Loading...
1perf-report(1)
2==============
3
4NAME
5----
6perf-report - Read perf.data (created by perf record) and display the profile
7
8SYNOPSIS
9--------
10[verse]
11'perf report' [-i <file> | --input=file]
12
13DESCRIPTION
14-----------
15This command displays the performance counter profile information recorded
16via perf record.
17
18OPTIONS
19-------
20-i::
21--input=::
22 Input file name. (default: perf.data unless stdin is a fifo)
23
24-v::
25--verbose::
26 Be more verbose. (show symbol address, etc)
27
28-q::
29--quiet::
30 Do not show any message. (Suppress -v)
31
32-n::
33--show-nr-samples::
34 Show the number of samples for each symbol
35
36--show-cpu-utilization::
37 Show sample percentage for different cpu modes.
38
39-T::
40--threads::
41 Show per-thread event counters. The input data file should be recorded
42 with -s option.
43-c::
44--comms=::
45 Only consider symbols in these comms. CSV that understands
46 file://filename entries. This option will affect the percentage of
47 the overhead column. See --percentage for more info.
48--pid=::
49 Only show events for given process ID (comma separated list).
50
51--tid=::
52 Only show events for given thread ID (comma separated list).
53-d::
54--dsos=::
55 Only consider symbols in these dsos. CSV that understands
56 file://filename entries. This option will affect the percentage of
57 the overhead column. See --percentage for more info.
58-S::
59--symbols=::
60 Only consider these symbols. CSV that understands
61 file://filename entries. This option will affect the percentage of
62 the overhead column. See --percentage for more info.
63
64--symbol-filter=::
65 Only show symbols that match (partially) with this filter.
66
67-U::
68--hide-unresolved::
69 Only display entries resolved to a symbol.
70
71-s::
72--sort=::
73 Sort histogram entries by given key(s) - multiple keys can be specified
74 in CSV format. Following sort keys are available:
75 pid, comm, dso, symbol, parent, cpu, socket, srcline, weight,
76 local_weight, cgroup_id.
77
78 Each key has following meaning:
79
80 - comm: command (name) of the task which can be read via /proc/<pid>/comm
81 - pid: command and tid of the task
82 - dso: name of library or module executed at the time of sample
83 - dso_size: size of library or module executed at the time of sample
84 - symbol: name of function executed at the time of sample
85 - symbol_size: size of function executed at the time of sample
86 - parent: name of function matched to the parent regex filter. Unmatched
87 entries are displayed as "[other]".
88 - cpu: cpu number the task ran at the time of sample
89 - socket: processor socket number the task ran at the time of sample
90 - srcline: filename and line number executed at the time of sample. The
91 DWARF debugging info must be provided.
92 - srcfile: file name of the source file of the samples. Requires dwarf
93 information.
94 - weight: Event specific weight, e.g. memory latency or transaction
95 abort cost. This is the global weight.
96 - local_weight: Local weight version of the weight above.
97 - cgroup_id: ID derived from cgroup namespace device and inode numbers.
98 - transaction: Transaction abort flags.
99 - overhead: Overhead percentage of sample
100 - overhead_sys: Overhead percentage of sample running in system mode
101 - overhead_us: Overhead percentage of sample running in user mode
102 - overhead_guest_sys: Overhead percentage of sample running in system mode
103 on guest machine
104 - overhead_guest_us: Overhead percentage of sample running in user mode on
105 guest machine
106 - sample: Number of sample
107 - period: Raw number of event count of sample
108 - time: Separate the samples by time stamp with the resolution specified by
109 --time-quantum (default 100ms). Specify with overhead and before it.
110
111 By default, comm, dso and symbol keys are used.
112 (i.e. --sort comm,dso,symbol)
113
114 If --branch-stack option is used, following sort keys are also
115 available:
116
117 - dso_from: name of library or module branched from
118 - dso_to: name of library or module branched to
119 - symbol_from: name of function branched from
120 - symbol_to: name of function branched to
121 - srcline_from: source file and line branched from
122 - srcline_to: source file and line branched to
123 - mispredict: "N" for predicted branch, "Y" for mispredicted branch
124 - in_tx: branch in TSX transaction
125 - abort: TSX transaction abort.
126 - cycles: Cycles in basic block
127
128 And default sort keys are changed to comm, dso_from, symbol_from, dso_to
129 and symbol_to, see '--branch-stack'.
130
131 When the sort key symbol is specified, columns "IPC" and "IPC Coverage"
132 are enabled automatically. Column "IPC" reports the average IPC per function
133 and column "IPC coverage" reports the percentage of instructions with
134 sampled IPC in this function. IPC means Instruction Per Cycle. If it's low,
135 it indicates there may be a performance bottleneck when the function is
136 executed, such as a memory access bottleneck. If a function has high overhead
137 and low IPC, it's worth further analyzing it to optimize its performance.
