__netpoll_rcu_free is used to free netpoll structures when the rtnl_lock is
already held. The mechanism is used to asynchronously call __netpoll_cleanup
outside of the holding of the rtnl_lock, so as to avoid deadlock.
Unfortunately, __netpoll_cleanup modifies pointers (dev->np), which means the
rtnl_lock must be held while calling it. Further, it cannot be held, because
rcu callbacks may be issued in softirq contexts, which cannot sleep.
Fix this by converting the rcu callback to a work queue that is guaranteed to
get scheduled in process context, so that we can hold the rtnl properly while
calling __netpoll_cleanup
Tested successfully by myself.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: "David S. Miller" <davem@davemloft.net>
CC: Cong Wang <amwang@redhat.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Create skb_panic() function in lieu of both skb_over_panic() and
skb_under_panic() so that code duplication would be avoided. Update type
and variable name where necessary.
Jiri Pirko suggested using wrappers so that we would be able to keep the
fruits of the original code.
Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Cc: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
Sparse spotted local function that could be static.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes sparse complaints about dropping __user in casts.
warning: cast removes address space of expression
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
And remove no longer used br->flags.
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cc: Erik Hugne <erik.hugne@ericsson.com>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
v2:
a) moved before multicast source address check
b) changed comment to netdev style
Cc: Erik Hugne <erik.hugne@ericsson.com>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reported-by: Erik Hugne <erik.hugne@ericsson.com>
Cc: Erik Hugne <erik.hugne@ericsson.com>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
To allow both of protocol-specific data and device-specific data
attached with neighbour entry, and to eliminate size calculation
cost when allocating entry, sizeof protocol-speicic data must be
multiple of NEIGH_PRIV_ALIGN. On 64bit archs,
sizeof(struct dn_neigh) is multiple of NEIGH_PRIV_ALIGN, but on
32bit archs, it was not.
Introduce NEIGH_ENTRY_SPACE() macro to ensure that protocol-specific
entry-size meets our requirement.
Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
RFC4291 (IPv6 addressing architecture) says that interface-Local scope
spans only a single interface on a node. We should not join L2 device
multicast list for addresses in interface-local (or smaller) scope.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Initial implementation of the Multiple VLAN Registration Protocol
(MVRP) from IEEE 802.1Q-2011, based on the existing implementation
of the GARP VLAN Registration Protocol (GVRP).
Signed-off-by: David Ward <david.ward@ll.mit.edu>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Initial implementation of the Multiple Registration Protocol (MRP)
from IEEE 802.1Q-2011, based on the existing implementation of the
Generic Attribute Registration Protocol (GARP).
Signed-off-by: David Ward <david.ward@ll.mit.edu>
Acked-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
VM Sockets allows communication between virtual machines and the hypervisor.
User level applications both in a virtual machine and on the host can use the
VM Sockets API, which facilitates fast and efficient communication between
guest virtual machines and their host. A socket address family, designed to be
compatible with UDP and TCP at the interface level, is provided.
Today, VM Sockets is used by various VMware Tools components inside the guest
for zero-config, network-less access to VMware host services. In addition to
this, VMware's users are using VM Sockets for various applications, where
network access of the virtual machine is restricted or non-existent. Examples
of this are VMs communicating with device proxies for proprietary hardware
running as host applications and automated testing of applications running
within virtual machines.
The VMware VM Sockets are similar to other socket types, like Berkeley UNIX
socket interface. The VM Sockets module supports both connection-oriented
stream sockets like TCP, and connectionless datagram sockets like UDP. The VM
Sockets protocol family is defined as "AF_VSOCK" and the socket operations
split for SOCK_DGRAM and SOCK_STREAM.
For additional information about the use of VM Sockets, please refer to the
VM Sockets Programming Guide available at:
https://www.vmware.com/support/developer/vmci-sdk/
Signed-off-by: George Zhang <georgezhang@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Andy king <acking@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Synchronize with 'net' in order to sort out some l2tp, wireless, and
ipv6 GRE fixes that will be built on top of in 'net-next'.
Signed-off-by: David S. Miller <davem@davemloft.net>
In sctp_auth_make_key_vector, we allocate a temporary sctp_auth_bytes
structure with kmalloc instead of the sctp_auth_create_key allocator.
