We must set the client call state to RXRPC_CALL_CLIENT_SEND_REQUEST before
attaching the call to the connection struct, not after, as it's liable to
receive errors and conn aborts as soon as the assignment is made - and
these will cause its state to be changed outside of the initiating thread's
control.
Signed-off-by: David Howells <dhowells@redhat.com>
The commit f9b2ee714c ("SUNRPC: Move UDP receive data path
into a workqueue context"), as a side effect, moved the
skb_free_datagram() call outside the scope of the related socket
lock, but UDP sockets require such lock to be held for proper
memory accounting.
Fix it by replacing skb_free_datagram() with
skb_free_datagram_locked().
Fixes: f9b2ee714c ("SUNRPC: Move UDP receive data path into a workqueue context")
Reported-and-tested-by: Jan Stancek <jstancek@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Cc: stable@vger.kernel.org # 4.4+
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Because of the risk of an excessive number of NACK messages and
retransissions, receivers have until now abstained from sending
broadcast NACKS directly upon detection of a packet sequence number
gap. We have instead relied on such gaps being detected by link
protocol STATE message exchange, something that by necessity delays
such detection and subsequent retransmissions.
With the introduction of unicast NACK transmission and rate control
of retransmissions we can now remove this limitation. We now allow
receiving nodes to send NACKS immediately, while coordinating the
permission to do so among the nodes in order to avoid NACK storms.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As cluster sizes grow, so does the amount of identical or overlapping
broadcast NACKs generated by the packet receivers. This often leads to
'NACK crunches' resulting in huge numbers of redundant retransmissions
of the same packet ranges.
In this commit, we introduce rate control of broadcast retransmissions,
so that a retransmitted range cannot be retransmitted again until after
at least 10 ms. This reduces the frequency of duplicate, redundant
retransmissions by an order of magnitude, while having a significant
positive impact on overall throughput and scalability.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we send broadcasts in clusters of more 70-80 nodes, we sometimes
see the broadcast link resetting because of an excessive number of
retransmissions. This is caused by a combination of two factors:
1) A 'NACK crunch", where loss of broadcast packets is discovered
and NACK'ed by several nodes simultaneously, leading to multiple
redundant broadcast retransmissions.
2) The fact that the NACKS as such also are sent as broadcast, leading
to excessive load and packet loss on the transmitting switch/bridge.
This commit deals with the latter problem, by moving sending of
broadcast nacks from the dedicated BCAST_PROTOCOL/NACK message type
to regular unicast LINK_PROTOCOL/STATE messages. We allocate 10 unused
bits in word 8 of the said message for this purpose, and introduce a
new capability bit, TIPC_BCAST_STATE_NACK in order to keep the change
backwards compatible.
Reviewed-by: Ying Xue <ying.xue@windriver.com>
Signed-off-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix the following uninitialised variable warning:
../net/rxrpc/call_event.c: In function 'rxrpc_process_call':
../net/rxrpc/call_event.c:879:58: warning: 'error' may be used uninitialized in this function [-Wmaybe-uninitialized]
_debug("post net error %d", error);
^
Signed-off-by: David Howells <dhowells@redhat.com>
gcc -Wmaybe-initialized correctly points out a newly introduced bug
through which we can end up calling rxrpc_queue_call() for a dead
connection:
net/rxrpc/call_object.c: In function 'rxrpc_mark_call_released':
net/rxrpc/call_object.c:600:5: error: 'sched' may be used uninitialized in this function [-Werror=maybe-uninitialized]
This sets the 'sched' variable to zero to restore the previous
behavior.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: f5c17aaeb2 ("rxrpc: Calls should only have one terminal state")
Signed-off-by: David Howells <dhowells@redhat.com>
Tunnel deletion is delayed by both a workqueue (l2tp_tunnel_delete -> wq
-> l2tp_tunnel_del_work) and RCU (sk_destruct -> RCU ->
l2tp_tunnel_destruct).
By the time l2tp_tunnel_destruct() runs to destroy the tunnel and finish
destroying the socket, the private data reserved via the net_generic
mechanism has already been freed, but l2tp_tunnel_destruct() actually
uses this data.
Make sure tunnel deletion for the netns has completed before returning
from l2tp_exit_net() by first flushing the tunnel removal workqueue, and
then waiting for RCU callbacks to complete.
