Dave observed number of machines hitting OOM on the UDP send
path. The workload seems to be sending large UDP packets over
loopback. Since loopback has MTU of 64k kernel will try to
allocate an skb with up to 64k of head space. This has a good
chance of failing under memory pressure. What's worse if
the message length is <32k the allocation may trigger an
OOM killer.
This is entirely avoidable, we can use an skb with page frags.
af_unix solves a similar problem by limiting the head
length to SKB_MAX_ALLOC. This seems like a good and simple
approach. It means that UDP messages > 16kB will now
use fragments if underlying device supports SG, if extra
allocator pressure causes regressions in real workloads
we can switch to trying the large allocation first and
falling back.
v4: pre-calculate all the additions to alloclen so
we can be sure it won't go over order-2
Reported-by: Dave Jones <dsj@fb.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Trivial conflicts in net/can/isotp.c and
tools/testing/selftests/net/mptcp/mptcp_connect.sh
scaled_ppm_to_ppb() was moved from drivers/ptp/ptp_clock.c
to include/linux/ptp_clock_kernel.h in -next so re-apply
the fix there.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
IETF RFC 8986 [1] includes the definition of SRv6 End.DT4, End.DT6, and
End.DT46 Behaviors.
The current SRv6 code in the Linux kernel only implements End.DT4 and
End.DT6 which can be used respectively to support IPv4-in-IPv6 and
IPv6-in-IPv6 VPNs. With End.DT4 and End.DT6 it is not possible to create a
single SRv6 VPN tunnel to carry both IPv4 and IPv6 traffic.
The proposed End.DT46 implementation is meant to support the decapsulation
of IPv4 and IPv6 traffic coming from a single SRv6 tunnel.
The implementation of the SRv6 End.DT46 Behavior in the Linux kernel
greatly simplifies the setup and operations of SRv6 VPNs.
The SRv6 End.DT46 Behavior leverages the infrastructure of SRv6 End.DT{4,6}
Behaviors implemented so far, because it makes use of a VRF device in
order to force the routing lookup into the associated routing table.
To make the End.DT46 work properly, it must be guaranteed that the routing
table used for routing lookup operations is bound to one and only one VRF
during the tunnel creation. Such constraint has to be enforced by enabling
the VRF strict_mode sysctl parameter, i.e.:
$ sysctl -wq net.vrf.strict_mode=1
Note that the same approach is used for the SRv6 End.DT4 Behavior and for
the End.DT6 Behavior in VRF mode.
The command used to instantiate an SRv6 End.DT46 Behavior is
straightforward, i.e.:
$ ip -6 route add 2001:db8::1 encap seg6local action End.DT46 vrftable 100 dev vrf100.
[1] https://www.rfc-editor.org/rfc/rfc8986.html#name-enddt46-decapsulation-and-s
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Performance and impact of SRv6 End.DT46 Behavior on the SRv6 Networking
=======================================================================
This patch aims to add the SRv6 End.DT46 Behavior with minimal impact on
the performance of SRv6 End.DT4 and End.DT6 Behaviors.
In order to verify this, we tested the performance of the newly introduced
SRv6 End.DT46 Behavior and compared it with the performance of SRv6
End.DT{4,6} Behaviors, considering both the patched kernel and the kernel
before applying the End.DT46 patch (referred to as vanilla kernel).
In details, the following decapsulation scenarios were considered:
1.a) IPv6 traffic in SRv6 End.DT46 Behavior on patched kernel;
1.b) IPv4 traffic in SRv6 End.DT46 Behavior on patched kernel;
2.a) SRv6 End.DT6 Behavior (VRF mode) on patched kernel;
2.b) SRv6 End.DT4 Behavior on patched kernel;
3.a) SRv6 End.DT6 Behavior (VRF mode) on vanilla kernel (without the
End.DT46 patch);
3.b) SRv6 End.DT4 Behavior on vanilla kernel (without the End.DT46 patch).
All tests were performed on a testbed deployed on the CloudLab [2]
facilities. We considered IPv{4,6} traffic handled by a single core (at 2.4
GHz on a Xeon(R) CPU E5-2630 v3) on kernel 5.13-rc1 using packets of size
~ 100 bytes.
