xref: /linux/tools/perf/Documentation/perf-config.txt (revision e5c86679d5e864947a52fb31e45a425dea3e7fa9)
1perf-config(1)
2==============
3
4NAME
5----
6perf-config - Get and set variables in a configuration file.
7
8SYNOPSIS
9--------
10[verse]
11'perf config' [<file-option>] [section.name[=value] ...]
12or
13'perf config' [<file-option>] -l | --list
14
15DESCRIPTION
16-----------
17You can manage variables in a configuration file with this command.
18
19OPTIONS
20-------
21
22-l::
23--list::
24	Show current config variables, name and value, for all sections.
25
26--user::
27	For writing and reading options: write to user
28	'$HOME/.perfconfig' file or read it.
29
30--system::
31	For writing and reading options: write to system-wide
32	'$(sysconfdir)/perfconfig' or read it.
33
34CONFIGURATION FILE
35------------------
36
37The perf configuration file contains many variables to change various
38aspects of each of its tools, including output, disk usage, etc.
39The '$HOME/.perfconfig' file is used to store a per-user configuration.
40The file '$(sysconfdir)/perfconfig' can be used to
41store a system-wide default configuration.
42
43When reading or writing, the values are read from the system and user
44configuration files by default, and options '--system' and '--user'
45can be used to tell the command to read from or write to only that location.
46
47Syntax
48~~~~~~
49
50The file consist of sections. A section starts with its name
51surrounded by square brackets and continues till the next section
52begins. Each variable must be in a section, and have the form
53'name = value', for example:
54
55	[section]
56		name1 = value1
57		name2 = value2
58
59Section names are case sensitive and can contain any characters except
60newline (double quote `"` and backslash have to be escaped as `\"` and `\\`,
61respectively). Section headers can't span multiple lines.
62
63Example
64~~~~~~~
65
66Given a $HOME/.perfconfig like this:
67
68#
69# This is the config file, and
70# a '#' and ';' character indicates a comment
71#
72
73	[colors]
74		# Color variables
75		top = red, default
76		medium = green, default
77		normal = lightgray, default
78		selected = white, lightgray
79		jump_arrows = blue, default
80		addr = magenta, default
81		root = white, blue
82
83	[tui]
84		# Defaults if linked with libslang
85		report = on
86		annotate = on
87		top = on
88
89	[buildid]
90		# Default, disable using /dev/null
91		dir = ~/.debug
92
93	[annotate]
94		# Defaults
95		hide_src_code = false
96		use_offset = true
97		jump_arrows = true
98		show_nr_jumps = false
99
100	[help]
101		# Format can be man, info, web or html
102		format = man
103		autocorrect = 0
104
105	[ui]
106		show-headers = true
107
108	[call-graph]
109		# fp (framepointer), dwarf
110		record-mode = fp
111		print-type = graph
112		order = caller
113		sort-key = function
114
115	[report]
116		# Defaults
117		sort-order = comm,dso,symbol
118		percent-limit = 0
119		queue-size = 0
120		children = true
121		group = true
122
123You can hide source code of annotate feature setting the config to false with
124
125	% perf config annotate.hide_src_code=true
126
127If you want to add or modify several config items, you can do like
128
129	% perf config ui.show-headers=false kmem.default=slab
130
131To modify the sort order of report functionality in user config file(i.e. `~/.perfconfig`), do
132
133	% perf config --user report sort-order=srcline
134
135To change colors of selected line to other foreground and background colors
136in system config file (i.e. `$(sysconf)/perfconfig`), do
137
138	% perf config --system colors.selected=yellow,green
139
140To query the record mode of call graph, do
141
142	% perf config call-graph.record-mode
143
144If you want to know multiple config key/value pairs, you can do like
145
146	% perf config report.queue-size call-graph.order report.children
147
148To query the config value of sort order of call graph in user config file (i.e. `~/.perfconfig`), do
149
150	% perf config --user call-graph.sort-order
151
152To query the config value of buildid directory in system config file (i.e. `$(sysconf)/perfconfig`), do
153
154	% perf config --system buildid.dir
155
156Variables
157~~~~~~~~~
158
159colors.*::
160	The variables for customizing the colors used in the output for the
161	'report', 'top' and 'annotate' in the TUI. They should specify the
162	foreground and background colors, separated by a comma, for example:
163
164		medium = green, lightgray
165
166	If you want to use the color configured for you terminal, just leave it
167	as 'default', for example:
168
169		medium = default, lightgray
170
171	Available colors:
172	red, yellow, green, cyan, gray, black, blue,
173	white, default, magenta, lightgray
174
175	colors.top::
176		'top' means a overhead percentage which is more than 5%.