Change this to sctp_auth_create_key as it is the case everywhere else,
so that we also can properly free it via sctp_auth_key_put. This makes
it easier for future code changes in the structure and allocator itself,
since a single API is consistently used for this purpose. Also, by
using sctp_auth_create_key we're doing sanity checks over the arguments.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes the following RCU warning:
[ 51.680236] ===============================
[ 51.681914] [ INFO: suspicious RCU usage. ]
[ 51.683610] 3.8.0-rc6-next-20130206-sasha-00028-g83214f7-dirty #276 Tainted: G W
[ 51.686703] -------------------------------
[ 51.688281] net/ipv6/ip6_flowlabel.c:671 suspicious rcu_dereference_check() usage!
we should use rcu_dereference_bh() when we hold rcu_read_lock_bh().
Reported-by: Sasha Levin <sasha.levin@oracle.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to address the fact that some devices cannot support the full 32K
frag size we need to have the value accessible somewhere so that we can use it
to do comparisons against what the device can support. As such I am moving
the values out of skbuff.c and into skbuff.h.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking fixes from David Miller:
1) Revert iwlwifi reclaimed packet tracking, it causes problems for a
bunch of folks. From Emmanuel Grumbach.
2) Work limiting code in brcmsmac wifi driver can clear tx status
without processing the event. From Arend van Spriel.
3) rtlwifi USB driver processes wrong SKB, fix from Larry Finger.
4) l2tp tunnel delete can race with close, fix from Tom Parkin.
5) pktgen_add_device() failures are not checked at all, fix from Cong
Wang.
6) Fix unintentional removal of carrier off from tun_detach(),
otherwise we confuse userspace, from Michael S. Tsirkin.
7) Don't leak socket reference counts and ubufs in vhost-net driver,
from Jason Wang.
8) vmxnet3 driver gets it's initial carrier state wrong, fix from Neil
Horman.
9) Protect against USB networking devices which spam the host with 0
length frames, from Bjørn Mork.
10) Prevent neighbour overflows in ipv6 for locally destined routes,
from Marcelo Ricardo. This is the best short-term fix for this, a
longer term fix has been implemented in net-next.
11) L2TP uses ipv4 datagram routines in it's ipv6 code, whoops. This
mistake is largely because the ipv6 functions don't even have some
kind of prefix in their names to suggest they are ipv6 specific.
From Tom Parkin.
12) Check SYN packet drops properly in tcp_rcv_fastopen_synack(), from
Yuchung Cheng.
13) Fix races and TX skb freeing bugs in via-rhine's NAPI support, from
Francois Romieu and your's truly.
14) Fix infinite loops and divides by zero in TCP congestion window
handling, from Eric Dumazet, Neal Cardwell, and Ilpo Järvinen.
15) AF_PACKET tx ring handling can leak kernel memory to userspace, fix
from Phil Sutter.
16) Fix error handling in ipv6 GRE tunnel transmit, from Tommi Rantala.
17) Protect XEN netback driver against hostile frontend putting garbage
into the rings, don't leak pages in TX GOP checking, and add proper
resource releasing in error path of xen_netbk_get_requests(). From
Ian Campbell.
18) SCTP authentication keys should be cleared out and released with
kzfree(), from Daniel Borkmann.
19) L2TP is a bit too clever trying to maintain skb->truesize, and ends
up corrupting socket memory accounting to the point where packet
sending is halted indefinitely. Just remove the adjustments
entirely, they aren't really needed. From Eric Dumazet.
20) ATM Iphase driver uses a data type with the same name as the S390
headers, rename to fix the build. From Heiko Carstens.
