This commit implements a new TCP congestion control algorithm: BBR
(Bottleneck Bandwidth and RTT). A detailed description of BBR will be
published in ACM Queue, Vol. 14 No. 5, September-October 2016, as
"BBR: Congestion-Based Congestion Control".
BBR has significantly increased throughput and reduced latency for
connections on Google's internal backbone networks and google.com and
YouTube Web servers.
BBR requires only changes on the sender side, not in the network or
the receiver side. Thus it can be incrementally deployed on today's
Internet, or in datacenters.
The Internet has predominantly used loss-based congestion control
(largely Reno or CUBIC) since the 1980s, relying on packet loss as the
signal to slow down. While this worked well for many years, loss-based
congestion control is unfortunately out-dated in today's networks. On
today's Internet, loss-based congestion control causes the infamous
bufferbloat problem, often causing seconds of needless queuing delay,
since it fills the bloated buffers in many last-mile links. On today's
high-speed long-haul links using commodity switches with shallow
buffers, loss-based congestion control has abysmal throughput because
it over-reacts to losses caused by transient traffic bursts.
In 1981 Kleinrock and Gale showed that the optimal operating point for
a network maximizes delivered bandwidth while minimizing delay and
loss, not only for single connections but for the network as a
whole. Finding that optimal operating point has been elusive, since
any single network measurement is ambiguous: network measurements are
the result of both bandwidth and propagation delay, and those two
cannot be measured simultaneously.
While it is impossible to disambiguate any single bandwidth or RTT
measurement, a connection's behavior over time tells a clearer
story. BBR uses a measurement strategy designed to resolve this
ambiguity. It combines these measurements with a robust servo loop
using recent control systems advances to implement a distributed
congestion control algorithm that reacts to actual congestion, not
packet loss or transient queue delay, and is designed to converge with
high probability to a point near the optimal operating point.
In a nutshell, BBR creates an explicit model of the network pipe by
sequentially probing the bottleneck bandwidth and RTT. On the arrival
of each ACK, BBR derives the current delivery rate of the last round
trip, and feeds it through a windowed max-filter to estimate the
bottleneck bandwidth. Conversely it uses a windowed min-filter to
estimate the round trip propagation delay. The max-filtered bandwidth
and min-filtered RTT estimates form BBR's model of the network pipe.
Using its model, BBR sets control parameters to govern sending
behavior. The primary control is the pacing rate: BBR applies a gain
multiplier to transmit faster or slower than the observed bottleneck
bandwidth. The conventional congestion window (cwnd) is now the
secondary control; the cwnd is set to a small multiple of the
estimated BDP (bandwidth-delay product) in order to allow full
utilization and bandwidth probing while bounding the potential amount
of queue at the bottleneck.
When a BBR connection starts, it enters STARTUP mode and applies a
high gain to perform an exponential search to quickly probe the
bottleneck bandwidth (doubling its sending rate each round trip, like
slow start). However, instead of continuing until it fills up the
buffer (i.e. a loss), or until delay or ACK spacing reaches some
threshold (like Hystart), it uses its model of the pipe to estimate
when that pipe is full: it estimates the pipe is full when it notices
the estimated bandwidth has stopped growing. At that point it exits
STARTUP and enters DRAIN mode, where it reduces its pacing rate to
drain the queue it estimates it has created.
Then BBR enters steady state. In steady state, PROBE_BW mode cycles
between first pacing faster to probe for more bandwidth, then pacing
slower to drain any queue that created if no more bandwidth was
available, and then cruising at the estimated bandwidth to utilize the
pipe without creating excess queue. Occasionally, on an as-needed
basis, it sends significantly slower to probe for RTT (PROBE_RTT
mode).
BBR has been fully deployed on Google's wide-area backbone networks
and we're experimenting with BBR on Google.com and YouTube on a global
scale. Replacing CUBIC with BBR has resulted in significant
improvements in network latency and application (RPC, browser, and
video) metrics. For more details please refer to our upcoming ACM
Queue publication.
Example performance results, to illustrate the difference between BBR
and CUBIC:
Resilience to random loss (e.g. from shallow buffers):
Consider a netperf TCP_STREAM test lasting 30 secs on an emulated
path with a 10Gbps bottleneck, 100ms RTT, and 1% packet loss
rate. CUBIC gets 3.27 Mbps, and BBR gets 9150 Mbps (2798x higher).
