1 .. SPDX-License-Identifier: GPL-2.0 2 3 ============== 4 Devlink Health 5 ============== 6 7 Background 8 ========== 9 10 The ``devlink`` health mechanism is targeted for Real Time Alerting, in 11 order to know when something bad happened to a PCI device. 12 13 * Provide alert debug information. 14 * Self healing. 15 * If problem needs vendor support, provide a way to gather all needed 16 debugging information. 17 18 Overview 19 ======== 20 21 The main idea is to unify and centralize driver health reports in the 22 generic ``devlink`` instance and allow the user to set different 23 attributes of the health reporting and recovery procedures. 24 25 The ``devlink`` health reporter: 26 Device driver creates a "health reporter" per each error/health type. 27 Error/Health type can be a known/generic (e.g. PCI error, fw error, rx/tx error) 28 or unknown (driver specific). 29 For each registered health reporter a driver can issue error/health reports 30 asynchronously. All health reports handling is done by ``devlink``. 31 Device driver can provide specific callbacks for each "health reporter", e.g.: 32 33 * Recovery procedures 34 * Diagnostics procedures 35 * Object dump procedures 36 * Out Of Box initial parameters 37 38 Different parts of the driver can register different types of health reporters 39 with different handlers. 40 41 Actions 42 ======= 43 44 Once an error is reported, devlink health will perform the following actions: 45 46 * A log is being send to the kernel trace events buffer 47 * Health status and statistics are being updated for the reporter instance 48 * Object dump is being taken and saved at the reporter instance (as long as 49 auto-dump is set and there is no other dump which is already stored) 50 * Auto recovery attempt is being done. Depends on: 51 52 - Auto-recovery configuration 53 - Grace period vs. time passed since last recover 54 55 Devlink formatted message 56 ========================= 57 58 To handle devlink health diagnose and health dump requests, devlink creates a 59 formatted message structure ``devlink_fmsg`` and send it to the driver's callback 60 to fill the data in using the devlink fmsg API. 61 62 Devlink fmsg is a mechanism to pass descriptors between drivers and devlink, in 63 json-like format. The API allows the driver to add nested attributes such as 64 object, object pair and value array, in addition to attributes such as name and 65 value. 66 67 Driver should use this API to fill the fmsg context in a format which will be 68 translated by the devlink to the netlink message later. When it needs to send 69 the data using SKBs to the netlink layer, it fragments the data between 70 different SKBs. In order to do this fragmentation, it uses virtual nests 71 attributes, to avoid actual nesting use which cannot be divided between 72 different SKBs. 73 74 User Interface 75 ============== 76 77 User can access/change each reporter's parameters and driver specific callbacks 78 via ``devlink``, e.g per error type (per health reporter): 79 80 * Configure reporter's generic parameters (like: disable/enable auto recovery) 81 * Invoke recovery procedure 82 * Run diagnostics 83 * Object dump 84 85 .. list-table:: List of devlink health interfaces 86 :widths: 10 90 87 88 * - Name 89 - Description 90 * - ``DEVLINK_CMD_HEALTH_REPORTER_GET`` 91 - Retrieves status and configuration info per DEV and reporter. 92 * - ``DEVLINK_CMD_HEALTH_REPORTER_SET`` 93 - Allows reporter-related configuration setting. 94 * - ``DEVLINK_CMD_HEALTH_REPORTER_RECOVER`` 95 - Triggers reporter's recovery procedure. 96 * - ``DEVLINK_CMD_HEALTH_REPORTER_TEST`` 97 - Triggers a fake health event on the reporter. The effects of the test 98 event in terms of recovery flow should follow closely that of a real 99 event. 100 * - ``DEVLINK_CMD_HEALTH_REPORTER_DIAGNOSE`` 101 - Retrieves current device state related to the reporter. 102 * - ``DEVLINK_CMD_HEALTH_REPORTER_DUMP_GET`` 103 - Retrieves the last stored dump. Devlink health 104 saves a single dump. If an dump is not already stored by devlink 105 for this reporter, devlink generates a new dump. 106 Dump output is defined by the reporter. 107 * - ``DEVLINK_CMD_HEALTH_REPORTER_DUMP_CLEAR`` 108 - Clears the last saved dump file for the specified reporter. 109 110 The following diagram provides a general overview of ``devlink-health``:: 111 112 netlink 113 +--------------------------+ 114 | | 115 | + | 116 | | | 117 +--------------------------+ 118 |request for ops 119 |(diagnose, 120 driver devlink |recover, 121 |dump) 122 +--------+ +--------------------------+ 123 | | | reporter| | 124 | | | +---------v----------+ | 125 | | ops execution | | | | 126 | <----------------------------------+ | | 127 | | | | | | 128 | | | + ^------------------+ | 129 | | | | request for ops | 130 | | | | (recover, dump) | 131 | | | | | 132 | | | +-+------------------+ | 133 | | health report | | health handler | | 134 | +-------------------------------> | | 135 | | | +--------------------+ | 136 | | health reporter create | | 137 | +----------------------------> | 138 +--------+ +--------------------------+
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.