Commit | Line | Data |
---|---|---|
832c927d TN |
1 | APM X-Gene SoC Performance Monitoring Unit (PMU) |
2 | ================================================ | |
3 | ||
4 | X-Gene SoC PMU consists of various independent system device PMUs such as | |
5 | L3 cache(s), I/O bridge(s), memory controller bridge(s) and memory | |
6 | controller(s). These PMU devices are loosely architected to follow the | |
7 | same model as the PMU for ARM cores. The PMUs share the same top level | |
8 | interrupt and status CSR region. | |
9 | ||
10 | PMU (perf) driver | |
11 | ----------------- | |
12 | ||
13 | The xgene-pmu driver registers several perf PMU drivers. Each of the perf | |
14 | driver provides description of its available events and configuration options | |
15 | in sysfs, see /sys/devices/<l3cX/iobX/mcbX/mcX>/. | |
16 | ||
17 | The "format" directory describes format of the config (event ID), | |
18 | config1 (agent ID) fields of the perf_event_attr structure. The "events" | |
19 | directory provides configuration templates for all supported event types that | |
20 | can be used with perf tool. For example, "l3c0/bank-fifo-full/" is an | |
21 | equivalent of "l3c0/config=0x0b/". | |
22 | ||
23 | Most of the SoC PMU has a specific list of agent ID used for monitoring | |
24 | performance of a specific datapath. For example, agents of a L3 cache can be | |
25 | a specific CPU or an I/O bridge. Each PMU has a set of 2 registers capable of | |
26 | masking the agents from which the request come from. If the bit with | |
27 | the bit number corresponding to the agent is set, the event is counted only if | |
28 | it is caused by a request from that agent. Each agent ID bit is inversely mapped | |
29 | to a corresponding bit in "config1" field. By default, the event will be | |
30 | counted for all agent requests (config1 = 0x0). For all the supported agents of | |
31 | each PMU, please refer to APM X-Gene User Manual. | |
32 | ||
33 | Each perf driver also provides a "cpumask" sysfs attribute, which contains a | |
34 | single CPU ID of the processor which will be used to handle all the PMU events. | |
35 | ||
36 | Example for perf tool use: | |
37 | ||
38 | / # perf list | grep -e l3c -e iob -e mcb -e mc | |
39 | l3c0/ackq-full/ [Kernel PMU event] | |
40 | <...> | |
41 | mcb1/mcb-csw-stall/ [Kernel PMU event] | |
42 | ||
43 | / # perf stat -a -e l3c0/read-miss/,mcb1/csw-write-request/ sleep 1 | |
44 | ||
45 | / # perf stat -a -e l3c0/read-miss,config1=0xfffffffffffffffe/ sleep 1 | |
46 | ||
47 | The driver does not support sampling, therefore "perf record" will | |
48 | not work. Per-task (without "-a") perf sessions are not supported. |