1PERFEVENT.CONF(5) File Formats Manual PERFEVENT.CONF(5)
2
6 perfevent.conf - the Performance Co-Pilot perfevent PMDA configuration
7 file
8
10 $PCP_PMDAS_DIRS/perfevent/perfevent.conf
11
13 The perfevent.conf configuration file defines which hardware perfor‐
14 mance counters are enabled by the perfevent PMDA.
15 The configuration file contains a list of events that should be enabled
17 for each PMU. The perfevent PMDA detects which PMUs present and pro‐
18 grams the configured events. If multiple PMUs are available, the per‐
19 fevent PMDA uses the order of the entries in the configuration file to
20 determine which PMU settings to use. The entries at the top of the con‐
21 figuration file have higher precedence. Only one PMU configuration
22 setting is used. The configuration file may also contain perf derived
23 events. A derived event has a list of events which must be listed as
24 base events. The values for the counters in the derived events are ag‐
25 gregated and displayed.
26
28 The file consists of one or more groups of PMUs and the events that go
29 with them.
30 The PMU identification string is enclosed in brackets and one or more
32 events are listed below, one per line.
33 ["PMU identification string"]
35 EVENT_NAME [CPU OPTION]
36 ...
37 For derived events, one or more groups of events can be specified
39 where, counters in one group can be specific to one architecture and
40 the counters in other group can be specific to some other architecture.
41 Else, only one group can be specified. EVENT_NAME_1, EVENT_NAME_2 etc.
42 belong to group 1 and EVENT_NAME_3, EVENT_NAME_4 belong to group 2 and
43 so on. If any of the event in the first group is not available, entire
44 group will be discarded. And, the next group will be checked. The first
45 group with all the events available will be activated and monitored.
46 The derived perf event is enclosed in brackets along with :derived as a
47 suffix. Note that the EVENT_NAME listed under derived event must be
48 present as a base event in the configuration file. And, all the events
49 under a particular derived perf event group must have the same CPU_OP‐
50 TION. || acts as a separator for different groups inside a single de‐
51 rived event. The syntax is :
52 [event:derived]
54 EVENT_NAME_1 [CPU_OPTION] [scale]
55 EVENT_NAME_2 [CPU_OPTION] [scale]
56 ...
57 ||
58 EVENT_NAME_3 [CPU_OPTION] [scale]
59 EVENT_NAME_4 [CPU_OPTION] [scale]
60 ...
61 The available event cpu options are as follows:
64 cpu to add the counter to all cpus in the system
66 cpu_rr to add to one of the cpus in a round-robin allocation
68 strategy
69 node to add the counter to the first cpu in each numa node
71 node_rr
73 to add the counter to the cpus in each numa node in a
74 round-robin allocation strategy
75 [N] where N is the numerical digit of the CPU index
77 if the CPU option is absent then the event is programmed on all
79 cpus.
80 scale is applicable for the counters in derived events. If scale
82 is given, the counters will be scaled to their respective val‐
83 ues. The options are as follows :
84 a float value
86 Can be a simple float value, or a value in scientific no‐
87 tation.
88 perf_scale
90 The scale will be taken from the kernel sysfs scale file
91 entry for that event. Its generally specified in
92 /sys/bus/event_source/device/events/<event_name>.scale
93 If no scale is given, the default scale will be taken as 1.0.
95 Blank lines are ignored. Lines that begin with the # sign are
97 ignored.
98 Multiple, comma separated, PMUs may be specified in the PMU def‐
100 inition.
101
103 pmdaperfevent(1).
104Performance Co-Pilot PCP PERFEVENT.CONF(5)