Fixes: 167eb17e0b ("l2tp: create tunnel sockets in the right namespace")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 8eb30be035 ("ipv6: Create ip6_tnl_xmit") unsets
flowi6_proto in ip4ip6_tnl_xmit() and ip6ip6_tnl_xmit().
Since xfrm_selector_match() relies on this info, IPv6 packets
sent by an ip6tunnel cannot be properly selected by their
protocols after removing it. This patch puts flowi6_proto back.
Cc: stable@vger.kernel.org
Fixes: 8eb30be035 ("ipv6: Create ip6_tnl_xmit")
Signed-off-by: Eli Cooper <elicooper@gmx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a per-port flag to control the unknown multicast flood, similar to the
unknown unicast flood flag and break a few long lines in the netlink flag
exports.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove the unicast flag and introduce an exact pkt_type. That would help us
for the upcoming per-port multicast flood flag and also slightly reduce the
tests in the input fast path.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The original codes depend on that the function parameters are evaluated from
left to right. But the parameter's evaluation order is not defined in C
standard actually.
When flow_keys_have_l4(&keys) is invoked before ___skb_get_hash(skb, &keys,
hashrnd) with some compilers or environment, the keys passed to
flow_keys_have_l4 is not initialized.
Fixes: 6db61d79c1 ("flow_dissector: Ignore flow dissector return value from ___skb_get_hash")
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
fdb dumps spanning multiple skb's currently restart from the first
interface again for every skb. This results in unnecessary
iterations on the already visited interfaces and their fdb
entries. In large scale setups, we have seen this to slow
down fdb dumps considerably. On a system with 30k macs we
see fdb dumps spanning across more than 300 skbs.
To fix the problem, this patch replaces the existing single fdb
marker with three markers: netdev hash entries, netdevs and fdb
index to continue where we left off instead of restarting from the
first netdev. This is consistent with link dumps.
In the process of fixing the performance issue, this patch also
re-implements fix done by
commit 472681d57a ("net: ndo_fdb_dump should report -EMSGSIZE to rtnl_fdb_dump")
(with an internal fix from Wilson Kok) in the following ways:
- change ndo_fdb_dump handlers to return error code instead
of the last fdb index
- use cb->args strictly for dump frag markers and not error codes.
This is consistent with other dump functions.
Below results were taken on a system with 1000 netdevs
and 35085 fdb entries:
before patch:
$time bridge fdb show | wc -l
15065
real 1m11.791s
user 0m0.070s
sys 1m8.395s
(existing code does not return all macs)
after patch:
$time bridge fdb show | wc -l
35085
real 0m2.017s
user 0m0.113s
sys 0m1.942s
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: Wilson Kok <wkok@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Don't expose skbs to in-kernel users, such as the AFS filesystem, but
instead provide a notification hook the indicates that a call needs
attention and another that indicates that there's a new call to be
collected.
This makes the following possibilities more achievable:
(1) Call refcounting can be made simpler if skbs don't hold refs to calls.
(2) skbs referring to non-data events will be able to be freed much sooner
rather than being queued for AFS to pick up as rxrpc_kernel_recv_data
will be able to consult the call state.
(3) We can shortcut the receive phase when a call is remotely aborted
because we don't have to go through all the packets to get to the one
cancelling the operation.
(4) It makes it easier to do encryption/decryption directly between AFS's
buffers and sk_buffs.
(5) Encryption/decryption can more easily be done in the AFS's thread
contexts - usually that of the userspace process that issued a syscall
- rather than in one of rxrpc's background threads on a workqueue.
(6) AFS will be able to wait synchronously on a call inside AF_RXRPC.
To make this work, the following interface function has been added:
int rxrpc_kernel_recv_data(
struct socket *sock, struct rxrpc_call *call,
void *buffer, size_t bufsize, size_t *_offset,
bool want_more, u32 *_abort_code);
This is the recvmsg equivalent. It allows the caller to find out about the
state of a specific call and to transfer received data into a buffer
piecemeal.
afs_extract_data() and rxrpc_kernel_recv_data() now do all the extraction
logic between them. They don't wait synchronously yet because the socket
lock needs to be dealt with.
Five interface functions have been removed:
rxrpc_kernel_is_data_last()
rxrpc_kernel_get_abort_code()
rxrpc_kernel_get_error_number()
rxrpc_kernel_free_skb()
rxrpc_kernel_data_consumed()
As a temporary hack, sk_buffs going to an in-kernel call are queued on the
rxrpc_call struct (->knlrecv_queue) rather than being handed over to the
in-kernel user. To process the queue internally, a temporary function,
temp_deliver_data() has been added. This will be replaced with common code
between the rxrpc_recvmsg() path and the kernel_rxrpc_recv_data() path in a
future patch.
