1======================== 2ftrace - Function Tracer 3======================== 4 5Copyright 2008 Red Hat Inc. 6 7:Author: Steven Rostedt <srostedt@redhat.com> 8:License: The GNU Free Documentation License, Version 1.2 9 (dual licensed under the GPL v2) 10:Original Reviewers: Elias Oltmanns, Randy Dunlap, Andrew Morton, 11 John Kacur, and David Teigland. 12 13- Written for: 2.6.28-rc2 14- Updated for: 3.10 15- Updated for: 4.13 - Copyright 2017 VMware Inc. Steven Rostedt 16- Converted to rst format - Changbin Du <changbin.du@intel.com> 17 18Introduction 19------------ 20 21Ftrace is an internal tracer designed to help out developers and 22designers of systems to find what is going on inside the kernel. 23It can be used for debugging or analyzing latencies and 24performance issues that take place outside of user-space. 25 26Although ftrace is typically considered the function tracer, it 27is really a framework of several assorted tracing utilities. 28There's latency tracing to examine what occurs between interrupts 29disabled and enabled, as well as for preemption and from a time 30a task is woken to the task is actually scheduled in. 31 32One of the most common uses of ftrace is the event tracing. 33Throughout the kernel is hundreds of static event points that 34can be enabled via the tracefs file system to see what is 35going on in certain parts of the kernel. 36 37See events.txt for more information. 38 39 40Implementation Details 41---------------------- 42 43See :doc:`ftrace-design` for details for arch porters and such. 44 45 46The File System 47--------------- 48 49Ftrace uses the tracefs file system to hold the control files as 50well as the files to display output. 51 52When tracefs is configured into the kernel (which selecting any ftrace 53option will do) the directory /sys/kernel/tracing will be created. To mount 54this directory, you can add to your /etc/fstab file:: 55 56 tracefs /sys/kernel/tracing tracefs defaults 0 0 57 58Or you can mount it at run time with:: 59 60 mount -t tracefs nodev /sys/kernel/tracing 61 62For quicker access to that directory you may want to make a soft link to 63it:: 64 65 ln -s /sys/kernel/tracing /tracing 66 67.. attention:: 68 69 Before 4.1, all ftrace tracing control files were within the debugfs 70 file system, which is typically located at /sys/kernel/debug/tracing. 71 For backward compatibility, when mounting the debugfs file system, 72 the tracefs file system will be automatically mounted at: 73 74 /sys/kernel/debug/tracing 75 76 All files located in the tracefs file system will be located in that 77 debugfs file system directory as well. 78 79.. attention:: 80 81 Any selected ftrace option will also create the tracefs file system. 82 The rest of the document will assume that you are in the ftrace directory 83 (cd /sys/kernel/tracing) and will only concentrate on the files within that 84 directory and not distract from the content with the extended 85 "/sys/kernel/tracing" path name. 86 87That's it! (assuming that you have ftrace configured into your kernel) 88 89After mounting tracefs you will have access to the control and output files 90of ftrace. Here is a list of some of the key files: 91 92 93 Note: all time values are in microseconds. 94 95 current_tracer: 96 97 This is used to set or display the current tracer 98 that is configured. Changing the current tracer clears 99 the ring buffer content as well as the "snapshot" buffer. 100 101 available_tracers: 102 103 This holds the different types of tracers that 104 have been compiled into the kernel. The 105 tracers listed here can be configured by 106 echoing their name into current_tracer. 107 108 tracing_on: 109 110 This sets or displays whether writing to the trace 111 ring buffer is enabled. Echo 0 into this file to disable 112 the tracer or 1 to enable it. Note, this only disables 113 writing to the ring buffer, the tracing overhead may 114 still be occurring. 115 116 The kernel function tracing_off() can be used within the 117 kernel to disable writing to the ring buffer, which will 118 set this file to "0". User space can re-enable tracing by 119 echoing "1" into the file. 120 121 Note, the function and event trigger "traceoff" will also 122 set this file to zero and stop tracing. Which can also 123 be re-enabled by user space using this file. 124 125 trace: 126 127 This file holds the output of the trace in a human 128 readable format (described below). Opening this file for 129 writing with the O_TRUNC flag clears the ring buffer content. 130 Note, this file is not a consumer. If tracing is off 131 (no tracer running, or tracing_on is zero), it will produce 132 the same output each time it is read. When tracing is on, 133 it may produce inconsistent results as it tries to read 134 the entire buffer without consuming it. 135 136 trace_pipe: 137 138 The output is the same as the "trace" file but this 139 file is meant to be streamed with live tracing. 140 Reads from this file will block until new data is 141 retrieved. Unlike the "trace" file, this file is a 142 consumer. This means reading from this file causes 143 sequential reads to display more current data. Once 144 data is read from this file, it is consumed, and 145 will not be read again with a sequential read. The 146 "trace" file is static, and if the tracer is not 147 adding more data, it will display the same 148 information every time it is read. 149 150 trace_options: 151 152 This file lets the user control the amount of data 153 that is displayed in one of the above output 154 files. Options also exist to modify how a tracer 155 or events work (stack traces, timestamps, etc). 156 157 options: 158 159 This is a directory that has a file for every available 160 trace option (also in trace_options). Options may also be set 161 or cleared by writing a "1" or "0" respectively into the 162 corresponding file with the option name. 163 164 tracing_max_latency: 165 166 Some of the tracers record the max latency. 167 For example, the maximum time that interrupts are disabled. 168 The maximum time is saved in this file. The max trace will also be 169 stored, and displayed by "trace". A new max trace will only be 170 recorded if the latency is greater than the value in this file 171 (in microseconds). 172 173 By echoing in a time into this file, no latency will be recorded 174 unless it is greater than the time in this file. 175 176 tracing_thresh: 177 178 Some latency tracers will record a trace whenever the 179 latency is greater than the number in this file. 180 Only active when the file contains a number greater than 0. 181 (in microseconds) 182 183 buffer_size_kb: 184 185 This sets or displays the number of kilobytes each CPU 186 buffer holds. By default, the trace buffers are the same size 187 for each CPU. The displayed number is the size of the 188 CPU buffer and not total size of all buffers. The 189 trace buffers are allocated in pages (blocks of memory 190 that the kernel uses for allocation, usually 4 KB in size). 191 A few extra pages may be allocated to accommodate buffer management 192 meta-data. If the last page allocated has room for more bytes 193 than requested, the rest of the page will be used, 194 making the actual allocation bigger than requested or shown. 195 ( Note, the size may not be a multiple of the page size 196 due to buffer management meta-data. ) 197 198 Buffer sizes for individual CPUs may vary 199 (see "per_cpu/cpu0/buffer_size_kb" below), and if they do 200 this file will show "X". 201 202 buffer_total_size_kb: 203 204 This displays the total combined size of all the trace buffers. 205 206 free_buffer: 207 208 If a process is performing tracing, and the ring buffer should be 209 shrunk "freed" when the process is finished, even if it were to be 210 killed by a signal, this file can be used for that purpose. On close 211 of this file, the ring buffer will be resized to its minimum size. 212 Having a process that is tracing also open this file, when the process 213 exits its file descriptor for this file will be closed, and in doing so, 214 the ring buffer will be "freed". 215 216 It may also stop tracing if disable_on_free option is set. 217 218 tracing_cpumask: 219 220 This is a mask that lets the user only trace on specified CPUs. 221 The format is a hex string representing the CPUs. 222 223 set_ftrace_filter: 224 225 When dynamic ftrace is configured in (see the 226 section below "dynamic ftrace"), the code is dynamically 227 modified (code text rewrite) to disable calling of the 228 function profiler (mcount). This lets tracing be configured 229 in with practically no overhead in performance. This also 230 has a side effect of enabling or disabling specific functions 231 to be traced. Echoing names of functions into this file 232 will limit the trace to only those functions. 233 This influences the tracers "function" and "function_graph" 234 and thus also function profiling (see "function_profile_enabled"). 235 236 The functions listed in "available_filter_functions" are what 237 can be written into this file. 238 239 This interface also allows for commands to be used. See the 240 "Filter commands" section for more details. 241 242 As a speed up, since processing strings can be quite expensive 243 and requires a check of all functions registered to tracing, instead 244 an index can be written into this file. A number (starting with "1") 245 written will instead select the same corresponding at the line position 246 of the "available_filter_functions" file. 247 248 set_ftrace_notrace: 249 250 This has an effect opposite to that of 251 set_ftrace_filter. Any function that is added here will not 252 be traced. If a function exists in both set_ftrace_filter 253 and set_ftrace_notrace, the function will _not_ be traced. 254 255 set_ftrace_pid: 256 257 Have the function tracer only trace the threads whose PID are 258 listed in this file. 259 260 If the "function-fork" option is set, then when a task whose 261 PID is listed in this file forks, the child's PID will 262 automatically be added to this file, and the child will be 263 traced by the function tracer as well. This option will also 264 cause PIDs of tasks that exit to be removed from the file. 265 266 set_ftrace_notrace_pid: 267 268 Have the function tracer ignore threads whose PID are listed in 269 this file. 270 271 If the "function-fork" option is set, then when a task whose 272 PID is listed in this file forks, the child's PID will 273 automatically be added to this file, and the child will not be 274 traced by the function tracer as well. This option will also 275 cause PIDs of tasks that exit to be removed from the file. 276 277 If a PID is in both this file and "set_ftrace_pid", then this 278 file takes precedence, and the thread will not be traced. 279 280 set_event_pid: 281 282 Have the events only trace a task with a PID listed in this file. 283 Note, sched_switch and sched_wake_up will also trace events 284 listed in this file. 285 286 To have the PIDs of children of tasks with their PID in this file 287 added on fork, enable the "event-fork" option. That option will also 288 cause the PIDs of tasks to be removed from this file when the task 289 exits. 290 291 set_event_notrace_pid: 292 293 Have the events not trace a task with a PID listed in this file. 294 Note, sched_switch and sched_wakeup will trace threads not listed 295 in this file, even if a thread's PID is in the file if the 296 sched_switch or sched_wakeup events also trace a thread that should 297 be traced. 298 299 To have the PIDs of children of tasks with their PID in this file 300 added on fork, enable the "event-fork" option. That option will also 301 cause the PIDs of tasks to be removed from this file when the task 302 exits. 303 304 set_graph_function: 305 306 Functions listed in this file will cause the function graph 307 tracer to only trace these functions and the functions that 308 they call. (See the section "dynamic ftrace" for more details). 309 Note, set_ftrace_filter and set_ftrace_notrace still affects 310 what functions are being traced. 311 312 set_graph_notrace: 313 314 Similar to set_graph_function, but will disable function graph 315 tracing when the function is hit until it exits the function. 316 This makes it possible to ignore tracing functions that are called 317 by a specific function. 318 319 available_filter_functions: 320 321 This lists the functions that ftrace has processed and can trace. 322 These are the function names that you can pass to 323 "set_ftrace_filter", "set_ftrace_notrace", 324 "set_graph_function", or "set_graph_notrace". 325 (See the section "dynamic ftrace" below for more details.) 326 327 dyn_ftrace_total_info: 328 329 This file is for debugging purposes. The number of functions that 330 have been converted to nops and are available to be traced. 331 332 enabled_functions: 333 334 This file is more for debugging ftrace, but can also be useful 335 in seeing if any function has a callback attached to it. 336 Not only does the trace infrastructure use ftrace function 337 trace utility, but other subsystems might too. This file 338 displays all functions that have a callback attached to them 339 as well as the number of callbacks that have been attached. 340 Note, a callback may also call multiple functions which will 341 not be listed in this count. 342 343 If the callback registered to be traced by a function with 344 the "save regs" attribute (thus even more overhead), a 'R' 345 will be displayed on the same line as the function that 346 is returning registers. 