linux/net/netlink/af_netlink.c

2170 lines
48 KiB
C
Raw Normal View History

/*
* NETLINK Kernel-user communication protocol.
*
* Authors: Alan Cox <alan@lxorguk.ukuu.org.uk>
* Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*
* Tue Jun 26 14:36:48 MEST 2001 Herbert "herp" Rosmanith
* added netlink_proto_exit
* Tue Jan 22 18:32:44 BRST 2002 Arnaldo C. de Melo <acme@conectiva.com.br>
* use nlk_sk, as sk->protinfo is on a diet 8)
* Fri Jul 22 19:51:12 MEST 2005 Harald Welte <laforge@gnumonks.org>
* - inc module use count of module that owns
* the kernel socket in case userspace opens
* socket of same protocol
* - remove all module support, since netlink is
* mandatory if CONFIG_NET=y these days
*/
#include <linux/module.h>
#include <linux/capability.h>
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/socket.h>
#include <linux/un.h>
#include <linux/fcntl.h>
#include <linux/termios.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/fs.h>
#include <linux/slab.h>
#include <asm/uaccess.h>
#include <linux/skbuff.h>
#include <linux/netdevice.h>
#include <linux/rtnetlink.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/notifier.h>
#include <linux/security.h>
#include <linux/jhash.h>
#include <linux/jiffies.h>
#include <linux/random.h>
#include <linux/bitops.h>
#include <linux/mm.h>
#include <linux/types.h>
#include <linux/audit.h>
#include <linux/mutex.h>
#include <net/net_namespace.h>
#include <net/sock.h>
#include <net/scm.h>
#include <net/netlink.h>
#define NLGRPSZ(x) (ALIGN(x, sizeof(unsigned long) * 8) / 8)
#define NLGRPLONGS(x) (NLGRPSZ(x)/sizeof(unsigned long))
struct netlink_sock {
/* struct sock has to be the first member of netlink_sock */
struct sock sk;
u32 pid;
u32 dst_pid;
u32 dst_group;
u32 flags;
u32 subscriptions;
u32 ngroups;
unsigned long *groups;
unsigned long state;
wait_queue_head_t wait;
struct netlink_callback *cb;
struct mutex *cb_mutex;
struct mutex cb_def_mutex;
void (*netlink_rcv)(struct sk_buff *skb);
struct module *module;
};
struct listeners {
struct rcu_head rcu;
unsigned long masks[0];
};
#define NETLINK_KERNEL_SOCKET 0x1
#define NETLINK_RECV_PKTINFO 0x2
#define NETLINK_BROADCAST_SEND_ERROR 0x4
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
#define NETLINK_RECV_NO_ENOBUFS 0x8
static inline struct netlink_sock *nlk_sk(struct sock *sk)
{
return container_of(sk, struct netlink_sock, sk);
}
static inline int netlink_is_kernel(struct sock *sk)
{
return nlk_sk(sk)->flags & NETLINK_KERNEL_SOCKET;
}
struct nl_pid_hash {
struct hlist_head *table;
unsigned long rehash_time;
unsigned int mask;
unsigned int shift;
unsigned int entries;
unsigned int max_shift;
u32 rnd;
};
struct netlink_table {
struct nl_pid_hash hash;
struct hlist_head mc_list;
struct listeners __rcu *listeners;
unsigned int nl_nonroot;
unsigned int groups;
struct mutex *cb_mutex;
struct module *module;
int registered;
};
static struct netlink_table *nl_table;
static DECLARE_WAIT_QUEUE_HEAD(nl_table_wait);
static int netlink_dump(struct sock *sk);
static void netlink_destroy_callback(struct netlink_callback *cb);
static DEFINE_RWLOCK(nl_table_lock);
static atomic_t nl_table_users = ATOMIC_INIT(0);
[PATCH] Notifier chain update: API changes The kernel's implementation of notifier chains is unsafe. There is no protection against entries being added to or removed from a chain while the chain is in use. The issues were discussed in this thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 We noticed that notifier chains in the kernel fall into two basic usage classes: "Blocking" chains are always called from a process context and the callout routines are allowed to sleep; "Atomic" chains can be called from an atomic context and the callout routines are not allowed to sleep. We decided to codify this distinction and make it part of the API. Therefore this set of patches introduces three new, parallel APIs: one for blocking notifiers, one for atomic notifiers, and one for "raw" notifiers (which is really just the old API under a new name). New kinds of data structures are used for the heads of the chains, and new routines are defined for registration, unregistration, and calling a chain. The three APIs are explained in include/linux/notifier.h and their implementation is in kernel/sys.c. With atomic and blocking chains, the implementation guarantees that the chain links will not be corrupted and that chain callers will not get messed up by entries being added or removed. For raw chains the implementation provides no guarantees at all; users of this API must provide their own protections. (The idea was that situations may come up where the assumptions of the atomic and blocking APIs are not appropriate, so it should be possible for users to handle these things in their own way.) There are some limitations, which should not be too hard to live with. For atomic/blocking chains, registration and unregistration must always be done in a process context since the chain is protected by a mutex/rwsem. Also, a callout routine for a non-raw chain must not try to register or unregister entries on its own chain. (This did happen in a couple of places and the code had to be changed to avoid it.) Since atomic chains may be called from within an NMI handler, they cannot use spinlocks for synchronization. Instead we use RCU. The overhead falls almost entirely in the unregister routine, which is okay since unregistration is much less frequent that calling a chain. Here is the list of chains that we adjusted and their classifications. None of them use the raw API, so for the moment it is only a placeholder. ATOMIC CHAINS ------------- arch/i386/kernel/traps.c: i386die_chain arch/ia64/kernel/traps.c: ia64die_chain arch/powerpc/kernel/traps.c: powerpc_die_chain arch/sparc64/kernel/traps.c: sparc64die_chain arch/x86_64/kernel/traps.c: die_chain drivers/char/ipmi/ipmi_si_intf.c: xaction_notifier_list kernel/panic.c: panic_notifier_list kernel/profile.c: task_free_notifier net/bluetooth/hci_core.c: hci_notifier net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_chain net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_expect_chain net/ipv6/addrconf.c: inet6addr_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_expect_chain net/netlink/af_netlink.c: netlink_chain BLOCKING CHAINS --------------- arch/powerpc/platforms/pseries/reconfig.c: pSeries_reconfig_chain arch/s390/kernel/process.c: idle_chain arch/x86_64/kernel/process.c idle_notifier drivers/base/memory.c: memory_chain drivers/cpufreq/cpufreq.c cpufreq_policy_notifier_list drivers/cpufreq/cpufreq.c cpufreq_transition_notifier_list drivers/macintosh/adb.c: adb_client_list drivers/macintosh/via-pmu.c sleep_notifier_list drivers/macintosh/via-pmu68k.c sleep_notifier_list drivers/macintosh/windfarm_core.c wf_client_list drivers/usb/core/notify.c usb_notifier_list drivers/video/fbmem.c fb_notifier_list kernel/cpu.c cpu_chain kernel/module.c module_notify_list kernel/profile.c munmap_notifier kernel/profile.c task_exit_notifier kernel/sys.c reboot_notifier_list net/core/dev.c netdev_chain net/decnet/dn_dev.c: dnaddr_chain net/ipv4/devinet.c: inetaddr_chain It's possible that some of these classifications are wrong. If they are, please let us know or submit a patch to fix them. Note that any chain that gets called very frequently should be atomic, because the rwsem read-locking used for blocking chains is very likely to incur cache misses on SMP systems. (However, if the chain's callout routines may sleep then the chain cannot be atomic.) The patch set was written by Alan Stern and Chandra Seetharaman, incorporating material written by Keith Owens and suggestions from Paul McKenney and Andrew Morton. [jes@sgi.com: restructure the notifier chain initialization macros] Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Jes Sorensen <jes@sgi.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 09:16:30 +00:00
static ATOMIC_NOTIFIER_HEAD(netlink_chain);
static u32 netlink_group_mask(u32 group)
{
return group ? 1 << (group - 1) : 0;
}
static struct hlist_head *nl_pid_hashfn(struct nl_pid_hash *hash, u32 pid)
{
return &hash->table[jhash_1word(pid, hash->rnd) & hash->mask];
}
static void netlink_sock_destruct(struct sock *sk)
{
struct netlink_sock *nlk = nlk_sk(sk);
if (nlk->cb) {
if (nlk->cb->done)
nlk->cb->done(nlk->cb);
netlink_destroy_callback(nlk->cb);
}
skb_queue_purge(&sk->sk_receive_queue);
if (!sock_flag(sk, SOCK_DEAD)) {
printk(KERN_ERR "Freeing alive netlink socket %p\n", sk);
return;
}
WARN_ON(atomic_read(&sk->sk_rmem_alloc));
WARN_ON(atomic_read(&sk->sk_wmem_alloc));
WARN_ON(nlk_sk(sk)->groups);
}
/* This lock without WQ_FLAG_EXCLUSIVE is good on UP and it is _very_ bad on
* SMP. Look, when several writers sleep and reader wakes them up, all but one
* immediately hit write lock and grab all the cpus. Exclusive sleep solves
* this, _but_ remember, it adds useless work on UP machines.