177		And values of this variable specify percentage colors.
178		Basic key values are foreground-color 'red' and
179		background-color 'default'.
180	colors.medium::
181		'medium' means a overhead percentage which has more than 0.5%.
182		Default values are 'green' and 'default'.
183	colors.normal::
184		'normal' means the rest of overhead percentages
185		except 'top', 'medium', 'selected'.
186		Default values are 'lightgray' and 'default'.
187	colors.selected::
188		This selects the colors for the current entry in a list of entries
189		from sub-commands (top, report, annotate).
190		Default values are 'black' and 'lightgray'.
191	colors.jump_arrows::
192		Colors for jump arrows on assembly code listings
193		such as 'jns', 'jmp', 'jane', etc.
194		Default values are 'blue', 'default'.
195	colors.addr::
196		This selects colors for addresses from 'annotate'.
197		Default values are 'magenta', 'default'.
198	colors.root::
199		Colors for headers in the output of a sub-commands (top, report).
200		Default values are 'white', 'blue'.
201
202tui.*, gtk.*::
203	Subcommands that can be configured here are 'top', 'report' and 'annotate'.
204	These values are booleans, for example:
205
206	[tui]
207		top = true
208
209	will make the TUI be the default for the 'top' subcommand. Those will be
210	available if the required libs were detected at tool build time.
211
212buildid.*::
213	buildid.dir::
214		Each executable and shared library in modern distributions comes with a
215		content based identifier that, if available, will be inserted in a
216		'perf.data' file header to, at analysis time find what is needed to do
217		symbol resolution, code annotation, etc.
218
219		The recording tools also stores a hard link or copy in a per-user
220		directory, $HOME/.debug/, of binaries, shared libraries, /proc/kallsyms
221		and /proc/kcore files to be used at analysis time.
222
223		The buildid.dir variable can be used to either change this directory
224		cache location, or to disable it altogether. If you want to disable it,
225		set buildid.dir to /dev/null. The default is $HOME/.debug
226
227annotate.*::
228	These options work only for TUI.
229	These are in control of addresses, jump function, source code
230	in lines of assembly code from a specific program.
231
232	annotate.hide_src_code::
233		If a program which is analyzed has source code,
234		this option lets 'annotate' print a list of assembly code with the source code.
235		For example, let's see a part of a program. There're four lines.
236		If this option is 'true', they can be printed
237		without source code from a program as below.
238
239		│        push   %rbp
240		│        mov    %rsp,%rbp
241		│        sub    $0x10,%rsp
242		│        mov    (%rdi),%rdx
243
244		But if this option is 'false', source code of the part
245		can be also printed as below. Default is 'false'.
246
247		│      struct rb_node *rb_next(const struct rb_node *node)
248		│      {
249		│        push   %rbp
250		│        mov    %rsp,%rbp
251		│        sub    $0x10,%rsp
252		│              struct rb_node *parent;
253254		│              if (RB_EMPTY_NODE(node))
255		│        mov    (%rdi),%rdx
256		│              return n;
257
258        annotate.use_offset::
259		Basing on a first address of a loaded function, offset can be used.
260		Instead of using original addresses of assembly code,
261		addresses subtracted from a base address can be printed.
262		Let's illustrate an example.
263		If a base address is 0XFFFFFFFF81624d50 as below,
264
265		ffffffff81624d50 <load0>
266
267		an address on assembly code has a specific absolute address as below
268
269		ffffffff816250b8:│  mov    0x8(%r14),%rdi
270
271		but if use_offset is 'true', an address subtracted from a base address is printed.
272		Default is true. This option is only applied to TUI.
273
274		             368:│  mov    0x8(%r14),%rdi
275
276	annotate.jump_arrows::
277		There can be jump instruction among assembly code.
278		Depending on a boolean value of jump_arrows,
279		arrows can be printed or not which represent
280		where do the instruction jump into as below.