347 348 If the callback registered to be traced by a function with 349 the "ip modify" attribute (thus the regs->ip can be changed), 350 an 'I' will be displayed on the same line as the function that 351 can be overridden. 352 353 If the architecture supports it, it will also show what callback 354 is being directly called by the function. If the count is greater 355 than 1 it most likely will be ftrace_ops_list_func(). 356 357 If the callback of the function jumps to a trampoline that is 358 specific to a the callback and not the standard trampoline, 359 its address will be printed as well as the function that the 360 trampoline calls. 361 362 function_profile_enabled: 363 364 When set it will enable all functions with either the function 365 tracer, or if configured, the function graph tracer. It will 366 keep a histogram of the number of functions that were called 367 and if the function graph tracer was configured, it will also keep 368 track of the time spent in those functions. The histogram 369 content can be displayed in the files: 370 371 trace_stat/function<cpu> ( function0, function1, etc). 372 373 trace_stat: 374 375 A directory that holds different tracing stats. 376 377 kprobe_events: 378 379 Enable dynamic trace points. See kprobetrace.txt. 380 381 kprobe_profile: 382 383 Dynamic trace points stats. See kprobetrace.txt. 384 385 max_graph_depth: 386 387 Used with the function graph tracer. This is the max depth 388 it will trace into a function. Setting this to a value of 389 one will show only the first kernel function that is called 390 from user space. 391 392 printk_formats: 393 394 This is for tools that read the raw format files. If an event in 395 the ring buffer references a string, only a pointer to the string 396 is recorded into the buffer and not the string itself. This prevents 397 tools from knowing what that string was. This file displays the string 398 and address for the string allowing tools to map the pointers to what 399 the strings were. 400 401 saved_cmdlines: 402 403 Only the pid of the task is recorded in a trace event unless 404 the event specifically saves the task comm as well. Ftrace 405 makes a cache of pid mappings to comms to try to display 406 comms for events. If a pid for a comm is not listed, then 407 "<...>" is displayed in the output. 408 409 If the option "record-cmd" is set to "0", then comms of tasks 410 will not be saved during recording. By default, it is enabled. 411 412 saved_cmdlines_size: 413 414 By default, 128 comms are saved (see "saved_cmdlines" above). To 415 increase or decrease the amount of comms that are cached, echo 416 the number of comms to cache into this file. 417 418 saved_tgids: 419 420 If the option "record-tgid" is set, on each scheduling context switch 421 the Task Group ID of a task is saved in a table mapping the PID of 422 the thread to its TGID. By default, the "record-tgid" option is 423 disabled. 424 425 snapshot: 426 427 This displays the "snapshot" buffer and also lets the user 428 take a snapshot of the current running trace. 429 See the "Snapshot" section below for more details. 430 431 stack_max_size: 432 433 When the stack tracer is activated, this will display the 434 maximum stack size it has encountered. 435 See the "Stack Trace" section below. 436 437 stack_trace: 438 439 This displays the stack back trace of the largest stack 440 that was encountered when the stack tracer is activated. 441 See the "Stack Trace" section below. 442 443 stack_trace_filter: 444 445 This is similar to "set_ftrace_filter" but it limits what 446 functions the stack tracer will check. 447 448 trace_clock: 449 450 Whenever an event is recorded into the ring buffer, a 451 "timestamp" is added. This stamp comes from a specified 452 clock. By default, ftrace uses the "local" clock. This 453 clock is very fast and strictly per cpu, but on some 454 systems it may not be monotonic with respect to other 455 CPUs. In other words, the local clocks may not be in sync 456 with local clocks on other CPUs. 457 458 Usual clocks for tracing:: 459 460 # cat trace_clock 461 [local] global counter x86-tsc 462 463 The clock with the square brackets around it is the one in effect. 464 465 local: 466 Default clock, but may not be in sync across CPUs 467 468 global: 469 This clock is in sync with all CPUs but may 470 be a bit slower than the local clock. 471 472 counter: 473 This is not a clock at all, but literally an atomic 474 counter. It counts up one by one, but is in sync 475 with all CPUs. This is useful when you need to 476 know exactly the order events occurred with respect to 477 each other on different CPUs. 478 479 uptime: 480 This uses the jiffies counter and the time stamp 481 is relative to the time since boot up. 482 483 perf: 484 This makes ftrace use the same clock that perf uses. 485 Eventually perf will be able to read ftrace buffers 486 and this will help out in interleaving the data. 487 488 x86-tsc: 489 Architectures may define their own clocks. For 490 example, x86 uses its own TSC cycle clock here. 491 492 ppc-tb: 493 This uses the powerpc timebase register value. 494 This is in sync across CPUs and can also be used 495 to correlate events across hypervisor/guest if 496 tb_offset is known. 497 498 mono: 499 This uses the fast monotonic clock (CLOCK_MONOTONIC) 500 which is monotonic and is subject to NTP rate adjustments. 501 502 mono_raw: 503 This is the raw monotonic clock (CLOCK_MONOTONIC_RAW) 504 which is monotonic but is not subject to any rate adjustments 505 and ticks at the same rate as the hardware clocksource. 506 507 boot: 508 This is the boot clock (CLOCK_BOOTTIME) and is based on the 509 fast monotonic clock, but also accounts for time spent in 510 suspend. Since the clock access is designed for use in 511 tracing in the suspend path, some side effects are possible 512 if clock is accessed after the suspend time is accounted before 513 the fast mono clock is updated. In this case, the clock update 514 appears to happen slightly sooner than it normally would have. 515 Also on 32-bit systems, it's possible that the 64-bit boot offset 516 sees a partial update. These effects are rare and post 517 processing should be able to handle them. See comments in the 518 ktime_get_boot_fast_ns() function for more information. 519 520 To set a clock, simply echo the clock name into this file:: 521 522 # echo global > trace_clock 523 524 Setting a clock clears the ring buffer content as well as the 525 "snapshot" buffer. 526 527 trace_marker: 528 529 This is a very useful file for synchronizing user space 530 with events happening in the kernel. Writing strings into 531 this file will be written into the ftrace buffer. 532 533 It is useful in applications to open this file at the start 534 of the application and just reference the file descriptor 535 for the file:: 536 537 void trace_write(const char *fmt, ...) 538 { 539 va_list ap; 540 char buf[256]; 541 int n; 542 543 if (trace_fd < 0) 544 return; 545 546 va_start(ap, fmt); 547 n = vsnprintf(buf, 256, fmt, ap); 548 va_end(ap); 549 550 write(trace_fd, buf, n); 551 } 552 553 start:: 554 555 trace_fd = open("trace_marker", WR_ONLY); 556 557 Note: Writing into the trace_marker file can also initiate triggers 558 that are written into /sys/kernel/tracing/events/ftrace/print/trigger 559 See "Event triggers" in Documentation/trace/events.rst and an 560 example in Documentation/trace/histogram.rst (Section 3.) 561 562 trace_marker_raw: 563 564 This is similar to trace_marker above, but is meant for for binary data 565 to be written to it, where a tool can be used to parse the data 566 from trace_pipe_raw. 567 568 uprobe_events: 569 570 Add dynamic tracepoints in programs. 571 See uprobetracer.txt 572 573 uprobe_profile: 574 575 Uprobe statistics. See uprobetrace.txt 576 577 instances: 578 579 This is a way to make multiple trace buffers where different 580 events can be recorded in different buffers. 581 See "Instances" section below. 582 583 events: 584 585 This is the trace event directory. It holds event tracepoints 586 (also known as static tracepoints) that have been compiled 587 into the kernel. It shows what event tracepoints exist 588 and how they are grouped by system. There are "enable" 589 files at various levels that can enable the tracepoints 590 when a "1" is written to them. 591 592 See events.txt for more information. 593 594 set_event: 595 596 By echoing in the event into this file, will enable that event. 597 598 See events.txt for more information. 599 600 available_events: 601 602 A list of events that can be enabled in tracing. 603 604 See events.txt for more information. 605 606 timestamp_mode: 607 608 Certain tracers may change the timestamp mode used when 609 logging trace events into the event buffer. Events with 610 different modes can coexist within a buffer but the mode in 611 effect when an event is logged determines which timestamp mode 612 is used for that event. The default timestamp mode is 613 'delta'. 614 615 Usual timestamp modes for tracing: 616 617 # cat timestamp_mode 618 [delta] absolute 619 620 The timestamp mode with the square brackets around it is the 621 one in effect. 622 623 delta: Default timestamp mode - timestamp is a delta against 624 a per-buffer timestamp. 625 626 absolute: The timestamp is a full timestamp, not a delta 627 against some other value. As such it takes up more 628 space and is less efficient. 629 630 hwlat_detector: 631 632 Directory for the Hardware Latency Detector. 633 See "Hardware Latency Detector" section below. 634 635 per_cpu: 636 637 This is a directory that contains the trace per_cpu information. 638 639 per_cpu/cpu0/buffer_size_kb: 640 641 The ftrace buffer is defined per_cpu. That is, there's a separate 642 buffer for each CPU to allow writes to be done atomically, 643 and free from cache bouncing. These buffers may have different 644 size buffers. This file is similar to the buffer_size_kb 645 file, but it only displays or sets the buffer size for the 646 specific CPU. (here cpu0). 647 648 per_cpu/cpu0/trace: 649 650 This is similar to the "trace" file, but it will only display 651 the data specific for the CPU. If written to, it only clears 652 the specific CPU buffer. 653 654 per_cpu/cpu0/trace_pipe 655 656 This is similar to the "trace_pipe" file, and is a consuming 657 read, but it will only display (and consume) the data specific 658 for the CPU. 659 660 per_cpu/cpu0/trace_pipe_raw 661 662 For tools that can parse the ftrace ring buffer binary format, 663 the trace_pipe_raw file can be used to extract the data 664 from the ring buffer directly. With the use of the splice() 665 system call, the buffer data can be quickly transferred to 666 a file or to the network where a server is collecting the 667 data. 668 669 Like trace_pipe, this is a consuming reader, where multiple 670 reads will always produce different data. 671 672 per_cpu/cpu0/snapshot: 673 674 This is similar to the main "snapshot" file, but will only 675 snapshot the current CPU (if supported). It only displays 676 the content of the snapshot for a given CPU, and if 677 written to, only clears this CPU buffer. 678 679 per_cpu/cpu0/snapshot_raw: 680 681 Similar to the trace_pipe_raw, but will read the binary format 682 from the snapshot buffer for the given CPU. 683 684 per_cpu/cpu0/stats: 685 686 This displays certain stats about the ring buffer: 687 688 entries: 689 The number of events that are still in the buffer. 690 691 overrun: 692 The number of lost events due to overwriting when 693 the buffer was full. 694 695 commit overrun: 696 Should always be zero. 697 This gets set if so many events happened within a nested 698 event (ring buffer is re-entrant), that it fills the 699 buffer and starts dropping events. 700 701 bytes: 702 Bytes actually read (not overwritten). 703 704 oldest event ts: 705 The oldest timestamp in the buffer 706 707 now ts: 708 The current timestamp 709 710 dropped events: 711 Events lost due to overwrite option being off. 712 713 read events: 714 The number of events read. 715 716The Tracers 717----------- 718 719Here is the list of current tracers that may be configured. 720 721 "function" 722 723 Function call tracer to trace all kernel functions. 724 725 "function_graph" 726 727 Similar to the function tracer except that the 728 function tracer probes the functions on their entry 729 whereas the function graph tracer traces on both entry 730 and exit of the functions. It then provides the ability 731 to draw a graph of function calls similar to C code 732 source. 733 734 "blk" 735 736 The block tracer. The tracer used by the blktrace user 737 application. 738 739 "hwlat" 740 741 The Hardware Latency tracer is used to detect if the hardware 742 produces any latency. See "Hardware Latency Detector" section 743 below. 744 745 "irqsoff" 746 747 Traces the areas that disable interrupts and saves 748 the trace with the longest max latency. 749 See tracing_max_latency. When a new max is recorded, 750 it replaces the old trace. It is best to view this 751 trace with the latency-format option enabled, which 752 happens automatically when the tracer is selected. 753 754 "preemptoff" 755 756 Similar to irqsoff but traces and records the amount of 757 time for which preemption is disabled. 758 759 "preemptirqsoff" 760 761 Similar to irqsoff and preemptoff, but traces and 762 records the largest time for which irqs and/or preemption 763 is disabled. 764 765 "wakeup" 766 767 Traces and records the max latency that it takes for 768 the highest priority task to get scheduled after 769 it has been woken up. 770 Traces all tasks as an average developer would expect. 