*/
void netlink_table_grab(void)
__acquires(nl_table_lock)
{
might_sleep();
write_lock_irq(&nl_table_lock);
if (atomic_read(&nl_table_users)) {
DECLARE_WAITQUEUE(wait, current);
add_wait_queue_exclusive(&nl_table_wait, &wait);
for (;;) {
set_current_state(TASK_UNINTERRUPTIBLE);
if (atomic_read(&nl_table_users) == 0)
break;
write_unlock_irq(&nl_table_lock);
schedule();
write_lock_irq(&nl_table_lock);
}
__set_current_state(TASK_RUNNING);
remove_wait_queue(&nl_table_wait, &wait);
}
}
void netlink_table_ungrab(void)
__releases(nl_table_lock)
{
write_unlock_irq(&nl_table_lock);
wake_up(&nl_table_wait);
}
static inline void
netlink_lock_table(void)
{
/* read_lock() synchronizes us to netlink_table_grab */
read_lock(&nl_table_lock);
atomic_inc(&nl_table_users);
read_unlock(&nl_table_lock);
}
static inline void
netlink_unlock_table(void)
{
if (atomic_dec_and_test(&nl_table_users))
wake_up(&nl_table_wait);
}
static inline struct sock *netlink_lookup(struct net *net, int protocol,
u32 pid)
{
struct nl_pid_hash *hash = &nl_table[protocol].hash;
struct hlist_head *head;
struct sock *sk;
struct hlist_node *node;
read_lock(&nl_table_lock);
head = nl_pid_hashfn(hash, pid);
sk_for_each(sk, node, head) {
if (net_eq(sock_net(sk), net) && (nlk_sk(sk)->pid == pid)) {
sock_hold(sk);
goto found;
}
}
sk = NULL;
found:
read_unlock(&nl_table_lock);
return sk;
}
static inline struct hlist_head *nl_pid_hash_zalloc(size_t size)
{
if (size <= PAGE_SIZE)
return kzalloc(size, GFP_ATOMIC);
else
return (struct hlist_head *)
__get_free_pages(GFP_ATOMIC | __GFP_ZERO,
get_order(size));
}
static inline void nl_pid_hash_free(struct hlist_head *table, size_t size)
{
if (size <= PAGE_SIZE)
kfree(table);
else
free_pages((unsigned long)table, get_order(size));
}
static int nl_pid_hash_rehash(struct nl_pid_hash *hash, int grow)
{
unsigned int omask, mask, shift;
size_t osize, size;
struct hlist_head *otable, *table;
int i;
omask = mask = hash->mask;
osize = size = (mask + 1) * sizeof(*table);
shift = hash->shift;
if (grow) {
if (++shift > hash->max_shift)
return 0;
mask = mask * 2 + 1;
size *= 2;
}
table = nl_pid_hash_zalloc(size);
if (!table)
return 0;
otable = hash->table;
hash->table = table;
hash->mask = mask;
hash->shift = shift;
get_random_bytes(&hash->rnd, sizeof(hash->rnd));
for (i = 0; i <= omask; i++) {
struct sock *sk;
struct hlist_node *node, *tmp;
sk_for_each_safe(sk, node, tmp, &otable[i])
__sk_add_node(sk, nl_pid_hashfn(hash, nlk_sk(sk)->pid));
}
nl_pid_hash_free(otable, osize);
hash->rehash_time = jiffies + 10 * 60 * HZ;
return 1;
}
static inline int nl_pid_hash_dilute(struct nl_pid_hash *hash, int len)
{
int avg = hash->entries >> hash->shift;
if (unlikely(avg > 1) && nl_pid_hash_rehash(hash, 1))
return 1;
if (unlikely(len > avg) && time_after(jiffies, hash->rehash_time)) {
nl_pid_hash_rehash(hash, 0);
return 1;
}
return 0;
}
static const struct proto_ops netlink_ops;
static void
netlink_update_listeners(struct sock *sk)
{
struct netlink_table *tbl = &nl_table[sk->sk_protocol];
struct hlist_node *node;
unsigned long mask;
unsigned int i;
for (i = 0; i < NLGRPLONGS(tbl->groups); i++) {
mask = 0;
sk_for_each_bound(sk, node, &tbl->mc_list) {
if (i < NLGRPLONGS(nlk_sk(sk)->ngroups))
mask |= nlk_sk(sk)->groups[i];
}
tbl->listeners->masks[i] = mask;
}
/* this function is only called with the netlink table "grabbed", which
* makes sure updates are visible before bind or setsockopt return. */
}
static int netlink_insert(struct sock *sk, struct net *net, u32 pid)
{
struct nl_pid_hash *hash = &nl_table[sk->sk_protocol].hash;
struct hlist_head *head;
int err = -EADDRINUSE;
struct sock *osk;
struct hlist_node *node;
int len;
netlink_table_grab();
head = nl_pid_hashfn(hash, pid);
len = 0;
sk_for_each(osk, node, head) {
if (net_eq(sock_net(osk), net) && (nlk_sk(osk)->pid == pid))
break;
len++;
}
if (node)
goto err;
err = -EBUSY;
if (nlk_sk(sk)->pid)
goto err;
err = -ENOMEM;
if (BITS_PER_LONG > 32 && unlikely(hash->entries >= UINT_MAX))
goto err;
if (len && nl_pid_hash_dilute(hash, len))
head = nl_pid_hashfn(hash, pid);
hash->entries++;
nlk_sk(sk)->pid = pid;
sk_add_node(sk, head);
err = 0;
err:
netlink_table_ungrab();
return err;
}
static void netlink_remove(struct sock *sk)
{
netlink_table_grab();
[NETLINK]: Fix two socket hashing bugs. 1) netlink_release() should only decrement the hash entry count if the socket was actually hashed. This was causing hash->entries to underflow, which resulting in all kinds of troubles. On 64-bit systems, this would cause the following conditional to erroneously trigger: err = -ENOMEM; if (BITS_PER_LONG > 32 && unlikely(hash->entries >= UINT_MAX)) goto err; 2) netlink_autobind() needs to propagate the error return from netlink_insert(). Otherwise, callers will not see the error as they should and thus try to operate on a socket with a zero pid, which is very bad. However, it should not propagate -EBUSY. If two threads race to autobind the socket, that is fine. This is consistent with the autobind behavior in other protocols. So bug #1 above, combined with this one, resulted in hangs on netlink_sendmsg() calls to the rtnetlink socket. We'd try to do the user sendmsg() with the socket's pid set to zero, later we do a socket lookup using that pid (via the value we stashed away in NETLINK_CB(skb).pid), but that won't give us the user socket, it will give us the rtnetlink socket. So when we try to wake up the receive queue, we dive back into rtnetlink_rcv() which tries to recursively take the rtnetlink semaphore. Thanks to Jakub Jelink for providing backtraces. Also, thanks to Herbert Xu for supplying debugging patches to help track this down, and also finding a mistake in an earlier version of this fix. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-26 22:31:51 +00:00
if (sk_del_node_init(sk))
nl_table[sk->sk_protocol].hash.entries--;
if (nlk_sk(sk)->subscriptions)
__sk_del_bind_node(sk);
netlink_table_ungrab();
}
static struct proto netlink_proto = {
.name = "NETLINK",
.owner = THIS_MODULE,
.obj_size = sizeof(struct netlink_sock),
};
static int __netlink_create(struct net *net, struct socket *sock,
struct mutex *cb_mutex, int protocol)
{
struct sock *sk;
struct netlink_sock *nlk;
sock->ops = &netlink_ops;
sk = sk_alloc(net, PF_NETLINK, GFP_KERNEL, &netlink_proto);
if (!sk)
return -ENOMEM;
sock_init_data(sock, sk);
nlk = nlk_sk(sk);
if (cb_mutex)
nlk->cb_mutex = cb_mutex;
else {
nlk->cb_mutex = &nlk->cb_def_mutex;
mutex_init(nlk->cb_mutex);
}
init_waitqueue_head(&nlk->wait);
sk->sk_destruct = netlink_sock_destruct;
sk->sk_protocol = protocol;
return 0;
}
static int netlink_create(struct net *net, struct socket *sock, int protocol,
int kern)
{
struct module *module = NULL;
struct mutex *cb_mutex;
struct netlink_sock *nlk;
int err = 0;
sock->state = SS_UNCONNECTED;
if (sock->type != SOCK_RAW && sock->type != SOCK_DGRAM)
return -ESOCKTNOSUPPORT;
if (protocol < 0 || protocol >= MAX_LINKS)
return -EPROTONOSUPPORT;
netlink_lock_table();
#ifdef CONFIG_MODULES
if (!nl_table[protocol].registered) {
netlink_unlock_table();
request_module("net-pf-%d-proto-%d", PF_NETLINK, protocol);
netlink_lock_table();
}
#endif
if (nl_table[protocol].registered &&
try_module_get(nl_table[protocol].module))
module = nl_table[protocol].module;
netlink: fix for too early rmmod Netlink code does module autoload if protocol userspace is asking for is not ready. However, module can dissapear right after it was autoloaded. Example: modprobe/rmmod stress-testing and xfrm_user.ko providing NETLINK_XFRM. netlink_create() in such situation _will_ create userspace socket and _will_not_ pin module. Now if module was removed and we're going to call ->netlink_rcv into nothing: BUG: unable to handle kernel paging request at ffffffffa02f842a ^^^^^^^^^^^^^^^^ modules are loaded near these addresses here IP: [<ffffffffa02f842a>] 0xffffffffa02f842a PGD 161f067 PUD 1623063 PMD baa12067 PTE 0 Oops: 0010 [#1] PREEMPT SMP DEBUG_PAGEALLOC last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host0/target0:0:0/0:0:0:0/block/sda/uevent CPU 1 Pid: 11515, comm: ip Not tainted 2.6.33-rc5-netns-00594-gaaa5728-dirty #6 P5E/P5E RIP: 0010:[<ffffffffa02f842a>] [<ffffffffa02f842a>] 0xffffffffa02f842a RSP: 0018:ffff8800baa3db48 EFLAGS: 00010292 RAX: ffff8800baa3dfd8 RBX: ffff8800be353640 RCX: 0000000000000000 RDX: ffffffff81959380 RSI: ffff8800bab7f130 RDI: 0000000000000001 RBP: ffff8800baa3db58 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000011 R13: ffff8800be353640 R14: ffff8800bcdec240 R15: ffff8800bd488010 FS: 00007f93749656f0(0000) GS:ffff880002300000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffffffffa02f842a CR3: 00000000ba82b000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process ip (pid: 11515, threadinfo ffff8800baa3c000, task ffff8800bab7eb30) Stack: ffffffff813637c0 ffff8800bd488000 ffff8800baa3dba8 ffffffff8136397d <0> 0000000000000000 ffffffff81344adc 7fffffffffffffff 0000000000000000 <0> ffff8800baa3ded8 ffff8800be353640 ffff8800bcdec240 0000000000000000 Call Trace: [<ffffffff813637c0>] ? netlink_unicast+0x100/0x2d0 [<ffffffff8136397d>] netlink_unicast+0x2bd/0x2d0 netlink_unicast_kernel: nlk->netlink_rcv(skb); [<ffffffff81344adc>] ? memcpy_fromiovec+0x6c/0x90 [<ffffffff81364263>] netlink_sendmsg+0x1d3/0x2d0 [<ffffffff8133975b>] sock_sendmsg+0xbb/0xf0 [<ffffffff8106cdeb>] ? __lock_acquire+0x27b/0xa60 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff8106db22>] ? __lock_release+0x82/0x170 [<ffffffff810a190e>] ? might_fault+0xbe/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff81344c77>] ? verify_iovec+0x47/0xd0 [<ffffffff8133a509>] sys_sendmsg+0x1a9/0x360 [<ffffffff813c2be5>] ? _raw_spin_unlock_irqrestore+0x65/0x70 [<ffffffff8106aced>] ? trace_hardirqs_on+0xd/0x10 [<ffffffff813c2bc2>] ? _raw_spin_unlock_irqrestore+0x42/0x70 [<ffffffff81197004>] ? __up_read+0x84/0xb0 [<ffffffff8106ac95>] ? trace_hardirqs_on_caller+0x145/0x190 [<ffffffff813c207f>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff8100262b>] system_call_fastpath+0x16/0x1b Code: Bad RIP value. RIP [<ffffffffa02f842a>] 0xffffffffa02f842a RSP <ffff8800baa3db48> CR2: ffffffffa02f842a If module was quickly removed after autoloading, return -E. Return -EPROTONOSUPPORT if module was quickly removed after autoloading. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-30 10:05:05 +00:00
else
err = -EPROTONOSUPPORT;
cb_mutex = nl_table[protocol].cb_mutex;
netlink_unlock_table();
netlink: fix for too early rmmod Netlink code does module autoload if protocol userspace is asking for is not ready. However, module can dissapear right after it was autoloaded. Example: modprobe/rmmod stress-testing and xfrm_user.ko providing NETLINK_XFRM. netlink_create() in such situation _will_ create userspace socket and _will_not_ pin module. Now if module was removed and we're going to call ->netlink_rcv into nothing: BUG: unable to handle kernel paging request at ffffffffa02f842a ^^^^^^^^^^^^^^^^ modules are loaded near these addresses here IP: [<ffffffffa02f842a>] 0xffffffffa02f842a PGD 161f067 PUD 1623063 PMD baa12067 PTE 0 Oops: 0010 [#1] PREEMPT SMP DEBUG_PAGEALLOC last sysfs file: /sys/devices/pci0000:00/0000:00:1f.2/host0/target0:0:0/0:0:0:0/block/sda/uevent CPU 1 Pid: 11515, comm: ip Not tainted 2.6.33-rc5-netns-00594-gaaa5728-dirty #6 P5E/P5E RIP: 0010:[<ffffffffa02f842a>] [<ffffffffa02f842a>] 0xffffffffa02f842a RSP: 0018:ffff8800baa3db48 EFLAGS: 00010292 RAX: ffff8800baa3dfd8 RBX: ffff8800be353640 RCX: 0000000000000000 RDX: ffffffff81959380 RSI: ffff8800bab7f130 RDI: 0000000000000001 RBP: ffff8800baa3db58 R08: 0000000000000001 R09: 0000000000000000 R10: 0000000000000001 R11: 0000000000000001 R12: 0000000000000011 R13: ffff8800be353640 R14: ffff8800bcdec240 R15: ffff8800bd488010 FS: 00007f93749656f0(0000) GS:ffff880002300000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffffffffa02f842a CR3: 00000000ba82b000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process ip (pid: 11515, threadinfo ffff8800baa3c000, task ffff8800bab7eb30) Stack: ffffffff813637c0 ffff8800bd488000 ffff8800baa3dba8 ffffffff8136397d <0> 0000000000000000 ffffffff81344adc 7fffffffffffffff 0000000000000000 <0> ffff8800baa3ded8 ffff8800be353640 ffff8800bcdec240 0000000000000000 Call Trace: [<ffffffff813637c0>] ? netlink_unicast+0x100/0x2d0 [<ffffffff8136397d>] netlink_unicast+0x2bd/0x2d0 netlink_unicast_kernel: nlk->netlink_rcv(skb); [<ffffffff81344adc>] ? memcpy_fromiovec+0x6c/0x90 [<ffffffff81364263>] netlink_sendmsg+0x1d3/0x2d0 [<ffffffff8133975b>] sock_sendmsg+0xbb/0xf0 [<ffffffff8106cdeb>] ? __lock_acquire+0x27b/0xa60 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff8106db22>] ? __lock_release+0x82/0x170 [<ffffffff810a190e>] ? might_fault+0xbe/0xd0 [<ffffffff810a18c3>] ? might_fault+0x73/0xd0 [<ffffffff81344c77>] ? verify_iovec+0x47/0xd0 [<ffffffff8133a509>] sys_sendmsg+0x1a9/0x360 [<ffffffff813c2be5>] ? _raw_spin_unlock_irqrestore+0x65/0x70 [<ffffffff8106aced>] ? trace_hardirqs_on+0xd/0x10 [<ffffffff813c2bc2>] ? _raw_spin_unlock_irqrestore+0x42/0x70 [<ffffffff81197004>] ? __up_read+0x84/0xb0 [<ffffffff8106ac95>] ? trace_hardirqs_on_caller+0x145/0x190 [<ffffffff813c207f>] ? trace_hardirqs_on_thunk+0x3a/0x3f [<ffffffff8100262b>] system_call_fastpath+0x16/0x1b Code: Bad RIP value. RIP [<ffffffffa02f842a>] 0xffffffffa02f842a RSP <ffff8800baa3db48> CR2: ffffffffa02f842a If module was quickly removed after autoloading, return -E. Return -EPROTONOSUPPORT if module was quickly removed after autoloading. Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2010-01-30 10:05:05 +00:00
if (err < 0)
goto out;
err = __netlink_create(net, sock, cb_mutex, protocol);
if (err < 0)
goto out_module;
local_bh_disable();
sock_prot_inuse_add(net, &netlink_proto, 1);
local_bh_enable();
nlk = nlk_sk(sock->sk);
nlk->module = module;
out:
return err;
out_module:
module_put(module);
goto out;
}
static int netlink_release(struct socket *sock)
{
struct sock *sk = sock->sk;
struct netlink_sock *nlk;
if (!sk)
return 0;
netlink_remove(sk);
sock_orphan(sk);
nlk = nlk_sk(sk);
/*
* OK. Socket is unlinked, any packets that arrive now
* will be purged.
