linux.git/net/ipv4, branch v3.14-rc5 Linux kernel source tree ipv4: fix counter in_slow_tot 2014-02-17T21:54:42+00:00 Duan Jiong duanj.fnst@cn.fujitsu.com 2014-02-17T07:23:43+00:00 a6254864c08109c66a194612585afc0439005286 since commit 89aef8921bf("ipv4: Delete routing cache."), the counter in_slow_tot can't work correctly. The counter in_slow_tot increase by one when fib_lookup() return successfully in ip_route_input_slow(), but actually the dst struct maybe not be created and cached, so we can increase in_slow_tot after the dst struct is created. Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
since commit 89aef8921bf("ipv4: Delete routing cache."), the counter
in_slow_tot can't work correctly.

The counter in_slow_tot increase by one when fib_lookup() return successfully
in ip_route_input_slow(), but actually the dst struct maybe not be created and
cached, so we can increase in_slow_tot after the dst struct is created.

Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ipv4: distinguish EHOSTUNREACH from the ENETUNREACH 2014-02-17T04:45:31+00:00 Duan Jiong duanj.fnst@cn.fujitsu.com 2014-02-14T10:26:22+00:00 cd0f0b95fd2cd2b716caf5f15db73ab76992789b since commit 251da413("ipv4: Cache ip_error() routes even when not forwarding."), the counter IPSTATS_MIB_INADDRERRORS can't work correctly, because the value of err was always set to ENETUNREACH. Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
since commit 251da413("ipv4: Cache ip_error() routes even when not forwarding."),
the counter IPSTATS_MIB_INADDRERRORS can't work correctly, because the value of
err was always set to ENETUNREACH.

Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ipv4: ipconfig.c: add parentheses in an if statement 2014-02-14T05:14:23+00:00 FX Le Bail fx.lebail@yahoo.com 2014-02-10T15:46:54+00:00 357137a4222b1984b3d867923c7e388669744ceb Even if the 'time_before' macro expand with parentheses, the look is bad. Signed-off-by: Francois-Xavier Le Bail <fx.lebail@yahoo.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
Even if the 'time_before' macro expand with parentheses, the look is bad.

Signed-off-by: Francois-Xavier Le Bail <fx.lebail@yahoo.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: ip, ipv6: handle gso skbs in forwarding path 2014-02-13T22:17:02+00:00 Florian Westphal fw@strlen.de 2014-02-13T22:09:12+00:00 fe6cc55f3a9a053482a76f5a6b2257cee51b4663 Marcelo Ricardo Leitner reported problems when the forwarding link path has a lower mtu than the incoming one if the inbound interface supports GRO. Given: Host <mtu1500> R1 <mtu1200> R2 Host sends tcp stream which is routed via R1 and R2. R1 performs GRO. In this case, the kernel will fail to send ICMP fragmentation needed messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu checks in forward path. Instead, Linux tries to send out packets exceeding the mtu. When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does not fragment the packets when forwarding, and again tries to send out packets exceeding R1-R2 link mtu. This alters the forwarding dstmtu checks to take the individual gso segment lengths into account. For ipv6, we send out pkt too big error for gso if the individual segments are too big. For ipv4, we either send icmp fragmentation needed, or, if the DF bit is not set, perform software segmentation and let the output path create fragments when the packet is leaving the machine. It is not 100% correct as the error message will contain the headers of the GRO skb instead of the original/segmented one, but it seems to work fine in my (limited) tests. Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid sofware segmentation. However it turns out that skb_segment() assumes skb nr_frags is related to mss size so we would BUG there. I don't want to mess with it considering Herbert and Eric disagree on what the correct behavior should be. Hannes Frederic Sowa notes that when we would shrink gso_size skb_segment would then also need to deal with the case where SKB_MAX_FRAGS would be exceeded. This uses sofware segmentation in the forward path when we hit ipv4 non-DF packets and the outgoing link mtu is too small. Its not perfect, but given the lack of bug reports wrt. GRO fwd being broken this is a rare case anyway. Also its not like this could not be improved later once the dust settles. Acked-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com> Signed-off-by: Florian Westphal <fw@strlen.de> Signed-off-by: David S. Miller <davem@davemloft.net> 
Marcelo Ricardo Leitner reported problems when the forwarding link path
has a lower mtu than the incoming one if the inbound interface supports GRO.

