Dmitry provided a syzkaller (http://github.com/google/syzkaller)
generated program that triggers the WARNING at
net/ipv4/tcp.c:1729 in tcp_recvmsg() :
WARN_ON(tp->copied_seq != tp->rcv_nxt &&
!(flags & (MSG_PEEK | MSG_TRUNC)));
His program is specifically attempting a Cross SYN TCP exchange,
that we support (for the pleasure of hackers ?), but it looks we
lack proper tcp->copied_seq initialization.
Thanks again Dmitry for your report and testings.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Tested-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_send_rcvq() is used for re-injecting data into tcp receive queue.
Problems :
- No check against size is performed, allowed user to fool kernel in
attempting very large memory allocations, eventually triggering
OOM when memory is fragmented.
- In case of fault during the copy we do not return correct errno.
Lets use alloc_skb_with_frags() to cook optimal skbs.
Fixes: 292e8d8c85 ("tcp: Move rcvq sending to tcp_input.c")
Fixes: c0e88ff0f2 ("tcp: Repair socket queues")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Pavel Emelyanov <xemul@parallels.com>
Acked-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch implements the second half of RACK that uses the the most
recent transmit time among all delivered packets to detect losses.
tcp_rack_mark_lost() is called upon receiving a dubious ACK.
It then checks if an not-yet-sacked packet was sent at least
"reo_wnd" prior to the sent time of the most recently delivered.
If so the packet is deemed lost.
The "reo_wnd" reordering window starts with 1msec for fast loss
detection and changes to min-RTT/4 when reordering is observed.
We found 1msec accommodates well on tiny degree of reordering
(<3 pkts) on faster links. We use min-RTT instead of SRTT because
reordering is more of a path property but SRTT can be inflated by
self-inflicated congestion. The factor of 4 is borrowed from the
delayed early retransmit and seems to work reasonably well.
Since RACK is still experimental, it is now used as a supplemental
loss detection on top of existing algorithms. It is only effective
after the fast recovery starts or after the timeout occurs. The
fast recovery is still triggered by FACK and/or dupack threshold
instead of RACK.
We introduce a new sysctl net.ipv4.tcp_recovery for future
experiments of loss recoveries. For now RACK can be disabled by
setting it to 0.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is the first half of the RACK loss recovery.
RACK loss recovery uses the notion of time instead
of packet sequence (FACK) or counts (dupthresh). It's inspired by the
previous FACK heuristic in tcp_mark_lost_retrans(): when a limited
transmit (new data packet) is sacked, then current retransmitted
sequence below the newly sacked sequence must been lost,
since at least one round trip time has elapsed.
But it has several limitations:
1) can't detect tail drops since it depends on limited transmit
2) is disabled upon reordering (assumes no reordering)
3) only enabled in fast recovery ut not timeout recovery
RACK (Recently ACK) addresses these limitations with the notion
of time instead: a packet P1 is lost if a later packet P2 is s/acked,
as at least one round trip has passed.
Since RACK cares about the time sequence instead of the data sequence
of packets, it can detect tail drops when later retransmission is
s/acked while FACK or dupthresh can't. For reordering RACK uses a
dynamically adjusted reordering window ("reo_wnd") to reduce false
positives on ever (small) degree of reordering.
This patch implements tcp_advanced_rack() which tracks the
most recent transmission time among the packets that have been
delivered (ACKed or SACKed) in tp->rack.mstamp. This timestamp
is the key to determine which packet has been lost.
Consider an example that the sender sends six packets:
T1: P1 (lost)
T2: P2
T3: P3
T4: P4
T100: sack of P2. rack.mstamp = T2
T101: retransmit P1
T102: sack of P2,P3,P4. rack.mstamp = T4
T205: ACK of P4 since the hole is repaired. rack.mstamp = T101
We need to be careful about spurious retransmission because it may
falsely advance tp->rack.mstamp by an RTT or an RTO, causing RACK
to falsely mark all packets lost, just like a spurious timeout.
We identify spurious retransmission by the ACK's TS echo value.