*/
sock->sk = NULL;
wake_up_interruptible_all(&nlk->wait);
skb_queue_purge(&sk->sk_write_queue);
if (nlk->pid) {
struct netlink_notify n = {
.net = sock_net(sk),
.protocol = sk->sk_protocol,
.pid = nlk->pid,
};
[PATCH] Notifier chain update: API changes The kernel's implementation of notifier chains is unsafe. There is no protection against entries being added to or removed from a chain while the chain is in use. The issues were discussed in this thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 We noticed that notifier chains in the kernel fall into two basic usage classes: "Blocking" chains are always called from a process context and the callout routines are allowed to sleep; "Atomic" chains can be called from an atomic context and the callout routines are not allowed to sleep. We decided to codify this distinction and make it part of the API. Therefore this set of patches introduces three new, parallel APIs: one for blocking notifiers, one for atomic notifiers, and one for "raw" notifiers (which is really just the old API under a new name). New kinds of data structures are used for the heads of the chains, and new routines are defined for registration, unregistration, and calling a chain. The three APIs are explained in include/linux/notifier.h and their implementation is in kernel/sys.c. With atomic and blocking chains, the implementation guarantees that the chain links will not be corrupted and that chain callers will not get messed up by entries being added or removed. For raw chains the implementation provides no guarantees at all; users of this API must provide their own protections. (The idea was that situations may come up where the assumptions of the atomic and blocking APIs are not appropriate, so it should be possible for users to handle these things in their own way.) There are some limitations, which should not be too hard to live with. For atomic/blocking chains, registration and unregistration must always be done in a process context since the chain is protected by a mutex/rwsem. Also, a callout routine for a non-raw chain must not try to register or unregister entries on its own chain. (This did happen in a couple of places and the code had to be changed to avoid it.) Since atomic chains may be called from within an NMI handler, they cannot use spinlocks for synchronization. Instead we use RCU. The overhead falls almost entirely in the unregister routine, which is okay since unregistration is much less frequent that calling a chain. Here is the list of chains that we adjusted and their classifications. None of them use the raw API, so for the moment it is only a placeholder. ATOMIC CHAINS ------------- arch/i386/kernel/traps.c: i386die_chain arch/ia64/kernel/traps.c: ia64die_chain arch/powerpc/kernel/traps.c: powerpc_die_chain arch/sparc64/kernel/traps.c: sparc64die_chain arch/x86_64/kernel/traps.c: die_chain drivers/char/ipmi/ipmi_si_intf.c: xaction_notifier_list kernel/panic.c: panic_notifier_list kernel/profile.c: task_free_notifier net/bluetooth/hci_core.c: hci_notifier net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_chain net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_expect_chain net/ipv6/addrconf.c: inet6addr_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_expect_chain net/netlink/af_netlink.c: netlink_chain BLOCKING CHAINS --------------- arch/powerpc/platforms/pseries/reconfig.c: pSeries_reconfig_chain arch/s390/kernel/process.c: idle_chain arch/x86_64/kernel/process.c idle_notifier drivers/base/memory.c: memory_chain drivers/cpufreq/cpufreq.c cpufreq_policy_notifier_list drivers/cpufreq/cpufreq.c cpufreq_transition_notifier_list drivers/macintosh/adb.c: adb_client_list drivers/macintosh/via-pmu.c sleep_notifier_list drivers/macintosh/via-pmu68k.c sleep_notifier_list drivers/macintosh/windfarm_core.c wf_client_list drivers/usb/core/notify.c usb_notifier_list drivers/video/fbmem.c fb_notifier_list kernel/cpu.c cpu_chain kernel/module.c module_notify_list kernel/profile.c munmap_notifier kernel/profile.c task_exit_notifier kernel/sys.c reboot_notifier_list net/core/dev.c netdev_chain net/decnet/dn_dev.c: dnaddr_chain net/ipv4/devinet.c: inetaddr_chain It's possible that some of these classifications are wrong. If they are, please let us know or submit a patch to fix them. Note that any chain that gets called very frequently should be atomic, because the rwsem read-locking used for blocking chains is very likely to incur cache misses on SMP systems. (However, if the chain's callout routines may sleep then the chain cannot be atomic.) The patch set was written by Alan Stern and Chandra Seetharaman, incorporating material written by Keith Owens and suggestions from Paul McKenney and Andrew Morton. [jes@sgi.com: restructure the notifier chain initialization macros] Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Jes Sorensen <jes@sgi.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 09:16:30 +00:00
atomic_notifier_call_chain(&netlink_chain,
NETLINK_URELEASE, &n);
}
module_put(nlk->module);
netlink_table_grab();
if (netlink_is_kernel(sk)) {
BUG_ON(nl_table[sk->sk_protocol].registered == 0);
if (--nl_table[sk->sk_protocol].registered == 0) {
kfree(nl_table[sk->sk_protocol].listeners);
nl_table[sk->sk_protocol].module = NULL;
nl_table[sk->sk_protocol].registered = 0;
}
} else if (nlk->subscriptions)
netlink_update_listeners(sk);
netlink_table_ungrab();
kfree(nlk->groups);
nlk->groups = NULL;
local_bh_disable();
sock_prot_inuse_add(sock_net(sk), &netlink_proto, -1);
local_bh_enable();
sock_put(sk);
return 0;
}
static int netlink_autobind(struct socket *sock)
{
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct nl_pid_hash *hash = &nl_table[sk->sk_protocol].hash;
struct hlist_head *head;
struct sock *osk;
struct hlist_node *node;
s32 pid = task_tgid_vnr(current);
int err;
static s32 rover = -4097;
retry:
cond_resched();
netlink_table_grab();
head = nl_pid_hashfn(hash, pid);
sk_for_each(osk, node, head) {
if (!net_eq(sock_net(osk), net))
continue;
if (nlk_sk(osk)->pid == pid) {
/* Bind collision, search negative pid values. */
pid = rover--;
if (rover > -4097)
rover = -4097;
netlink_table_ungrab();
goto retry;
}
}
netlink_table_ungrab();
err = netlink_insert(sk, net, pid);
if (err == -EADDRINUSE)
goto retry;
[NETLINK]: Fix two socket hashing bugs. 1) netlink_release() should only decrement the hash entry count if the socket was actually hashed. This was causing hash->entries to underflow, which resulting in all kinds of troubles. On 64-bit systems, this would cause the following conditional to erroneously trigger: err = -ENOMEM; if (BITS_PER_LONG > 32 && unlikely(hash->entries >= UINT_MAX)) goto err; 2) netlink_autobind() needs to propagate the error return from netlink_insert(). Otherwise, callers will not see the error as they should and thus try to operate on a socket with a zero pid, which is very bad. However, it should not propagate -EBUSY. If two threads race to autobind the socket, that is fine. This is consistent with the autobind behavior in other protocols. So bug #1 above, combined with this one, resulted in hangs on netlink_sendmsg() calls to the rtnetlink socket. We'd try to do the user sendmsg() with the socket's pid set to zero, later we do a socket lookup using that pid (via the value we stashed away in NETLINK_CB(skb).pid), but that won't give us the user socket, it will give us the rtnetlink socket. So when we try to wake up the receive queue, we dive back into rtnetlink_rcv() which tries to recursively take the rtnetlink semaphore. Thanks to Jakub Jelink for providing backtraces. Also, thanks to Herbert Xu for supplying debugging patches to help track this down, and also finding a mistake in an earlier version of this fix. Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-26 22:31:51 +00:00
/* If 2 threads race to autobind, that is fine. */
if (err == -EBUSY)
err = 0;
return err;
}
static inline int netlink_capable(struct socket *sock, unsigned int flag)
{
return (nl_table[sock->sk->sk_protocol].nl_nonroot & flag) ||
capable(CAP_NET_ADMIN);
}
static void
netlink_update_subscriptions(struct sock *sk, unsigned int subscriptions)
{
struct netlink_sock *nlk = nlk_sk(sk);
if (nlk->subscriptions && !subscriptions)
__sk_del_bind_node(sk);
else if (!nlk->subscriptions && subscriptions)
sk_add_bind_node(sk, &nl_table[sk->sk_protocol].mc_list);
nlk->subscriptions = subscriptions;
}
static int netlink_realloc_groups(struct sock *sk)
{
struct netlink_sock *nlk = nlk_sk(sk);
unsigned int groups;
unsigned long *new_groups;
int err = 0;
netlink_table_grab();
groups = nl_table[sk->sk_protocol].groups;
if (!nl_table[sk->sk_protocol].registered) {
err = -ENOENT;
goto out_unlock;
}
if (nlk->ngroups >= groups)
goto out_unlock;
new_groups = krealloc(nlk->groups, NLGRPSZ(groups), GFP_ATOMIC);
if (new_groups == NULL) {
err = -ENOMEM;
goto out_unlock;
}
memset((char *)new_groups + NLGRPSZ(nlk->ngroups), 0,
NLGRPSZ(groups) - NLGRPSZ(nlk->ngroups));
nlk->groups = new_groups;
nlk->ngroups = groups;
out_unlock:
netlink_table_ungrab();
return err;
}
static int netlink_bind(struct socket *sock, struct sockaddr *addr,
int addr_len)
{
struct sock *sk = sock->sk;
struct net *net = sock_net(sk);
struct netlink_sock *nlk = nlk_sk(sk);
struct sockaddr_nl *nladdr = (struct sockaddr_nl *)addr;
int err;
if (nladdr->nl_family != AF_NETLINK)
return -EINVAL;
/* Only superuser is allowed to listen multicasts */
if (nladdr->nl_groups) {
if (!netlink_capable(sock, NL_NONROOT_RECV))
return -EPERM;
err = netlink_realloc_groups(sk);
if (err)
return err;
}
if (nlk->pid) {
if (nladdr->nl_pid != nlk->pid)
return -EINVAL;
} else {
err = nladdr->nl_pid ?
