X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=net%2Fcore%2Fsock.c;h=3ca5f8d4c0197b54366955549aa56bf1c61fe86b;hb=97bf2856c6014879bd04983a3e9dfcdac1e7fe85;hp=cb5861e219d0b7fdd4c9a6440a4e5e017386f491;hpb=43bc926fffd92024b46cafaf7350d669ba9ca884;p=linux-2.6.git diff --git a/net/core/sock.c b/net/core/sock.c index cb5861e21..3ca5f8d4c 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -92,7 +92,6 @@ */ #include -#include #include #include #include @@ -112,6 +111,7 @@ #include #include #include +#include #include #include @@ -133,6 +133,53 @@ #include #endif +/* + * Each address family might have different locking rules, so we have + * one slock key per address family: + */ +static struct lock_class_key af_family_keys[AF_MAX]; +static struct lock_class_key af_family_slock_keys[AF_MAX]; + +#ifdef CONFIG_DEBUG_LOCK_ALLOC +/* + * Make lock validator output more readable. (we pre-construct these + * strings build-time, so that runtime initialization of socket + * locks is fast): + */ +static const char *af_family_key_strings[AF_MAX+1] = { + "sk_lock-AF_UNSPEC", "sk_lock-AF_UNIX" , "sk_lock-AF_INET" , + "sk_lock-AF_AX25" , "sk_lock-AF_IPX" , "sk_lock-AF_APPLETALK", + "sk_lock-AF_NETROM", "sk_lock-AF_BRIDGE" , "sk_lock-AF_ATMPVC" , + "sk_lock-AF_X25" , "sk_lock-AF_INET6" , "sk_lock-AF_ROSE" , + "sk_lock-AF_DECnet", "sk_lock-AF_NETBEUI" , "sk_lock-AF_SECURITY" , + "sk_lock-AF_KEY" , "sk_lock-AF_NETLINK" , "sk_lock-AF_PACKET" , + "sk_lock-AF_ASH" , "sk_lock-AF_ECONET" , "sk_lock-AF_ATMSVC" , + "sk_lock-21" , "sk_lock-AF_SNA" , "sk_lock-AF_IRDA" , + "sk_lock-AF_PPPOX" , "sk_lock-AF_WANPIPE" , "sk_lock-AF_LLC" , + "sk_lock-27" , "sk_lock-28" , "sk_lock-29" , + "sk_lock-AF_TIPC" , "sk_lock-AF_BLUETOOTH", "sk_lock-AF_MAX" +}; +static const char *af_family_slock_key_strings[AF_MAX+1] = { + "slock-AF_UNSPEC", "slock-AF_UNIX" , "slock-AF_INET" , + "slock-AF_AX25" , "slock-AF_IPX" , "slock-AF_APPLETALK", + "slock-AF_NETROM", "slock-AF_BRIDGE" , "slock-AF_ATMPVC" , + "slock-AF_X25" , "slock-AF_INET6" , "slock-AF_ROSE" , + "slock-AF_DECnet", "slock-AF_NETBEUI" , "slock-AF_SECURITY" , + "slock-AF_KEY" , "slock-AF_NETLINK" , "slock-AF_PACKET" , + "slock-AF_ASH" , "slock-AF_ECONET" , "slock-AF_ATMSVC" , + "slock-21" , "slock-AF_SNA" , "slock-AF_IRDA" , + "slock-AF_PPPOX" , "slock-AF_WANPIPE" , "slock-AF_LLC" , + "slock-27" , "slock-28" , "slock-29" , + "slock-AF_TIPC" , "slock-AF_BLUETOOTH", "slock-AF_MAX" +}; +#endif + +/* + * sk_callback_lock locking rules are per-address-family, + * so split the lock classes by using a per-AF key: + */ +static struct lock_class_key af_callback_keys[AF_MAX]; + /* Take into consideration the size of the struct sk_buff overhead in the * determination of these values, since that is non-constant across * platforms. This makes socket queueing behavior and performance @@ -144,13 +191,13 @@ #define SK_RMEM_MAX (_SK_MEM_OVERHEAD * _SK_MEM_PACKETS) /* Run time adjustable parameters. */ -__u32 sysctl_wmem_max = SK_WMEM_MAX; -__u32 sysctl_rmem_max = SK_RMEM_MAX; -__u32 sysctl_wmem_default = SK_WMEM_MAX; -__u32 sysctl_rmem_default = SK_RMEM_MAX; +__u32 sysctl_wmem_max __read_mostly = SK_WMEM_MAX; +__u32 sysctl_rmem_max __read_mostly = SK_RMEM_MAX; +__u32 sysctl_wmem_default __read_mostly = SK_WMEM_MAX; +__u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX; /* Maximal space eaten by iovec or ancilliary data plus some space */ -int sysctl_optmem_max = sizeof(unsigned long)*(2*UIO_MAXIOV + 512); +int sysctl_optmem_max __read_mostly = sizeof(unsigned long)*(2*UIO_MAXIOV+512); static int sock_set_timeout(long *timeo_p, char __user *optval, int optlen) { @@ -204,11 +251,7 @@ int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) goto out; } - /* It would be deadlock, if sock_queue_rcv_skb is used - with socket lock! We assume that users of this - function are lock free. - */ - err = sk_filter(sk, skb, 1); + err = sk_filter(sk, skb); if (err) goto out; @@ -231,19 +274,29 @@ out: } EXPORT_SYMBOL(sock_queue_rcv_skb); -int sk_receive_skb(struct sock *sk, struct sk_buff *skb) +int sk_receive_skb(struct sock *sk, struct sk_buff *skb, const int nested) { int rc = NET_RX_SUCCESS; - if (sk_filter(sk, skb, 0)) + if (sk_filter(sk, skb)) goto discard_and_relse; skb->dev = NULL; - bh_lock_sock(sk); - if (!sock_owned_by_user(sk)) - rc = sk->sk_backlog_rcv(sk, skb); + if (nested) + bh_lock_sock_nested(sk); else + bh_lock_sock(sk); + if (!sock_owned_by_user(sk)) { + /* + * trylock + unlock semantics: + */ + mutex_acquire(&sk->sk_lock.dep_map, 0, 1, _RET_IP_); + + rc = sk->sk_backlog_rcv(sk, skb); + + mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_); + } else sk_add_backlog(sk, skb); bh_unlock_sock(sk); out: @@ -556,18 +609,25 @@ set_rcvbuf: break; case SO_DETACH_FILTER: - spin_lock_bh(&sk->sk_lock.slock); - filter = sk->sk_filter; + rcu_read_lock_bh(); + filter = rcu_dereference(sk->sk_filter); if (filter) { - sk->sk_filter = NULL; - spin_unlock_bh(&sk->sk_lock.slock); + rcu_assign_pointer(sk->sk_filter, NULL); sk_filter_release(sk, filter); + rcu_read_unlock_bh(); break; } - spin_unlock_bh(&sk->sk_lock.slock); + rcu_read_unlock_bh(); ret = -ENONET; break; + case SO_PASSSEC: + if (valbool) + set_bit(SOCK_PASSSEC, &sock->flags); + else + clear_bit(SOCK_PASSSEC, &sock->flags); + break; + /* We implement the SO_SNDLOWAT etc to not be settable (1003.1g 5.3) */ default: @@ -726,6 +786,10 @@ int sock_getsockopt(struct socket *sock, int level, int optname, v.val = sk->sk_state == TCP_LISTEN; break; + case SO_PASSSEC: + v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0; + break; + case SO_PEERSEC: return security_socket_getpeersec_stream(sock, optval, optlen, len); @@ -742,6 +806,20 @@ lenout: return 0; } +/* + * Initialize an sk_lock. + * + * (We also register the sk_lock with the lock validator.) + */ +static void inline sock_lock_init(struct sock *sk) +{ + sock_lock_init_class_and_name(sk, + af_family_slock_key_strings[sk->sk_family], + af_family_slock_keys + sk->sk_family, + af_family_key_strings[sk->sk_family], + af_family_keys + sk->sk_family); +} + /** * sk_alloc - All socket objects are allocated here * @family: protocol family @@ -753,7 +831,7 @@ struct sock *sk_alloc(int family, gfp_t priority, struct proto *prot, int zero_it) { struct sock *sk = NULL; - kmem_cache_t *slab = prot->slab; + struct kmem_cache *slab = prot->slab; if (slab != NULL) sk = kmem_cache_alloc(slab, priority); @@ -798,10 +876,10 @@ void sk_free(struct sock *sk) if (sk->sk_destruct) sk->sk_destruct(sk); - filter = sk->sk_filter; + filter = rcu_dereference(sk->sk_filter); if (filter) { sk_filter_release(sk, filter); - sk->sk_filter = NULL; + rcu_assign_pointer(sk->sk_filter, NULL); } sock_disable_timestamp(sk); @@ -823,14 +901,14 @@ void sk_free(struct sock *sk) module_put(owner); } -struct sock *sk_clone(struct sock *sk, const gfp_t priority) +struct sock *sk_clone(const struct sock *sk, const gfp_t priority) { struct sock *newsk = sk_alloc(sk->sk_family, priority, sk->sk_prot, 