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Yuchung noticed that on the first TFO server data packet sent after
the (TFO) handshake, the server echoed the TCP timestamp value in the
SYN/data instead of the timestamp value in the final ACK of the
handshake. This problem did not happen on regular opens.
The tcp_replace_ts_recent() logic that decides whether to remember an
incoming TS value needs tp->rcv_wup to hold the latest receive
sequence number that we have ACKed (latest tp->rcv_nxt we have
ACKed). This commit fixes this issue by ensuring that a TFO server
properly updates tp->rcv_wup to match tp->rcv_nxt at the time it sends
a SYN/ACK for the SYN/data.
Reported-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Fixes: 168a8f5805 ("tcp: TCP Fast Open Server - main code path")
Signed-off-by: David S. Miller <davem@davemloft.net>
The 'default' value was not advertised.
Fixes: f3a1bfb11c ("rtnl/ipv6: use netconf msg to advertise forwarding status")
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In a dual bearer configuration, if the second tipc link becomes
active while the first link still has pending nametable "bulk"
updates, it randomly leads to reset of the second link.
When a link is established, the function named_distribute(),
fills the skb based on node mtu (allows room for TUNNEL_PROTOCOL)
with NAME_DISTRIBUTOR message for each PUBLICATION.
However, the function named_distribute() allocates the buffer by
increasing the node mtu by INT_H_SIZE (to insert NAME_DISTRIBUTOR).
This consumes the space allocated for TUNNEL_PROTOCOL.
When establishing the second link, the link shall tunnel all the
messages in the first link queue including the "bulk" update.
As size of the NAME_DISTRIBUTOR messages while tunnelling, exceeds
the link mtu the transmission fails (-EMSGSIZE).
Thus, the synch point based on the message count of the tunnel
packets is never reached leading to link timeout.
In this commit, we adjust the size of name distributor message so that
they can be tunnelled.
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Signed-off-by: Parthasarathy Bhuvaragan <parthasarathy.bhuvaragan@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Dmitry reported a double free on kcm socket, which could
be easily reproduced by:
#include <unistd.h>
#include <sys/syscall.h>
int main()
{
int fd = syscall(SYS_socket, 0x29ul, 0x5ul, 0x0ul, 0, 0, 0);
syscall(SYS_ioctl, fd, 0x89e2ul, 0x20a98000ul, 0, 0, 0);
return 0;
}
This is because on the error path, after we install
the new socket file, we call sock_release() to clean
up the socket, which leaves the fd pointing to a freed
socket. Fix this by calling sys_close() on that fd
directly.
Fixes: ab7ac4eb98 ("kcm: Kernel Connection Multiplexor module")
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
commit bc8c20acae ("bridge: multicast: treat igmpv3 report with
INCLUDE and no sources as a leave") seems to have accidentally reverted
commit 47cc84ce0c ("bridge: fix parsing of MLDv2 reports"). This
commit brings back a change to br_ip6_multicast_mld2_report() where
parsing of MLDv2 reports stops when the first group is successfully
added to the MDB cache.
Fixes: bc8c20acae ("bridge: multicast: treat igmpv3 report with INCLUDE and no sources as a leave")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Acked-by: Thadeu Lima de Souza Cascardo <cascardo@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As reported by Lennert the MPLS GSO code is failing to properly segment
large packets. There are a couple of problems:
1. the inner protocol is not set so the gso segment functions for inner
protocol layers are not getting run, and
2 MPLS labels for packets that use the "native" (non-OVS) MPLS code
are not properly accounted for in mpls_gso_segment.
The MPLS GSO code was added for OVS. It is re-using skb_mac_gso_segment
to call the gso segment functions for the higher layer protocols. That
means skb_mac_gso_segment is called twice -- once with the network
protocol set to MPLS and again with the network protocol set to the
inner protocol.
This patch sets the inner skb protocol addressing item 1 above and sets
the network_header and inner_network_header to mark where the MPLS labels
start and end. The MPLS code in OVS is also updated to set the two
network markers.
>From there the MPLS GSO code uses the difference between the network
header and the inner network header to know the size of the MPLS header
that was pushed. It then pulls the MPLS header, resets the mac_len and
protocol for the inner protocol and then calls skb_mac_gso_segment
to segment the skb.
