1 ====================================== 1 ======================================
2 HNS3 Performance Monitoring Unit (PMU) 2 HNS3 Performance Monitoring Unit (PMU)
3 ====================================== 3 ======================================
4 4
5 HNS3(HiSilicon network system 3) Performance M 5 HNS3(HiSilicon network system 3) Performance Monitoring Unit (PMU) is an
6 End Point device to collect performance statis 6 End Point device to collect performance statistics of HiSilicon SoC NIC.
7 On Hip09, each SICL(Super I/O cluster) has one 7 On Hip09, each SICL(Super I/O cluster) has one PMU device.
8 8
9 HNS3 PMU supports collection of performance st 9 HNS3 PMU supports collection of performance statistics such as bandwidth,
10 latency, packet rate and interrupt rate. 10 latency, packet rate and interrupt rate.
11 11
12 Each HNS3 PMU supports 8 hardware events. 12 Each HNS3 PMU supports 8 hardware events.
13 13
14 HNS3 PMU driver 14 HNS3 PMU driver
15 =============== 15 ===============
16 16
17 The HNS3 PMU driver registers a perf PMU with 17 The HNS3 PMU driver registers a perf PMU with the name of its sicl id.::
18 18
19 /sys/bus/event_source/devices/hns3_pmu_sicl_ !! 19 /sys/devices/hns3_pmu_sicl_<sicl_id>
20 20
21 PMU driver provides description of available e 21 PMU driver provides description of available events, filter modes, format,
22 identifier and cpumask in sysfs. 22 identifier and cpumask in sysfs.
23 23
24 The "events" directory describes the event cod 24 The "events" directory describes the event code of all supported events
25 shown in perf list. 25 shown in perf list.
26 26
27 The "filtermode" directory describes the suppo 27 The "filtermode" directory describes the supported filter modes of each
28 event. 28 event.
29 29
30 The "format" directory describes all formats o 30 The "format" directory describes all formats of the config (events) and
31 config1 (filter options) fields of the perf_ev 31 config1 (filter options) fields of the perf_event_attr structure.
32 32
33 The "identifier" file shows version of PMU har 33 The "identifier" file shows version of PMU hardware device.
34 34
35 The "bdf_min" and "bdf_max" files show the sup 35 The "bdf_min" and "bdf_max" files show the supported bdf range of each
36 pmu device. 36 pmu device.
37 37
38 The "hw_clk_freq" file shows the hardware cloc 38 The "hw_clk_freq" file shows the hardware clock frequency of each pmu
39 device. 39 device.
40 40
41 Example usage of checking event code and subev 41 Example usage of checking event code and subevent code::
42 42
43 $# cat /sys/bus/event_source/devices/hns3_pm !! 43 $# cat /sys/devices/hns3_pmu_sicl_0/events/dly_tx_normal_to_mac_time
44 config=0x00204 44 config=0x00204
45 $# cat /sys/bus/event_source/devices/hns3_pm !! 45 $# cat /sys/devices/hns3_pmu_sicl_0/events/dly_tx_normal_to_mac_packet_num
46 config=0x10204 46 config=0x10204
47 47
48 Each performance statistic has a pair of event 48 Each performance statistic has a pair of events to get two values to
49 calculate real performance data in userspace. 49 calculate real performance data in userspace.
50 50
51 The bits 0~15 of config (here 0x0204) are the 51 The bits 0~15 of config (here 0x0204) are the true hardware event code. If
52 two events have same value of bits 0~15 of con 52 two events have same value of bits 0~15 of config, that means they are
53 event pair. And the bit 16 of config indicates 53 event pair. And the bit 16 of config indicates getting counter 0 or
54 counter 1 of hardware event. 54 counter 1 of hardware event.
