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>
Keep a bitmask of multicast groups with subscribed listeners to let
netlink users check for listeners before generating multicast
messages.
Queries don't perform any locking, which may result in false
positives, it is guaranteed however that any new subscriptions are
visible before bind() or setsockopt() return.
Signed-off-by: Patrick McHardy <kaber@trash.net>
ACKed-by: Jamal Hadi Salim<hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>
The skb given to netlink_cmsg_recv_pktinfo is already freed, move it up
a few lines.
Coverity #948
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
netlink overrun was broken while improvement of netlink.
Destination socket is used in the place where it was meant to be source socket,
so that now overrun is never sent to user netlink sockets, when it should be,
and it even can be set on kernel socket, which results in complete deadlock
of rtnetlink.
Suggested fix is to restore status quo passing source socket as additional
argument to netlink_attachskb().
A little explanation: overrun is set on a socket, when it failed
to receive some message and sender of this messages does not or even
have no way to handle this error. This happens in two cases:
1. when kernel sends something. Kernel never retransmits and cannot
wait for buffer space.
2. when user sends a broadcast and the message was not delivered
to some recipients.
Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: Use <linux/capability.h> where capable() is used.
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
From: Martin Murray <murrayma@citi.umich.edu>
Sanity check nlmsg_len during netlink_rcv_skb. An nlmsg_len == 0 can
cause infinite loop in kernel, effectively DoSing machine. Noted by
Matin Murray.
Signed-off-by: Chris Wright <chrisw@sous-sol.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
I noticed that some of 'struct proto_ops' used in the kernel may share
a cache line used by locks or other heavily modified data. (default
linker alignement is 32 bytes, and L1_CACHE_LINE is 64 or 128 at
least)
This patch makes sure a 'struct proto_ops' can be declared as const,
so that all cpus can share all parts of it without false sharing.
This is not mandatory : a driver can still use a read/write structure
if it needs to (and eventually a __read_mostly)
I made a global stubstitute to change all existing occurences to make
them const.
This should reduce the possibility of false sharing on SMP, and
speedup some socket system calls.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduces netlink_run_queue() to handle the receive queue of
a netlink socket in a generic way. Processes as much as there
was in the queue upon entry and invokes a callback function
for each netlink message found. The callback function may
refuse a message by returning a negative error code but setting
the error pointer to 0 in which case netlink_run_queue() will
return with a qlen != 0.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Most netlink families make no use of the done() callback, making
it optional gets rid of all unnecessary dummy implementations.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove the variable nlk & call to nlk_sk as it does not have any side effect.
Signed-off-by: Jayachandran C. <c.jayachandran at gmail.com>
Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
- added typedef unsigned int __nocast gfp_t;
- replaced __nocast uses for gfp flags with gfp_t - it gives exactly
the same warnings as far as sparse is concerned, doesn't change
generated code (from gcc point of view we replaced unsigned int with
typedef) and documents what's going on far better.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
This broke the pam audit module which includes an incorrect check for
-ENOENT instead of -EPROTONOTSUPP.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
NETLINK_ADD_MEMBERSHIP/NETLINK_DROP_MEMBERSHIP are used to join/leave
groups, NETLINK_PKTINFO is used to enable nl_pktinfo control messages
for received packets to get the extended destination group number.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is necessary for dynamic number of netlink groups to make sure we know
the number of possible groups before bind() is called. With this change pure
userspace communication using unused netlink protocols becomes impossible.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Using the group number allows increasing the number of groups without
beeing limited by the size of the bitmask. It introduces one limitation
for netlink users: messages can't be broadcasted to multiple groups anymore,
however this feature was never used inside the kernel.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use-after-free: the struct proto_ops containing the module pointer
is freed when a socket with pid=0 is released, which besides for kernel
sockets is true for all unbound sockets.
Module refcount leak: when the kernel socket is closed before all user
sockets have been closed the proto_ops struct for this family is
replaced by the generic one and the module refcount can't be dropped.
