1 What: /sys/kernel/debug/accel/<paren 1 What: /sys/kernel/debug/accel/<parent_device>/addr
2 Date: Jan 2019 2 Date: Jan 2019
3 KernelVersion: 5.1 3 KernelVersion: 5.1
4 Contact: ogabbay@kernel.org 4 Contact: ogabbay@kernel.org
5 Description: Sets the device address to be 5 Description: Sets the device address to be used for read or write through
6 PCI bar, or the device VA of a 6 PCI bar, or the device VA of a host mapped memory to be read or
7 written directly from the host 7 written directly from the host. The latter option is allowed
8 only when the IOMMU is disable 8 only when the IOMMU is disabled.
9 The acceptable value is a stri 9 The acceptable value is a string that starts with "0x"
10 10
11 What: /sys/kernel/debug/accel/<paren 11 What: /sys/kernel/debug/accel/<parent_device>/clk_gate
12 Date: May 2020 12 Date: May 2020
13 KernelVersion: 5.8 13 KernelVersion: 5.8
14 Contact: ogabbay@kernel.org 14 Contact: ogabbay@kernel.org
15 Description: This setting is now deprecated 15 Description: This setting is now deprecated as clock gating is handled solely by the f/w
16 16
17 What: /sys/kernel/debug/accel/<paren 17 What: /sys/kernel/debug/accel/<parent_device>/command_buffers
18 Date: Jan 2019 18 Date: Jan 2019
19 KernelVersion: 5.1 19 KernelVersion: 5.1
20 Contact: ogabbay@kernel.org 20 Contact: ogabbay@kernel.org
21 Description: Displays a list with informati 21 Description: Displays a list with information about the currently allocated
22 command buffers 22 command buffers
23 23
24 What: /sys/kernel/debug/accel/<paren 24 What: /sys/kernel/debug/accel/<parent_device>/command_submission
25 Date: Jan 2019 25 Date: Jan 2019
26 KernelVersion: 5.1 26 KernelVersion: 5.1
27 Contact: ogabbay@kernel.org 27 Contact: ogabbay@kernel.org
28 Description: Displays a list with informati 28 Description: Displays a list with information about the currently active
29 command submissions 29 command submissions
30 30
31 What: /sys/kernel/debug/accel/<paren 31 What: /sys/kernel/debug/accel/<parent_device>/command_submission_jobs
32 Date: Jan 2019 32 Date: Jan 2019
33 KernelVersion: 5.1 33 KernelVersion: 5.1
34 Contact: ogabbay@kernel.org 34 Contact: ogabbay@kernel.org
35 Description: Displays a list with detailed 35 Description: Displays a list with detailed information about each JOB (CB) of
36 each active command submission 36 each active command submission
37 37
38 What: /sys/kernel/debug/accel/<paren 38 What: /sys/kernel/debug/accel/<parent_device>/data32
39 Date: Jan 2019 39 Date: Jan 2019
40 KernelVersion: 5.1 40 KernelVersion: 5.1
41 Contact: ogabbay@kernel.org 41 Contact: ogabbay@kernel.org
42 Description: Allows the root user to read o 42 Description: Allows the root user to read or write directly through the
43 device's PCI bar. Writing to t 43 device's PCI bar. Writing to this file generates a write
44 transaction while reading from 44 transaction while reading from the file generates a read
45 transaction. This custom inter 45 transaction. This custom interface is needed (instead of using
46 the generic Linux user-space P 46 the generic Linux user-space PCI mapping) because the DDR bar
47 is very small compared to the 47 is very small compared to the DDR memory and only the driver can
48 move the bar before and after 48 move the bar before and after the transaction.