netlink_insert(sk, net, nladdr->nl_pid) :
netlink_autobind(sock);
if (err)
return err;
}
if (!nladdr->nl_groups && (nlk->groups == NULL || !(u32)nlk->groups[0]))
return 0;
netlink_table_grab();
netlink_update_subscriptions(sk, nlk->subscriptions +
hweight32(nladdr->nl_groups) -
hweight32(nlk->groups[0]));
nlk->groups[0] = (nlk->groups[0] & ~0xffffffffUL) | nladdr->nl_groups;
netlink_update_listeners(sk);
netlink_table_ungrab();
return 0;
}
static int netlink_connect(struct socket *sock, struct sockaddr *addr,
int alen, int flags)
{
int err = 0;
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
struct sockaddr_nl *nladdr = (struct sockaddr_nl *)addr;
if (alen < sizeof(addr->sa_family))
return -EINVAL;
if (addr->sa_family == AF_UNSPEC) {
sk->sk_state = NETLINK_UNCONNECTED;
nlk->dst_pid = 0;
nlk->dst_group = 0;
return 0;
}
if (addr->sa_family != AF_NETLINK)
return -EINVAL;
/* Only superuser is allowed to send multicasts */
if (nladdr->nl_groups && !netlink_capable(sock, NL_NONROOT_SEND))
return -EPERM;
if (!nlk->pid)
err = netlink_autobind(sock);
if (err == 0) {
sk->sk_state = NETLINK_CONNECTED;
nlk->dst_pid = nladdr->nl_pid;
nlk->dst_group = ffs(nladdr->nl_groups);
}
return err;
}
static int netlink_getname(struct socket *sock, struct sockaddr *addr,
int *addr_len, int peer)
{
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
DECLARE_SOCKADDR(struct sockaddr_nl *, nladdr, addr);
nladdr->nl_family = AF_NETLINK;
nladdr->nl_pad = 0;
*addr_len = sizeof(*nladdr);
if (peer) {
nladdr->nl_pid = nlk->dst_pid;
nladdr->nl_groups = netlink_group_mask(nlk->dst_group);
} else {
nladdr->nl_pid = nlk->pid;
nladdr->nl_groups = nlk->groups ? nlk->groups[0] : 0;
}
return 0;
}
static void netlink_overrun(struct sock *sk)
{
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
struct netlink_sock *nlk = nlk_sk(sk);
if (!(nlk->flags & NETLINK_RECV_NO_ENOBUFS)) {
if (!test_and_set_bit(0, &nlk_sk(sk)->state)) {
sk->sk_err = ENOBUFS;
sk->sk_error_report(sk);
}
}
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
atomic_inc(&sk->sk_drops);
}
static struct sock *netlink_getsockbypid(struct sock *ssk, u32 pid)
{
struct sock *sock;
struct netlink_sock *nlk;
sock = netlink_lookup(sock_net(ssk), ssk->sk_protocol, pid);
if (!sock)
return ERR_PTR(-ECONNREFUSED);
/* Don't bother queuing skb if kernel socket has no input function */
nlk = nlk_sk(sock);
if (sock->sk_state == NETLINK_CONNECTED &&
nlk->dst_pid != nlk_sk(ssk)->pid) {
sock_put(sock);
return ERR_PTR(-ECONNREFUSED);
}
return sock;
}
struct sock *netlink_getsockbyfilp(struct file *filp)
{
struct inode *inode = filp->f_path.dentry->d_inode;
struct sock *sock;
if (!S_ISSOCK(inode->i_mode))
return ERR_PTR(-ENOTSOCK);
sock = SOCKET_I(inode)->sk;
if (sock->sk_family != AF_NETLINK)
return ERR_PTR(-EINVAL);
sock_hold(sock);
return sock;
}
/*
* Attach a skb to a netlink socket.
* The caller must hold a reference to the destination socket. On error, the
* reference is dropped. The skb is not send to the destination, just all
* all error checks are performed and memory in the queue is reserved.
* Return values:
* < 0: error. skb freed, reference to sock dropped.
* 0: continue
* 1: repeat lookup - reference dropped while waiting for socket memory.
*/
int netlink_attachskb(struct sock *sk, struct sk_buff *skb,
long *timeo, struct sock *ssk)
{
struct netlink_sock *nlk;
nlk = nlk_sk(sk);
if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf ||
test_bit(0, &nlk->state)) {
DECLARE_WAITQUEUE(wait, current);
if (!*timeo) {
if (!ssk || netlink_is_kernel(ssk))
netlink_overrun(sk);
sock_put(sk);
kfree_skb(skb);
return -EAGAIN;
}
__set_current_state(TASK_INTERRUPTIBLE);
add_wait_queue(&nlk->wait, &wait);
if ((atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf ||
test_bit(0, &nlk->state)) &&
!sock_flag(sk, SOCK_DEAD))
*timeo = schedule_timeout(*timeo);
__set_current_state(TASK_RUNNING);
remove_wait_queue(&nlk->wait, &wait);
sock_put(sk);
if (signal_pending(current)) {
kfree_skb(skb);
return sock_intr_errno(*timeo);
}
return 1;
}
skb_set_owner_r(skb, sk);
return 0;
}
int netlink_sendskb(struct sock *sk, struct sk_buff *skb)
{
int len = skb->len;
skb_queue_tail(&sk->sk_receive_queue, skb);
sk->sk_data_ready(sk, len);
sock_put(sk);
return len;
}
void netlink_detachskb(struct sock *sk, struct sk_buff *skb)
{
kfree_skb(skb);
sock_put(sk);
}
static inline struct sk_buff *netlink_trim(struct sk_buff *skb,
gfp_t allocation)
{
int delta;
skb_orphan(skb);
delta = skb->end - skb->tail;
if (delta * 2 < skb->truesize)
return skb;
if (skb_shared(skb)) {
struct sk_buff *nskb = skb_clone(skb, allocation);
if (!nskb)
return skb;
kfree_skb(skb);
skb = nskb;
}
if (!pskb_expand_head(skb, 0, -delta, allocation))
skb->truesize -= delta;
return skb;
}
static inline void netlink_rcv_wake(struct sock *sk)
{
struct netlink_sock *nlk = nlk_sk(sk);
if (skb_queue_empty(&sk->sk_receive_queue))
clear_bit(0, &nlk->state);
if (!test_bit(0, &nlk->state))
wake_up_interruptible(&nlk->wait);
}
static inline int netlink_unicast_kernel(struct sock *sk, struct sk_buff *skb)
{
int ret;
struct netlink_sock *nlk = nlk_sk(sk);
ret = -ECONNREFUSED;
if (nlk->netlink_rcv != NULL) {
ret = skb->len;
skb_set_owner_r(skb, sk);
nlk->netlink_rcv(skb);
}
kfree_skb(skb);
sock_put(sk);
return ret;
}
int netlink_unicast(struct sock *ssk, struct sk_buff *skb,
u32 pid, int nonblock)
{
struct sock *sk;
int err;
long timeo;
skb = netlink_trim(skb, gfp_any());
timeo = sock_sndtimeo(ssk, nonblock);
retry:
sk = netlink_getsockbypid(ssk, pid);
if (IS_ERR(sk)) {
kfree_skb(skb);
return PTR_ERR(sk);
}
if (netlink_is_kernel(sk))
return netlink_unicast_kernel(sk, skb);
if (sk_filter(sk, skb)) {
err = skb->len;
kfree_skb(skb);
sock_put(sk);
return err;
}
err = netlink_attachskb(sk, skb, &timeo, ssk);
if (err == 1)
goto retry;
if (err)
return err;
return netlink_sendskb(sk, skb);
}
EXPORT_SYMBOL(netlink_unicast);
int netlink_has_listeners(struct sock *sk, unsigned int group)
{
int res = 0;
struct listeners *listeners;
BUG_ON(!netlink_is_kernel(sk));
rcu_read_lock();
listeners = rcu_dereference(nl_table[sk->sk_protocol].listeners);
if (group - 1 < nl_table[sk->sk_protocol].groups)
res = test_bit(group - 1, listeners->masks);
rcu_read_unlock();
return res;
}
EXPORT_SYMBOL_GPL(netlink_has_listeners);
static inline int netlink_broadcast_deliver(struct sock *sk,
struct sk_buff *skb)
{
struct netlink_sock *nlk = nlk_sk(sk);
if (atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf &&
!test_bit(0, &nlk->state)) {
skb_set_owner_r(skb, sk);
skb_queue_tail(&sk->sk_receive_queue, skb);
sk->sk_data_ready(sk, skb->len);
return atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf;
}
return -1;
}
struct netlink_broadcast_data {
struct sock *exclude_sk;
struct net *net;
u32 pid;
u32 group;
int failure;
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-06 07:56:36 +00:00
int delivery_failure;
int congested;
int delivered;
gfp_t allocation;
struct sk_buff *skb, *skb2;
int (*tx_filter)(struct sock *dsk, struct sk_buff *skb, void *data);
void *tx_data;
};
static inline int do_one_broadcast(struct sock *sk,
struct netlink_broadcast_data *p)
{
struct netlink_sock *nlk = nlk_sk(sk);
int val;
if (p->exclude_sk == sk)
goto out;
if (nlk->pid == p->pid || p->group - 1 >= nlk->ngroups ||
!test_bit(p->group - 1, nlk->groups))
goto out;
if (!net_eq(sock_net(sk), p->net))
goto out;
if (p->failure) {
netlink_overrun(sk);
goto out;
}
sock_hold(sk);
if (p->skb2 == NULL) {
if (skb_shared(p->skb)) {
p->skb2 = skb_clone(p->skb, p->allocation);
} else {
p->skb2 = skb_get(p->skb);
/*
* skb ownership may have been set when
* delivered to a previous socket.