Low latency with the bloated buffers common in today's last-mile links:
Consider a netperf TCP_STREAM test lasting 120 secs on an emulated
path with a 10Mbps bottleneck, 40ms RTT, and 1000-packet bottleneck
buffer. Both fully utilize the bottleneck bandwidth, but BBR
achieves this with a median RTT 25x lower (43 ms instead of 1.09
secs).
Our long-term goal is to improve the congestion control algorithms
used on the Internet. We are hopeful that BBR can help advance the
efforts toward this goal, and motivate the community to do further
research.
Test results, performance evaluations, feedback, and BBR-related
discussions are very welcome in the public e-mail list for BBR:
https://groups.google.com/forum/#!forum/bbr-dev
NOTE: BBR *must* be used with the fq qdisc ("man tc-fq") with pacing
enabled, since pacing is integral to the BBR design and
implementation. BBR without pacing would not function properly, and
may incur unnecessary high packet loss rates.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit introduces an optional new "omnipotent" hook,
cong_control(), for congestion control modules. The cong_control()
function is called at the end of processing an ACK (i.e., after
updating sequence numbers, the SACK scoreboard, and loss
detection). At that moment we have precise delivery rate information
the congestion control module can use to control the sending behavior
(using cwnd, TSO skb size, and pacing rate) in any CA state.
This function can also be used by a congestion control that prefers
not to use the default cwnd reduction approach (i.e., the PRR
algorithm) during CA_Recovery to control the cwnd and sending rate
during loss recovery.
We take advantage of the fact that recent changes defer the
retransmission or transmission of new data (e.g. by F-RTO) in recovery
until the new tcp_cong_control() function is run.
With this commit, we only run tcp_update_pacing_rate() if the
congestion control is not using this new API. New congestion controls
which use the new API do not want the TCP stack to run the default
pacing rate calculation and overwrite whatever pacing rate they have
chosen at initialization time.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently the TCP send buffer expands to twice cwnd, in order to allow
limited transmits in the CA_Recovery state. This assumes that cwnd
does not increase in the CA_Recovery.
For some congestion control algorithms, like the upcoming BBR module,
if the losses in recovery do not indicate congestion then we may
continue to raise cwnd multiplicatively in recovery. In such cases the
current multiplier will falsely limit the sending rate, much as if it
were limited by the application.
This commit adds an optional congestion control callback to use a
different multiplier to expand the TCP send buffer. For congestion
control modules that do not specificy this callback, TCP continues to
use the previous default of 2.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Export tcp_mss_to_mtu(), so that congestion control modules can use
this to help calculate a pacing rate.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To allow congestion control modules to use the default TSO auto-sizing
algorithm as one of the ingredients in their own decision about TSO sizing:
1) Export tcp_tso_autosize() so that CC modules can use it.
2) Change tcp_tso_autosize() to allow callers to specify a minimum
number of segments per TSO skb, in case the congestion control
module has a different notion of the best floor for TSO skbs for
the connection right now. For very low-rate paths or policed
connections it can be appropriate to use smaller TSO skbs.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the tso_segs_goal() function in tcp_congestion_ops to allow the
congestion control module to specify the number of segments that
should be in a TSO skb sent by tcp_write_xmit() and
tcp_xmit_retransmit_queue(). The congestion control module can either
request a particular number of segments in TSO skb that we transmit,
or return 0 if it doesn't care.
This allows the upcoming BBR congestion control module to select small
TSO skb sizes if the module detects that the bottleneck bandwidth is
very low, or that the connection is policed to a low rate.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit export two new fields in struct tcp_info:
tcpi_delivery_rate: The most recent goodput, as measured by
tcp_rate_gen(). If the socket is limited by the sending
application (e.g., no data to send), it reports the highest
measurement instead of the most recent. The unit is bytes per
second (like other rate fields in tcp_info).
tcpi_delivery_rate_app_limited: A boolean indicating if the goodput
was measured when the socket's throughput was limited by the
sending application.
This delivery rate information can be useful for applications that
want to know the current throughput the TCP connection is seeing,
e.g. adaptive bitrate video streaming. It can also be very useful for
debugging or troubleshooting.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit adds code to track whether the delivery rate represented
by each rate_sample was limited by the application.
Upon each transmit, we store in the is_app_limited field in the skb a
boolean bit indicating whether there is a known "bubble in the pipe":
a point in the rate sample interval where the sender was
application-limited, and did not transmit even though the cwnd and
pacing rate allowed it.
