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 - symbol: name of function executed at the time of sample 84 - symbol_size: size of function executed at the time of sample 85 - parent: name of function matched to the parent regex filter. Unmatched 86 entries are displayed as "[other]". 87 - cpu: cpu number the task ran at the time of sample 88 - socket: processor socket number the task ran at the time of sample 89 - srcline: filename and line number executed at the time of sample. The 90 DWARF debugging info must be provided. 91 - srcfile: file name of the source file of the same. Requires dwarf 92 information. 93 - weight: Event specific weight, e.g. memory latency or transaction 94 abort cost. This is the global weight. 95 - local_weight: Local weight version of the weight above. 96 - cgroup_id: ID derived from cgroup namespace device and inode numbers. 97 - transaction: Transaction abort flags. 98 - overhead: Overhead percentage of sample 99 - overhead_sys: Overhead percentage of sample running in system mode 100 - overhead_us: Overhead percentage of sample running in user mode 101 - overhead_guest_sys: Overhead percentage of sample running in system mode 102 on guest machine 103 - overhead_guest_us: Overhead percentage of sample running in user mode on 104 guest machine 105 - sample: Number of sample 106 - period: Raw number of event count of sample 107 108 By default, comm, dso and symbol keys are used. 109 (i.e. --sort comm,dso,symbol) 110 111 If --branch-stack option is used, following sort keys are also 112 available: 113 114 - dso_from: name of library or module branched from 115 - dso_to: name of library or module branched to 116 - symbol_from: name of function branched from 117 - symbol_to: name of function branched to 118 - srcline_from: source file and line branched from 119 - srcline_to: source file and line branched to 120 - mispredict: "N" for predicted branch, "Y" for mispredicted branch 121 - in_tx: branch in TSX transaction 122 - abort: TSX transaction abort. 123 - cycles: Cycles in basic block 124 125 And default sort keys are changed to comm, dso_from, symbol_from, dso_to 126 and symbol_to, see '--branch-stack'. 127 128 If the --mem-mode option is used, the following sort keys are also available 129 (incompatible with --branch-stack): 130 symbol_daddr, dso_daddr, locked, tlb, mem, snoop, dcacheline. 131 132 - symbol_daddr: name of data symbol being executed on at the time of sample 133 - dso_daddr: name of library or module containing the data being executed 134 on at the time of the sample 135 - locked: whether the bus was locked at the time of the sample 136 - tlb: type of tlb access for the data at the time of the sample 137 - mem: type of memory access for the data at the time of the sample 138 - snoop: type of snoop (if any) for the data at the time of the sample 139 - dcacheline: the cacheline the data address is on at the time of the sample 140 - phys_daddr: physical address of data being executed on at the time of sample 141 142 And the default sort keys are changed to local_weight, mem, sym, dso, 143 symbol_daddr, dso_daddr, snoop, tlb, locked, see '--mem-mode'. 144 145 If the data file has tracepoint event(s), following (dynamic) sort keys 146 are also available: 147 trace, trace_fields, [<event>.]<field>[/raw] 148 149 - trace: pretty printed trace output in a single column 150 - trace_fields: fields in tracepoints in separate columns 151 - <field name>: optional event and field name for a specific field 152 153 The last form consists of event and field names. If event name is 154 omitted, it searches all events for matching field name. The matched 155 field will be shown only for the event has the field. The event name 156 supports substring match so user doesn't need to specify full subsystem 157 and event name everytime. For example, 'sched:sched_switch' event can 158 be shortened to 'switch' as long as it's not ambiguous. Also event can 159 be specified by its index (starting from 1) preceded by the '%'. 160 So '%1' is the first event, '%2' is the second, and so on. 161 162 The field name can have '/raw' suffix which disables pretty printing 163 and shows raw field value like hex numbers. The --raw-trace option 164 has the same effect for all dynamic sort keys. 165 166 The default sort keys are changed to 'trace' if all events in the data 167 file are tracepoint. 168 169-F:: 170--fields=:: 171 Specify output field - multiple keys can be specified in CSV format. 172 Following fields are available: 173 overhead, overhead_sys, overhead_us, overhead_children, sample and period. 174 Also it can contain any sort key(s). 175 176 By default, every sort keys not specified in -F will be appended 177 automatically. 178 179 If the keys starts with a prefix '+', then it will append the specified 180 field(s) to the default field order. For example: perf report -F +period,sample. 181 182-p:: 183--parent=<regex>:: 184 A regex filter to identify parent. The parent is a caller of this 185 function and searched through the callchain, thus it requires callchain 186 information recorded. The pattern is in the extended regex format and 187 defaults to "\^sys_|^do_page_fault", see '--sort parent'. 188 189-x:: 190--exclude-other:: 191 Only display entries with parent-match. 192 193-w:: 194--column-widths=<width[,width...]>:: 195 Force each column width to the provided list, for large terminal 196 readability. 0 means no limit (default behavior). 197 198-t:: 199--field-separator=:: 200 Use a special separator character and don't pad with spaces, replacing 201 all occurrences of this separator in symbol names (and other output) 202 with a '.' character, that thus it's the only non valid separator. 203 204-D:: 205--dump-raw-trace:: 206 Dump raw trace in ASCII. 207 208-g:: 209--call-graph=<print_type,threshold[,print_limit],order,sort_key[,branch],value>:: 210 Display call chains using type, min percent threshold, print limit, 211 call order, sort key, optional branch and value. Note that ordering 212 is not fixed so any parameter can be given in an arbitrary order. 213 One exception is the print_limit which should be preceded by threshold. 214 215 print_type can be either: 216 - flat: single column, linear exposure of call chains. 217 - graph: use a graph tree, displaying absolute overhead rates. (default) 218 - fractal: like graph, but displays relative rates. Each branch of 219 the tree is considered as a new profiled object. 220 - folded: call chains are displayed in a line, separated by semicolons 221 - none: disable call chain display. 222 223 threshold is a percentage value which specifies a minimum percent to be 224 included in the output call graph. Default is 0.5 (%). 225 226 print_limit is only applied when stdio interface is used. It's to limit 227 number of call graph entries in a single hist entry. Note that it needs 228 to be given after threshold (but not necessarily consecutive). 229 Default is 0 (unlimited). 230 231 order can be either: 232 - callee: callee based call graph. 233 - caller: inverted caller based call graph. 234 Default is 'caller' when --children is used, otherwise 'callee'. 235 236 sort_key can be: 237 - function: compare on functions (default) 238 - address: compare on individual code addresses 239 - srcline: compare on source filename and line number 240 241 branch can be: 242 - branch: include last branch information in callgraph when available. 243 Usually more convenient to use --branch-history for this. 244 245 value can be: 246 - percent: diplay overhead percent (default) 247 - period: display event period 248 - count: display event count 249 250--children:: 251 Accumulate callchain of children to parent entry so that then can 252 show up in the output. The output will have a new "Children" column 253 and will be sorted on the data. It requires callchains are recorded. 254 See the `overhead calculation' section for more details. Enabled by 255 default, disable with --no-children. 256 257--max-stack:: 258 Set the stack depth limit when parsing the callchain, anything 259 beyond the specified depth will be ignored. This is a trade-off 260 between information loss and faster processing especially for 261 workloads that can have a very long callchain stack. 262 Note that when using the --itrace option the synthesized callchain size 263 will override this value if the synthesized callchain size is bigger. 264 265 Default: 127 266 267-G:: 268--inverted:: 269 alias for inverted caller based call graph. 270 271--ignore-callees=<regex>:: 272 Ignore callees of the function(s) matching the given regex. 273 This has the effect of collecting the callers of each such 274 function into one place in the call-graph tree. 275 276--pretty=<key>:: 277 Pretty printing style. key: normal, raw 278 279--stdio:: Use the stdio interface. 280 281--stdio-color:: 282 'always', 'never' or 'auto', allowing configuring color output 283 via the command line, in addition to via "color.ui" .perfconfig. 284 Use '--stdio-color always' to generate color even when redirecting 285 to a pipe or file. Using just '--stdio-color' is equivalent to 286 using 'always'. 287 288--tui:: Use the TUI interface, that is integrated with annotate and allows 289 zooming into DSOs or threads, among other features. Use of --tui 290 requires a tty, if one is not present, as when piping to other 291 commands, the stdio interface is used. 292 293--gtk:: Use the GTK2 interface. 294 295-k:: 296--vmlinux=<file>:: 297 vmlinux pathname 298 299--ignore-vmlinux:: 300 Ignore vmlinux files. 301 302--kallsyms=<file>:: 303 kallsyms pathname 304 305-m:: 306--modules:: 307 Load module symbols. WARNING: This should only be used with -k and 308 a LIVE kernel. 309 310-f:: 311--force:: 312 Don't do ownership validation. 313 314--symfs=<directory>:: 315 Look for files with symbols relative to this directory. 316 317-C:: 318--cpu:: Only report samples for the list of CPUs provided. Multiple CPUs can 319 be provided as a comma-separated list with no space: 0,1. Ranges of 320 CPUs are specified with -: 0-2. Default is to report samples on all 321 CPUs. 322 323-M:: 324--disassembler-style=:: Set disassembler style for objdump. 325 326--source:: 327 Interleave source code with assembly code. Enabled by default, 328 disable with --no-source. 329 330--asm-raw:: 331 Show raw instruction encoding of assembly instructions. 332 333--show-total-period:: Show a column with the sum of periods. 334 335-I:: 336--show-info:: 337 Display extended information about the perf.data file. This adds 338 information which may be very large and thus may clutter the display. 339 It currently includes: cpu and numa topology of the host system. 340 341-b:: 342--branch-stack:: 343 Use the addresses of sampled taken branches instead of the instruction 344 address to build the histograms. To generate meaningful output, the 345 perf.data file must have been obtained using perf record -b or 346 perf record --branch-filter xxx where xxx is a branch filter option. 347 perf report is able to auto-detect whether a perf.data file contains 348 branch stacks and it will automatically switch to the branch view mode, 349 unless --no-branch-stack is used. 350 351--branch-history:: 352 Add the addresses of sampled taken branches to the callstack. 353 This allows to examine the path the program took to each sample. 354 The data collection must have used -b (or -j) and -g. 355 356--objdump=<path>:: 357 Path to objdump binary. 358 359--group:: 360 Show event group information together. It forces group output also 361 if there are no groups defined in data file. 362 363--demangle:: 364 Demangle symbol names to human readable form. It's enabled by default, 365 disable with --no-demangle. 366 367--demangle-kernel:: 368 Demangle kernel symbol names to human readable form (for C++ kernels). 369 370--mem-mode:: 371 Use the data addresses of samples in addition to instruction addresses 372 to build the histograms. To generate meaningful output, the perf.data 373 file must have been obtained using perf record -d -W and using a 374 special event -e cpu/mem-loads/p or -e cpu/mem-stores/p. See 375 'perf mem' for simpler access. 376 377--percent-limit:: 378 Do not show entries which have an overhead under that percent. 379 (Default: 0). Note that this option also sets the percent limit (threshold) 380 of callchains. However the default value of callchain threshold is 381 different than the default value of hist entries. Please see the 382 --call-graph option for details. 383 384--percentage:: 385 Determine how to display the overhead percentage of filtered entries. 386 Filters can be applied by --comms, --dsos and/or --symbols options and 387 Zoom operations on the TUI (thread, dso, etc). 388 389 "relative" means it's relative to filtered entries only so that the 390 sum of shown entries will be always 100%. "absolute" means it retains 391 the original value before and after the filter is applied. 392 393--header:: 394 Show header information in the perf.data file. This includes 395 various information like hostname, OS and perf version, cpu/mem 396 info, perf command line, event list and so on. Currently only 397 --stdio output supports this feature. 398 399--header-only:: 400 Show only perf.data header (forces --stdio). 401 402--time:: 403 Only analyze samples within given time window: <start>,<stop>. Times 404 have the format seconds.microseconds. If start is not given (i.e., time 405 string is ',x.y') then analysis starts at the beginning of the file. If 406 stop time is not given (i.e, time string is 'x.y,') then analysis goes 407 to end of file. 408 409 Also support time percent with multiple time range. Time string is 410 'a%/n,b%/m,...' or 'a%-b%,c%-%d,...'. 411 412 For example: 413 Select the second 10% time slice: 414 415 perf report --time 10%/2 416 417 Select from 0% to 10% time slice: 418 419 perf report --time 0%-10% 420 421 Select the first and second 10% time slices: 422 423 perf report --time 10%/1,10%/2 424 425 Select from 0% to 10% and 30% to 40% slices: 426 427 perf report --time 0%-10%,30%-40% 428 429--itrace:: 430 Options for decoding instruction tracing data. The options are: 431 432include::itrace.txt[] 433 434 To disable decoding entirely, use --no-itrace. 435 436--full-source-path:: 437 Show the full path for source files for srcline output. 438 439--show-ref-call-graph:: 440 When multiple events are sampled, it may not be needed to collect 441 callgraphs for all of them. The sample sites are usually nearby, 442 and it's enough to collect the callgraphs on a reference event. 443 So user can use "call-graph=no" event modifier to disable callgraph 444 for other events to reduce the overhead. 445 However, perf report cannot show callgraphs for the event which 446 disable the callgraph. 447 This option extends the perf report to show reference callgraphs, 448 which collected by reference event, in no callgraph event. 449 450--socket-filter:: 451 Only report the samples on the processor socket that match with this filter 452 453--raw-trace:: 454 When displaying traceevent output, do not use print fmt or plugins. 455 456--hierarchy:: 457 Enable hierarchical output. 458 459--inline:: 460 If a callgraph address belongs to an inlined function, the inline stack 461 will be printed. Each entry is function name or file/line. Enabled by 462 default, disable with --no-inline. 463 464--mmaps:: 465 Show --tasks output plus mmap information in a format similar to 466 /proc/<PID>/maps. 467 468 Please note that not all mmaps are stored, options affecting which ones 469 are include 'perf record --data', for instance. 470 471--stats:: 472 Display overall events statistics without any further processing. 473 (like the one at the end of the perf report -D command) 474 475--tasks:: 476 Display monitored tasks stored in perf data. Displaying pid/tid/ppid 477 plus the command string aligned to distinguish parent and child tasks. 478 479include::callchain-overhead-calculation.txt[] 480 481SEE ALSO 482-------- 483linkperf:perf-stat[1], linkperf:perf-annotate[1], linkperf:perf-record[1] 484