49 49
50 If the IOMMU is disabled, it a 50 If the IOMMU is disabled, it also allows the root user to read
51 or write from the host a devic 51 or write from the host a device VA of a host mapped memory
52 52
53 What: /sys/kernel/debug/accel/<paren 53 What: /sys/kernel/debug/accel/<parent_device>/data64
54 Date: Jan 2020 54 Date: Jan 2020
55 KernelVersion: 5.6 55 KernelVersion: 5.6
56 Contact: ogabbay@kernel.org 56 Contact: ogabbay@kernel.org
57 Description: Allows the root user to read o 57 Description: Allows the root user to read or write 64 bit data directly
58 through the device's PCI bar. 58 through the device's PCI bar. Writing to this file generates a
59 write transaction while readin 59 write transaction while reading from the file generates a read
60 transaction. This custom inter 60 transaction. This custom interface is needed (instead of using
61 the generic Linux user-space P 61 the generic Linux user-space PCI mapping) because the DDR bar
62 is very small compared to the 62 is very small compared to the DDR memory and only the driver can
63 move the bar before and after 63 move the bar before and after the transaction.
64 64
65 If the IOMMU is disabled, it a 65 If the IOMMU is disabled, it also allows the root user to read
66 or write from the host a devic 66 or write from the host a device VA of a host mapped memory
67 67
68 What: /sys/kernel/debug/accel/<paren 68 What: /sys/kernel/debug/accel/<parent_device>/data_dma
69 Date: Apr 2021 69 Date: Apr 2021
70 KernelVersion: 5.13 70 KernelVersion: 5.13
71 Contact: ogabbay@kernel.org 71 Contact: ogabbay@kernel.org
72 Description: Allows the root user to read f 72 Description: Allows the root user to read from the device's internal
73 memory (DRAM/SRAM) through a D 73 memory (DRAM/SRAM) through a DMA engine.
74 This property is a binary blob 74 This property is a binary blob that contains the result of the
75 DMA transfer. 75 DMA transfer.
76 This custom interface is neede 76 This custom interface is needed (instead of using the generic
77 Linux user-space PCI mapping) 77 Linux user-space PCI mapping) because the amount of internal
78 memory is huge (>32GB) and rea 78 memory is huge (>32GB) and reading it via the PCI bar will take
79 a very long time. 79 a very long time.
80 This interface doesn't support 80 This interface doesn't support concurrency in the same device.
81 In GAUDI and GOYA, this action 81 In GAUDI and GOYA, this action can cause undefined behavior
82 in case it is done while the d 82 in case it is done while the device is executing user
83 workloads. 83 workloads.
84 Only supported on GAUDI at thi 84 Only supported on GAUDI at this stage.
85 85
86 What: /sys/kernel/debug/accel/<paren 86 What: /sys/kernel/debug/accel/<parent_device>/device
87 Date: Jan 2019 87 Date: Jan 2019
88 KernelVersion: 5.1 88 KernelVersion: 5.1
89 Contact: ogabbay@kernel.org 89 Contact: ogabbay@kernel.org
90 Description: Enables the root user to set t 90 Description: Enables the root user to set the device to specific state.
91 Valid values are "disable", "e 91 Valid values are "disable", "enable", "suspend", "resume".
92 User can read this property to 92 User can read this property to see the valid values
93 93
94 What: /sys/kernel/debug/accel/<paren 94 What: /sys/kernel/debug/accel/<parent_device>/device_release_watchdog_timeout
95 Date: Oct 2022 95 Date: Oct 2022
96 KernelVersion: 6.2 96 KernelVersion: 6.2
97 Contact: ttayar@habana.ai 97 Contact: ttayar@habana.ai
98 Description: The watchdog timeout value in 98 Description: The watchdog timeout value in seconds for a device release upon
99 certain error cases, after whi 99 certain error cases, after which the device is reset.
100 100
101 What: /sys/kernel/debug/accel/<paren 101 What: /sys/kernel/debug/accel/<parent_device>/dma_size
102 Date: Apr 2021 102 Date: Apr 2021
103 KernelVersion: 5.13 103 KernelVersion: 5.13
104 Contact: ogabbay@kernel.org 104 Contact: ogabbay@kernel.org
105 Description: Specify the size of the DMA tr 105 Description: Specify the size of the DMA transaction when using DMA to read
106 from the device's internal mem 106 from the device's internal memory. The value can not be larger
107 than 128MB. Writing to this va 107 than 128MB. Writing to this value initiates the DMA transfer.
