linux-stable.git/net/packet, branch v3.2.87

packet: Do not call fanout_release from atomic contexts

2017-03-16T02:18:51+00:00

commit 2bd624b4611ffee36422782d16e1c944d1351e98 upstream.

Commit 6664498280cf ("packet: call fanout_release, while UNREGISTERING a
netdev"), unfortunately, introduced the following issues.

1. calling mutex_lock(&fanout_mutex) (fanout_release()) from inside
rcu_read-side critical section. rcu_read_lock disables preemption, most often,
which prohibits calling sleeping functions.

[  ] include/linux/rcupdate.h:560 Illegal context switch in RCU read-side critical section!
[  ]
[  ] rcu_scheduler_active = 1, debug_locks = 0
[  ] 4 locks held by ovs-vswitchd/1969:
[  ]  #0:  (cb_lock){++++++}, at: [] genl_rcv+0x19/0x40
[  ]  #1:  (ovs_mutex){+.+.+.}, at: [] ovs_vport_cmd_del+0x4a/0x100 [openvswitch]
[  ]  #2:  (rtnl_mutex){+.+.+.}, at: [] rtnl_lock+0x17/0x20
[  ]  #3:  (rcu_read_lock){......}, at: [] packet_notifier+0x5/0x3f0
[  ]
[  ] Call Trace:
[  ]  [] dump_stack+0x85/0xc4
[  ]  [] lockdep_rcu_suspicious+0x107/0x110
[  ]  [] ___might_sleep+0x57/0x210
[  ]  [] __might_sleep+0x70/0x90
[  ]  [] mutex_lock_nested+0x3c/0x3a0
[  ]  [] ? vprintk_default+0x1f/0x30
[  ]  [] ? printk+0x4d/0x4f
[  ]  [] fanout_release+0x1d/0xe0
[  ]  [] packet_notifier+0x2f9/0x3f0

2. calling mutex_lock(&fanout_mutex) inside spin_lock(&po->bind_lock).
"sleeping function called from invalid context"

[  ] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:620
[  ] in_atomic(): 1, irqs_disabled(): 0, pid: 1969, name: ovs-vswitchd
[  ] INFO: lockdep is turned off.
[  ] Call Trace:
[  ]  [] dump_stack+0x85/0xc4
[  ]  [] ___might_sleep+0x202/0x210
[  ]  [] __might_sleep+0x70/0x90
[  ]  [] mutex_lock_nested+0x3c/0x3a0
[  ]  [] fanout_release+0x1d/0xe0
[  ]  [] packet_notifier+0x2f9/0x3f0

3. calling dev_remove_pack(&fanout->prot_hook), from inside
spin_lock(&po->bind_lock) or rcu_read-side critical-section. dev_remove_pack()
-> synchronize_net(), which might sleep.

[  ] BUG: scheduling while atomic: ovs-vswitchd/1969/0x00000002
[  ] INFO: lockdep is turned off.
[  ] Call Trace:
[  ]  [] dump_stack+0x85/0xc4
[  ]  [] __schedule_bug+0x64/0x73
[  ]  [] __schedule+0x6b/0xd10
[  ]  [] schedule+0x6b/0x80
[  ]  [] schedule_timeout+0x38d/0x410
[  ]  [] synchronize_sched_expedited+0x53d/0x810
[  ]  [] synchronize_rcu_expedited+0xe/0x10
[  ]  [] synchronize_net+0x35/0x50
[  ]  [] dev_remove_pack+0x13/0x20
[  ]  [] fanout_release+0xbe/0xe0
[  ]  [] packet_notifier+0x2f9/0x3f0

4. fanout_release() races with calls from different CPU.

To fix the above problems, remove the call to fanout_release() under
rcu_read_lock(). Instead, call __dev_remove_pack(&fanout->prot_hook) and
netdev_run_todo will be happy that &dev->ptype_specific list is empty. In order
to achieve this, I moved dev_{add,remove}_pack() out of fanout_{add,release} to
__fanout_{link,unlink}. So, call to {,__}unregister_prot_hook() will make sure
fanout->prot_hook is removed as well.

Fixes: 6664498280cf ("packet: call fanout_release, while UNREGISTERING a netdev")
Reported-by: Eric Dumazet 
Signed-off-by: Anoob Soman 
Acked-by: Eric Dumazet 
Signed-off-by: David S. Miller 
[bwh: Backported to 3.2:
 - Don't call fanout_release_data()
 - Adjust context]
Signed-off-by: Ben Hutchings

packet: call fanout_release, while UNREGISTERING a netdev

2017-03-16T02:18:50+00:00

commit 6664498280cf17a59c3e7cf1a931444c02633ed1 upstream.