771 772 "wakeup_rt" 773 774 Traces and records the max latency that it takes for just 775 RT tasks (as the current "wakeup" does). This is useful 776 for those interested in wake up timings of RT tasks. 777 778 "wakeup_dl" 779 780 Traces and records the max latency that it takes for 781 a SCHED_DEADLINE task to be woken (as the "wakeup" and 782 "wakeup_rt" does). 783 784 "mmiotrace" 785 786 A special tracer that is used to trace binary module. 787 It will trace all the calls that a module makes to the 788 hardware. Everything it writes and reads from the I/O 789 as well. 790 791 "branch" 792 793 This tracer can be configured when tracing likely/unlikely 794 calls within the kernel. It will trace when a likely and 795 unlikely branch is hit and if it was correct in its prediction 796 of being correct. 797 798 "nop" 799 800 This is the "trace nothing" tracer. To remove all 801 tracers from tracing simply echo "nop" into 802 current_tracer. 803 804Error conditions 805---------------- 806 807 For most ftrace commands, failure modes are obvious and communicated 808 using standard return codes. 809 810 For other more involved commands, extended error information may be 811 available via the tracing/error_log file. For the commands that 812 support it, reading the tracing/error_log file after an error will 813 display more detailed information about what went wrong, if 814 information is available. The tracing/error_log file is a circular 815 error log displaying a small number (currently, 8) of ftrace errors 816 for the last (8) failed commands. 817 818 The extended error information and usage takes the form shown in 819 this example:: 820 821 # echo xxx > /sys/kernel/debug/tracing/events/sched/sched_wakeup/trigger 822 echo: write error: Invalid argument 823 824 # cat /sys/kernel/debug/tracing/error_log 825 [ 5348.887237] location: error: Couldn't yyy: zzz 826 Command: xxx 827 ^ 828 [ 7517.023364] location: error: Bad rrr: sss 829 Command: ppp qqq 830 ^ 831 832 To clear the error log, echo the empty string into it:: 833 834 # echo > /sys/kernel/debug/tracing/error_log 835 836Examples of using the tracer 837---------------------------- 838 839Here are typical examples of using the tracers when controlling 840them only with the tracefs interface (without using any 841user-land utilities). 842 843Output format: 844-------------- 845 846Here is an example of the output format of the file "trace":: 847 848 # tracer: function 849 # 850 # entries-in-buffer/entries-written: 140080/250280 #P:4 851 # 852 # _-----=> irqs-off 853 # / _----=> need-resched 854 # | / _---=> hardirq/softirq 855 # || / _--=> preempt-depth 856 # ||| / delay 857 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 858 # | | | |||| | | 859 bash-1977 [000] .... 17284.993652: sys_close <-system_call_fastpath 860 bash-1977 [000] .... 17284.993653: __close_fd <-sys_close 861 bash-1977 [000] .... 17284.993653: _raw_spin_lock <-__close_fd 862 sshd-1974 [003] .... 17284.993653: __srcu_read_unlock <-fsnotify 863 bash-1977 [000] .... 17284.993654: add_preempt_count <-_raw_spin_lock 864 bash-1977 [000] ...1 17284.993655: _raw_spin_unlock <-__close_fd 865 bash-1977 [000] ...1 17284.993656: sub_preempt_count <-_raw_spin_unlock 866 bash-1977 [000] .... 17284.993657: filp_close <-__close_fd 867 bash-1977 [000] .... 17284.993657: dnotify_flush <-filp_close 868 sshd-1974 [003] .... 17284.993658: sys_select <-system_call_fastpath 869 .... 870 871A header is printed with the tracer name that is represented by 872the trace. In this case the tracer is "function". Then it shows the 873number of events in the buffer as well as the total number of entries 874that were written. The difference is the number of entries that were 875lost due to the buffer filling up (250280 - 140080 = 110200 events 876lost). 877 878The header explains the content of the events. Task name "bash", the task 879PID "1977", the CPU that it was running on "000", the latency format 880(explained below), the timestamp in <secs>.<usecs> format, the 881function name that was traced "sys_close" and the parent function that 882called this function "system_call_fastpath". The timestamp is the time 883at which the function was entered. 884 885Latency trace format 886-------------------- 887 888When the latency-format option is enabled or when one of the latency 889tracers is set, the trace file gives somewhat more information to see 890why a latency happened. Here is a typical trace:: 891 892 # tracer: irqsoff 893 # 894 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 895 # -------------------------------------------------------------------- 896 # latency: 259 us, #4/4, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 897 # ----------------- 898 # | task: ps-6143 (uid:0 nice:0 policy:0 rt_prio:0) 899 # ----------------- 900 # => started at: __lock_task_sighand 901 # => ended at: _raw_spin_unlock_irqrestore 902 # 903 # 904 # _------=> CPU# 905 # / _-----=> irqs-off 906 # | / _----=> need-resched 907 # || / _---=> hardirq/softirq 908 # ||| / _--=> preempt-depth 909 # |||| / delay 910 # cmd pid ||||| time | caller 911 # \ / ||||| \ | / 912 ps-6143 2d... 0us!: trace_hardirqs_off <-__lock_task_sighand 913 ps-6143 2d..1 259us+: trace_hardirqs_on <-_raw_spin_unlock_irqrestore 914 ps-6143 2d..1 263us+: time_hardirqs_on <-_raw_spin_unlock_irqrestore 915 ps-6143 2d..1 306us : <stack trace> 916 => trace_hardirqs_on_caller 917 => trace_hardirqs_on 918 => _raw_spin_unlock_irqrestore 919 => do_task_stat 920 => proc_tgid_stat 921 => proc_single_show 922 => seq_read 923 => vfs_read 924 => sys_read 925 => system_call_fastpath 926 927 928This shows that the current tracer is "irqsoff" tracing the time 929for which interrupts were disabled. It gives the trace version (which 930never changes) and the version of the kernel upon which this was executed on 931(3.8). Then it displays the max latency in microseconds (259 us). The number 932of trace entries displayed and the total number (both are four: #4/4). 933VP, KP, SP, and HP are always zero and are reserved for later use. 934#P is the number of online CPUs (#P:4). 935 936The task is the process that was running when the latency 937occurred. (ps pid: 6143). 938 939The start and stop (the functions in which the interrupts were 940disabled and enabled respectively) that caused the latencies: 941 942 - __lock_task_sighand is where the interrupts were disabled. 943 - _raw_spin_unlock_irqrestore is where they were enabled again. 944 945The next lines after the header are the trace itself. The header 946explains which is which. 947 948 cmd: The name of the process in the trace. 949 950 pid: The PID of that process. 951 952 CPU#: The CPU which the process was running on. 953 954 irqs-off: 'd' interrupts are disabled. '.' otherwise. 955 .. caution:: If the architecture does not support a way to 956 read the irq flags variable, an 'X' will always 957 be printed here. 958 959 need-resched: 960 - 'N' both TIF_NEED_RESCHED and PREEMPT_NEED_RESCHED is set, 961 - 'n' only TIF_NEED_RESCHED is set, 962 - 'p' only PREEMPT_NEED_RESCHED is set, 963 - '.' otherwise. 964 965 hardirq/softirq: 966 - 'Z' - NMI occurred inside a hardirq 967 - 'z' - NMI is running 968 - 'H' - hard irq occurred inside a softirq. 969 - 'h' - hard irq is running 970 - 's' - soft irq is running 971 - '.' - normal context. 972 973 preempt-depth: The level of preempt_disabled 974 975The above is mostly meaningful for kernel developers. 976 977 time: 978 When the latency-format option is enabled, the trace file 979 output includes a timestamp relative to the start of the 980 trace. This differs from the output when latency-format 981 is disabled, which includes an absolute timestamp. 982 983 delay: 984 This is just to help catch your eye a bit better. And 985 needs to be fixed to be only relative to the same CPU. 986 The marks are determined by the difference between this 987 current trace and the next trace. 988 989 - '$' - greater than 1 second 990 - '@' - greater than 100 millisecond 991 - '*' - greater than 10 millisecond 992 - '#' - greater than 1000 microsecond 993 - '!' - greater than 100 microsecond 994 - '+' - greater than 10 microsecond 995 - ' ' - less than or equal to 10 microsecond. 996 997 The rest is the same as the 'trace' file. 998 999 Note, the latency tracers will usually end with a back trace 1000 to easily find where the latency occurred. 1001 1002trace_options 1003------------- 1004 1005The trace_options file (or the options directory) is used to control 1006what gets printed in the trace output, or manipulate the tracers. 1007To see what is available, simply cat the file:: 1008 1009 cat trace_options 1010 print-parent 1011 nosym-offset 1012 nosym-addr 1013 noverbose 1014 noraw 1015 nohex 1016 nobin 1017 noblock 1018 trace_printk 1019 annotate 1020 nouserstacktrace 1021 nosym-userobj 1022 noprintk-msg-only 1023 context-info 1024 nolatency-format 1025 record-cmd 1026 norecord-tgid 1027 overwrite 1028 nodisable_on_free 1029 irq-info 1030 markers 1031 noevent-fork 1032 function-trace 1033 nofunction-fork 1034 nodisplay-graph 1035 nostacktrace 1036 nobranch 1037 1038To disable one of the options, echo in the option prepended with 1039"no":: 1040 1041 echo noprint-parent > trace_options 1042 1043To enable an option, leave off the "no":: 1044 1045 echo sym-offset > trace_options 1046 1047Here are the available options: 1048 1049 print-parent 1050 On function traces, display the calling (parent) 1051 function as well as the function being traced. 1052 :: 1053 1054 print-parent: 1055 bash-4000 [01] 1477.606694: simple_strtoul <-kstrtoul 1056 1057 noprint-parent: 1058 bash-4000 [01] 1477.606694: simple_strtoul 1059 1060 1061 sym-offset 1062 Display not only the function name, but also the 1063 offset in the function. For example, instead of 1064 seeing just "ktime_get", you will see 1065 "ktime_get+0xb/0x20". 1066 :: 1067 1068 sym-offset: 1069 bash-4000 [01] 1477.606694: simple_strtoul+0x6/0xa0 1070 1071 sym-addr 1072 This will also display the function address as well 1073 as the function name. 1074 :: 1075 1076 sym-addr: 1077 bash-4000 [01] 1477.606694: simple_strtoul <c0339346> 1078 1079 verbose 1080 This deals with the trace file when the 1081 latency-format option is enabled. 1082 :: 1083 1084 bash 4000 1 0 00000000 00010a95 [58127d26] 1720.415ms \ 1085 (+0.000ms): simple_strtoul (kstrtoul) 1086 1087 raw 1088 This will display raw numbers. This option is best for 1089 use with user applications that can translate the raw 1090 numbers better than having it done in the kernel. 1091 1092 hex 1093 Similar to raw, but the numbers will be in a hexadecimal format. 1094 1095 bin 1096 This will print out the formats in raw binary. 1097 1098 block 1099 When set, reading trace_pipe will not block when polled. 1100 1101 trace_printk 1102 Can disable trace_printk() from writing into the buffer. 1103 1104 annotate 1105 It is sometimes confusing when the CPU buffers are full 1106 and one CPU buffer had a lot of events recently, thus 1107 a shorter time frame, were another CPU may have only had 1108 a few events, which lets it have older events. When 1109 the trace is reported, it shows the oldest events first, 1110 and it may look like only one CPU ran (the one with the 1111 oldest events). When the annotate option is set, it will 1112 display when a new CPU buffer started:: 1113 1114 <idle>-0 [001] dNs4 21169.031481: wake_up_idle_cpu <-add_timer_on 1115 <idle>-0 [001] dNs4 21169.031482: _raw_spin_unlock_irqrestore <-add_timer_on 1116 <idle>-0 [001] .Ns4 21169.031484: sub_preempt_count <-_raw_spin_unlock_irqrestore 1117 ##### CPU 2 buffer started #### 1118 <idle>-0 [002] .N.1 21169.031484: rcu_idle_exit <-cpu_idle 1119 <idle>-0 [001] .Ns3 21169.031484: _raw_spin_unlock <-clocksource_watchdog 1120 <idle>-0 [001] .Ns3 21169.031485: sub_preempt_count <-_raw_spin_unlock 1121 1122 userstacktrace 1123 This option changes the trace. It records a 1124 stacktrace of the current user space thread after 1125 each trace event. 1126 1127 sym-userobj 1128 when user stacktrace are enabled, look up which 1129 object the address belongs to, and print a 1130 relative address. This is especially useful when 1131 ASLR is on, otherwise you don't get a chance to 1132 resolve the address to object/file/line after 1133 the app is no longer running 1134 1135 The lookup is performed when you read 1136 trace,trace_pipe. Example:: 1137 1138 a.out-1623 [000] 40874.465068: /root/a.out[+0x480] <-/root/a.out[+0 1139 x494] <- /root/a.out[+0x4a8] <- /lib/libc-2.7.so[+0x1e1a6] 1140 1141 1142 printk-msg-only 1143 When set, trace_printk()s will only show the format 1144 and not their parameters (if trace_bprintk() or 1145 trace_bputs() was used to save the trace_printk()). 1146 1147 context-info 1148 Show only the event data. Hides the comm, PID, 1149 timestamp, CPU, and other useful data. 1150 1151 latency-format 1152 This option changes the trace output. When it is enabled, 1153 the trace displays additional information about the 1154 latency, as described in "Latency trace format". 1155 1156 pause-on-trace 1157 When set, opening the trace file for read, will pause 1158 writing to the ring buffer (as if tracing_on was set to zero). 1159 This simulates the original behavior of the trace file. 1160 When the file is closed, tracing will be enabled again. 