If TS option is not applicable but the retransmission is acknowledged
less than min-RTT ago, it is likely to be spurious. We refrain from
using the transmission time of these spurious retransmissions.
The second half is implemented in the next patch that marks packet
lost using RACK timestamp.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
a helper to prepare the main RACK patch
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove the existing lost retransmit detection because RACK subsumes
it completely. This also stops the overloading the ack_seq field of
the skb control block.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Kathleen Nichols' algorithm for tracking the minimum RTT of a
data stream over some measurement window. It uses constant space
and constant time per update. Yet it almost always delivers
the same minimum as an implementation that has to keep all
the data in the window. The measurement window is tunable via
sysctl.net.ipv4.tcp_min_rtt_wlen with a default value of 5 minutes.
The algorithm keeps track of the best, 2nd best & 3rd best min
values, maintaining an invariant that the measurement time of
the n'th best >= n-1'th best. It also makes sure that the three
values are widely separated in the time window since that bounds
the worse case error when that data is monotonically increasing
over the window.
Upon getting a new min, we can forget everything earlier because
it has no value - the new min is less than everything else in the
window by definition and it's the most recent. So we restart fresh
on every new min and overwrites the 2nd & 3rd choices. The same
property holds for the 2nd & 3rd best.
Therefore we have to maintain two invariants to maximize the
information in the samples, one on values (1st.v <= 2nd.v <=
3rd.v) and the other on times (now-win <=1st.t <= 2nd.t <= 3rd.t <=
now). These invariants determine the structure of the code
The RTT input to the windowed filter is the minimum RTT measured
from ACK or SACK, or as the last resort from TCP timestamps.
The accessor tcp_min_rtt() returns the minimum RTT seen in the
window. ~0U indicates it is not available. The minimum is 1usec
even if the true RTT is below that.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently ca_seq_rtt_us does not use Kern's check. Fix that by
checking if any packet acked is a retransmit, for both RTT used
for RTT estimation and congestion control.
Fixes: 5b08e47ca ("tcp: prefer packet timing to TS-ECR for RTT")
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
At the time of commit fff3269907 ("tcp: reflect SYN queue_mapping into
SYNACK packets") we had little ways to cope with SYN floods.
We no longer need to reflect incoming skb queue mappings, and instead
can pick a TX queue based on cpu cooking the SYNACK, with normal XPS
affinities.
Note that all SYNACK retransmits were picking TX queue 0, this no longer
is a win given that SYNACK rtx are now distributed on all cpus.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
One 32bit hole is following skc_refcnt, use it.
skc_incoming_cpu can also be an union for request_sock rcv_wnd.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
inet_reqsk_alloc() is used to allocate a temporary request
in order to generate a SYNACK with a cookie. Then later,
syncookie validation also uses a temporary request.
These paths already took a reference on listener refcount,
we can avoid a couple of atomic operations.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are multiple races that need fixes :
1) skb_get() + queue skb + kfree_skb() is racy
An accept() can be done on another cpu, data consumed immediately.
tcp_recvmsg() uses __kfree_skb() as it is assumed all skb found in
socket receive queue are private.
Then the kfree_skb() in tcp_rcv_state_process() uses an already freed skb
2) tcp_reqsk_record_syn() needs to be done before tcp_try_fastopen()
for the same reasons.
3) We want to send the SYNACK before queueing child into accept queue,
otherwise we might reintroduce the ooo issue fixed in
commit 7c85af8810 ("tcp: avoid reorders for TFO passive connections")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If a listen backlog is very big (to avoid syncookies), then
the listener sk->sk_wmem_alloc is the main source of false
sharing, as we need to touch it twice per SYNACK re-transmit
and TX completion.
(One SYN packet takes listener lock once, but up to 6 SYNACK
are generated)
By attaching the skb to the request socket, we remove this
source of contention.
Tested:
listen(fd, 10485760); // single listener (no SO_REUSEPORT)
16 RX/TX queue NIC
Sustain a SYNFLOOD attack of ~320,000 SYN per second,
Sending ~1,400,000 SYNACK per second.