281
282		│     ┌──jmp    1333
283		│     │  xchg   %ax,%ax
284		│1330:│  mov    %r15,%r10
285		│1333:└─→cmp    %r15,%r14
286
287		If jump_arrow is 'false', the arrows isn't printed as below.
288		Default is 'false'.
289
290		│      ↓ jmp    1333
291		│        xchg   %ax,%ax
292		│1330:   mov    %r15,%r10
293		│1333:   cmp    %r15,%r14
294
295        annotate.show_linenr::
296		When showing source code if this option is 'true',
297		line numbers are printed as below.
298
299		│1628         if (type & PERF_SAMPLE_IDENTIFIER) {
300		│     ↓ jne    508
301		│1628                 data->id = *array;
302		│1629                 array++;
303		│1630         }
304
305		However if this option is 'false', they aren't printed as below.
306		Default is 'false'.
307
308		│             if (type & PERF_SAMPLE_IDENTIFIER) {
309		│     ↓ jne    508
310		│                     data->id = *array;
311		│                     array++;
312		│             }
313
314        annotate.show_nr_jumps::
315		Let's see a part of assembly code.
316
317		│1382:   movb   $0x1,-0x270(%rbp)
318
319		If use this, the number of branches jumping to that address can be printed as below.
320		Default is 'false'.
321
322		│1 1382:   movb   $0x1,-0x270(%rbp)
323
324        annotate.show_total_period::
325		To compare two records on an instruction base, with this option
326		provided, display total number of samples that belong to a line
327		in assembly code. If this option is 'true', total periods are printed
328		instead of percent values as below.
329
330		  302 │      mov    %eax,%eax
331
332		But if this option is 'false', percent values for overhead are printed i.e.
333		Default is 'false'.
334
335		99.93 │      mov    %eax,%eax
336
337hist.*::
338	hist.percentage::
339		This option control the way to calculate overhead of filtered entries -
340		that means the value of this option is effective only if there's a
341		filter (by comm, dso or symbol name). Suppose a following example:
342
343		       Overhead  Symbols
344		       ........  .......
345		        33.33%     foo
346		        33.33%     bar
347		        33.33%     baz
348
349	       This is an original overhead and we'll filter out the first 'foo'
350	       entry. The value of 'relative' would increase the overhead of 'bar'
351	       and 'baz' to 50.00% for each, while 'absolute' would show their
352	       current overhead (33.33%).
353
354ui.*::
355	ui.show-headers::
356		This option controls display of column headers (like 'Overhead' and 'Symbol')
357		in 'report' and 'top'. If this option is false, they are hidden.
358		This option is only applied to TUI.
359
360call-graph.*::
361	When sub-commands 'top' and 'report' work with -g/—-children
362	there're options in control of call-graph.
363
364	call-graph.record-mode::
365		The record-mode can be 'fp' (frame pointer), 'dwarf' and 'lbr'.
366		The value of 'dwarf' is effective only if perf detect needed library
367		(libunwind or a recent version of libdw).
368		'lbr' only work for cpus that support it.
369
370	call-graph.dump-size::
371		The size of stack to dump in order to do post-unwinding. Default is 8192 (byte).
372		When using dwarf into record-mode, the default size will be used if omitted.
373
374	call-graph.print-type::
375		The print-types can be graph (graph absolute), fractal (graph relative),
376		flat and folded. This option controls a way to show overhead for each callchain
377		entry. Suppose a following example.
378
379                Overhead  Symbols
380                ........  .......
381                  40.00%  foo
382                          |
383                          ---foo
384                             |
385                             |--50.00%--bar
386                             |          main
387                             |
388                              --50.00%--baz
389                                        main
390
391		This output is a 'fractal' format. The 'foo' came from 'bar' and 'baz' exactly
392		half and half so 'fractal' shows 50.00% for each
393		(meaning that it assumes 100% total overhead of 'foo').
394
395		The 'graph' uses absolute overhead value of 'foo' as total so each of
396		'bar' and 'baz' callchain will have 20.00% of overhead.
397		If 'flat' is used, single column and linear exposure of call chains.
398		'folded' mean call chains are displayed in a line, separated by semicolons.
399
400	call-graph.order::
401		This option controls print order of callchains. The default is
402		'callee' which means callee is printed at top and then followed by its
403		caller and so on. The 'caller' prints it in reverse order.
