Connection tracking helpers (specifically FTP) need to be called
before NAT sequence numbers adjustments are performed to be able
to compare them against previously seen ones. We've introduced
two new hooks around 2.6.11 to maintain this ordering when NAT
modules were changed to get called from conntrack helpers directly.
The cost of netfilter hooks is quite high and sequence number
adjustments are only rarely needed however. Add a RCU-protected
sequence number adjustment function pointer and call it from
IPv4 conntrack after calling the helper.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Adding extensions to confirmed conntracks is not allowed to avoid races
on reallocation. Don't setup NAT for confirmed conntracks in case NAT
module is loaded late.
The has one side-effect, the connections existing before the NAT module
was loaded won't enter the bysource hash. The only case where this actually
makes a difference is in case of SNAT to a multirange where the IP before
NAT is also part of the range. Since old connections don't enter the
bysource hash the first new connection from the IP will have a new address
selected. This shouldn't matter at all.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Locally generated ICMP packets have a reference to the conntrack entry
of the original packet manually attached by icmp_send(). Therefore the
check for locally originated untracked ICMP redirects can never be
true.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Move the UDP-Lite conntrack checksum validation to a generic helper
similar to nf_checksum() and make it fall back to nf_checksum()
in case the full packet is to be checksummed and hardware checksums
are available. This is to be used by DCCP conntrack, which also
needs to verify partial checksums.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Move responsibility for setting the IP_NAT_RANGE_PROTO_SPECIFIED flag
to the NAT protocol, properly propagate errors and get rid of ugly
return value convention.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Move to nf_nat_proto_common and rename to nf_nat_proto_... since they're
also used by protocols that don't have port numbers.
Signed-off-by: Patrick McHardy <kaber@trash.net>
The port rover should not get overwritten when using random mode,
otherwise other rules will also use more or less random ports.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Rule dumping is performed in two steps: first userspace gets the
ruleset size using getsockopt(SO_GET_INFO) and allocates memory,
then it calls getsockopt(SO_GET_ENTRIES) to actually dump the
ruleset. When another process changes the ruleset in between the
sizes from the first getsockopt call doesn't match anymore and
the kernel aborts. Unfortunately it returns EAGAIN, as for multiple
other possible errors, so userspace can't distinguish this case
from real errors.
Return EAGAIN so userspace can retry the operation.
Fixes (with current iptables SVN version) netfilter bugzilla #104.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Commit 9335f047fe aka
"[NETFILTER]: ip_tables: per-netns FILTER, MANGLE, RAW"
added per-netns _view_ of iptables rules. They were shown to user, but
ignored by filtering code. Now that it's possible to at least ping loopback,
per-netns tables can affect filtering decisions.
netns is taken in case of
PRE_ROUTING, LOCAL_IN -- from in device,
POST_ROUTING, LOCAL_OUT -- from out device,
FORWARD -- from in device which should be equal to out device's netns.
This code is relatively new, so BUG_ON was plugged.
Wrappers were added to a) keep code the same from CONFIG_NET_NS=n users
(overwhelming majority), b) consolidate code in one place -- similar
changes will be done in ipv6 and arp netfilter code.
Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Dump the mark value in log messages similar to nfnetlink_log. This
is useful for debugging complex setups where marks are used for
routing or traffic classification.
Signed-off-by: Patrick McHardy <kaber@trash.net>
Consider we are putting a clusterip_config entry with the "entries"
count == 1, and on the other CPU there's a clusterip_config_find_get
in progress:
CPU1: CPU2:
clusterip_config_entry_put: clusterip_config_find_get:
if (atomic_dec_and_test(&c->entries)) {
/* true */
read_lock_bh(&clusterip_lock);
c = __clusterip_config_find(clusterip);
/* found - it's still in list */
...
atomic_inc(&c->entries);
read_unlock_bh(&clusterip_lock);
write_lock_bh(&clusterip_lock);
list_del(&c->list);
write_unlock_bh(&clusterip_lock);
...
dev_put(c->dev);
Oops! We have an entry returned by the clusterip_config_find_get,
which is a) not in list b) has a stale dev pointer.
The problems will happen when the CPU2 will release the entry - it
will remove it from the list for the 2nd time, thus spoiling it, and
will put a stale dev pointer.
The fix is to make atomic_dec_and_test under the clusterip_lock.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
This expresses __skb_append in terms of __skb_queue_after, exploiting that
__skb_append(old, new, list) = __skb_queue_after(list, old, new).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace seq_open with seq_open_net and remove tcp_seq_release
completely. seq_release_net will do this job just fine.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
No need to create seq_operations for each instance of 'netstat'.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_proc_register/tcp_proc_unregister are called with a static pointer only.
Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
sk_reuse is declared as "unsigned char", but is set as type valbool in net/core/sock.c.
There is no other place in net/ where sk->sk_reuse is set to a value > 1, so the test
"sk_reuse > 1" can not be true.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Documentation/ is a little large, and filesystems/ seems an obvious
place for this file.
Signed-off-by: J. Bruce Fields <bfields@citi.umich.edu>
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Copy the network namespace from the socket to the timewait socket.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Acked-by: Mark Lord <mlord@pobox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The comparison in ip_route_input is a hot path, by recoding the C
"and" as bit operations, fewer conditional branches get generated
so the code should be faster. Maybe someday Gcc will be smart
enough to do this?
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Avoid unneeded test in the case where object to be freed
has to be a leaf. Don't need to use the generic tnode_free()
function, instead just setup leaf to be freed.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The trie pointer is passed down to flush_list and flush_leaf
but never used.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow the use of SACK and window scaling when syncookies are used
and the client supports tcp timestamps. Options are encoded into
the timestamp sent in the syn-ack and restored from the timestamp
echo when the ack is received.