This logic marks the flow app-limited on a write if *all* of the
following are true:
1) There is less than 1 MSS of unsent data in the write queue
available to transmit.
2) There is no packet in the sender's queues (e.g. in fq or the NIC
tx queue).
3) The connection is not limited by cwnd.
4) There are no lost packets to retransmit.
The tcp_rate_check_app_limited() code in tcp_rate.c determines whether
the connection is application-limited at the moment. If the flow is
application-limited, it sets the tp->app_limited field. If the flow is
application-limited then that means there is effectively a "bubble" of
silence in the pipe now, and this silence will be reflected in a lower
bandwidth sample for any rate samples from now until we get an ACK
indicating this bubble has exited the pipe: specifically, until we get
an ACK for the next packet we transmit.
When we send every skb we record in scb->tx.is_app_limited whether the
resulting rate sample will be application-limited.
The code in tcp_rate_gen() checks to see when it is safe to mark all
known application-limited bubbles of silence as having exited the
pipe. It does this by checking to see when the delivered count moves
past the tp->app_limited marker. At this point it zeroes the
tp->app_limited marker, as all known bubbles are out of the pipe.
We make room for the tx.is_app_limited bit in the skb by borrowing a
bit from the in_flight field used by NV to record the number of bytes
in flight. The receive window in the TCP header is 16 bits, and the
max receive window scaling shift factor is 14 (RFC 1323). So the max
receive window offered by the TCP protocol is 2^(16+14) = 2^30. So we
only need 30 bits for the tx.in_flight used by NV.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch generates data delivery rate (throughput) samples on a
per-ACK basis. These rate samples can be used by congestion control
modules, and specifically will be used by TCP BBR in later patches in
this series.
Key state:
tp->delivered: Tracks the total number of data packets (original or not)
delivered so far. This is an already-existing field.
tp->delivered_mstamp: the last time tp->delivered was updated.
Algorithm:
A rate sample is calculated as (d1 - d0)/(t1 - t0) on a per-ACK basis:
d1: the current tp->delivered after processing the ACK
t1: the current time after processing the ACK
d0: the prior tp->delivered when the acked skb was transmitted
t0: the prior tp->delivered_mstamp when the acked skb was transmitted
When an skb is transmitted, we snapshot d0 and t0 in its control
block in tcp_rate_skb_sent().
When an ACK arrives, it may SACK and ACK some skbs. For each SACKed
or ACKed skb, tcp_rate_skb_delivered() updates the rate_sample struct
to reflect the latest (d0, t0).
Finally, tcp_rate_gen() generates a rate sample by storing
(d1 - d0) in rs->delivered and (t1 - t0) in rs->interval_us.
One caveat: if an skb was sent with no packets in flight, then
tp->delivered_mstamp may be either invalid (if the connection is
starting) or outdated (if the connection was idle). In that case,
we'll re-stamp tp->delivered_mstamp.
At first glance it seems t0 should always be the time when an skb was
transmitted, but actually this could over-estimate the rate due to
phase mismatch between transmit and ACK events. To track the delivery
rate, we ensure that if packets are in flight then t0 and and t1 are
times at which packets were marked delivered.
If the initial and final RTTs are different then one may be corrupted
by some sort of noise. The noise we see most often is sending gaps
caused by delayed, compressed, or stretched acks. This either affects
both RTTs equally or artificially reduces the final RTT. We approach
this by recording the info we need to compute the initial RTT
(duration of the "send phase" of the window) when we recorded the
associated inflight. Then, for a filter to avoid bandwidth
overestimates, we generalize the per-sample bandwidth computation
from:
bw = delivered / ack_phase_rtt
to the following:
bw = delivered / max(send_phase_rtt, ack_phase_rtt)
In large-scale experiments, this filtering approach incorporating
send_phase_rtt is effective at avoiding bandwidth overestimates due to
ACK compression or stretched ACKs.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Count the number of packets that a TCP connection marks lost.
Congestion control modules can use this loss rate information for more
intelligent decisions about how fast to send.
Specifically, this is used in TCP BBR policer detection. BBR uses a
high packet loss rate as one signal in its policer detection and
policer bandwidth estimation algorithm.