108 When the write is finished, th 108 When the write is finished, the user can read the "data_dma"
109 blob 109 blob
110 110
111 What: /sys/kernel/debug/accel/<paren 111 What: /sys/kernel/debug/accel/<parent_device>/dump_razwi_events
112 Date: Aug 2022 112 Date: Aug 2022
113 KernelVersion: 5.20 113 KernelVersion: 5.20
114 Contact: fkassabri@habana.ai 114 Contact: fkassabri@habana.ai
115 Description: Dumps all razwi events to dmes 115 Description: Dumps all razwi events to dmesg if exist.
116 After reading the status regis 116 After reading the status register of an existing event
117 the routine will clear the sta 117 the routine will clear the status register.
118 Usage: cat dump_razwi_events 118 Usage: cat dump_razwi_events
119 119
120 What: /sys/kernel/debug/accel/<paren 120 What: /sys/kernel/debug/accel/<parent_device>/dump_security_violations
121 Date: Jan 2021 121 Date: Jan 2021
122 KernelVersion: 5.12 122 KernelVersion: 5.12
123 Contact: ogabbay@kernel.org 123 Contact: ogabbay@kernel.org
124 Description: Dumps all security violations 124 Description: Dumps all security violations to dmesg. This will also ack
125 all security violations meanin 125 all security violations meanings those violations will not be
126 dumped next time user calls th 126 dumped next time user calls this API
127 127
128 What: /sys/kernel/debug/accel/<paren 128 What: /sys/kernel/debug/accel/<parent_device>/engines
129 Date: Jul 2019 129 Date: Jul 2019
130 KernelVersion: 5.3 130 KernelVersion: 5.3
131 Contact: ogabbay@kernel.org 131 Contact: ogabbay@kernel.org
132 Description: Displays the status registers 132 Description: Displays the status registers values of the device engines and
133 their derived idle status 133 their derived idle status
134 134
135 What: /sys/kernel/debug/accel/<paren 135 What: /sys/kernel/debug/accel/<parent_device>/i2c_addr
136 Date: Jan 2019 136 Date: Jan 2019
137 KernelVersion: 5.1 137 KernelVersion: 5.1
138 Contact: ogabbay@kernel.org 138 Contact: ogabbay@kernel.org
139 Description: Sets I2C device address for I2 139 Description: Sets I2C device address for I2C transaction that is generated
140 by the device's CPU, Not avail 140 by the device's CPU, Not available when device is loaded with secured
141 firmware 141 firmware
142 142
143 What: /sys/kernel/debug/accel/<paren 143 What: /sys/kernel/debug/accel/<parent_device>/i2c_bus
144 Date: Jan 2019 144 Date: Jan 2019
145 KernelVersion: 5.1 145 KernelVersion: 5.1
146 Contact: ogabbay@kernel.org 146 Contact: ogabbay@kernel.org
147 Description: Sets I2C bus address for I2C t 147 Description: Sets I2C bus address for I2C transaction that is generated by
148 the device's CPU, Not availabl 148 the device's CPU, Not available when device is loaded with secured
149 firmware 149 firmware
150 150
151 What: /sys/kernel/debug/accel/<paren 151 What: /sys/kernel/debug/accel/<parent_device>/i2c_data
152 Date: Jan 2019 152 Date: Jan 2019
153 KernelVersion: 5.1 153 KernelVersion: 5.1
154 Contact: ogabbay@kernel.org 154 Contact: ogabbay@kernel.org
155 Description: Triggers an I2C transaction th 155 Description: Triggers an I2C transaction that is generated by the device's
156 CPU. Writing to this file gene 156 CPU. Writing to this file generates a write transaction while
157 reading from the file generate 157 reading from the file generates a read transaction, Not available
158 when device is loaded with sec 158 when device is loaded with secured firmware
159 159
160 What: /sys/kernel/debug/accel/<paren 160 What: /sys/kernel/debug/accel/<parent_device>/i2c_len
161 Date: Dec 2021 161 Date: Dec 2021
162 KernelVersion: 5.17 162 KernelVersion: 5.17
163 Contact: obitton@habana.ai 163 Contact: obitton@habana.ai
