mlx5 devlink support

mlx5 devlink 지원

이 문서는 `GPL-2.0` 라이선스를 따릅니다.

mlx5 devlink 지원

이 문서는 `mlx5` 장치 드라이버가 구현한 devlink 기능을 설명합니다.

.. SPDX-License-Identifier: GPL-2.0

====================
mlx5 devlink support
====================

This document describes the devlink features implemented by the ``mlx5``
device driver.

공통 매개변수

매개변수

`enable_sriov`와 `total_vfs` 같은 영구 매개변수는 적용하려면 펌웨어 재설정이 필요합니다.

# setup parameters
devlink dev param set pci/0000:01:00.0 name enable_sriov value true cmode permanent
devlink dev param set pci/0000:01:00.0 name total_vfs value 8 cmode permanent

# Fw reset
devlink dev reload pci/0000:01:00.0 action fw_activate

# for PCI related config such as sriov PCI reset/rescan is required:
echo 1 >/sys/bus/pci/devices/0000:01:00.0/remove
echo 1 >/sys/bus/pci/rescan
grep ^ /sys/bus/pci/devices/0000:01:00.0/sriov_*

Parameters
==========

.. list-table:: Generic parameters implemented

   * - Name
     - Mode
     - Validation
     - Notes
   * - ``enable_roce``
     - driverinit
     - Boolean
     - If the device supports RoCE disablement, RoCE enablement state controls
       device support for RoCE capability. Otherwise, the control occurs in the
       driver stack. When RoCE is disabled at the driver level, only raw
       ethernet QPs are supported.
   * - ``io_eq_size``
     - driverinit
     - The range is between 64 and 4096.
     -
   * - ``event_eq_size``
     - driverinit
     - The range is between 64 and 4096.
     -
   * - ``max_macs``
     - driverinit
     - The range is between 1 and 2^31. Only power of 2 values are supported.
     -
   * - ``enable_sriov``
     - permanent
     - Boolean
     - Applies to each physical function (PF) independently, if the device
       supports it. Otherwise, it applies symmetrically to all PFs.
   * - ``total_vfs``
     - permanent
     - The range is between 1 and a device-specific max.
     - Applies to each physical function (PF) independently, if the device
       supports it. Otherwise, it applies symmetrically to all PFs.

Note: permanent parameters such as ``enable_sriov`` and ``total_vfs`` require FW reset to take effect

.. code-block:: bash

   # setup parameters
   devlink dev param set pci/0000:01:00.0 name enable_sriov value true cmode permanent
   devlink dev param set pci/0000:01:00.0 name total_vfs value 8 cmode permanent

   # Fw reset
   devlink dev reload pci/0000:01:00.0 action fw_activate

   # for PCI related config such as sriov PCI reset/rescan is required:
   echo 1 >/sys/bus/pci/devices/0000:01:00.0/remove
   echo 1 >/sys/bus/pci/rescan
   grep ^ /sys/bus/pci/devices/0000:01:00.0/sriov_*

   * - ``num_doorbells``
     - driverinit
     - This controls the number of channel doorbells used by the netdev. In all
       cases, an additional doorbell is allocated and used for non-channel
       communication (e.g. for PTP, HWS, etc.). Supported values are:

       - 0: No channel-specific doorbells, use the global one for everything.
       - [1, max_num_channels]: Spread netdev channels equally across these
         doorbells.

흐름 조정과 E-Switch 매개변수

`mlx5` 드라이버는 다음과 같은 드라이버 전용 매개변수도 구현합니다.

향후 ConnectX-8 이상 장치에서는 단일 E-Switch 환경에서 모든 NIC 포트 간 전달을 허용하기 위해 `esw_multiport`가 결국 기본값이 되고, 이중 E-Switch 모드는 사용 중단될 가능성이 큽니다. 현재 기본값은 비활성화입니다.

The ``mlx5`` driver also implements the following driver-specific
parameters.

