Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 15 11月, 2019 8 次提交
    • D
      Merge branch 'Rework-mt762x-GDM-setup-flow' · 839554b7
      由 David S. Miller 提交于 
      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: NDavid S. Miller <davem@davemloft.net>
      839554b7
    • M
      net: ethernet: mediatek: Enable GDM GDMA_DROP_ALL mode · 8d66a818
      由 MarkLee 提交于 
      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: NMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      8d66a818
    • M
      net: ethernet: mediatek: Refine the timing of GDM/PSE setup · 5ac9eda0
      由 MarkLee 提交于 
      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: NMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      5ac9eda0
    • M
      net: ethernet: mediatek: Integrate GDM/PSE setup operations · 8d3f4a95
      由 MarkLee 提交于 
      Integrate GDM/PSE setup operations into single function "mtk_gdm_config"
Signed-off-by: NMarkLee <Mark-MC.Lee@mediatek.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      8d3f4a95
    • V
      net: dsa: sja1105: Simplify reset handling · abfb228a
      由 Vladimir Oltean 提交于 
      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: NVladimir Oltean <olteanv@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      abfb228a
    • D
      Merge branch 'PTP-clock-source-for-SJA1105-tc-taprio-offload' · ccb68993
      由 David S. Miller 提交于 
      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: NDavid S. Miller <davem@davemloft.net>
      ccb68993
    • V
      net: dsa: sja1105: Implement state machine for TAS with PTP clock source · 86db36a3
      由 Vladimir Oltean 提交于 
      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: NVladimir Oltean <olteanv@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      86db36a3
    • V
      net: dsa: sja1105: Make the PTP command read-write · 41603d78
      由 Vladimir Oltean 提交于 
      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: NVladimir Oltean <olteanv@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
      41603d78
  3. 14 11月, 2019 13 次提交
  5. 13 11月, 2019 19 次提交