Given:
Host <mtu1500> R1 <mtu1200> R2

Host sends tcp stream which is routed via R1 and R2.  R1 performs GRO.

In this case, the kernel will fail to send ICMP fragmentation needed
messages (or pkt too big for ipv6), as GSO packets currently bypass dstmtu
checks in forward path. Instead, Linux tries to send out packets exceeding
the mtu.

When locking route MTU on Host (i.e., no ipv4 DF bit set), R1 does
not fragment the packets when forwarding, and again tries to send out
packets exceeding R1-R2 link mtu.

This alters the forwarding dstmtu checks to take the individual gso
segment lengths into account.

For ipv6, we send out pkt too big error for gso if the individual
segments are too big.

For ipv4, we either send icmp fragmentation needed, or, if the DF bit
is not set, perform software segmentation and let the output path
create fragments when the packet is leaving the machine.
It is not 100% correct as the error message will contain the headers of
the GRO skb instead of the original/segmented one, but it seems to
work fine in my (limited) tests.

Eric Dumazet suggested to simply shrink mss via ->gso_size to avoid
sofware segmentation.

However it turns out that skb_segment() assumes skb nr_frags is related
to mss size so we would BUG there.  I don't want to mess with it considering
Herbert and Eric disagree on what the correct behavior should be.

Hannes Frederic Sowa notes that when we would shrink gso_size
skb_segment would then also need to deal with the case where
SKB_MAX_FRAGS would be exceeded.

This uses sofware segmentation in the forward path when we hit ipv4
non-DF packets and the outgoing link mtu is too small.  Its not perfect,
but given the lack of bug reports wrt. GRO fwd being broken this is a
rare case anyway.  Also its not like this could not be improved later
once the dust settles.

Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Reported-by: Marcelo Ricardo Leitner <mleitner@redhat.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp: tsq: fix nonagle handling 2014-02-10T23:23:39+00:00 John Ogness john.ogness@linutronix.de 2014-02-10T02:40:11+00:00 bf06200e732de613a1277984bf34d1a21c2de03d Commit 46d3ceabd8d9 ("tcp: TCP Small Queues") introduced a possible regression for applications using TCP_NODELAY. If TCP session is throttled because of tsq, we should consult tp->nonagle when TX completion is done and allow us to send additional segment, especially if this segment is not a full MSS. Otherwise this segment is sent after an RTO. [edumazet] : Cooked the changelog, added another fix about testing sk_wmem_alloc twice because TX completion can happen right before setting TSQ_THROTTLED bit. This problem is particularly visible with recent auto corking, but might also be triggered with low tcp_limit_output_bytes values or NIC drivers delaying TX completion by hundred of usec, and very low rtt. Thomas Glanzmann for example reported an iscsi regression, caused by tcp auto corking making this bug quite visible. Fixes: 46d3ceabd8d9 ("tcp: TCP Small Queues") Signed-off-by: John Ogness <john.ogness@linutronix.de> Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Thomas Glanzmann <thomas@glanzmann.de> Signed-off-by: David S. Miller <davem@davemloft.net> 
Commit 46d3ceabd8d9 ("tcp: TCP Small Queues") introduced a possible
regression for applications using TCP_NODELAY.

If TCP session is throttled because of tsq, we should consult
tp->nonagle when TX completion is done and allow us to send additional
segment, especially if this segment is not a full MSS.
Otherwise this segment is sent after an RTO.

[edumazet] : Cooked the changelog, added another fix about testing
sk_wmem_alloc twice because TX completion can happen right before
setting TSQ_THROTTLED bit.

This problem is particularly visible with recent auto corking,
but might also be triggered with low tcp_limit_output_bytes
values or NIC drivers delaying TX completion by hundred of usec,
and very low rtt.

Thomas Glanzmann for example reported an iscsi regression, caused
by tcp auto corking making this bug quite visible.