Scenario (1.a): average 684.70 kpps; std. dev. 0.7 kpps;
Scenario (1.b): average 711.69 kpps; std. dev. 1.2 kpps;
Scenario (2.a): average 690.70 kpps; std. dev. 1.2 kpps;
Scenario (2.b): average 722.22 kpps; std. dev. 1.7 kpps;
Scenario (3.a): average 690.02 kpps; std. dev. 2.6 kpps;
Scenario (3.b): average 721.91 kpps; std. dev. 1.2 kpps;
Considering the results for the patched kernel (1.a, 1.b, 2.a, 2.b) we
observe that the performance degradation incurred in using End.DT46 rather
than End.DT6 and End.DT4 respectively for IPv6 and IPv4 traffic is minimal,
around 0.9% and 1.5%. Such very minimal performance degradation is the
price to be paid if one prefers to use a single tunnel capable of handling
both types of traffic (IPv4 and IPv6).
Comparing the results for End.DT4 and End.DT6 under the patched and the
vanilla kernel (2.a, 2.b, 3.a, 3.b) we observe that the introduction of the
End.DT46 patch has no impact on the performance of End.DT4 and End.DT6.
[2] https://www.cloudlab.us
Signed-off-by: Andrea Mayer <andrea.mayer@uniroma2.it>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Borkmann says:
====================
pull-request: bpf-next 2021-06-17
The following pull-request contains BPF updates for your *net-next* tree.
We've added 50 non-merge commits during the last 25 day(s) which contain
a total of 148 files changed, 4779 insertions(+), 1248 deletions(-).
The main changes are:
1) BPF infrastructure to migrate TCP child sockets from a listener to another
in the same reuseport group/map, from Kuniyuki Iwashima.
2) Add a provably sound, faster and more precise algorithm for tnum_mul() as
noted in https://arxiv.org/abs/2105.05398, from Harishankar Vishwanathan.
3) Streamline error reporting changes in libbpf as planned out in the
'libbpf: the road to v1.0' effort, from Andrii Nakryiko.
4) Add broadcast support to xdp_redirect_map(), from Hangbin Liu.
5) Extends bpf_map_lookup_and_delete_elem() functionality to 4 more map
types, that is, {LRU_,PERCPU_,LRU_PERCPU_,}HASH, from Denis Salopek.
6) Support new LLVM relocations in libbpf to make them more linker friendly,
also add a doc to describe the BPF backend relocations, from Yonghong Song.
7) Silence long standing KUBSAN complaints on register-based shifts in
interpreter, from Daniel Borkmann and Eric Biggers.
8) Add dummy PT_REGS macros in libbpf to fail BPF program compilation when
target arch cannot be determined, from Lorenz Bauer.
9) Extend AF_XDP to support large umems with 1M+ pages, from Magnus Karlsson.
10) Fix two minor libbpf tc BPF API issues, from Kumar Kartikeya Dwivedi.
11) Move libbpf BPF_SEQ_PRINTF/BPF_SNPRINTF macros that can be used by BPF
programs to bpf_helpers.h header, from Florent Revest.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch also changes the code to call reuseport_migrate_sock() and
inet_reqsk_clone(), but unlike the other cases, we do not call
inet_reqsk_clone() right after reuseport_migrate_sock().
Currently, in the receive path for TCP_NEW_SYN_RECV sockets, its listener
has three kinds of refcnt:
(A) for listener itself
(B) carried by reuqest_sock
(C) sock_hold() in tcp_v[46]_rcv()
While processing the req, (A) may disappear by close(listener). Also, (B)
can disappear by accept(listener) once we put the req into the accept
queue. So, we have to hold another refcnt (C) for the listener to prevent
use-after-free.
For socket migration, we call reuseport_migrate_sock() to select a listener
with (A) and to increment the new listener's refcnt in tcp_v[46]_rcv().
This refcnt corresponds to (C) and is cleaned up later in tcp_v[46]_rcv().
Thus we have to take another refcnt (B) for the newly cloned request_sock.
In inet_csk_complete_hashdance(), we hold the count (B), clone the req, and
try to put the new req into the accept queue. By migrating req after
winning the "own_req" race, we can avoid such a worst situation:
CPU 1 looks up req1
CPU 2 looks up req1, unhashes it, then CPU 1 loses the race
CPU 3 looks up req2, unhashes it, then CPU 2 loses the race
...
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.co.jp>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/20210612123224.12525-8-kuniyu@amazon.co.jp
If link mtu is too big, mld_newpack() allocates high-order page.
But most mld packets don't need high-order page.
So, it might waste unnecessary pages.
To avoid this, it makes mld_newpack() try to allocate order-0 page.
Suggested-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The variable err is being initialized with a value that is never read, the
assignment is redundant and can be removed.
Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for net:
1) Fix a crash when stateful expression with its own gc callback
is used in a set definition.
2) Skip IPv6 packets from any link-local address in IPv6 fib expression.