55 55
56 After getting two values of event pair in user 56 After getting two values of event pair in userspace, the formula of
57 computation to calculate real performance data 57 computation to calculate real performance data is:::
58 58
59 counter 0 / counter 1 59 counter 0 / counter 1
60 60
61 Example usage of checking supported filter mod 61 Example usage of checking supported filter mode::
62 62
63 $# cat /sys/bus/event_source/devices/hns3_pm !! 63 $# cat /sys/devices/hns3_pmu_sicl_0/filtermode/bw_ssu_rpu_byte_num
64 filter mode supported: global/port/port-tc/f 64 filter mode supported: global/port/port-tc/func/func-queue/
65 65
66 Example usage of perf:: 66 Example usage of perf::
67 67
68 $# perf list 68 $# perf list
69 hns3_pmu_sicl_0/bw_ssu_rpu_byte_num/ [kernel 69 hns3_pmu_sicl_0/bw_ssu_rpu_byte_num/ [kernel PMU event]
70 hns3_pmu_sicl_0/bw_ssu_rpu_time/ [kernel 70 hns3_pmu_sicl_0/bw_ssu_rpu_time/ [kernel PMU event]
71 ------------------------------------------ 71 ------------------------------------------
72 72
73 $# perf stat -g -e hns3_pmu_sicl_0/bw_ssu_rp 73 $# perf stat -g -e hns3_pmu_sicl_0/bw_ssu_rpu_byte_num,global=1/ -e hns3_pmu_sicl_0/bw_ssu_rpu_time,global=1/ -I 1000
74 or 74 or
75 $# perf stat -g -e hns3_pmu_sicl_0/config=0x 75 $# perf stat -g -e hns3_pmu_sicl_0/config=0x00002,global=1/ -e hns3_pmu_sicl_0/config=0x10002,global=1/ -I 1000
76 76
77 77
78 Filter modes 78 Filter modes
79 -------------- 79 --------------
80 80
81 1. global mode 81 1. global mode
82 PMU collect performance statistics for all HNS 82 PMU collect performance statistics for all HNS3 PCIe functions of IO DIE.
83 Set the "global" filter option to 1 will enabl 83 Set the "global" filter option to 1 will enable this mode.
84 Example usage of perf:: 84 Example usage of perf::
85 85
86 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 86 $# perf stat -a -e hns3_pmu_sicl_0/config=0x1020F,global=1/ -I 1000
87 87
88 2. port mode 88 2. port mode
89 PMU collect performance statistic of one whole 89 PMU collect performance statistic of one whole physical port. The port id
90 is same as mac id. The "tc" filter option must 90 is same as mac id. The "tc" filter option must be set to 0xF in this mode,
91 here tc stands for traffic class. 91 here tc stands for traffic class.
92 92
93 Example usage of perf:: 93 Example usage of perf::
94 94
95 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 95 $# perf stat -a -e hns3_pmu_sicl_0/config=0x1020F,port=0,tc=0xF/ -I 1000
96 96
97 3. port-tc mode 97 3. port-tc mode
98 PMU collect performance statistic of one tc of 98 PMU collect performance statistic of one tc of physical port. The port id
99 is same as mac id. The "tc" filter option must 99 is same as mac id. The "tc" filter option must be set to 0 ~ 7 in this
100 mode. 100 mode.
101 Example usage of perf:: 101 Example usage of perf::
102 102
103 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 103 $# perf stat -a -e hns3_pmu_sicl_0/config=0x1020F,port=0,tc=0/ -I 1000
104 104
105 4. func mode 105 4. func mode
106 PMU collect performance statistic of one PF/VF 106 PMU collect performance statistic of one PF/VF. The function id is BDF of
107 PF/VF, its conversion formula:: 107 PF/VF, its conversion formula::
108 108
109 func = (bus << 8) + (device << 3) + (functio 109 func = (bus << 8) + (device << 3) + (function)
110 110
111 for example: 111 for example:
112 BDF func 112 BDF func
113 35:00.0 0x3500 113 35:00.0 0x3500
114 35:00.1 0x3501 114 35:00.1 0x3501
115 35:01.0 0x3508 115 35:01.0 0x3508
116 116
117 In this mode, the "queue" filter option must b 117 In this mode, the "queue" filter option must be set to 0xFFFF.
118 Example usage of perf:: 118 Example usage of perf::
119 119
120 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 120 $# perf stat -a -e hns3_pmu_sicl_0/config=0x1020F,bdf=0x3500,queue=0xFFFF/ -I 1000
121 121
122 5. func-queue mode 122 5. func-queue mode
123 PMU collect performance statistic of one queue 123 PMU collect performance statistic of one queue of PF/VF. The function id
124 is BDF of PF/VF, the "queue" filter option mus 124 is BDF of PF/VF, the "queue" filter option must be set to the exact queue
125 id of function. 125 id of function.
126 Example usage of perf:: 126 Example usage of perf::
127 127
128 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 128 $# perf stat -a -e hns3_pmu_sicl_0/config=0x1020F,bdf=0x3500,queue=0/ -I 1000
129 129
130 6. func-intr mode 130 6. func-intr mode
131 PMU collect performance statistic of one inter 131 PMU collect performance statistic of one interrupt of PF/VF. The function
132 id is BDF of PF/VF, the "intr" filter option m 132 id is BDF of PF/VF, the "intr" filter option must be set to the exact
133 interrupt id of function. 133 interrupt id of function.
134 Example usage of perf:: 134 Example usage of perf::
135 135
136 $# perf stat -a -e hns3_pmu_sicl_0/config=0x 136 $# perf stat -a -e hns3_pmu_sicl_0/config=0x00301,bdf=0x3500,intr=0/ -I 1000
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.