.. list-table:: Driver-specific parameters implemented
   :widths: 5 5 5 85

   * - Name
     - Type
     - Mode
     - Description
   * - ``flow_steering_mode``
     - string
     - runtime
     - Controls the flow steering mode of the driver

       * ``dmfs`` Device managed flow steering. In DMFS mode, the HW
         steering entities are created and managed through firmware.
       * ``smfs`` Software managed flow steering. In SMFS mode, the HW
         steering entities are created and manage through the driver without
         firmware intervention.
       * ``hmfs`` Hardware managed flow steering. In HMFS mode, the driver
         is configuring steering rules directly to the HW using Work Queues with
         a special new type of WQE (Work Queue Element).

       SMFS mode is faster and provides better rule insertion rate compared to
       default DMFS mode.
   * - ``fdb_large_groups``
     - u32
     - driverinit
     - Control the number of large groups (size > 1) in the FDB table.

       * The default value is 15, and the range is between 1 and 1024.
   * - ``esw_multiport``
     - Boolean
     - runtime
     - Control MultiPort E-Switch shared fdb mode.

       An experimental mode where a single E-Switch is used and all the vports
       and physical ports on the NIC are connected to it.

       An example is to send traffic from a VF that is created on PF0 to an
       uplink that is natively associated with the uplink of PF1

       Note: Future devices, ConnectX-8 and onward, will eventually have this
       as the default to allow forwarding between all NIC ports in a single
       E-switch environment and the dual E-switch mode will likely get
       deprecated.

       Default: disabled

메타데이터와 hairpin 매개변수

메타데이터 태깅은 다중 포트 RoCE, representor 사이의 장애 조치, 스택 장치에 필요합니다. 지원 장치의 E-Switch에서는 기본적으로 켜져 있습니다. 이 설정은 switchdev 모드의 E-Switch에만 적용되며, 다음 기능을 하나도 사용하지 않을 때만 꺼야 합니다: HCA의 이중·다중 포트 RoCE 모드, 보통 라이브 마이그레이션에 쓰이는 VF/SF representor bonding, 스택 장치.

메타데이터를 끈 상태에서 위 기능을 활성화하려 하면 초기화에 실패합니다. 매개변수 변경은 즉시 적용되지 않습니다. legacy 모드에서 설정한 다음 switchdev 모드를 활성화해야 E-Switch 포트 메타데이터 설정이 적용됩니다.

   * - ``esw_port_metadata``
     - Boolean
     - runtime
     - When applicable, disabling eswitch metadata can increase packet rate up
       to 20% depending on the use case and packet sizes.

       Eswitch port metadata state controls whether to internally tag packets
       with metadata. Metadata tagging must be enabled for multi-port RoCE,
       failover between representors and stacked devices. By default metadata is
       enabled on the supported devices in E-switch. Metadata is applicable only
       for E-switch in switchdev mode and users may disable it when NONE of the
       below use cases will be in use:
       1. HCA is in Dual/multi-port RoCE mode.
       2. VF/SF representor bonding (Usually used for Live migration)
       3. Stacked devices

       When metadata is disabled, the above use cases will fail to initialize if
       users try to enable them.

       Note: Setting this parameter does not take effect immediately. Setting
       must happen in legacy mode and eswitch port metadata takes effect after
       enabling switchdev mode.
   * - ``hairpin_num_queues``
     - u32
     - driverinit
     - We refer to a TC NIC rule that involves forwarding as "hairpin".
       Hairpin queues are mlx5 hardware specific implementation for hardware
       forwarding of such packets.

       Control the number of hairpin queues.
   * - ``hairpin_queue_size``
     - u32
     - driverinit
     - Control the size (in packets) of the hairpin queues.

PCIe 혼잡과 CQE 압축 매개변수

   * - ``pcie_cong_inbound_high``
     - u16
     - driverinit
     - High threshold configuration for PCIe congestion events. The firmware
       will send an event once device side inbound PCIe traffic went
       above the configured high threshold for a long enough period (at least
       200ms).

       See pci_bw_inbound_high ethtool stat.