The BBR policer detection algorithm cannot simply track retransmits,
because a retransmit can be (and often is) an indicator of packets
lost long, long ago. This is particularly true in a long CA_Loss
period that repairs the initial massive losses when a policer kicks
in.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Revert to the tcp_skb_cb size check that tcp_init() had before commit
b4772ef879 ("net: use common macro for assering skb->cb[] available
size in protocol families"). As related commit 744d5a3e9f ("net:
move skb->dropcount to skb->cb[]") explains, the
sock_skb_cb_check_size() mechanism was added to ensure that there is
space for dropcount, "for protocol families using it". But TCP is not
a protocol using dropcount, so tcp_init() doesn't need to provision
space for dropcount in the skb->cb[], and thus we can revert to the
older form of the tcp_skb_cb size check. Doing so allows TCP to use 4
more bytes of the skb->cb[] space.
Fixes: b4772ef879 ("net: use common macro for assering skb->cb[] available size in protocol families")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Refactor the TCP min_rtt code to reuse the new win_minmax library in
lib/win_minmax.c to simplify the TCP code.
This is a pure refactor: the functionality is exactly the same. We
just moved the windowed min code to make TCP easier to read and
maintain, and to allow other parts of the kernel to use the windowed
min/max filter code.
Signed-off-by: Van Jacobson <vanj@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Soheil Hassas Yeganeh <soheil@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The upcoming change "lib/win_minmax: windowed min or max estimator"
introduces a struct called minmax, which is then included in
include/linux/tcp.h in the upcoming change "tcp: use windowed min
filter library for TCP min_rtt estimation". This would create a
compilation error for tcp_cdg.c, which defines its own minmax
struct. To avoid this naming conflict (and potentially others in the
future), this commit renames the version used in tcp_cdg.c to
cdg_minmax.
Signed-off-by: Soheil Hassas Yeganeh <soheil@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>
Cc: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit 8a29111c7 ("net: gro: allow to build full sized skb")
gro may build buffers with a frag_list. This can hurt forwarding
because most NICs can't offload such packets, they need to be
segmented in software. This patch splits buffers with a frag_list
at the frag_list pointer into buffers that can be TSO offloaded.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Acked-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Similar to gre, vxlan, geneve tunnels allow IPIP tunnels to
operate in 'collect metadata' mode.
bpf_skb_[gs]et_tunnel_key() helpers can make use of it right away.
ovs can use it as well in the future (once appropriate ovs-vport
abstractions and user apis are added).
Note that just like in other tunnels we cannot cache the dst,
since tunnel_info metadata can be different for every packet.
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
With large BDP TCP flows and lossy networks, it is very important
to keep a low number of skbs in the write queue.
RACK and SACK processing can perform a linear scan of it.
We should avoid putting any payload in skb->head, so that SACK
shifting can be done if needed.
With this patch, we allow to pack ~0.5 MB per skb instead of
the 64KB initially cooked at tcp_sendmsg() time.
This gives a reduction of number of skbs in write queue by eight.
tcp_rack_detect_loss() likes this.
We still allow payload in skb->head for first skb put in the queue,
to not impact RPC workloads.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When skb replaces another one in ooo queue, I forgot to also
update tp->ooo_last_skb as well, if the replaced skb was the last one
in the queue.
To fix this, we simply can re-use the code that runs after an insertion,
trying to merge skbs at the right of current skb.
This not only fixes the bug, but also remove all small skbs that might
be a subset of the new one.
Example:
We receive segments 2001:3001, 4001:5001
Then we receive 2001:8001 : We should replace 2001:3001 with the big
skb, but also remove 4001:50001 from the queue to save space.
packetdrill test demonstrating the bug
0.000 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
+0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
+0 bind(3, ..., ...) = 0
+0 listen(3, 1) = 0
+0 < S 0:0(0) win 32792 <mss 1000,sackOK,nop,nop,nop,wscale 7>
+0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 7>
+0.100 < . 1:1(0) ack 1 win 1024
+0 accept(3, ..., ...) = 4
+0.01 < . 1001:2001(1000) ack 1 win 1024
+0 > . 1:1(0) ack 1 <nop,nop, sack 1001:2001>
+0.01 < . 1001:3001(2000) ack 1 win 1024
+0 > . 1:1(0) ack 1 <nop,nop, sack 1001:2001 1001:3001>
Fixes: 9f5afeae51 ("tcp: use an RB tree for ooo receive queue")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Yuchung Cheng <ycheng@google.com>
Cc: Yaogong Wang <wygivan@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/mediatek/mtk_eth_soc.c
drivers/net/ethernet/qlogic/qed/qed_dcbx.c
drivers/net/phy/Kconfig
All conflicts were cases of overlapping commits.
