1 .. SPDX-License-Identifier: GPL-2.0 1 .. SPDX-License-Identifier: GPL-2.0
2 2
3 ==================== 3 ====================
4 mlx5 devlink support 4 mlx5 devlink support
5 ==================== 5 ====================
6 6
7 This document describes the devlink features i 7 This document describes the devlink features implemented by the ``mlx5``
8 device driver. 8 device driver.
9 9
10 Parameters 10 Parameters
11 ========== 11 ==========
12 12
13 .. list-table:: Generic parameters implemented 13 .. list-table:: Generic parameters implemented
14 14
15 * - Name 15 * - Name
16 - Mode 16 - Mode
17 - Validation 17 - Validation
18 * - ``enable_roce`` 18 * - ``enable_roce``
19 - driverinit 19 - driverinit
20 - Type: Boolean 20 - Type: Boolean
21 21
22 If the device supports RoCE disablement 22 If the device supports RoCE disablement, RoCE enablement state controls
23 device support for RoCE capability. Oth 23 device support for RoCE capability. Otherwise, the control occurs in the
24 driver stack. When RoCE is disabled at 24 driver stack. When RoCE is disabled at the driver level, only raw
25 ethernet QPs are supported. 25 ethernet QPs are supported.
26 * - ``io_eq_size`` 26 * - ``io_eq_size``
27 - driverinit 27 - driverinit
28 - The range is between 64 and 4096. 28 - The range is between 64 and 4096.
29 * - ``event_eq_size`` 29 * - ``event_eq_size``
30 - driverinit 30 - driverinit
31 - The range is between 64 and 4096. 31 - The range is between 64 and 4096.
32 * - ``max_macs`` 32 * - ``max_macs``
33 - driverinit 33 - driverinit
34 - The range is between 1 and 2^31. Only p 34 - The range is between 1 and 2^31. Only power of 2 values are supported.
35 35
36 The ``mlx5`` driver also implements the follow 36 The ``mlx5`` driver also implements the following driver-specific
37 parameters. 37 parameters.
38 38
39 .. list-table:: Driver-specific parameters imp 39 .. list-table:: Driver-specific parameters implemented
40 :widths: 5 5 5 85 40 :widths: 5 5 5 85
41 41
42 * - Name 42 * - Name
43 - Type 43 - Type
44 - Mode 44 - Mode
45 - Description 45 - Description
46 * - ``flow_steering_mode`` 46 * - ``flow_steering_mode``
47 - string 47 - string
48 - runtime 48 - runtime
49 - Controls the flow steering mode of the 49 - Controls the flow steering mode of the driver
50 50
51 * ``dmfs`` Device managed flow steering 51 * ``dmfs`` Device managed flow steering. In DMFS mode, the HW
52 steering entities are created and man 52 steering entities are created and managed through firmware.
53 * ``smfs`` Software managed flow steeri 53 * ``smfs`` Software managed flow steering. In SMFS mode, the HW
54 steering entities are created and man 54 steering entities are created and manage through the driver without
55 firmware intervention. 55 firmware intervention.
56 56
57 SMFS mode is faster and provides better 57 SMFS mode is faster and provides better rule insertion rate compared to
58 default DMFS mode. 58 default DMFS mode.
59 * - ``fdb_large_groups`` 59 * - ``fdb_large_groups``
60 - u32 60 - u32
61 - driverinit 61 - driverinit
62 - Control the number of large groups (siz 62 - Control the number of large groups (size > 1) in the FDB table.
63 63
64 * The default value is 15, and the rang 64 * The default value is 15, and the range is between 1 and 1024.
65 * - ``esw_multiport`` 65 * - ``esw_multiport``
66 - Boolean 66 - Boolean
67 - runtime 67 - runtime
68 - Control MultiPort E-Switch shared fdb m 68 - Control MultiPort E-Switch shared fdb mode.
69 69
70 An experimental mode where a single E-S 70 An experimental mode where a single E-Switch is used and all the vports
71 and physical ports on the NIC are conne 71 and physical ports on the NIC are connected to it.
72 72
73 An example is to send traffic from a VF 73 An example is to send traffic from a VF that is created on PF0 to an
74 uplink that is natively associated with 74 uplink that is natively associated with the uplink of PF1
75 75
76 Note: Future devices, ConnectX-8 and on 76 Note: Future devices, ConnectX-8 and onward, will eventually have this
77 as the default to allow forwarding betw 77 as the default to allow forwarding between all NIC ports in a single
78 E-switch environment and the dual E-swi 78 E-switch environment and the dual E-switch mode will likely get
79 deprecated. 79 deprecated.
