Currently RACK would mark loss before the undo operations in TCP
loss recovery. This could incorrectly identify real losses as
spurious. For example a sender first experiences a delay spike and
then eventually some packets were lost due to buffer overrun.
In this case, the sender should perform fast recovery b/c not all
the packets were lost.

But the sender may first trigger a (spurious) RTO and reset
cwnd to 1. The following ACKs may used to mark real losses by
tcp_rack_mark_lost. Then in tcp_process_loss this ACK could trigger
F-RTO undo condition and unmark real losses and revert the cwnd
reduction. If there are no more ACKs coming back, eventually the
sender would timeout again instead of performing fast recovery.

The patch fixes this incorrect process by always performing
the undo checks before detecting losses.

Fixes: 4f41b1c5 ("tcp: use RACK to detect losses")
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The packets inside a jumbo skb (e.g., TSO) share the same skb
timestamp, even though they are sent sequentially on the wire. Since
RACK is based on time, it can not detect some packets inside the
same skb are lost.  However, we can leverage the packet sequence
numbers as extended timestamps to detect losses. Therefore, when
RACK timestamp is identical to skb's timestamp (i.e., one of the
packets of the skb is acked or sacked), we use the sequence numbers
of the acked and unacked packets to break ties.

We can use the same sequence logic to advance RACK xmit time as
well to detect more losses and avoid timeout.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch makes RACK install a reordering timer when it suspects
some packets might be lost, but wants to delay the decision
a little bit to accomodate reordering.

It does not create a new timer but instead repurposes the existing
RTO timer, because both are meant to retransmit packets.
Specifically it arms a timer ICSK_TIME_REO_TIMEOUT when
the RACK timing check fails. The wait time is set to

  RACK.RTT + RACK.reo_wnd - (NOW - Packet.xmit_time) + fudge

This translates to expecting a packet (Packet) should take
(RACK.RTT + RACK.reo_wnd + fudge) to deliver after it was sent.

When there are multiple packets that need a timer, we use one timer
with the maximum timeout. Therefore the timer conservatively uses
the maximum window to expire N packets by one timeout, instead of
N timeouts to expire N packets sent at different times.

The fudge factor is 2 jiffies to ensure when the timer fires, all
the suspected packets would exceed the deadline and be marked lost
by tcp_rack_detect_loss(). It has to be at least 1 jiffy because the
clock may tick between calling icsk_reset_xmit_timer(timeout) and
actually hang the timer. The next jiffy is to lower-bound the timeout
to 2 jiffies when reo_wnd is < 1ms.

When the reordering timer fires (tcp_rack_reo_timeout): If we aren't
in Recovery we'll enter fast recovery and force fast retransmit.
This is very similar to the early retransmit (RFC5827) except RACK
is not constrained to only enter recovery for small outstanding
flights.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Record the most recent RTT in RACK. It is often identical to the
"ca_rtt_us" values in tcp_clean_rtx_queue. But when the packet has
been retransmitted, RACK choses to believe the ACK is for the
(latest) retransmitted packet if the RTT is over minimum RTT.

This requires passing the arrival time of the most recent ACK to
RACK routines. The timestamp is now recorded in the "ack_time"
in tcp_sacktag_state during the ACK processing.

This patch does not change the RACK algorithm itself. It only adds
the RTT variable to prepare the next main patch.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Create a new helper tcp_rack_detect_loss to prepare the upcoming
RACK reordering timer patch.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Create a new helper tcp_rack_mark_skb_lost to prepare the
upcoming RACK reordering timer support.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove assignment to ndo_select_queue so that fallback is used for
selecting txq.  Also remove the now-useless function that used to be
assigned to ndo_select_queue.
Signed-off-by: NSatanand Burla <satananda.burla@cavium.com>
Signed-off-by: NFelix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: NDerek Chickles <derek.chickles@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The Marvell 6352 chip has a 8-bit address/16-bit data EEPROM access.
The Marvell 6390 chip has a 16-bit address/8-bit data EEPROM access.

This patch implements the 8-bit data EEPROM access in the mv88e6xxx
driver and adds its support to chips of the 6390 family.
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Current allocations are not NUMA aware, and lack proper
cleanup in case of error.

It is perfectly fine to use static per cpu allocations for 256 bytes
per cpu.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: David Lebrun <david.lebrun@uclouvain.be>
Acked-by: NDavid Lebrun <david.lebrun@uclouvain.be>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Recently we started using ipmr with thousands of entries and easily hit
soft lockups on smaller devices. The reason is that the hash function
uses the high order bits from the src and dst, but those don't change in
many common cases, also the hash table  is only 64 elements so with
thousands it doesn't scale at all.
This patch migrates the hash table to rhashtable, and in particular the
rhl interface which allows for duplicate elements to be chained because
of the MFC_PROXY support (*,G; *,*,oif cases) which allows for multiple
duplicate entries to be added with different interfaces (IMO wrong, but
it's been in for a long time).

