9 files changed, 159 insertions, 30 deletions

diff --git a/net/8021q/vlan.c b/net/8021q/vlan.c
index 3948949a609a..458031bfff55 100644
--- a/net/8021q/vlan.c
+++ b/net/8021q/vlan.c
@@ -364,6 +364,14 @@ static void vlan_transfer_operstate(const struct net_device *dev, struct net_dev
         }
 }
 
+/*
+ * vlan network devices have devices nesting below it, and are a special
+ * "super class" of normal network devices; split their locks off into a
+ * separate class since they always nest.
+ */
+static struct lock_class_key vlan_netdev_xmit_lock_key;
+
+
 /*  Attach a VLAN device to a mac address (ie Ethernet Card).
  *  Returns the device that was created, or NULL if there was
  *  an error of some kind.
@@ -460,6 +468,7 @@ static struct net_device *register_vlan_device(const char *eth_IF_name,
                     
         new_dev = alloc_netdev(sizeof(struct vlan_dev_info), name,
                                vlan_setup);
+
         if (new_dev == NULL)
                 goto out_unlock;
 
@@ -518,6 +527,8 @@ static struct net_device *register_vlan_device(const char *eth_IF_name,
         if (register_netdevice(new_dev))
                 goto out_free_newdev;
 
+        lockdep_set_class(&new_dev->_xmit_lock, &vlan_netdev_xmit_lock_key);
+
         new_dev->iflink = real_dev->ifindex;
         vlan_transfer_operstate(real_dev, new_dev);
         linkwatch_fire_event(new_dev); /* _MUST_ call rfc2863_policy() */
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 7cfbdb215ba2..44f6a181a754 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -71,6 +71,13 @@ static kmem_cache_t *skbuff_head_cache __read_mostly;
 static kmem_cache_t *skbuff_fclone_cache __read_mostly;
 
 /*
+ * lockdep: lock class key used by skb_queue_head_init():
+ */
+struct lock_class_key skb_queue_lock_key;
+
+EXPORT_SYMBOL(skb_queue_lock_key);
+
+/*
  *      Keep out-of-line to prevent kernel bloat.
  *      __builtin_return_address is not used because it is not always
  *      reliable.
diff --git a/net/core/sock.c b/net/core/sock.c
index 533b9317144b..51fcfbc041a7 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -129,6 +129,53 @@
 #include <net/tcp.h>
 #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
@@ -237,9 +284,16 @@ int sk_receive_skb(struct sock *sk, struct sk_buff *skb)
         skb->dev = NULL;
 
         bh_lock_sock(sk);
-        if (!sock_owned_by_user(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);
-        else
+
+                mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
+        } else
                 sk_add_backlog(sk, skb);
         bh_unlock_sock(sk);
 out:
@@ -749,6 +803,33 @@ 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)
+{
+        spin_lock_init(&sk->sk_lock.slock);
+        sk->sk_lock.owner = NULL;
+        init_waitqueue_head(&sk->sk_lock.wq);
+        /*
+         * Make sure we are not reinitializing a held lock:
+         */
+        debug_check_no_locks_freed((void *)&sk->sk_lock, sizeof(sk->sk_lock));
+
+        /*
+         * Mark both the sk_lock and the sk_lock.slock as a
+         * per-address-family lock class:
+         */
+        lockdep_set_class_and_name(&sk->sk_lock.slock,
+                                   af_family_slock_keys + sk->sk_family,
+                                   af_family_slock_key_strings[sk->sk_family]);
+        lockdep_init_map(&sk->sk_lock.dep_map,
+                         af_family_key_strings[sk->sk_family],
+                         af_family_keys + sk->sk_family);
+}
+
 /**
  *      sk_alloc - All socket objects are allocated here
  *      @family: protocol family
@@ -848,6 +929,8 @@ struct sock *sk_clone(const struct sock *sk, const gfp_t priority)
 
                 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;
@@ -1422,6 +1505,8 @@ 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;
@@ -1449,24 +1534,34 @@ void sock_init_data(struct socket *sock, struct sock *sk)
 void fastcall lock_sock(struct sock *sk)
 {
         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, 0, 0, _RET_IP_);
+        local_bh_enable();
 }
 