80 80
81 Default: disabled 81 Default: disabled
82 * - ``esw_port_metadata`` 82 * - ``esw_port_metadata``
83 - Boolean 83 - Boolean
84 - runtime 84 - runtime
85 - When applicable, disabling eswitch meta 85 - When applicable, disabling eswitch metadata can increase packet rate up
86 to 20% depending on the use case and pa 86 to 20% depending on the use case and packet sizes.
87 87
88 Eswitch port metadata state controls wh 88 Eswitch port metadata state controls whether to internally tag packets
89 with metadata. Metadata tagging must be 89 with metadata. Metadata tagging must be enabled for multi-port RoCE,
90 failover between representors and stack 90 failover between representors and stacked devices. By default metadata is
91 enabled on the supported devices in E-s 91 enabled on the supported devices in E-switch. Metadata is applicable only
92 for E-switch in switchdev mode and user 92 for E-switch in switchdev mode and users may disable it when NONE of the
93 below use cases will be in use: 93 below use cases will be in use:
94 1. HCA is in Dual/multi-port RoCE mode. 94 1. HCA is in Dual/multi-port RoCE mode.
95 2. VF/SF representor bonding (Usually u 95 2. VF/SF representor bonding (Usually used for Live migration)
96 3. Stacked devices 96 3. Stacked devices
97 97
98 When metadata is disabled, the above us 98 When metadata is disabled, the above use cases will fail to initialize if
99 users try to enable them. 99 users try to enable them.
100 100
101 Note: Setting this parameter does not t 101 Note: Setting this parameter does not take effect immediately. Setting
102 must happen in legacy mode and eswitch 102 must happen in legacy mode and eswitch port metadata takes effect after
103 enabling switchdev mode. 103 enabling switchdev mode.
104 * - ``hairpin_num_queues`` 104 * - ``hairpin_num_queues``
105 - u32 105 - u32
106 - driverinit 106 - driverinit
107 - We refer to a TC NIC rule that involves 107 - We refer to a TC NIC rule that involves forwarding as "hairpin".
108 Hairpin queues are mlx5 hardware specif 108 Hairpin queues are mlx5 hardware specific implementation for hardware
109 forwarding of such packets. 109 forwarding of such packets.
110 110
111 Control the number of hairpin queues. 111 Control the number of hairpin queues.
112 * - ``hairpin_queue_size`` 112 * - ``hairpin_queue_size``
113 - u32 113 - u32
114 - driverinit 114 - driverinit
115 - Control the size (in packets) of the ha 115 - Control the size (in packets) of the hairpin queues.
116 116
117 The ``mlx5`` driver supports reloading via ``D 117 The ``mlx5`` driver supports reloading via ``DEVLINK_CMD_RELOAD``
118 118
119 Info versions 119 Info versions
120 ============= 120 =============
121 121
122 The ``mlx5`` driver reports the following vers 122 The ``mlx5`` driver reports the following versions
123 123
124 .. list-table:: devlink info versions implemen 124 .. list-table:: devlink info versions implemented
125 :widths: 5 5 90 125 :widths: 5 5 90
126 126
127 * - Name 127 * - Name
128 - Type 128 - Type
129 - Description 129 - Description
130 * - ``fw.psid`` 130 * - ``fw.psid``
131 - fixed 131 - fixed
132 - Used to represent the board id of the d 132 - Used to represent the board id of the device.
133 * - ``fw.version`` 133 * - ``fw.version``
134 - stored, running 134 - stored, running
135 - Three digit major.minor.subminor firmwa 135 - Three digit major.minor.subminor firmware version number.
136 136
137 Health reporters 137 Health reporters
138 ================ 138 ================
139 139
140 tx reporter 140 tx reporter
141 ----------- 141 -----------
142 The tx reporter is responsible for reporting a 142 The tx reporter is responsible for reporting and recovering of the following three error scenarios:
143 143
144 - tx timeout 144 - tx timeout
145 Report on kernel tx timeout detection. 145 Report on kernel tx timeout detection.
146 Recover by searching lost interrupts. 146 Recover by searching lost interrupts.
147 - tx error completion 147 - tx error completion
148 Report on error tx completion. 148 Report on error tx completion.
149 Recover by flushing the tx queue and reset 149 Recover by flushing the tx queue and reset it.
150 - tx PTP port timestamping CQ unhealthy 150 - tx PTP port timestamping CQ unhealthy
151 Report too many CQEs never delivered on po 151 Report too many CQEs never delivered on port ts CQ.
152 Recover by flushing and re-creating all PT 152 Recover by flushing and re-creating all PTP channels.
153 153
154 tx reporter also support on demand diagnose ca 154 tx reporter also support on demand diagnose callback, on which it provides
155 real time information of its send queues statu 155 real time information of its send queues status.