Add a selftest for this scenario, from Florian Westphal.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Kaustubh reported and diagnosed a panic in udp_lib_lookup().
The root cause is udp_abort() racing with close(). Both
racing functions acquire the socket lock, but udp{v6}_destroy_sock()
release it before performing destructive actions.
We can't easily extend the socket lock scope to avoid the race,
instead use the SOCK_DEAD flag to prevent udp_abort from doing
any action when the critical race happens.
Diagnosed-and-tested-by: Kaustubh Pandey <kapandey@codeaurora.org>
Fixes: 5d77dca828 ("net: diag: support SOCK_DESTROY for UDP sockets")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The ip6tables rpfilter match has an extra check to skip packets with
"::" source address.
Extend this to ipv6 fib expression. Else ipv6 duplicate address detection
packets will fail rpf route check -- lookup returns -ENETUNREACH.
While at it, extend the prerouting check to also cover the ingress hook.
Closes: https://bugzilla.netfilter.org/show_bug.cgi?id=1543
Fixes: f6d0cbcf09 ("netfilter: nf_tables: add fib expression")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
When 'nla_parse_nested_deprecated' failed, it's no need to
BUG() here, return -EINVAL is ok.
Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter updates for net-next
The following patchset contains Netfilter updates for net-next:
1) Support for SCTP chunks matching on nf_tables, from Phil Sutter.
2) Skip LDMXCSR, we don't need a valid MXCSR state. From Stefano Brivio.
3) CONFIG_RETPOLINE for nf_tables set lookups, from Florian Westphal.
4) A few Kconfig leading spaces removal, from Juerg Haefliger.
5) Remove spinlock from xt_limit, from Jason Baron.
6) Remove useless initialization in xt_CT, oneliner from Yang Li.
7) Tree-wide replacement of netlink_unicast() by nfnetlink_unicast().
8) Reduce footprint of several structures: xt_action_param,
nft_pktinfo and nf_hook_state, from Florian.
10) Add nft_thoff() and nft_sk() helpers and use them, also from Florian.
11) Fix documentation in nf_tables pipapo avx2, from Florian Westphal.
12) Fix clang-12 fmt string warnings, also from Florian.
====================
This is a complement to commit aa6dd211e4 ("inet: use bigger hash
table for IP ID generation"), but focusing on some specific aspects
of IPv6.
Contary to IPv4, IPv6 only uses packet IDs with fragments, and with a
minimum MTU of 1280, it's much less easy to force a remote peer to
produce many fragments to explore its ID sequence. In addition packet
IDs are 32-bit in IPv6, which further complicates their analysis. On
the other hand, it is often easier to choose among plenty of possible
source addresses and partially work around the bigger hash table the
commit above permits, which leaves IPv6 partially exposed to some
possibilities of remote analysis at the risk of weakening some
protocols like DNS if some IDs can be predicted with a good enough
probability.
Given the wide range of permitted IDs, the risk of collision is extremely
low so there's no need to rely on the positive increment algorithm that
is shared with the IPv4 code via ip_idents_reserve(). We have a fast
PRNG, so let's simply call prandom_u32() and be done with it.
Performance measurements at 10 Gbps couldn't show any difference with
the previous code, even when using a single core, because due to the
large fragments, we're limited to only ~930 kpps at 10 Gbps and the cost
of the random generation is completely offset by other operations and by
the network transfer time. In addition, this change removes the need to
update a shared entry in the idents table so it may even end up being
slightly faster on large scale systems where this matters.
The risk of at least one collision here is about 1/80 million among
10 IDs, 1/850k among 100 IDs, and still only 1/8.5k among 1000 IDs,
which remains very low compared to IPv4 where all IDs are reused
every 4 to 80ms on a 10 Gbps flow depending on packet sizes.
Reported-by: Amit Klein <aksecurity@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20210529110746.6796-1-w@1wt.eu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This allows to change storage placement later on without changing readers.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The fragment offset in ipv4/ipv6 is a 16bit field, so use
u16 instead of unsigned int.
On 64bit: 40 bytes to 32 bytes. By extension this also reduces
nft_pktinfo (56 to 48 byte).
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Commit dbd1759e6a ("ipv6: on reassembly, record frag_max_size")
filled the frag_max_size field in IP6CB in the input path.
The field should also be filled in case of atomic fragments.
Fixes: dbd1759e6a ('ipv6: on reassembly, record frag_max_size')
Signed-off-by: Francesco Ruggeri <fruggeri@arista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In-kernel notifications are already sent when the multipath hash policy
itself changes, but not when the multipath hash fields change.