Perf profiles now show listener spinlock being next bottleneck.
20.29% [kernel] [k] queued_spin_lock_slowpath
10.06% [kernel] [k] __inet_lookup_established
5.12% [kernel] [k] reqsk_timer_handler
3.22% [kernel] [k] get_next_timer_interrupt
3.00% [kernel] [k] tcp_make_synack
2.77% [kernel] [k] ipt_do_table
2.70% [kernel] [k] run_timer_softirq
2.50% [kernel] [k] ip_finish_output
2.04% [kernel] [k] cascade
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In this patch, we insert request sockets into TCP/DCCP
regular ehash table (where ESTABLISHED and TIMEWAIT sockets
are) instead of using the per listener hash table.
ACK packets find SYN_RECV pseudo sockets without having
to find and lock the listener.
In nominal conditions, this halves pressure on listener lock.
Note that this will allow for SO_REUSEPORT refinements,
so that we can select a listener using cpu/numa affinities instead
of the prior 'consistent hash', since only SYN packets will
apply this selection logic.
We will shrink listen_sock in the following patch to ease
code review.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Ying Cai <ycai@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
long term plan is to remove struct listen_sock when its hash
table is no longer there.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_syn_flood_action() will soon be called with unlocked socket.
In order to avoid SYN flood warning being emitted multiple times,
use xchg().
Extend max_qlen_log and synflood_warned fields in struct listen_sock
to u32
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Factorize code to get tcp header from skb. It makes no sense
to duplicate code in callers.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Once we realize tcp_rcv_synsent_state_process() does not use
its 'len' argument and we get rid of it, then it becomes clear
this argument is no longer used in tcp_rcv_state_process()
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We found that a TCP Fast Open passive connection was vulnerable
to reorders, as the exchange might look like
[1] C -> S S <FO ...> <request>
[2] S -> C S. ack request <options>
[3] S -> C . <answer>
packets [2] and [3] can be generated at almost the same time.
If C receives the 3rd packet before the 2nd, it will drop it as
the socket is in SYN_SENT state and expects a SYNACK.
S will have to retransmit the answer.
Current OOO avoidance in linux is defeated because SYNACK
packets are attached to the LISTEN socket, while DATA packets
are attached to the children. They might be sent by different cpus,
and different TX queues might be selected.
It turns out that for TFO, we created a child, which is a
full blown socket in TCP_SYN_RECV state, and we simply can attach
the SYNACK packet to this socket.
This means that at the time tcp_sendmsg() pushes DATA packet,
skb->ooo_okay will be set iff the SYNACK packet had been sent
and TX completed.
This removes the reorder source at the host level.
We also removed the export of tcp_try_fastopen(), as it is no
longer called from IPv6.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently SYN/ACK RTT is measured in jiffies. For LAN the SYN/ACK
RTT is often measured as 0ms or sometimes 1ms, which would affect
RTT estimation and min RTT samping used by some congestion control.
This patch improves SYN/ACK RTT to be usec resolution if platform
supports it. While the timestamping of SYN/ACK is done in request
sock, the RTT measurement is carefully arranged to avoid storing
another u64 timestamp in tcp_sock.
For regular handshake w/o SYNACK retransmission, the RTT is sampled
right after the child socket is created and right before the request
sock is released (tcp_check_req() in tcp_minisocks.c)
For Fast Open the child socket is already created when SYN/ACK was
sent, the RTT is sampled in tcp_rcv_state_process() after processing
the final ACK an right before the request socket is released.
If the SYN/ACK was retransmistted or SYN-cookie was used, we rely
on TCP timestamps to measure the RTT. The sample is taken at the
same place in tcp_rcv_state_process() after the timestamp values
are validated in tcp_validate_incoming(). Note that we do not store
TS echo value in request_sock for SYN-cookies, because the value
is already stored in tp->rx_opt used by tcp_ack_update_rtt().