 EXPORT_SYMBOL(lock_sock);
 
 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);
 
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index da44fabf4dc5..2dc6dbb28467 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -205,21 +205,27 @@ __u8 ip_tos2prio[16] = {
 struct rt_hash_bucket {
         struct rtable   *chain;
 };
-#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK)
+#if defined(CONFIG_SMP) || defined(CONFIG_DEBUG_SPINLOCK) || \
+        defined(CONFIG_PROVE_LOCKING)
 /*
  * Instead of using one spinlock for each rt_hash_bucket, we use a table of spinlocks
  * The size of this table is a power of two and depends on the number of CPUS.
+ * (on lockdep we have a quite big spinlock_t, so keep the size down there)
  */
-#if NR_CPUS >= 32
-#define RT_HASH_LOCK_SZ 4096
-#elif NR_CPUS >= 16
-#define RT_HASH_LOCK_SZ 2048
-#elif NR_CPUS >= 8
-#define RT_HASH_LOCK_SZ 1024
-#elif NR_CPUS >= 4
-#define RT_HASH_LOCK_SZ 512
+#ifdef CONFIG_LOCKDEP
+# define RT_HASH_LOCK_SZ        256
 #else
-#define RT_HASH_LOCK_SZ 256
+# if NR_CPUS >= 32
+#  define RT_HASH_LOCK_SZ       4096
+# elif NR_CPUS >= 16
+#  define RT_HASH_LOCK_SZ       2048
+# elif NR_CPUS >= 8
+#  define RT_HASH_LOCK_SZ       1024
+# elif NR_CPUS >= 4
+#  define RT_HASH_LOCK_SZ       512
+# else
+#  define RT_HASH_LOCK_SZ       256
+# endif
 #endif
 
 static spinlock_t       *rt_hash_locks;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 8355b729fa95..5a886e6efbbe 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -90,7 +90,7 @@ static struct socket *tcp_socket;
 void tcp_v4_send_check(struct sock *sk, int len, struct sk_buff *skb);
 
 struct inet_hashinfo __cacheline_aligned tcp_hashinfo = {
-        .lhash_lock     = RW_LOCK_UNLOCKED,
+        .lhash_lock     = __RW_LOCK_UNLOCKED(tcp_hashinfo.lhash_lock),
         .lhash_users    = ATOMIC_INIT(0),
         .lhash_wait     = __WAIT_QUEUE_HEAD_INITIALIZER(tcp_hashinfo.lhash_wait),
 };
@@ -1090,7 +1090,7 @@ process:
 
         skb->dev = NULL;
 
-        bh_lock_sock(sk);
+        bh_lock_sock_nested(sk);
         ret = 0;
         if (!sock_owned_by_user(sk)) {
 #ifdef CONFIG_NET_DMA
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index e0851697ad5e..0ccb7cb22b15 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -40,7 +40,7 @@ int sysctl_tcp_abort_on_overflow;
 struct inet_timewait_death_row tcp_death_row = {
         .sysctl_max_tw_buckets = NR_FILE * 2,
         .period         = TCP_TIMEWAIT_LEN / INET_TWDR_TWKILL_SLOTS,
-        .death_lock     = SPIN_LOCK_UNLOCKED,
+        .death_lock     = __SPIN_LOCK_UNLOCKED(tcp_death_row.death_lock),
         .hashinfo       = &tcp_hashinfo,
         .tw_timer       = TIMER_INITIALIZER(inet_twdr_hangman, 0,
                                             (unsigned long)&tcp_death_row),
diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
index 70cee82a98bf..55c0adc8f115 100644
--- a/net/netlink/af_netlink.c
+++ b/net/netlink/af_netlink.c
@@ -156,7 +156,7 @@ static void netlink_sock_destruct(struct sock *sk)
 
 static void netlink_table_grab(void)
 {
-        write_lock_bh(&nl_table_lock);
+        write_lock_irq(&nl_table_lock);
 
         if (atomic_read(&nl_table_users)) {
                 DECLARE_WAITQUEUE(wait, current);
@@ -166,9 +166,9 @@ static void netlink_table_grab(void)
                         set_current_state(TASK_UNINTERRUPTIBLE);
                         if (atomic_read(&nl_table_users) == 0)
                                 break;
-                        write_unlock_bh(&nl_table_lock);
+                        write_unlock_irq(&nl_table_lock);
                         schedule();
-                        write_lock_bh(&nl_table_lock);
+                        write_lock_irq(&nl_table_lock);
                 }
 