164 Description: Sets I2C length in bytes for I 164 Description: Sets I2C length in bytes for I2C transaction that is generated by
165 the device's CPU, Not availabl 165 the device's CPU, Not available when device is loaded with secured
166 firmware 166 firmware
167 167
168 What: /sys/kernel/debug/accel/<paren 168 What: /sys/kernel/debug/accel/<parent_device>/i2c_reg
169 Date: Jan 2019 169 Date: Jan 2019
170 KernelVersion: 5.1 170 KernelVersion: 5.1
171 Contact: ogabbay@kernel.org 171 Contact: ogabbay@kernel.org
172 Description: Sets I2C register id for I2C t 172 Description: Sets I2C register id for I2C transaction that is generated by
173 the device's CPU, Not availabl 173 the device's CPU, Not available when device is loaded with secured
174 firmware 174 firmware
175 175
176 What: /sys/kernel/debug/accel/<paren 176 What: /sys/kernel/debug/accel/<parent_device>/led0
177 Date: Jan 2019 177 Date: Jan 2019
178 KernelVersion: 5.1 178 KernelVersion: 5.1
179 Contact: ogabbay@kernel.org 179 Contact: ogabbay@kernel.org
180 Description: Sets the state of the first S/ 180 Description: Sets the state of the first S/W led on the device, Not available
181 when device is loaded with sec 181 when device is loaded with secured firmware
182 182
183 What: /sys/kernel/debug/accel/<paren 183 What: /sys/kernel/debug/accel/<parent_device>/led1
184 Date: Jan 2019 184 Date: Jan 2019
185 KernelVersion: 5.1 185 KernelVersion: 5.1
186 Contact: ogabbay@kernel.org 186 Contact: ogabbay@kernel.org
187 Description: Sets the state of the second S 187 Description: Sets the state of the second S/W led on the device, Not available
188 when device is loaded with sec 188 when device is loaded with secured firmware
189 189
190 What: /sys/kernel/debug/accel/<paren 190 What: /sys/kernel/debug/accel/<parent_device>/led2
191 Date: Jan 2019 191 Date: Jan 2019
192 KernelVersion: 5.1 192 KernelVersion: 5.1
193 Contact: ogabbay@kernel.org 193 Contact: ogabbay@kernel.org
194 Description: Sets the state of the third S/ 194 Description: Sets the state of the third S/W led on the device, Not available
195 when device is loaded with sec 195 when device is loaded with secured firmware
196 196
197 What: /sys/kernel/debug/accel/<paren 197 What: /sys/kernel/debug/accel/<parent_device>/memory_scrub
198 Date: May 2022 198 Date: May 2022
199 KernelVersion: 5.19 199 KernelVersion: 5.19
200 Contact: dhirschfeld@habana.ai 200 Contact: dhirschfeld@habana.ai
201 Description: Allows the root user to scrub 201 Description: Allows the root user to scrub the dram memory. The scrubbing
202 value can be set using the deb 202 value can be set using the debugfs file memory_scrub_val.
203 203
204 What: /sys/kernel/debug/accel/<paren 204 What: /sys/kernel/debug/accel/<parent_device>/memory_scrub_val
205 Date: May 2022 205 Date: May 2022
206 KernelVersion: 5.19 206 KernelVersion: 5.19
207 Contact: dhirschfeld@habana.ai 207 Contact: dhirschfeld@habana.ai
208 Description: The value to which the dram wi 208 Description: The value to which the dram will be set to when the user
209 scrubs the dram using 'memory_ 209 scrubs the dram using 'memory_scrub' debugfs file and
210 the scrubbing value when using 210 the scrubbing value when using module param 'memory_scrub'
211 211
212 What: /sys/kernel/debug/accel/<paren 212 What: /sys/kernel/debug/accel/<parent_device>/mmu
213 Date: Jan 2019 213 Date: Jan 2019
214 KernelVersion: 5.1 214 KernelVersion: 5.1
215 Contact: ogabbay@kernel.org 215 Contact: ogabbay@kernel.org
216 Description: Displays the hop values and ph 216 Description: Displays the hop values and physical address for a given ASID
217 and virtual address. The user 217 and virtual address. The user should write the ASID and VA into
218 the file and then read the fil 218 the file and then read the file to get the result.