One side benefit is that the RTT measurement now happens before
initializing congestion control (of the passive side). Therefore
the congestion control can use the SYN/ACK RTT.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit b73c3d0e4f ("net: Save TX flow hash in sock and set in skbuf
on xmit"), Tom provided a l4 hash to most outgoing TCP packets.
We'd like to provide one as well for SYNACK packets, so that all packets
of a given flow share same txhash, to later enable bonding driver to
also use skb->hash to perform slave selection.
Note that a SYNACK retransmit shuffles the tx hash, as Tom did
in commit 265f94ff54 ("net: Recompute sk_txhash on negative routing
advice") for established sockets.
This has nice effect making TCP flows resilient to some kind of black
holes, even at connection establish phase.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tom Herbert <tom@herbertland.com>
Cc: Mahesh Bandewar <maheshb@google.com>
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the following case doesn't use DCTCP, even if it should:
A responder has f.e. Cubic as system wide default, but for a specific
route to the initiating host, DCTCP is being set in RTAX_CC_ALGO. The
initiating host then uses DCTCP as congestion control, but since the
initiator sets ECT(0), tcp_ecn_create_request() doesn't set ecn_ok,
and we have to fall back to Reno after 3WHS completes.
We were thinking on how to solve this in a minimal, non-intrusive
way without bloating tcp_ecn_create_request() needlessly: lets cache
the CA ecn option flag in RTAX_FEATURES. In other words, when ECT(0)
is set on the SYN packet, set ecn_ok=1 iff route RTAX_FEATURES
contains the unexposed (internal-only) DST_FEATURE_ECN_CA. This allows
to only do a single metric feature lookup inside tcp_ecn_create_request().
Joint work with Florian Westphal.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
When TCP pacing was added back in linux-3.12, we chose
to apply a fixed ratio of 200 % against current rate,
to allow probing for optimal throughput even during
slow start phase, where cwnd can be doubled every other gRTT.
At Google, we found it was better applying a different ratio
while in Congestion Avoidance phase.
This ratio was set to 120 %.
We've used the normal tcp_in_slow_start() helper for a while,
then tuned the condition to select the conservative ratio
as soon as cwnd >= ssthresh/2 :
- After cwnd reduction, it is safer to ramp up more slowly,
as we approach optimal cwnd.
- Initial ramp up (ssthresh == INFINITY) still allows doubling
cwnd every other RTT.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
slow start after idle might reduce cwnd, but we perform this
after first packet was cooked and sent.
With TSO/GSO, it means that we might send a full TSO packet
even if cwnd should have been reduced to IW10.
Moving the SSAI check in skb_entail() makes sense, because
we slightly reduce number of times this check is done,
especially for large send() and TCP Small queue callbacks from
softirq context.
As Neal pointed out, we also need to perform the check
if/when receive window opens.
Tested:
Following packetdrill test demonstrates the problem
// Test of slow start after idle
`sysctl -q net.ipv4.tcp_slow_start_after_idle=1`
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 65535 <mss 1000,sackOK,nop,nop,nop,wscale 7>
+0 > S. 0:0(0) ack 1 <mss 1460,nop,nop,sackOK,nop,wscale 6>
+.100 < . 1:1(0) ack 1 win 511
+0 accept(3, ..., ...) = 4
+0 setsockopt(4, SOL_SOCKET, SO_SNDBUF, [200000], 4) = 0
+0 write(4, ..., 26000) = 26000
+0 > . 1:5001(5000) ack 1
+0 > . 5001:10001(5000) ack 1
+0 %{ assert tcpi_snd_cwnd == 10 }%
+.100 < . 1:1(0) ack 10001 win 511
+0 %{ assert tcpi_snd_cwnd == 20, tcpi_snd_cwnd }%
+0 > . 10001:20001(10000) ack 1
+0 > P. 20001:26001(6000) ack 1
+.100 < . 1:1(0) ack 26001 win 511
+0 %{ assert tcpi_snd_cwnd == 36, tcpi_snd_cwnd }%
+4 write(4, ..., 20000) = 20000
// If slow start after idle works properly, we should send 5 MSS here (cwnd/2)
+0 > . 26001:31001(5000) ack 1
+0 %{ assert tcpi_snd_cwnd == 10, tcpi_snd_cwnd }%
+0 > . 31001:36001(5000) ack 1
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
net/bridge/br_mdb.c
br_mdb.c conflict was a function call being removed to fix a bug in
'net' but whose signature was changed in 'net-next'.