138
139 If the --mem-mode option is used, the following sort keys are also available
140 (incompatible with --branch-stack):
141 symbol_daddr, dso_daddr, locked, tlb, mem, snoop, dcacheline.
142
143 - symbol_daddr: name of data symbol being executed on at the time of sample
144 - dso_daddr: name of library or module containing the data being executed
145 on at the time of the sample
146 - locked: whether the bus was locked at the time of the sample
147 - tlb: type of tlb access for the data at the time of the sample
148 - mem: type of memory access for the data at the time of the sample
149 - snoop: type of snoop (if any) for the data at the time of the sample
150 - dcacheline: the cacheline the data address is on at the time of the sample
151 - phys_daddr: physical address of data being executed on at the time of sample
152
153 And the default sort keys are changed to local_weight, mem, sym, dso,
154 symbol_daddr, dso_daddr, snoop, tlb, locked, see '--mem-mode'.
155
156 If the data file has tracepoint event(s), following (dynamic) sort keys
157 are also available:
158 trace, trace_fields, [<event>.]<field>[/raw]
159
160 - trace: pretty printed trace output in a single column
161 - trace_fields: fields in tracepoints in separate columns
162 - <field name>: optional event and field name for a specific field
163
164 The last form consists of event and field names. If event name is
165 omitted, it searches all events for matching field name. The matched
166 field will be shown only for the event has the field. The event name
167 supports substring match so user doesn't need to specify full subsystem
168 and event name everytime. For example, 'sched:sched_switch' event can
169 be shortened to 'switch' as long as it's not ambiguous. Also event can
170 be specified by its index (starting from 1) preceded by the '%'.
171 So '%1' is the first event, '%2' is the second, and so on.
172
173 The field name can have '/raw' suffix which disables pretty printing
174 and shows raw field value like hex numbers. The --raw-trace option
175 has the same effect for all dynamic sort keys.
176
177 The default sort keys are changed to 'trace' if all events in the data
178 file are tracepoint.
179
180-F::
181--fields=::
182 Specify output field - multiple keys can be specified in CSV format.
183 Following fields are available:
184 overhead, overhead_sys, overhead_us, overhead_children, sample and period.
185 Also it can contain any sort key(s).
186
187 By default, every sort keys not specified in -F will be appended
188 automatically.
189
190 If the keys starts with a prefix '+', then it will append the specified
191 field(s) to the default field order. For example: perf report -F +period,sample.
192
193-p::
194--parent=<regex>::
195 A regex filter to identify parent. The parent is a caller of this
196 function and searched through the callchain, thus it requires callchain
197 information recorded. The pattern is in the extended regex format and
198 defaults to "\^sys_|^do_page_fault", see '--sort parent'.
199
200-x::
201--exclude-other::
202 Only display entries with parent-match.
203
204-w::
205--column-widths=<width[,width...]>::
206 Force each column width to the provided list, for large terminal
207 readability. 0 means no limit (default behavior).
208
209-t::
210--field-separator=::
211 Use a special separator character and don't pad with spaces, replacing
212 all occurrences of this separator in symbol names (and other output)
213 with a '.' character, that thus it's the only non valid separator.
214
215-D::
216--dump-raw-trace::
217 Dump raw trace in ASCII.
218
219-g::
220--call-graph=<print_type,threshold[,print_limit],order,sort_key[,branch],value>::
221 Display call chains using type, min percent threshold, print limit,
222 call order, sort key, optional branch and value. Note that ordering
223 is not fixed so any parameter can be given in an arbitrary order.
224 One exception is the print_limit which should be preceded by threshold.
225
226 print_type can be either:
227 - flat: single column, linear exposure of call chains.
228 - graph: use a graph tree, displaying absolute overhead rates. (default)
229 - fractal: like graph, but displays relative rates. Each branch of
230 the tree is considered as a new profiled object.
231 - folded: call chains are displayed in a line, separated by semicolons
232 - none: disable call chain display.
233
234 threshold is a percentage value which specifies a minimum percent to be
235 included in the output call graph. Default is 0.5 (%).
236
237 print_limit is only applied when stdio interface is used. It's to limit
238 number of call graph entries in a single hist entry. Note that it needs
239 to be given after threshold (but not necessarily consecutive).
240 Default is 0 (unlimited).
241
242 order can be either:
243 - callee: callee based call graph.
244 - caller: inverted caller based call graph.
245 Default is 'caller' when --children is used, otherwise 'callee'.
246
247 sort_key can be:
248 - function: compare on functions (default)
249 - address: compare on individual code addresses
250 - srcline: compare on source filename and line number
251
252 branch can be:
253 - branch: include last branch information in callgraph when available.
254 Usually more convenient to use --branch-history for this.
255
256 value can be:
257 - percent: display overhead percent (default)
258 - period: display event period
259 - count: display event count
260
261--children::
262 Accumulate callchain of children to parent entry so that then can
263 show up in the output. The output will have a new "Children" column
264 and will be sorted on the data. It requires callchains are recorded.