       Units are 0.01 %. Accepted values are in range [0, 10000].
       pcie_cong_inbound_low < pcie_cong_inbound_high.
       Default value: 9000 (Corresponds to 90%).
   * - ``pcie_cong_inbound_low``
     - u16
     - driverinit
     - Low threshold configuration for PCIe congestion events. The firmware
       will send an event once device side inbound PCIe traffic went
       below the configured low threshold, only after having been previously in
       a congested state.

       See pci_bw_inbound_low ethtool stat.

       Units are 0.01 %. Accepted values are in range [0, 10000].
       pcie_cong_inbound_low < pcie_cong_inbound_high.
       Default value: 7500.
   * - ``pcie_cong_outbound_high``
     - u16
     - driverinit
     - High threshold configuration for PCIe congestion events. The firmware
       will send an event once device side outbound PCIe traffic went
       above the configured high threshold for a long enough period (at least
       200ms).

       See pci_bw_outbound_high ethtool stat.

       Units are 0.01 %. Accepted values are in range [0, 10000].
       pcie_cong_outbound_low < pcie_cong_outbound_high.
       Default value: 9000 (Corresponds to 90%).
   * - ``pcie_cong_outbound_low``
     - u16
     - driverinit
     - Low threshold configuration for PCIe congestion events. The firmware
       will send an event once device side outbound PCIe traffic went
       below the configured low threshold, only after having been previously in
       a congested state.

       See pci_bw_outbound_low ethtool stat.

       Units are 0.01 %. Accepted values are in range [0, 10000].
       pcie_cong_outbound_low < pcie_cong_outbound_high.
       Default value: 7500.

   * - ``cqe_compress_type``
     - string
     - permanent
     - Configure which mechanism/algorithm should be used by the NIC that will
       affect the rate (aggressiveness) of compressed CQEs depending on PCIe bus
       conditions and other internal NIC factors. This mode affects all queues
       that enable compression.
       * ``balanced`` : Merges fewer CQEs, resulting in a moderate compression ratio but maintaining a balance between bandwidth savings and performance
       * ``aggressive`` : Merges more CQEs into a single entry, achieving a higher compression rate and maximizing performance, particularly under high traffic loads

다시 불러오기

`mlx5` 드라이버는 `DEVLINK_CMD_RELOAD`를 통한 다시 불러오기를 지원합니다.

The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``

정보 버전

정보 버전

`mlx5` 드라이버는 다음 버전을 보고합니다.

Info versions
=============

The ``mlx5`` driver reports the following versions

.. list-table:: devlink info versions implemented
   :widths: 5 5 90

   * - Name
     - Type
     - Description
   * - ``fw.psid``
     - fixed
     - Used to represent the board id of the device.
   * - ``fw.version``
     - stored, running
     - Three digit major.minor.subminor firmware version number.

TX 상태 보고자

상태 보고자

TX 보고자

TX 보고자는 다음 세 오류 상황을 보고하고 복구합니다.

TX 보고자는 요청 시 실행하는 diagnose 콜백도 지원하며, 송신 큐 상태의 실시간 정보를 제공합니다.

송신 큐 상태를 진단합니다.

$ devlink health diagnose pci/0000:82:00.0 reporter tx

이 명령은 인터페이스가 올라와 있을 때만 유효한 출력을 내며, 내려가 있으면 출력이 비어 있습니다.

표시된 TX 오류 수, 성공적으로 끝난 복구 흐름 수, 자동 복구 활성화 여부와 마지막 복구 이후의 유예 기간을 표시합니다.

$ devlink health show pci/0000:82:00.0 reporter tx

Health reporters
================

tx reporter
-----------
The tx reporter is responsible for reporting and recovering of the following three error scenarios:

- tx timeout
    Report on kernel tx timeout detection.
    Recover by searching lost interrupts.
- tx error completion
    Report on error tx completion.
    Recover by flushing the tx queue and reset it.
- tx PTP port timestamping CQ unhealthy
    Report too many CQEs never delivered on port ts CQ.
    Recover by flushing and re-creating all PTP channels.

tx reporter also support on demand diagnose callback, on which it provides
real time information of its send queues status.

User commands examples:

- Diagnose send queues status::

    $ devlink health diagnose pci/0000:82:00.0 reporter tx

.. note::
   This command has valid output only when interface is up, otherwise the command has empty output.