156 156
157 User commands examples: 157 User commands examples:
158 158
159 - Diagnose send queues status:: 159 - Diagnose send queues status::
160 160
161 $ devlink health diagnose pci/0000:82:00.0 161 $ devlink health diagnose pci/0000:82:00.0 reporter tx
162 162
163 .. note:: 163 .. note::
164 This command has valid output only when int 164 This command has valid output only when interface is up, otherwise the command has empty output.
165 165
166 - Show number of tx errors indicated, number o 166 - Show number of tx errors indicated, number of recover flows ended successfully,
167 is autorecover enabled and graceful period f 167 is autorecover enabled and graceful period from last recover::
168 168
169 $ devlink health show pci/0000:82:00.0 rep 169 $ devlink health show pci/0000:82:00.0 reporter tx
170 170
171 rx reporter 171 rx reporter
172 ----------- 172 -----------
173 The rx reporter is responsible for reporting a 173 The rx reporter is responsible for reporting and recovering of the following two error scenarios:
174 174
175 - rx queues' initialization (population) timeo 175 - rx queues' initialization (population) timeout
176 Population of rx queues' descriptors on ri 176 Population of rx queues' descriptors on ring initialization is done
177 in napi context via triggering an irq. In 177 in napi context via triggering an irq. In case of a failure to get
178 the minimum amount of descriptors, a timeo 178 the minimum amount of descriptors, a timeout would occur, and
179 descriptors could be recovered by polling 179 descriptors could be recovered by polling the EQ (Event Queue).
180 - rx completions with errors (reported by HW o 180 - rx completions with errors (reported by HW on interrupt context)
181 Report on rx completion error. 181 Report on rx completion error.
182 Recover (if needed) by flushing the relate 182 Recover (if needed) by flushing the related queue and reset it.
183 183
184 rx reporter also supports on demand diagnose c 184 rx reporter also supports on demand diagnose callback, on which it
185 provides real time information of its receive 185 provides real time information of its receive queues' status.
186 186
187 - Diagnose rx queues' status and corresponding 187 - Diagnose rx queues' status and corresponding completion queue::
188 188
189 $ devlink health diagnose pci/0000:82:00.0 189 $ devlink health diagnose pci/0000:82:00.0 reporter rx
190 190
191 .. note:: 191 .. note::
192 This command has valid output only when int 192 This command has valid output only when interface is up. Otherwise, the command has empty output.
193 193
194 - Show number of rx errors indicated, number o 194 - Show number of rx errors indicated, number of recover flows ended successfully,
195 is autorecover enabled, and graceful period 195 is autorecover enabled, and graceful period from last recover::
196 196
197 $ devlink health show pci/0000:82:00.0 rep 197 $ devlink health show pci/0000:82:00.0 reporter rx
198 198
199 fw reporter 199 fw reporter
200 ----------- 200 -----------
201 The fw reporter implements `diagnose` and `dum 201 The fw reporter implements `diagnose` and `dump` callbacks.
202 It follows symptoms of fw error such as fw syn 202 It follows symptoms of fw error such as fw syndrome by triggering
203 fw core dump and storing it into the dump buff 203 fw core dump and storing it into the dump buffer.
204 The fw reporter diagnose command can be trigge 204 The fw reporter diagnose command can be triggered any time by the user to check
205 current fw status. 205 current fw status.
206 206
207 User commands examples: 207 User commands examples:
208 208
209 - Check fw heath status:: 209 - Check fw heath status::
210 210
211 $ devlink health diagnose pci/0000:82:00.0 211 $ devlink health diagnose pci/0000:82:00.0 reporter fw
212 212
213 - Read FW core dump if already stored or trigg 213 - Read FW core dump if already stored or trigger new one::
214 214
215 $ devlink health dump show pci/0000:82:00. 215 $ devlink health dump show pci/0000:82:00.0 reporter fw
216 216
217 .. note:: 217 .. note::
218 This command can run only on the PF which h 218 This command can run only on the PF which has fw tracer ownership,
219 running it on other PF or any VF will retur 219 running it on other PF or any VF will return "Operation not permitted".
220 220
221 fw fatal reporter 221 fw fatal reporter
222 ----------------- 222 -----------------
223 The fw fatal reporter implements `dump` and `r 223 The fw fatal reporter implements `dump` and `recover` callbacks.
224 It follows fatal errors indications by CR-spac 224 It follows fatal errors indications by CR-space dump and recover flow.