1161 1162 record-cmd 1163 When any event or tracer is enabled, a hook is enabled 1164 in the sched_switch trace point to fill comm cache 1165 with mapped pids and comms. But this may cause some 1166 overhead, and if you only care about pids, and not the 1167 name of the task, disabling this option can lower the 1168 impact of tracing. See "saved_cmdlines". 1169 1170 record-tgid 1171 When any event or tracer is enabled, a hook is enabled 1172 in the sched_switch trace point to fill the cache of 1173 mapped Thread Group IDs (TGID) mapping to pids. See 1174 "saved_tgids". 1175 1176 overwrite 1177 This controls what happens when the trace buffer is 1178 full. If "1" (default), the oldest events are 1179 discarded and overwritten. If "0", then the newest 1180 events are discarded. 1181 (see per_cpu/cpu0/stats for overrun and dropped) 1182 1183 disable_on_free 1184 When the free_buffer is closed, tracing will 1185 stop (tracing_on set to 0). 1186 1187 irq-info 1188 Shows the interrupt, preempt count, need resched data. 1189 When disabled, the trace looks like:: 1190 1191 # tracer: function 1192 # 1193 # entries-in-buffer/entries-written: 144405/9452052 #P:4 1194 # 1195 # TASK-PID CPU# TIMESTAMP FUNCTION 1196 # | | | | | 1197 <idle>-0 [002] 23636.756054: ttwu_do_activate.constprop.89 <-try_to_wake_up 1198 <idle>-0 [002] 23636.756054: activate_task <-ttwu_do_activate.constprop.89 1199 <idle>-0 [002] 23636.756055: enqueue_task <-activate_task 1200 1201 1202 markers 1203 When set, the trace_marker is writable (only by root). 1204 When disabled, the trace_marker will error with EINVAL 1205 on write. 1206 1207 event-fork 1208 When set, tasks with PIDs listed in set_event_pid will have 1209 the PIDs of their children added to set_event_pid when those 1210 tasks fork. Also, when tasks with PIDs in set_event_pid exit, 1211 their PIDs will be removed from the file. 1212 1213 This affects PIDs listed in set_event_notrace_pid as well. 1214 1215 function-trace 1216 The latency tracers will enable function tracing 1217 if this option is enabled (default it is). When 1218 it is disabled, the latency tracers do not trace 1219 functions. This keeps the overhead of the tracer down 1220 when performing latency tests. 1221 1222 function-fork 1223 When set, tasks with PIDs listed in set_ftrace_pid will 1224 have the PIDs of their children added to set_ftrace_pid 1225 when those tasks fork. Also, when tasks with PIDs in 1226 set_ftrace_pid exit, their PIDs will be removed from the 1227 file. 1228 1229 This affects PIDs in set_ftrace_notrace_pid as well. 1230 1231 display-graph 1232 When set, the latency tracers (irqsoff, wakeup, etc) will 1233 use function graph tracing instead of function tracing. 1234 1235 stacktrace 1236 When set, a stack trace is recorded after any trace event 1237 is recorded. 1238 1239 branch 1240 Enable branch tracing with the tracer. This enables branch 1241 tracer along with the currently set tracer. Enabling this 1242 with the "nop" tracer is the same as just enabling the 1243 "branch" tracer. 1244 1245.. tip:: Some tracers have their own options. They only appear in this 1246 file when the tracer is active. They always appear in the 1247 options directory. 1248 1249 1250Here are the per tracer options: 1251 1252Options for function tracer: 1253 1254 func_stack_trace 1255 When set, a stack trace is recorded after every 1256 function that is recorded. NOTE! Limit the functions 1257 that are recorded before enabling this, with 1258 "set_ftrace_filter" otherwise the system performance 1259 will be critically degraded. Remember to disable 1260 this option before clearing the function filter. 1261 1262Options for function_graph tracer: 1263 1264 Since the function_graph tracer has a slightly different output 1265 it has its own options to control what is displayed. 1266 1267 funcgraph-overrun 1268 When set, the "overrun" of the graph stack is 1269 displayed after each function traced. The 1270 overrun, is when the stack depth of the calls 1271 is greater than what is reserved for each task. 1272 Each task has a fixed array of functions to 1273 trace in the call graph. If the depth of the 1274 calls exceeds that, the function is not traced. 1275 The overrun is the number of functions missed 1276 due to exceeding this array. 1277 1278 funcgraph-cpu 1279 When set, the CPU number of the CPU where the trace 1280 occurred is displayed. 1281 1282 funcgraph-overhead 1283 When set, if the function takes longer than 1284 A certain amount, then a delay marker is 1285 displayed. See "delay" above, under the 1286 header description. 1287 1288 funcgraph-proc 1289 Unlike other tracers, the process' command line 1290 is not displayed by default, but instead only 1291 when a task is traced in and out during a context 1292 switch. Enabling this options has the command 1293 of each process displayed at every line. 1294 1295 funcgraph-duration 1296 At the end of each function (the return) 1297 the duration of the amount of time in the 1298 function is displayed in microseconds. 1299 1300 funcgraph-abstime 1301 When set, the timestamp is displayed at each line. 1302 1303 funcgraph-irqs 1304 When disabled, functions that happen inside an 1305 interrupt will not be traced. 1306 1307 funcgraph-tail 1308 When set, the return event will include the function 1309 that it represents. By default this is off, and 1310 only a closing curly bracket "}" is displayed for 1311 the return of a function. 1312 1313 sleep-time 1314 When running function graph tracer, to include 1315 the time a task schedules out in its function. 1316 When enabled, it will account time the task has been 1317 scheduled out as part of the function call. 1318 1319 graph-time 1320 When running function profiler with function graph tracer, 1321 to include the time to call nested functions. When this is 1322 not set, the time reported for the function will only 1323 include the time the function itself executed for, not the 1324 time for functions that it called. 1325 1326Options for blk tracer: 1327 1328 blk_classic 1329 Shows a more minimalistic output. 1330 1331 1332irqsoff 1333------- 1334 1335When interrupts are disabled, the CPU can not react to any other 1336external event (besides NMIs and SMIs). This prevents the timer 1337interrupt from triggering or the mouse interrupt from letting 1338the kernel know of a new mouse event. The result is a latency 1339with the reaction time. 1340 1341The irqsoff tracer tracks the time for which interrupts are 1342disabled. When a new maximum latency is hit, the tracer saves 1343the trace leading up to that latency point so that every time a 1344new maximum is reached, the old saved trace is discarded and the 1345new trace is saved. 1346 1347To reset the maximum, echo 0 into tracing_max_latency. Here is 1348an example:: 1349 1350 # echo 0 > options/function-trace 1351 # echo irqsoff > current_tracer 1352 # echo 1 > tracing_on 1353 # echo 0 > tracing_max_latency 1354 # ls -ltr 1355 [...] 1356 # echo 0 > tracing_on 1357 # cat trace 1358 # tracer: irqsoff 1359 # 1360 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 1361 # -------------------------------------------------------------------- 1362 # latency: 16 us, #4/4, CPU#0 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1363 # ----------------- 1364 # | task: swapper/0-0 (uid:0 nice:0 policy:0 rt_prio:0) 1365 # ----------------- 1366 # => started at: run_timer_softirq 1367 # => ended at: run_timer_softirq 1368 # 1369 # 1370 # _------=> CPU# 1371 # / _-----=> irqs-off 1372 # | / _----=> need-resched 1373 # || / _---=> hardirq/softirq 1374 # ||| / _--=> preempt-depth 1375 # |||| / delay 1376 # cmd pid ||||| time | caller 1377 # \ / ||||| \ | / 1378 <idle>-0 0d.s2 0us+: _raw_spin_lock_irq <-run_timer_softirq 1379 <idle>-0 0dNs3 17us : _raw_spin_unlock_irq <-run_timer_softirq 1380 <idle>-0 0dNs3 17us+: trace_hardirqs_on <-run_timer_softirq 1381 <idle>-0 0dNs3 25us : <stack trace> 1382 => _raw_spin_unlock_irq 1383 => run_timer_softirq 1384 => __do_softirq 1385 => call_softirq 1386 => do_softirq 1387 => irq_exit 1388 => smp_apic_timer_interrupt 1389 => apic_timer_interrupt 1390 => rcu_idle_exit 1391 => cpu_idle 1392 => rest_init 1393 => start_kernel 1394 => x86_64_start_reservations 1395 => x86_64_start_kernel 1396 1397Here we see that that we had a latency of 16 microseconds (which is 1398very good). The _raw_spin_lock_irq in run_timer_softirq disabled 1399interrupts. The difference between the 16 and the displayed 1400timestamp 25us occurred because the clock was incremented 1401between the time of recording the max latency and the time of 1402recording the function that had that latency. 1403 1404Note the above example had function-trace not set. If we set 1405function-trace, we get a much larger output:: 1406 1407 with echo 1 > options/function-trace 1408 1409 # tracer: irqsoff 1410 # 1411 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 1412 # -------------------------------------------------------------------- 1413 # latency: 71 us, #168/168, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1414 # ----------------- 1415 # | task: bash-2042 (uid:0 nice:0 policy:0 rt_prio:0) 1416 # ----------------- 1417 # => started at: ata_scsi_queuecmd 1418 # => ended at: ata_scsi_queuecmd 1419 # 1420 # 1421 # _------=> CPU# 1422 # / _-----=> irqs-off 1423 # | / _----=> need-resched 1424 # || / _---=> hardirq/softirq 1425 # ||| / _--=> preempt-depth 1426 # |||| / delay 1427 # cmd pid ||||| time | caller 1428 # \ / ||||| \ | / 1429 bash-2042 3d... 0us : _raw_spin_lock_irqsave <-ata_scsi_queuecmd 1430 bash-2042 3d... 0us : add_preempt_count <-_raw_spin_lock_irqsave 1431 bash-2042 3d..1 1us : ata_scsi_find_dev <-ata_scsi_queuecmd 1432 bash-2042 3d..1 1us : __ata_scsi_find_dev <-ata_scsi_find_dev 1433 bash-2042 3d..1 2us : ata_find_dev.part.14 <-__ata_scsi_find_dev 1434 bash-2042 3d..1 2us : ata_qc_new_init <-__ata_scsi_queuecmd 1435 bash-2042 3d..1 3us : ata_sg_init <-__ata_scsi_queuecmd 1436 bash-2042 3d..1 4us : ata_scsi_rw_xlat <-__ata_scsi_queuecmd 1437 bash-2042 3d..1 4us : ata_build_rw_tf <-ata_scsi_rw_xlat 1438 [...] 1439 bash-2042 3d..1 67us : delay_tsc <-__delay 1440 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc 1441 bash-2042 3d..2 67us : sub_preempt_count <-delay_tsc 1442 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc 1443 bash-2042 3d..2 68us : sub_preempt_count <-delay_tsc 1444 bash-2042 3d..1 68us+: ata_bmdma_start <-ata_bmdma_qc_issue 1445 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1446 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1447 bash-2042 3d..1 72us+: trace_hardirqs_on <-ata_scsi_queuecmd 1448 bash-2042 3d..1 120us : <stack trace> 1449 => _raw_spin_unlock_irqrestore 1450 => ata_scsi_queuecmd 1451 => scsi_dispatch_cmd 1452 => scsi_request_fn 1453 => __blk_run_queue_uncond 1454 => __blk_run_queue 1455 => blk_queue_bio 1456 => generic_make_request 1457 => submit_bio 1458 => submit_bh 1459 => __ext3_get_inode_loc 1460 => ext3_iget 1461 => ext3_lookup 1462 => lookup_real 1463 => __lookup_hash 1464 => walk_component 1465 => lookup_last 1466 => path_lookupat 1467 => filename_lookup 1468 => user_path_at_empty 1469 => user_path_at 1470 => vfs_fstatat 1471 => vfs_stat 1472 => sys_newstat 1473 => system_call_fastpath 1474 1475 1476Here we traced a 71 microsecond latency. But we also see all the 1477functions that were called during that time. Note that by 1478enabling function tracing, we incur an added overhead. This 1479overhead may extend the latency times. But nevertheless, this 1480trace has provided some very helpful debugging information. 1481 1482If we prefer function graph output instead of function, we can set 1483display-graph option:: 1484 1485 with echo 1 > options/display-graph 1486 1487 # tracer: irqsoff 1488 # 1489 # irqsoff latency trace v1.1.5 on 4.20.0-rc6+ 1490 # -------------------------------------------------------------------- 1491 # latency: 3751 us, #274/274, CPU#0 | (M:desktop VP:0, KP:0, SP:0 HP:0 #P:4) 1492 # ----------------- 1493 # | task: bash-1507 (uid:0 nice:0 policy:0 rt_prio:0) 1494 # ----------------- 1495 # => started at: free_debug_processing 1496 # => ended at: return_to_handler 1497 # 1498 # 1499 # _-----=> irqs-off 1500 # / _----=> need-resched 1501 # | / _---=> hardirq/softirq 1502 # || / _--=> preempt-depth 1503 # ||| / 1504 # REL TIME CPU TASK/PID |||| DURATION FUNCTION CALLS 1505 # | | | | |||| | | | | | | 1506 0 us | 0) bash-1507 | d... | 0.000 us | _raw_spin_lock_irqsave(); 1507 0 us | 0) bash-1507 | d..1 | 0.378 us | do_raw_spin_trylock(); 1508 1 us | 0) bash-1507 | d..2 | | set_track() { 1509 2 us | 0) bash-1507 | d..2 | | save_stack_trace() { 1510 2 us | 0) bash-1507 | d..2 | | __save_stack_trace() { 1511 3 us | 0) bash-1507 | d..2 | | __unwind_start() { 1512 3 us | 0) bash-1507 | d..2 | | get_stack_info() { 1513 3 us | 0) bash-1507 | d..2 | 0.351 us | in_task_stack(); 1514 4 us | 0) bash-1507 | d..2 | 1.107 us | } 1515 [...] 1516 3750 us | 0) bash-1507 | d..1 | 0.516 us | do_raw_spin_unlock(); 1517 3750 us | 0) bash-1507 | d..1 | 0.000 us | _raw_spin_unlock_irqrestore(); 1518 3764 us | 0) bash-1507 | d..1 | 0.000 us | tracer_hardirqs_on(); 1519 bash-1507 0d..1 3792us : <stack trace> 1520 => free_debug_processing 1521 => __slab_free 1522 => kmem_cache_free 1523 => vm_area_free 1524 => remove_vma 1525 => exit_mmap 1526 => mmput 1527 => begin_new_exec 1528 => load_elf_binary 1529 => search_binary_handler 1530 => __do_execve_file.isra.32 1531 => __x64_sys_execve 1532 => do_syscall_64 1533 => entry_SYSCALL_64_after_hwframe 1534 1535preemptoff 1536---------- 1537 1538When preemption is disabled, we may be able to receive 1539interrupts but the task cannot be preempted and a higher 1540priority task must wait for preemption to be enabled again 1541before it can preempt a lower priority task. 1542 1543The preemptoff tracer traces the places that disable preemption. 