From dd4f10722aeb10f4f582948839f066bebe44e5fb Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 3 Mar 2017 21:01:02 -0800 Subject: net: fix socket refcounting in skb_complete_wifi_ack() TX skbs do not necessarily hold a reference on skb->sk->sk_refcnt By the time TX completion happens, sk_refcnt might be already 0. sock_hold()/sock_put() would then corrupt critical state, like sk_wmem_alloc. Fixes: bf7fa551e0ce ("mac80211: Resolve sk_refcnt/sk_wmem_alloc issue in wifi ack path") Signed-off-by: Eric Dumazet Cc: Alexander Duyck Cc: Johannes Berg Cc: Soheil Hassas Yeganeh Cc: Willem de Bruijn Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/core/skbuff.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'net/core') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index f3557958e9bf..e2f37a560ec4 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3893,7 +3893,7 @@ void skb_complete_wifi_ack(struct sk_buff *skb, bool acked) { struct sock *sk = skb->sk; struct sock_exterr_skb *serr; - int err; + int err = 1; skb->wifi_acked_valid = 1; skb->wifi_acked = acked; @@ -3903,14 +3903,15 @@ void skb_complete_wifi_ack(struct sk_buff *skb, bool acked) serr->ee.ee_errno = ENOMSG; serr->ee.ee_origin = SO_EE_ORIGIN_TXSTATUS; - /* take a reference to prevent skb_orphan() from freeing the socket */ - sock_hold(sk); - - err = sock_queue_err_skb(sk, skb); + /* Take a reference to prevent skb_orphan() from freeing the socket, + * but only if the socket refcount is not zero. + */ + if (likely(atomic_inc_not_zero(&sk->sk_refcnt))) { + err = sock_queue_err_skb(sk, skb); + sock_put(sk); + } if (err) kfree_skb(skb); - - sock_put(sk); } EXPORT_SYMBOL_GPL(skb_complete_wifi_ack); -- cgit From 9ac25fc063751379cb77434fef9f3b088cd3e2f7 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Fri, 3 Mar 2017 21:01:03 -0800 Subject: net: fix socket refcounting in skb_complete_tx_timestamp() TX skbs do not necessarily hold a reference on skb->sk->sk_refcnt By the time TX completion happens, sk_refcnt might be already 0. sock_hold()/sock_put() would then corrupt critical state, like sk_wmem_alloc and lead to leaks or use after free. Fixes: 62bccb8cdb69 ("net-timestamp: Make the clone operation stand-alone from phy timestamping") Signed-off-by: Eric Dumazet Cc: Alexander Duyck Cc: Johannes Berg Cc: Soheil Hassas Yeganeh Cc: Willem de Bruijn Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/core/skbuff.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'net/core') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index e2f37a560ec4..cd4ba8c6b609 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3828,13 +3828,14 @@ void skb_complete_tx_timestamp(struct sk_buff *skb, if (!skb_may_tx_timestamp(sk, false)) return; - /* take a reference to prevent skb_orphan() from freeing the socket */ - sock_hold(sk); - - *skb_hwtstamps(skb) = *hwtstamps; - __skb_complete_tx_timestamp(skb, sk, SCM_TSTAMP_SND); - - sock_put(sk); + /* Take a reference to prevent skb_orphan() from freeing the socket, + * but only if the socket refcount is not zero. + */ + if (likely(atomic_inc_not_zero(&sk->sk_refcnt))) { + *skb_hwtstamps(skb) = *hwtstamps; + __skb_complete_tx_timestamp(skb, sk, SCM_TSTAMP_SND); + sock_put(sk); + } } EXPORT_SYMBOL_GPL(skb_complete_tx_timestamp); -- cgit From cdfbabfb2f0ce983fdaa42f20e5f7842178fc01e Mon Sep 17 00:00:00 2001 From: David Howells Date: Thu, 9 Mar 2017 08:09:05 +0000 Subject: net: Work around lockdep limitation in sockets that use sockets Lockdep issues a circular dependency warning when AFS issues an