265 See the `overhead calculation' section for more details. Enabled by
266 default, disable with --no-children.
267
268--max-stack::
269 Set the stack depth limit when parsing the callchain, anything
270 beyond the specified depth will be ignored. This is a trade-off
271 between information loss and faster processing especially for
272 workloads that can have a very long callchain stack.
273 Note that when using the --itrace option the synthesized callchain size
274 will override this value if the synthesized callchain size is bigger.
275
276 Default: 127
277
278-G::
279--inverted::
280 alias for inverted caller based call graph.
281
282--ignore-callees=<regex>::
283 Ignore callees of the function(s) matching the given regex.
284 This has the effect of collecting the callers of each such
285 function into one place in the call-graph tree.
286
287--pretty=<key>::
288 Pretty printing style. key: normal, raw
289
290--stdio:: Use the stdio interface.
291
292--stdio-color::
293 'always', 'never' or 'auto', allowing configuring color output
294 via the command line, in addition to via "color.ui" .perfconfig.
295 Use '--stdio-color always' to generate color even when redirecting
296 to a pipe or file. Using just '--stdio-color' is equivalent to
297 using 'always'.
298
299--tui:: Use the TUI interface, that is integrated with annotate and allows
300 zooming into DSOs or threads, among other features. Use of --tui
301 requires a tty, if one is not present, as when piping to other
302 commands, the stdio interface is used.
303
304--gtk:: Use the GTK2 interface.
305
306-k::
307--vmlinux=<file>::
308 vmlinux pathname
309
310--ignore-vmlinux::
311 Ignore vmlinux files.
312
313--kallsyms=<file>::
314 kallsyms pathname
315
316-m::
317--modules::
318 Load module symbols. WARNING: This should only be used with -k and
319 a LIVE kernel.
320
321-f::
322--force::
323 Don't do ownership validation.
324
325--symfs=<directory>::
326 Look for files with symbols relative to this directory.
327
328-C::
329--cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
330 be provided as a comma-separated list with no space: 0,1. Ranges of
331 CPUs are specified with -: 0-2. Default is to report samples on all
332 CPUs.
333
334-M::
335--disassembler-style=:: Set disassembler style for objdump.
336
337--source::
338 Interleave source code with assembly code. Enabled by default,
339 disable with --no-source.
340
341--asm-raw::
342 Show raw instruction encoding of assembly instructions.
343
344--show-total-period:: Show a column with the sum of periods.
345
346-I::
347--show-info::
348 Display extended information about the perf.data file. This adds
349 information which may be very large and thus may clutter the display.
350 It currently includes: cpu and numa topology of the host system.
351
352-b::
353--branch-stack::
354 Use the addresses of sampled taken branches instead of the instruction
355 address to build the histograms. To generate meaningful output, the
356 perf.data file must have been obtained using perf record -b or
357 perf record --branch-filter xxx where xxx is a branch filter option.
358 perf report is able to auto-detect whether a perf.data file contains
359 branch stacks and it will automatically switch to the branch view mode,
360 unless --no-branch-stack is used.
361
362--branch-history::
363 Add the addresses of sampled taken branches to the callstack.
364 This allows to examine the path the program took to each sample.
365 The data collection must have used -b (or -j) and -g.
366
367--objdump=<path>::
368 Path to objdump binary.
369
370--group::
371 Show event group information together. It forces group output also
372 if there are no groups defined in data file.
373
374--demangle::
375 Demangle symbol names to human readable form. It's enabled by default,
376 disable with --no-demangle.
377
378--demangle-kernel::
379 Demangle kernel symbol names to human readable form (for C++ kernels).
380
381--mem-mode::
382 Use the data addresses of samples in addition to instruction addresses
383 to build the histograms. To generate meaningful output, the perf.data
384 file must have been obtained using perf record -d -W and using a
385 special event -e cpu/mem-loads/p or -e cpu/mem-stores/p. See
386 'perf mem' for simpler access.
387
388--percent-limit::
389 Do not show entries which have an overhead under that percent.
390 (Default: 0). Note that this option also sets the percent limit (threshold)
391 of callchains. However the default value of callchain threshold is
392 different than the default value of hist entries. Please see the
393 --call-graph option for details.
394
395--percentage::
396 Determine how to display the overhead percentage of filtered entries.
397 Filters can be applied by --comms, --dsos and/or --symbols options and
398 Zoom operations on the TUI (thread, dso, etc).
399
400 "relative" means it's relative to filtered entries only so that the
401 sum of shown entries will be always 100%. "absolute" means it retains
402 the original value before and after the filter is applied.
403
404--header::
405 Show header information in the perf.data file. This includes
406 various information like hostname, OS and perf version, cpu/mem
407 info, perf command line, event list and so on. Currently only
408 --stdio output supports this feature.