And here are some results from tests I've run in a VM:
 mr_table size (default, allocated for all namespaces):
  Before                    After
   49304 bytes               2400 bytes

 Add 65000 routes (the diff is much larger on smaller devices):
  Before                    After
   1m42s                     58s

 Forwarding 256 byte packets with 65000 routes (test done in a VM):
  Before                    After
   3 Mbps / ~1465 pps        122 Mbps / ~59000 pps

As a bonus we no longer see the soft lockups on smaller devices which
showed up even with 2000 entries before.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes following warnings :

net/core/secure_seq.c:125:28: warning: incorrect type in argument 1
(different base types)
net/core/secure_seq.c:125:28:    expected unsigned int const [unsigned]
[usertype] a
net/core/secure_seq.c:125:28:    got restricted __be32 [usertype] saddr
net/core/secure_seq.c:125:35: warning: incorrect type in argument 2
(different base types)
net/core/secure_seq.c:125:35:    expected unsigned int const [unsigned]
[usertype] b
net/core/secure_seq.c:125:35:    got restricted __be32 [usertype] daddr
net/core/secure_seq.c:125:43: warning: cast from restricted __be16
net/core/secure_seq.c:125:61: warning: restricted __be16 degrades to
integer

Fixes: 7cd23e53 ("secure_seq: use SipHash in place of MD5")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reviewed-by: NJason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Reduce the load time of the VF driver by decreasing the wait time between
iterations of the loop that polls for a mailbox response from the PF. Also
change the wait time units from jiffies to milliseconds.
Signed-off-by: NPrasad Kanneganti <prasad.kanneganti@cavium.com>
Signed-off-by: NFelix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: NRaghu Vatsavayi <raghu.vatsavayi@cavium.com>
Signed-off-by: NDerek Chickles <derek.chickles@cavium.com>
Signed-off-by: NSatanand Burla <satananda.burla@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove code that's no longer needed.  It used to serve a purpose, which was
to fix a link-related bug.  For a while now, the NIC firmware has had a
more elegant fix for that bug.
Signed-off-by: NFelix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: NDerek Chickles <derek.chickles@cavium.com>
Signed-off-by: NSatanand Burla <satananda.burla@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit bc1f4470 ("net: make ndo_get_stats64 a void function")
mistakenly used a return value for this void conversion.

Fix it.
Signed-off-by: NJoe Perches <joe@perches.com>
cc: stephen hemminger <stephen@networkplumber.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Florian Fainelli says:

====================
net: mdio-gpio: Use modern GPIO helpers

This patch series modernizes the mdio-gpio and makes it switch to the
latest and greatest API for manipulating GPIO lines, thus allowing
some simplifications in the driver.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

gpiod functions support handling low-active pins, so we can move
thos code out of this driver into the gpio subsystem and simplify
the code a bit.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using gpiod functions lets us use functionality which is not available
with gpio functions.

There is no gpiod function to match devm_gpio_request_one, so leave it
in place and use gpio_to_desc() to convert absolute pin numbers to gpio
descriptors.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using devm_gpio_request_one lets us request gpio pins with initial state
in one go.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes the following sparse warning:

drivers/net/usb/cdc_ether.c:469:6: warning:
 symbol 'usbnet_cdc_zte_status' was not declared. Should it be static?
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The filter added by sock_setfilter is intended to only permit
packets matching the pattern set up by create_payload(), but
we only check the ip_len, and a single test-character in
the IP packet to ensure this condition.

Harden the filter by adding additional constraints so that we only
permit UDP/IPv4 packets that meet the ip_len and test-character
requirements. Include the bpf_asm src as a comment, in case this
needs to be enhanced in the future
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes the following sparse warning:

net/core/lwt_bpf.c:355:5: warning:
 symbol 'bpf_lwt_prog_cmp' was not declared. Should it be static?
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Ursula Braun says:

====================
s390: qeth patches

yesterday I came up with 13 qeth patches. Since you have not been
happy with the 13th patch, I want to make sure that at least the
remaining 12 qeth patches can be applied to net-next. Here is the
resend of them.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

qeth devices in layer3 mode need a separate handling of vipa and proxy-arp
addresses. vipa and proxy-arp addresses processed by qeth can be read from
userspace. Introduced with commit 5f78e29c ("qeth: optimize IP handling
in rx_mode callback") the retrieval of vipa and proxy-arp addresses is
broken, if more than one vipa or proxy-arp address are set.

