1 What: /sys/kernel/debug/qat_<device> 1 What: /sys/kernel/debug/qat_<device>_<BDF>/fw_counters
2 Date: November 2023 2 Date: November 2023
3 KernelVersion: 6.6 3 KernelVersion: 6.6
4 Contact: qat-linux@intel.com 4 Contact: qat-linux@intel.com
5 Description: (RO) Read returns the number o 5 Description: (RO) Read returns the number of requests sent to the FW and the number of responses
6 received from the FW for each 6 received from the FW for each Acceleration Engine
7 Reported firmware counters:: 7 Reported firmware counters::
8 8
9 <N>: Number of request 9 <N>: Number of requests sent from Acceleration Engine N to FW and responses
10 Acceleration Engi 10 Acceleration Engine N received from FW
11 11
12 What: /sys/kernel/debug/qat_<device> 12 What: /sys/kernel/debug/qat_<device>_<BDF>/heartbeat/config
13 Date: November 2023 13 Date: November 2023
14 KernelVersion: 6.6 14 KernelVersion: 6.6
15 Contact: qat-linux@intel.com 15 Contact: qat-linux@intel.com
16 Description: (RW) Read returns value of the 16 Description: (RW) Read returns value of the Heartbeat update period.
17 Write to the file changes this 17 Write to the file changes this period value.
18 18
19 This period should reflect pla 19 This period should reflect planned polling interval of device
20 health status. High frequency 20 health status. High frequency Heartbeat monitoring wastes CPU cycles
21 but minimizes the customer’s 21 but minimizes the customer’s system downtime. Also, if there are
22 large service requests that ta 22 large service requests that take some time to complete, high frequency
23 Heartbeat monitoring could res 23 Heartbeat monitoring could result in false reports of unresponsiveness
24 and in those cases, period nee 24 and in those cases, period needs to be increased.
25 25
26 This parameter is effective on 26 This parameter is effective only for c3xxx, c62x, dh895xcc devices.
27 4xxx has this value internally 27 4xxx has this value internally fixed to 200ms.
28 28
29 Default value is set to 500. M 29 Default value is set to 500. Minimal allowed value is 200.
30 All values are expressed in mi 30 All values are expressed in milliseconds.
31 31
32 What: /sys/kernel/debug/qat_<device> 32 What: /sys/kernel/debug/qat_<device>_<BDF>/heartbeat/queries_failed
33 Date: November 2023 33 Date: November 2023
34 KernelVersion: 6.6 34 KernelVersion: 6.6
35 Contact: qat-linux@intel.com 35 Contact: qat-linux@intel.com
36 Description: (RO) Read returns the number o 36 Description: (RO) Read returns the number of times the device became unresponsive.
37 37
38 Attribute returns value of the 38 Attribute returns value of the counter which is incremented when
39 status query results negative. 39 status query results negative.
40 40
41 What: /sys/kernel/debug/qat_<device> 41 What: /sys/kernel/debug/qat_<device>_<BDF>/heartbeat/queries_sent
42 Date: November 2023 42 Date: November 2023
43 KernelVersion: 6.6 43 KernelVersion: 6.6
44 Contact: qat-linux@intel.com 44 Contact: qat-linux@intel.com
45 Description: (RO) Read returns the number o 45 Description: (RO) Read returns the number of times the control process checked
46 if the device is responsive. 46 if the device is responsive.
47 47
48 Attribute returns value of the 48 Attribute returns value of the counter which is incremented on
49 every status query. 49 every status query.
50 50
51 What: /sys/kernel/debug/qat_<device> 51 What: /sys/kernel/debug/qat_<device>_<BDF>/heartbeat/status
52 Date: November 2023 52 Date: November 2023
53 KernelVersion: 6.6 53 KernelVersion: 6.6
54 Contact: qat-linux@intel.com 54 Contact: qat-linux@intel.com
55 Description: (RO) Read returns the device h 55 Description: (RO) Read returns the device health status.
56 56
57 Returns 0 when device is healt 57 Returns 0 when device is healthy or -1 when is unresponsive
58 or the query failed to send. 58 or the query failed to send.
59 59
60 The driver does not monitor fo 60 The driver does not monitor for Heartbeat. It is left for a user
61 to poll the status periodicall 61 to poll the status periodically.
62 62
63 What: /sys/kernel/debug/qat_<device> 63 What: /sys/kernel/debug/qat_<device>_<BDF>/pm_status
64 Date: January 2024 64 Date: January 2024
65 KernelVersion: 6.7 65 KernelVersion: 6.7
66 Contact: qat-linux@intel.com 66 Contact: qat-linux@intel.com
67 Description: (RO) Read returns power manage 67 Description: (RO) Read returns power management information specific to the
68 QAT device. 68 QAT device.
69 69
70 This attribute is only availab 70 This attribute is only available for qat_4xxx devices.
71 71
72 What: /sys/kernel/debug/qat_<device> 72 What: /sys/kernel/debug/qat_<device>_<BDF>/cnv_errors
73 Date: January 2024 73 Date: January 2024
74 KernelVersion: 6.7 74 KernelVersion: 6.7
75 Contact: qat-linux@intel.com 75 Contact: qat-linux@intel.com
76 Description: (RO) Read returns, for each Ac 76 Description: (RO) Read returns, for each Acceleration Engine (AE), the number
77 of errors and the type of the 77 of errors and the type of the last error detected by the device
78 when performing verified compr 78 when performing verified compression.
79 Reported counters:: 79 Reported counters::
80 80
81 <N>: Number of Compres 81 <N>: Number of Compress and Verify (CnV) errors and type
82 of the last CnV e 82 of the last CnV error detected by Acceleration
83 Engine N. 83 Engine N.
84 84
85 What: /sys/kernel/debug/qat_<device> 85 What: /sys/kernel/debug/qat_<device>_<BDF>/heartbeat/inject_error
86 Date: March 2024 86 Date: March 2024
87 KernelVersion: 6.8 87 KernelVersion: 6.8
88 Contact: qat-linux@intel.com 88 Contact: qat-linux@intel.com
89 Description: (WO) Write to inject an error 89 Description: (WO) Write to inject an error that simulates an heartbeat
90 failure. This is to be used fo 90 failure. This is to be used for testing purposes.
91 91
92 After writing this file, the d 92 After writing this file, the driver stops arbitration on a
93 random engine and disables the 93 random engine and disables the fetching of heartbeat counters.
94 If a workload is running on th 94 If a workload is running on the device, a job submitted to the
95 accelerator might not get a re 95 accelerator might not get a response and a read of the
96 `heartbeat/status` attribute m 96 `heartbeat/status` attribute might report -1, i.e. device
97 unresponsive. 97 unresponsive.
98 The error is unrecoverable thu 98 The error is unrecoverable thus the device must be restarted to
99 restore its functionality. 99 restore its functionality.
100 100
101 This attribute is available on 101 This attribute is available only when the kernel is built with
102 CONFIG_CRYPTO_DEV_QAT_ERROR_IN 102 CONFIG_CRYPTO_DEV_QAT_ERROR_INJECTION=y.
103 103
104 A write of 1 enables error inj 104 A write of 1 enables error injection.
105 105
106 The following example shows ho 106 The following example shows how to enable error injection::
107 107
108 # cd /sys/kernel/debug 108 # cd /sys/kernel/debug/qat_<device>_<BDF>
109 # echo 1 > heartbeat/i 109 # echo 1 > heartbeat/inject_error
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.