy net device. + * @sizeof_priv: size of private data to allocate space for + * + * Return: the allocated net_device on success, NULL otherwise + */ +struct net_device *alloc_netdev_dummy(int sizeof_priv) +{ + return alloc_netdev(sizeof_priv, "dummy#", NET_NAME_UNKNOWN, + init_dummy_netdev_core); +} +EXPORT_SYMBOL_GPL(alloc_netdev_dummy); + +/** * synchronize_net - Synchronize with packet receive processing * * Wait for packets currently being received to be done. @@ -11004,6 +11162,7 @@ void unregister_netdevice_many_notify(struct list_head *head, { struct net_device *dev, *tmp; LIST_HEAD(close_head); + int cnt = 0; BUG_ON(dev_boot_phase); ASSERT_RTNL(); @@ -11035,10 +11194,8 @@ void unregister_netdevice_many_notify(struct list_head *head, list_for_each_entry(dev, head, unreg_list) { /* And unlink it from device chain. */ - write_lock(&dev_base_lock); - unlist_netdevice(dev, false); - dev->reg_state = NETREG_UNREGISTERING; - write_unlock(&dev_base_lock); + unlist_netdevice(dev); + WRITE_ONCE(dev->reg_state, NETREG_UNREGISTERING); } flush_all_backlogs(); @@ -11100,7 +11257,9 @@ void unregister_netdevice_many_notify(struct list_head *head, list_for_each_entry(dev, head, unreg_list) { netdev_put(dev, &dev->dev_registered_tracker); net_set_todo(dev); + cnt++; } + atomic_add(cnt, &dev_unreg_count); list_del(head); } @@ -11218,7 +11377,7 @@ int __dev_change_net_namespace(struct net_device *dev, struct net *net, dev_close(dev); /* And unlink it from device chain */ - unlist_netdevice(dev, true); + unlist_netdevice(dev); synchronize_net(); @@ -11257,8 +11416,12 @@ int __dev_change_net_namespace(struct net_device *dev, struct net *net, dev_net_set(dev, net); dev->ifindex = new_ifindex; - if (new_name[0]) /* Rename the netdev to prepared name */ + if (new_name[0]) { + /* Rename the netdev to prepared name */ + write_seqlock(&netdev_rename_lock); strscpy(dev->name, new_name, IFNAMSIZ); + write_sequnlock(&netdev_rename_lock); + } /* Fixup kobjects */ dev_set_uevent_suppress(&dev->dev, 1); @@ -11333,7 +11496,7 @@ static int dev_cpu_dead(unsigned int oldcpu) list_del_init(&napi->poll_list); if (napi->poll == process_backlog) - napi->state = 0; + napi->state &= NAPIF_STATE_THREADED; else ____napi_schedule(sd, napi); } @@ -11341,21 +11504,23 @@ static int dev_cpu_dead(unsigned int oldcpu) raise_softirq_irqoff(NET_TX_SOFTIRQ); local_irq_enable(); + if (!use_backlog_threads()) { #ifdef CONFIG_RPS - remsd = oldsd->rps_ipi_list; - oldsd->rps_ipi_list = NULL; + remsd = oldsd->rps_ipi_list; + oldsd->rps_ipi_list = NULL; #endif - /* send out pending IPI's on offline CPU */ - net_rps_send_ipi(remsd); + /* send out pending IPI's on offline CPU */ + net_rps_send_ipi(remsd); + } /* Process offline CPU's input_pkt_queue */ while ((skb = __skb_dequeue(&oldsd->process_queue))) { netif_rx(skb); - input_queue_head_incr(oldsd); + rps_input_queue_head_incr(oldsd); } while ((skb = skb_dequeue(&oldsd->input_pkt_queue))) { netif_rx(skb); - input_queue_head_incr(oldsd); + rps_input_queue_head_incr(oldsd); } return 0; @@ -11554,11 +11719,8 @@ static void __net_exit default_device_exit_net(struct net *net) snprintf(fb_name, IFNAMSIZ, "dev%%d"); netdev_for_each_altname_safe(dev, name_node, tmp) - if (netdev_name_in_use(&init_net, name_node->name)) { - netdev_name_node_del(name_node); - synchronize_rcu(); + if (netdev_name_in_use(&init_net, name_node->name)) __netdev_name_node_alt_destroy(name_node); - } err = dev_change_net_namespace(dev, &init_net, fb_name); if (err) { @@ -11631,11 +11793,12 @@ static void __init net_dev_struct_check(void) /* TXRX read-mostly hotpath */ CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, lstats); + CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, state); CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, flags); CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, hard_header_len); CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, features); CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_txrx, ip6_ptr); - CACHELINE_ASSERT_GROUP_SIZE(struct net_device, net_device_read_txrx, 38); + CACHELINE_ASSERT_GROUP_SIZE(struct net_device, net_device_read_txrx, 46); /* RX read-mostly hotpath */ CACHELINE_ASSERT_GROUP_MEMBER(struct net_device, net_device_read_rx, ptype_specific); @@ -11665,6 +11828,60 @@ static void __init net_dev_struct_check(void) * */ +/* We allocate 256 pages for each CPU if PAGE_SHIFT is 12 */ +#define SYSTEM_PERCPU_PAGE_POOL_SIZE ((1 << 20) / PAGE_SIZE) + +static int net_page_pool_create(int cpuid) +{ +#if IS_ENABLED(CONFIG_PAGE_POOL) + struct page_pool_params page_pool_params = { + .pool_size = SYSTEM_PERCPU_PAGE_POOL_SIZE, + .flags = PP_FLAG_SYSTEM_POOL, + .nid = cpu_to_mem(cpuid), + }; + struct page_pool *pp_ptr; + + pp_ptr = page_pool_create_percpu(&page_pool_params, cpuid); + if (IS_ERR(pp_ptr)) + return -ENOMEM; + + per_cpu(system_page_pool, cpuid) = pp_ptr; +#endif + return 0; +} + +static int backlog_napi_should_run(unsigned int cpu) +{ + struct softnet_data *sd = per_cpu_ptr(&softnet_data, cpu); + struct napi_struct *napi = &sd->backlog; + + return test_bit(NAPI_STATE_SCHED_THREADED, &napi->state); +} + +static void run_backlog_napi(unsigned int cpu) +{ + struct softnet_data *sd = per_cpu_ptr(&softnet_data, cpu); + + napi_threaded_poll_loop(&sd->backlog); +} + +static void backlog_napi_setup(unsigned int cpu) +{ + struct softnet_data *sd = per_cpu_ptr(&softnet_data, cpu); + struct napi_struct *napi = &sd->backlog; + + napi->thread = this_cpu_read(backlog_napi); + set_bit(NAPI_STATE_THREADED, &napi->state); +} + +static struct smp_hotplug_thread backlog_threads = { + .store = &backlog_napi, + .thread_should_run = backlog_napi_should_run, + .thread_fn = run_backlog_napi, + .thread_comm = "backlog_napi/%u", + .setup = backlog_napi_setup, +}; + /* * This is called single threaded during boot, so no need * to take the rtnl semaphore. @@ -11683,7 +11900,6 @@ static int __init net_dev_init(void) if (netdev_kobject_init()) goto out; - INIT_LIST_HEAD(&ptype_all); for (i = 0; i < PTYPE_HASH_SIZE; i++) INIT_LIST_HEAD(&ptype_base[i]); @@ -11717,7 +11933,13 @@ static int __init net_dev_init(void) init_gro_hash(&sd->backlog); sd->backlog.poll = process_backlog; sd->backlog.weight = weight_p; + INIT_LIST_HEAD(&sd->backlog.poll_list); + + if (net_page_pool_create(i)) + goto out; } + if (use_backlog_threads()) + smpboot_register_percpu_thread(&backlog_threads); dev_boot_phase = 0; @@ -11743,7 +11965,24 @@ static int __init net_dev_init(void) NULL, dev_cpu_dead); WARN_ON(rc < 0); rc = 0; + + /* avoid static key IPIs to isolated CPUs */ + if (housekeeping_enabled(HK_TYPE_MISC)) + net_enable_timestamp(); out: + if (rc < 0) { + for_each_possible_cpu(i) { + struct page_pool *pp_ptr; + + pp_ptr = per_cpu(system_page_pool, i); + if (!pp_ptr) + continue; + + page_pool_destroy(pp_ptr); + per_cpu(system_page_pool, i) = NULL; + } + } + return rc; } |