*/
skb_orphan(p->skb2);
}
}
if (p->skb2 == NULL) {
netlink_overrun(sk);
/* Clone failed. Notify ALL listeners. */
p->failure = 1;
if (nlk->flags & NETLINK_BROADCAST_SEND_ERROR)
p->delivery_failure = 1;
} else if (p->tx_filter && p->tx_filter(sk, p->skb2, p->tx_data)) {
kfree_skb(p->skb2);
p->skb2 = NULL;
} else if (sk_filter(sk, p->skb2)) {
kfree_skb(p->skb2);
p->skb2 = NULL;
} else if ((val = netlink_broadcast_deliver(sk, p->skb2)) < 0) {
netlink_overrun(sk);
if (nlk->flags & NETLINK_BROADCAST_SEND_ERROR)
p->delivery_failure = 1;
} else {
p->congested |= val;
p->delivered = 1;
p->skb2 = NULL;
}
sock_put(sk);
out:
return 0;
}
int netlink_broadcast_filtered(struct sock *ssk, struct sk_buff *skb, u32 pid,
u32 group, gfp_t allocation,
int (*filter)(struct sock *dsk, struct sk_buff *skb, void *data),
void *filter_data)
{
struct net *net = sock_net(ssk);
struct netlink_broadcast_data info;
struct hlist_node *node;
struct sock *sk;
skb = netlink_trim(skb, allocation);
info.exclude_sk = ssk;
info.net = net;
info.pid = pid;
info.group = group;
info.failure = 0;
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-06 07:56:36 +00:00
info.delivery_failure = 0;
info.congested = 0;
info.delivered = 0;
info.allocation = allocation;
info.skb = skb;
info.skb2 = NULL;
info.tx_filter = filter;
info.tx_data = filter_data;
/* While we sleep in clone, do not allow to change socket list */
netlink_lock_table();
sk_for_each_bound(sk, node, &nl_table[ssk->sk_protocol].mc_list)
do_one_broadcast(sk, &info);
consume_skb(skb);
netlink_unlock_table();
if (info.delivery_failure) {
kfree_skb(info.skb2);
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-06 07:56:36 +00:00
return -ENOBUFS;
} else
consume_skb(info.skb2);
netlink: change return-value logic of netlink_broadcast() Currently, netlink_broadcast() reports errors to the caller if no messages at all were delivered: 1) If, at least, one message has been delivered correctly, returns 0. 2) Otherwise, if no messages at all were delivered due to skb_clone() failure, return -ENOBUFS. 3) Otherwise, if there are no listeners, return -ESRCH. With this patch, the caller knows if the delivery of any of the messages to the listeners have failed: 1) If it fails to deliver any message (for whatever reason), return -ENOBUFS. 2) Otherwise, if all messages were delivered OK, returns 0. 3) Otherwise, if no listeners, return -ESRCH. In the current ctnetlink code and in Netfilter in general, we can add reliable logging and connection tracking event delivery by dropping the packets whose events were not successfully delivered over Netlink. Of course, this option would be settable via /proc as this approach reduces performance (in terms of filtered connections per seconds by a stateful firewall) but providing reliable logging and event delivery (for conntrackd) in return. This patch also changes some clients of netlink_broadcast() that may report ENOBUFS errors via printk. This error handling is not of any help. Instead, the userspace daemons that are listening to those netlink messages should resync themselves with the kernel-side if they hit ENOBUFS. BTW, netlink_broadcast() clients include those that call cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they internally call netlink_broadcast() and return its error value. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-06 07:56:36 +00:00
if (info.delivered) {
if (info.congested && (allocation & __GFP_WAIT))
yield();
return 0;
}
return -ESRCH;
}
EXPORT_SYMBOL(netlink_broadcast_filtered);
int netlink_broadcast(struct sock *ssk, struct sk_buff *skb, u32 pid,
u32 group, gfp_t allocation)
{
return netlink_broadcast_filtered(ssk, skb, pid, group, allocation,
NULL, NULL);
}
EXPORT_SYMBOL(netlink_broadcast);
struct netlink_set_err_data {
struct sock *exclude_sk;
u32 pid;
u32 group;
int code;
};
static inline int do_one_set_err(struct sock *sk,
struct netlink_set_err_data *p)
{
struct netlink_sock *nlk = nlk_sk(sk);
int ret = 0;
if (sk == p->exclude_sk)
goto out;
if (!net_eq(sock_net(sk), sock_net(p->exclude_sk)))
goto out;
if (nlk->pid == p->pid || p->group - 1 >= nlk->ngroups ||
!test_bit(p->group - 1, nlk->groups))
goto out;
if (p->code == ENOBUFS && nlk->flags & NETLINK_RECV_NO_ENOBUFS) {
ret = 1;
goto out;
}
sk->sk_err = p->code;
sk->sk_error_report(sk);
out:
return ret;
}
/**
* netlink_set_err - report error to broadcast listeners
* @ssk: the kernel netlink socket, as returned by netlink_kernel_create()
* @pid: the PID of a process that we want to skip (if any)
* @groups: the broadcast group that will notice the error
* @code: error code, must be negative (as usual in kernelspace)
*
* This function returns the number of broadcast listeners that have set the
* NETLINK_RECV_NO_ENOBUFS socket option.
*/
int netlink_set_err(struct sock *ssk, u32 pid, u32 group, int code)
{
struct netlink_set_err_data info;
struct hlist_node *node;
struct sock *sk;
int ret = 0;
info.exclude_sk = ssk;
info.pid = pid;
info.group = group;
/* sk->sk_err wants a positive error value */
info.code = -code;
read_lock(&nl_table_lock);
sk_for_each_bound(sk, node, &nl_table[ssk->sk_protocol].mc_list)
ret += do_one_set_err(sk, &info);
read_unlock(&nl_table_lock);
return ret;
}
EXPORT_SYMBOL(netlink_set_err);
/* must be called with netlink table grabbed */
static void netlink_update_socket_mc(struct netlink_sock *nlk,
unsigned int group,
int is_new)
{
int old, new = !!is_new, subscriptions;
old = test_bit(group - 1, nlk->groups);
subscriptions = nlk->subscriptions - old + new;
if (new)
__set_bit(group - 1, nlk->groups);
else
__clear_bit(group - 1, nlk->groups);
netlink_update_subscriptions(&nlk->sk, subscriptions);
netlink_update_listeners(&nlk->sk);
}
static int netlink_setsockopt(struct socket *sock, int level, int optname,
char __user *optval, unsigned int optlen)
{
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
unsigned int val = 0;
int err;
if (level != SOL_NETLINK)
return -ENOPROTOOPT;
if (optlen >= sizeof(int) &&
get_user(val, (unsigned int __user *)optval))
return -EFAULT;
switch (optname) {
case NETLINK_PKTINFO:
if (val)
nlk->flags |= NETLINK_RECV_PKTINFO;
else
nlk->flags &= ~NETLINK_RECV_PKTINFO;
err = 0;
break;
case NETLINK_ADD_MEMBERSHIP:
case NETLINK_DROP_MEMBERSHIP: {
if (!netlink_capable(sock, NL_NONROOT_RECV))
return -EPERM;
err = netlink_realloc_groups(sk);
if (err)
return err;
if (!val || val - 1 >= nlk->ngroups)
return -EINVAL;
netlink_table_grab();
netlink_update_socket_mc(nlk, val,
optname == NETLINK_ADD_MEMBERSHIP);
netlink_table_ungrab();
err = 0;
break;
}
case NETLINK_BROADCAST_ERROR:
if (val)
nlk->flags |= NETLINK_BROADCAST_SEND_ERROR;
else
nlk->flags &= ~NETLINK_BROADCAST_SEND_ERROR;
err = 0;
break;
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
case NETLINK_NO_ENOBUFS:
if (val) {
nlk->flags |= NETLINK_RECV_NO_ENOBUFS;
clear_bit(0, &nlk->state);
wake_up_interruptible(&nlk->wait);
} else
nlk->flags &= ~NETLINK_RECV_NO_ENOBUFS;
err = 0;
break;
default:
err = -ENOPROTOOPT;
}
return err;
}
static int netlink_getsockopt(struct socket *sock, int level, int optname,
char __user *optval, int __user *optlen)
{
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
int len, val, err;
if (level != SOL_NETLINK)
return -ENOPROTOOPT;
if (get_user(len, optlen))
return -EFAULT;
if (len < 0)
return -EINVAL;
switch (optname) {
case NETLINK_PKTINFO:
if (len < sizeof(int))
return -EINVAL;
len = sizeof(int);
val = nlk->flags & NETLINK_RECV_PKTINFO ? 1 : 0;
if (put_user(len, optlen) ||
put_user(val, optval))
return -EFAULT;
err = 0;
break;
case NETLINK_BROADCAST_ERROR:
if (len < sizeof(int))
return -EINVAL;
len = sizeof(int);
val = nlk->flags & NETLINK_BROADCAST_SEND_ERROR ? 1 : 0;
if (put_user(len, optlen) ||
put_user(val, optval))
return -EFAULT;
err = 0;
break;
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
case NETLINK_NO_ENOBUFS:
if (len < sizeof(int))
return -EINVAL;
len = sizeof(int);
val = nlk->flags & NETLINK_RECV_NO_ENOBUFS ? 1 : 0;
if (put_user(len, optlen) ||
put_user(val, optval))
return -EFAULT;
err = 0;
break;
default:
err = -ENOPROTOOPT;
}
return err;
}
static void netlink_cmsg_recv_pktinfo(struct msghdr *msg, struct sk_buff *skb)
{
struct nl_pktinfo info;
info.group = NETLINK_CB(skb).dst_group;
put_cmsg(msg, SOL_NETLINK, NETLINK_PKTINFO, sizeof(info), &info);
}
static int netlink_sendmsg(struct kiocb *kiocb, struct socket *sock,
struct msghdr *msg, size_t len)
{
struct sock_iocb *siocb = kiocb_to_siocb(kiocb);
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
struct sockaddr_nl *addr = msg->msg_name;
u32 dst_pid;
u32 dst_group;
struct sk_buff *skb;
int err;
struct scm_cookie scm;
if (msg->msg_flags&MSG_OOB)
return -EOPNOTSUPP;
if (NULL == siocb->scm) {
siocb->scm = &scm;
memset(&scm, 0, sizeof(scm));
}
err = scm_send(sock, msg, siocb->scm);
if (err < 0)
return err;
if (msg->msg_namelen) {
err = -EINVAL;
if (addr->nl_family != AF_NETLINK)
goto out;
dst_pid = addr->nl_pid;
dst_group = ffs(addr->nl_groups);
err = -EPERM;
if (dst_group && !netlink_capable(sock, NL_NONROOT_SEND))
goto out;
} else {
dst_pid = nlk->dst_pid;
dst_group = nlk->dst_group;
}
if (!nlk->pid) {
err = netlink_autobind(sock);
if (err)
goto out;
}
err = -EMSGSIZE;
if (len > sk->sk_sndbuf - 32)
goto out;
err = -ENOBUFS;
skb = alloc_skb(len, GFP_KERNEL);
if (skb == NULL)
goto out;
NETLINK_CB(skb).pid = nlk->pid;
NETLINK_CB(skb).dst_group = dst_group;
memcpy(NETLINK_CREDS(skb), &siocb->scm->creds, sizeof(struct ucred));
err = -EFAULT;
if (memcpy_fromiovec(skb_put(skb, len), msg->msg_iov, len)) {
kfree_skb(skb);
goto out;
}
err = security_netlink_send(sk, skb);
if (err) {
kfree_skb(skb);
goto out;
}
if (dst_group) {
atomic_inc(&skb->users);
netlink_broadcast(sk, skb, dst_pid, dst_group, GFP_KERNEL);
}
err = netlink_unicast(sk, skb, dst_pid, msg->msg_flags&MSG_DONTWAIT);
out:
scm_destroy(siocb->scm);
return err;
}
static int netlink_recvmsg(struct kiocb *kiocb, struct socket *sock,
struct msghdr *msg, size_t len,
int flags)
{
struct sock_iocb *siocb = kiocb_to_siocb(kiocb);
struct scm_cookie scm;
struct sock *sk = sock->sk;
struct netlink_sock *nlk = nlk_sk(sk);
int noblock = flags&MSG_DONTWAIT;
size_t copied;
struct sk_buff *skb, *data_skb;
int err, ret;
if (flags&MSG_OOB)
return -EOPNOTSUPP;
copied = 0;
skb = skb_recv_datagram(sk, flags, noblock, &err);
if (skb == NULL)
goto out;
data_skb = skb;
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
#ifdef CONFIG_COMPAT_NETLINK_MESSAGES
if (unlikely(skb_shinfo(skb)->frag_list)) {
/*
* If this skb has a frag_list, then here that means that we
* will have to use the frag_list skb's data for compat tasks
* and the regular skb's data for normal (non-compat) tasks.
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
*
* If we need to send the compat skb, assign it to the
* 'data_skb' variable so that it will be used below for data
* copying. We keep 'skb' for everything else, including
* freeing both later.