21) Fix a typo in copying the inner network header offset from one SKB
to another, from Pravin B Shelar.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (56 commits)
net: sctp: sctp_endpoint_free: zero out secret key data
net: sctp: sctp_setsockopt_auth_key: use kzfree instead of kfree
atm/iphase: rename fregt_t -> ffreg_t
net: usb: fix regression from FLAG_NOARP code
l2tp: dont play with skb->truesize
net: sctp: sctp_auth_key_put: use kzfree instead of kfree
netback: correct netbk_tx_err to handle wrap around.
xen/netback: free already allocated memory on failure in xen_netbk_get_requests
xen/netback: don't leak pages on failure in xen_netbk_tx_check_gop.
xen/netback: shutdown the ring if it contains garbage.
net: qmi_wwan: add more Huawei devices, including E320
net: cdc_ncm: add another Huawei vendor specific device
ipv6/ip6_gre: fix error case handling in ip6gre_tunnel_xmit()
tcp: fix for zero packets_in_flight was too broad
brcmsmac: rework of mac80211 .flush() callback operation
ssb: unregister gpios before unloading ssb
bcma: unregister gpios before unloading bcma
rtlwifi: Fix scheduling while atomic bug
net: usbnet: fix tx_dropped statistics
tcp: ipv6: Update MIB counters for drops
...
On sctp_endpoint_destroy, previously used sensitive keying material
should be zeroed out before the memory is returned, as we already do
with e.g. auth keys when released.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In sctp_setsockopt_auth_key, we create a temporary copy of the user
passed shared auth key for the endpoint or association and after
internal setup, we free it right away. Since it's sensitive data, we
should zero out the key before returning the memory back to the
allocator. Thus, use kzfree instead of kfree, just as we do in
sctp_auth_key_put().
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Andrew Savchenko reported a DNS failure and we diagnosed that
some UDP sockets were unable to send more packets because their
sk_wmem_alloc was corrupted after a while (tx_queue column in
following trace)
$ cat /proc/net/udp
sl local_address rem_address st tx_queue rx_queue tr tm->when retrnsmt uid timeout inode ref pointer drops
...
459: 00000000:0270 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 4507 2 ffff88003d612380 0
466: 00000000:0277 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 4802 2 ffff88003d613180 0
470: 076A070A:007B 00000000:0000 07 FFFF4600:00000000 00:00000000 00000000 123 0 5552 2 ffff880039974380 0
470: 010213AC:007B 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 4986 2 ffff88003dbd3180 0
470: 010013AC:007B 00000000:0000 07 00000000:00000000 00:00000000 00000000 0 0 4985 2 ffff88003dbd2e00 0
470: 00FCA8C0:007B 00000000:0000 07 FFFFFB00:00000000 00:00000000 00000000 0 0 4984 2 ffff88003dbd2a80 0
...
Playing with skb->truesize is tricky, especially when
skb is attached to a socket, as we can fool memory charging.
Just remove this code, its not worth trying to be ultra
precise in xmit path.
Reported-by: Andrew Savchenko <bircoph@gmail.com>
Tested-by: Andrew Savchenko <bircoph@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of calling 3 times ntohs(random->param_hdr.length), 2 times
ntohs(hmacs->param_hdr.length), and 3 times ntohs(chunks->param_hdr.length)
within the same function, we only call each once and store it in a
variable.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For sensitive data like keying material, it is common practice to zero
out keys before returning the memory back to the allocator. Thus, use
kzfree instead of kfree.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jesse Gross says:
====================
One bug fix for net/3.8 for a long standing problem that was reported a few
times recently.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
ip6gre_tunnel_xmit() is leaking the skb when we hit this error branch,
and the -1 return value from this function is bogus. Use the error
handling we already have in place in ip6gre_tunnel_xmit() for this error
case to fix this.
Signed-off-by: Tommi Rantala <tt.rantala@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
On 64 bit arches :
There is a off-by-one error in qdisc_pkt_len_init() because
mac_header is not set in xmit path.
skb_mac_header() returns an out of bound value that was
harmless because hdr_len is an 'unsigned int'
On 32bit arches, the error is abysmal.
This patch is also a prereq for "macvlan: add multicast filter"
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ben Greear <greearb@candelatech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
skb_gso_segment() is almost always called in tx path,
except for openvswitch. It calls this function when
it receives the packet and tries to queue it to user-space.
In this special case, the ->ip_summed check inside
skb_gso_segment() is no longer true, as ->ip_summed value
has different meanings on rx path.
This patch adjusts skb_gso_segment() so that we can at least
avoid such warnings on checksum.
Cc: Jesse Gross <jesse@nicira.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
head buffer is only temporary available in mac802154_header_create.
So it's not necessary to put it on the heap.
Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
head buffer is only temporary available in lowpan_header_create.
So it's not necessary to put it on the heap.
Also fixed a comment codestyle issue.
Signed-off-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are transients during normal FRTO procedure during which
the packets_in_flight can go to zero between write_queue state
updates and firing the resulting segments out. As FRTO processing
occurs during that window the check must be more precise to
not match "spuriously" :-). More specificly, e.g., when
packets_in_flight is zero but FLAG_DATA_ACKED is true the problematic
branch that set cwnd into zero would not be taken and new segments
might be sent out later.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Tested-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ivan Vercera was recently backporting commit
9c13cb8bb4 to a RHEL kernel, and I noticed that,
while this patch protects the tg3 driver from having its ndo_poll_controller
routine called during device initalization, it does nothing for the driver
during shutdown. I.e. it would be entirely possible to have the
ndo_poll_controller method (or subsequently the ndo_poll) routine called for a
driver in the netpoll path on CPU A while in parallel on CPU B, the ndo_close or
ndo_open routine could be called. Given that the two latter routines tend to
initizlize and free many data structures that the former two rely on, the result
can easily be data corruption or various other crashes. Furthermore, it seems
that this is potentially a problem with all net drivers that support netpoll,
and so this should ideally be fixed in a common path.
As Ben H Pointed out to me, we can't preform dev_open/dev_close in atomic
context, so I've come up with this solution. We can use a mutex to sleep in
open/close paths and just do a mutex_trylock in the napi poll path and abandon
the poll attempt if we're locked, as we'll just retry the poll on the next send
anyway.
I've tested this here by flooding netconsole with messages on a system whos nic
driver I modfied to periodically return NETDEV_TX_BUSY, so that the netpoll tx
workqueue would be forced to send frames and poll the device. While this was
going on I rapidly ifdown/up'ed the interface and watched for any problems.
I've not found any.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: Ivan Vecera <ivecera@redhat.com>
CC: "David S. Miller" <davem@davemloft.net>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Francois Romieu <romieu@fr.zoreil.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Calling icmpv6_send() on a local message size error leads to an
incorrect update of the path mtu in the case when IPsec is used.
So use ipv6_local_error() instead to notify the socket about the
error.
Reported-by: Jiri Bohac <jbohac@suse.cz>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
alloc failures already get standardized OOM
messages and a dump_stack.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP Appropriate Byte Count was added by me, but later disabled.
There is no point in maintaining it since it is a potential source
of bugs and Linux already implements other better window protection
heuristics.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
All in-tree ipv4 protocol implementations are now namespace
aware. Therefore all the run-time checks are superfluous.
Reject registry of any non-namespace aware ipv4 protocol.
Eventually we'll remove prot->netns_ok and this registry
time check as well.
Signed-off-by: David S. Miller <davem@davemloft.net>
The infrastructure is already pretty much entirely there
to allow this conversion.
The tunnel and session lookups have per-namespace tables,
and the ipv4 bind lookup includes the namespace in the
lookup key.
Set netns_ok in l2tp_ip_protocol.
Signed-off-by: David S. Miller <davem@davemloft.net>
When creating unmanaged tunnel sockets we should honour the network namespace
passed to l2tp_tunnel_create. Furthermore, unmanaged tunnel sockets should
not hold a reference to the network namespace lest they accidentally keep
alive a namespace which should otherwise have been released.
Unmanaged tunnel sockets now drop their namespace reference via sk_change_net,
and are released in a new pernet exit callback, l2tp_exit_net.
Signed-off-by: Tom Parkin <tparkin@katalix.com>
Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
l2tp_tunnel_create is passed a pointer to the network namespace for the
tunnel, along with an optional file descriptor for the tunnel which may
be passed in from userspace via. netlink.
In the case where the file descriptor is defined, ensure that the namespace
associated with that socket matches the namespace explicitly passed to
l2tp_tunnel_create.
Signed-off-by: Tom Parkin <tparkin@katalix.com>
Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The L2TP netlink code can run in namespaces. Set the netnsok flag in
genl_family to true to reflect that fact.
Signed-off-by: Tom Parkin <tparkin@katalix.com>
Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To allow l2tp_tunnel_delete to be called from an atomic context, place the
tunnel socket release calls on a workqueue for asynchronous execution.