1544Like the irqsoff tracer, it records the maximum latency for 1545which preemption was disabled. The control of preemptoff tracer 1546is much like the irqsoff tracer. 1547:: 1548 1549 # echo 0 > options/function-trace 1550 # echo preemptoff > current_tracer 1551 # echo 1 > tracing_on 1552 # echo 0 > tracing_max_latency 1553 # ls -ltr 1554 [...] 1555 # echo 0 > tracing_on 1556 # cat trace 1557 # tracer: preemptoff 1558 # 1559 # preemptoff latency trace v1.1.5 on 3.8.0-test+ 1560 # -------------------------------------------------------------------- 1561 # latency: 46 us, #4/4, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1562 # ----------------- 1563 # | task: sshd-1991 (uid:0 nice:0 policy:0 rt_prio:0) 1564 # ----------------- 1565 # => started at: do_IRQ 1566 # => ended at: do_IRQ 1567 # 1568 # 1569 # _------=> CPU# 1570 # / _-----=> irqs-off 1571 # | / _----=> need-resched 1572 # || / _---=> hardirq/softirq 1573 # ||| / _--=> preempt-depth 1574 # |||| / delay 1575 # cmd pid ||||| time | caller 1576 # \ / ||||| \ | / 1577 sshd-1991 1d.h. 0us+: irq_enter <-do_IRQ 1578 sshd-1991 1d..1 46us : irq_exit <-do_IRQ 1579 sshd-1991 1d..1 47us+: trace_preempt_on <-do_IRQ 1580 sshd-1991 1d..1 52us : <stack trace> 1581 => sub_preempt_count 1582 => irq_exit 1583 => do_IRQ 1584 => ret_from_intr 1585 1586 1587This has some more changes. Preemption was disabled when an 1588interrupt came in (notice the 'h'), and was enabled on exit. 1589But we also see that interrupts have been disabled when entering 1590the preempt off section and leaving it (the 'd'). We do not know if 1591interrupts were enabled in the mean time or shortly after this 1592was over. 1593:: 1594 1595 # tracer: preemptoff 1596 # 1597 # preemptoff latency trace v1.1.5 on 3.8.0-test+ 1598 # -------------------------------------------------------------------- 1599 # latency: 83 us, #241/241, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1600 # ----------------- 1601 # | task: bash-1994 (uid:0 nice:0 policy:0 rt_prio:0) 1602 # ----------------- 1603 # => started at: wake_up_new_task 1604 # => ended at: task_rq_unlock 1605 # 1606 # 1607 # _------=> CPU# 1608 # / _-----=> irqs-off 1609 # | / _----=> need-resched 1610 # || / _---=> hardirq/softirq 1611 # ||| / _--=> preempt-depth 1612 # |||| / delay 1613 # cmd pid ||||| time | caller 1614 # \ / ||||| \ | / 1615 bash-1994 1d..1 0us : _raw_spin_lock_irqsave <-wake_up_new_task 1616 bash-1994 1d..1 0us : select_task_rq_fair <-select_task_rq 1617 bash-1994 1d..1 1us : __rcu_read_lock <-select_task_rq_fair 1618 bash-1994 1d..1 1us : source_load <-select_task_rq_fair 1619 bash-1994 1d..1 1us : source_load <-select_task_rq_fair 1620 [...] 1621 bash-1994 1d..1 12us : irq_enter <-smp_apic_timer_interrupt 1622 bash-1994 1d..1 12us : rcu_irq_enter <-irq_enter 1623 bash-1994 1d..1 13us : add_preempt_count <-irq_enter 1624 bash-1994 1d.h1 13us : exit_idle <-smp_apic_timer_interrupt 1625 bash-1994 1d.h1 13us : hrtimer_interrupt <-smp_apic_timer_interrupt 1626 bash-1994 1d.h1 13us : _raw_spin_lock <-hrtimer_interrupt 1627 bash-1994 1d.h1 14us : add_preempt_count <-_raw_spin_lock 1628 bash-1994 1d.h2 14us : ktime_get_update_offsets <-hrtimer_interrupt 1629 [...] 1630 bash-1994 1d.h1 35us : lapic_next_event <-clockevents_program_event 1631 bash-1994 1d.h1 35us : irq_exit <-smp_apic_timer_interrupt 1632 bash-1994 1d.h1 36us : sub_preempt_count <-irq_exit 1633 bash-1994 1d..2 36us : do_softirq <-irq_exit 1634 bash-1994 1d..2 36us : __do_softirq <-call_softirq 1635 bash-1994 1d..2 36us : __local_bh_disable <-__do_softirq 1636 bash-1994 1d.s2 37us : add_preempt_count <-_raw_spin_lock_irq 1637 bash-1994 1d.s3 38us : _raw_spin_unlock <-run_timer_softirq 1638 bash-1994 1d.s3 39us : sub_preempt_count <-_raw_spin_unlock 1639 bash-1994 1d.s2 39us : call_timer_fn <-run_timer_softirq 1640 [...] 1641 bash-1994 1dNs2 81us : cpu_needs_another_gp <-rcu_process_callbacks 1642 bash-1994 1dNs2 82us : __local_bh_enable <-__do_softirq 1643 bash-1994 1dNs2 82us : sub_preempt_count <-__local_bh_enable 1644 bash-1994 1dN.2 82us : idle_cpu <-irq_exit 1645 bash-1994 1dN.2 83us : rcu_irq_exit <-irq_exit 1646 bash-1994 1dN.2 83us : sub_preempt_count <-irq_exit 1647 bash-1994 1.N.1 84us : _raw_spin_unlock_irqrestore <-task_rq_unlock 1648 bash-1994 1.N.1 84us+: trace_preempt_on <-task_rq_unlock 1649 bash-1994 1.N.1 104us : <stack trace> 1650 => sub_preempt_count 1651 => _raw_spin_unlock_irqrestore 1652 => task_rq_unlock 1653 => wake_up_new_task 1654 => do_fork 1655 => sys_clone 1656 => stub_clone 1657 1658 1659The above is an example of the preemptoff trace with 1660function-trace set. Here we see that interrupts were not disabled 1661the entire time. The irq_enter code lets us know that we entered 1662an interrupt 'h'. Before that, the functions being traced still 1663show that it is not in an interrupt, but we can see from the 1664functions themselves that this is not the case. 1665 1666preemptirqsoff 1667-------------- 1668 1669Knowing the locations that have interrupts disabled or 1670preemption disabled for the longest times is helpful. But 1671sometimes we would like to know when either preemption and/or 1672interrupts are disabled. 1673 1674Consider the following code:: 1675 1676 local_irq_disable(); 1677 call_function_with_irqs_off(); 1678 preempt_disable(); 1679 call_function_with_irqs_and_preemption_off(); 1680 local_irq_enable(); 1681 call_function_with_preemption_off(); 1682 preempt_enable(); 1683 1684The irqsoff tracer will record the total length of 1685call_function_with_irqs_off() and 1686call_function_with_irqs_and_preemption_off(). 1687 1688The preemptoff tracer will record the total length of 1689call_function_with_irqs_and_preemption_off() and 1690call_function_with_preemption_off(). 1691 1692But neither will trace the time that interrupts and/or 1693preemption is disabled. This total time is the time that we can 1694not schedule. To record this time, use the preemptirqsoff 1695tracer. 1696 1697Again, using this trace is much like the irqsoff and preemptoff 1698tracers. 1699:: 1700 1701 # echo 0 > options/function-trace 1702 # echo preemptirqsoff > current_tracer 1703 # echo 1 > tracing_on 1704 # echo 0 > tracing_max_latency 1705 # ls -ltr 1706 [...] 1707 # echo 0 > tracing_on 1708 # cat trace 1709 # tracer: preemptirqsoff 1710 # 1711 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+ 1712 # -------------------------------------------------------------------- 1713 # latency: 100 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1714 # ----------------- 1715 # | task: ls-2230 (uid:0 nice:0 policy:0 rt_prio:0) 1716 # ----------------- 1717 # => started at: ata_scsi_queuecmd 1718 # => ended at: ata_scsi_queuecmd 1719 # 1720 # 1721 # _------=> CPU# 1722 # / _-----=> irqs-off 1723 # | / _----=> need-resched 1724 # || / _---=> hardirq/softirq 1725 # ||| / _--=> preempt-depth 1726 # |||| / delay 1727 # cmd pid ||||| time | caller 1728 # \ / ||||| \ | / 1729 ls-2230 3d... 0us+: _raw_spin_lock_irqsave <-ata_scsi_queuecmd 1730 ls-2230 3...1 100us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1731 ls-2230 3...1 101us+: trace_preempt_on <-ata_scsi_queuecmd 1732 ls-2230 3...1 111us : <stack trace> 1733 => sub_preempt_count 1734 => _raw_spin_unlock_irqrestore 1735 => ata_scsi_queuecmd 1736 => scsi_dispatch_cmd 1737 => scsi_request_fn 1738 => __blk_run_queue_uncond 1739 => __blk_run_queue 1740 => blk_queue_bio 1741 => generic_make_request 1742 => submit_bio 1743 => submit_bh 1744 => ext3_bread 1745 => ext3_dir_bread 1746 => htree_dirblock_to_tree 1747 => ext3_htree_fill_tree 1748 => ext3_readdir 1749 => vfs_readdir 1750 => sys_getdents 1751 => system_call_fastpath 1752 1753 1754The trace_hardirqs_off_thunk is called from assembly on x86 when 1755interrupts are disabled in the assembly code. Without the 1756function tracing, we do not know if interrupts were enabled 1757within the preemption points. We do see that it started with 1758preemption enabled. 1759 1760Here is a trace with function-trace set:: 1761 1762 # tracer: preemptirqsoff 1763 # 1764 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+ 1765 # -------------------------------------------------------------------- 1766 # latency: 161 us, #339/339, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1767 # ----------------- 1768 # | task: ls-2269 (uid:0 nice:0 policy:0 rt_prio:0) 1769 # ----------------- 1770 # => started at: schedule 1771 # => ended at: mutex_unlock 1772 # 1773 # 1774 # _------=> CPU# 1775 # / _-----=> irqs-off 1776 # | / _----=> need-resched 1777 # || / _---=> hardirq/softirq 1778 # ||| / _--=> preempt-depth 1779 # |||| / delay 1780 # cmd pid ||||| time | caller 1781 # \ / ||||| \ | / 1782 kworker/-59 3...1 0us : __schedule <-schedule 1783 kworker/-59 3d..1 0us : rcu_preempt_qs <-rcu_note_context_switch 1784 kworker/-59 3d..1 1us : add_preempt_count <-_raw_spin_lock_irq 1785 kworker/-59 3d..2 1us : deactivate_task <-__schedule 1786 kworker/-59 3d..2 1us : dequeue_task <-deactivate_task 1787 kworker/-59 3d..2 2us : update_rq_clock <-dequeue_task 1788 kworker/-59 3d..2 2us : dequeue_task_fair <-dequeue_task 1789 kworker/-59 3d..2 2us : update_curr <-dequeue_task_fair 1790 kworker/-59 3d..2 2us : update_min_vruntime <-update_curr 1791 kworker/-59 3d..2 3us : cpuacct_charge <-update_curr 1792 kworker/-59 3d..2 3us : __rcu_read_lock <-cpuacct_charge 1793 kworker/-59 3d..2 3us : __rcu_read_unlock <-cpuacct_charge 1794 kworker/-59 3d..2 3us : update_cfs_rq_blocked_load <-dequeue_task_fair 1795 kworker/-59 3d..2 4us : clear_buddies <-dequeue_task_fair 1796 kworker/-59 3d..2 4us : account_entity_dequeue <-dequeue_task_fair 1797 kworker/-59 3d..2 4us : update_min_vruntime <-dequeue_task_fair 1798 kworker/-59 3d..2 4us : update_cfs_shares <-dequeue_task_fair 1799 kworker/-59 3d..2 5us : hrtick_update <-dequeue_task_fair 1800 kworker/-59 3d..2 5us : wq_worker_sleeping <-__schedule 1801 kworker/-59 3d..2 5us : kthread_data <-wq_worker_sleeping 1802 kworker/-59 3d..2 5us : put_prev_task_fair <-__schedule 1803 kworker/-59 3d..2 6us : pick_next_task_fair <-pick_next_task 1804 kworker/-59 3d..2 6us : clear_buddies <-pick_next_task_fair 1805 kworker/-59 3d..2 6us : set_next_entity <-pick_next_task_fair 1806 kworker/-59 3d..2 6us : update_stats_wait_end <-set_next_entity 1807 ls-2269 3d..2 7us : finish_task_switch <-__schedule 1808 ls-2269 3d..2 7us : _raw_spin_unlock_irq <-finish_task_switch 1809 ls-2269 3d..2 8us : do_IRQ <-ret_from_intr 1810 ls-2269 3d..2 8us : irq_enter <-do_IRQ 1811 ls-2269 3d..2 8us : rcu_irq_enter <-irq_enter 1812 ls-2269 3d..2 9us : add_preempt_count <-irq_enter 1813 ls-2269 3d.h2 9us : exit_idle <-do_IRQ 1814 [...] 1815 ls-2269 3d.h3 20us : sub_preempt_count <-_raw_spin_unlock 1816 ls-2269 3d.h2 20us : irq_exit <-do_IRQ 1817 ls-2269 3d.h2 21us : sub_preempt_count <-irq_exit 1818 ls-2269 3d..3 21us : do_softirq <-irq_exit 1819 ls-2269 3d..3 21us : __do_softirq <-call_softirq 1820 ls-2269 3d..3 21us+: __local_bh_disable <-__do_softirq 1821 ls-2269 3d.s4 29us : sub_preempt_count <-_local_bh_enable_ip 1822 ls-2269 3d.s5 29us : sub_preempt_count <-_local_bh_enable_ip 1823 ls-2269 3d.s5 31us : do_IRQ <-ret_from_intr 1824 ls-2269 3d.s5 31us : irq_enter <-do_IRQ 1825 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter 1826 [...] 1827 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter 1828 ls-2269 3d.s5 32us : add_preempt_count <-irq_enter 1829 ls-2269 3d.H5 32us : exit_idle <-do_IRQ 1830 ls-2269 3d.H5 32us : handle_irq <-do_IRQ 1831 ls-2269 3d.H5 32us : irq_to_desc <-handle_irq 1832 ls-2269 3d.H5 33us : handle_fasteoi_irq <-handle_irq 1833 [...] 1834 ls-2269 3d.s5 158us : _raw_spin_unlock_irqrestore <-rtl8139_poll 1835 ls-2269 3d.s3 158us : net_rps_action_and_irq_enable.isra.65 <-net_rx_action 1836 ls-2269 3d.s3 159us : __local_bh_enable <-__do_softirq 1837 ls-2269 3d.s3 159us : sub_preempt_count <-__local_bh_enable 1838 ls-2269 3d..3 159us : idle_cpu <-irq_exit 1839 ls-2269 3d..3 159us : rcu_irq_exit <-irq_exit 1840 ls-2269 3d..3 160us : sub_preempt_count <-irq_exit 1841 ls-2269 3d... 161us : __mutex_unlock_slowpath <-mutex_unlock 1842 ls-2269 3d... 162us+: trace_hardirqs_on <-mutex_unlock 1843 ls-2269 3d... 186us : <stack trace> 1844 => __mutex_unlock_slowpath 1845 => mutex_unlock 1846 => process_output 1847 => n_tty_write 1848 => tty_write 1849 => vfs_write 1850 => sys_write 1851 => system_call_fastpath 1852 1853This is an interesting trace. It started with kworker running and 1854scheduling out and ls taking over. But as soon as ls released the 1855rq lock and enabled interrupts (but not preemption) an interrupt 1856triggered. When the interrupt finished, it started running softirqs. 1857But while the softirq was running, another interrupt triggered. 1858When an interrupt is running inside a softirq, the annotation is 'H'. 1859 1860 1861wakeup 1862------ 1863 1864One common case that people are interested in tracing is the 1865time it takes for a task that is woken to actually wake up. 1866Now for non Real-Time tasks, this can be arbitrary. But tracing 1867it none the less can be interesting. 