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
*/
if (flags & MSG_CMSG_COMPAT)
data_skb = skb_shinfo(skb)->frag_list;
net/compat/wext: send different messages to compat tasks Wireless extensions have the unfortunate problem that events are multicast netlink messages, and are not independent of pointer size. Thus, currently 32-bit tasks on 64-bit platforms cannot properly receive events and fail with all kinds of strange problems, for instance wpa_supplicant never notices disassociations, due to the way the 64-bit event looks (to a 32-bit process), the fact that the address is all zeroes is lost, it thinks instead it is 00:00:00:00:01:00. The same problem existed with the ioctls, until David Miller fixed those some time ago in an heroic effort. A different problem caused by this is that we cannot send the ASSOCREQIE/ASSOCRESPIE events because sending them causes a 32-bit wpa_supplicant on a 64-bit system to overwrite its internal information, which is worse than it not getting the information at all -- so we currently resort to sending a custom string event that it then parses. This, however, has a severe size limitation we are frequently hitting with modern access points; this limitation would can be lifted after this patch by sending the correct binary, not custom, event. A similar problem apparently happens for some other netlink users on x86_64 with 32-bit tasks due to the alignment for 64-bit quantities. In order to fix these problems, I have implemented a way to send compat messages to tasks. When sending an event, we send the non-compat event data together with a compat event data in skb_shinfo(main_skb)->frag_list. Then, when the event is read from the socket, the netlink code makes sure to pass out only the skb that is compatible with the task. This approach was suggested by David Miller, my original approach required always sending two skbs but that had various small problems. To determine whether compat is needed or not, I have used the MSG_CMSG_COMPAT flag, and adjusted the call path for recv and recvfrom to include it, even if those calls do not have a cmsg parameter. I have not solved one small part of the problem, and I don't think it is necessary to: if a 32-bit application uses read() rather than any form of recvmsg() it will still get the wrong (64-bit) event. However, neither do applications actually do this, nor would it be a regression. Signed-off-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-07-01 11:26:02 +00:00
}
#endif
msg->msg_namelen = 0;
copied = data_skb->len;
if (len < copied) {
msg->msg_flags |= MSG_TRUNC;
copied = len;
}
skb_reset_transport_header(data_skb);
err = skb_copy_datagram_iovec(data_skb, 0, msg->msg_iov, copied);
if (msg->msg_name) {
struct sockaddr_nl *addr = (struct sockaddr_nl *)msg->msg_name;
addr->nl_family = AF_NETLINK;
addr->nl_pad = 0;
addr->nl_pid = NETLINK_CB(skb).pid;
addr->nl_groups = netlink_group_mask(NETLINK_CB(skb).dst_group);
msg->msg_namelen = sizeof(*addr);
}
if (nlk->flags & NETLINK_RECV_PKTINFO)
netlink_cmsg_recv_pktinfo(msg, skb);
if (NULL == siocb->scm) {
memset(&scm, 0, sizeof(scm));
siocb->scm = &scm;
}
siocb->scm->creds = *NETLINK_CREDS(skb);
if (flags & MSG_TRUNC)
copied = data_skb->len;
skb_free_datagram(sk, skb);
if (nlk->cb && atomic_read(&sk->sk_rmem_alloc) <= sk->sk_rcvbuf / 2) {
ret = netlink_dump(sk);
if (ret) {
sk->sk_err = ret;
sk->sk_error_report(sk);
}
}
scm_recv(sock, msg, siocb->scm, flags);
out:
netlink_rcv_wake(sk);
return err ? : copied;
}
static void netlink_data_ready(struct sock *sk, int len)
{
BUG();
}
/*
* We export these functions to other modules. They provide a
* complete set of kernel non-blocking support for message
* queueing.
*/
struct sock *
netlink_kernel_create(struct net *net, int unit, unsigned int groups,
void (*input)(struct sk_buff *skb),
struct mutex *cb_mutex, struct module *module)
{
struct socket *sock;
struct sock *sk;
struct netlink_sock *nlk;
struct listeners *listeners = NULL;
BUG_ON(!nl_table);
if (unit < 0 || unit >= MAX_LINKS)
return NULL;
if (sock_create_lite(PF_NETLINK, SOCK_DGRAM, unit, &sock))
return NULL;
[NETNS]: Fix race between put_net() and netlink_kernel_create(). The comment about "race free view of the set of network namespaces" was a bit hasty. Look (there even can be only one CPU, as discovered by Alexey Dobriyan and Denis Lunev): put_net() if (atomic_dec_and_test(&net->refcnt)) /* true */ __put_net(net); queue_work(...); /* * note: the net now has refcnt 0, but still in * the global list of net namespaces */ == re-schedule == register_pernet_subsys(&some_ops); register_pernet_operations(&some_ops); (*some_ops)->init(net); /* * we call netlink_kernel_create() here * in some places */ netlink_kernel_create(); sk_alloc(); get_net(net); /* refcnt = 1 */ /* * now we drop the net refcount not to * block the net namespace exit in the * future (or this can be done on the * error path) */ put_net(sk->sk_net); if (atomic_dec_and_test(&...)) /* * true. BOOOM! The net is * scheduled for release twice */ When thinking on this problem, I decided, that getting and putting the net in init callback is wrong. If some init callback needs to have a refcount-less reference on the struct net, _it_ has to be careful himself, rather than relying on the infrastructure to handle this correctly. In case of netlink_kernel_create(), the problem is that the sk_alloc() gets the given namespace, but passing the info that we don't want to get it inside this call is too heavy. Instead, I propose to crate the socket inside an init_net namespace and then re-attach it to the desired one right after the socket is created. After doing this, we also have to be careful on error paths not to drop the reference on the namespace, we didn't get the one on. Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Denis Lunev <den@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-31 03:31:06 +00:00
/*
* We have to just have a reference on the net from sk, but don't
* get_net it. Besides, we cannot get and then put the net here.
* So we create one inside init_net and the move it to net.
*/
if (__netlink_create(&init_net, sock, cb_mutex, unit) < 0)
goto out_sock_release_nosk;
sk = sock->sk;
sk_change_net(sk, net);
if (groups < 32)
groups = 32;
listeners = kzalloc(sizeof(*listeners) + NLGRPSZ(groups), GFP_KERNEL);
if (!listeners)
goto out_sock_release;
sk->sk_data_ready = netlink_data_ready;
if (input)
nlk_sk(sk)->netlink_rcv = input;
if (netlink_insert(sk, net, 0))
goto out_sock_release;
nlk = nlk_sk(sk);
nlk->flags |= NETLINK_KERNEL_SOCKET;
netlink_table_grab();
if (!nl_table[unit].registered) {
nl_table[unit].groups = groups;
rcu_assign_pointer(nl_table[unit].listeners, listeners);
nl_table[unit].cb_mutex = cb_mutex;
nl_table[unit].module = module;
nl_table[unit].registered = 1;
} else {
kfree(listeners);
nl_table[unit].registered++;
}
netlink_table_ungrab();
return sk;
out_sock_release:
kfree(listeners);
netlink_kernel_release(sk);
[NETNS]: Fix race between put_net() and netlink_kernel_create(). The comment about "race free view of the set of network namespaces" was a bit hasty. Look (there even can be only one CPU, as discovered by Alexey Dobriyan and Denis Lunev): put_net() if (atomic_dec_and_test(&net->refcnt)) /* true */ __put_net(net); queue_work(...); /* * note: the net now has refcnt 0, but still in * the global list of net namespaces */ == re-schedule == register_pernet_subsys(&some_ops); register_pernet_operations(&some_ops); (*some_ops)->init(net); /* * we call netlink_kernel_create() here * in some places */ netlink_kernel_create(); sk_alloc(); get_net(net); /* refcnt = 1 */ /* * now we drop the net refcount not to * block the net namespace exit in the * future (or this can be done on the * error path) */ put_net(sk->sk_net); if (atomic_dec_and_test(&...)) /* * true. BOOOM! The net is * scheduled for release twice */ When thinking on this problem, I decided, that getting and putting the net in init callback is wrong. If some init callback needs to have a refcount-less reference on the struct net, _it_ has to be careful himself, rather than relying on the infrastructure to handle this correctly. In case of netlink_kernel_create(), the problem is that the sk_alloc() gets the given namespace, but passing the info that we don't want to get it inside this call is too heavy. Instead, I propose to crate the socket inside an init_net namespace and then re-attach it to the desired one right after the socket is created. After doing this, we also have to be careful on error paths not to drop the reference on the namespace, we didn't get the one on. Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Denis Lunev <den@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2008-01-31 03:31:06 +00:00
return NULL;
out_sock_release_nosk:
sock_release(sock);
return NULL;
}
EXPORT_SYMBOL(netlink_kernel_create);
void
netlink_kernel_release(struct sock *sk)
{
sk_release_kernel(sk);
}
EXPORT_SYMBOL(netlink_kernel_release);
static void listeners_free_rcu(struct rcu_head *head)
{
kfree(container_of(head, struct listeners, rcu));
}
int __netlink_change_ngroups(struct sock *sk, unsigned int groups)
{
struct listeners *new, *old;
struct netlink_table *tbl = &nl_table[sk->sk_protocol];
if (groups < 32)
groups = 32;
if (NLGRPSZ(tbl->groups) < NLGRPSZ(groups)) {
new = kzalloc(sizeof(*new) + NLGRPSZ(groups), GFP_ATOMIC);
if (!new)
return -ENOMEM;
old = rcu_dereference_raw(tbl->listeners);
memcpy(new->masks, old->masks, NLGRPSZ(tbl->groups));
rcu_assign_pointer(tbl->listeners, new);
call_rcu(&old->rcu, listeners_free_rcu);
}
tbl->groups = groups;
return 0;
}
/**
* netlink_change_ngroups - change number of multicast groups
*
* This changes the number of multicast groups that are available
* on a certain netlink family. Note that it is not possible to
* change the number of groups to below 32. Also note that it does
* not implicitly call netlink_clear_multicast_users() when the
* number of groups is reduced.
*
* @sk: The kernel netlink socket, as returned by netlink_kernel_create().
* @groups: The new number of groups.
*/
int netlink_change_ngroups(struct sock *sk, unsigned int groups)
{
int err;
netlink_table_grab();
err = __netlink_change_ngroups(sk, groups);
netlink_table_ungrab();
return err;
}
void __netlink_clear_multicast_users(struct sock *ksk, unsigned int group)
{
struct sock *sk;
struct hlist_node *node;
struct netlink_table *tbl = &nl_table[ksk->sk_protocol];
sk_for_each_bound(sk, node, &tbl->mc_list)
netlink_update_socket_mc(nlk_sk(sk), group, 0);
}
/**
* netlink_clear_multicast_users - kick off multicast listeners
*
* This function removes all listeners from the given group.
* @ksk: The kernel netlink socket, as returned by
* netlink_kernel_create().
* @group: The multicast group to clear.
*/
void netlink_clear_multicast_users(struct sock *ksk, unsigned int group)
{
netlink_table_grab();
__netlink_clear_multicast_users(ksk, group);
netlink_table_ungrab();
}
void netlink_set_nonroot(int protocol, unsigned int flags)
{
if ((unsigned int)protocol < MAX_LINKS)
nl_table[protocol].nl_nonroot = flags;
}
EXPORT_SYMBOL(netlink_set_nonroot);
static void netlink_destroy_callback(struct netlink_callback *cb)
{
kfree_skb(cb->skb);
kfree(cb);
}
/*
* It looks a bit ugly.
* It would be better to create kernel thread.