0); if (newsk != NULL) { struct sk_filter *filter; - memcpy(newsk, sk, sk->sk_prot->obj_size); + sock_copy(newsk, sk); /* SANITY */ sock_vx_init(newsk); @@ -844,9 +922,14 @@ struct sock *sk_clone(struct sock *sk, const gfp_t priority) atomic_set(&newsk->sk_omem_alloc, 0); skb_queue_head_init(&newsk->sk_receive_queue); skb_queue_head_init(&newsk->sk_write_queue); +#ifdef CONFIG_NET_DMA + skb_queue_head_init(&newsk->sk_async_wait_queue); +#endif rwlock_init(&newsk->sk_dst_lock); rwlock_init(&newsk->sk_callback_lock); + lockdep_set_class(&newsk->sk_callback_lock, + af_callback_keys + newsk->sk_family); newsk->sk_dst_cache = NULL; newsk->sk_wmem_queued = 0; @@ -862,9 +945,6 @@ struct sock *sk_clone(struct sock *sk, const gfp_t priority) if (filter != NULL) sk_filter_charge(newsk, filter); - if (sk->sk_create_child) - sk->sk_create_child(sk, newsk); - if (unlikely(xfrm_sk_clone_policy(newsk))) { /* It is still raw copy of parent, so invalidate * destructor and make plain sk_free() */ @@ -1089,7 +1169,7 @@ static struct sk_buff *sock_alloc_send_pskb(struct sock *sk, goto failure; if (atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) { - skb = alloc_skb(header_len, sk->sk_allocation); + skb = alloc_skb(header_len, gfp_mask); if (skb) { int npages; int i; @@ -1404,6 +1484,9 @@ void sock_init_data(struct socket *sock, struct sock *sk) skb_queue_head_init(&sk->sk_receive_queue); skb_queue_head_init(&sk->sk_write_queue); skb_queue_head_init(&sk->sk_error_queue); +#ifdef CONFIG_NET_DMA + skb_queue_head_init(&sk->sk_async_wait_queue); +#endif sk->sk_send_head = NULL; @@ -1427,6 +1510,8 @@ void sock_init_data(struct socket *sock, struct sock *sk) rwlock_init(&sk->sk_dst_lock); rwlock_init(&sk->sk_callback_lock); + lockdep_set_class(&sk->sk_callback_lock, + af_callback_keys + sk->sk_family); sk->sk_state_change = sock_def_wakeup; sk->sk_data_ready = sock_def_readable; @@ -1456,27 +1541,37 @@ void sock_init_data(struct socket *sock, struct sock *sk) atomic_set(&sk->sk_refcnt, 1); } -void fastcall lock_sock(struct sock *sk) +void fastcall lock_sock_nested(struct sock *sk, int subclass) { might_sleep(); - spin_lock_bh(&(sk->sk_lock.slock)); + spin_lock_bh(&sk->sk_lock.slock); if (sk->sk_lock.owner) __lock_sock(sk); sk->sk_lock.owner = (void *)1; - spin_unlock_bh(&(sk->sk_lock.slock)); + spin_unlock(&sk->sk_lock.slock); + /* + * The sk_lock has mutex_lock() semantics here: + */ + mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_); + local_bh_enable(); } -EXPORT_SYMBOL(lock_sock); +EXPORT_SYMBOL(lock_sock_nested); void fastcall release_sock(struct sock *sk) { - spin_lock_bh(&(sk->sk_lock.slock)); + /* + * The sk_lock has mutex_unlock() semantics: + */ + mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_); + + spin_lock_bh(&sk->sk_lock.slock); if (sk->sk_backlog.tail) __release_sock(sk); sk->sk_lock.owner = NULL; - if (waitqueue_active(&(sk->sk_lock.wq))) - wake_up(&(sk->sk_lock.wq)); - spin_unlock_bh(&(sk->sk_lock.slock)); + if (waitqueue_active(&sk->sk_lock.wq)) + wake_up(&sk->sk_lock.wq); + spin_unlock_bh(&sk->sk_lock.slock); } EXPORT_SYMBOL(release_sock); @@ -1525,7 +1620,7 @@ int compat_sock_common_getsockopt(struct socket *sock, int level, int optname, { struct sock *sk = sock->sk; - if (sk->sk_prot->compat_setsockopt != NULL) + if (sk->sk_prot->compat_getsockopt != NULL) return sk->sk_prot->compat_getsockopt(sk, level, optname, optval, optlen); return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);