The second problem can't be solved cleanly using module refcounting in the
generic socket code, so this patch adds explicit refcounting to
netlink_create/netlink_release.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
- Remove bogus code for compiling netlink as module
- Add module refcounting support for modules implementing a netlink
protocol
- Add support for autoloading modules that implement a netlink protocol
as soon as someone opens a socket for that protocol
Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
From: Victor Fusco <victor@cetuc.puc-rio.br>
Fix the sparse warning "implicit cast to nocast type"
Signed-off-by: Victor Fusco <victor@cetuc.puc-rio.br>
Signed-off-by: Domen Puncer <domen@coderock.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This is part of the grand scheme to eliminate the qlen
member of skb_queue_head, and subsequently remove the
'list' member of sk_buff.
Most users of skb_queue_len() want to know if the queue is
empty or not, and that's trivially done with skb_queue_empty()
which doesn't use the skb_queue_head->qlen member and instead
uses the queue list emptyness as the test.
Signed-off-by: David S. Miller <davem@davemloft.net>
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>
Introduces a new macro NLMSG_NEW which extends NLMSG_PUT but takes
a flags argument. NLMSG_PUT stays there for compatibility but now
calls NLMSG_NEW with flags == 0. NLMSG_PUT_ANSWER is renamed to
NLMSG_NEW_ANSWER which now also takes a flags argument.
Also converts the users of NLMSG_PUT_ANSWER to use NLMSG_NEW_ANSWER
and fixes the two direct users of __nlmsg_put to either provide
the flags or use NLMSG_NEW(_ANSWER).
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
In netlink_broadcast() we're sending shared skb's to netlink listeners
when possible (saves some copying). This is OK, since we hold the only
other reference to the skb.
However, this implies that we must drop our reference on the skb, before
allowing a receiving socket to disappear. Otherwise, the socket buffer
accounting is disrupted.
Signed-off-by: Tommy S. Christensen <tommy.christensen@tpack.net>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cloned packets don't need the orphan call.
Signed-off-by: Tommy S. Christensen <tommy.christensen@tpack.net>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
This bug causes:
assertion (!atomic_read(&sk->sk_rmem_alloc)) failed at net/netlink/af_netlink.c (122)
What's happening is that:
1) The skb is sent to socket 1.
2) Someone does a recvmsg on socket 1 and drops the ref on the skb.
Note that the rmalloc is not returned at this point since the
skb is still referenced.
3) The same skb is now sent to socket 2.
This version of the fix resurrects the skb_orphan call that was moved
out, last time we had 'shared-skb troubles'. It is practically a no-op
in the common case, but still prevents the possible race with recvmsg.
Signed-off-by: Tommy S. Christensen <tommy.christensen@tpack.net>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Here is a little optimisation for the cb_lock used by netlink_dump.
While fixing that race earlier, I noticed that the reference count
held by cb_lock is completely useless. The reason is that in order
to obtain the protection of the reference count, you have to take
the cb_lock. But the only way to take the cb_lock is through
dereferencing the socket.
That is, you must already possess a reference count on the socket
before you can take advantage of the reference count held by cb_lock.
As a corollary, we can remve the reference count held by the cb_lock.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
scumbags!
net/netlink/af_netlink.c: In function `netlink_sendmsg':
net/netlink/af_netlink.c:908: warning: implicit declaration of function `audit_get_loginuid'
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Most audit control messages are sent over netlink.In order to properly
log the identity of the sender of audit control messages, we would like
to add the loginuid to the netlink_creds structure, as per the attached
patch.
Signed-off-by: Serge Hallyn <serue@us.ibm.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
A lot of places in there are including major.h for no reason whatsoever.
Removed. And yes, it still builds.
The history of that stuff is often amusing. E.g. for net/core/sock.c
the story looks so, as far as I've been able to reconstruct it: we used
to need major.h in net/socket.c circa 1.1.early. In 1.1.13 that need
had disappeared, along with register_chrdev(SOCKET_MAJOR, "socket",
&net_fops) in sock_init(). Include had not. When 1.2 -> 1.3 reorg of
net/* had moved a lot of stuff from net/socket.c to net/core/sock.c,
this crap had followed...
Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Initial git repository build. I'm not bothering with the full history,
even though we have it. We can create a separate "historical" git
archive of that later if we want to, and in the meantime it's about
3.2GB when imported into git - space that would just make the early
git days unnecessarily complicated, when we don't have a lot of good
infrastructure for it.
Let it rip!
Alan Stern