6 perf-list - List all symbolic event types
11 'perf list' [--no-desc] [--long-desc]
12 [hw|sw|cache|tracepoint|pmu|sdt|metric|metricgroup|event_glob]
16 This command displays the symbolic event types which can be selected in the
17 various perf commands with the -e option.
23 Print extra event descriptions. (default)
26 Don't print descriptions.
30 Print longer event descriptions.
33 Enable debugging output.
36 Print how named events are resolved internally into perf events, and also
37 any extra expressions computed by perf stat.
40 Print deprecated events. By default the deprecated events are hidden.
46 Events can optionally have a modifier by appending a colon and one or
47 more modifiers. Modifiers allow the user to restrict the events to be
48 counted. The following modifiers exist:
50 u - user-space counting
52 h - hypervisor counting
54 G - guest counting (in KVM guests)
55 H - host counting (not in KVM guests)
57 P - use maximum detected precise level
58 S - read sample value (PERF_SAMPLE_READ)
59 D - pin the event to the PMU
60 W - group is weak and will fallback to non-group if not schedulable,
62 The 'p' modifier can be used for specifying how precise the instruction
63 address should be. The 'p' modifier can be specified multiple times:
65 0 - SAMPLE_IP can have arbitrary skid
66 1 - SAMPLE_IP must have constant skid
67 2 - SAMPLE_IP requested to have 0 skid
68 3 - SAMPLE_IP must have 0 skid, or uses randomization to avoid
69 sample shadowing effects.
71 For Intel systems precise event sampling is implemented with PEBS
72 which supports up to precise-level 2, and precise level 3 for
75 On AMD systems it is implemented using IBS (up to precise-level 2).
76 The precise modifier works with event types 0x76 (cpu-cycles, CPU
77 clocks not halted) and 0xC1 (micro-ops retired). Both events map to
78 IBS execution sampling (IBS op) with the IBS Op Counter Control bit
79 (IbsOpCntCtl) set respectively (see AMD64 Architecture Programmer’s
80 Manual Volume 2: System Programming, 13.3 Instruction-Based
81 Sampling). Examples to use IBS:
83 perf record -a -e cpu-cycles:p ... # use ibs op counting cycles
84 perf record -a -e r076:p ... # same as -e cpu-cycles:p
85 perf record -a -e r0C1:p ... # use ibs op counting micro-ops
87 RAW HARDWARE EVENT DESCRIPTOR
88 -----------------------------
89 Even when an event is not available in a symbolic form within perf right now,
90 it can be encoded in a per processor specific way.
92 For instance For x86 CPUs NNN represents the raw register encoding with the
93 layout of IA32_PERFEVTSELx MSRs (see [Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide] Figure 30-1 Layout
94 of IA32_PERFEVTSELx MSRs) or AMD's PerfEvtSeln (see [AMD64 Architecture Programmer’s Manual Volume 2: System Programming], Page 344,
95 Figure 13-7 Performance Event-Select Register (PerfEvtSeln)).
97 Note: Only the following bit fields can be set in x86 counter
98 registers: event, umask, edge, inv, cmask. Esp. guest/host only and
99 OS/user mode flags must be setup using <<EVENT_MODIFIERS, EVENT
104 If the Intel docs for a QM720 Core i7 describe an event as:
106 Event Umask Event Mask
107 Num. Value Mnemonic Description Comment
109 A8H 01H LSD.UOPS Counts the number of micro-ops Use cmask=1 and
110 delivered by loop stream detector invert to count
113 raw encoding of 0x1A8 can be used:
115 perf stat -e r1a8 -a sleep 1
116 perf record -e r1a8 ...
118 You should refer to the processor specific documentation for getting these
119 details. Some of them are referenced in the SEE ALSO section below.
124 perf also supports an extended syntax for specifying raw parameters
125 to PMUs. Using this typically requires looking up the specific event
126 in the CPU vendor specific documentation.
128 The available PMUs and their raw parameters can be listed with
130 ls /sys/devices/*/format
132 For example the raw event "LSD.UOPS" core pmu event above could
135 perf stat -e cpu/event=0xa8,umask=0x1,name=LSD.UOPS_CYCLES,cmask=0x1/ ...
137 or using extended name syntax
139 perf stat -e cpu/event=0xa8,umask=0x1,cmask=0x1,name=\'LSD.UOPS_CYCLES:cmask=0x1\'/ ...
144 Some PMUs are not associated with a core, but with a whole CPU socket.
145 Events on these PMUs generally cannot be sampled, but only counted globally
146 with perf stat -a. They can be bound to one logical CPU, but will measure
147 all the CPUs in the same socket.
149 This example measures memory bandwidth every second
150 on the first memory controller on socket 0 of a Intel Xeon system
152 perf stat -C 0 -a uncore_imc_0/cas_count_read/,uncore_imc_0/cas_count_write/ -I 1000 ...