1868 1869Without function tracing:: 1870 1871 # echo 0 > options/function-trace 1872 # echo wakeup > current_tracer 1873 # echo 1 > tracing_on 1874 # echo 0 > tracing_max_latency 1875 # chrt -f 5 sleep 1 1876 # echo 0 > tracing_on 1877 # cat trace 1878 # tracer: wakeup 1879 # 1880 # wakeup latency trace v1.1.5 on 3.8.0-test+ 1881 # -------------------------------------------------------------------- 1882 # latency: 15 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1883 # ----------------- 1884 # | task: kworker/3:1H-312 (uid:0 nice:-20 policy:0 rt_prio:0) 1885 # ----------------- 1886 # 1887 # _------=> CPU# 1888 # / _-----=> irqs-off 1889 # | / _----=> need-resched 1890 # || / _---=> hardirq/softirq 1891 # ||| / _--=> preempt-depth 1892 # |||| / delay 1893 # cmd pid ||||| time | caller 1894 # \ / ||||| \ | / 1895 <idle>-0 3dNs7 0us : 0:120:R + [003] 312:100:R kworker/3:1H 1896 <idle>-0 3dNs7 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up 1897 <idle>-0 3d..3 15us : __schedule <-schedule 1898 <idle>-0 3d..3 15us : 0:120:R ==> [003] 312:100:R kworker/3:1H 1899 1900The tracer only traces the highest priority task in the system 1901to avoid tracing the normal circumstances. Here we see that 1902the kworker with a nice priority of -20 (not very nice), took 1903just 15 microseconds from the time it woke up, to the time it 1904ran. 1905 1906Non Real-Time tasks are not that interesting. A more interesting 1907trace is to concentrate only on Real-Time tasks. 1908 1909wakeup_rt 1910--------- 1911 1912In a Real-Time environment it is very important to know the 1913wakeup time it takes for the highest priority task that is woken 1914up to the time that it executes. This is also known as "schedule 1915latency". I stress the point that this is about RT tasks. It is 1916also important to know the scheduling latency of non-RT tasks, 1917but the average schedule latency is better for non-RT tasks. 1918Tools like LatencyTop are more appropriate for such 1919measurements. 1920 1921Real-Time environments are interested in the worst case latency. 1922That is the longest latency it takes for something to happen, 1923and not the average. We can have a very fast scheduler that may 1924only have a large latency once in a while, but that would not 1925work well with Real-Time tasks. The wakeup_rt tracer was designed 1926to record the worst case wakeups of RT tasks. Non-RT tasks are 1927not recorded because the tracer only records one worst case and 1928tracing non-RT tasks that are unpredictable will overwrite the 1929worst case latency of RT tasks (just run the normal wakeup 1930tracer for a while to see that effect). 1931 1932Since this tracer only deals with RT tasks, we will run this 1933slightly differently than we did with the previous tracers. 1934Instead of performing an 'ls', we will run 'sleep 1' under 1935'chrt' which changes the priority of the task. 1936:: 1937 1938 # echo 0 > options/function-trace 1939 # echo wakeup_rt > current_tracer 1940 # echo 1 > tracing_on 1941 # echo 0 > tracing_max_latency 1942 # chrt -f 5 sleep 1 1943 # echo 0 > tracing_on 1944 # cat trace 1945 # tracer: wakeup 1946 # 1947 # tracer: wakeup_rt 1948 # 1949 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 1950 # -------------------------------------------------------------------- 1951 # latency: 5 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1952 # ----------------- 1953 # | task: sleep-2389 (uid:0 nice:0 policy:1 rt_prio:5) 1954 # ----------------- 1955 # 1956 # _------=> CPU# 1957 # / _-----=> irqs-off 1958 # | / _----=> need-resched 1959 # || / _---=> hardirq/softirq 1960 # ||| / _--=> preempt-depth 1961 # |||| / delay 1962 # cmd pid ||||| time | caller 1963 # \ / ||||| \ | / 1964 <idle>-0 3d.h4 0us : 0:120:R + [003] 2389: 94:R sleep 1965 <idle>-0 3d.h4 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up 1966 <idle>-0 3d..3 5us : __schedule <-schedule 1967 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep 1968 1969 1970Running this on an idle system, we see that it only took 5 microseconds 1971to perform the task switch. Note, since the trace point in the schedule 1972is before the actual "switch", we stop the tracing when the recorded task 1973is about to schedule in. This may change if we add a new marker at the 1974end of the scheduler. 1975 1976Notice that the recorded task is 'sleep' with the PID of 2389 1977and it has an rt_prio of 5. This priority is user-space priority 1978and not the internal kernel priority. The policy is 1 for 1979SCHED_FIFO and 2 for SCHED_RR. 1980 1981Note, that the trace data shows the internal priority (99 - rtprio). 1982:: 1983 1984 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep 1985 1986The 0:120:R means idle was running with a nice priority of 0 (120 - 120) 1987and in the running state 'R'. The sleep task was scheduled in with 19882389: 94:R. That is the priority is the kernel rtprio (99 - 5 = 94) 1989and it too is in the running state. 1990 1991Doing the same with chrt -r 5 and function-trace set. 1992:: 1993 1994 echo 1 > options/function-trace 1995 1996 # tracer: wakeup_rt 1997 # 1998 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 1999 # -------------------------------------------------------------------- 2000 # latency: 29 us, #85/85, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2001 # ----------------- 2002 # | task: sleep-2448 (uid:0 nice:0 policy:1 rt_prio:5) 2003 # ----------------- 2004 # 2005 # _------=> CPU# 2006 # / _-----=> irqs-off 2007 # | / _----=> need-resched 2008 # || / _---=> hardirq/softirq 2009 # ||| / _--=> preempt-depth 2010 # |||| / delay 2011 # cmd pid ||||| time | caller 2012 # \ / ||||| \ | / 2013 <idle>-0 3d.h4 1us+: 0:120:R + [003] 2448: 94:R sleep 2014 <idle>-0 3d.h4 2us : ttwu_do_activate.constprop.87 <-try_to_wake_up 2015 <idle>-0 3d.h3 3us : check_preempt_curr <-ttwu_do_wakeup 2016 <idle>-0 3d.h3 3us : resched_curr <-check_preempt_curr 2017 <idle>-0 3dNh3 4us : task_woken_rt <-ttwu_do_wakeup 2018 <idle>-0 3dNh3 4us : _raw_spin_unlock <-try_to_wake_up 2019 <idle>-0 3dNh3 4us : sub_preempt_count <-_raw_spin_unlock 2020 <idle>-0 3dNh2 5us : ttwu_stat <-try_to_wake_up 2021 <idle>-0 3dNh2 5us : _raw_spin_unlock_irqrestore <-try_to_wake_up 2022 <idle>-0 3dNh2 6us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2023 <idle>-0 3dNh1 6us : _raw_spin_lock <-__run_hrtimer 2024 <idle>-0 3dNh1 6us : add_preempt_count <-_raw_spin_lock 2025 <idle>-0 3dNh2 7us : _raw_spin_unlock <-hrtimer_interrupt 2026 <idle>-0 3dNh2 7us : sub_preempt_count <-_raw_spin_unlock 2027 <idle>-0 3dNh1 7us : tick_program_event <-hrtimer_interrupt 2028 <idle>-0 3dNh1 7us : clockevents_program_event <-tick_program_event 2029 <idle>-0 3dNh1 8us : ktime_get <-clockevents_program_event 2030 <idle>-0 3dNh1 8us : lapic_next_event <-clockevents_program_event 2031 <idle>-0 3dNh1 8us : irq_exit <-smp_apic_timer_interrupt 2032 <idle>-0 3dNh1 9us : sub_preempt_count <-irq_exit 2033 <idle>-0 3dN.2 9us : idle_cpu <-irq_exit 2034 <idle>-0 3dN.2 9us : rcu_irq_exit <-irq_exit 2035 <idle>-0 3dN.2 10us : rcu_eqs_enter_common.isra.45 <-rcu_irq_exit 2036 <idle>-0 3dN.2 10us : sub_preempt_count <-irq_exit 2037 <idle>-0 3.N.1 11us : rcu_idle_exit <-cpu_idle 2038 <idle>-0 3dN.1 11us : rcu_eqs_exit_common.isra.43 <-rcu_idle_exit 2039 <idle>-0 3.N.1 11us : tick_nohz_idle_exit <-cpu_idle 2040 <idle>-0 3dN.1 12us : menu_hrtimer_cancel <-tick_nohz_idle_exit 2041 <idle>-0 3dN.1 12us : ktime_get <-tick_nohz_idle_exit 2042 <idle>-0 3dN.1 12us : tick_do_update_jiffies64 <-tick_nohz_idle_exit 2043 <idle>-0 3dN.1 13us : cpu_load_update_nohz <-tick_nohz_idle_exit 2044 <idle>-0 3dN.1 13us : _raw_spin_lock <-cpu_load_update_nohz 2045 <idle>-0 3dN.1 13us : add_preempt_count <-_raw_spin_lock 2046 <idle>-0 3dN.2 13us : __cpu_load_update <-cpu_load_update_nohz 2047 <idle>-0 3dN.2 14us : sched_avg_update <-__cpu_load_update 2048 <idle>-0 3dN.2 14us : _raw_spin_unlock <-cpu_load_update_nohz 2049 <idle>-0 3dN.2 14us : sub_preempt_count <-_raw_spin_unlock 2050 <idle>-0 3dN.1 15us : calc_load_nohz_stop <-tick_nohz_idle_exit 2051 <idle>-0 3dN.1 15us : touch_softlockup_watchdog <-tick_nohz_idle_exit 2052 <idle>-0 3dN.1 15us : hrtimer_cancel <-tick_nohz_idle_exit 2053 <idle>-0 3dN.1 15us : hrtimer_try_to_cancel <-hrtimer_cancel 2054 <idle>-0 3dN.1 16us : lock_hrtimer_base.isra.18 <-hrtimer_try_to_cancel 2055 <idle>-0 3dN.1 16us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18 2056 <idle>-0 3dN.1 16us : add_preempt_count <-_raw_spin_lock_irqsave 2057 <idle>-0 3dN.2 17us : __remove_hrtimer <-remove_hrtimer.part.16 2058 <idle>-0 3dN.2 17us : hrtimer_force_reprogram <-__remove_hrtimer 2059 <idle>-0 3dN.2 17us : tick_program_event <-hrtimer_force_reprogram 2060 <idle>-0 3dN.2 18us : clockevents_program_event <-tick_program_event 2061 <idle>-0 3dN.2 18us : ktime_get <-clockevents_program_event 2062 <idle>-0 3dN.2 18us : lapic_next_event <-clockevents_program_event 2063 <idle>-0 3dN.2 19us : _raw_spin_unlock_irqrestore <-hrtimer_try_to_cancel 2064 <idle>-0 3dN.2 19us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2065 <idle>-0 3dN.1 19us : hrtimer_forward <-tick_nohz_idle_exit 2066 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward 2067 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward 2068 <idle>-0 3dN.1 20us : hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11 2069 <idle>-0 3dN.1 20us : __hrtimer_start_range_ns <-hrtimer_start_range_ns 2070 <idle>-0 3dN.1 21us : lock_hrtimer_base.isra.18 <-__hrtimer_start_range_ns 2071 <idle>-0 3dN.1 21us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18 2072 <idle>-0 3dN.1 21us : add_preempt_count <-_raw_spin_lock_irqsave 2073 <idle>-0 3dN.2 22us : ktime_add_safe <-__hrtimer_start_range_ns 2074 <idle>-0 3dN.2 22us : enqueue_hrtimer <-__hrtimer_start_range_ns 2075 <idle>-0 3dN.2 22us : tick_program_event <-__hrtimer_start_range_ns 2076 <idle>-0 3dN.2 23us : clockevents_program_event <-tick_program_event 2077 <idle>-0 3dN.2 23us : ktime_get <-clockevents_program_event 2078 <idle>-0 3dN.2 23us : lapic_next_event <-clockevents_program_event 2079 <idle>-0 3dN.2 24us : _raw_spin_unlock_irqrestore <-__hrtimer_start_range_ns 2080 <idle>-0 3dN.2 24us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2081 <idle>-0 3dN.1 24us : account_idle_ticks <-tick_nohz_idle_exit 2082 <idle>-0 3dN.1 24us : account_idle_time <-account_idle_ticks 2083 <idle>-0 3.N.1 25us : sub_preempt_count <-cpu_idle 2084 <idle>-0 3.N.. 25us : schedule <-cpu_idle 2085 <idle>-0 3.N.. 25us : __schedule <-preempt_schedule 2086 <idle>-0 3.N.. 26us : add_preempt_count <-__schedule 2087 <idle>-0 3.N.1 26us : rcu_note_context_switch <-__schedule 2088 <idle>-0 3.N.1 26us : rcu_sched_qs <-rcu_note_context_switch 2089 <idle>-0 3dN.1 27us : rcu_preempt_qs <-rcu_note_context_switch 2090 <idle>-0 3.N.1 27us : _raw_spin_lock_irq <-__schedule 2091 <idle>-0 3dN.1 27us : add_preempt_count <-_raw_spin_lock_irq 2092 <idle>-0 3dN.2 28us : put_prev_task_idle <-__schedule 2093 <idle>-0 3dN.2 28us : pick_next_task_stop <-pick_next_task 2094 <idle>-0 3dN.2 28us : pick_next_task_rt <-pick_next_task 2095 <idle>-0 3dN.2 29us : dequeue_pushable_task <-pick_next_task_rt 2096 <idle>-0 3d..3 29us : __schedule <-preempt_schedule 2097 <idle>-0 3d..3 30us : 0:120:R ==> [003] 2448: 94:R sleep 2098 2099This isn't that big of a trace, even with function tracing enabled, 2100so I included the entire trace. 2101 2102The interrupt went off while when the system was idle. Somewhere 2103before task_woken_rt() was called, the NEED_RESCHED flag was set, 2104this is indicated by the first occurrence of the 'N' flag. 2105 2106Latency tracing and events 2107-------------------------- 2108As function tracing can induce a much larger latency, but without 2109seeing what happens within the latency it is hard to know what 2110caused it. There is a middle ground, and that is with enabling 2111events. 2112:: 2113 2114 # echo 0 > options/function-trace 2115 # echo wakeup_rt > current_tracer 2116 # echo 1 > events/enable 2117 # echo 1 > tracing_on 2118 # echo 0 > tracing_max_latency 2119 # chrt -f 5 sleep 1 2120 # echo 0 > tracing_on 2121 # cat trace 2122 # tracer: wakeup_rt 2123 # 2124 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 2125 # -------------------------------------------------------------------- 2126 # latency: 6 us, #12/12, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2127 # ----------------- 2128 # | task: sleep-5882 (uid:0 nice:0 policy:1 rt_prio:5) 2129 # ----------------- 2130 # 2131 # _------=> CPU# 2132 # / _-----=> irqs-off 2133 # | / _----=> need-resched 2134 # || / _---=> hardirq/softirq 2135 # ||| / _--=> preempt-depth 2136 # |||| / delay 2137 # cmd pid ||||| time | caller 2138 # \ / ||||| \ | / 2139 <idle>-0 2d.h4 0us : 0:120:R + [002] 5882: 94:R sleep 2140 <idle>-0 2d.h4 0us : ttwu_do_activate.constprop.87 <-try_to_wake_up 2141 <idle>-0 2d.h4 1us : sched_wakeup: comm=sleep pid=5882 prio=94 success=1 target_cpu=002 2142 <idle>-0 2dNh2 1us : hrtimer_expire_exit: hrtimer=ffff88007796feb8 2143 <idle>-0 2.N.2 2us : power_end: cpu_id=2 2144 <idle>-0 2.N.2 3us : cpu_idle: state=4294967295 cpu_id=2 2145 <idle>-0 2dN.3 4us : hrtimer_cancel: hrtimer=ffff88007d50d5e0 2146 <idle>-0 2dN.3 4us : hrtimer_start: hrtimer=ffff88007d50d5e0 function=tick_sched_timer expires=34311211000000 softexpires=34311211000000 2147 <idle>-0 2.N.2 5us : rcu_utilization: Start context switch 2148 <idle>-0 2.N.2 5us : rcu_utilization: End context switch 2149 <idle>-0 2d..3 6us : __schedule <-schedule 2150 <idle>-0 2d..3 6us : 0:120:R ==> [002] 5882: 94:R sleep 2151 2152 2153Hardware Latency Detector 2154------------------------- 2155 2156The hardware latency detector is executed by enabling the "hwlat" tracer. 2157 2158NOTE, this tracer will affect the performance of the system as it will 2159periodically make a CPU constantly busy with interrupts disabled. 