operation through AF_RXRPC from a context in which the VFS/VM holds the mmap_sem. The theory lockdep comes up with is as follows: (1) If the pagefault handler decides it needs to read pages from AFS, it calls AFS with mmap_sem held and AFS begins an AF_RXRPC call, but creating a call requires the socket lock: mmap_sem must be taken before sk_lock-AF_RXRPC (2) afs_open_socket() opens an AF_RXRPC socket and binds it. rxrpc_bind() binds the underlying UDP socket whilst holding its socket lock. inet_bind() takes its own socket lock: sk_lock-AF_RXRPC must be taken before sk_lock-AF_INET (3) Reading from a TCP socket into a userspace buffer might cause a fault and thus cause the kernel to take the mmap_sem, but the TCP socket is locked whilst doing this: sk_lock-AF_INET must be taken before mmap_sem However, lockdep's theory is wrong in this instance because it deals only with lock classes and not individual locks. The AF_INET lock in (2) isn't really equivalent to the AF_INET lock in (3) as the former deals with a socket entirely internal to the kernel that never sees userspace. This is a limitation in the design of lockdep. Fix the general case by: (1) Double up all the locking keys used in sockets so that one set are used if the socket is created by userspace and the other set is used if the socket is created by the kernel. (2) Store the kern parameter passed to sk_alloc() in a variable in the sock struct (sk_kern_sock). This informs sock_lock_init(), sock_init_data() and sk_clone_lock() as to the lock keys to be used. Note that the child created by sk_clone_lock() inherits the parent's kern setting. (3) Add a 'kern' parameter to ->accept() that is analogous to the one passed in to ->create() that distinguishes whether kernel_accept() or sys_accept4() was the caller and can be passed to sk_alloc(). Note that a lot of accept functions merely dequeue an already allocated socket. I haven't touched these as the new socket already exists before we get the parameter. Note also that there are a couple of places where I've made the accepted socket unconditionally kernel-based: irda_accept() rds_rcp_accept_one() tcp_accept_from_sock() because they follow a sock_create_kern() and accept off of that. Whilst creating this, I noticed that lustre and ocfs don't create sockets through sock_create_kern() and thus they aren't marked as for-kernel, though they appear to be internal. I wonder if these should do that so that they use the new set of lock keys. Signed-off-by: David Howells Signed-off-by: David S. Miller --- net/core/sock.c | 106 ++++++++++++++++++++++++++++++-------------------------- 1 file changed, 57 insertions(+), 49 deletions(-) (limited to 'net/core') diff --git a/net/core/sock.c b/net/core/sock.c index f6fd79f33097..a96d5f7a5734 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -197,66 +197,55 @@ EXPORT_SYMBOL(sk_net_capable); /* * Each address family might have different locking rules, so we have - * one slock key per address family: + * one slock key per address family and separate keys for internal and + * userspace sockets. */ static struct lock_class_key af_family_keys[AF_MAX]; +static struct lock_class_key af_family_kern_keys[AF_MAX]; static struct lock_class_key af_family_slock_keys[AF_MAX]; +static struct lock_class_key af_family_kern_slock_keys[AF_MAX]; /* * Make lock validator output more readable. (we pre-construct these * strings build-time, so that runtime initialization of socket * locks is fast): */ + +#define _sock_locks(x) \ + x "AF_UNSPEC", x "AF_UNIX" , x "AF_INET" , \ + x "AF_AX25" , x "AF_IPX" , x "AF_APPLETALK", \ + x "AF_NETROM", x "AF_BRIDGE" , x "AF_ATMPVC" , \ + x "AF_X25" , x "AF_INET6" , x "AF_ROSE" , \ + x "AF_DECnet", x "AF_NETBEUI" , x "AF_SECURITY" , \ + x "AF_KEY" , x "AF_NETLINK" , x "AF_PACKET" , \ + x "AF_ASH" , x "AF_ECONET" , x "AF_ATMSVC" , \ + x "AF_RDS" , x "AF_SNA" , x "AF_IRDA" , \ + x "AF_PPPOX" , x "AF_WANPIPE" , x "AF_LLC" , \ + x "27" , x "28" , x "AF_CAN" , \ + x "AF_TIPC" , x "AF_BLUETOOTH", x "IUCV" , \ + x "AF_RXRPC" , x "AF_ISDN" , x "AF_PHONET" , \ + x "AF_IEEE802154", x "AF_CAIF" , x "AF_ALG" , \ + x "AF_NFC" , x "AF_VSOCK" , x "AF_KCM" , \ + x "AF_QIPCRTR", x "AF_SMC" , x "AF_MAX" + static const char *const 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-AF_RDS" , "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-AF_CAN" , - "sk_lock-AF_TIPC" , "sk_lock-AF_BLUETOOTH", "sk_lock-IUCV" , - "sk_lock-AF_RXRPC" , "sk_lock-AF_ISDN" , "sk_lock-AF_PHONET" , - "sk_lock-AF_IEEE802154", "sk_lock-AF_CAIF" , "sk_lock-AF_ALG" , - "sk_lock-AF_NFC" , "sk_lock-AF_VSOCK" , "sk_lock-AF_KCM" , - "sk_lock-AF_QIPCRTR", "sk_lock-AF_SMC" , "sk_lock-AF_MAX" + _sock_locks("sk_lock-") }; static const char *const 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-AF_RDS" , "slock-AF_SNA" , "slock-AF_IRDA" , - "slock-AF_PPPOX" , "slock-AF_WANPIPE" , "slock-AF_LLC" , - "slock-27" , "slock-28" , "slock-AF_CAN" , - "slock-AF_TIPC" , "slock-AF_BLUETOOTH", "slock-AF_IUCV" , - "slock-AF_RXRPC" , "slock-AF_ISDN" , "slock-AF_PHONET" , - "slock-AF_IEEE802154", "slock-AF_CAIF" , "slock-AF_ALG" , - "slock-AF_NFC" , "slock-AF_VSOCK" ,"slock-AF_KCM" , - "slock-AF_QIPCRTR", "slock-AF_SMC" , "slock-AF_MAX" + _sock_locks("slock-") }; static const char *const af_family_clock_key_strings[AF_MAX+1] = { - "clock-AF_UNSPEC", "clock-AF_UNIX" , "clock-AF_INET" , - "clock-AF_AX25" , "clock-AF_IPX" , "clock-AF_APPLETALK", - "clock-AF_NETROM", "clock-AF_BRIDGE" , "clock-AF_ATMPVC" , - "clock-AF_X25" , "clock-AF_INET6" , "clock-AF_ROSE" , - "clock-AF_DECnet", "clock-AF_NETBEUI" , "clock-AF_SECURITY" , - "clock-AF_KEY" , "clock-AF_NETLINK" , "clock-AF_PACKET" , - "clock-AF_ASH" , "clock-AF_ECONET" , "clock-AF_ATMSVC" , - "clock-AF_RDS" , "clock-AF_SNA" , "clock-AF_IRDA" , - "clock-AF_PPPOX" , "clock-AF_WANPIPE" , "clock-AF_LLC" , - "clock-27" , "clock-28" , "clock-AF_CAN" , - "clock-AF_TIPC" , "clock-AF_BLUETOOTH", "clock-AF_IUCV" , - "clock-AF_RXRPC" , "clock-AF_ISDN" , "clock-AF_PHONET" , - "clock-AF_IEEE802154", "clock-AF_CAIF" , "clock-AF_ALG" , - "clock-AF_NFC" , "clock-AF_VSOCK" , "clock-AF_KCM" , - "clock-AF_QIPCRTR", "clock-AF_SMC" , "clock-AF_MAX" + _sock_locks("clock-") +}; + +static const char *const af_family_kern_key_strings[AF_MAX+1] = { + _sock_locks("k-sk_lock-") +}; +static const char *const af_family_kern_slock_key_strings[AF_MAX+1] = { + _sock_locks("k-slock-") +}; +static const char *const af_family_kern_clock_key_strings[AF_MAX+1] = { + _sock_locks("k-clock-") }; /* @@ -264,6 +253,7 @@ static const char *const af_family_clock_key_strings[AF_MAX+1] = { * so split the lock classes by using a per-AF key: */ static struct lock_class_key af_callback_keys[AF_MAX]; +static struct lock_class_key af_kern_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 @@ -1293,7 +1283,16 @@ lenout: */ static inline void sock_lock_init(struct sock *sk) { - sock_lock_init_class_and_name(sk, + if (sk->sk_kern_sock) + sock_lock_init_class_and_name( + sk, + af_family_kern_slock_key_strings[sk->sk_family], + af_family_kern_slock_keys + sk->sk_family, + af_family_kern_key_strings[sk->sk_family], + af_family_kern_keys + sk->sk_family); + else + 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], @@ -1399,6 +1398,7 @@ struct sock *sk_alloc(struct net *net, int family, gfp_t priority, * why we need sk_prot_creator -acme */ sk->sk_prot = sk->sk_prot_creator = prot; + sk->sk_kern_sock = kern; sock_lock_init(sk); sk->sk_net_refcnt = kern ? 