409
410--header-only::
411 Show only perf.data header (forces --stdio).
412
413--time::
414 Only analyze samples within given time window: <start>,<stop>. Times
415 have the format seconds.nanoseconds. If start is not given (i.e. time
416 string is ',x.y') then analysis starts at the beginning of the file. If
417 stop time is not given (i.e. time string is 'x.y,') then analysis goes
418 to end of file. Multiple ranges can be separated by spaces, which
419 requires the argument to be quoted e.g. --time "1234.567,1234.789 1235,"
420
421 Also support time percent with multiple time ranges. Time string is
422 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
423
424 For example:
425 Select the second 10% time slice:
426
427 perf report --time 10%/2
428
429 Select from 0% to 10% time slice:
430
431 perf report --time 0%-10%
432
433 Select the first and second 10% time slices:
434
435 perf report --time 10%/1,10%/2
436
437 Select from 0% to 10% and 30% to 40% slices:
438
439 perf report --time 0%-10%,30%-40%
440
441--switch-on EVENT_NAME::
442 Only consider events after this event is found.
443
444 This may be interesting to measure a workload only after some initialization
445 phase is over, i.e. insert a perf probe at that point and then using this
446 option with that probe.
447
448--switch-off EVENT_NAME::
449 Stop considering events after this event is found.
450
451--show-on-off-events::
452 Show the --switch-on/off events too. This has no effect in 'perf report' now
453 but probably we'll make the default not to show the switch-on/off events
454 on the --group mode and if there is only one event besides the off/on ones,
455 go straight to the histogram browser, just like 'perf report' with no events
456 explicitely specified does.
457
458--itrace::
459 Options for decoding instruction tracing data. The options are:
460
461include::itrace.txt[]
462
463 To disable decoding entirely, use --no-itrace.
464
465--full-source-path::
466 Show the full path for source files for srcline output.
467
468--show-ref-call-graph::
469 When multiple events are sampled, it may not be needed to collect
470 callgraphs for all of them. The sample sites are usually nearby,
471 and it's enough to collect the callgraphs on a reference event.
472 So user can use "call-graph=no" event modifier to disable callgraph
473 for other events to reduce the overhead.
474 However, perf report cannot show callgraphs for the event which
475 disable the callgraph.
476 This option extends the perf report to show reference callgraphs,
477 which collected by reference event, in no callgraph event.
478
479--socket-filter::
480 Only report the samples on the processor socket that match with this filter
481
482--samples=N::
483 Save N individual samples for each histogram entry to show context in perf
484 report tui browser.
485
486--raw-trace::
487 When displaying traceevent output, do not use print fmt or plugins.
488
489--hierarchy::
490 Enable hierarchical output.
491
492--inline::
493 If a callgraph address belongs to an inlined function, the inline stack
494 will be printed. Each entry is function name or file/line. Enabled by
495 default, disable with --no-inline.
496
497--mmaps::
498 Show --tasks output plus mmap information in a format similar to
499 /proc/<PID>/maps.
500
501 Please note that not all mmaps are stored, options affecting which ones
502 are include 'perf record --data', for instance.
503
504--ns::
505 Show time stamps in nanoseconds.
506
507--stats::
508 Display overall events statistics without any further processing.
509 (like the one at the end of the perf report -D command)
510
511--tasks::
512 Display monitored tasks stored in perf data. Displaying pid/tid/ppid
513 plus the command string aligned to distinguish parent and child tasks.
514
515--percent-type::
516 Set annotation percent type from following choices:
517 global-period, local-period, global-hits, local-hits
518
519 The local/global keywords set if the percentage is computed
520 in the scope of the function (local) or the whole data (global).
521 The period/hits keywords set the base the percentage is computed
522 on - the samples period or the number of samples (hits).
523
524--time-quantum::
525 Configure time quantum for time sort key. Default 100ms.
526 Accepts s, us, ms, ns units.
527
528include::callchain-overhead-calculation.txt[]
529
530SEE ALSO
531--------
532linkperf:perf-stat[1], linkperf:perf-annotate[1], linkperf:perf-record[1]
1perf-report(1)
2==============
3
4NAME
5----
6perf-report - Read perf.data (created by perf record) and display the profile
7
8SYNOPSIS
9--------
10[verse]
11'perf report' [-i <file> | --input=file]
12
13DESCRIPTION
14-----------
15This command displays the performance counter profile information recorded
16via perf record.
17
18OPTIONS
19-------
20-i::
21--input=::
22 Input file name. (default: perf.data unless stdin is a fifo)
23
24-v::
25--verbose::
26 Be more verbose. (show symbol address, etc)
27
28-q::
29--quiet::
30 Do not show any message. (Suppress -v)
31
32-n::
33--show-nr-samples::
34 Show the number of samples for each symbol
35
36--show-cpu-utilization::
37 Show sample percentage for different cpu modes.
38
39-T::
40--threads::
41 Show per-thread event counters. The input data file should be recorded
42 with -s option.
43-c::
44--comms=::
45 Only consider symbols in these comms. CSV that understands
46 file://filename entries. This option will affect the percentage of
47 the overhead column. See --percentage for more info.