Tunnel memory is eventually freed in the tunnel socket destructor.
Signed-off-by: Tom Parkin <tparkin@katalix.com>
Signed-off-by: James Chapman <jchapman@katalix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/intel/e1000e/ethtool.c
drivers/net/vmxnet3/vmxnet3_drv.c
drivers/net/wireless/iwlwifi/dvm/tx.c
net/ipv6/route.c
The ipv6 route.c conflict is simple, just ignore the 'net' side change
as we fixed the same problem in 'net-next' by eliminating cached
neighbours from ipv6 routes.
The e1000e conflict is an addition of a new statistic in the ethtool
code, trivial.
The vmxnet3 conflict is about one change in 'net' removing a guarding
conditional, whilst in 'net-next' we had a netdev_info() conversion.
The iwlwifi conflict is dealing with a WARN_ON() conversion in
'net-next' vs. a revert happening in 'net'.
Signed-off-by: David S. Miller <davem@davemloft.net>
With the loop, don't check 'rv' twice in a row. Without the loop, 'rv'
doesn't even need to be checked.
Make the comment more grammar-friendly.
Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As in del_timer() there has already placed a timer_pending() function
to check whether the timer to be deleted is pending or not, it's
unnecessary to check timer pending state again before del_timer() is
called.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It seems due to RCU usage, i.e. within SCTP's address binding list,
a, say, ``behavioral change'' was introduced which does actually
not conform to the RFC anymore. In particular consider the following
(fictional) scenario to demonstrate this:
do:
Two SOCK_SEQPACKET-style sockets are opened (S1, S2)
S1 is bound to 127.0.0.1, port 1024 [server]
S2 is bound to 127.0.0.1, port 1025 [client]
listen(2) is invoked on S1
From S2 we call one sendmsg(2) with msg.msg_name and
msg.msg_namelen parameters set to the server's
address
S1, S2 are closed
goto do
The first pass of this loop passes successful, while the second round
fails during binding of S1 (address still in use). What is happening?
In the first round, the initial handshake is being done, and, at the
time close(2) is called on S1, a non-graceful shutdown is performed via
ABORT since in S1's receive queue an unprocessed packet is present,
thus stating an error condition. This can be considered as a correct
behavior.
During close also all bound addresses are freed, thus nothing *must*
be active anymore. In reference to RFC2960:
After checking the Verification Tag, the receiving endpoint shall
remove the association from its record, and shall report the
termination to its upper layer. (9.1 Abort of an Association)
Also, no half-open states are supported, thus after an ungraceful
shutdown, we leave nothing behind. However, this seems not to be
happening though. In a real-world scenario, this is exactly where
it breaks the lksctp-tools functional test suite, *for instance*:
./test_sockopt
test_sockopt.c 1 PASS : getsockopt(SCTP_STATUS) on a socket with no assoc
test_sockopt.c 2 PASS : getsockopt(SCTP_STATUS)
test_sockopt.c 3 PASS : getsockopt(SCTP_STATUS) with invalid associd
test_sockopt.c 4 PASS : getsockopt(SCTP_STATUS) with NULL associd
test_sockopt.c 5 BROK : bind: Address already in use
The underlying problem is that sctp_endpoint_destroy() hasn't been
triggered yet while the next bind attempt is being done. It will be
triggered eventually (but too late) by sctp_transport_destroy_rcu()
after one RCU grace period:
sctp_transport_destroy()
sctp_transport_destroy_rcu() ----.
sctp_association_put() [*] <--+--> sctp_packet_free()
sctp_association_destroy() [...]
sctp_endpoint_put() skb->destructor
sctp_endpoint_destroy() sctp_wfree()
sctp_bind_addr_free() sctp_association_put() [*]
Thus, we move out the condition with sctp_association_put() as well as
the sctp_packet_free() invocation and the issue can be solved. We also
better free the SCTP chunks first before putting the ref of the association.
With this patch, the example above (which simulates a similar scenario
as in the implementation of this test case) and therefore also the test
suite run successfully through. Tested by myself.
Cc: Vlad Yasevich <vyasevich@gmail.com>
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Acked-by: Vlad Yasevich <vyasevich@gmail.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
