Skip to content

L
linux
Switch branch/tag
  1. 15 Nov, 2019 3 commits
    • Dan Murphy's avatar
      net: phy: dp83869: Add TI dp83869 phy · 01db923e
      Dan Murphy authored 
      Add support for the TI DP83869 Gigabit ethernet phy
device.

The DP83869 is a robust, low power, fully featured
Physical Layer transceiver with integrated PMD
sublayers to support 10BASE-T, 100BASE-TX and
1000BASE-T Ethernet protocols.
Signed-off-by: default avatarDan Murphy <dmurphy@ti.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      01db923e
    • Dan Murphy's avatar
      dt-bindings: net: dp83869: Add TI dp83869 phy · 4d66c56f
      Dan Murphy authored 
      Add dt bindings for the TI dp83869 Gigabit ethernet phy
device.
Signed-off-by: default avatarDan Murphy <dmurphy@ti.com>
CC: Rob Herring <robh+dt@kernel.org>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      4d66c56f
    • Tonghao Zhang's avatar
      net: openvswitch: add hash info to upcall · bd1903b7
      Tonghao Zhang authored 
      When using the kernel datapath, the upcall don't
include skb hash info relatived. That will introduce
some problem, because the hash of skb is important
in kernel stack. For example, VXLAN module uses
it to select UDP src port. The tx queue selection
may also use the hash in stack.

Hash is computed in different ways. Hash is random
for a TCP socket, and hash may be computed in hardware,
or software stack. Recalculation hash is not easy.

Hash of TCP socket is computed:
tcp_v4_connect
    -> sk_set_txhash (is random)

__tcp_transmit_skb
    -> skb_set_hash_from_sk

There will be one upcall, without information of skb
hash, to ovs-vswitchd, for the first packet of a TCP
session. The rest packets will be processed in Open vSwitch
modules, hash kept. If this tcp session is forward to
VXLAN module, then the UDP src port of first tcp packet
is different from rest packets.

TCP packets may come from the host or dockers, to Open vSwitch.
To fix it, we store the hash info to upcall, and restore hash
when packets sent back.

+---------------+          +-------------------------+
|   Docker/VMs  |          |     ovs-vswitchd        |
+----+----------+          +-+--------------------+--+
     |                       ^                    |
     |                       |                    |
     |                       |  upcall            v restore packet hash (not recalculate)
     |                     +-+--------------------+--+
     |  tap netdev         |                         |   vxlan module
     +--------------->     +-->  Open vSwitch ko     +-->
       or internal type    |                         |
                           +-------------------------+

Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2019-October/364062.htmlSigned-off-by: default avatarTonghao Zhang <xiangxia.m.yue@gmail.com>
Acked-by: default avatarPravin B Shelar <pshelar@ovn.org>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      bd1903b7
  3. 14 Nov, 2019 8 commits
    • David S. Miller's avatar
      Merge branch 'Rework-mt762x-GDM-setup-flow' · 839554b7
      David S. Miller authored 
      MarkLee says:

====================
Rework mt762x GDM setup flow

The mt762x GDM block is mainly used to setup the HW internal
rx path from GMAC to RX DMA engine(PDMA) and the packet
switching engine(PSE) is responsed to do the data forward
following the GDM configuration.

This patch set have three goals :

1. Integrate GDM/PSE setup operations into single function "mtk_gdm_config"

2. Refine the timing of GDM/PSE setup, move it from mtk_hw_init
   to mtk_open

3. Enable GDM GDMA_DROP_ALL mode to drop all packet during the
   stop operation
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      839554b7
    • MarkLee's avatar
      net: ethernet: mediatek: Enable GDM GDMA_DROP_ALL mode · 8d66a818
      MarkLee authored 
      Enable GDM GDMA_DROP_ALL mode to drop all packet during the
stop operation. This is recommended by the mt762x HW design
to drop all packet from GMAC before stopping PDMA.
Signed-off-by: default avatarMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      8d66a818
    • MarkLee's avatar
      net: ethernet: mediatek: Refine the timing of GDM/PSE setup · 5ac9eda0
      MarkLee authored 
      Refine the timing of GDM/PSE setup, move it from mtk_hw_init
to mtk_open. This is recommended by the mt762x HW design to
do GDM/PSE setup only after PDMA has been started.