48--pid=::
49 Only show events for given process ID (comma separated list).
50
51--tid=::
52 Only show events for given thread ID (comma separated list).
53-d::
54--dsos=::
55 Only consider symbols in these dsos. CSV that understands
56 file://filename entries. This option will affect the percentage of
57 the overhead column. See --percentage for more info.
58-S::
59--symbols=::
60 Only consider these symbols. CSV that understands
61 file://filename entries. This option will affect the percentage of
62 the overhead column. See --percentage for more info.
63
64--symbol-filter=::
65 Only show symbols that match (partially) with this filter.
66
67-U::
68--hide-unresolved::
69 Only display entries resolved to a symbol.
70
71-s::
72--sort=::
73 Sort histogram entries by given key(s) - multiple keys can be specified
74 in CSV format. Following sort keys are available:
75 pid, comm, dso, symbol, parent, cpu, socket, srcline, weight,
76 local_weight, cgroup_id.
77
78 Each key has following meaning:
79
80 - comm: command (name) of the task which can be read via /proc/<pid>/comm
81 - pid: command and tid of the task
82 - dso: name of library or module executed at the time of sample
83 - dso_size: size of library or module executed at the time of sample
84 - symbol: name of function executed at the time of sample
85 - symbol_size: size of function executed at the time of sample
86 - parent: name of function matched to the parent regex filter. Unmatched
87 entries are displayed as "[other]".
88 - cpu: cpu number the task ran at the time of sample
89 - socket: processor socket number the task ran at the time of sample
90 - srcline: filename and line number executed at the time of sample. The
91 DWARF debugging info must be provided.
92 - srcfile: file name of the source file of the samples. Requires dwarf
93 information.
94 - weight: Event specific weight, e.g. memory latency or transaction
95 abort cost. This is the global weight.
96 - local_weight: Local weight version of the weight above.
97 - cgroup_id: ID derived from cgroup namespace device and inode numbers.
98 - cgroup: cgroup pathname in the cgroupfs.
99 - transaction: Transaction abort flags.
100 - overhead: Overhead percentage of sample
101 - overhead_sys: Overhead percentage of sample running in system mode
102 - overhead_us: Overhead percentage of sample running in user mode
103 - overhead_guest_sys: Overhead percentage of sample running in system mode
104 on guest machine
105 - overhead_guest_us: Overhead percentage of sample running in user mode on
106 guest machine
107 - sample: Number of sample
108 - period: Raw number of event count of sample
109 - time: Separate the samples by time stamp with the resolution specified by
110 --time-quantum (default 100ms). Specify with overhead and before it.
111 - code_page_size: the code page size of sampled code address (ip)
112 - ins_lat: Instruction latency in core cycles. This is the global instruction
113 latency
114 - local_ins_lat: Local instruction latency version
115 - p_stage_cyc: On powerpc, this presents the number of cycles spent in a
116 pipeline stage. And currently supported only on powerpc.
117
118 By default, comm, dso and symbol keys are used.
119 (i.e. --sort comm,dso,symbol)
120
121 If --branch-stack option is used, following sort keys are also
122 available:
123
124 - dso_from: name of library or module branched from
125 - dso_to: name of library or module branched to
126 - symbol_from: name of function branched from
127 - symbol_to: name of function branched to
128 - srcline_from: source file and line branched from
129 - srcline_to: source file and line branched to
130 - mispredict: "N" for predicted branch, "Y" for mispredicted branch
131 - in_tx: branch in TSX transaction
132 - abort: TSX transaction abort.
133 - cycles: Cycles in basic block
134
135 And default sort keys are changed to comm, dso_from, symbol_from, dso_to
136 and symbol_to, see '--branch-stack'.
137
138 When the sort key symbol is specified, columns "IPC" and "IPC Coverage"
139 are enabled automatically. Column "IPC" reports the average IPC per function
140 and column "IPC coverage" reports the percentage of instructions with
141 sampled IPC in this function. IPC means Instruction Per Cycle. If it's low,
142 it indicates there may be a performance bottleneck when the function is
143 executed, such as a memory access bottleneck. If a function has high overhead
144 and low IPC, it's worth further analyzing it to optimize its performance.
145
146 If the --mem-mode option is used, the following sort keys are also available
147 (incompatible with --branch-stack):
148 symbol_daddr, dso_daddr, locked, tlb, mem, snoop, dcacheline, blocked.
149
150 - symbol_daddr: name of data symbol being executed on at the time of sample
151 - dso_daddr: name of library or module containing the data being executed
152 on at the time of the sample
153 - locked: whether the bus was locked at the time of the sample
154 - tlb: type of tlb access for the data at the time of the sample
155 - mem: type of memory access for the data at the time of the sample
156 - snoop: type of snoop (if any) for the data at the time of the sample
157 - dcacheline: the cacheline the data address is on at the time of the sample
158 - phys_daddr: physical address of data being executed on at the time of sample
159 - data_page_size: the data page size of data being executed on at the time of sample
160 - blocked: reason of blocked load access for the data at the time of the sample
161
162 And the default sort keys are changed to local_weight, mem, sym, dso,
163 symbol_daddr, dso_daddr, snoop, tlb, locked, blocked, local_ins_lat,
164 see '--mem-mode'.