2160:: 2161 2162 # echo hwlat > current_tracer 2163 # sleep 100 2164 # cat trace 2165 # tracer: hwlat 2166 # 2167 # entries-in-buffer/entries-written: 13/13 #P:8 2168 # 2169 # _-----=> irqs-off 2170 # / _----=> need-resched 2171 # | / _---=> hardirq/softirq 2172 # || / _--=> preempt-depth 2173 # ||| / delay 2174 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2175 # | | | |||| | | 2176 <...>-1729 [001] d... 678.473449: #1 inner/outer(us): 11/12 ts:1581527483.343962693 count:6 2177 <...>-1729 [004] d... 689.556542: #2 inner/outer(us): 16/9 ts:1581527494.889008092 count:1 2178 <...>-1729 [005] d... 714.756290: #3 inner/outer(us): 16/16 ts:1581527519.678961629 count:5 2179 <...>-1729 [001] d... 718.788247: #4 inner/outer(us): 9/17 ts:1581527523.889012713 count:1 2180 <...>-1729 [002] d... 719.796341: #5 inner/outer(us): 13/9 ts:1581527524.912872606 count:1 2181 <...>-1729 [006] d... 844.787091: #6 inner/outer(us): 9/12 ts:1581527649.889048502 count:2 2182 <...>-1729 [003] d... 849.827033: #7 inner/outer(us): 18/9 ts:1581527654.889013793 count:1 2183 <...>-1729 [007] d... 853.859002: #8 inner/outer(us): 9/12 ts:1581527658.889065736 count:1 2184 <...>-1729 [001] d... 855.874978: #9 inner/outer(us): 9/11 ts:1581527660.861991877 count:1 2185 <...>-1729 [001] d... 863.938932: #10 inner/outer(us): 9/11 ts:1581527668.970010500 count:1 nmi-total:7 nmi-count:1 2186 <...>-1729 [007] d... 878.050780: #11 inner/outer(us): 9/12 ts:1581527683.385002600 count:1 nmi-total:5 nmi-count:1 2187 <...>-1729 [007] d... 886.114702: #12 inner/outer(us): 9/12 ts:1581527691.385001600 count:1 2188 2189 2190The above output is somewhat the same in the header. All events will have 2191interrupts disabled 'd'. Under the FUNCTION title there is: 2192 2193 #1 2194 This is the count of events recorded that were greater than the 2195 tracing_threshold (See below). 2196 2197 inner/outer(us): 11/11 2198 2199 This shows two numbers as "inner latency" and "outer latency". The test 2200 runs in a loop checking a timestamp twice. The latency detected within 2201 the two timestamps is the "inner latency" and the latency detected 2202 after the previous timestamp and the next timestamp in the loop is 2203 the "outer latency". 2204 2205 ts:1581527483.343962693 2206 2207 The absolute timestamp that the first latency was recorded in the window. 2208 2209 count:6 2210 2211 The number of times a latency was detected during the window. 2212 2213 nmi-total:7 nmi-count:1 2214 2215 On architectures that support it, if an NMI comes in during the 2216 test, the time spent in NMI is reported in "nmi-total" (in 2217 microseconds). 2218 2219 All architectures that have NMIs will show the "nmi-count" if an 2220 NMI comes in during the test. 2221 2222hwlat files: 2223 2224 tracing_threshold 2225 This gets automatically set to "10" to represent 10 2226 microseconds. This is the threshold of latency that 2227 needs to be detected before the trace will be recorded. 2228 2229 Note, when hwlat tracer is finished (another tracer is 2230 written into "current_tracer"), the original value for 2231 tracing_threshold is placed back into this file. 2232 2233 hwlat_detector/width 2234 The length of time the test runs with interrupts disabled. 2235 2236 hwlat_detector/window 2237 The length of time of the window which the test 2238 runs. That is, the test will run for "width" 2239 microseconds per "window" microseconds 2240 2241 tracing_cpumask 2242 When the test is started. A kernel thread is created that 2243 runs the test. This thread will alternate between CPUs 2244 listed in the tracing_cpumask between each period 2245 (one "window"). To limit the test to specific CPUs 2246 set the mask in this file to only the CPUs that the test 2247 should run on. 2248 2249function 2250-------- 2251 2252This tracer is the function tracer. Enabling the function tracer 2253can be done from the debug file system. Make sure the 2254ftrace_enabled is set; otherwise this tracer is a nop. 2255See the "ftrace_enabled" section below. 2256:: 2257 2258 # sysctl kernel.ftrace_enabled=1 2259 # echo function > current_tracer 2260 # echo 1 > tracing_on 2261 # usleep 1 2262 # echo 0 > tracing_on 2263 # cat trace 2264 # tracer: function 2265 # 2266 # entries-in-buffer/entries-written: 24799/24799 #P:4 2267 # 2268 # _-----=> irqs-off 2269 # / _----=> need-resched 2270 # | / _---=> hardirq/softirq 2271 # || / _--=> preempt-depth 2272 # ||| / delay 2273 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2274 # | | | |||| | | 2275 bash-1994 [002] .... 3082.063030: mutex_unlock <-rb_simple_write 2276 bash-1994 [002] .... 3082.063031: __mutex_unlock_slowpath <-mutex_unlock 2277 bash-1994 [002] .... 3082.063031: __fsnotify_parent <-fsnotify_modify 2278 bash-1994 [002] .... 3082.063032: fsnotify <-fsnotify_modify 2279 bash-1994 [002] .... 3082.063032: __srcu_read_lock <-fsnotify 2280 bash-1994 [002] .... 3082.063032: add_preempt_count <-__srcu_read_lock 2281 bash-1994 [002] ...1 3082.063032: sub_preempt_count <-__srcu_read_lock 2282 bash-1994 [002] .... 3082.063033: __srcu_read_unlock <-fsnotify 2283 [...] 2284 2285 2286Note: function tracer uses ring buffers to store the above 2287entries. The newest data may overwrite the oldest data. 2288Sometimes using echo to stop the trace is not sufficient because 2289the tracing could have overwritten the data that you wanted to 2290record. For this reason, it is sometimes better to disable 2291tracing directly from a program. This allows you to stop the 2292tracing at the point that you hit the part that you are 2293interested in. To disable the tracing directly from a C program, 2294something like following code snippet can be used:: 2295 2296 int trace_fd; 2297 [...] 2298 int main(int argc, char *argv[]) { 2299 [...] 2300 trace_fd = open(tracing_file("tracing_on"), O_WRONLY); 2301 [...] 2302 if (condition_hit()) { 2303 write(trace_fd, "0", 1); 2304 } 2305 [...] 2306 } 2307 2308 2309Single thread tracing 2310--------------------- 2311 2312By writing into set_ftrace_pid you can trace a 2313single thread. For example:: 2314 2315 # cat set_ftrace_pid 2316 no pid 2317 # echo 3111 > set_ftrace_pid 2318 # cat set_ftrace_pid 2319 3111 2320 # echo function > current_tracer 2321 # cat trace | head 2322 # tracer: function 2323 # 2324 # TASK-PID CPU# TIMESTAMP FUNCTION 2325 # | | | | | 2326 yum-updatesd-3111 [003] 1637.254676: finish_task_switch <-thread_return 2327 yum-updatesd-3111 [003] 1637.254681: hrtimer_cancel <-schedule_hrtimeout_range 2328 yum-updatesd-3111 [003] 1637.254682: hrtimer_try_to_cancel <-hrtimer_cancel 2329 yum-updatesd-3111 [003] 1637.254683: lock_hrtimer_base <-hrtimer_try_to_cancel 2330 yum-updatesd-3111 [003] 1637.254685: fget_light <-do_sys_poll 2331 yum-updatesd-3111 [003] 1637.254686: pipe_poll <-do_sys_poll 2332 # echo > set_ftrace_pid 2333 # cat trace |head 2334 # tracer: function 2335 # 2336 # TASK-PID CPU# TIMESTAMP FUNCTION 2337 # | | | | | 2338 ##### CPU 3 buffer started #### 2339 yum-updatesd-3111 [003] 1701.957688: free_poll_entry <-poll_freewait 2340 yum-updatesd-3111 [003] 1701.957689: remove_wait_queue <-free_poll_entry 2341 yum-updatesd-3111 [003] 1701.957691: fput <-free_poll_entry 2342 yum-updatesd-3111 [003] 1701.957692: audit_syscall_exit <-sysret_audit 2343 yum-updatesd-3111 [003] 1701.957693: path_put <-audit_syscall_exit 2344 2345If you want to trace a function when executing, you could use 2346something like this simple program. 2347:: 2348 2349 #include <stdio.h> 2350 #include <stdlib.h> 2351 #include <sys/types.h> 2352 #include <sys/stat.h> 2353 #include <fcntl.h> 2354 #include <unistd.h> 2355 #include <string.h> 2356 2357 #define _STR(x) #x 2358 #define STR(x) _STR(x) 2359 #define MAX_PATH 256 2360 2361 const char *find_tracefs(void) 2362 { 2363 static char tracefs[MAX_PATH+1]; 2364 static int tracefs_found; 2365 char type[100]; 2366 FILE *fp; 2367 2368 if (tracefs_found) 2369 return tracefs; 2370 2371 if ((fp = fopen("/proc/mounts","r")) == NULL) { 2372 perror("/proc/mounts"); 2373 return NULL; 2374 } 2375 2376 while (fscanf(fp, "%*s %" 2377 STR(MAX_PATH) 2378 "s %99s %*s %*d %*d\n", 2379 tracefs, type) == 2) { 2380 if (strcmp(type, "tracefs") == 0) 2381 break; 2382 } 2383 fclose(fp); 2384 2385 if (strcmp(type, "tracefs") != 0) { 2386 fprintf(stderr, "tracefs not mounted"); 2387 return NULL; 2388 } 2389 2390 strcat(tracefs, "/tracing/"); 2391 tracefs_found = 1; 2392 2393 return tracefs; 2394 } 2395 2396 const char *tracing_file(const char *file_name) 2397 { 2398 static char trace_file[MAX_PATH+1]; 2399 snprintf(trace_file, MAX_PATH, "%s/%s", find_tracefs(), file_name); 2400 return trace_file; 2401 } 2402 2403 int main (int argc, char **argv) 2404 { 2405 if (argc < 1) 2406 exit(-1); 2407 2408 if (fork() > 0) { 2409 int fd, ffd; 2410 char line[64]; 2411 int s; 2412 2413 ffd = open(tracing_file("current_tracer"), O_WRONLY); 2414 if (ffd < 0) 2415 exit(-1); 2416 write(ffd, "nop", 3); 2417 2418 fd = open(tracing_file("set_ftrace_pid"), O_WRONLY); 2419 s = sprintf(line, "%d\n", getpid()); 2420 write(fd, line, s); 2421 2422 write(ffd, "function", 8); 2423 2424 close(fd); 2425 close(ffd); 2426 2427 execvp(argv[1], argv+1); 2428 } 2429 2430 return 0; 2431 } 2432 2433Or this simple script! 2434:: 2435 2436 #!/bin/bash 2437 2438 tracefs=`sed -ne 's/^tracefs \(.*\) tracefs.*/\1/p' /proc/mounts` 2439 echo nop > $tracefs/tracing/current_tracer 2440 echo 0 > $tracefs/tracing/tracing_on 2441 echo $$ > $tracefs/tracing/set_ftrace_pid 2442 echo function > $tracefs/tracing/current_tracer 2443 echo 1 > $tracefs/tracing/tracing_on 2444 exec "$@" 2445 2446 2447function graph tracer 2448--------------------------- 2449 2450This tracer is similar to the function tracer except that it 2451probes a function on its entry and its exit. This is done by 2452using a dynamically allocated stack of return addresses in each 2453task_struct. On function entry the tracer overwrites the return 2454address of each function traced to set a custom probe. Thus the 2455original return address is stored on the stack of return address 2456in the task_struct. 2457 2458Probing on both ends of a function leads to special features 2459such as: 2460 2461- measure of a function's time execution 2462- having a reliable call stack to draw function calls graph 2463 2464This tracer is useful in several situations: 2465 2466- you want to find the reason of a strange kernel behavior and 2467 need to see what happens in detail on any areas (or specific 2468 ones). 2469 2470- you are experiencing weird latencies but it's difficult to 2471 find its origin. 2472 2473- you want to find quickly which path is taken by a specific 2474 function 2475 2476- you just want to peek inside a working kernel and want to see 2477 what happens there. 2478 2479:: 2480 2481 # tracer: function_graph 2482 # 2483 # CPU DURATION FUNCTION CALLS 2484 # | | | | | | | 2485 2486 0) | sys_open() { 2487 0) | do_sys_open() { 2488 0) | getname() { 2489 0) | kmem_cache_alloc() { 2490 0) 1.382 us | __might_sleep(); 2491 0) 2.478 us | } 2492 0) | strncpy_from_user() { 2493 0) | might_fault() { 2494 0) 1.389 us | __might_sleep(); 2495 0) 2.553 us | } 2496 0) 3.807 us | } 2497 0) 7.876 us | } 2498 0) | alloc_fd() { 2499 0) 0.668 us | _spin_lock(); 2500 0) 0.570 us | expand_files(); 2501 0) 0.586 us | _spin_unlock(); 2502 2503 2504There are several columns that can be dynamically 2505enabled/disabled. You can use every combination of options you 2506want, depending on your needs. 2507 2508- The cpu number on which the function executed is default 2509 enabled. It is sometimes better to only trace one cpu (see 2510 tracing_cpu_mask file) or you might sometimes see unordered 2511 function calls while cpu tracing switch. 2512 2513 - hide: echo nofuncgraph-cpu > trace_options 2514 - show: echo funcgraph-cpu > trace_options 2515 2516- The duration (function's time of execution) is displayed on 2517 the closing bracket line of a function or on the same line 2518 than the current function in case of a leaf one. It is default 2519 enabled. 2520 2521 - hide: echo nofuncgraph-duration > trace_options 2522 - show: echo funcgraph-duration > trace_options 2523 2524- The overhead field precedes the duration field in case of 2525 reached duration thresholds. 2526 2527 - hide: echo nofuncgraph-overhead > trace_options 2528 - show: echo funcgraph-overhead > trace_options 2529 - depends on: funcgraph-duration 2530 2531 ie:: 2532 2533 3) # 1837.709 us | } /* __switch_to */ 2534 3) | finish_task_switch() { 2535 3) 0.313 us | _raw_spin_unlock_irq(); 2536 3) 3.177 us | } 2537 3) # 1889.063 us | } /* __schedule */ 2538 3) ! 140.417 us | } /* __schedule */ 2539 3) # 2034.948 us | } /* schedule */ 2540 3) * 33998.59 us | } /* schedule_preempt_disabled */ 2541 2542 [...] 2543 2544 1) 0.260 us | msecs_to_jiffies(); 2545 1) 0.313 us | __rcu_read_unlock(); 2546 1) + 61.770 us | } 2547 1) + 64.479 us | } 2548 1) 0.313 us | rcu_bh_qs(); 2549 1) 0.313 us | __local_bh_enable(); 2550 1) ! 217.240 us | } 2551 1) 0.365 us | idle_cpu(); 2552 1) | rcu_irq_exit() { 2553 1) 0.417 us | rcu_eqs_enter_common.isra.47(); 2554 1) 3.125 us | } 2555 1) ! 227.812 us | } 2556 1) ! 457.395 us | } 2557 1) @ 119760.2 us | } 2558 2559 [...] 2560 2561 2) | handle_IPI() { 2562 1) 6.979 us | } 2563 2) 0.417 us | scheduler_ipi(); 2564 1) 9.791 us | } 2565 1) + 12.917 us | } 2566 2) 3.490 us | } 2567 1) + 15.729 us | } 2568 1) + 18.542 us | } 2569 2) $ 3594274 us | } 2570 2571Flags:: 2572 2573 + means that the function exceeded 10 usecs. 2574 ! means that the function exceeded 100 usecs. 2575 # means that the function exceeded 1000 usecs. 