Signed-off-by: David S. Miller <davem@davemloft.net>
A previous patch added l3mdev flow update making these hooks
redundant. Remove them.
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Flip the IPv4 output path to use the l3mdev tx out hook. The VRF dst
is not returned on the first FIB lookup. Instead, the dst on the
skb is switched at the beginning of the IPv4 output processing to
send the packet to the VRF driver on xmit.
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allow an L3 master device to act as the loopback for that L3 domain.
For IPv4 the device can also have the address 127.0.0.1.
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the infrastructure to the output path to pass an skb
to an l3mdev device if it has a hook registered. This is the Tx parallel
to l3mdev_ip{6}_rcv in the receive path and is the basis for removing
the existing hook that returns the vrf dst on the fib lookup.
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add l3mdev hook to set FLOWI_FLAG_SKIP_NH_OIF flag and update oif/iif
in flow struct if its oif or iif points to a device enslaved to an L3
Master device. Only 1 needs to be converted to match the l3mdev FIB
rule. This moves the flow adjustment for l3mdev to a single point
catching all lookups. It is redundant for existing hooks (those are
removed in later patches) but is needed for missed lookups such as
PMTU updates.
Signed-off-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Willem noticed that we could avoid an rbtree lookup if the
the attempt to coalesce incoming skb to the last skb failed
for some reason.
Since most ooo additions are at the tail, this is definitely
worth adding a test and fast path.
Suggested-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yaogong Wang <wygivan@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
The IS_ENABLED() macro checks if a Kconfig symbol has been enabled either
built-in or as a module, use that macro instead of open coding the same.
Using the macro makes the code more readable by helping abstract away some
of the Kconfig built-in and module enable details.
Signed-off-by: Javier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add utility functions to convert a 32 bits key into a 64 bits tunnel and
vice versa.
These functions will be used instead of cloning code in GRE and VXLAN,
and in tc act_iptunnel which will be introduced in a following patch in
this patchset.
Signed-off-by: Amir Vadai <amir@vadai.me>
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.com>
Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Acked-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If skb has a valid l4 hash, there is no point clearing hash and force
a further flow dissection when a tunnel encapsulation is added.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
fib_table_insert() inconsistently fills the nlmsg_flags field in its
notification messages.
Since commit b8f5583135 ("[RTNETLINK]: Fix sending netlink message
when replace route."), the netlink message has its nlmsg_flags set to
NLM_F_REPLACE if the route replaced a preexisting one.
Then commit a2bb6d7d6f ("ipv4: include NLM_F_APPEND flag in append
route notifications") started setting nlmsg_flags to NLM_F_APPEND if
the route matched a preexisting one but was appended.
In other cases (exclusive creation or prepend), nlmsg_flags is 0.
This patch sets ->nlmsg_flags in all situations, preserving the
semantic of the NLM_F_* bits:
* NLM_F_CREATE: a new fib entry has been created for this route.
* NLM_F_EXCL: no other fib entry existed for this route.
* NLM_F_REPLACE: this route has overwritten a preexisting fib entry.
* NLM_F_APPEND: the new fib entry was added after other entries for
the same route.
As a result, the possible flag combination can now be reported
(iproute2's terminology into parentheses):
* NLM_F_CREATE | NLM_F_EXCL: route didn't exist, exclusive creation
("add").
* NLM_F_CREATE | NLM_F_APPEND: route did already exist, new route
added after preexisting ones ("append").
* NLM_F_CREATE: route did already exist, new route added before
preexisting ones ("prepend").
* NLM_F_REPLACE: route did already exist, new route replaced the
first preexisting one ("change").
Signed-off-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit f02db315b8 ("ipv4: IP_TOS and IP_TTL can be specified as
ancillary data") Francesco added IP_TOS values specified as integer.
However, kernel sends to userspace (at recvmsg() time) an IP_TOS value
in a single byte, when IP_RECVTOS is set on the socket.
It can be very useful to reflect all ancillary options as given by the
kernel in a subsequent sendmsg(), instead of aborting the sendmsg() with
EINVAL after Francesco patch.
So this patch extends IP_TOS ancillary to accept an u8, so that an UDP
server can simply reuse same ancillary block without having to mangle
it.