Afterward the inner protocol segmentation is done the skb protocol
is set to mpls for each segment and the network and mac headers
restored.
Reported-by: Lennert Buytenhek <buytenh@wantstofly.org>
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Today mpls iptunnel lwtunnel_output redirect expects the tunnel
output function to handle fragmentation. This is ok but can be
avoided if we did not do the mpls output redirect too early.
ie we could wait until ip fragmentation is done and then call
mpls output for each ip fragment.
To make this work we will need,
1) the lwtunnel state to carry encap headroom
2) and do the redirect to the encap output handler on the ip fragment
(essentially do the output redirect after fragmentation)
This patch adds tunnel headroom in lwtstate to make sure we
account for tunnel data in mtu calculations during fragmentation
and adds new xmit redirect handler to redirect to lwtunnel xmit func
after ip fragmentation.
This includes IPV6 and some mtu fixes and testing from David Ahern.
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After commit 145dd5f9c8 ("net: flush the softnet backlog in process
context"), we can easily batch calls to flush_all_backlogs() for all
devices processed in rollback_registered_many()
Tested:
Before patch, on an idle host.
modprobe dummy numdummies=10000
perf stat -e context-switches -a rmmod dummy
Performance counter stats for 'system wide':
1,211,798 context-switches
1.302137465 seconds time elapsed
After patch:
perf stat -e context-switches -a rmmod dummy
Performance counter stats for 'system wide':
225,523 context-switches
0.721623566 seconds time elapsed
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for your net tree,
they are:
1) Allow nf_tables reject expression from input, forward and output hooks,
since only there the routing information is available, otherwise we crash.
2) Fix unsafe list iteration when flushing timeout and accouting objects.
3) Fix refcount leak on timeout policy parsing failure.
4) Unlink timeout object for unconfirmed conntracks too
5) Missing validation of pkttype mangling from bridge family.
6) Fix refcount leak on ebtables on second lookup for the specific
bridge match extension, this patch from Sabrina Dubroca.
7) Remove unnecessary ip_hdr() in nf_tables_netdev family.
Patches from 1-5 and 7 from Liping Zhang.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes Berg says:
====================
Three little fixes:
* revert a recent wext patch, which Ben Hutchings noticed was
wrong, and it turns out not to be necessary for any driver
* fix an infinite loop that can occur under certain conditions
in mac80211's TDLS code (depending on regulatory information)
* add a cfg80211_get_station() static inline when cfg80211 isn't
built, to allow other modules to not have to depend on it for it
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull NFS client bugfixes from Trond Myklebust:
"Highlights include:
Stable patches:
- Fix a refcount leak in nfs_callback_up_net
- Fix an Oopsable condition when the flexfile pNFS driver connection
to the DS fails
- Fix an Oopsable condition in NFSv4.1 server callback races
- Ensure pNFS clients stop doing I/O to the DS if their lease has
expired, as required by the NFSv4.1 protocol
Bugfixes:
- Fix potential looping in the NFSv4.x migration code
- Patch series to close callback races for OPEN, LAYOUTGET and
LAYOUTRETURN
- Silence WARN_ON when NFSv4.1 over RDMA is in use
- Fix a LAYOUTCOMMIT race in the pNFS/blocks client
- Fix pNFS timeout issues when the DS fails"
* tag 'nfs-for-4.8-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
NFSv4.x: Fix a refcount leak in nfs_callback_up_net
NFS4: Avoid migration loops
pNFS/flexfiles: Fix an Oopsable condition when connection to the DS fails
NFSv4.1: Remove obsolete and incorrrect assignment in nfs4_callback_sequence
NFSv4.1: Close callback races for OPEN, LAYOUTGET and LAYOUTRETURN
NFSv4.1: Defer bumping the slot sequence number until we free the slot
NFSv4.1: Delay callback processing when there are referring triples
NFSv4.1: Fix Oopsable condition in server callback races
SUNRPC: Silence WARN_ON when NFSv4.1 over RDMA is in use
pnfs/blocklayout: update last_write_offset atomically with extents
pNFS: The client must not do I/O to the DS if it's lease has expired
pNFS: Handle NFS4ERR_OLD_STATEID correctly in LAYOUTSTAT calls
pNFS/flexfiles: Set reasonable default retrans values for the data channel
NFS: Allow the mount option retrans=0
pNFS/flexfiles: Fix layoutstat periodic reporting
Pass struct socket * to more rxrpc kernel interface functions. They should
be starting from this rather than the socket pointer in the rxrpc_call
struct if they need to access the socket.