Based on earlier work by Glenn Griffin.
This patch avoids increasing the size of structs by encoding TCP
options into the least significant bits of the timestamp and
by not using any 'timestamp offset'.
The downside is that the timestamp sent in the packet after the synack
will increase by several seconds.
changes since v1:
don't duplicate timestamp echo decoding function, put it into ipv4/syncookie.c
and have ipv6/syncookies.c use it.
Feedback from Glenn Griffin: fix line indented with spaces, kill redundant if ()
Reviewed-by: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use vmalloc rather than alloc_pages to avoid wasting memory.
The problem is that tnode structure has a power of 2 sized array,
plus a header. So the current code wastes almost half the memory
allocated because it always needs the next bigger size to hold
that small header.
This is similar to an earlier patch by Eric, but instead of a list
and lock, I used a workqueue to handle the fact that vfree can't
be done in interrupt context.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
No urgency on the rehash interval timer, so mark it as deferrable.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since route hash is a triple, use jhash_3words rather doing the mixing
directly. This should be as fast and give better distribution.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Don't mark functions that are large as inline, let compiler decide.
Also, use inline rather than __inline__.
Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes kernel bugzilla 10371.
As reported by M.Piechaczek@osmosys.tv, if we try to grab a
char sized socket option value, as in:
unsigned char ttl = 255;
socklen_t len = sizeof(ttl);
setsockopt(socket, IPPROTO_IP, IP_MULTICAST_TTL, &ttl, &len);
getsockopt(socket, IPPROTO_IP, IP_MULTICAST_TTL, &ttl, &len);
The ttl returned will be wrong on big-endian, and on both little-
endian and big-endian the next three bytes in userspace are written
with garbage.
It's because of this test in do_ip_getsockopt():
if (len < sizeof(int) && len > 0 && val>=0 && val<255) {
It should allow a 'val' of 255 to pass here, but it doesn't so it
copies a full 'int' back to userspace.
On little-endian that will write the correct value into the location
but it spams on the next three bytes in userspace. On big endian it
writes the wrong value into the location and spams the next three
bytes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Without this patch, the generic L3 tracker would kick in
if nf_conntrack_ipv4 was not loaded before nf_nat, which
would lead to translation problems with ICMP errors.
NAT does not make sense without IPv4 connection tracking
anyway, so just add a call to need_ipv4_conntrack().
Signed-off-by: Jan Engelhardt <jengelh@computergmbh.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
MTU probe can cause some remedies for FRTO because the normal
packet ordering may be violated allowing FRTO to make a wrong
decision (it might not be that serious threat for anything
though). Thus it's safer to not run FRTO while MTU probe is
underway.
It seems that the basic FRTO variant should also look for an
skb at probe_seq.start to check if that's retransmitted one
but I didn't implement it now (plain seqno in window check
isn't robust against wraparounds).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes Bugzilla #10384
tcp_simple_retransmit does L increment without any checking
whatsoever for overflowing S+L when Reno is in use.
The simplest scenario I can currently think of is rather
complex in practice (there might be some more straightforward
cases though). Ie., if mss is reduced during mtu probing, it
may end up marking everything lost and if some duplicate ACKs
arrived prior to that sacked_out will be non-zero as well,
leading to S+L > packets_out, tcp_clean_rtx_queue on the next
cumulative ACK or tcp_fastretrans_alert on the next duplicate
ACK will fix the S counter.
More straightforward (but questionable) solution would be to
just call tcp_reset_reno_sack() in tcp_simple_retransmit but
it would negatively impact the probe's retransmission, ie.,
the retransmissions would not occur if some duplicate ACKs
had arrived.
So I had to add reno sacked_out reseting to CA_Loss state
when the first cumulative ACK arrives (this stale sacked_out
might actually be the explanation for the reports of left_out
overflows in kernel prior to 2.6.23 and S+L overflow reports
of 2.6.24). However, this alone won't be enough to fix kernel
before 2.6.24 because it is building on top of the commit
1b6d427bb7 ([TCP]: Reduce sacked_out with reno when purging
write_queue) to keep the sacked_out from overflowing.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Reported-by: Alessandro Suardi <alessandro.suardi@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes a long-standing bug which makes NewReno recovery crippled.
With GSO the whole head skb was marked as LOST which is in
violation of NewReno procedure that only wants to mark one packet
and ended up breaking our TCP code by causing counter overflow
because our code was built on top of assumption about valid
NewReno procedure. This manifested as triggering a WARN_ON for
the overflow in a number of places.
It seems relatively safe alternative to just do nothing if
tcp_fragment fails due to oom because another duplicate ACK is
likely to be received soon and the fragmentation will be retried.
Special thanks goes to Soeren Sonnenburg <kernel@nn7.de> who was
lucky enough to be able to reproduce this so that the warning
for the overflow was hit. It's not as easy task as it seems even
if this bug happens quite often because the amount of outstanding
data is pretty significant for the mismarkings to lead to an
overflow.
Because it's very late in 2.6.25-rc cycle (if this even makes in
time), I didn't want to touch anything with SACK enabled here.
Fragmenting might be useful for it as well but it's more or less
a policy decision rather than mandatory fix. Thus there's no need
to rush and we can postpone considering tcp_fragment with SACK
for 2.6.26.
In 2.6.24 and earlier, this very same bug existed but the effect
is slightly different because of a small changes in the if
conditions that fit to the patch's context. With them nothing
got lost marker and thus no retransmissions happened.
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The fast retransmission can be forced locally to the rfc3517
branch in tcp_update_scoreboard instead of making such fragile
constructs deeper in tcp_mark_head_lost.
This is necessary for the next patch which must not have
loopholes for cnt > packets check. As one can notice,
readability got some improvements too because of this :-).
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Patrick McHardy