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently F-RTO may repeatedly send new data packets on non-recurring
timeouts in CA_Loss mode. This is a bug because F-RTO (RFC5682)
should only be used on either new recovery or recurring timeouts.
This exacerbates the recovery progress during frequent timeout &
repair, because we prioritize sending new data packets instead of
repairing the holes when the bandwidth is already scarce.
Fix it by correcting the test of a new recovery episode.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The congestion state and cwnd can be updated in the wrong order.
For example, upon receiving a dubious ACK, we incorrectly raise
the cwnd first (tcp_may_raise_cwnd()/tcp_cong_avoid()) because
the state is still Open, then enter recovery state to reduce cwnd.
For another example, if the ACK indicates spurious timeout or
retransmits, we first revert the cwnd reduction and congestion
state back to Open state. But we don't raise the cwnd even though
the ACK does not indicate any congestion.
To fix this problem we should first call tcp_fastretrans_alert() to
process the dubious ACK and update the congestion state, then call
tcp_may_raise_cwnd() that raises cwnd based on the current state.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
V1 of this patch contains Eric Dumazet's suggestion to move the per
dst RTAX_QUICKACK check into tcp_in_quickack_mode(). Thanks Eric.
I ran some tests and after setting the "ip route change quickack 1"
knob there were still many delayed ACKs sent. This occured
because when icsk_ack.quick=0 the !icsk_ack.pingpong value is
subsequently ignored as tcp_in_quickack_mode() checks both these
values. The condition for a quick ack to trigger requires
that both icsk_ack.quick != 0 and icsk_ack.pingpong=0. Currently
only icsk_ack.pingpong is controlled by the knob. But the
icsk_ack.quick value changes dynamically depending on heuristics.
The crux of the matter is that delayed acks still cannot be entirely
disabled even with the RTAX_QUICKACK per dst knob enabled. This
patch ensures that a quick ack is always sent when the RTAX_QUICKACK
per dst knob is turned on.
The "ip route change quickack 1" knob was recently added to enable
quickacks. It was modeled around the TCP_QUICKACK setsockopt() option.
This issue is that even with "ip route change quickack 1" enabled
we still see delayed ACKs under some conditions. It would be nice
to be able to completely disable delayed ACKs.
Here is an example:
# netstat -s|grep dela
3 delayed acks sent
For all routes enable the knob
# ip route change quickack 1
Generate some traffic across a slow link and we still see the delayed
acks.
# netstat -s|grep dela
106 delayed acks sent
1 delayed acks further delayed because of locked socket
The issue is that both the "ip route change quickack 1" knob and
the TCP_QUICKACK option set the icsk_ack.pingpong variable to 0.
However at the business end in the __tcp_ack_snd_check() routine,
tcp_in_quickack_mode() checks that both icsk_ack.quick != 0
and icsk_ack.pingpong=0 in order to trigger a quickack. As
icsk_ack.quick is determined by heuristics it can be 0. When
that occurs the icsk_ack.pingpong value is ignored and a delayed
ACK is sent regardless.
This patch moves the RTAX_QUICKACK per dst check into the
tcp_in_quickack_mode() routine which ensures that a quickack is
always sent when the quickack knob is enabled for that dst.
Signed-off-by: Jon Maxwell <jmaxwell37@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
PRR slow start is often too aggressive especially when drops are
caused by traffic policers. The policers mainly use token bucket
to enforce the rate so sending (twice) faster than the delivery
rate causes excessive drops.
This patch changes PRR to the conservative reduction bound
(CRB) mode in RFC 6937 by default. CRB follows the packet
conservation rule to send at most the delivery rate by default.