Add these notifications, so that interested listeners (e.g., switch ASIC
drivers) could perform the necessary configuration.
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a new multipath hash policy where the packet fields used for hash
calculation are determined by user space via the
fib_multipath_hash_fields sysctl that was introduced in the previous
patch.
The current set of available packet fields includes both outer and inner
fields, which requires two invocations of the flow dissector. Avoid
unnecessary dissection of the outer or inner flows by skipping
dissection if none of the outer or inner fields are required.
In accordance with the existing policies, when an skb is not available,
packet fields are extracted from the provided flow key. In which case,
only outer fields are considered.
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
A subsequent patch will add a new multipath hash policy where the packet
fields used for multipath hash calculation are determined by user space.
This patch adds a sysctl that allows user space to set these fields.
The packet fields are represented using a bitmask and are common between
IPv4 and IPv6 to allow user space to use the same numbering across both
protocols. For example, to hash based on standard 5-tuple:
# sysctl -w net.ipv6.fib_multipath_hash_fields=0x0037
net.ipv6.fib_multipath_hash_fields = 0x0037
To avoid introducing holes in 'struct netns_sysctl_ipv6', move the
'bindv6only' field after the multipath hash fields.
The kernel rejects unknown fields, for example:
# sysctl -w net.ipv6.fib_multipath_hash_fields=0x1000
sysctl: setting key "net.ipv6.fib_multipath_hash_fields": Invalid argument
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
A subsequent patch will add another multipath hash policy where the
multipath hash is calculated directly by the policy specific code and
not outside of the switch statement.
Prepare for this change by moving the multipath hash calculation inside
the switch statement.
No functional changes intended.
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 'out_timer' label was added in commit 63152fc0de ("[NETNS][IPV6]
ip6_fib - gc timer per namespace") when the timer was allocated on the
heap.
Commit 417f28bb34 ("netns: dont alloc ipv6 fib timer list") removed
the allocation, but kept the label name.
Rename it to a more suitable name.
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
mld_newpack() doesn't allow to allocate high order page,
only order-0 allocation is allowed.
If headroom size is too large, a kernel panic could occur in skb_put().
Test commands:
ip netns del A
ip netns del B
ip netns add A
ip netns add B
ip link add veth0 type veth peer name veth1
ip link set veth0 netns A
ip link set veth1 netns B
ip netns exec A ip link set lo up
ip netns exec A ip link set veth0 up
ip netns exec A ip -6 a a 2001:db8:0::1/64 dev veth0
ip netns exec B ip link set lo up
ip netns exec B ip link set veth1 up
ip netns exec B ip -6 a a 2001:db8:0::2/64 dev veth1
for i in {1..99}
do
let A=$i-1
ip netns exec A ip link add ip6gre$i type ip6gre \
local 2001:db8:$A::1 remote 2001:db8:$A::2 encaplimit 100
ip netns exec A ip -6 a a 2001:db8:$i::1/64 dev ip6gre$i
ip netns exec A ip link set ip6gre$i up
ip netns exec B ip link add ip6gre$i type ip6gre \
local 2001:db8:$A::2 remote 2001:db8:$A::1 encaplimit 100
ip netns exec B ip -6 a a 2001:db8:$i::2/64 dev ip6gre$i
ip netns exec B ip link set ip6gre$i up
done
Splat looks like:
kernel BUG at net/core/skbuff.c:110!
invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
CPU: 0 PID: 7 Comm: kworker/0:1 Not tainted 5.12.0+ #891
Workqueue: ipv6_addrconf addrconf_dad_work
RIP: 0010:skb_panic+0x15d/0x15f
Code: 92 fe 4c 8b 4c 24 10 53 8b 4d 70 45 89 e0 48 c7 c7 00 ae 79 83
41 57 41 56 41 55 48 8b 54 24 a6 26 f9 ff <0f> 0b 48 8b 6c 24 20 89
34 24 e8 4a 4e 92 fe 8b 34 24 48 c7 c1 20
RSP: 0018:ffff88810091f820 EFLAGS: 00010282
RAX: 0000000000000089 RBX: ffff8881086e9000 RCX: 0000000000000000
RDX: 0000000000000089 RSI: 0000000000000008 RDI: ffffed1020123efb
RBP: ffff888005f6eac0 R08: ffffed1022fc0031 R09: ffffed1022fc0031
R10: ffff888117e00187 R11: ffffed1022fc0030 R12: 0000000000000028
R13: ffff888008284eb0 R14: 0000000000000ed8 R15: 0000000000000ec0
FS: 0000000000000000(0000) GS:ffff888117c00000(0000)
knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8b801c5640 CR3: 0000000033c2c006 CR4: 00000000003706f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
? ip6_mc_hdr.isra.26.constprop.46+0x12a/0x600
? ip6_mc_hdr.isra.26.constprop.46+0x12a/0x600
skb_put.cold.104+0x22/0x22
ip6_mc_hdr.isra.26.constprop.46+0x12a/0x600
? rcu_read_lock_sched_held+0x91/0xc0
mld_newpack+0x398/0x8f0
? ip6_mc_hdr.isra.26.constprop.46+0x600/0x600
? lock_contended+0xc40/0xc40
add_grhead.isra.33+0x280/0x380
add_grec+0x5ca/0xff0
? mld_sendpack+0xf40/0xf40
? lock_downgrade+0x690/0x690
mld_send_initial_cr.part.34+0xb9/0x180
ipv6_mc_dad_complete+0x15d/0x1b0
addrconf_dad_completed+0x8d2/0xbb0
? lock_downgrade+0x690/0x690
? addrconf_rs_timer+0x660/0x660
? addrconf_dad_work+0x73c/0x10e0
addrconf_dad_work+0x73c/0x10e0
Allowing high order page allocation could fix this problem.
Fixes: 72e09ad107 ("ipv6: avoid high order allocations")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a tracepoint for capturing TCP segments with
a bad checksum. This makes it easy to identify
sources of bad frames in the fleet (e.g. machines
with faulty NICs).
It should also help tools like IOvisor's tcpdrop.py
which are used today to get detailed information
about such packets.
We don't have a socket in many cases so we must
open code the address extraction based just on
the skb.
v2: add missing export for ipv6=m
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Variable 'err' is set to -ENOMEM but this value is never read as it is
overwritten with a new value later on, hence the 'If statements' and
assignments are redundantand and can be removed.
Cleans up the following clang-analyzer warning:
net/ipv6/seg6.c:126:4: warning: Value stored to 'err' is never read
[clang-analyzer-deadcode.DeadStores]
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch provides counters for SRv6 Behaviors as defined in [1],
section 6. For each SRv6 Behavior instance, counters defined in [1] are:
- the total number of packets that have been correctly processed;
- the total amount of traffic in bytes of all packets that have been
correctly processed;
In addition, this patch introduces a new counter that counts the number of
packets that have NOT been properly processed (i.e. errors) by an SRv6
Behavior instance.
Counters are not only interesting for network monitoring purposes (i.e.
counting the number of packets processed by a given behavior) but they also
provide a simple tool for checking whether a behavior instance is working
as we expect or not.
Counters can be useful for troubleshooting misconfigured SRv6 networks.
Indeed, an SRv6 Behavior can silently drop packets for very different
reasons (i.e. wrong SID configuration, interfaces set with SID addresses,
etc) without any notification/message to the user.
Due to the nature of SRv6 networks, diagnostic tools such as ping and
traceroute may be ineffective: paths used for reaching a given router can
be totally different from the ones followed by probe packets. In addition,
paths are often asymmetrical and this makes it even more difficult to keep
up with the journey of the packets and to understand which behaviors are
actually processing our traffic.
When counters are enabled on an SRv6 Behavior instance, it is possible to
verify if packets are actually processed by such behavior and what is the
outcome of the processing. Therefore, the counters for SRv6 Behaviors offer
an non-invasive observability point which can be leveraged for both traffic
monitoring and troubleshooting purposes.
[1] https://www.rfc-editor.org/rfc/rfc8986.html#name-counters
Troubleshooting using SRv6 Behavior counters
--------------------------------------------
Let's make a brief example to see how helpful counters can be for SRv6
networks. Let's consider a node where an SRv6 End Behavior receives an SRv6
packet whose Segment Left (SL) is equal to 0. In this case, the End
Behavior (which accepts only packets with SL >= 1) discards the packet and
increases the error counter.
This information can be leveraged by the network operator for
troubleshooting. Indeed, the error counter is telling the user that the
packet:
(i) arrived at the node;
(ii) the packet has been taken into account by the SRv6 End behavior;
(iii) but an error has occurred during the processing.
The error (iii) could be caused by different reasons, such as wrong route
settings on the node or due to an invalid SID List carried by the SRv6
packet. Anyway, the error counter is used to exclude that the packet did
not arrive at the node or it has not been processed by the behavior at
all.