225 The CR-space dump uses vsc interface which is 225 The CR-space dump uses vsc interface which is valid even if the FW command
226 interface is not functional, which is the case 226 interface is not functional, which is the case in most FW fatal errors.
227 The recover function runs recover flow which r 227 The recover function runs recover flow which reloads the driver and triggers fw
228 reset if needed. 228 reset if needed.
229 On firmware error, the health buffer is dumped 229 On firmware error, the health buffer is dumped into the dmesg. The log
230 level is derived from the error's severity (gi 230 level is derived from the error's severity (given in health buffer).
231 231
232 User commands examples: 232 User commands examples:
233 233
234 - Run fw recover flow manually:: 234 - Run fw recover flow manually::
235 235
236 $ devlink health recover pci/0000:82:00.0 236 $ devlink health recover pci/0000:82:00.0 reporter fw_fatal
237 237
238 - Read FW CR-space dump if already stored or t 238 - Read FW CR-space dump if already stored or trigger new one::
239 239
240 $ devlink health dump show pci/0000:82:00. 240 $ devlink health dump show pci/0000:82:00.1 reporter fw_fatal
241 241
242 .. note:: 242 .. note::
243 This command can run only on PF. 243 This command can run only on PF.
244 244
245 vnic reporter 245 vnic reporter
246 ------------- 246 -------------
247 The vnic reporter implements only the `diagnos 247 The vnic reporter implements only the `diagnose` callback.
248 It is responsible for querying the vnic diagno 248 It is responsible for querying the vnic diagnostic counters from fw and displaying
249 them in realtime. 249 them in realtime.
250 250
251 Description of the vnic counters: 251 Description of the vnic counters:
252 252
253 - total_error_queues 253 - total_error_queues
254 number of queues in an error state due 254 number of queues in an error state due to
255 an async error or errored command. 255 an async error or errored command.
256 - send_queue_priority_update_flow 256 - send_queue_priority_update_flow
257 number of QP/SQ priority/SL update eve 257 number of QP/SQ priority/SL update events.
258 - cq_overrun 258 - cq_overrun
259 number of times CQ entered an error st 259 number of times CQ entered an error state due to an overflow.
260 - async_eq_overrun 260 - async_eq_overrun
261 number of times an EQ mapped to async 261 number of times an EQ mapped to async events was overrun.
262 - comp_eq_overrun 262 - comp_eq_overrun
263 number of times an EQ mapped to comple 263 number of times an EQ mapped to completion events was
264 overrun. 264 overrun.
265 - quota_exceeded_command 265 - quota_exceeded_command
266 number of commands issued and failed d 266 number of commands issued and failed due to quota exceeded.
267 - invalid_command 267 - invalid_command
268 number of commands issued and failed d 268 number of commands issued and failed dues to any reason other than quota
269 exceeded. 269 exceeded.
270 - nic_receive_steering_discard 270 - nic_receive_steering_discard
271 number of packets that completed RX fl 271 number of packets that completed RX flow
272 steering but were discarded due to a m 272 steering but were discarded due to a mismatch in flow table.
273 - generated_pkt_steering_fail 273 - generated_pkt_steering_fail
274 number of packets generated by the VNI 274 number of packets generated by the VNIC experiencing unexpected steering
275 failure (at any point in steering flow 275 failure (at any point in steering flow).
276 - handled_pkt_steering_fail 276 - handled_pkt_steering_fail
277 number of packets handled by the VNIC 277 number of packets handled by the VNIC experiencing unexpected steering
278 failure (at any point in steering flow 278 failure (at any point in steering flow owned by the VNIC, including the FDB
279 for the eswitch owner). 279 for the eswitch owner).
280 280
281 User commands examples: 281 User commands examples:
282 282
283 - Diagnose PF/VF vnic counters:: 283 - Diagnose PF/VF vnic counters::
284 284
285 $ devlink health diagnose pci/0000:82: 285 $ devlink health diagnose pci/0000:82:00.1 reporter vnic
286 286
287 - Diagnose representor vnic counters (performe 287 - Diagnose representor vnic counters (performed by supplying devlink port of the
288 representor, which can be obtained via devli 288 representor, which can be obtained via devlink port command)::
289 289
290 $ devlink health diagnose pci/0000:82: 290 $ devlink health diagnose pci/0000:82:00.1/65537 reporter vnic
291 291
292 .. note:: 292 .. note::
293 This command can run over all interfaces su 293 This command can run over all interfaces such as PF/VF and representor ports.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.