But if many packets are lost and the pipe is empty, CRB may take N
round trips to repair N losses. We conditionally turn on slow start
mode if all these conditions are made to speed up the recovery:
1) on the second round or later in recovery
2) retransmission sent in the previous round is delivered on this ACK
3) no retransmission is marked lost on this ACK
By using packet conservation by default, this change reduces the loss
retransmits signicantly on networks that deploy traffic policers,
up to 20% reduction of overall loss rate.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If the retransmission in CA_Loss is lost again, we should not
continue to slow start or raise cwnd in congestion avoidance mode.
Instead we should enter fast recovery and use PRR to reduce cwnd,
following the principle in RFC5681:
"... or the loss of a retransmission, should be taken as two
indications of congestion and, therefore, cwnd (and ssthresh) MUST
be lowered twice in this case."
This is especially important to reduce loss when the CA_Loss
state was caused by a traffic policer dropping the entire inflight.
The CA_Loss state has a problem where a loss of L packets causes the
sender to send a burst of L packets. So a policer that's dropping
most packets in a given RTT can cause a huge retransmit storm. By
contrast, PRR includes logic to bound the number of outbound packets
that result from a given ACK. So switching to CA_Recovery on lost
retransmits in CA_Loss avoids this retransmit storm problem when
in CA_Loss.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit cd7d8498c9 ("tcp: change tcp_skb_pcount() location") we stored
gso_segs in a temporary cache hot location.
This patch does the same for gso_size.
This allows to save 2 cache line misses in tcp xmit path for
the last packet that is considered but not sent because of
various conditions (cwnd, tso defer, receiver window, TSQ...)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Our goal is to touch skb_shinfo(skb) only when absolutely needed,
to avoid two cache line misses in TCP output path for last skb
that is considered but not sent because of various conditions
(cwnd, tso defer, receiver window, TSQ...)
A packet is GSO only when skb_shinfo(skb)->gso_size is not zero.
We can set skb_shinfo(skb)->gso_type to sk->sk_gso_type even for
non GSO packets.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Upcoming tcp_cdg uses tcp_enter_cwr() to initiate PRR. Export this
function so that CDG can be compiled as a module.
Cc: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: David Hayes <davihay@ifi.uio.no>
Cc: Andreas Petlund <apetlund@simula.no>
Cc: Dave Taht <dave.taht@bufferbloat.net>
Cc: Nicolas Kuhn <nicolas.kuhn@telecom-bretagne.eu>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/cadence/macb.c
drivers/net/phy/phy.c
include/linux/skbuff.h
net/ipv4/tcp.c
net/switchdev/switchdev.c
Switchdev was a case of RTNH_H_{EXTERNAL --> OFFLOAD}
renaming overlapping with net-next changes of various
sorts.
phy.c was a case of two changes, one adding a local
variable to a function whilst the second was removing
one.
tcp.c overlapped a deadlock fix with the addition of new tcp_info
statistic values.
macb.c involved the addition of two zyncq device entries.
skbuff.h involved adding back ipv4_daddr to nf_bridge_info
whilst net-next changes put two other existing members of
that struct into a union.
Signed-off-by: David S. Miller <davem@davemloft.net>
Taking socket spinlock in tcp_get_info() can deadlock, as
inet_diag_dump_icsk() holds the &hashinfo->ehash_locks[i],
while packet processing can use the reverse locking order.
We could avoid this locking for TCP_LISTEN states, but lockdep would
certainly get confused as all TCP sockets share same lockdep classes.
[ 523.722504] ======================================================
[ 523.728706] [ INFO: possible circular locking dependency detected ]
[ 523.734990] 4.1.0-dbg-DEV #1676 Not tainted
[ 523.739202] -------------------------------------------------------
[ 523.745474] ss/18032 is trying to acquire lock:
[ 523.750002] (slock-AF_INET){+.-...}, at: [<ffffffff81669d44>] tcp_get_info+0x2c4/0x360
[ 523.758129]
[ 523.758129] but task is already holding lock:
[ 523.763968] (&(&hashinfo->ehash_locks[i])->rlock){+.-...}, at: [<ffffffff816bcb75>] inet_diag_dump_icsk+0x1d5/0x6c0
[ 523.774661]
[ 523.774661] which lock already depends on the new lock.