0 : 1; if (likely(sk->sk_net_refcnt)) @@ -2277,7 +2277,8 @@ int sock_no_socketpair(struct socket *sock1, struct socket *sock2) } EXPORT_SYMBOL(sock_no_socketpair); -int sock_no_accept(struct socket *sock, struct socket *newsock, int flags) +int sock_no_accept(struct socket *sock, struct socket *newsock, int flags, + bool kern) { return -EOPNOTSUPP; } @@ -2481,7 +2482,14 @@ void sock_init_data(struct socket *sock, struct sock *sk) } rwlock_init(&sk->sk_callback_lock); - lockdep_set_class_and_name(&sk->sk_callback_lock, + if (sk->sk_kern_sock) + lockdep_set_class_and_name( + &sk->sk_callback_lock, + af_kern_callback_keys + sk->sk_family, + af_family_kern_clock_key_strings[sk->sk_family]); + else + lockdep_set_class_and_name( + &sk->sk_callback_lock, af_callback_keys + sk->sk_family, af_family_clock_key_strings[sk->sk_family]); -- cgit From 91864f5852f9996210fad400cf70fb85af091243 Mon Sep 17 00:00:00 2001 From: Andrey Vagin Date: Sun, 12 Mar 2017 21:36:18 -0700 Subject: net: use net->count to check whether a netns is alive or not MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The previous idea was to check whether a net namespace is in net_exit_list or not. It doesn't work, because net->exit_list is used in __register_pernet_operations and __unregister_pernet_operations where all namespaces are added to a temporary list to make cleanup in a error case, so list_empty(&net->exit_list) always returns false. Reported-by: Mantas Mikulėnas Fixes: 002d8a1a6c11 ("net: skip genenerating uevents for network namespaces that are exiting") Signed-off-by: Andrei Vagin Signed-off-by: David S. Miller --- net/core/net-sysfs.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'net/core') diff --git a/net/core/net-sysfs.c b/net/core/net-sysfs.c index 3945821e9c1f..65ea0ff4017c 100644 --- a/net/core/net-sysfs.c +++ b/net/core/net-sysfs.c @@ -953,7 +953,7 @@ net_rx_queue_update_kobjects(struct net_device *dev, int old_num, int new_num) while (--i >= new_num) { struct kobject *kobj = &dev->_rx[i].kobj; - if (!list_empty(&dev_net(dev)->exit_list)) + if (!atomic_read(&dev_net(dev)->count)) kobj->uevent_suppress = 1; if (dev->sysfs_rx_queue_group) sysfs_remove_group(kobj, dev->sysfs_rx_queue_group); @@ -1371,7 +1371,7 @@ netdev_queue_update_kobjects(struct net_device *dev, int old_num, int new_num) while (--i >= new_num) { struct netdev_queue *queue = dev->_tx + i; - if (!list_empty(&dev_net(dev)->exit_list)) + if (!atomic_read(&dev_net(dev)->count)) queue->kobj.uevent_suppress = 1; #ifdef CONFIG_BQL sysfs_remove_group(&queue->kobj, &dql_group); @@ -1558,7 +1558,7 @@ void netdev_unregister_kobject(struct net_device *ndev) { struct device *dev = &(ndev->dev); - if (!list_empty(&dev_net(ndev)->exit_list)) + if (!atomic_read(&dev_net(ndev)->count)) dev_set_uevent_suppress(dev, 1); kobject_get(&dev->kobj); -- cgit From 37c343b4f4e70e9dc328ab04903c0ec8d154c1a4 Mon Sep 17 00:00:00 2001 From: Vlad Yasevich Date: Tue, 14 Mar 2017 08:58:08 -0400 Subject: net: Resend IGMP memberships upon peer notification. When we notify peers of potential changes, it's also good to update IGMP memberships. For example, during VM migration, updating IGMP memberships