Turning on/off counters for SRv6 Behaviors
------------------------------------------
Each SRv6 Behavior instance can be configured, at the time of its creation,
to make use of counters.
This is done through iproute2 which allows the user to create an SRv6
Behavior instance specifying the optional "count" attribute as shown in the
following example:
$ ip -6 route add 2001:db8::1 encap seg6local action End count dev eth0
per-behavior counters can be shown by adding "-s" to the iproute2 command
line, i.e.:
$ ip -s -6 route show 2001:db8::1
2001:db8::1 encap seg6local action End packets 0 bytes 0 errors 0 dev eth0
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Impact of counters for SRv6 Behaviors on performance
====================================================
To determine the performance impact due to the introduction of counters in
the SRv6 Behavior subsystem, we have carried out extensive tests.
We chose to test the throughput achieved by the SRv6 End.DX2 Behavior
because, among all the other behaviors implemented so far, it reaches the
highest throughput which is around 1.5 Mpps (per core at 2.4 GHz on a
Xeon(R) CPU E5-2630 v3) on kernel 5.12-rc2 using packets of size ~ 100
bytes.
Three different tests were conducted in order to evaluate the overall
throughput of the SRv6 End.DX2 Behavior in the following scenarios:
1) vanilla kernel (without the SRv6 Behavior counters patch) and a single
instance of an SRv6 End.DX2 Behavior;
2) patched kernel with SRv6 Behavior counters and a single instance of
an SRv6 End.DX2 Behavior with counters turned off;
3) patched kernel with SRv6 Behavior counters and a single instance of
SRv6 End.DX2 Behavior with counters turned on.
All tests were performed on a testbed deployed on the CloudLab facilities
[2], a flexible infrastructure dedicated to scientific research on the
future of Cloud Computing.
Results of tests are shown in the following table:
Scenario (1): average 1504764,81 pps (~1504,76 kpps); std. dev 3956,82 pps
Scenario (2): average 1501469,78 pps (~1501,47 kpps); std. dev 2979,85 pps
Scenario (3): average 1501315,13 pps (~1501,32 kpps); std. dev 2956,00 pps
As can be observed, throughputs achieved in scenarios (2),(3) did not
suffer any observable degradation compared to scenario (1).
Thanks to Jakub Kicinski and David Ahern for their valuable suggestions
and comments provided during the discussion of the proposed RFCs.
[2] https://www.cloudlab.us
Signed-off-by: Andrea Mayer <andrea.mayer@uniroma2.it>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The IPv6 Multicast Router Advertisements parsing has the following two
issues:
For one thing, ICMPv6 MRD Advertisements are smaller than ICMPv6 MLD
messages (ICMPv6 MRD Adv.: 8 bytes vs. ICMPv6 MLDv1/2: >= 24 bytes,
assuming MLDv2 Reports with at least one multicast address entry).
When ipv6_mc_check_mld_msg() tries to parse an Multicast Router
Advertisement its MLD length check will fail - and it will wrongly
return -EINVAL, even if we have a valid MRD Advertisement. With the
returned -EINVAL the bridge code will assume a broken packet and will
wrongly discard it, potentially leading to multicast packet loss towards
multicast routers.
The second issue is the MRD header parsing in
br_ip6_multicast_mrd_rcv(): It wrongly checks for an ICMPv6 header
immediately after the IPv6 header (IPv6 next header type). However
according to RFC4286, section 2 all MRD messages contain a Router Alert
option (just like MLD). So instead there is an IPv6 Hop-by-Hop option
for the Router Alert between the IPv6 and ICMPv6 header, again leading
to the bridge wrongly discarding Multicast Router Advertisements.
To fix these two issues, introduce a new return value -ENODATA to
ipv6_mc_check_mld() to indicate a valid ICMPv6 packet with a hop-by-hop
option which is not an MLD but potentially an MRD packet. This also
simplifies further parsing in the bridge code, as ipv6_mc_check_mld()
already fully checks the ICMPv6 header and hop-by-hop option.
These issues were found and fixed with the help of the mrdisc tool
(https://github.com/troglobit/mrdisc).
Fixes: 4b3087c7e3 ("bridge: Snoop Multicast Router Advertisements")
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>
The compat layer needs to parse untrusted input (the ruleset)
to translate it to a 64bit compatible format.
We had a number of bugs in this department in the past, so allow users
to turn this feature off.
Add CONFIG_NETFILTER_XTABLES_COMPAT kconfig knob and make it default to y
to keep existing behaviour.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Same patch as the ip_tables one: removal of all accesses to ip6_tables
xt_table pointers. After this patch the struct net xt_table anchors
can be removed.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
This changes how ip(6)table nat passes the ruleset/table to the
evaluation loop.