219 e.g. to display info about VA 219 e.g. to display info about VA 0x1000 for ASID 1 you need to do:
220 echo "1 0x1000" > /sys/kernel/ 220 echo "1 0x1000" > /sys/kernel/debug/accel/<parent_device>/mmu
221 221
222 What: /sys/kernel/debug/accel/<paren 222 What: /sys/kernel/debug/accel/<parent_device>/mmu_error
223 Date: Mar 2021 223 Date: Mar 2021
224 KernelVersion: 5.12 224 KernelVersion: 5.12
225 Contact: fkassabri@habana.ai 225 Contact: fkassabri@habana.ai
226 Description: Check and display page fault o 226 Description: Check and display page fault or access violation mmu errors for
227 all MMUs specified in mmu_cap_ 227 all MMUs specified in mmu_cap_mask.
228 e.g. to display error info for 228 e.g. to display error info for MMU hw cap bit 9, you need to do:
229 echo "0x200" > /sys/kernel/deb 229 echo "0x200" > /sys/kernel/debug/accel/<parent_device>/mmu_error
230 cat /sys/kernel/debug/accel/<p 230 cat /sys/kernel/debug/accel/<parent_device>/mmu_error
231 231
232 What: /sys/kernel/debug/accel/<paren 232 What: /sys/kernel/debug/accel/<parent_device>/monitor_dump
233 Date: Mar 2022 233 Date: Mar 2022
234 KernelVersion: 5.19 234 KernelVersion: 5.19
235 Contact: osharabi@habana.ai 235 Contact: osharabi@habana.ai
236 Description: Allows the root user to dump m 236 Description: Allows the root user to dump monitors status from the device's
237 protected config space. 237 protected config space.
238 This property is a binary blob 238 This property is a binary blob that contains the result of the
239 monitors registers dump. 239 monitors registers dump.
240 This custom interface is neede 240 This custom interface is needed (instead of using the generic
241 Linux user-space PCI mapping) 241 Linux user-space PCI mapping) because this space is protected
242 and cannot be accessed using P 242 and cannot be accessed using PCI read.
243 This interface doesn't support 243 This interface doesn't support concurrency in the same device.
244 Only supported on GAUDI. 244 Only supported on GAUDI.
245 245
246 What: /sys/kernel/debug/accel/<paren 246 What: /sys/kernel/debug/accel/<parent_device>/monitor_dump_trig
247 Date: Mar 2022 247 Date: Mar 2022
248 KernelVersion: 5.19 248 KernelVersion: 5.19
249 Contact: osharabi@habana.ai 249 Contact: osharabi@habana.ai
250 Description: Triggers dump of monitor data. 250 Description: Triggers dump of monitor data. The value to trigger the operation
251 must be 1. Triggering the moni 251 must be 1. Triggering the monitor dump operation initiates dump of
252 current registers values of al 252 current registers values of all monitors.
253 When the write is finished, th 253 When the write is finished, the user can read the "monitor_dump"
254 blob 254 blob
255 255
256 What: /sys/kernel/debug/accel/<paren 256 What: /sys/kernel/debug/accel/<parent_device>/server_type
257 Date: Feb 2024 257 Date: Feb 2024
258 KernelVersion: 6.11 258 KernelVersion: 6.11
259 Contact: trisin@habana.ai 259 Contact: trisin@habana.ai
260 Description: Exposes the device's server ty 260 Description: Exposes the device's server type, maps to enum hl_server_type.
261 261
262 What: /sys/kernel/debug/accel/<paren 262 What: /sys/kernel/debug/accel/<parent_device>/set_power_state
263 Date: Jan 2019 263 Date: Jan 2019
264 KernelVersion: 5.1 264 KernelVersion: 5.1
265 Contact: ogabbay@kernel.org 265 Contact: ogabbay@kernel.org
266 Description: Sets the PCI power state. Vali 266 Description: Sets the PCI power state. Valid values are "1" for D0 and "2"
267 for D3Hot 267 for D3Hot
268 268
269 What: /sys/kernel/debug/accel/<paren 269 What: /sys/kernel/debug/accel/<parent_device>/skip_reset_on_timeout
270 Date: Jun 2021 270 Date: Jun 2021
271 KernelVersion: 5.13 271 KernelVersion: 5.13
272 Contact: ynudelman@habana.ai 272 Contact: ynudelman@habana.ai
273 Description: Sets the skip reset on timeout 273 Description: Sets the skip reset on timeout option for the device. Value of
274 "0" means device will be reset 274 "0" means device will be reset in case some CS has timed out,
275 otherwise it will not be reset 275 otherwise it will not be reset.