Jesper can then augment
https://github.com/netoptimizer/network-testing/blob/master/src/udp_example02.c
to add TOS reflection ;)
Fixes: f02db315b8 ("ipv4: IP_TOS and IP_TTL can be specified as ancillary data")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Francesco Fusco <ffusco@redhat.com>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Over the years, TCP BDP has increased by several orders of magnitude,
and some people are considering to reach the 2 Gbytes limit.
Even with current window scale limit of 14, ~1 Gbytes maps to ~740,000
MSS.
In presence of packet losses (or reorders), TCP stores incoming packets
into an out of order queue, and number of skbs sitting there waiting for
the missing packets to be received can be in the 10^5 range.
Most packets are appended to the tail of this queue, and when
packets can finally be transferred to receive queue, we scan the queue
from its head.
However, in presence of heavy losses, we might have to find an arbitrary
point in this queue, involving a linear scan for every incoming packet,
throwing away cpu caches.
This patch converts it to a RB tree, to get bounded latencies.
Yaogong wrote a preliminary patch about 2 years ago.
Eric did the rebase, added ofo_last_skb cache, polishing and tests.
Tested with network dropping between 1 and 10 % packets, with good
success (about 30 % increase of throughput in stress tests)
Next step would be to also use an RB tree for the write queue at sender
side ;)
Signed-off-by: Yaogong Wang <wygivan@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-By: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 76174004a0
(tcp: do not slow start when cwnd equals ssthresh )
introduced regression in TCP YeAH. Using 100ms delay 1% loss virtual
ethernet link kernel 4.2 shows bandwidth ~500KB/s for single TCP
connection and kernel 4.3 and above (including 4.8-rc4) shows bandwidth
~100KB/s.
That is caused by stalled cwnd when cwnd equals ssthresh. This patch
fixes it by proper increasing cwnd in this case.
Signed-off-by: Artem Germanov <agermanov@anchorfree.com>
Acked-by: Dmitry Adamushko <d.adamushko@anchorfree.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This adds the capability for a process that has CAP_NET_ADMIN on
a socket to see the socket mark in socket dumps.
Commit a52e95abf7 ("net: diag: allow socket bytecode filters to
match socket marks") recently gave privileged processes the
ability to filter socket dumps based on mark. This patch is
complementary: it ensures that the mark is also passed to
userspace in the socket's netlink attributes. It is useful for
tools like ss which display information about sockets.
Tested: https://android-review.googlesource.com/270210
Signed-off-by: Lorenzo Colitti <lorenzo@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When DATA and/or FIN are carried in a SYN/ACK message or SYN message,
we append an skb in socket receive queue, but we forget to call
sk_forced_mem_schedule().
Effect is that the socket has a negative sk->sk_forward_alloc as long as
the message is not read by the application.
Josh Hunt fixed a similar issue in commit d22e153718 ("tcp: fix tcp
fin memory accounting")
Fixes: 168a8f5805 ("tcp: TCP Fast Open Server - main code path")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Josh Hunt <johunt@akamai.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Steffen Klassert says:
====================
ipsec 2016-09-08
1) Fix a crash when xfrm_dump_sa returns an error.
From Vegard Nossum.
2) Remove some incorrect WARN() on normal error handling.
From Vegard Nossum.
3) Ignore socket policies when rebuilding hash tables,
socket policies are not inserted into the hash tables.
From Tobias Brunner.
4) Initialize and check tunnel pointers properly before
we use it. From Alexey Kodanev.
5) Fix l3mdev oif setting on xfrm dst lookups.
From David Ahern.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
udp_diag_destroy does look up the IPv4 UDP hashtable for mapped
addresses, but it gets the IPv4 address to look up from the
beginning of the IPv6 address instead of the end.
Tested: https://android-review.googlesource.com/269874
Fixes: 5d77dca828 ("net: diag: support SOCK_DESTROY for UDP sockets")
Signed-off-by: Lorenzo Colitti <lorenzo@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When deleting an IP address from an interface, there is a clean-up of
routes which refer to this local address. However, there was no check to
see that the VRF matched. This meant that deletion wasn't confined to
the VRF it should have been.
To solve this, a new field has been added to fib_info to hold a table
id. When removing fib entries corresponding to a local ip address, this
table id is also used in the comparison.
The table id is populated when the fib_info is created. This was already
done in some places, but not in ip_rt_ioctl(). This has now been fixed.