We exclude mt7628 in mtk_gdm_config function since it is a old IP
and there is no GDM/PSE block on it.
Signed-off-by: default avatarMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      5ac9eda0
    • MarkLee's avatar
      net: ethernet: mediatek: Integrate GDM/PSE setup operations · 8d3f4a95
      MarkLee authored 
      Integrate GDM/PSE setup operations into single function "mtk_gdm_config"
Signed-off-by: default avatarMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      8d3f4a95
    • Vladimir Oltean's avatar
      net: dsa: sja1105: Simplify reset handling · abfb228a
      Vladimir Oltean authored 
      We don't really need 10k species of reset. Remove everything except cold
reset which is what is actually used. Too bad the hardware designers
couldn't agree to use the same bit field for rev 1 and rev 2, so the
(*reset_cmd) function pointer is there to stay.

However let's simplify the prototype and give it a struct dsa_switch (we
want to avoid forward-declarations of structures, in this case struct
sja1105_private, wherever we can).
Signed-off-by: default avatarVladimir Oltean <olteanv@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      abfb228a
    • David S. Miller's avatar
      Merge branch 'PTP-clock-source-for-SJA1105-tc-taprio-offload' · ccb68993
      David S. Miller authored 
      Vladimir Oltean says:

====================
PTP clock source for SJA1105 tc-taprio offload

This series makes the IEEE 802.1Qbv egress scheduler of the sja1105
switch use a time reference that is synchronized to the network. This
enables quite a few real Time Sensitive Networking use cases, since in
this mode the switch can offer its clients a TDMA sort of access to the
network, and guaranteed latency for frames that are properly scheduled
based on the common PTP time.

The driver needs to do a 2-part activity:
- Program the gate control list into the static config and upload it
  over SPI to the switch (already supported)
- Write the activation time of the scheduler (base-time) into the
  PTPSCHTM register, and set the PTPSTRTSCH bit.
- Monitor the activation of the scheduler at the planned time and its
  health.

Ok, 3 parts.

The time-aware scheduler cannot be programmed to activate at a time in
the past, and there is some logic to avoid that.

PTPCLKCORP is one of those "black magic" registers that just need to be
written to the length of the cycle. There is a 40-line long comment in
the second patch which explains why.
====================
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      ccb68993
    • Vladimir Oltean's avatar
      net: dsa: sja1105: Implement state machine for TAS with PTP clock source · 86db36a3
      Vladimir Oltean authored 
      Tested using the following bash script and the tc from iproute2-next:

	#!/bin/bash

	set -e -u -o pipefail

	NSEC_PER_SEC="1000000000"

	gatemask() {
		local tc_list="$1"
		local mask=0

		for tc in ${tc_list}; do
			mask=$((${mask} | (1 << ${tc})))
		done

		printf "%02x" ${mask}
	}

	if ! systemctl is-active --quiet ptp4l; then
		echo "Please start the ptp4l service"
		exit
	fi

	now=$(phc_ctl /dev/ptp1 get | gawk '/clock time is/ { print $5; }')
	# Phase-align the base time to the start of the next second.
	sec=$(echo "${now}" | gawk -F. '{ print $1; }')
	base_time="$(((${sec} + 1) * ${NSEC_PER_SEC}))"

	tc qdisc add dev swp5 parent root handle 100 taprio \
		num_tc 8 \
		map 0 1 2 3 5 6 7 \
		queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 \
		base-time ${base_time} \
		sched-entry S $(gatemask 7) 100000 \
		sched-entry S $(gatemask "0 1 2 3 4 5 6") 400000 \
		clockid CLOCK_TAI flags 2

The "state machine" is a workqueue invoked after each manipulation
command on the PTP clock (reset, adjust time, set time, adjust
frequency) which checks over the state of the time-aware scheduler.
So it is not monitored periodically, only in reaction to a PTP command
typically triggered from a userspace daemon (linuxptp). Otherwise there
is no reason for things to go wrong.

Now that the timecounter/cyclecounter has been replaced with hardware
operations on the PTP clock, the TAS Kconfig now depends upon PTP and
the standalone clocksource operating mode has been removed.
Signed-off-by: default avatarVladimir Oltean <olteanv@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      86db36a3
    • Vladimir Oltean's avatar
      net: dsa: sja1105: Make the PTP command read-write · 41603d78
      Vladimir Oltean authored 
      The PTPSTRTSCH and PTPSTOPSCH bits are actually readable and indicate
whether the time-aware scheduler is running or not. We will be using
that for monitoring the scheduler in the next patch, so refactor the PTP
command API in order to allow that.
Signed-off-by: default avatarVladimir Oltean <olteanv@gmail.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
      41603d78
  5. 13 Nov, 2019 29 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7