276 276
277 What: /sys/kernel/debug/accel/<paren 277 What: /sys/kernel/debug/accel/<parent_device>/state_dump
278 Date: Oct 2021 278 Date: Oct 2021
279 KernelVersion: 5.15 279 KernelVersion: 5.15
280 Contact: ynudelman@habana.ai 280 Contact: ynudelman@habana.ai
281 Description: Gets the state dump occurring 281 Description: Gets the state dump occurring on a CS timeout or failure.
282 State dump is used for debug a 282 State dump is used for debug and is created each time in case of
283 a problem in a CS execution, b 283 a problem in a CS execution, before reset.
284 Reading from the node returns 284 Reading from the node returns the newest state dump available.
285 Writing an integer X discards 285 Writing an integer X discards X state dumps, so that the
286 next read would return X+1-st 286 next read would return X+1-st newest state dump.
287 287
288 What: /sys/kernel/debug/accel/<paren 288 What: /sys/kernel/debug/accel/<parent_device>/stop_on_err
289 Date: Mar 2020 289 Date: Mar 2020
290 KernelVersion: 5.6 290 KernelVersion: 5.6
291 Contact: ogabbay@kernel.org 291 Contact: ogabbay@kernel.org
292 Description: Sets the stop-on_error option 292 Description: Sets the stop-on_error option for the device engines. Value of
293 "0" is for disable, otherwise 293 "0" is for disable, otherwise enable.
294 Relevant only for GOYA and GAU 294 Relevant only for GOYA and GAUDI.
295 295
296 What: /sys/kernel/debug/accel/<paren 296 What: /sys/kernel/debug/accel/<parent_device>/timeout_locked
297 Date: Sep 2021 297 Date: Sep 2021
298 KernelVersion: 5.16 298 KernelVersion: 5.16
299 Contact: obitton@habana.ai 299 Contact: obitton@habana.ai
300 Description: Sets the command submission ti 300 Description: Sets the command submission timeout value in seconds.
301 301
302 What: /sys/kernel/debug/accel/<paren 302 What: /sys/kernel/debug/accel/<parent_device>/userptr
303 Date: Jan 2019 303 Date: Jan 2019
304 KernelVersion: 5.1 304 KernelVersion: 5.1
305 Contact: ogabbay@kernel.org 305 Contact: ogabbay@kernel.org
306 Description: Displays a list with informati 306 Description: Displays a list with information about the current user
307 pointers (user virtual address 307 pointers (user virtual addresses) that are pinned and mapped
308 to DMA addresses 308 to DMA addresses
309 309
310 What: /sys/kernel/debug/accel/<paren 310 What: /sys/kernel/debug/accel/<parent_device>/userptr_lookup
311 Date: Oct 2021 311 Date: Oct 2021
312 KernelVersion: 5.15 312 KernelVersion: 5.15
313 Contact: ogabbay@kernel.org 313 Contact: ogabbay@kernel.org
314 Description: Allows to search for specific 314 Description: Allows to search for specific user pointers (user virtual
315 addresses) that are pinned and 315 addresses) that are pinned and mapped to DMA addresses, and see
316 their resolution to the specif 316 their resolution to the specific dma address.
317 317
318 What: /sys/kernel/debug/accel/<paren 318 What: /sys/kernel/debug/accel/<parent_device>/vm
319 Date: Jan 2019 319 Date: Jan 2019
320 KernelVersion: 5.1 320 KernelVersion: 5.1
321 Contact: ogabbay@kernel.org 321 Contact: ogabbay@kernel.org
322 Description: Displays a list with informati 322 Description: Displays a list with information about all the active virtual
323 address mappings per ASID and 323 address mappings per ASID and all user mappings of HW blocks
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.