154 Each memory controller has its own PMU. Measuring the complete system
155 bandwidth would require specifying all imc PMUs (see perf list output),
156 and adding the values together. To simplify creation of multiple events,
157 prefix and glob matching is supported in the PMU name, and the prefix
158 'uncore_' is also ignored when performing the match. So the command above
159 can be expanded to all memory controllers by using the syntaxes:
161 perf stat -C 0 -a imc/cas_count_read/,imc/cas_count_write/ -I 1000 ...
162 perf stat -C 0 -a *imc*/cas_count_read/,*imc*/cas_count_write/ -I 1000 ...
164 This example measures the combined core power every second
166 perf stat -I 1000 -e power/energy-cores/ -a
171 For non root users generally only context switched PMU events are available.
172 This is normally only the events in the cpu PMU, the predefined events
173 like cycles and instructions and some software events.
175 Other PMUs and global measurements are normally root only.
176 Some event qualifiers, such as "any", are also root only.
178 This can be overridden by setting the kernel.perf_event_paranoid
179 sysctl to -1, which allows non root to use these events.
181 For accessing trace point events perf needs to have read access to
182 /sys/kernel/debug/tracing, even when perf_event_paranoid is in a relaxed
188 Some PMUs control advanced hardware tracing capabilities, such as Intel PT,
189 that allows low overhead execution tracing. These are described in a separate
190 intel-pt.txt document.
195 Some pmu events listed by 'perf-list' will be displayed with '?' in them. For
198 hv_gpci/dtbp_ptitc,phys_processor_idx=?/
200 This means that when provided as an event, a value for '?' must
201 also be supplied. For example:
203 perf stat -C 0 -e 'hv_gpci/dtbp_ptitc,phys_processor_idx=0x2/' ...
207 It is also possible to add extra qualifiers to an event:
211 Sums up the event counts for all hardware threads in a core, e.g.:
214 perf stat -e cpu/event=0,umask=0x3,percore=1/
220 Perf supports time based multiplexing of events, when the number of events
221 active exceeds the number of hardware performance counters. Multiplexing
222 can cause measurement errors when the workload changes its execution
225 When metrics are computed using formulas from event counts, it is useful to
226 ensure some events are always measured together as a group to minimize multiplexing
227 errors. Event groups can be specified using { }.
229 perf stat -e '{instructions,cycles}' ...
231 The number of available performance counters depend on the CPU. A group
232 cannot contain more events than available counters.
233 For example Intel Core CPUs typically have four generic performance counters
234 for the core, plus three fixed counters for instructions, cycles and
235 ref-cycles. Some special events have restrictions on which counter they
236 can schedule, and may not support multiple instances in a single group.
237 When too many events are specified in the group some of them will not
240 Globally pinned events can limit the number of counters available for
241 other groups. On x86 systems, the NMI watchdog pins a counter by default.
242 The nmi watchdog can be disabled as root with
244 echo 0 > /proc/sys/kernel/nmi_watchdog
246 Events from multiple different PMUs cannot be mixed in a group, with
247 some exceptions for software events.
252 perf also supports group leader sampling using the :S specifier.
254 perf record -e '{cycles,instructions}:S' ...
257 Normally all events in an event group sample, but with :S only
258 the first event (the leader) samples, and it only reads the values of the
259 other events in the group.
264 Without options all known events will be listed.
266 To limit the list use:
268 . 'hw' or 'hardware' to list hardware events such as cache-misses, etc.
270 . 'sw' or 'software' to list software events such as context switches, etc.
272 . 'cache' or 'hwcache' to list hardware cache events such as L1-dcache-loads, etc.
274 . 'tracepoint' to list all tracepoint events, alternatively use
275 'subsys_glob:event_glob' to filter by tracepoint subsystems such as sched,
278 . 'pmu' to print the kernel supplied PMU events.
280 . 'sdt' to list all Statically Defined Tracepoint events.
282 . 'metric' to list metrics
284 . 'metricgroup' to list metricgroups with metrics.
286 . If none of the above is matched, it will apply the supplied glob to all
287 events, printing the ones that match.
289 . As a last resort, it will do a substring search in all event names.
291 One or more types can be used at the same time, listing the events for the
296 . '--raw-dump', shows the raw-dump of all the events.
297 . '--raw-dump [hw|sw|cache|tracepoint|pmu|event_glob]', shows the raw-dump of
298 a certain kind of events.
302 linkperf:perf-stat[1], linkperf:perf-top[1],
303 linkperf:perf-record[1],
304 http://www.intel.com/sdm/[Intel® 64 and IA-32 Architectures Software Developer's Manual Volume 3B: System Programming Guide],
305 http://support.amd.com/us/Processor_TechDocs/24593_APM_v2.pdf[AMD64 Architecture Programmer’s Manual Volume 2: System Programming]