Fixes: 46d3ceabd8d9 ("tcp: TCP Small Queues")
Signed-off-by: John Ogness <john.ogness@linutronix.de>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Thomas Glanzmann <thomas@glanzmann.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp: correct code comment stating 3 min timeout for FIN_WAIT2, we only do 1 min 2014-02-10T03:14:23+00:00 Jesper Juhl jj@chaosbits.net 2014-02-09T22:30:32+00:00 b10bd54c05a6c3c96549f4642d7de23c99b9853b As far as I can tell we have used a default of 60 seconds for FIN_WAIT2 timeout for ages (since 2.x times??). In any case, the timeout these days is 60 seconds, so the 3 min comment is wrong (and cost me a few minutes of my life when I was debugging a FIN_WAIT2 related problem in a userspace application and checked the kernel source for details). Signed-off-by: Jesper Juhl <jj@chaosbits.net> Signed-off-by: David S. Miller <davem@davemloft.net> 
As far as I can tell we have used a default of 60 seconds for
FIN_WAIT2 timeout for ages (since 2.x times??).

In any case, the timeout these days is 60 seconds, so the 3 min
comment is wrong (and cost me a few minutes of my life when I was
debugging a FIN_WAIT2 related problem in a userspace application and
checked the kernel source for details).

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf 2014-02-09T22:20:00+00:00 David S. Miller davem@davemloft.net 2014-02-09T22:20:00+00:00 f41f03196041f91acad2b6d2b3e1f800aed60100 Pablo Neira Ayuso says: ==================== Netfilter/nftables/IPVS fixes for net The following patchset contains Netfilter/IPVS fixes, mostly nftables fixes, most relevantly they are: * Fix a crash in the h323 conntrack NAT helper due to expectation list corruption, from Alexey Dobriyan. * A couple of RCU race fixes for conntrack, one manifests by hitting BUG_ON in nf_nat_setup_info() and the destroy path, patches from Andrey Vagin and me. * Dump direction attribute in nft_ct only if it is set, from Arturo Borrero. * Fix IPVS bug in its own connection tracking system that may lead to copying only 4 bytes of the IPv6 address when initializing the ip_vs_conn object, from Michal Kubecek. * Fix -EBUSY errors in nftables when deleting the rules, chain and tables in a row due mixture of asynchronous and synchronous object releasing, from me. * Three fixes for the nf_tables set infrastructure when using intervals and mappings, from me. * Four patches to fixing the nf_tables log, reject and ct expressions from the new inet table, from Patrick McHardy. * Fix memory overrun in the map that is used to dynamically allocate names from anonymous sets, also from Patrick. * Fix a potential oops if you dump a set with NFPROTO_UNSPEC and a table name, from Patrick McHardy. ==================== Signed-off-by: David S. Miller <davem@davemloft.net> 
Pablo Neira Ayuso says:

====================
Netfilter/nftables/IPVS fixes for net

The following patchset contains Netfilter/IPVS fixes, mostly nftables
fixes, most relevantly they are:

* Fix a crash in the h323 conntrack NAT helper due to expectation list
  corruption, from Alexey Dobriyan.

* A couple of RCU race fixes for conntrack, one manifests by hitting BUG_ON
  in nf_nat_setup_info() and the destroy path, patches from Andrey Vagin and
  me.

* Dump direction attribute in nft_ct only if it is set, from Arturo
  Borrero.

* Fix IPVS bug in its own connection tracking system that may lead to
  copying only 4 bytes of the IPv6 address when initializing the
  ip_vs_conn object, from Michal Kubecek.

* Fix -EBUSY errors in nftables when deleting the rules, chain and tables
  in a row due mixture of asynchronous and synchronous object releasing,
  from me.

* Three fixes for the nf_tables set infrastructure when using intervals and
  mappings, from me.

* Four patches to fixing the nf_tables log, reject and ct expressions from
  the new inet table, from Patrick McHardy.

* Fix memory overrun in the map that is used to dynamically allocate names
  from anonymous sets, also from Patrick.