At the moment, it will fetch the table from struct net.
This change stores the table in the hook_ops 'priv' argument
instead.
This requires to duplicate the hook_ops for each netns, so
they can store the (per-net) xt_table structure.
The dupliated nat hook_ops get stored in net_generic data area.
They are free'd in the namespace exit path.
This is a pre-requisite to remove the xt_table/ruleset pointers
from struct net.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
No need for these.
There is only one caller, the xtables core, when the table is registered
for the first time with a particular network namespace.
After ->table_init() call, the table is linked into the tables[af] list,
so next call to that function will skip the ->table_init().
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Its the same function as ipt_unregister_table_exit.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
When I changed defrag hooks to no longer get registered by default I
intentionally made it so that registration can only be un-done by unloading
the nf_defrag_ipv4/6 module.
In hindsight this was too conservative; there is no reason to keep defrag
on while there is no feature dependency anymore.
Moreover, this won't work if user isn't allowed to remove nf_defrag module.
This adds the disable() functions for both ipv4 and ipv6 and calls them
from conntrack, TPROXY and the xtables socket module.
ipvs isn't converted here, it will behave as before this patch and
will need module removal.
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Pablo Neira Ayuso says:
====================
Netfilter updates for net-next
The following patchset contains Netfilter updates for net-next:
1) Add vlan match and pop actions to the flowtable offload,
patches from wenxu.
2) Reduce size of the netns_ct structure, which itself is
embedded in struct net Make netns_ct a read-mostly structure.
Patches from Florian Westphal.
3) Add FLOW_OFFLOAD_XMIT_UNSPEC to skip dst check from garbage
collector path, as required by the tc CT action. From Roi Dayan.
4) VLAN offload fixes for nftables: Allow for matching on both s-vlan
and c-vlan selectors. Fix match of VLAN id due to incorrect
byteorder. Add a new routine to properly populate flow dissector
ethertypes.
5) Missing keys in ip{6}_route_me_harder() results in incorrect
routes. This includes an update for selftest infra. Patches
from Ido Schimmel.
6) Add counter hardware offload support through FLOW_CLS_STATS.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Netfilter tries to reroute mangled packets as a different route might
need to be used following the mangling. When this happens, netfilter
does not populate the IP protocol, the source port and the destination
port in the flow key. Therefore, FIB rules that match on these fields
are ignored and packets can be misrouted.
Solve this by dissecting the outer flow and populating the flow key
before rerouting the packet. Note that flow dissection only happens when
FIB rules that match on these fields are installed, so in the common
case there should not be a penalty.
Reported-by: Michal Soltys <msoltyspl@yandex.pl>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
- keep the ZC code, drop the code related to reinit
net/bridge/netfilter/ebtables.c
- fix build after move to net_generic
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
__ipv6_dev_mc_dec() internally uses sleepable functions so that caller
must not acquire atomic locks. But caller, which is addrconf_verify_rtnl()
acquires rcu_read_lock_bh().
So this warning occurs in the __ipv6_dev_mc_dec().
Test commands:
ip netns add A
ip link add veth0 type veth peer name veth1
ip link set veth1 netns A
ip link set veth0 up
ip netns exec A ip link set veth1 up
ip a a 2001:db8::1/64 dev veth0 valid_lft 2 preferred_lft 1
Splat looks like:
============================
WARNING: suspicious RCU usage
5.12.0-rc6+ #515 Not tainted
-----------------------------
kernel/sched/core.c:8294 Illegal context switch in RCU-bh read-side
critical section!