165
166 If the data file has tracepoint event(s), following (dynamic) sort keys
167 are also available:
168 trace, trace_fields, [<event>.]<field>[/raw]
169
170 - trace: pretty printed trace output in a single column
171 - trace_fields: fields in tracepoints in separate columns
172 - <field name>: optional event and field name for a specific field
173
174 The last form consists of event and field names. If event name is
175 omitted, it searches all events for matching field name. The matched
176 field will be shown only for the event has the field. The event name
177 supports substring match so user doesn't need to specify full subsystem
178 and event name everytime. For example, 'sched:sched_switch' event can
179 be shortened to 'switch' as long as it's not ambiguous. Also event can
180 be specified by its index (starting from 1) preceded by the '%'.
181 So '%1' is the first event, '%2' is the second, and so on.
182
183 The field name can have '/raw' suffix which disables pretty printing
184 and shows raw field value like hex numbers. The --raw-trace option
185 has the same effect for all dynamic sort keys.
186
187 The default sort keys are changed to 'trace' if all events in the data
188 file are tracepoint.
189
190-F::
191--fields=::
192 Specify output field - multiple keys can be specified in CSV format.
193 Following fields are available:
194 overhead, overhead_sys, overhead_us, overhead_children, sample and period.
195 Also it can contain any sort key(s).
196
197 By default, every sort keys not specified in -F will be appended
198 automatically.
199
200 If the keys starts with a prefix '+', then it will append the specified
201 field(s) to the default field order. For example: perf report -F +period,sample.
202
203-p::
204--parent=<regex>::
205 A regex filter to identify parent. The parent is a caller of this
206 function and searched through the callchain, thus it requires callchain
207 information recorded. The pattern is in the extended regex format and
208 defaults to "\^sys_|^do_page_fault", see '--sort parent'.
209
210-x::
211--exclude-other::
212 Only display entries with parent-match.
213
214-w::
215--column-widths=<width[,width...]>::
216 Force each column width to the provided list, for large terminal
217 readability. 0 means no limit (default behavior).
218
219-t::
220--field-separator=::
221 Use a special separator character and don't pad with spaces, replacing
222 all occurrences of this separator in symbol names (and other output)
223 with a '.' character, that thus it's the only non valid separator.
224
225-D::
226--dump-raw-trace::
227 Dump raw trace in ASCII.
228
229--disable-order::
230 Disable raw trace ordering.
231
232-g::
233--call-graph=<print_type,threshold[,print_limit],order,sort_key[,branch],value>::
234 Display call chains using type, min percent threshold, print limit,
235 call order, sort key, optional branch and value. Note that ordering
236 is not fixed so any parameter can be given in an arbitrary order.
237 One exception is the print_limit which should be preceded by threshold.
238
239 print_type can be either:
240 - flat: single column, linear exposure of call chains.
241 - graph: use a graph tree, displaying absolute overhead rates. (default)
242 - fractal: like graph, but displays relative rates. Each branch of
243 the tree is considered as a new profiled object.
244 - folded: call chains are displayed in a line, separated by semicolons
245 - none: disable call chain display.
246
247 threshold is a percentage value which specifies a minimum percent to be
248 included in the output call graph. Default is 0.5 (%).
249
250 print_limit is only applied when stdio interface is used. It's to limit
251 number of call graph entries in a single hist entry. Note that it needs
252 to be given after threshold (but not necessarily consecutive).
253 Default is 0 (unlimited).
254
255 order can be either:
256 - callee: callee based call graph.
257 - caller: inverted caller based call graph.
258 Default is 'caller' when --children is used, otherwise 'callee'.
259
260 sort_key can be:
261 - function: compare on functions (default)
262 - address: compare on individual code addresses
263 - srcline: compare on source filename and line number
264
265 branch can be:
266 - branch: include last branch information in callgraph when available.
267 Usually more convenient to use --branch-history for this.
268
269 value can be:
270 - percent: display overhead percent (default)
271 - period: display event period
272 - count: display event count
273
274--children::
275 Accumulate callchain of children to parent entry so that then can
276 show up in the output. The output will have a new "Children" column
277 and will be sorted on the data. It requires callchains are recorded.
278 See the `overhead calculation' section for more details. Enabled by
279 default, disable with --no-children.
280
281--max-stack::
282 Set the stack depth limit when parsing the callchain, anything
283 beyond the specified depth will be ignored. This is a trade-off
284 between information loss and faster processing especially for
285 workloads that can have a very long callchain stack.
286 Note that when using the --itrace option the synthesized callchain size
287 will override this value if the synthesized callchain size is bigger.
288
289 Default: 127
290
291-G::
292--inverted::
293 alias for inverted caller based call graph.