I have left:
rxrpc_kernel_is_data_last()
rxrpc_kernel_get_abort_code()
rxrpc_kernel_get_error_number()
rxrpc_kernel_free_skb()
rxrpc_kernel_data_consumed()
unmodified as they're all about to be removed (and, in any case, don't
touch the socket).
Signed-off-by: David Howells <dhowells@redhat.com>
Provide a function so that kernel users, such as AFS, can ask for the peer
address of a call:
void rxrpc_kernel_get_peer(struct rxrpc_call *call,
struct sockaddr_rxrpc *_srx);
In the future the kernel service won't get sk_buffs to look inside.
Further, this allows us to hide any canonicalisation inside AF_RXRPC for
when IPv6 support is added.
Also propagate this through to afs_find_server() and issue a warning if we
can't handle the address family yet.
Signed-off-by: David Howells <dhowells@redhat.com>
Condense the terminal states of a call state machine to a single state,
plus a separate completion type value. The value is then set, along with
error and abort code values, only when the call is transitioned to the
completion state.
Helpers are provided to simplify this.
Signed-off-by: David Howells <dhowells@redhat.com>
The call pointer in a channel on a connection will be NULL if there's no
active call on that channel. rxrpc_abort_calls() needs to check for this
before trying to take the call's state_lock.
Signed-off-by: David Howells <dhowells@redhat.com>
The nf_log_set is an interface function, so it should do the strict sanity
check of parameters. Convert the return value of nf_log_set as int instead
of void. When the pf is invalid, return -EOPNOTSUPP.
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
There is one macro ARPHRD_ETHER which defines the ethernet proto for ARP,
so we could use it instead of the literal number '1'.
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
After timer removal this just calls nf_ct_delete so remove the __ prefix
version and make nf_ct_kill a shorthand for nf_ct_delete.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
If we evicted a large fraction of the scanned conntrack entries re-schedule
the next gc cycle for immediate execution.
This triggers during tests where load is high, then drops to zero and
many connections will be in TW/CLOSE state with < 30 second timeouts.
Without this change it will take several minutes until conntrack count
comes back to normal.
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Conntrack gc worker to evict stale entries.
GC happens once every 5 seconds, but we only scan at most 1/64th of the
table (and not more than 8k) buckets to avoid hogging cpu.
This means that a complete scan of the table will take several minutes
of wall-clock time.
Considering that the gc run will never have to evict any entries
during normal operation because those will happen from packet path
this should be fine.
We only need gc to make sure userspace (conntrack event listeners)
eventually learn of the timeout, and for resource reclaim in case the
system becomes idle.
We do not disable BH and cond_resched for every bucket so this should
not introduce noticeable latencies either.
A followup patch will add a small change to speed up GC for the extreme
case where most entries are timed out on an otherwise idle system.
v2: Use cond_resched_rcu_qs & add comment wrt. missing restart on
nulls value change in gc worker, suggested by Eric Dumazet.
v3: don't call cancel_delayed_work_sync twice (again, Eric).
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
When dumping we already have to look at the entire table, so we might
as well toss those entries whose timeout value is in the past.
We also look at every entry during resize operations.
However, eviction there is not as simple because we hold the
global resize lock so we can't evict without adding a 'expired' list
to drop from later. Considering that resizes are very rare it doesn't
seem worth doing it.
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
With stats enabled this eats 80 bytes on x86_64 per nf_conn entry, as
Eric Dumazet pointed out during netfilter workshop 2016.
Eric also says: "Another reason was the fact that Thomas was about to
change max timer range [..]" (500462a9de, 'timers: Switch to
a non-cascading wheel').
Remove the timer and use a 32bit jiffies value containing timestamp until
entry is valid.
During conntrack lookup, even before doing tuple comparision, check
the timeout value and evict the entry in case it is too old.
The dying bit is used as a synchronization point to avoid races where
multiple cpus try to evict the same entry.
Because lookup is always lockless, we need to bump the refcnt once
when we evict, else we could try to evict already-dead entry that
is being recycled.
This is the standard/expected way when conntrack entries are destroyed.
Followup patches will introduce garbage colliction via work queue
and further places where we can reap obsoleted entries (e.g. during
netlink dumps), this is needed to avoid expired conntracks from hanging
around for too long when lookup rate is low after a busy period.
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