- Show number of tx errors indicated, number of recover flows ended successfully,
  is autorecover enabled and graceful period from last recover::

    $ devlink health show pci/0000:82:00.0 reporter tx

RX 상태 보고자

RX 보고자

RX 보고자는 다음 두 오류 상황을 보고하고 복구합니다.

RX 보고자는 요청 시 diagnose 콜백도 지원하며 수신 큐와 대응 completion queue의 실시간 상태를 제공합니다.

$ devlink health diagnose pci/0000:82:00.0 reporter rx

이 명령도 인터페이스가 올라와 있을 때만 유효한 출력을 내며, 내려가 있으면 출력이 비어 있습니다.

표시된 RX 오류 수, 성공적으로 끝난 복구 흐름 수, 자동 복구 활성화 여부와 마지막 복구 이후의 유예 기간을 표시합니다.

$ devlink health show pci/0000:82:00.0 reporter rx

rx reporter
-----------
The rx reporter is responsible for reporting and recovering of the following two error scenarios:

- rx queues' initialization (population) timeout
    Population of rx queues' descriptors on ring initialization is done
    in napi context via triggering an irq. In case of a failure to get
    the minimum amount of descriptors, a timeout would occur, and
    descriptors could be recovered by polling the EQ (Event Queue).
- rx completions with errors (reported by HW on interrupt context)
    Report on rx completion error.
    Recover (if needed) by flushing the related queue and reset it.

rx reporter also supports on demand diagnose callback, on which it
provides real time information of its receive queues' status.

- Diagnose rx queues' status and corresponding completion queue::

    $ devlink health diagnose pci/0000:82:00.0 reporter rx

.. note::
   This command has valid output only when interface is up. Otherwise, the command has empty output.

- Show number of rx errors indicated, number of recover flows ended successfully,
  is autorecover enabled, and graceful period from last recover::

    $ devlink health show pci/0000:82:00.0 reporter rx

펌웨어 상태 보고자

펌웨어 보고자

FW 보고자는 `diagnose`와 `dump` 콜백을 구현합니다. FW syndrome 같은 펌웨어 오류 징후를 추적하고, 펌웨어 core dump를 트리거하여 dump buffer에 저장합니다. 사용자는 언제든 diagnose 명령을 실행해 현재 펌웨어 상태를 확인할 수 있습니다.

펌웨어 상태를 확인합니다.

$ devlink health diagnose pci/0000:82:00.0 reporter fw

이미 저장된 펌웨어 core dump를 읽거나 새 dump를 트리거합니다.

$ devlink health dump show pci/0000:82:00.0 reporter fw

이 명령은 FW tracer 소유권을 가진 PF에서만 실행할 수 있습니다. 다른 PF나 어떤 VF에서든 실행하면 `Operation not permitted`가 반환됩니다.

fw reporter
-----------
The fw reporter implements `diagnose` and `dump` callbacks.
It follows symptoms of fw error such as fw syndrome by triggering
fw core dump and storing it into the dump buffer.
The fw reporter diagnose command can be triggered any time by the user to check
current fw status.

User commands examples:

- Check fw heath status::

    $ devlink health diagnose pci/0000:82:00.0 reporter fw

- Read FW core dump if already stored or trigger new one::

    $ devlink health dump show pci/0000:82:00.0 reporter fw

.. note::
   This command can run only on the PF which has fw tracer ownership,
   running it on other PF or any VF will return "Operation not permitted".

치명적 펌웨어 상태 보고자

치명적 펌웨어 보고자

FW fatal 보고자는 `dump`와 `recover` 콜백을 구현합니다. CR-space dump와 복구 흐름으로 치명적 오류 표시를 추적합니다. CR-space dump는 FW 명령 인터페이스가 동작하지 않아도 유효한 VSC 인터페이스를 사용합니다. 대부분의 치명적 FW 오류가 바로 이런 상황입니다.

복구 함수는 드라이버를 다시 불러오고 필요하면 펌웨어 재설정을 트리거하는 복구 흐름을 실행합니다. 펌웨어 오류가 나면 health buffer를 dmesg에 덤프하며, 로그 수준은 health buffer가 제공하는 오류 심각도에서 정합니다.