*/
static int netlink_dump(struct sock *sk)
{
struct netlink_sock *nlk = nlk_sk(sk);
struct netlink_callback *cb;
struct sk_buff *skb;
struct nlmsghdr *nlh;
int len, err = -ENOBUFS;
skb = sock_rmalloc(sk, NLMSG_GOODSIZE, 0, GFP_KERNEL);
if (!skb)
goto errout;
mutex_lock(nlk->cb_mutex);
cb = nlk->cb;
if (cb == NULL) {
err = -EINVAL;
goto errout_skb;
}
len = cb->dump(skb, cb);
if (len > 0) {
mutex_unlock(nlk->cb_mutex);
if (sk_filter(sk, skb))
kfree_skb(skb);
else {
skb_queue_tail(&sk->sk_receive_queue, skb);
sk->sk_data_ready(sk, skb->len);
}
return 0;
}
nlh = nlmsg_put_answer(skb, cb, NLMSG_DONE, sizeof(len), NLM_F_MULTI);
if (!nlh)
goto errout_skb;
memcpy(nlmsg_data(nlh), &len, sizeof(len));
if (sk_filter(sk, skb))
kfree_skb(skb);
else {
skb_queue_tail(&sk->sk_receive_queue, skb);
sk->sk_data_ready(sk, skb->len);
}
if (cb->done)
cb->done(cb);
nlk->cb = NULL;
mutex_unlock(nlk->cb_mutex);
netlink_destroy_callback(cb);
return 0;
errout_skb:
mutex_unlock(nlk->cb_mutex);
kfree_skb(skb);
errout:
return err;
}
int netlink_dump_start(struct sock *ssk, struct sk_buff *skb,
const struct nlmsghdr *nlh,
int (*dump)(struct sk_buff *skb,
struct netlink_callback *),
int (*done)(struct netlink_callback *))
{
struct netlink_callback *cb;
struct sock *sk;
struct netlink_sock *nlk;
int ret;
cb = kzalloc(sizeof(*cb), GFP_KERNEL);
if (cb == NULL)
return -ENOBUFS;
cb->dump = dump;
cb->done = done;
cb->nlh = nlh;
atomic_inc(&skb->users);
cb->skb = skb;
sk = netlink_lookup(sock_net(ssk), ssk->sk_protocol, NETLINK_CB(skb).pid);
if (sk == NULL) {
netlink_destroy_callback(cb);
return -ECONNREFUSED;
}
nlk = nlk_sk(sk);
/* A dump is in progress... */
mutex_lock(nlk->cb_mutex);
if (nlk->cb) {
mutex_unlock(nlk->cb_mutex);
netlink_destroy_callback(cb);
sock_put(sk);
return -EBUSY;
}
nlk->cb = cb;
mutex_unlock(nlk->cb_mutex);
ret = netlink_dump(sk);
sock_put(sk);
if (ret)
return ret;
/* We successfully started a dump, by returning -EINTR we
* signal not to send ACK even if it was requested.
*/
return -EINTR;
}
EXPORT_SYMBOL(netlink_dump_start);
void netlink_ack(struct sk_buff *in_skb, struct nlmsghdr *nlh, int err)
{
struct sk_buff *skb;
struct nlmsghdr *rep;
struct nlmsgerr *errmsg;
size_t payload = sizeof(*errmsg);
/* error messages get the original request appened */
if (err)
payload += nlmsg_len(nlh);
skb = nlmsg_new(payload, GFP_KERNEL);
if (!skb) {
struct sock *sk;
sk = netlink_lookup(sock_net(in_skb->sk),
in_skb->sk->sk_protocol,
NETLINK_CB(in_skb).pid);
if (sk) {
sk->sk_err = ENOBUFS;
sk->sk_error_report(sk);
sock_put(sk);
}
return;
}
rep = __nlmsg_put(skb, NETLINK_CB(in_skb).pid, nlh->nlmsg_seq,
NLMSG_ERROR, payload, 0);
errmsg = nlmsg_data(rep);
errmsg->error = err;
memcpy(&errmsg->msg, nlh, err ? nlh->nlmsg_len : sizeof(*nlh));
netlink_unicast(in_skb->sk, skb, NETLINK_CB(in_skb).pid, MSG_DONTWAIT);
}
EXPORT_SYMBOL(netlink_ack);
int netlink_rcv_skb(struct sk_buff *skb, int (*cb)(struct sk_buff *,
struct nlmsghdr *))
{
struct nlmsghdr *nlh;
int err;
while (skb->len >= nlmsg_total_size(0)) {
int msglen;
nlh = nlmsg_hdr(skb);
err = 0;
if (nlh->nlmsg_len < NLMSG_HDRLEN || skb->len < nlh->nlmsg_len)
return 0;
/* Only requests are handled by the kernel */
if (!(nlh->nlmsg_flags & NLM_F_REQUEST))
goto ack;
/* Skip control messages */
if (nlh->nlmsg_type < NLMSG_MIN_TYPE)
goto ack;
err = cb(skb, nlh);
if (err == -EINTR)
goto skip;
ack:
if (nlh->nlmsg_flags & NLM_F_ACK || err)
netlink_ack(skb, nlh, err);
skip:
msglen = NLMSG_ALIGN(nlh->nlmsg_len);
if (msglen > skb->len)
msglen = skb->len;
skb_pull(skb, msglen);
}
return 0;
}
EXPORT_SYMBOL(netlink_rcv_skb);
/**
* nlmsg_notify - send a notification netlink message
* @sk: netlink socket to use
* @skb: notification message
* @pid: destination netlink pid for reports or 0
* @group: destination multicast group or 0
* @report: 1 to report back, 0 to disable
* @flags: allocation flags
*/
int nlmsg_notify(struct sock *sk, struct sk_buff *skb, u32 pid,
unsigned int group, int report, gfp_t flags)
{
int err = 0;
if (group) {
int exclude_pid = 0;
if (report) {
atomic_inc(&skb->users);
exclude_pid = pid;
}
2009-02-25 07:18:28 +00:00
/* errors reported via destination sk->sk_err, but propagate
* delivery errors if NETLINK_BROADCAST_ERROR flag is set */
err = nlmsg_multicast(sk, skb, exclude_pid, group, flags);
}
2009-02-25 07:18:28 +00:00
if (report) {
int err2;
err2 = nlmsg_unicast(sk, skb, pid);
if (!err || err == -ESRCH)
err = err2;
}
return err;
}
EXPORT_SYMBOL(nlmsg_notify);
#ifdef CONFIG_PROC_FS
struct nl_seq_iter {
struct seq_net_private p;
int link;
int hash_idx;
};
static struct sock *netlink_seq_socket_idx(struct seq_file *seq, loff_t pos)
{
struct nl_seq_iter *iter = seq->private;
int i, j;
struct sock *s;
struct hlist_node *node;
loff_t off = 0;
for (i = 0; i < MAX_LINKS; i++) {
struct nl_pid_hash *hash = &nl_table[i].hash;
for (j = 0; j <= hash->mask; j++) {
sk_for_each(s, node, &hash->table[j]) {
if (sock_net(s) != seq_file_net(seq))
continue;
if (off == pos) {
iter->link = i;
iter->hash_idx = j;
return s;
}
++off;
}
}
}
return NULL;
}
static void *netlink_seq_start(struct seq_file *seq, loff_t *pos)
__acquires(nl_table_lock)
{
read_lock(&nl_table_lock);
return *pos ? netlink_seq_socket_idx(seq, *pos - 1) : SEQ_START_TOKEN;
}
static void *netlink_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
struct sock *s;
struct nl_seq_iter *iter;
int i, j;
++*pos;
if (v == SEQ_START_TOKEN)
return netlink_seq_socket_idx(seq, 0);
iter = seq->private;
s = v;
do {
s = sk_next(s);
} while (s && sock_net(s) != seq_file_net(seq));
if (s)
return s;
i = iter->link;
j = iter->hash_idx + 1;
do {
struct nl_pid_hash *hash = &nl_table[i].hash;
for (; j <= hash->mask; j++) {
s = sk_head(&hash->table[j]);
while (s && sock_net(s) != seq_file_net(seq))
s = sk_next(s);
if (s) {
iter->link = i;
iter->hash_idx = j;
return s;
}
}
j = 0;
} while (++i < MAX_LINKS);
return NULL;
}
static void netlink_seq_stop(struct seq_file *seq, void *v)
__releases(nl_table_lock)
{
read_unlock(&nl_table_lock);
}
static int netlink_seq_show(struct seq_file *seq, void *v)
{
if (v == SEQ_START_TOKEN)
seq_puts(seq,
"sk Eth Pid Groups "
"Rmem Wmem Dump Locks Drops Inode\n");
else {
struct sock *s = v;
struct netlink_sock *nlk = nlk_sk(s);
seq_printf(seq, "%p %-3d %-6d %08x %-8d %-8d %p %-8d %-8d %-8lu\n",
s,
s->sk_protocol,
nlk->pid,
nlk->groups ? (u32)nlk->groups[0] : 0,
sk_rmem_alloc_get(s),
sk_wmem_alloc_get(s),
nlk->cb,
netlink: add NETLINK_NO_ENOBUFS socket flag This patch adds the NETLINK_NO_ENOBUFS socket flag. This flag can be used by unicast and broadcast listeners to avoid receiving ENOBUFS errors. Generally speaking, ENOBUFS errors are useful to notify two things to the listener: a) You may increase the receiver buffer size via setsockopt(). b) You have lost messages, you may be out of sync. In some cases, ignoring ENOBUFS errors can be useful. For example: a) nfnetlink_queue: this subsystem does not have any sort of resync method and you can decide to ignore ENOBUFS once you have set a given buffer size. b) ctnetlink: you can use this together with the socket flag NETLINK_BROADCAST_SEND_ERROR to stop getting ENOBUFS errors as you do not need to resync (packets whose event are not delivered are drop to provide reliable logging and state-synchronization). Moreover, the use of NETLINK_NO_ENOBUFS also reduces a "go up, go down" effect in terms of performance which is due to the netlink congestion control when the listener cannot back off. The effect is the following: 1) throughput rate goes up and netlink messages are inserted in the receiver buffer. 2) Then, netlink buffer fills and overruns (set on nlk->state bit 0). 3) While the listener empties the receiver buffer, netlink keeps dropping messages. Thus, throughput goes dramatically down. 4) Then, once the listener has emptied the buffer (nlk->state bit 0 is set off), goto step 1. This effect is easy to trigger with netlink broadcast under heavy load, and it is more noticeable when using a big receiver buffer. You can find some results in [1] that show this problem. [1] http://1984.lsi.us.es/linux/netlink/ This patch also includes the use of sk_drop to account the number of netlink messages drop due to overrun. This value is shown in /proc/net/netlink. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: David S. Miller <davem@davemloft.net>
2009-03-24 23:37:55 +00:00
atomic_read(&s->sk_refcnt),
atomic_read(&s->sk_drops),
sock_i_ino(s)
);
}
return 0;
}
static const struct seq_operations netlink_seq_ops = {
.start = netlink_seq_start,
.next = netlink_seq_next,
.stop = netlink_seq_stop,
.show = netlink_seq_show,
};
static int netlink_seq_open(struct inode *inode, struct file *file)
{
return seq_open_net(inode, file, &netlink_seq_ops,
sizeof(struct nl_seq_iter));
}
static const struct file_operations netlink_seq_fops = {
.owner = THIS_MODULE,
.open = netlink_seq_open,
.read = seq_read,
.llseek = seq_lseek,
.release = seq_release_net,
};
#endif
int netlink_register_notifier(struct notifier_block *nb)
{
[PATCH] Notifier chain update: API changes The kernel's implementation of notifier chains is unsafe. There is no protection against entries being added to or removed from a chain while the chain is in use. The issues were discussed in this thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 We noticed that notifier chains in the kernel fall into two basic usage classes: "Blocking" chains are always called from a process context and the callout routines are allowed to sleep; "Atomic" chains can be called from an atomic context and the callout routines are not allowed to sleep. We decided to codify this distinction and make it part of the API. Therefore this set of patches introduces three new, parallel APIs: one for blocking notifiers, one for atomic notifiers, and one for "raw" notifiers (which is really just the old API under a new name). New kinds of data structures are used for the heads of the chains, and new routines are defined for registration, unregistration, and calling a chain. The three APIs are explained in include/linux/notifier.h and their implementation is in kernel/sys.c. With atomic and blocking chains, the implementation guarantees that the chain links will not be corrupted and that chain callers will not get messed up by entries being added or removed. For raw chains the implementation provides no guarantees at all; users of this API must provide their own protections. (The idea was that situations may come up where the assumptions of the atomic and blocking APIs are not appropriate, so it should be possible for users to handle these things in their own way.) There are some limitations, which should not be too hard to live with. For atomic/blocking chains, registration and unregistration must always be done in a process context since the chain is protected by a mutex/rwsem. Also, a callout routine for a non-raw chain must not try to