                 __set_current_state(TASK_RUNNING);
@@ -178,7 +178,7 @@ static void netlink_table_grab(void)
 
 static __inline__ void netlink_table_ungrab(void)
 {
-        write_unlock_bh(&nl_table_lock);
+        write_unlock_irq(&nl_table_lock);
         wake_up(&nl_table_wait);
 }
 
diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c
index 6db6006616c6..dc6cb93c8830 100644
--- a/net/sunrpc/rpc_pipe.c
+++ b/net/sunrpc/rpc_pipe.c
@@ -515,7 +515,7 @@ rpc_depopulate(struct dentry *parent)
         struct dentry *dentry, *dvec[10];
         int n = 0;
 
-        mutex_lock(&dir->i_mutex);
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_CHILD);
 repeat:
         spin_lock(&dcache_lock);
         list_for_each_safe(pos, next, &parent->d_subdirs) {
@@ -631,7 +631,7 @@ rpc_lookup_negative(char *path, struct nameidata *nd)
         if ((error = rpc_lookup_parent(path, nd)) != 0)
                 return ERR_PTR(error);
         dir = nd->dentry->d_inode;
-        mutex_lock(&dir->i_mutex);
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
         dentry = lookup_one_len(nd->last.name, nd->dentry, nd->last.len);
         if (IS_ERR(dentry))
                 goto out_err;
@@ -693,7 +693,7 @@ rpc_rmdir(char *path)
         if ((error = rpc_lookup_parent(path, &nd)) != 0)
                 return error;
         dir = nd.dentry->d_inode;
-        mutex_lock(&dir->i_mutex);
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
         dentry = lookup_one_len(nd.last.name, nd.dentry, nd.last.len);
         if (IS_ERR(dentry)) {
                 error = PTR_ERR(dentry);
@@ -754,7 +754,7 @@ rpc_unlink(char *path)
         if ((error = rpc_lookup_parent(path, &nd)) != 0)
                 return error;
         dir = nd.dentry->d_inode;
-        mutex_lock(&dir->i_mutex);
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
         dentry = lookup_one_len(nd.last.name, nd.dentry, nd.last.len);
         if (IS_ERR(dentry)) {
                 error = PTR_ERR(dentry);
diff --git a/net/unix/af_unix.c b/net/unix/af_unix.c
index aca650109425..e9a287bc3142 100644
--- a/net/unix/af_unix.c
+++ b/net/unix/af_unix.c
@@ -565,6 +565,14 @@ static struct proto unix_proto = {
         .obj_size = sizeof(struct unix_sock),
 };
 
+/*
+ * AF_UNIX sockets do not interact with hardware, hence they
+ * dont trigger interrupts - so it's safe for them to have
+ * bh-unsafe locking for their sk_receive_queue.lock. Split off
+ * this special lock-class by reinitializing the spinlock key:
+ */
+static struct lock_class_key af_unix_sk_receive_queue_lock_key;
+
 static struct sock * unix_create1(struct socket *sock)
 {
         struct sock *sk = NULL;
@@ -580,6 +588,8 @@ static struct sock * unix_create1(struct socket *sock)
         atomic_inc(&unix_nr_socks);
 
         sock_init_data(sock,sk);
+        lockdep_set_class(&sk->sk_receive_queue.lock,
+                                &af_unix_sk_receive_queue_lock_key);
 
         sk->sk_write_space      = unix_write_space;
         sk->sk_max_ack_backlog  = sysctl_unix_max_dgram_qlen;
@@ -1045,7 +1055,7 @@ restart:
                 goto out_unlock;
         }
 
-        unix_state_wlock(sk);
+        unix_state_wlock_nested(sk);
 
         if (sk->sk_state != st) {
                 unix_state_wunlock(sk);