펌웨어 복구 흐름을 수동으로 실행합니다.

$ devlink health recover pci/0000:82:00.0 reporter fw_fatal

이미 저장된 펌웨어 CR-space dump를 읽거나 새 dump를 트리거합니다.

$ devlink health dump show pci/0000:82:00.1 reporter fw_fatal

이 명령은 PF에서만 실행할 수 있습니다.

fw fatal reporter
-----------------
The fw fatal reporter implements `dump` and `recover` callbacks.
It follows fatal errors indications by CR-space dump and recover flow.
The CR-space dump uses vsc interface which is valid even if the FW command
interface is not functional, which is the case in most FW fatal errors.
The recover function runs recover flow which reloads the driver and triggers fw
reset if needed.
On firmware error, the health buffer is dumped into the dmesg. The log
level is derived from the error's severity (given in health buffer).

User commands examples:

- Run fw recover flow manually::

    $ devlink health recover pci/0000:82:00.0 reporter fw_fatal

- Read FW CR-space dump if already stored or trigger new one::

    $ devlink health dump show pci/0000:82:00.1 reporter fw_fatal

.. note::
   This command can run only on PF.

VNIC 상태 보고자

VNIC 보고자

VNIC 보고자는 `diagnose` 콜백만 구현합니다. 펌웨어에서 VNIC 진단 카운터를 조회하여 실시간으로 표시합니다.

PF 또는 VF의 VNIC 카운터를 진단합니다.

$ devlink health diagnose pci/0000:82:00.1 reporter vnic

representor의 devlink port를 지정하여 representor VNIC 카운터를 진단할 수도 있습니다. 이 포트는 `devlink port` 명령으로 확인합니다.

$ devlink health diagnose pci/0000:82:00.1/65537 reporter vnic

이 명령은 PF, VF, representor port를 포함한 모든 인터페이스에서 실행할 수 있습니다.

vnic reporter
-------------
The vnic reporter implements only the `diagnose` callback.
It is responsible for querying the vnic diagnostic counters from fw and displaying
them in realtime.

Description of the vnic counters:

- total_error_queues
        number of queues in an error state due to
        an async error or errored command.
- send_queue_priority_update_flow
        number of QP/SQ priority/SL update events.
- cq_overrun
        number of times CQ entered an error state due to an overflow.
- async_eq_overrun
        number of times an EQ mapped to async events was overrun.
- comp_eq_overrun
        number of times an EQ mapped to completion events was
        overrun.
- quota_exceeded_command
        number of commands issued and failed due to quota exceeded.
- invalid_command
        number of commands issued and failed dues to any reason other than quota
        exceeded.
- nic_receive_steering_discard
        number of packets that completed RX flow
        steering but were discarded due to a mismatch in flow table.
- generated_pkt_steering_fail
        number of packets generated by the VNIC experiencing unexpected steering
        failure (at any point in steering flow).
- handled_pkt_steering_fail
        number of packets handled by the VNIC experiencing unexpected steering
        failure (at any point in steering flow owned by the VNIC, including the FDB
        for the eswitch owner).
- icm_consumption
        amount of Interconnect Host Memory (ICM) consumed by the vnic in
        granularity of 4KB. ICM is host memory allocated by SW upon HCA request
        and is used for storing data structures that control HCA operation.
- bar_uar_access
        number of WRITE or READ access operations to the UAR on the PCIe BAR.
- odp_local_triggered_page_fault
        number of locally-triggered page-faults due to ODP.
- odp_remote_triggered_page_fault
        number of remotly-triggered page-faults due to ODP.

User commands examples:

- Diagnose PF/VF vnic counters::

        $ devlink health diagnose pci/0000:82:00.1 reporter vnic

- Diagnose representor vnic counters (performed by supplying devlink port of the
  representor, which can be obtained via devlink port command)::

        $ devlink health diagnose pci/0000:82:00.1/65537 reporter vnic

.. note::
   This command can run over all interfaces such as PF/VF and representor ports.