Fixes: 021dd3b8a1 ("net: Add routes to the table associated with the device")
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Tested-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: Mark Tomlinson <mark.tomlinson@alliedtelesis.co.nz>
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 your net-next
tree. Most relevant updates are the removal of per-conntrack timers to
use a workqueue/garbage collection approach instead from Florian
Westphal, the hash and numgen expression for nf_tables from Laura
Garcia, updates on nf_tables hash set to honor the NLM_F_EXCL flag,
removal of ip_conntrack sysctl and many other incremental updates on our
Netfilter codebase.
More specifically, they are:
1) Retrieve only 4 bytes to fetch ports in case of non-linear skb
transport area in dccp, sctp, tcp, udp and udplite protocol
conntrackers, from Gao Feng.
2) Missing whitespace on error message in physdev match, from Hangbin Liu.
3) Skip redundant IPv4 checksum calculation in nf_dup_ipv4, from Liping Zhang.
4) Add nf_ct_expires() helper function and use it, from Florian Westphal.
5) Replace opencoded nf_ct_kill() call in IPVS conntrack support, also
from Florian.
6) Rename nf_tables set implementation to nft_set_{name}.c
7) Introduce the hash expression to allow arbitrary hashing of selector
concatenations, from Laura Garcia Liebana.
8) Remove ip_conntrack sysctl backward compatibility code, this code has
been around for long time already, and we have two interfaces to do
this already: nf_conntrack sysctl and ctnetlink.
9) Use nf_conntrack_get_ht() helper function whenever possible, instead
of opencoding fetch of hashtable pointer and size, patch from Liping Zhang.
10) Add quota expression for nf_tables.
11) Add number generator expression for nf_tables, this supports
incremental and random generators that can be combined with maps,
very useful for load balancing purpose, again from Laura Garcia Liebana.
12) Fix a typo in a debug message in FTP conntrack helper, from Colin Ian King.
13) Introduce a nft_chain_parse_hook() helper function to parse chain hook
configuration, this is used by a follow up patch to perform better chain
update validation.
14) Add rhashtable_lookup_get_insert_key() to rhashtable and use it from the
nft_set_hash implementation to honor the NLM_F_EXCL flag.
15) Missing nulls check in nf_conntrack from nf_conntrack_tuple_taken(),
patch from Florian Westphal.
16) Don't use the DYING bit to know if the conntrack event has been already
delivered, instead a state variable to track event re-delivery
states, also from Florian.
17) Remove the per-conntrack timer, use the workqueue approach that was
discussed during the NFWS, from Florian Westphal.
18) Use the netlink conntrack table dump path to kill stale entries,
again from Florian.
19) Add a garbage collector to get rid of stale conntracks, from
Florian.
20) Reschedule garbage collector if eviction rate is high.
21) Get rid of the __nf_ct_kill_acct() helper.
22) Use ARPHRD_ETHER instead of hardcoded 1 from ARP logger.
23) Make nf_log_set() interface assertive on unsupported families.
====================
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>
All changes are notified, but the initial state was missing.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.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>
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>
When TCP operates in lossy environments (between 1 and 10 % packet
losses), many SACK blocks can be exchanged, and I noticed we could
drop them on busy senders, if these SACK blocks have to be queued
into the socket backlog.
While the main cause is the poor performance of RACK/SACK processing,
we can try to avoid these drops of valuable information that can lead to
spurious timeouts and retransmits.
Cause of the drops is the skb->truesize overestimation caused by :
- drivers allocating ~2048 (or more) bytes as a fragment to hold an
Ethernet frame.
- various pskb_may_pull() calls bringing the headers into skb->head
might have pulled all the frame content, but skb->truesize could
not be lowered, as the stack has no idea of each fragment truesize.
The backlog drops are also more visible on bidirectional flows, since
their sk_rmem_alloc can be quite big.
Let's add some room for the backlog, as only the socket owner
can selectively take action to lower memory needs, like collapsing
receive queues or partial ofo pruning.
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>
In inet_stream_ops we set read_sock to tcp_read_sock and peek_len to
tcp_peek_len (which is just a stub function that calls tcp_inq).
Signed-off-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adds SNMP counter for drops caused by MD5 mismatches.
The current syslog might help, but a counter is more precise and helps
monitoring.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
TCP MD5 mismatches do increment sk_drops counter in all states but
SYN_RECV.
This is very unlikely to happen in the real world, but worth adding
to help diagnostics.
We increase the parent (listener) sk_drops.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