404
405		If this option is not set and report.children or top.children is
406		set to true (or the equivalent command line option is given),
407		the default value of this option is changed to 'caller' for the
408		execution of 'perf report' or 'perf top'. Other commands will
409		still default to 'callee'.
410
411	call-graph.sort-key::
412		The callchains are merged if they contain same information.
413		The sort-key option determines a way to compare the callchains.
414		A value of 'sort-key' can be 'function' or 'address'.
415		The default is 'function'.
416
417	call-graph.threshold::
418		When there're many callchains it'd print tons of lines. So perf omits
419		small callchains under a certain overhead (threshold) and this option
420		control the threshold. Default is 0.5 (%). The overhead is calculated
421		by value depends on call-graph.print-type.
422
423	call-graph.print-limit::
424		This is a maximum number of lines of callchain printed for a single
425		histogram entry. Default is 0 which means no limitation.
426
427report.*::
428	report.sort_order::
429		Allows changing the default sort order from "comm,dso,symbol" to
430		some other default, for instance "sym,dso" may be more fitting for
431		kernel developers.
432	report.percent-limit::
433		This one is mostly the same as call-graph.threshold but works for
434		histogram entries. Entries having an overhead lower than this
435		percentage will not be printed. Default is '0'. If percent-limit
436		is '10', only entries which have more than 10% of overhead will be
437		printed.
438
439	report.queue-size::
440		This option sets up the maximum allocation size of the internal
441		event queue for ordering events. Default is 0, meaning no limit.
442
443	report.children::
444		'Children' means functions called from another function.
445		If this option is true, 'perf report' cumulates callchains of children
446		and show (accumulated) total overhead as well as 'Self' overhead.
447		Please refer to the 'perf report' manual. The default is 'true'.
448
449	report.group::
450		This option is to show event group information together.
451		Example output with this turned on, notice that there is one column
452		per event in the group, ref-cycles and cycles:
453
454		# group: {ref-cycles,cycles}
455		# ========
456		#
457		# Samples: 7K of event 'anon group { ref-cycles, cycles }'
458		# Event count (approx.): 6876107743
459		#
460		#         Overhead  Command      Shared Object               Symbol
461		# ................  .......  .................  ...................
462		#
463		    99.84%  99.76%  noploop  noploop            [.] main
464		     0.07%   0.00%  noploop  ld-2.15.so         [.] strcmp
465		     0.03%   0.00%  noploop  [kernel.kallsyms]  [k] timerqueue_del
466
467top.*::
468	top.children::
469		Same as 'report.children'. So if it is enabled, the output of 'top'
470		command will have 'Children' overhead column as well as 'Self' overhead
471		column by default.
472		The default is 'true'.
473
474man.*::
475	man.viewer::
476		This option can assign a tool to view manual pages when 'help'
477		subcommand was invoked. Supported tools are 'man', 'woman'
478		(with emacs client) and 'konqueror'. Default is 'man'.
479
480		New man viewer tool can be also added using 'man.<tool>.cmd'
481		or use different path using 'man.<tool>.path' config option.
482
483pager.*::
484	pager.<subcommand>::
485		When the subcommand is run on stdio, determine whether it uses
486		pager or not based on this value. Default is 'unspecified'.
487
488kmem.*::
489	kmem.default::
490		This option decides which allocator is to be analyzed if neither
491		'--slab' nor '--page' option is used. Default is 'slab'.
492
493record.*::
494	record.build-id::
495		This option can be 'cache', 'no-cache' or 'skip'.
496		'cache' is to post-process data and save/update the binaries into
497		the build-id cache (in ~/.debug). This is the default.
498		But if this option is 'no-cache', it will not update the build-id cache.
499		'skip' skips post-processing and does not update the cache.
500
501diff.*::
502	diff.order::
503		This option sets the number of columns to sort the result.
504		The default is 0, which means sorting by baseline.
505		Setting it to 1 will sort the result by delta (or other
506		compute method selected).
507
508	diff.compute::
509		This options sets the method for computing the diff result.
510		Possible values are 'delta', 'delta-abs', 'ratio' and
511		'wdiff'.  Default is 'delta'.
512
513SEE ALSO
514--------
515linkperf:perf[1]
516