[ 523.774661]
[ 523.782850]
[ 523.782850] the existing dependency chain (in reverse order) is:
[ 523.790326]
-> #1 (&(&hashinfo->ehash_locks[i])->rlock){+.-...}:
[ 523.796599] [<ffffffff811126bb>] lock_acquire+0xbb/0x270
[ 523.802565] [<ffffffff816f5868>] _raw_spin_lock+0x38/0x50
[ 523.808628] [<ffffffff81665af8>] __inet_hash_nolisten+0x78/0x110
[ 523.815273] [<ffffffff816819db>] tcp_v4_syn_recv_sock+0x24b/0x350
[ 523.822067] [<ffffffff81684d41>] tcp_check_req+0x3c1/0x500
[ 523.828199] [<ffffffff81682d09>] tcp_v4_do_rcv+0x239/0x3d0
[ 523.834331] [<ffffffff816842fe>] tcp_v4_rcv+0xa8e/0xc10
[ 523.840202] [<ffffffff81658fa3>] ip_local_deliver_finish+0x133/0x3e0
[ 523.847214] [<ffffffff81659a9a>] ip_local_deliver+0xaa/0xc0
[ 523.853440] [<ffffffff816593b8>] ip_rcv_finish+0x168/0x5c0
[ 523.859624] [<ffffffff81659db7>] ip_rcv+0x307/0x420
Lets use u64_sync infrastructure instead. As a bonus, 64bit
arches get optimized, as these are nop for them.
Fixes: 0df48c26d8 ("tcp: add tcpi_bytes_acked to tcp_info")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
After sending the new data packets to probe (step 2), F-RTO may
incorrectly send more probes if the next ACK advances SND_UNA and
does not sack new packet. However F-RTO RFC 5682 probes at most
once. This bug may cause sender to always send new data instead of
repairing holes, inducing longer HoL blocking on the receiver for
the application.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Undo based on TCP timestamps should only happen on ACKs that advance
SND_UNA, according to the Eifel algorithm in RFC 3522:
Section 3.2:
(4) If the value of the Timestamp Echo Reply field of the
acceptable ACK's Timestamps option is smaller than the
value of RetransmitTS, then proceed to step (5),
Section Terminology:
We use the term 'acceptable ACK' as defined in [RFC793]. That is an
ACK that acknowledges previously unacknowledged data.
This is because upon receiving an out-of-order packet, the receiver
returns the last timestamp that advances RCV_NXT, not the current
timestamp of the packet in the DUPACK. Without checking the flag,
the DUPACK will cause tcp_packet_delayed() to return true and
tcp_try_undo_loss() will revert cwnd reduction.
Note that we check the condition in CA_Recovery already by only
calling tcp_try_undo_partial() if FLAG_SND_UNA_ADVANCED is set or
tcp_try_undo_recovery() if snd_una crosses high_seq.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While testing tight tcp_mem settings, I found tcp sessions could be
stuck because we do not allow even one skb to be received on them.
By allowing one skb to be received, we introduce fairness and
eventuallu force memory hogs to release their allocation.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Introduce an optimized version of sk_under_memory_pressure()
for TCP. Our intent is to use it in fast paths.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the advent of small rto timers in datacenter TCP,
(ip route ... rto_min x), the following can happen :
1) Qdisc is full, transmit fails.
TCP sets a timer based on icsk_rto to retry the transmit, without
exponential backoff.
With low icsk_rto, and lot of sockets, all cpus are servicing timer
interrupts like crazy.
Intent of the code was to retry with a timer between 200 (TCP_RTO_MIN)
and 500ms (TCP_RESOURCE_PROBE_INTERVAL)
2) Receivers can send zero windows if they don't drain their receive queue.
TCP sends zero window probes, based on icsk_rto current value, with
exponential backoff.