If a socket has FANOUT sockopt set, a new proto_hook is registered
as part of fanout_add(). When processing a NETDEV_UNREGISTER event in
af_packet, __fanout_unlink is called for all sockets, but prot_hook which was
registered as part of fanout_add is not removed. Call fanout_release, on a
NETDEV_UNREGISTER, which removes prot_hook and removes fanout from the
fanout_list.

This fixes BUG_ON(!list_empty(&dev->ptype_specific)) in netdev_run_todo()

Signed-off-by: Anoob Soman 
Signed-off-by: David S. Miller 
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings

packet: fix races in fanout_add()

2017-03-16T02:18:50+00:00

commit d199fab63c11998a602205f7ee7ff7c05c97164b upstream.

Multiple threads can call fanout_add() at the same time.

We need to grab fanout_mutex earlier to avoid races that could
lead to one thread freeing po->rollover that was set by another thread.

Do the same in fanout_release(), for peace of mind, and to help us
finding lockdep issues earlier.

Fixes: dc99f600698d ("packet: Add fanout support.")
Fixes: 0648ab70afe6 ("packet: rollover prepare: per-socket state")
Signed-off-by: Eric Dumazet 
Cc: Willem de Bruijn 
Signed-off-by: David S. Miller 
[bwh: Backported to 3.2:
 - No rollover queue stats
 - Adjust context]
Signed-off-by: Ben Hutchings

packet: fix race condition in packet_set_ring

2017-02-23T03:51:01+00:00

commit 84ac7260236a49c79eede91617700174c2c19b0c upstream.

When packet_set_ring creates a ring buffer it will initialize a
struct timer_list if the packet version is TPACKET_V3. This value
can then be raced by a different thread calling setsockopt to
set the version to TPACKET_V1 before packet_set_ring has finished.

This leads to a use-after-free on a function pointer in the
struct timer_list when the socket is closed as the previously
initialized timer will not be deleted.

The bug is fixed by taking lock_sock(sk) in packet_setsockopt when
changing the packet version while also taking the lock at the start
of packet_set_ring.

Fixes: f6fb8f100b80 ("af-packet: TPACKET_V3 flexible buffer implementation.")
Signed-off-by: Philip Pettersson 
Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings

packet: fix match_fanout_group()

2015-11-27T12:48:20+00:00

commit 161642e24fee40fba2c5bc2ceacc00d118a22d65 upstream.

Recent TCP listener patches exposed a prior af_packet bug :
match_fanout_group() blindly assumes it is always safe
to cast sk to a packet socket to compare fanout with af_packet_priv

But SYNACK packets can be sent while attached to request_sock, which
are smaller than a "struct sock".

We can read non existent memory and crash.

Fixes: c0de08d04215 ("af_packet: don't emit packet on orig fanout group")
Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener")
Signed-off-by: Eric Dumazet 
Cc: Willem de Bruijn 
Cc: Eric Leblond 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings

packet: avoid out of bounds read in round robin fanout

2015-08-06T23:32:17+00:00

[ Upstream commit 468479e6043c84f5a65299cc07cb08a22a28c2b1 ]

PACKET_FANOUT_LB computes f->rr_cur such that it is modulo
f->num_members. It returns the old value unconditionally, but
f->num_members may have changed since the last store. Ensure
that the return value is always < num.

When modifying the logic, simplify it further by replacing the loop
with an unconditional atomic increment.

Fixes: dc99f600698d ("packet: Add fanout support.")
Suggested-by: Eric Dumazet 
Signed-off-by: Willem de Bruijn 
Acked-by: Eric Dumazet 
Signed-off-by: David S. Miller 
[bwh: Backported to 3.2:
 - Adjust context
 - Demux functions return a sock pointer, not an index]
Signed-off-by: Ben Hutchings

packet: read num_members once in packet_rcv_fanout()

2015-08-06T23:32:17+00:00

[ Upstream commit f98f4514d07871da7a113dd9e3e330743fd70ae4 ]

We need to tell compiler it must not read f->num_members multiple
times. Otherwise testing if num is not zero is flaky, and we could
attempt an invalid divide by 0 in fanout_demux_cpu()

Note bug was present in packet_rcv_fanout_hash() and
packet_rcv_fanout_lb() but final 3.1 had a simple location
after commit 95ec3eb417115fb ("packet: Add 'cpu' fanout policy.")