register or unregister entries on its own chain. (This did happen in a couple of places and the code had to be changed to avoid it.) Since atomic chains may be called from within an NMI handler, they cannot use spinlocks for synchronization. Instead we use RCU. The overhead falls almost entirely in the unregister routine, which is okay since unregistration is much less frequent that calling a chain. Here is the list of chains that we adjusted and their classifications. None of them use the raw API, so for the moment it is only a placeholder. ATOMIC CHAINS ------------- arch/i386/kernel/traps.c: i386die_chain arch/ia64/kernel/traps.c: ia64die_chain arch/powerpc/kernel/traps.c: powerpc_die_chain arch/sparc64/kernel/traps.c: sparc64die_chain arch/x86_64/kernel/traps.c: die_chain drivers/char/ipmi/ipmi_si_intf.c: xaction_notifier_list kernel/panic.c: panic_notifier_list kernel/profile.c: task_free_notifier net/bluetooth/hci_core.c: hci_notifier net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_chain net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_expect_chain net/ipv6/addrconf.c: inet6addr_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_expect_chain net/netlink/af_netlink.c: netlink_chain BLOCKING CHAINS --------------- arch/powerpc/platforms/pseries/reconfig.c: pSeries_reconfig_chain arch/s390/kernel/process.c: idle_chain arch/x86_64/kernel/process.c idle_notifier drivers/base/memory.c: memory_chain drivers/cpufreq/cpufreq.c cpufreq_policy_notifier_list drivers/cpufreq/cpufreq.c cpufreq_transition_notifier_list drivers/macintosh/adb.c: adb_client_list drivers/macintosh/via-pmu.c sleep_notifier_list drivers/macintosh/via-pmu68k.c sleep_notifier_list drivers/macintosh/windfarm_core.c wf_client_list drivers/usb/core/notify.c usb_notifier_list drivers/video/fbmem.c fb_notifier_list kernel/cpu.c cpu_chain kernel/module.c module_notify_list kernel/profile.c munmap_notifier kernel/profile.c task_exit_notifier kernel/sys.c reboot_notifier_list net/core/dev.c netdev_chain net/decnet/dn_dev.c: dnaddr_chain net/ipv4/devinet.c: inetaddr_chain It's possible that some of these classifications are wrong. If they are, please let us know or submit a patch to fix them. Note that any chain that gets called very frequently should be atomic, because the rwsem read-locking used for blocking chains is very likely to incur cache misses on SMP systems. (However, if the chain's callout routines may sleep then the chain cannot be atomic.) The patch set was written by Alan Stern and Chandra Seetharaman, incorporating material written by Keith Owens and suggestions from Paul McKenney and Andrew Morton. [jes@sgi.com: restructure the notifier chain initialization macros] Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Jes Sorensen <jes@sgi.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 09:16:30 +00:00
return atomic_notifier_chain_register(&netlink_chain, nb);
}
EXPORT_SYMBOL(netlink_register_notifier);
int netlink_unregister_notifier(struct notifier_block *nb)
{
[PATCH] Notifier chain update: API changes The kernel's implementation of notifier chains is unsafe. There is no protection against entries being added to or removed from a chain while the chain is in use. The issues were discussed in this thread: http://marc.theaimsgroup.com/?l=linux-kernel&m=113018709002036&w=2 We noticed that notifier chains in the kernel fall into two basic usage classes: "Blocking" chains are always called from a process context and the callout routines are allowed to sleep; "Atomic" chains can be called from an atomic context and the callout routines are not allowed to sleep. We decided to codify this distinction and make it part of the API. Therefore this set of patches introduces three new, parallel APIs: one for blocking notifiers, one for atomic notifiers, and one for "raw" notifiers (which is really just the old API under a new name). New kinds of data structures are used for the heads of the chains, and new routines are defined for registration, unregistration, and calling a chain. The three APIs are explained in include/linux/notifier.h and their implementation is in kernel/sys.c. With atomic and blocking chains, the implementation guarantees that the chain links will not be corrupted and that chain callers will not get messed up by entries being added or removed. For raw chains the implementation provides no guarantees at all; users of this API must provide their own protections. (The idea was that situations may come up where the assumptions of the atomic and blocking APIs are not appropriate, so it should be possible for users to handle these things in their own way.) There are some limitations, which should not be too hard to live with. For atomic/blocking chains, registration and unregistration must always be done in a process context since the chain is protected by a mutex/rwsem. Also, a callout routine for a non-raw chain must not try to register or unregister entries on its own chain. (This did happen in a couple of places and the code had to be changed to avoid it.) Since atomic chains may be called from within an NMI handler, they cannot use spinlocks for synchronization. Instead we use RCU. The overhead falls almost entirely in the unregister routine, which is okay since unregistration is much less frequent that calling a chain. Here is the list of chains that we adjusted and their classifications. None of them use the raw API, so for the moment it is only a placeholder. ATOMIC CHAINS ------------- arch/i386/kernel/traps.c: i386die_chain arch/ia64/kernel/traps.c: ia64die_chain arch/powerpc/kernel/traps.c: powerpc_die_chain arch/sparc64/kernel/traps.c: sparc64die_chain arch/x86_64/kernel/traps.c: die_chain drivers/char/ipmi/ipmi_si_intf.c: xaction_notifier_list kernel/panic.c: panic_notifier_list kernel/profile.c: task_free_notifier net/bluetooth/hci_core.c: hci_notifier net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_chain net/ipv4/netfilter/ip_conntrack_core.c: ip_conntrack_expect_chain net/ipv6/addrconf.c: inet6addr_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_chain net/netfilter/nf_conntrack_core.c: nf_conntrack_expect_chain net/netlink/af_netlink.c: netlink_chain BLOCKING CHAINS --------------- arch/powerpc/platforms/pseries/reconfig.c: pSeries_reconfig_chain arch/s390/kernel/process.c: idle_chain arch/x86_64/kernel/process.c idle_notifier drivers/base/memory.c: memory_chain drivers/cpufreq/cpufreq.c cpufreq_policy_notifier_list drivers/cpufreq/cpufreq.c cpufreq_transition_notifier_list drivers/macintosh/adb.c: adb_client_list drivers/macintosh/via-pmu.c sleep_notifier_list drivers/macintosh/via-pmu68k.c sleep_notifier_list drivers/macintosh/windfarm_core.c wf_client_list drivers/usb/core/notify.c usb_notifier_list drivers/video/fbmem.c fb_notifier_list kernel/cpu.c cpu_chain kernel/module.c module_notify_list kernel/profile.c munmap_notifier kernel/profile.c task_exit_notifier kernel/sys.c reboot_notifier_list net/core/dev.c netdev_chain net/decnet/dn_dev.c: dnaddr_chain net/ipv4/devinet.c: inetaddr_chain It's possible that some of these classifications are wrong. If they are, please let us know or submit a patch to fix them. Note that any chain that gets called very frequently should be atomic, because the rwsem read-locking used for blocking chains is very likely to incur cache misses on SMP systems. (However, if the chain's callout routines may sleep then the chain cannot be atomic.) The patch set was written by Alan Stern and Chandra Seetharaman, incorporating material written by Keith Owens and suggestions from Paul McKenney and Andrew Morton. [jes@sgi.com: restructure the notifier chain initialization macros] Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com> Signed-off-by: Jes Sorensen <jes@sgi.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-03-27 09:16:30 +00:00
return atomic_notifier_chain_unregister(&netlink_chain, nb);
}
EXPORT_SYMBOL(netlink_unregister_notifier);
static const struct proto_ops netlink_ops = {
.family = PF_NETLINK,
.owner = THIS_MODULE,
.release = netlink_release,
.bind = netlink_bind,
.connect = netlink_connect,
.socketpair = sock_no_socketpair,
.accept = sock_no_accept,
.getname = netlink_getname,
.poll = datagram_poll,
.ioctl = sock_no_ioctl,
.listen = sock_no_listen,
.shutdown = sock_no_shutdown,
.setsockopt = netlink_setsockopt,
.getsockopt = netlink_getsockopt,
.sendmsg = netlink_sendmsg,
.recvmsg = netlink_recvmsg,
.mmap = sock_no_mmap,
.sendpage = sock_no_sendpage,
};
static const struct net_proto_family netlink_family_ops = {
.family = PF_NETLINK,
.create = netlink_create,
.owner = THIS_MODULE, /* for consistency 8) */
};
static int __net_init netlink_net_init(struct net *net)
{
#ifdef CONFIG_PROC_FS
if (!proc_net_fops_create(net, "netlink", 0, &netlink_seq_fops))
return -ENOMEM;
#endif
return 0;
}
static void __net_exit netlink_net_exit(struct net *net)
{
#ifdef CONFIG_PROC_FS
proc_net_remove(net, "netlink");
#endif
}
static void __init netlink_add_usersock_entry(void)
{
struct listeners *listeners;
int groups = 32;
listeners = kzalloc(sizeof(*listeners) + NLGRPSZ(groups), GFP_KERNEL);
if (!listeners)
panic("netlink_add_usersock_entry: Cannot allocate listeners\n");
netlink_table_grab();
nl_table[NETLINK_USERSOCK].groups = groups;
rcu_assign_pointer(nl_table[NETLINK_USERSOCK].listeners, listeners);
nl_table[NETLINK_USERSOCK].module = THIS_MODULE;
nl_table[NETLINK_USERSOCK].registered = 1;
netlink_table_ungrab();
}
static struct pernet_operations __net_initdata netlink_net_ops = {
.init = netlink_net_init,
.exit = netlink_net_exit,
};
static int __init netlink_proto_init(void)
{
struct sk_buff *dummy_skb;
int i;
unsigned long limit;
unsigned int order;
int err = proto_register(&netlink_proto, 0);
if (err != 0)
goto out;
BUILD_BUG_ON(sizeof(struct netlink_skb_parms) > sizeof(dummy_skb->cb));
nl_table = kcalloc(MAX_LINKS, sizeof(*nl_table), GFP_KERNEL);
if (!nl_table)
goto panic;
if (totalram_pages >= (128 * 1024))
limit = totalram_pages >> (21 - PAGE_SHIFT);
else
limit = totalram_pages >> (23 - PAGE_SHIFT);
order = get_bitmask_order(limit) - 1 + PAGE_SHIFT;
limit = (1UL << order) / sizeof(struct hlist_head);
order = get_bitmask_order(min(limit, (unsigned long)UINT_MAX)) - 1;
for (i = 0; i < MAX_LINKS; i++) {
struct nl_pid_hash *hash = &nl_table[i].hash;
hash->table = nl_pid_hash_zalloc(1 * sizeof(*hash->table));
if (!hash->table) {
while (i-- > 0)
nl_pid_hash_free(nl_table[i].hash.table,
1 * sizeof(*hash->table));
kfree(nl_table);
goto panic;
}
hash->max_shift = order;
hash->shift = 0;
hash->mask = 0;
hash->rehash_time = jiffies;
}
netlink_add_usersock_entry();
sock_register(&netlink_family_ops);
register_pernet_subsys(&netlink_net_ops);
/* The netlink device handler may be needed early. */
rtnetlink_init();
out:
return err;
panic:
panic("netlink_init: Cannot allocate nl_table\n");
}
core_initcall(netlink_proto_init);