With /proc/sys/net/ipv4/tcp_retries2 being 15 (or even smaller in
some cases), sender can abort in less than one or two minutes !
If receiver stops the sender, it obviously doesn't care of very tight
rto. Probability of dropping the ACK reopening the window is not
worth the risk.
Lets change the base timer to be at least 200ms (TCP_RTO_MIN) for these
events (but not normal RTO based retransmits)
A followup patch adds a new SNMP counter, as it would have helped a lot
diagnosing this issue.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows a server application to get the TCP SYN headers for
its passive connections. This is useful if the server is doing
fingerprinting of clients based on SYN packet contents.
Two socket options are added: TCP_SAVE_SYN and TCP_SAVED_SYN.
The first is used on a socket to enable saving the SYN headers
for child connections. This can be set before or after the listen()
call.
The latter is used to retrieve the SYN headers for passive connections,
if the parent listener has enabled TCP_SAVE_SYN.
TCP_SAVED_SYN is read once, it frees the saved SYN headers.
The data returned in TCP_SAVED_SYN are network (IPv4/IPv6) and TCP
headers.
Original patch was written by Tom Herbert, I changed it to not hold
a full skb (and associated dst and conntracking reference).
We have used such patch for about 3 years at Google.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Invoking pkts_acked is currently conditioned on FLAG_ACKED:
receiving a cumulative ACK of new data, or ACK with SYN flag set.
Remove this condition so that CC may get RTT measurements from all SACKs.
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_sacktag_one() always picks the earliest sequence SACKed for RTT.
This might not make sense for congestion control in cases where:
1. ACKs are lost, i.e. a SACK following a lost SACK covers both
new and old segments at the receiver.
2. The receiver disregards the RFC 5681 recommendation to immediately
ACK out-of-order segments.
Give congestion control a RTT for the latest segment SACKed, which is the
most accurate RTT estimate, but preserve the conservative RTT for RTO.
Removes the call to skb_mstamp_get() in tcp_sacktag_one().
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Later patch passes two values set in tcp_sacktag_one() to
tcp_clean_rtx_queue(). Prepare passing them via struct tcp_sacktag_state.
Acked-by: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
tcp_mark_lost_retrans is not used when FACK is disabled. Since
tcp_update_reordering may disable FACK, it should be called first
before tcp_mark_lost_retrans.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch tracks total number of payload bytes received on a TCP socket.
This is the sum of all changes done to tp->rcv_nxt
RFC4898 named this : tcpEStatsAppHCThruOctetsReceived
This is a 64bit field, and can be fetched both from TCP_INFO
getsockopt() if one has a handle on a TCP socket, or from inet_diag
netlink facility (iproute2/ss patch will follow)
Note that tp->bytes_received was placed near tp->rcv_nxt for
best data locality and minimal performance impact.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch tracks total number of bytes acked for a TCP socket.
This is the sum of all changes done to tp->snd_una, and allows
for precise tracking of delivered data.
RFC4898 named this : tcpEStatsAppHCThruOctetsAcked
This is a 64bit field, and can be fetched both from TCP_INFO
getsockopt() if one has a handle on a TCP socket, or from inet_diag
netlink facility (iproute2/ss patch will follow)
Note that tp->bytes_acked was placed near tp->snd_una for
best data locality and minimal performance impact.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ensure that we either see that the buffer has write space
in tcp_poll() or that we perform a wakeup from the input
side. Did not run into any actual problem here, but thought
that we should make things explicit.
Signed-off-by: Jason Baron <jbaron@akamai.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since retransmitted segments are not used for RTT estimation, previously
SACKed segments present in the rtx queue are used. This estimation can be
several times larger than the actual RTT. When a cumulative ack covers both
previously SACKed and retransmitted segments, CC may thus get a bogus RTT.
Such segments previously had an RTT estimation in tcp_sacktag_one(), so it
seems reasonable to not reuse them in tcp_clean_rtx_queue() at all.
Afaik, this has had no effect on SRTT/RTO because of Karn's check.
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet