0001 perf-config(1)
0002 ==============
0003
0004 NAME
0005 ----
0006 perf-config - Get and set variables in a configuration file.
0007
0008 SYNOPSIS
0009 --------
0010 [verse]
0011 'perf config' [<file-option>] [section.name[=value] ...]
0012 or
0013 'perf config' [<file-option>] -l | --list
0014
0015 DESCRIPTION
0016 -----------
0017 You can manage variables in a configuration file with this command.
0018
0019 OPTIONS
0020 -------
0021
0022 -l::
0023 --list::
0024 Show current config variables, name and value, for all sections.
0025
0026 --user::
0027 For writing and reading options: write to user
0028 '$HOME/.perfconfig' file or read it.
0029
0030 --system::
0031 For writing and reading options: write to system-wide
0032 '$(sysconfdir)/perfconfig' or read it.
0033
0034 CONFIGURATION FILE
0035 ------------------
0036
0037 The perf configuration file contains many variables to change various
0038 aspects of each of its tools, including output, disk usage, etc.
0039 The '$HOME/.perfconfig' file is used to store a per-user configuration.
0040 The file '$(sysconfdir)/perfconfig' can be used to
0041 store a system-wide default configuration.
0042
0043 One an disable reading config files by setting the PERF_CONFIG environment
0044 variable to /dev/null, or provide an alternate config file by setting that
0045 variable.
0046
0047 When reading or writing, the values are read from the system and user
0048 configuration files by default, and options '--system' and '--user'
0049 can be used to tell the command to read from or write to only that location.
0050
0051 Syntax
0052 ~~~~~~
0053
0054 The file consist of sections. A section starts with its name
0055 surrounded by square brackets and continues till the next section
0056 begins. Each variable must be in a section, and have the form
0057 'name = value', for example:
0058
0059 [section]
0060 name1 = value1
0061 name2 = value2
0062
0063 Section names are case sensitive and can contain any characters except
0064 newline (double quote `"` and backslash have to be escaped as `\"` and `\\`,
0065 respectively). Section headers can't span multiple lines.
0066
0067 Example
0068 ~~~~~~~
0069
0070 Given a $HOME/.perfconfig like this:
0071
0072 #
0073 # This is the config file, and
0074 # a '#' and ';' character indicates a comment
0075 #
0076
0077 [colors]
0078 # Color variables
0079 top = red, default
0080 medium = green, default
0081 normal = lightgray, default
0082 selected = white, lightgray
0083 jump_arrows = blue, default
0084 addr = magenta, default
0085 root = white, blue
0086
0087 [tui]
0088 # Defaults if linked with libslang
0089 report = on
0090 annotate = on
0091 top = on
0092
0093 [buildid]
0094 # Default, disable using /dev/null
0095 dir = ~/.debug
0096
0097 [annotate]
0098 # Defaults
0099 hide_src_code = false
0100 use_offset = true
0101 jump_arrows = true
0102 show_nr_jumps = false
0103
0104 [help]
0105 # Format can be man, info, web or html
0106 format = man
0107 autocorrect = 0
0108
0109 [ui]
0110 show-headers = true
0111
0112 [call-graph]
0113 # fp (framepointer), dwarf
0114 record-mode = fp
0115 print-type = graph
0116 order = caller
0117 sort-key = function
0118
0119 [report]
0120 # Defaults
0121 sort_order = comm,dso,symbol
0122 percent-limit = 0
0123 queue-size = 0
0124 children = true
0125 group = true
0126 skip-empty = true
0127
0128 [llvm]
0129 dump-obj = true
0130 clang-opt = -g
0131
0132 You can hide source code of annotate feature setting the config to false with
0133
0134 % perf config annotate.hide_src_code=true
0135
0136 If you want to add or modify several config items, you can do like
0137
0138 % perf config ui.show-headers=false kmem.default=slab
0139
0140 To modify the sort order of report functionality in user config file(i.e. `~/.perfconfig`), do
0141
0142 % perf config --user report.sort-order=srcline
0143
0144 To change colors of selected line to other foreground and background colors
0145 in system config file (i.e. `$(sysconf)/perfconfig`), do
0146
0147 % perf config --system colors.selected=yellow,green
0148
0149 To query the record mode of call graph, do
0150
0151 % perf config call-graph.record-mode
0152
0153 If you want to know multiple config key/value pairs, you can do like
0154
0155 % perf config report.queue-size call-graph.order report.children
0156
0157 To query the config value of sort order of call graph in user config file (i.e. `~/.perfconfig`), do
0158
0159 % perf config --user call-graph.sort-order
0160
0161 To query the config value of buildid directory in system config file (i.e. `$(sysconf)/perfconfig`), do
0162
0163 % perf config --system buildid.dir
0164
0165 Variables
0166 ~~~~~~~~~
0167
0168 colors.*::
0169 The variables for customizing the colors used in the output for the
0170 'report', 'top' and 'annotate' in the TUI. They should specify the
0171 foreground and background colors, separated by a comma, for example:
0172
0173 medium = green, lightgray
0174
0175 If you want to use the color configured for you terminal, just leave it
0176 as 'default', for example:
0177
0178 medium = default, lightgray
0179
0180 Available colors:
0181 red, yellow, green, cyan, gray, black, blue,
0182 white, default, magenta, lightgray
0183
0184 colors.top::
0185 'top' means a overhead percentage which is more than 5%.
0186 And values of this variable specify percentage colors.
0187 Basic key values are foreground-color 'red' and
0188 background-color 'default'.
0189 colors.medium::
0190 'medium' means a overhead percentage which has more than 0.5%.
0191 Default values are 'green' and 'default'.
0192 colors.normal::
0193 'normal' means the rest of overhead percentages
0194 except 'top', 'medium', 'selected'.
0195 Default values are 'lightgray' and 'default'.
0196 colors.selected::
0197 This selects the colors for the current entry in a list of entries
0198 from sub-commands (top, report, annotate).
0199 Default values are 'black' and 'lightgray'.
0200 colors.jump_arrows::
0201 Colors for jump arrows on assembly code listings
0202 such as 'jns', 'jmp', 'jane', etc.
0203 Default values are 'blue', 'default'.
0204 colors.addr::
0205 This selects colors for addresses from 'annotate'.
0206 Default values are 'magenta', 'default'.
0207 colors.root::
0208 Colors for headers in the output of a sub-commands (top, report).
0209 Default values are 'white', 'blue'.
0210
0211 core.*::
0212 core.proc-map-timeout::
0213 Sets a timeout (in milliseconds) for parsing /proc/<pid>/maps files.
0214 Can be overridden by the --proc-map-timeout option on supported
0215 subcommands. The default timeout is 500ms.
0216
0217 tui.*, gtk.*::
0218 Subcommands that can be configured here are 'top', 'report' and 'annotate'.
0219 These values are booleans, for example:
0220
0221 [tui]
0222 top = true
0223
0224 will make the TUI be the default for the 'top' subcommand. Those will be
0225 available if the required libs were detected at tool build time.
0226
0227 buildid.*::
0228 buildid.dir::
0229 Each executable and shared library in modern distributions comes with a
0230 content based identifier that, if available, will be inserted in a
0231 'perf.data' file header to, at analysis time find what is needed to do
0232 symbol resolution, code annotation, etc.
0233
0234 The recording tools also stores a hard link or copy in a per-user
0235 directory, $HOME/.debug/, of binaries, shared libraries, /proc/kallsyms
0236 and /proc/kcore files to be used at analysis time.
0237
0238 The buildid.dir variable can be used to either change this directory
0239 cache location, or to disable it altogether. If you want to disable it,
0240 set buildid.dir to /dev/null. The default is $HOME/.debug
0241
0242 buildid-cache.*::
0243 buildid-cache.debuginfod=URLs
0244 Specify debuginfod URLs to be used when retrieving perf.data binaries,
0245 it follows the same syntax as the DEBUGINFOD_URLS variable, like:
0246
0247 buildid-cache.debuginfod=http://192.168.122.174:8002
0248
0249 annotate.*::
0250 These are in control of addresses, jump function, source code
0251 in lines of assembly code from a specific program.
0252
0253 annotate.disassembler_style:
0254 Use this to change the default disassembler style to some other value
0255 supported by binutils, such as "intel", see the '-M' option help in the
0256 'objdump' man page.
0257
0258 annotate.hide_src_code::
0259 If a program which is analyzed has source code,
0260 this option lets 'annotate' print a list of assembly code with the source code.
0261 For example, let's see a part of a program. There're four lines.
0262 If this option is 'true', they can be printed
0263 without source code from a program as below.
0264
0265 │ push %rbp
0266 │ mov %rsp,%rbp
0267 │ sub $0x10,%rsp
0268 │ mov (%rdi),%rdx
0269
0270 But if this option is 'false', source code of the part
0271 can be also printed as below. Default is 'false'.
0272
0273 │ struct rb_node *rb_next(const struct rb_node *node)
0274 │ {
0275 │ push %rbp
0276 │ mov %rsp,%rbp
0277 │ sub $0x10,%rsp
0278 │ struct rb_node *parent;
0279 │
0280 │ if (RB_EMPTY_NODE(node))
0281 │ mov (%rdi),%rdx
0282 │ return n;
0283
0284 This option works with tui, stdio2 browsers.
0285
0286 annotate.use_offset::
0287 Basing on a first address of a loaded function, offset can be used.
0288 Instead of using original addresses of assembly code,
0289 addresses subtracted from a base address can be printed.
0290 Let's illustrate an example.
0291 If a base address is 0XFFFFFFFF81624d50 as below,
0292
0293 ffffffff81624d50 <load0>
0294
0295 an address on assembly code has a specific absolute address as below
0296
0297 ffffffff816250b8:│ mov 0x8(%r14),%rdi
0298
0299 but if use_offset is 'true', an address subtracted from a base address is printed.
0300 Default is true. This option is only applied to TUI.
0301
0302 368:│ mov 0x8(%r14),%rdi
0303
0304 This option works with tui, stdio2 browsers.
0305
0306 annotate.jump_arrows::
0307 There can be jump instruction among assembly code.
0308 Depending on a boolean value of jump_arrows,
0309 arrows can be printed or not which represent
0310 where do the instruction jump into as below.
0311
0312 │ ┌──jmp 1333
0313 │ │ xchg %ax,%ax
0314 │1330:│ mov %r15,%r10
0315 │1333:└─→cmp %r15,%r14
0316
0317 If jump_arrow is 'false', the arrows isn't printed as below.
0318 Default is 'false'.
0319
0320 │ ↓ jmp 1333
0321 │ xchg %ax,%ax
0322 │1330: mov %r15,%r10
0323 │1333: cmp %r15,%r14
0324
0325 This option works with tui browser.
0326
0327 annotate.show_linenr::
0328 When showing source code if this option is 'true',
0329 line numbers are printed as below.
0330
0331 │1628 if (type & PERF_SAMPLE_IDENTIFIER) {
0332 │ ↓ jne 508
0333 │1628 data->id = *array;
0334 │1629 array++;
0335 │1630 }
0336
0337 However if this option is 'false', they aren't printed as below.
0338 Default is 'false'.
0339
0340 │ if (type & PERF_SAMPLE_IDENTIFIER) {
0341 │ ↓ jne 508
0342 │ data->id = *array;
0343 │ array++;
0344 │ }
0345
0346 This option works with tui, stdio2 browsers.
0347
0348 annotate.show_nr_jumps::
0349 Let's see a part of assembly code.
0350
0351 │1382: movb $0x1,-0x270(%rbp)
0352
0353 If use this, the number of branches jumping to that address can be printed as below.
0354 Default is 'false'.
0355
0356 │1 1382: movb $0x1,-0x270(%rbp)
0357
0358 This option works with tui, stdio2 browsers.
0359
0360 annotate.show_total_period::
0361 To compare two records on an instruction base, with this option
0362 provided, display total number of samples that belong to a line
0363 in assembly code. If this option is 'true', total periods are printed
0364 instead of percent values as below.
0365
0366 302 │ mov %eax,%eax
0367
0368 But if this option is 'false', percent values for overhead are printed i.e.
0369 Default is 'false'.
0370
0371 99.93 │ mov %eax,%eax
0372
0373 This option works with tui, stdio2, stdio browsers.
0374
0375 annotate.show_nr_samples::
0376 By default perf annotate shows percentage of samples. This option
0377 can be used to print absolute number of samples. Ex, when set as
0378 false:
0379
0380 Percent│
0381 74.03 │ mov %fs:0x28,%rax
0382
0383 When set as true:
0384
0385 Samples│
0386 6 │ mov %fs:0x28,%rax
0387
0388 This option works with tui, stdio2, stdio browsers.
0389
0390 annotate.offset_level::
0391 Default is '1', meaning just jump targets will have offsets show right beside
0392 the instruction. When set to '2' 'call' instructions will also have its offsets
0393 shown, 3 or higher will show offsets for all instructions.
0394
0395 This option works with tui, stdio2 browsers.
0396
0397 annotate.demangle::
0398 Demangle symbol names to human readable form. Default is 'true'.
0399
0400 annotate.demangle_kernel::
0401 Demangle kernel symbol names to human readable form. Default is 'true'.
0402
0403 hist.*::
0404 hist.percentage::
0405 This option control the way to calculate overhead of filtered entries -
0406 that means the value of this option is effective only if there's a
0407 filter (by comm, dso or symbol name). Suppose a following example:
0408
0409 Overhead Symbols
0410 ........ .......
0411 33.33% foo
0412 33.33% bar
0413 33.33% baz
0414
0415 This is an original overhead and we'll filter out the first 'foo'
0416 entry. The value of 'relative' would increase the overhead of 'bar'
0417 and 'baz' to 50.00% for each, while 'absolute' would show their
0418 current overhead (33.33%).
0419
0420 ui.*::
0421 ui.show-headers::
0422 This option controls display of column headers (like 'Overhead' and 'Symbol')
0423 in 'report' and 'top'. If this option is false, they are hidden.
0424 This option is only applied to TUI.
0425
0426 call-graph.*::
0427 The following controls the handling of call-graphs (obtained via the
0428 -g/--call-graph options).
0429
0430 call-graph.record-mode::
0431 The mode for user space can be 'fp' (frame pointer), 'dwarf'
0432 and 'lbr'. The value 'dwarf' is effective only if libunwind
0433 (or a recent version of libdw) is present on the system;
0434 the value 'lbr' only works for certain cpus. The method for
0435 kernel space is controlled not by this option but by the
0436 kernel config (CONFIG_UNWINDER_*).
0437
0438 call-graph.dump-size::
0439 The size of stack to dump in order to do post-unwinding. Default is 8192 (byte).
0440 When using dwarf into record-mode, the default size will be used if omitted.
0441
0442 call-graph.print-type::
0443 The print-types can be graph (graph absolute), fractal (graph relative),
0444 flat and folded. This option controls a way to show overhead for each callchain
0445 entry. Suppose a following example.
0446
0447 Overhead Symbols
0448 ........ .......
0449 40.00% foo
0450 |
0451 ---foo
0452 |
0453 |--50.00%--bar
0454 | main
0455 |
0456 --50.00%--baz
0457 main
0458
0459 This output is a 'fractal' format. The 'foo' came from 'bar' and 'baz' exactly
0460 half and half so 'fractal' shows 50.00% for each
0461 (meaning that it assumes 100% total overhead of 'foo').
0462
0463 The 'graph' uses absolute overhead value of 'foo' as total so each of
0464 'bar' and 'baz' callchain will have 20.00% of overhead.
0465 If 'flat' is used, single column and linear exposure of call chains.
0466 'folded' mean call chains are displayed in a line, separated by semicolons.
0467
0468 call-graph.order::
0469 This option controls print order of callchains. The default is
0470 'callee' which means callee is printed at top and then followed by its
0471 caller and so on. The 'caller' prints it in reverse order.
0472
0473 If this option is not set and report.children or top.children is
0474 set to true (or the equivalent command line option is given),
0475 the default value of this option is changed to 'caller' for the
0476 execution of 'perf report' or 'perf top'. Other commands will
0477 still default to 'callee'.
0478
0479 call-graph.sort-key::
0480 The callchains are merged if they contain same information.
0481 The sort-key option determines a way to compare the callchains.
0482 A value of 'sort-key' can be 'function' or 'address'.
0483 The default is 'function'.
0484
0485 call-graph.threshold::
0486 When there're many callchains it'd print tons of lines. So perf omits
0487 small callchains under a certain overhead (threshold) and this option
0488 control the threshold. Default is 0.5 (%). The overhead is calculated
0489 by value depends on call-graph.print-type.
0490
0491 call-graph.print-limit::
0492 This is a maximum number of lines of callchain printed for a single
0493 histogram entry. Default is 0 which means no limitation.
0494
0495 report.*::
0496 report.sort_order::
0497 Allows changing the default sort order from "comm,dso,symbol" to
0498 some other default, for instance "sym,dso" may be more fitting for
0499 kernel developers.
0500 report.percent-limit::
0501 This one is mostly the same as call-graph.threshold but works for
0502 histogram entries. Entries having an overhead lower than this
0503 percentage will not be printed. Default is '0'. If percent-limit
0504 is '10', only entries which have more than 10% of overhead will be
0505 printed.
0506
0507 report.queue-size::
0508 This option sets up the maximum allocation size of the internal
0509 event queue for ordering events. Default is 0, meaning no limit.
0510
0511 report.children::
0512 'Children' means functions called from another function.
0513 If this option is true, 'perf report' cumulates callchains of children
0514 and show (accumulated) total overhead as well as 'Self' overhead.
0515 Please refer to the 'perf report' manual. The default is 'true'.
0516
0517 report.group::
0518 This option is to show event group information together.
0519 Example output with this turned on, notice that there is one column
0520 per event in the group, ref-cycles and cycles:
0521
0522 # group: {ref-cycles,cycles}
0523 # ========
0524 #
0525 # Samples: 7K of event 'anon group { ref-cycles, cycles }'
0526 # Event count (approx.): 6876107743
0527 #
0528 # Overhead Command Shared Object Symbol
0529 # ................ ....... ................. ...................
0530 #
0531 99.84% 99.76% noploop noploop [.] main
0532 0.07% 0.00% noploop ld-2.15.so [.] strcmp
0533 0.03% 0.00% noploop [kernel.kallsyms] [k] timerqueue_del
0534
0535 report.skip-empty::
0536 This option can change default stat behavior with empty results.
0537 If it's set true, 'perf report --stat' will not show 0 stats.
0538
0539 top.*::
0540 top.children::
0541 Same as 'report.children'. So if it is enabled, the output of 'top'
0542 command will have 'Children' overhead column as well as 'Self' overhead
0543 column by default.
0544 The default is 'true'.
0545
0546 top.call-graph::
0547 This is identical to 'call-graph.record-mode', except it is
0548 applicable only for 'top' subcommand. This option ONLY setup
0549 the unwind method. To enable 'perf top' to actually use it,
0550 the command line option -g must be specified.
0551
0552 man.*::
0553 man.viewer::
0554 This option can assign a tool to view manual pages when 'help'
0555 subcommand was invoked. Supported tools are 'man', 'woman'
0556 (with emacs client) and 'konqueror'. Default is 'man'.
0557
0558 New man viewer tool can be also added using 'man.<tool>.cmd'
0559 or use different path using 'man.<tool>.path' config option.
0560
0561 pager.*::
0562 pager.<subcommand>::
0563 When the subcommand is run on stdio, determine whether it uses
0564 pager or not based on this value. Default is 'unspecified'.
0565
0566 kmem.*::
0567 kmem.default::
0568 This option decides which allocator is to be analyzed if neither
0569 '--slab' nor '--page' option is used. Default is 'slab'.
0570
0571 record.*::
0572 record.build-id::
0573 This option can be 'cache', 'no-cache', 'skip' or 'mmap'.
0574 'cache' is to post-process data and save/update the binaries into
0575 the build-id cache (in ~/.debug). This is the default.
0576 But if this option is 'no-cache', it will not update the build-id cache.
0577 'skip' skips post-processing and does not update the cache.
0578 'mmap' skips post-processing and reads build-ids from MMAP events.
0579
0580 record.call-graph::
0581 This is identical to 'call-graph.record-mode', except it is
0582 applicable only for 'record' subcommand. This option ONLY setup
0583 the unwind method. To enable 'perf record' to actually use it,
0584 the command line option -g must be specified.
0585
0586 record.aio::
0587 Use 'n' control blocks in asynchronous (Posix AIO) trace writing
0588 mode ('n' default: 1, max: 4).
0589
0590 record.debuginfod::
0591 Specify debuginfod URL to be used when cacheing perf.data binaries,
0592 it follows the same syntax as the DEBUGINFOD_URLS variable, like:
0593
0594 http://192.168.122.174:8002
0595
0596 If the URLs is 'system', the value of DEBUGINFOD_URLS system environment
0597 variable is used.
0598
0599 diff.*::
0600 diff.order::
0601 This option sets the number of columns to sort the result.
0602 The default is 0, which means sorting by baseline.
0603 Setting it to 1 will sort the result by delta (or other
0604 compute method selected).
0605
0606 diff.compute::
0607 This options sets the method for computing the diff result.
0608 Possible values are 'delta', 'delta-abs', 'ratio' and
0609 'wdiff'. Default is 'delta'.
0610
0611 trace.*::
0612 trace.add_events::
0613 Allows adding a set of events to add to the ones specified
0614 by the user, or use as a default one if none was specified.
0615 The initial use case is to add augmented_raw_syscalls.o to
0616 activate the 'perf trace' logic that looks for syscall
0617 pointer contents after the normal tracepoint payload.
0618
0619 trace.args_alignment::
0620 Number of columns to align the argument list, default is 70,
0621 use 40 for the strace default, zero to no alignment.
0622
0623 trace.no_inherit::
0624 Do not follow children threads.
0625
0626 trace.show_arg_names::
0627 Should syscall argument names be printed? If not then trace.show_zeros
0628 will be set.
0629
0630 trace.show_duration::
0631 Show syscall duration.
0632
0633 trace.show_prefix::
0634 If set to 'yes' will show common string prefixes in tables. The default
0635 is to remove the common prefix in things like "MAP_SHARED", showing just "SHARED".
0636
0637 trace.show_timestamp::
0638 Show syscall start timestamp.
0639
0640 trace.show_zeros::
0641 Do not suppress syscall arguments that are equal to zero.
0642
0643 trace.tracepoint_beautifiers::
0644 Use "libtraceevent" to use that library to augment the tracepoint arguments,
0645 "libbeauty", the default, to use the same argument beautifiers used in the
0646 strace-like sys_enter+sys_exit lines.
0647
0648 ftrace.*::
0649 ftrace.tracer::
0650 Can be used to select the default tracer when neither -G nor
0651 -F option is not specified. Possible values are 'function' and
0652 'function_graph'.
0653
0654 llvm.*::
0655 llvm.clang-path::
0656 Path to clang. If omit, search it from $PATH.
0657
0658 llvm.clang-bpf-cmd-template::
0659 Cmdline template. Below lines show its default value. Environment
0660 variable is used to pass options.
0661 "$CLANG_EXEC -D__KERNEL__ -D__NR_CPUS__=$NR_CPUS "\
0662 "-DLINUX_VERSION_CODE=$LINUX_VERSION_CODE " \
0663 "$CLANG_OPTIONS $PERF_BPF_INC_OPTIONS $KERNEL_INC_OPTIONS " \
0664 "-Wno-unused-value -Wno-pointer-sign " \
0665 "-working-directory $WORKING_DIR " \
0666 "-c \"$CLANG_SOURCE\" -target bpf $CLANG_EMIT_LLVM -O2 -o - $LLVM_OPTIONS_PIPE"
0667
0668 llvm.clang-opt::
0669 Options passed to clang.
0670
0671 llvm.kbuild-dir::
0672 kbuild directory. If not set, use /lib/modules/`uname -r`/build.
0673 If set to "" deliberately, skip kernel header auto-detector.
0674
0675 llvm.kbuild-opts::
0676 Options passed to 'make' when detecting kernel header options.
0677
0678 llvm.dump-obj::
0679 Enable perf dump BPF object files compiled by LLVM.
0680
0681 llvm.opts::
0682 Options passed to llc.
0683
0684 samples.*::
0685
0686 samples.context::
0687 Define how many ns worth of time to show
0688 around samples in perf report sample context browser.
0689
0690 scripts.*::
0691
0692 Any option defines a script that is added to the scripts menu
0693 in the interactive perf browser and whose output is displayed.
0694 The name of the option is the name, the value is a script command line.
0695 The script gets the same options passed as a full perf script,
0696 in particular -i perfdata file, --cpu, --tid
0697
0698 convert.*::
0699
0700 convert.queue-size::
0701 Limit the size of ordered_events queue, so we could control
0702 allocation size of perf data files without proper finished
0703 round events.
0704 stat.*::
0705
0706 stat.big-num::
0707 (boolean) Change the default for "--big-num". To make
0708 "--no-big-num" the default, set "stat.big-num=false".
0709
0710 intel-pt.*::
0711
0712 intel-pt.cache-divisor::
0713
0714 intel-pt.mispred-all::
0715 If set, Intel PT decoder will set the mispred flag on all
0716 branches.
0717
0718 intel-pt.max-loops::
0719 If set and non-zero, the maximum number of unconditional
0720 branches decoded without consuming any trace packets. If
0721 the maximum is exceeded there will be a "Never-ending loop"
0722 error. The default is 100000.
0723
0724 auxtrace.*::
0725
0726 auxtrace.dumpdir::
0727 s390 only. The directory to save the auxiliary trace buffer
0728 can be changed using this option. Ex, auxtrace.dumpdir=/tmp.
0729 If the directory does not exist or has the wrong file type,
0730 the current directory is used.
0731
0732 daemon.*::
0733
0734 daemon.base::
0735 Base path for daemon data. All sessions data are stored under
0736 this path.
0737
0738 session-<NAME>.*::
0739
0740 session-<NAME>.run::
0741
0742 Defines new record session for daemon. The value is record's
0743 command line without the 'record' keyword.
0744
0745
0746 SEE ALSO
0747 --------
0748 linkperf:perf[1]