will redirect existing multicast streams to the VM at the new location. Signed-off-by: Vladislav Yasevich Acked-by: Michael S. Tsirkin Signed-off-by: David S. Miller --- net/core/dev.c | 1 + 1 file changed, 1 insertion(+) (limited to 'net/core') diff --git a/net/core/dev.c b/net/core/dev.c index 8637b2b71f3d..7869ae3837ca 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -1304,6 +1304,7 @@ void netdev_notify_peers(struct net_device *dev) { rtnl_lock(); call_netdevice_notifiers(NETDEV_NOTIFY_PEERS, dev); + call_netdevice_notifiers(NETDEV_RESEND_IGMP, dev); rtnl_unlock(); } EXPORT_SYMBOL(netdev_notify_peers); -- cgit From 22a0e18eac7a9e986fec76c60fa4a2926d1291e2 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 15 Mar 2017 13:21:28 -0700 Subject: net: properly release sk_frag.page I mistakenly added the code to release sk->sk_frag in sk_common_release() instead of sk_destruct() TCP sockets using sk->sk_allocation == GFP_ATOMIC do no call sk_common_release() at close time, thus leaking one (order-3) page. iSCSI is using such sockets. Fixes: 5640f7685831 ("net: use a per task frag allocator") Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/core/sock.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'net/core') diff --git a/net/core/sock.c b/net/core/sock.c index a96d5f7a5734..acb0d4137499 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -1442,6 +1442,11 @@ static void __sk_destruct(struct rcu_head *head) pr_debug("%s: optmem leakage (%d bytes) detected\n", __func__, atomic_read(&sk->sk_omem_alloc)); + if (sk->sk_frag.page) { + put_page(sk->sk_frag.page); + sk->sk_frag.page = NULL; + } + if (sk->sk_peer_cred) put_cred(sk->sk_peer_cred); put_pid(sk->sk_peer_pid); @@ -2787,11 +2792,6 @@ void sk_common_release(struct sock *sk) sk_refcnt_debug_release(sk); - if (sk->sk_frag.page) { - put_page(sk->sk_frag.page); - sk->sk_frag.page = NULL; - } - sock_put(sk); } EXPORT_SYMBOL(sk_common_release); -- cgit From 8605330aac5a5785630aec8f64378a54891937cc Mon Sep 17 00:00:00 2001 From: Soheil Hassas Yeganeh Date: Sat, 18 Mar 2017 17:02:59 -0400 Subject: tcp: fix SCM_TIMESTAMPING_OPT_STATS for normal skbs __sock_recv_timestamp can be called for both normal skbs (for receive timestamps) and for skbs on the error queue (for transmit timestamps). Commit 1c885808e456 (tcp: SOF_TIMESTAMPING_OPT_STATS option for SO_TIMESTAMPING) assumes any skb passed to __sock_recv_timestamp are from the error queue, containing OPT_STATS in the content of the skb. This results in accessing invalid memory or generating junk data. To fix this, set skb->pkt_type to PACKET_OUTGOING for packets on the error queue. This is safe because on the receive path on local sockets skb->pkt_type is never set to PACKET_OUTGOING. With that, copy OPT_STATS from a packet, only if its pkt_type is PACKET_OUTGOING. Fixes: 1c885808e456 ("tcp: SOF_TIMESTAMPING_OPT_STATS option for SO_TIMESTAMPING") Reported-by: JongHwan Kim Signed-off-by: Soheil Hassas Yeganeh Signed-off-by: Eric Dumazet Signed-off-by: Willem de Bruijn Signed-off-by: David S. Miller --- net/core/skbuff.