294
295--ignore-callees=<regex>::
296 Ignore callees of the function(s) matching the given regex.
297 This has the effect of collecting the callers of each such
298 function into one place in the call-graph tree.
299
300--pretty=<key>::
301 Pretty printing style. key: normal, raw
302
303--stdio:: Use the stdio interface.
304
305--stdio-color::
306 'always', 'never' or 'auto', allowing configuring color output
307 via the command line, in addition to via "color.ui" .perfconfig.
308 Use '--stdio-color always' to generate color even when redirecting
309 to a pipe or file. Using just '--stdio-color' is equivalent to
310 using 'always'.
311
312--tui:: Use the TUI interface, that is integrated with annotate and allows
313 zooming into DSOs or threads, among other features. Use of --tui
314 requires a tty, if one is not present, as when piping to other
315 commands, the stdio interface is used.
316
317--gtk:: Use the GTK2 interface.
318
319-k::
320--vmlinux=<file>::
321 vmlinux pathname
322
323--ignore-vmlinux::
324 Ignore vmlinux files.
325
326--kallsyms=<file>::
327 kallsyms pathname
328
329-m::
330--modules::
331 Load module symbols. WARNING: This should only be used with -k and
332 a LIVE kernel.
333
334-f::
335--force::
336 Don't do ownership validation.
337
338--symfs=<directory>::
339 Look for files with symbols relative to this directory.
340
341-C::
342--cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can
343 be provided as a comma-separated list with no space: 0,1. Ranges of
344 CPUs are specified with -: 0-2. Default is to report samples on all
345 CPUs.
346
347-M::
348--disassembler-style=:: Set disassembler style for objdump.
349
350--source::
351 Interleave source code with assembly code. Enabled by default,
352 disable with --no-source.
353
354--asm-raw::
355 Show raw instruction encoding of assembly instructions.
356
357--show-total-period:: Show a column with the sum of periods.
358
359-I::
360--show-info::
361 Display extended information about the perf.data file. This adds
362 information which may be very large and thus may clutter the display.
363 It currently includes: cpu and numa topology of the host system.
364
365-b::
366--branch-stack::
367 Use the addresses of sampled taken branches instead of the instruction
368 address to build the histograms. To generate meaningful output, the
369 perf.data file must have been obtained using perf record -b or
370 perf record --branch-filter xxx where xxx is a branch filter option.
371 perf report is able to auto-detect whether a perf.data file contains
372 branch stacks and it will automatically switch to the branch view mode,
373 unless --no-branch-stack is used.
374
375--branch-history::
376 Add the addresses of sampled taken branches to the callstack.
377 This allows to examine the path the program took to each sample.
378 The data collection must have used -b (or -j) and -g.
379
380--objdump=<path>::
381 Path to objdump binary.
382
383--prefix=PREFIX::
384--prefix-strip=N::
385 Remove first N entries from source file path names in executables
386 and add PREFIX. This allows to display source code compiled on systems
387 with different file system layout.
388
389--group::
390 Show event group information together. It forces group output also
391 if there are no groups defined in data file.
392
393--group-sort-idx::
394 Sort the output by the event at the index n in group. If n is invalid,
395 sort by the first event. It can support multiple groups with different
396 amount of events. WARNING: This should be used on grouped events.
397
398--demangle::
399 Demangle symbol names to human readable form. It's enabled by default,
400 disable with --no-demangle.
401
402--demangle-kernel::
403 Demangle kernel symbol names to human readable form (for C++ kernels).
404
405--mem-mode::
406 Use the data addresses of samples in addition to instruction addresses
407 to build the histograms. To generate meaningful output, the perf.data
408 file must have been obtained using perf record -d -W and using a
409 special event -e cpu/mem-loads/p or -e cpu/mem-stores/p. See
410 'perf mem' for simpler access.
411
412--percent-limit::
413 Do not show entries which have an overhead under that percent.
414 (Default: 0). Note that this option also sets the percent limit (threshold)
415 of callchains. However the default value of callchain threshold is
416 different than the default value of hist entries. Please see the
417 --call-graph option for details.
418
419--percentage::
420 Determine how to display the overhead percentage of filtered entries.
421 Filters can be applied by --comms, --dsos and/or --symbols options and
422 Zoom operations on the TUI (thread, dso, etc).
423
424 "relative" means it's relative to filtered entries only so that the
425 sum of shown entries will be always 100%. "absolute" means it retains
426 the original value before and after the filter is applied.
427
428--header::
429 Show header information in the perf.data file. This includes
430 various information like hostname, OS and perf version, cpu/mem
431 info, perf command line, event list and so on. Currently only
432 --stdio output supports this feature.
