linux-stable.git/net/netfilter, branch linux-3.15.y Linux kernel stable tree inetpeer: get rid of ip_id_count 2014-08-14T01:51:48+00:00 Eric Dumazet edumazet@google.com 2014-06-02T12:26:03+00:00 6ea4adaf4cd1183e44b81e62f3226b1816749c77 [ Upstream commit 73f156a6e8c1074ac6327e0abd1169e95eb66463 ] Ideally, we would need to generate IP ID using a per destination IP generator. linux kernels used inet_peer cache for this purpose, but this had a huge cost on servers disabling MTU discovery. 1) each inet_peer struct consumes 192 bytes 2) inetpeer cache uses a binary tree of inet_peer structs, with a nominal size of ~66000 elements under load. 3) lookups in this tree are hitting a lot of cache lines, as tree depth is about 20. 4) If server deals with many tcp flows, we have a high probability of not finding the inet_peer, allocating a fresh one, inserting it in the tree with same initial ip_id_count, (cf secure_ip_id()) 5) We garbage collect inet_peer aggressively. IP ID generation do not have to be 'perfect' Goal is trying to avoid duplicates in a short period of time, so that reassembly units have a chance to complete reassembly of fragments belonging to one message before receiving other fragments with a recycled ID. We simply use an array of generators, and a Jenkin hash using the dst IP as a key. ipv6_select_ident() is put back into net/ipv6/ip6_output.c where it belongs (it is only used from this file) secure_ip_id() and secure_ipv6_id() no longer are needed. Rename ip_select_ident_more() to ip_select_ident_segs() to avoid unnecessary decrement/increment of the number of segments. Signed-off-by: Eric Dumazet <edumazet@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
[ Upstream commit 73f156a6e8c1074ac6327e0abd1169e95eb66463 ]

Ideally, we would need to generate IP ID using a per destination IP
generator.

linux kernels used inet_peer cache for this purpose, but this had a huge
cost on servers disabling MTU discovery.

1) each inet_peer struct consumes 192 bytes

2) inetpeer cache uses a binary tree of inet_peer structs,
   with a nominal size of ~66000 elements under load.

3) lookups in this tree are hitting a lot of cache lines, as tree depth
   is about 20.

4) If server deals with many tcp flows, we have a high probability of
   not finding the inet_peer, allocating a fresh one, inserting it in
   the tree with same initial ip_id_count, (cf secure_ip_id())

5) We garbage collect inet_peer aggressively.

IP ID generation do not have to be 'perfect'

Goal is trying to avoid duplicates in a short period of time,
so that reassembly units have a chance to complete reassembly of
fragments belonging to one message before receiving other fragments
with a recycled ID.

We simply use an array of generators, and a Jenkin hash using the dst IP
as a key.

ipv6_select_ident() is put back into net/ipv6/ip6_output.c where it
belongs (it is only used from this file)

secure_ip_id() and secure_ipv6_id() no longer are needed.

Rename ip_select_ident_more() to ip_select_ident_segs() to avoid
unnecessary decrement/increment of the number of segments.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
netfilter: ctnetlink: fix refcnt leak in dying/unconfirmed list dumper 2014-07-09T18:21:32+00:00 Florian Westphal fw@strlen.de 2014-06-08T09:41:23+00:00 f211ea001a9c90e02a3a0aa5a90e3a3160301c50 commit cd5f336f1780cb20e83146cde64d3d5779e175e6 upstream. 'last' keeps track of the ct that had its refcnt bumped during previous dump cycle. Thus it must not be overwritten until end-of-function. Another (unrelated, theoretical) issue: Don't attempt to bump refcnt of a conntrack whose reference count is already 0. Such conntrack is being destroyed right now, its memory is freed once we release the percpu dying spinlock. Fixes: b7779d06 ('netfilter: conntrack: spinlock per cpu to protect special lists.') Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
commit cd5f336f1780cb20e83146cde64d3d5779e175e6 upstream.

'last' keeps track of the ct that had its refcnt bumped during previous
dump cycle.  Thus it must not be overwritten until end-of-function.

Another (unrelated, theoretical) issue: Don't attempt to bump refcnt of a conntrack
whose reference count is already 0.  Such conntrack is being destroyed
right now, its memory is freed once we release the percpu dying spinlock.