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'net/core') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index cd4ba8c6b609..b1fbd1958eb6 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3694,6 +3694,15 @@ static void sock_rmem_free(struct sk_buff *skb) atomic_sub(skb->truesize, &sk->sk_rmem_alloc); } +static void skb_set_err_queue(struct sk_buff *skb) +{ + /* pkt_type of skbs received on local sockets is never PACKET_OUTGOING. + * So, it is safe to (mis)use it to mark skbs on the error queue. + */ + skb->pkt_type = PACKET_OUTGOING; + BUILD_BUG_ON(PACKET_OUTGOING == 0); +} + /* * Note: We dont mem charge error packets (no sk_forward_alloc changes) */ @@ -3707,6 +3716,7 @@ int sock_queue_err_skb(struct sock *sk, struct sk_buff *skb) skb->sk = sk; skb->destructor = sock_rmem_free; atomic_add(skb->truesize, &sk->sk_rmem_alloc); + skb_set_err_queue(skb); /* before exiting rcu section, make sure dst is refcounted */ skb_dst_force(skb); -- cgit From 4ef1b2869447411ad3ef91ad7d4891a83c1a509a Mon Sep 17 00:00:00 2001 From: Soheil Hassas Yeganeh Date: Sat, 18 Mar 2017 17:03:00 -0400 Subject: tcp: mark skbs with SCM_TIMESTAMPING_OPT_STATS SOF_TIMESTAMPING_OPT_STATS can be enabled and disabled while packets are collected on the error queue. So, checking SOF_TIMESTAMPING_OPT_STATS in sk->sk_tsflags is not enough to safely assume that the skb contains OPT_STATS data. Add a bit in sock_exterr_skb to indicate whether the skb contains opt_stats data. Fixes: 1c885808e456 ("tcp: SOF_TIMESTAMPING_OPT_STATS option for SO_TIMESTAMPING") Reported-by: JongHwan Kim Signed-off-by: Soheil Hassas Yeganeh Signed-off-by: Eric Dumazet Signed-off-by: Willem de Bruijn Signed-off-by: David S. Miller --- net/core/skbuff.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) (limited to 'net/core') diff --git a/net/core/skbuff.c b/net/core/skbuff.c index b1fbd1958eb6..9f781092fda9 100644 --- a/net/core/skbuff.c +++ b/net/core/skbuff.c @@ -3793,16 +3793,20 @@ EXPORT_SYMBOL(skb_clone_sk); static void __skb_complete_tx_timestamp(struct sk_buff *skb, struct sock *sk, - int tstype) + int tstype, + bool opt_stats) { struct sock_exterr_skb *serr; int err; + BUILD_BUG_ON(sizeof(struct sock_exterr_skb) > sizeof(skb->cb)); + serr = SKB_EXT_ERR(skb); memset(serr, 0, sizeof(*serr)); serr->ee.ee_errno = ENOMSG; serr->ee.ee_origin = SO_EE_ORIGIN_TIMESTAMPING; serr->ee.ee_info = tstype; + serr->opt_stats = opt_stats; if (sk->sk_tsflags & SOF_TIMESTAMPING_OPT_ID) { serr->ee.ee_data = skb_shinfo(skb)->tskey; if (sk->sk_protocol == IPPROTO_TCP && @@ -3843,7 +3847,7 @@ void skb_complete_tx_timestamp(struct sk_buff *skb, */ if (likely(atomic_inc_not_zero(&sk->sk_refcnt))) { *skb_hwtstamps(skb) = *hwtstamps; - __skb_complete_tx_timestamp(skb, sk, SCM_TSTAMP_SND); + __skb_complete_tx_timestamp(skb, sk, SCM_TSTAMP_SND, false); sock_put(sk); } } @@ -3854,7 +3858,7 @@ void __skb_tstamp_tx(struct sk_buff *orig_skb, struct sock *sk, int tstype) { struct sk_buff *skb; - bool tsonly; + bool tsonly, opt_stats = false; if (!sk) return; @@ -3867,9 +3871,10 @@ void __skb_tstamp_tx(struct sk_buff *orig_skb, #ifdef CONFIG_INET if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_STATS) && sk->sk_protocol == IPPROTO_TCP && - sk->sk_type == SOCK_STREAM) + sk->sk_type == SOCK_STREAM) { skb = tcp_get_timestamping_opt_stats(sk); - else + opt_stats = true; + } else #endif skb = alloc_skb(0, GFP_ATOMIC); } else { @@ -3888,7 +3893,7 @@ void __skb_tstamp_tx(struct sk_buff *orig_skb, else skb->tstamp = ktime_get_real(); - __skb_complete_tx_timestamp(skb, sk, tstype); + __skb_complete_tx_timestamp(skb, sk, tstype, opt_stats); } EXPORT_SYMBOL_GPL(__skb_tstamp_tx); -- cgit From a05d4fd9176003e0c1f9c3d083f4dac19fd346ab Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 14 Mar 2017 19:25:56 -0400 Subject: cgroup, net_cls: iterate the fds of only the tasks which are being migrated The net_cls controller controls the classid field of each socket which is associated with the cgroup. Because the classid is per-socket