other info that might help us debug this:
rcu_scheduler_active = 2, debug_locks = 1
4 locks held by kworker/4:0/1997:
#0: ffff88810bd72d48 ((wq_completion)ipv6_addrconf){+.+.}-{0:0}, at:
process_one_work+0x761/0x1440
#1: ffff888105c8fe00 ((addr_chk_work).work){+.+.}-{0:0}, at:
process_one_work+0x795/0x1440
#2: ffffffffb9279fb0 (rtnl_mutex){+.+.}-{3:3}, at:
addrconf_verify_work+0xa/0x20
#3: ffffffffb8e30860 (rcu_read_lock_bh){....}-{1:2}, at:
addrconf_verify_rtnl+0x23/0xc60
stack backtrace:
CPU: 4 PID: 1997 Comm: kworker/4:0 Not tainted 5.12.0-rc6+ #515
Workqueue: ipv6_addrconf addrconf_verify_work
Call Trace:
dump_stack+0xa4/0xe5
___might_sleep+0x27d/0x2b0
__mutex_lock+0xc8/0x13f0
? lock_downgrade+0x690/0x690
? __ipv6_dev_mc_dec+0x49/0x2a0
? mark_held_locks+0xb7/0x120
? mutex_lock_io_nested+0x1270/0x1270
? lockdep_hardirqs_on_prepare+0x12c/0x3e0
? _raw_spin_unlock_irqrestore+0x47/0x50
? trace_hardirqs_on+0x41/0x120
? __wake_up_common_lock+0xc9/0x100
? __wake_up_common+0x620/0x620
? memset+0x1f/0x40
? netlink_broadcast_filtered+0x2c4/0xa70
? __ipv6_dev_mc_dec+0x49/0x2a0
__ipv6_dev_mc_dec+0x49/0x2a0
? netlink_broadcast_filtered+0x2f6/0xa70
addrconf_leave_solict.part.64+0xad/0xf0
? addrconf_join_solict.part.63+0xf0/0xf0
? nlmsg_notify+0x63/0x1b0
__ipv6_ifa_notify+0x22c/0x9c0
? inet6_fill_ifaddr+0xbe0/0xbe0
? lockdep_hardirqs_on_prepare+0x12c/0x3e0
? __local_bh_enable_ip+0xa5/0xf0
? ipv6_del_addr+0x347/0x870
ipv6_del_addr+0x3b1/0x870
? addrconf_ifdown+0xfe0/0xfe0
? rcu_read_lock_any_held.part.27+0x20/0x20
addrconf_verify_rtnl+0x8a9/0xc60
addrconf_verify_work+0xf/0x20
process_one_work+0x84c/0x1440
In order to avoid this problem, it uses rcu_read_unlock_bh() for
a short time. RCU is used for avoiding freeing
ifp(struct *inet6_ifaddr) while ifp is being used. But this will
not be released even if rcu_read_unlock_bh() is used.
Because before rcu_read_unlock_bh(), it uses in6_ifa_hold(ifp).
So this is safe.
Fixes: 63ed8de4be ("mld: add mc_lock for protecting per-interface mld data")
Suggested-by: Eric Dumazet <edumazet@google.com>
Reported-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Steffen Klassert says:
====================
pull request (net-next): ipsec-next 2021-04-14
Not much this time:
1) Simplification of some variable calculations in esp4 and esp6.
From Jiapeng Chong and Junlin Yang.
2) Fix a clang Wformat warning in esp6 and ah6.
From Arnd Bergmann.
Please pull or let me know if there are problems.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The current icmp_rcv function drops all unknown ICMP types, including
ICMP_EXT_ECHOREPLY (type 43). In order to parse Extended Echo Reply messages, we have
to pass these packets to the ping_rcv function, which does not do any
other filtering and passes the packet to the designated socket.
Pass incoming RFC 8335 ICMP Extended Echo Reply packets to the ping_rcv
handler instead of discarding the packet.
Signed-off-by: Andreas Roeseler <andreas.a.roeseler@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similarly to the sit case, we need to remove the tunnels with no
addresses that have been moved to another network namespace.
Fixes: 0bd8762824 ("ip6tnl: add x-netns support")
Signed-off-by: Hristo Venev <hristo@venev.name>
Signed-off-by: David S. Miller <davem@davemloft.net>
A sit interface created without a local or a remote address is linked
into the `sit_net::tunnels_wc` list of its original namespace. When
deleting a network namespace, delete the devices that have been moved.
The following script triggers a null pointer dereference if devices
linked in a deleted `sit_net` remain:
for i in `seq 1 30`; do
ip netns add ns-test
ip netns exec ns-test ip link add dev veth0 type veth peer veth1
ip netns exec ns-test ip link add dev sit$i type sit dev veth0
ip netns exec ns-test ip link set dev sit$i netns $$
ip netns del ns-test
done
for i in `seq 1 30`; do
ip link del dev sit$i
done
Fixes: 5e6700b3bf ("sit: add support of x-netns")
Signed-off-by: Hristo Venev <hristo@venev.name>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for net:
1) Fix NAT IPv6 offload in the flowtable.
2) icmpv6 is printed as unknown in /proc/net/nf_conntrack.
3) Use div64_u64() in nft_limit, from Eric Dumazet.
4) Use pre_exit to unregister ebtables and arptables hooks,
from Florian Westphal.
5) Fix out-of-bound memset in x_tables compat match/target,
also from Florian.
6) Clone set elements expression to ensure proper initialization.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