The qeth code used local variable "int i" for 2 different purposes. This
patch now spends 2 separate local variables of type "int".
While touching these functions hash_for_each_safe() is converted to
hash_for_each(), since there is no removal of hash entries.
Signed-off-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Reviewed-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reference-ID: RQM 3524
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

STARTLAN needs to be the first IPA command after MPC initialization
completes.
So move the qeth_send_startlan() call from the layer disciplines
into the core path, right after the MPC handshake.
While at it, replace the magic LAN OFFLINE return code
with the existing enum.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Move all MAC utility functions in one place, and drop the
forward declarations.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This matches qeth_l2_write_mac().
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Consolidate errno handling for MAC management: Instead of doing this in every
caller, do it in one place.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Suggested-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

qeth_l2_send_groupmac() already translates the return code, so
calling qeth_setdel_makerc() a second time only produces garbage.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The only caller passes del = 0, so remove both the parameter and
the code that handles != 0.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Acked-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove unused define QETH_IP_HEADER_SIZE.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Acked-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Accessing the current hsuid via card->options.hsuid is perfectly
fine, even when the card is DOWN.
Signed-off-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Acked-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When RX/TX checksum offloading is turned on and the adapter is
an OSA 3 card in layer 3 mode, the checksum offloading is only
performed when both peers use different adapters. If both peers
share an OSA 3 card, communication is a memory copy and
checksum offloading is not performed.

This patch adds a warning to inform the administrator.

OSA 3 in layer 2 mode does not offer the RX/TX checksum
offload feature.
Signed-off-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Turning on receive and/or transmit checksum offload support
on the OSA card requires 2 commands:
1. start command which replies with available features
2. enable command to turn on selected features.

The current version does not check the reply of the start
command and simply uses the returned value to enable
offload features. When the start command returns zero, this
leads to a situation where no checksum offload
is turned on by the hardware. Even worse no error
indication is returned. The Linux kernel assumes
the OSA card performs RX/TX checksum offload, but the hardware
does not perform any checksum verification at all.

This patch checks the return of the start and enable
command responses from the hardware and turns off
checksum offloading if the commands fails or does not
respond with the correct bit setting.
Signed-off-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Rework the RX/TX checksum offloading command sequence to use
the provided function call back mechanims to return card
data to the device driver.
Signed-off-by: NThomas Richter <tmricht@linux.vnet.ibm.com>
Reviewed-by: NJulian Wiedmann <jwi@linux.vnet.ibm.com>
Reviewed-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
More flexible BPF cb access

This patch improves BPF's cb access by allowing b/h/w/dw
access variants on it. For details, please see individual
patches.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When structs are used to store temporary state in cb[] buffer that is
used with programs and among tail calls, then the generated code will
not always access the buffer in bpf_w chunks. We can ease programming
of it and let this act more natural by allowing for aligned b/h/w/dw
sized access for cb[] ctx member. Various test cases are attached as
well for the selftest suite. Potentially, this can also be reused for
other program types to pass data around.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, when calling convert_ctx_access() callback for the various
program types, we pass in insn->dst_reg, insn->src_reg, insn->off from
the original instruction. This information is needed to rewrite the
instruction that is based on the user ctx structure into a kernel
representation for the ctx. As we'd like to allow access size beyond
just BPF_W, we'd need also insn->code for that in order to decode the
original access size. Given that, lets just pass insn directly to the
convert_ctx_access() callback and work on that to not clutter the
callback with even more arguments we need to pass when everything is
already contained in insn. So lets go through that once, no functional
change.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Ursula Braun says:

====================
net/smc: fix typo and clc-bug

I received 2 bug reports for my new AF_SMC-code. Here are the fixes for them.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When creating an SMC connection, there is a CLC (connection layer control)
handshake to prepare for RDMA traffic. The corresponding code is part of
commit 0cfdd8f9 ("smc: connection and link group creation").
Mac addresses to be exchanged in the handshake are copied with a wrong
length of 12 instead of 6 bytes. Following code overwrites the wrongly
copied code, but nevertheless the correct length should already be used for
the preceding mac address copying. Use ETH_ALEN for the memcpy length with
mac addresses.
Signed-off-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Fixes: 0cfdd8f9 ("smc: connection and link group creation")
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When introducing the new socket family AF_SMC in
commit ac713874 ("smc: establish new socket family"),
a typo in af_family_clock_key_strings has slipped in.
This patch repairs it.
Signed-off-by: NUrsula Braun <ubraun@linux.vnet.ibm.com>
Fixes: ac713874 ("smc: establish new socket family")
Reported-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>