attribute, when a task migrates to another cgroup or the configured classid of the cgroup changes, the controller needs to walk all sockets and update the classid value, which was implemented by 3b13758f51de ("cgroups: Allow dynamically changing net_classid"). While the approach is not scalable, migrating tasks which have a lot of fds attached to them is rare and the cost is born by the ones initiating the operations. However, for simplicity, both the migration and classid config change paths call update_classid() which scans all fds of all tasks in the target css. This is an overkill for the migration path which only needs to cover a much smaller subset of tasks which are actually getting migrated in. On cgroup v1, this can lead to unexpected scalability issues when one tries to migrate a task or process into a net_cls cgroup which already contains a lot of fds. Even if the migration traget doesn't have many to get scanned, update_classid() ends up scanning all fds in the target cgroup which can be extremely numerous. Unfortunately, on cgroup v2 which doesn't use net_cls, the problem is even worse. Before bfc2cf6f61fc ("cgroup: call subsys->*attach() only for subsystems which are actually affected by migration"), cgroup core would call the ->css_attach callback even for controllers which don't see actual migration to a different css. As net_cls is always disabled but still mounted on cgroup v2, whenever a process is migrated on the cgroup v2 hierarchy, net_cls sees identity migration from root to root and cgroup core used to call ->css_attach callback for those. The net_cls ->css_attach ends up calling update_classid() on the root net_cls css to which all processes on the system belong to as the controller isn't used. This makes any cgroup v2 migration O(total_number_of_fds_on_the_system) which is horrible and easily leads to noticeable stalls triggering RCU stall warnings and so on. The worst symptom is already fixed in upstream by bfc2cf6f61fc ("cgroup: call subsys->*attach() only for subsystems which are actually affected by migration"); however, backporting that commit is too invasive and we want to avoid other cases too. This patch updates net_cls's cgrp_attach() to iterate fds of only the processes which are actually getting migrated. This removes the surprising migration cost which is dependent on the total number of fds in the target cgroup. As this leaves write_classid() the only user of update_classid(), open-code the helper into write_classid(). Reported-by: David Goode Fixes: 3b13758f51de ("cgroups: Allow dynamically changing net_classid") Cc: stable@vger.kernel.org # v4.4+ Cc: Nina Schiff Cc: David S. Miller Signed-off-by: Tejun Heo Signed-off-by: David S. Miller --- net/core/netclassid_cgroup.c | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-) (limited to 'net/core') diff --git a/net/core/netclassid_cgroup.c b/net/core/netclassid_cgroup.c index 6ae56037bb13..029a61ac6cdd 100644 --- a/net/core/netclassid_cgroup.c +++ b/net/core/netclassid_cgroup.c @@ -71,27 +71,17 @@ static int update_classid_sock(const void *v, struct file *file, unsigned n) return 0; } -static void update_classid(struct cgroup_subsys_state *css, void *v) +static void cgrp_attach(struct cgroup_taskset *tset) { - struct css_task_iter it; + struct cgroup_subsys_state *css; struct task_struct *p; - css_task_iter_start(css, &it); - while ((p = css_task_iter_next(&it))) { + cgroup_taskset_for_each(p, css, tset) { task_lock(p); - iterate_fd(p->files, 0, update_classid_sock, v); + iterate_fd(p->files, 