* Fix a potential oops if you dump a set with NFPROTO_UNSPEC and a table
  name, from Patrick McHardy.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
tcp: remove 1ms offset in srtt computation 2014-02-07T05:28:06+00:00 Eric Dumazet edumazet@google.com 2014-02-06T23:57:10+00:00 4a5ab4e224288403b0b4b6b8c4d339323150c312 TCP pacing depends on an accurate srtt estimation. Current srtt estimation is using jiffie resolution, and has an artificial offset of at least 1 ms, which can produce slowdowns when FQ/pacing is used, especially in DC world, where typical rtt is below 1 ms. We are planning a switch to usec resolution for linux-3.15, but in the meantime, this patch removes the 1 ms offset. All we need is to have tp->srtt minimal value of 1 to differentiate the case of srtt being initialized or not, not 8. The problematic behavior was observed on a 40Gbit testbed, where 32 concurrent netperf were reaching 12Gbps of aggregate speed, instead of line speed. This patch also has the effect of reporting more accurate srtt and send rates to iproute2 ss command as in : $ ss -i dst cca2 Netid State Recv-Q Send-Q Local Address:Port Peer Address:Port tcp ESTAB 0 0 10.244.129.1:56984 10.244.129.2:12865 cubic wscale:6,6 rto:200 rtt:0.25/0.25 ato:40 mss:1448 cwnd:10 send 463.4Mbps rcv_rtt:1 rcv_space:29200 tcp ESTAB 0 390960 10.244.129.1:60247 10.244.129.2:50204 cubic wscale:6,6 rto:200 rtt:0.875/0.75 mss:1448 cwnd:73 ssthresh:51 send 966.4Mbps unacked:73 retrans:0/121 rcv_space:29200 Reported-by: Vytautas Valancius <valas@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Cc: Yuchung Cheng <ycheng@google.com> Cc: Neal Cardwell <ncardwell@google.com> Acked-by: Neal Cardwell <ncardwell@google.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
TCP pacing depends on an accurate srtt estimation.

Current srtt estimation is using jiffie resolution,
and has an artificial offset of at least 1 ms, which can produce
slowdowns when FQ/pacing is used, especially in DC world,
where typical rtt is below 1 ms.

We are planning a switch to usec resolution for linux-3.15,
but in the meantime, this patch removes the 1 ms offset.

All we need is to have tp->srtt minimal value of 1 to differentiate
the case of srtt being initialized or not, not 8.

The problematic behavior was observed on a 40Gbit testbed,
where 32 concurrent netperf were reaching 12Gbps of aggregate
speed, instead of line speed.

This patch also has the effect of reporting more accurate srtt and send
rates to iproute2 ss command as in :

$ ss -i dst cca2
Netid  State      Recv-Q Send-Q          Local Address:Port
Peer Address:Port
tcp    ESTAB      0      0                10.244.129.1:56984
10.244.129.2:12865
	 cubic wscale:6,6 rto:200 rtt:0.25/0.25 ato:40 mss:1448 cwnd:10 send
463.4Mbps rcv_rtt:1 rcv_space:29200
tcp    ESTAB      0      390960           10.244.129.1:60247
10.244.129.2:50204
	 cubic wscale:6,6 rto:200 rtt:0.875/0.75 mss:1448 cwnd:73 ssthresh:51
send 966.4Mbps unacked:73 retrans:0/121 rcv_space:29200

Reported-by: Vytautas Valancius <valas@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ipv4: Fix runtime WARNING in rtmsg_ifa() 2014-02-07T04:02:15+00:00 Geert Uytterhoeven geert@linux-m68k.org 2014-02-05T07:38:25+00:00 63b5f152eb4a5bb79b9caf7ec37b4201d12f6e66 On m68k/ARAnyM: WARNING: CPU: 0 PID: 407 at net/ipv4/devinet.c:1599 0x316a99() Modules linked in: CPU: 0 PID: 407 Comm: ifconfig Not tainted 3.13.0-atari-09263-g0c71d68014d1 #1378 Stack from 10c4fdf0: 10c4fdf0 002ffabb 000243e8 00000000 008ced6c 00024416 00316a99 0000063f 00316a99 00000009 00000000 002501b4 00316a99 0000063f c0a86117 00000080 c0a86117 00ad0c90 00250a5a 00000014 00ad0c90 00000000 00000000 00000001 00b02dd0 00356594 00000000 00356594 c0a86117 eff6c9e4 008ced6c 00000002 008ced60 0024f9b4 00250b52 00ad0c90 00000000 00000000 00252390 00ad0c90 eff6c9e4 0000004f 00000000 00000000 eff6c9e4 8000e25c eff6c9e4 80001020 Call Trace: [<000243e8>] warn_slowpath_common+0x52/0x6c [<00024416>] warn_slowpath_null+0x14/0x1a [<002501b4>] rtmsg_ifa+0xdc/0xf0 [<00250a5a>] __inet_insert_ifa+0xd6/0x1c2 [<0024f9b4>] inet_abc_len+0x0/0x42 [<00250b52>] inet_insert_ifa+0xc/0x12 [<00252390>] devinet_ioctl+0x2ae/0x5d6 Adding some debugging code reveals that net_fill_ifaddr() fails in put_cacheinfo(skb, ifa->ifa_cstamp, ifa->ifa_tstamp, preferred, valid)) nla_put complains: lib/nlattr.c:454: skb_tailroom(skb) = 12, nla_total_size(attrlen) = 20 Apparently commit 5c766d642bcaffd0c2a5b354db2068515b3846cf ("ipv4: introduce address lifetime") forgot to take into account the addition of struct ifa_cacheinfo in inet_nlmsg_size(). Hence add it, like is already done for ipv6. Suggested-by: Cong Wang <cwang@twopensource.com> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Cong Wang <cwang@twopensource.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
On m68k/ARAnyM:

WARNING: CPU: 0 PID: 407 at net/ipv4/devinet.c:1599 0x316a99()
Modules linked in:
CPU: 0 PID: 407 Comm: ifconfig Not tainted
3.13.0-atari-09263-g0c71d68014d1 #1378
Stack from 10c4fdf0:
        10c4fdf0 002ffabb 000243e8 00000000 008ced6c 00024416 00316a99 0000063f
        00316a99 00000009 00000000 002501b4 00316a99 0000063f c0a86117 00000080
        c0a86117 00ad0c90 00250a5a 00000014 00ad0c90 00000000 00000000 00000001
        00b02dd0 00356594 00000000 00356594 c0a86117 eff6c9e4 008ced6c 00000002
        008ced60 0024f9b4 00250b52 00ad0c90 00000000 00000000 00252390 00ad0c90
        eff6c9e4 0000004f 00000000 00000000 eff6c9e4 8000e25c eff6c9e4 80001020
Call Trace: [<000243e8>] warn_slowpath_common+0x52/0x6c
 [<00024416>] warn_slowpath_null+0x14/0x1a
 [<002501b4>] rtmsg_ifa+0xdc/0xf0
 [<00250a5a>] __inet_insert_ifa+0xd6/0x1c2
 [<0024f9b4>] inet_abc_len+0x0/0x42
 [<00250b52>] inet_insert_ifa+0xc/0x12
 [<00252390>] devinet_ioctl+0x2ae/0x5d6

Adding some debugging code reveals that net_fill_ifaddr() fails in

    put_cacheinfo(skb, ifa->ifa_cstamp, ifa->ifa_tstamp,
                              preferred, valid))

nla_put complains:

    lib/nlattr.c:454: skb_tailroom(skb) = 12, nla_total_size(attrlen) = 20

Apparently commit 5c766d642bcaffd0c2a5b354db2068515b3846cf ("ipv4:
introduce address lifetime") forgot to take into account the addition of
struct ifa_cacheinfo in inet_nlmsg_size(). Hence add it, like is already
done for ipv6.

Suggested-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
netfilter: nf_tables: add reject module for NFPROTO_INET 2014-02-06T08:44:18+00:00 Patrick McHardy kaber@trash.net 2014-02-05T15:03:39+00:00 05513e9e33dbded8124567466a444d32173eecc6 Add a reject module for NFPROTO_INET. It does nothing but dispatch to the AF-specific modules based on the hook family. Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> 
Add a reject module for NFPROTO_INET. It does nothing but dispatch
to the AF-specific modules based on the hook family.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>