Fixes: dc99f600698dc ("packet: Add fanout support.")
Signed-off-by: Eric Dumazet 
Cc: Willem de Bruijn 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings

af_packet: block BH in prb_shutdown_retire_blk_timer()

2014-01-03T04:33:34+00:00

[ Upstream commit ec6f809ff6f19fafba3212f6aff0dda71dfac8e8 ]

Currently we're using plain spin_lock() in prb_shutdown_retire_blk_timer(),
however the timer might fire right in the middle and thus try to re-aquire
the same spinlock, leaving us in a endless loop.

To fix that, use the spin_lock_bh() to block it.

Fixes: f6fb8f100b80 ("af-packet: TPACKET_V3 flexible buffer implementation.")
CC: "David S. Miller" 
CC: Daniel Borkmann 
CC: Willem de Bruijn 
CC: Phil Sutter 
CC: Eric Dumazet 
Reported-by: Jan Stancek 
Tested-by: Jan Stancek 
Signed-off-by: Veaceslav Falico 
Acked-by: Daniel Borkmann 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings

packet: fix use after free race in send path when dev is released

2014-01-03T04:33:34+00:00

[ Upstream commit e40526cb20b5ee53419452e1f03d97092f144418 ]

Salam reported a use after free bug in PF_PACKET that occurs when
we're sending out frames on a socket bound device and suddenly the
net device is being unregistered. It appears that commit 827d9780
introduced a possible race condition between {t,}packet_snd() and
packet_notifier(). In the case of a bound socket, packet_notifier()
can drop the last reference to the net_device and {t,}packet_snd()
might end up suddenly sending a packet over a freed net_device.

To avoid reverting 827d9780 and thus introducing a performance
regression compared to the current state of things, we decided to
hold a cached RCU protected pointer to the net device and maintain
it on write side via bind spin_lock protected register_prot_hook()
and __unregister_prot_hook() calls.

In {t,}packet_snd() path, we access this pointer under rcu_read_lock
through packet_cached_dev_get() that holds reference to the device
to prevent it from being freed through packet_notifier() while
we're in send path. This is okay to do as dev_put()/dev_hold() are
per-cpu counters, so this should not be a performance issue. Also,
the code simplifies a bit as we don't need need_rls_dev anymore.

Fixes: 827d978037d7 ("af-packet: Use existing netdev reference for bound sockets.")
Reported-by: Salam Noureddine 
Signed-off-by: Daniel Borkmann 
Signed-off-by: Salam Noureddine 
Cc: Ben Greear 
Cc: Eric Dumazet 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings

net: rework recvmsg handler msg_name and msg_namelen logic

2014-01-03T04:33:33+00:00

[ Upstream commit f3d3342602f8bcbf37d7c46641cb9bca7618eb1c ]

This patch now always passes msg->msg_namelen as 0. recvmsg handlers must
set msg_namelen to the proper size <= sizeof(struct sockaddr_storage)
to return msg_name to the user.

This prevents numerous uninitialized memory leaks we had in the
recvmsg handlers and makes it harder for new code to accidentally leak
uninitialized memory.

Optimize for the case recvfrom is called with NULL as address. We don't
need to copy the address at all, so set it to NULL before invoking the
recvmsg handler. We can do so, because all the recvmsg handlers must
cope with the case a plain read() is called on them. read() also sets
msg_name to NULL.

Also document these changes in include/linux/net.h as suggested by David
Miller.

Changes since RFC:

Set msg->msg_name = NULL if user specified a NULL in msg_name but had a
non-null msg_namelen in verify_iovec/verify_compat_iovec. This doesn't
affect sendto as it would bail out earlier while trying to copy-in the
address. It also more naturally reflects the logic by the callers of
verify_iovec.

With this change in place I could remove "
if (!uaddr || msg_sys->msg_namelen == 0)
	msg->msg_name = NULL
".

This change does not alter the user visible error logic as we ignore
msg_namelen as long as msg_name is NULL.

Also remove two unnecessary curly brackets in ___sys_recvmsg and change
comments to netdev style.

Cc: David Miller 
Suggested-by: Eric Dumazet 
Signed-off-by: Hannes Frederic Sowa 
Signed-off-by: David S. Miller 
Signed-off-by: Ben Hutchings