0, update_classid_sock, + (void *)(unsigned long)css_cls_state(css)->classid); task_unlock(p); } - css_task_iter_end(&it); -} - -static void cgrp_attach(struct cgroup_taskset *tset) -{ - struct cgroup_subsys_state *css; - - cgroup_taskset_first(tset, &css); - update_classid(css, - (void *)(unsigned long)css_cls_state(css)->classid); } static u64 read_classid(struct cgroup_subsys_state *css, struct cftype *cft) @@ -103,12 +93,22 @@ static int write_classid(struct cgroup_subsys_state *css, struct cftype *cft, u64 value) { struct cgroup_cls_state *cs = css_cls_state(css); + struct css_task_iter it; + struct task_struct *p; cgroup_sk_alloc_disable(); cs->classid = (u32)value; - update_classid(css, (void *)(unsigned long)cs->classid); + css_task_iter_start(css, &it); + while ((p = css_task_iter_next(&it))) { + task_lock(p); + iterate_fd(p->files, 0, update_classid_sock, + (void *)(unsigned long)cs->classid); + task_unlock(p); + } + css_task_iter_end(&it); + return 0; } -- cgit From a97e50cc4cb67e1e7bff56f6b41cda62ca832336 Mon Sep 17 00:00:00 2001 From: Daniel Borkmann Date: Wed, 22 Mar 2017 13:08:08 +0100 Subject: socket, bpf: fix sk_filter use after free in sk_clone_lock In sk_clone_lock(), we create a new socket and inherit most of the parent's members via sock_copy() which memcpy()'s various sections. Now, in case the parent socket had a BPF socket filter attached, then newsk->sk_filter points to the same instance as the original sk->sk_filter. sk_filter_charge() is then called on the newsk->sk_filter to take a reference and should that fail due to hitting max optmem, we bail out and release the newsk instance. The issue is that commit 278571baca2a ("net: filter: simplify socket charging") wrongly combined the dismantle path with the failure path of xfrm_sk_clone_policy(). This means, even when charging failed, we call sk_free_unlock_clone() on the newsk, which then still points to the same sk_filter as the original sk. Thus, sk_free_unlock_clone() calls into __sk_destruct() eventually where it tests for present sk_filter and calls sk_filter_uncharge() on it, which potentially lets sk_omem_alloc wrap around and releases the eBPF prog and sk_filter structure from the (still intact) parent. Fix it by making sure that when sk_filter_charge() failed, we reset newsk->sk_filter back to NULL before passing to sk_free_unlock_clone(), so that we don't mess with the parents sk_filter. Only if xfrm_sk_clone_policy() fails, we did reach the point where either the parent's filter was NULL and as a result newsk's as well or where we previously had a successful sk_filter_charge(), thus for that case, we do need sk_filter_uncharge() to release the prior taken reference on sk_filter. Fixes: 278571baca2a ("net: filter: simplify socket charging") Signed-off-by: Daniel Borkmann Acked-by: Alexei Starovoitov Signed-off-by: David S. Miller --- net/core/sock.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'net/core') diff --git a/net/core/sock.c b/net/core/sock.c index acb0d4137499..2c4f574168fb 100644 --- a/net/core/sock.c +++ b/net/core/sock.c @@ -1544,6 +1544,12 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority) is_charged = sk_filter_charge(newsk, filter); if (unlikely(!is_charged || xfrm_sk_clone_policy(newsk, sk))) { + /* We need to make sure that we don't uncharge the new + * socket if we couldn't charge it in the first place + * as otherwise we uncharge the parent's filter. + */ + if (!is_charged) + RCU_INIT_POINTER(newsk->sk_filter, NULL); sk_free_unlock_clone(newsk); newsk = NULL; goto out; -- cgit