433
434--header-only::
435 Show only perf.data header (forces --stdio).
436
437--time::
438 Only analyze samples within given time window: <start>,<stop>. Times
439 have the format seconds.nanoseconds. If start is not given (i.e. time
440 string is ',x.y') then analysis starts at the beginning of the file. If
441 stop time is not given (i.e. time string is 'x.y,') then analysis goes
442 to end of file. Multiple ranges can be separated by spaces, which
443 requires the argument to be quoted e.g. --time "1234.567,1234.789 1235,"
444
445 Also support time percent with multiple time ranges. Time string is
446 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'.
447
448 For example:
449 Select the second 10% time slice:
450
451 perf report --time 10%/2
452
453 Select from 0% to 10% time slice:
454
455 perf report --time 0%-10%
456
457 Select the first and second 10% time slices:
458
459 perf report --time 10%/1,10%/2
460
461 Select from 0% to 10% and 30% to 40% slices:
462
463 perf report --time 0%-10%,30%-40%
464
465--switch-on EVENT_NAME::
466 Only consider events after this event is found.
467
468 This may be interesting to measure a workload only after some initialization
469 phase is over, i.e. insert a perf probe at that point and then using this
470 option with that probe.
471
472--switch-off EVENT_NAME::
473 Stop considering events after this event is found.
474
475--show-on-off-events::
476 Show the --switch-on/off events too. This has no effect in 'perf report' now
477 but probably we'll make the default not to show the switch-on/off events
478 on the --group mode and if there is only one event besides the off/on ones,
479 go straight to the histogram browser, just like 'perf report' with no events
480 explicitly specified does.
481
482--itrace::
483 Options for decoding instruction tracing data. The options are:
484
485include::itrace.txt[]
486
487 To disable decoding entirely, use --no-itrace.
488
489--full-source-path::
490 Show the full path for source files for srcline output.
491
492--show-ref-call-graph::
493 When multiple events are sampled, it may not be needed to collect
494 callgraphs for all of them. The sample sites are usually nearby,
495 and it's enough to collect the callgraphs on a reference event.
496 So user can use "call-graph=no" event modifier to disable callgraph
497 for other events to reduce the overhead.
498 However, perf report cannot show callgraphs for the event which
499 disable the callgraph.
500 This option extends the perf report to show reference callgraphs,
501 which collected by reference event, in no callgraph event.
502
503--stitch-lbr::
504 Show callgraph with stitched LBRs, which may have more complete
505 callgraph. The perf.data file must have been obtained using
506 perf record --call-graph lbr.
507 Disabled by default. In common cases with call stack overflows,
508 it can recreate better call stacks than the default lbr call stack
509 output. But this approach is not full proof. There can be cases
510 where it creates incorrect call stacks from incorrect matches.
511 The known limitations include exception handing such as
512 setjmp/longjmp will have calls/returns not match.
513
514--socket-filter::
515 Only report the samples on the processor socket that match with this filter
516
517--samples=N::
518 Save N individual samples for each histogram entry to show context in perf
519 report tui browser.
520
521--raw-trace::
522 When displaying traceevent output, do not use print fmt or plugins.
523
524--hierarchy::
525 Enable hierarchical output.
526
527--inline::
528 If a callgraph address belongs to an inlined function, the inline stack
529 will be printed. Each entry is function name or file/line. Enabled by
530 default, disable with --no-inline.
531
532--mmaps::
533 Show --tasks output plus mmap information in a format similar to
534 /proc/<PID>/maps.
535
536 Please note that not all mmaps are stored, options affecting which ones
537 are include 'perf record --data', for instance.
538
539--ns::
540 Show time stamps in nanoseconds.
541
542--stats::
543 Display overall events statistics without any further processing.
544 (like the one at the end of the perf report -D command)
545
546--tasks::
547 Display monitored tasks stored in perf data. Displaying pid/tid/ppid
548 plus the command string aligned to distinguish parent and child tasks.
549
550--percent-type::
551 Set annotation percent type from following choices:
552 global-period, local-period, global-hits, local-hits
553
554 The local/global keywords set if the percentage is computed
555 in the scope of the function (local) or the whole data (global).
556 The period/hits keywords set the base the percentage is computed
557 on - the samples period or the number of samples (hits).
558
559--time-quantum::
560 Configure time quantum for time sort key. Default 100ms.
561 Accepts s, us, ms, ns units.
562
563--total-cycles::
564 When --total-cycles is specified, it supports sorting for all blocks by
565 'Sampled Cycles%'. This is useful to concentrate on the globally hottest
566 blocks. In output, there are some new columns:
567
568 'Sampled Cycles%' - block sampled cycles aggregation / total sampled cycles
569 'Sampled Cycles' - block sampled cycles aggregation
570 'Avg Cycles%' - block average sampled cycles / sum of total block average
571 sampled cycles
572 'Avg Cycles' - block average sampled cycles
573
574--skip-empty::
575 Do not print 0 results in the --stat output.
576
577include::callchain-overhead-calculation.txt[]
578
579SEE ALSO
580--------
581linkperf:perf-stat[1], linkperf:perf-annotate[1], linkperf:perf-record[1],
582linkperf:perf-intel-pt[1]