Fixes: b7779d06 ('netfilter: conntrack: spinlock per cpu to protect special lists.')
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
netfilter: nf_nat: fix oops on netns removal 2014-07-09T18:21:32+00:00 Florian Westphal fw@strlen.de 2014-06-07T19:17:04+00:00 23517eb3d184cf7731d74215786cd7ae39da7557 commit 945b2b2d259d1a4364a2799e80e8ff32f8c6ee6f upstream. Quoting Samu Kallio: Basically what's happening is, during netns cleanup, nf_nat_net_exit gets called before ipv4_net_exit. As I understand it, nf_nat_net_exit is supposed to kill any conntrack entries which have NAT context (through nf_ct_iterate_cleanup), but for some reason this doesn't happen (perhaps something else is still holding refs to those entries?). When ipv4_net_exit is called, conntrack entries (including those with NAT context) are cleaned up, but the nat_bysource hashtable is long gone - freed in nf_nat_net_exit. The bug happens when attempting to free a conntrack entry whose NAT hash 'prev' field points to a slot in the freed hash table (head for that bin). We ignore conntracks with null nat bindings. But this is wrong, as these are in bysource hash table as well. Restore nat-cleaning for the netns-is-being-removed case. bug: https://bugzilla.kernel.org/show_bug.cgi?id=65191 Fixes: c2d421e1718 ('netfilter: nf_nat: fix race when unloading protocol modules') Reported-by: Samu Kallio <samu.kallio@aberdeencloud.com> Debugged-by: Samu Kallio <samu.kallio@aberdeencloud.com> Signed-off-by: Florian Westphal <fw@strlen.de> Tested-by: Samu Kallio <samu.kallio@aberdeencloud.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
commit 945b2b2d259d1a4364a2799e80e8ff32f8c6ee6f upstream.

Quoting Samu Kallio:

 Basically what's happening is, during netns cleanup,
 nf_nat_net_exit gets called before ipv4_net_exit. As I understand
 it, nf_nat_net_exit is supposed to kill any conntrack entries which
 have NAT context (through nf_ct_iterate_cleanup), but for some
 reason this doesn't happen (perhaps something else is still holding
 refs to those entries?).

 When ipv4_net_exit is called, conntrack entries (including those
 with NAT context) are cleaned up, but the
 nat_bysource hashtable is long gone - freed in nf_nat_net_exit. The
 bug happens when attempting to free a conntrack entry whose NAT hash
 'prev' field points to a slot in the freed hash table (head for that
 bin).

We ignore conntracks with null nat bindings.  But this is wrong,
as these are in bysource hash table as well.

Restore nat-cleaning for the netns-is-being-removed case.

bug:
https://bugzilla.kernel.org/show_bug.cgi?id=65191

Fixes: c2d421e1718 ('netfilter: nf_nat: fix race when unloading protocol modules')
Reported-by: Samu Kallio <samu.kallio@aberdeencloud.com>
Debugged-by: Samu Kallio <samu.kallio@aberdeencloud.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Tested-by: Samu Kallio <samu.kallio@aberdeencloud.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
netfilter: ctnetlink: fix dumping of dying/unconfirmed conntracks 2014-07-09T18:21:31+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-06-05T12:28:44+00:00 2b4ba2df0337d3d8d0a1ae3bcf9a73e947b68cdd commit 266155b2de8fb721ae353688529b2f8bcdde2f90 upstream. The dumping prematurely stops, it seems the callback argument that indicates that all entries have been dumped is set after iterating on the first cpu list. The dumping also may stop before the entire per-cpu list content is also dumped. With this patch, conntrack -L dying now shows the dying list content again. Fixes: b7779d06 ("netfilter: conntrack: spinlock per cpu to protect special lists.") Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> 
commit 266155b2de8fb721ae353688529b2f8bcdde2f90 upstream.

The dumping prematurely stops, it seems the callback argument that
indicates that all entries have been dumped is set after iterating
on the first cpu list. The dumping also may stop before the entire
per-cpu list content is also dumped.

With this patch, conntrack -L dying now shows the dying list content
again.

Fixes: b7779d06 ("netfilter: conntrack: spinlock per cpu to protect special lists.")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
ipvs: Fix panic due to non-linear skb 2014-05-26T01:22:46+00:00 Peter Christensen pch@ordbogen.com 2014-05-24T19:40:12+00:00 f44a5f45f544561302e855e7bd104e5f506ec01b Receiving a ICMP response to an IPIP packet in a non-linear skb could cause a kernel panic in __skb_pull. The problem was introduced in commit f2edb9f7706dcb2c0d9a362b2ba849efe3a97f5e ("ipvs: implement passive PMTUD for IPIP packets"). Signed-off-by: Peter Christensen <pch@ordbogen.com> Acked-by: Julian Anastasov <ja@ssi.bg> Signed-off-by: Simon Horman <horms@verge.net.au> 
Receiving a ICMP response to an IPIP packet in a non-linear skb could
cause a kernel panic in __skb_pull.

The problem was introduced in
commit f2edb9f7706dcb2c0d9a362b2ba849efe3a97f5e ("ipvs: implement
passive PMTUD for IPIP packets").

Signed-off-by: Peter Christensen <pch@ordbogen.com>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
netfilter: nf_tables: fix trace of matching non-terminal rule 2014-05-15T17:44:20+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-05-15T15:18:26+00:00 3b084e99a3fabaeb0f9c65a0806cde30f0b2835e Add the corresponding trace if we have a full match in a non-terminal rule. Note that the traces will look slightly different than in x_tables since the log message after all expressions have been evaluated (contrary to x_tables, that emits it before the target action). This manifests in two differences in nf_tables wrt. x_tables: 1) The rule that enables the tracing is included in the trace. 2) If the rule emits some log message, that is shown before the trace log message. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
Add the corresponding trace if we have a full match in a non-terminal
rule. Note that the traces will look slightly different than in
x_tables since the log message after all expressions have been
evaluated (contrary to x_tables, that emits it before the target
action). This manifests in two differences in nf_tables wrt. x_tables:

1) The rule that enables the tracing is included in the trace.

2) If the rule emits some log message, that is shown before the
   trace log message.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nf_tables: fix missing return trace at the end of non-base chain 2014-05-12T14:33:11+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-05-11T15:14:49+00:00 7e9bc10db275b22a9db0f976b33b5aeed288da73 Display "return" for implicit rule at the end of a non-base chain, instead of when popping chain from the stack. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
Display "return" for implicit rule at the end of a non-base chain,
instead of when popping chain from the stack.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nf_tables: fix bogus rulenum after goto action 2014-05-12T14:33:10+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-05-10T16:46:02+00:00 f7e7e39b21c285ad73a62fac0736191b8d830704 After returning from the chain that we just went to with no matchings, we get a bogus rule number in the trace. To fix this, we would need to iterate over the list of remaining rules in the chain to update the rule number counter. Patrick suggested to set this to the maximum value since the default base chain policy is the very last action when the processing the base chain is over. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
After returning from the chain that we just went to with no matchings,
we get a bogus rule number in the trace. To fix this, we would need
to iterate over the list of remaining rules in the chain to update the
rule number counter.

Patrick suggested to set this to the maximum value since the default
base chain policy is the very last action when the processing the base
chain is over.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nf_tables: fix tracing of the goto action 2014-05-12T14:33:08+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-05-10T16:42:57+00:00 7b9d5ef932297413adcbd8be98fe612b9527a312 Add missing code to trace goto actions. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
Add missing code to trace goto actions.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: nf_tables: fix goto action 2014-05-12T14:32:41+00:00 Pablo Neira Ayuso pablo@netfilter.org 2014-05-10T16:33:11+00:00 5467a512216753d54f757314c73dbc60f659f9e6 This patch fixes a crash when trying to access the counters and the default chain policy from the non-base chain that we have reached via the goto chain. Fix this by falling back on the original base chain after returning from the custom chain. While fixing this, kill the inline function to account chain statistics to improve source code readability. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
This patch fixes a crash when trying to access the counters and the
default chain policy from the non-base chain that we have reached
via the goto chain. Fix this by falling back on the original base
chain after returning from the custom chain.

While fixing this, kill the inline function to account chain statistics
to improve source code readability.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>