2576 * means that the function exceeded 10 msecs. 2577 @ means that the function exceeded 100 msecs. 2578 $ means that the function exceeded 1 sec. 2579 2580 2581- The task/pid field displays the thread cmdline and pid which 2582 executed the function. It is default disabled. 2583 2584 - hide: echo nofuncgraph-proc > trace_options 2585 - show: echo funcgraph-proc > trace_options 2586 2587 ie:: 2588 2589 # tracer: function_graph 2590 # 2591 # CPU TASK/PID DURATION FUNCTION CALLS 2592 # | | | | | | | | | 2593 0) sh-4802 | | d_free() { 2594 0) sh-4802 | | call_rcu() { 2595 0) sh-4802 | | __call_rcu() { 2596 0) sh-4802 | 0.616 us | rcu_process_gp_end(); 2597 0) sh-4802 | 0.586 us | check_for_new_grace_period(); 2598 0) sh-4802 | 2.899 us | } 2599 0) sh-4802 | 4.040 us | } 2600 0) sh-4802 | 5.151 us | } 2601 0) sh-4802 | + 49.370 us | } 2602 2603 2604- The absolute time field is an absolute timestamp given by the 2605 system clock since it started. A snapshot of this time is 2606 given on each entry/exit of functions 2607 2608 - hide: echo nofuncgraph-abstime > trace_options 2609 - show: echo funcgraph-abstime > trace_options 2610 2611 ie:: 2612 2613 # 2614 # TIME CPU DURATION FUNCTION CALLS 2615 # | | | | | | | | 2616 360.774522 | 1) 0.541 us | } 2617 360.774522 | 1) 4.663 us | } 2618 360.774523 | 1) 0.541 us | __wake_up_bit(); 2619 360.774524 | 1) 6.796 us | } 2620 360.774524 | 1) 7.952 us | } 2621 360.774525 | 1) 9.063 us | } 2622 360.774525 | 1) 0.615 us | journal_mark_dirty(); 2623 360.774527 | 1) 0.578 us | __brelse(); 2624 360.774528 | 1) | reiserfs_prepare_for_journal() { 2625 360.774528 | 1) | unlock_buffer() { 2626 360.774529 | 1) | wake_up_bit() { 2627 360.774529 | 1) | bit_waitqueue() { 2628 360.774530 | 1) 0.594 us | __phys_addr(); 2629 2630 2631The function name is always displayed after the closing bracket 2632for a function if the start of that function is not in the 2633trace buffer. 2634 2635Display of the function name after the closing bracket may be 2636enabled for functions whose start is in the trace buffer, 2637allowing easier searching with grep for function durations. 2638It is default disabled. 2639 2640 - hide: echo nofuncgraph-tail > trace_options 2641 - show: echo funcgraph-tail > trace_options 2642 2643 Example with nofuncgraph-tail (default):: 2644 2645 0) | putname() { 2646 0) | kmem_cache_free() { 2647 0) 0.518 us | __phys_addr(); 2648 0) 1.757 us | } 2649 0) 2.861 us | } 2650 2651 Example with funcgraph-tail:: 2652 2653 0) | putname() { 2654 0) | kmem_cache_free() { 2655 0) 0.518 us | __phys_addr(); 2656 0) 1.757 us | } /* kmem_cache_free() */ 2657 0) 2.861 us | } /* putname() */ 2658 2659You can put some comments on specific functions by using 2660trace_printk() For example, if you want to put a comment inside 2661the __might_sleep() function, you just have to include 2662<linux/ftrace.h> and call trace_printk() inside __might_sleep():: 2663 2664 trace_printk("I'm a comment!\n") 2665 2666will produce:: 2667 2668 1) | __might_sleep() { 2669 1) | /* I'm a comment! */ 2670 1) 1.449 us | } 2671 2672 2673You might find other useful features for this tracer in the 2674following "dynamic ftrace" section such as tracing only specific 2675functions or tasks. 2676 2677dynamic ftrace 2678-------------- 2679 2680If CONFIG_DYNAMIC_FTRACE is set, the system will run with 2681virtually no overhead when function tracing is disabled. The way 2682this works is the mcount function call (placed at the start of 2683every kernel function, produced by the -pg switch in gcc), 2684starts of pointing to a simple return. (Enabling FTRACE will 2685include the -pg switch in the compiling of the kernel.) 2686 2687At compile time every C file object is run through the 2688recordmcount program (located in the scripts directory). This 2689program will parse the ELF headers in the C object to find all 2690the locations in the .text section that call mcount. Starting 2691with gcc version 4.6, the -mfentry has been added for x86, which 2692calls "__fentry__" instead of "mcount". Which is called before 2693the creation of the stack frame. 2694 2695Note, not all sections are traced. They may be prevented by either 2696a notrace, or blocked another way and all inline functions are not 2697traced. Check the "available_filter_functions" file to see what functions 2698can be traced. 2699 2700A section called "__mcount_loc" is created that holds 2701references to all the mcount/fentry call sites in the .text section. 2702The recordmcount program re-links this section back into the 2703original object. The final linking stage of the kernel will add all these 2704references into a single table. 2705 2706On boot up, before SMP is initialized, the dynamic ftrace code 2707scans this table and updates all the locations into nops. It 2708also records the locations, which are added to the 2709available_filter_functions list. Modules are processed as they 2710are loaded and before they are executed. When a module is 2711unloaded, it also removes its functions from the ftrace function 2712list. This is automatic in the module unload code, and the 2713module author does not need to worry about it. 2714 2715When tracing is enabled, the process of modifying the function 2716tracepoints is dependent on architecture. The old method is to use 2717kstop_machine to prevent races with the CPUs executing code being 2718modified (which can cause the CPU to do undesirable things, especially 2719if the modified code crosses cache (or page) boundaries), and the nops are 2720patched back to calls. But this time, they do not call mcount 2721(which is just a function stub). They now call into the ftrace 2722infrastructure. 2723 2724The new method of modifying the function tracepoints is to place 2725a breakpoint at the location to be modified, sync all CPUs, modify 2726the rest of the instruction not covered by the breakpoint. Sync 2727all CPUs again, and then remove the breakpoint with the finished 2728version to the ftrace call site. 2729 2730Some archs do not even need to monkey around with the synchronization, 2731and can just slap the new code on top of the old without any 2732problems with other CPUs executing it at the same time. 2733 2734One special side-effect to the recording of the functions being 2735traced is that we can now selectively choose which functions we 2736wish to trace and which ones we want the mcount calls to remain 2737as nops. 2738 2739Two files are used, one for enabling and one for disabling the 2740tracing of specified functions. They are: 2741 2742 set_ftrace_filter 2743 2744and 2745 2746 set_ftrace_notrace 2747 2748A list of available functions that you can add to these files is 2749listed in: 2750 2751 available_filter_functions 2752 2753:: 2754 2755 # cat available_filter_functions 2756 put_prev_task_idle 2757 kmem_cache_create 2758 pick_next_task_rt 2759 get_online_cpus 2760 pick_next_task_fair 2761 mutex_lock 2762 [...] 2763 2764If I am only interested in sys_nanosleep and hrtimer_interrupt:: 2765 2766 # echo sys_nanosleep hrtimer_interrupt > set_ftrace_filter 2767 # echo function > current_tracer 2768 # echo 1 > tracing_on 2769 # usleep 1 2770 # echo 0 > tracing_on 2771 # cat trace 2772 # tracer: function 2773 # 2774 # entries-in-buffer/entries-written: 5/5 #P:4 2775 # 2776 # _-----=> irqs-off 2777 # / _----=> need-resched 2778 # | / _---=> hardirq/softirq 2779 # || / _--=> preempt-depth 2780 # ||| / delay 2781 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2782 # | | | |||| | | 2783 usleep-2665 [001] .... 4186.475355: sys_nanosleep <-system_call_fastpath 2784 <idle>-0 [001] d.h1 4186.475409: hrtimer_interrupt <-smp_apic_timer_interrupt 2785 usleep-2665 [001] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt 2786 <idle>-0 [003] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt 2787 <idle>-0 [002] d.h1 4186.475427: hrtimer_interrupt <-smp_apic_timer_interrupt 2788 2789To see which functions are being traced, you can cat the file: 2790:: 2791 2792 # cat set_ftrace_filter 2793 hrtimer_interrupt 2794 sys_nanosleep 2795 2796 2797Perhaps this is not enough. The filters also allow glob(7) matching. 2798 2799 ``<match>*`` 2800 will match functions that begin with <match> 2801 ``*<match>`` 2802 will match functions that end with <match> 2803 ``*<match>*`` 2804 will match functions that have <match> in it 2805 ``<match1>*<match2>`` 2806 will match functions that begin with <match1> and end with <match2> 2807 2808.. note:: 2809 It is better to use quotes to enclose the wild cards, 2810 otherwise the shell may expand the parameters into names 2811 of files in the local directory. 2812 2813:: 2814 2815 # echo 'hrtimer_*' > set_ftrace_filter 2816 2817Produces:: 2818 2819 # tracer: function 2820 # 2821 # entries-in-buffer/entries-written: 897/897 #P:4 2822 # 2823 # _-----=> irqs-off 2824 # / _----=> need-resched 2825 # | / _---=> hardirq/softirq 2826 # || / _--=> preempt-depth 2827 # ||| / delay 2828 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2829 # | | | |||| | | 2830 <idle>-0 [003] dN.1 4228.547803: hrtimer_cancel <-tick_nohz_idle_exit 2831 <idle>-0 [003] dN.1 4228.547804: hrtimer_try_to_cancel <-hrtimer_cancel 2832 <idle>-0 [003] dN.2 4228.547805: hrtimer_force_reprogram <-__remove_hrtimer 2833 <idle>-0 [003] dN.1 4228.547805: hrtimer_forward <-tick_nohz_idle_exit 2834 <idle>-0 [003] dN.1 4228.547805: hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11 2835 <idle>-0 [003] d..1 4228.547858: hrtimer_get_next_event <-get_next_timer_interrupt 2836 <idle>-0 [003] d..1 4228.547859: hrtimer_start <-__tick_nohz_idle_enter 2837 <idle>-0 [003] d..2 4228.547860: hrtimer_force_reprogram <-__rem 2838 2839Notice that we lost the sys_nanosleep. 2840:: 2841 2842 # cat set_ftrace_filter 2843 hrtimer_run_queues 2844 hrtimer_run_pending 2845 hrtimer_init 2846 hrtimer_cancel 2847 hrtimer_try_to_cancel 2848 hrtimer_forward 2849 hrtimer_start 2850 hrtimer_reprogram 2851 hrtimer_force_reprogram 2852 hrtimer_get_next_event 2853 hrtimer_interrupt 2854 hrtimer_nanosleep 2855 hrtimer_wakeup 2856 hrtimer_get_remaining 2857 hrtimer_get_res 2858 hrtimer_init_sleeper 2859 2860 2861This is because the '>' and '>>' act just like they do in bash. 2862To rewrite the filters, use '>' 2863To append to the filters, use '>>' 2864 2865To clear out a filter so that all functions will be recorded 2866again:: 2867 2868 # echo > set_ftrace_filter 2869 # cat set_ftrace_filter 2870 # 2871 2872Again, now we want to append. 2873 2874:: 2875 2876 # echo sys_nanosleep > set_ftrace_filter 2877 # cat set_ftrace_filter 2878 sys_nanosleep 2879 # echo 'hrtimer_*' >> set_ftrace_filter 2880 # cat set_ftrace_filter 2881 hrtimer_run_queues 2882 hrtimer_run_pending 2883 hrtimer_init 2884 hrtimer_cancel 2885 hrtimer_try_to_cancel 2886 hrtimer_forward 2887 hrtimer_start 2888 hrtimer_reprogram 2889 hrtimer_force_reprogram 2890 hrtimer_get_next_event 2891 hrtimer_interrupt 2892 sys_nanosleep 2893 hrtimer_nanosleep 2894 hrtimer_wakeup 2895 hrtimer_get_remaining 2896 hrtimer_get_res 2897 hrtimer_init_sleeper 2898 2899 2900The set_ftrace_notrace prevents those functions from being 2901traced. 2902:: 2903 2904 # echo '*preempt*' '*lock*' > set_ftrace_notrace 2905 2906Produces:: 2907 2908 # tracer: function 2909 # 2910 # entries-in-buffer/entries-written: 39608/39608 #P:4 2911 # 2912 # _-----=> irqs-off 2913 # / _----=> need-resched 2914 # | / _---=> hardirq/softirq 2915 # || / _--=> preempt-depth 2916 # ||| / delay 2917 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2918 # | | | |||| | | 2919 bash-1994 [000] .... 4342.324896: file_ra_state_init <-do_dentry_open 2920 bash-1994 [000] .... 4342.324897: open_check_o_direct <-do_last 2921 bash-1994 [000] .... 4342.324897: ima_file_check <-do_last 2922 bash-1994 [000] .... 4342.324898: process_measurement <-ima_file_check 2923 bash-1994 [000] .... 4342.324898: ima_get_action <-process_measurement 2924 bash-1994 [000] .... 4342.324898: ima_match_policy <-ima_get_action 2925 bash-1994 [000] .... 4342.324899: do_truncate <-do_last 2926 bash-1994 [000] .... 4342.324899: should_remove_suid <-do_truncate 2927 bash-1994 [000] .... 4342.324899: notify_change <-do_truncate 2928 bash-1994 [000] .... 4342.324900: current_fs_time <-notify_change 2929 bash-1994 [000] .... 4342.324900: current_kernel_time <-current_fs_time 2930 bash-1994 [000] .... 4342.324900: timespec_trunc <-current_fs_time 2931 2932We can see that there's no more lock or preempt tracing. 2933 2934Selecting function filters via index 2935------------------------------------ 2936 2937Because processing of strings is expensive (the address of the function 2938needs to be looked up before comparing to the string being passed in), 2939an index can be used as well to enable functions. This is useful in the 2940case of setting thousands of specific functions at a time. By passing 2941in a list of numbers, no string processing will occur. Instead, the function 2942at the specific location in the internal array (which corresponds to the 2943functions in the "available_filter_functions" file), is selected. 2944 2945:: 2946 2947 # echo 1 > set_ftrace_filter 2948 2949Will select the first function listed in "available_filter_functions" 2950 2951:: 2952 2953 # head -1 available_filter_functions 2954 trace_initcall_finish_cb 2955 2956 # cat set_ftrace_filter 2957 trace_initcall_finish_cb 2958 2959 # head -50 available_filter_functions | tail -1 2960 x86_pmu_commit_txn 2961 2962 # echo 1 50 > set_ftrace_filter 2963 # cat set_ftrace_filter 2964 trace_initcall_finish_cb 2965 x86_pmu_commit_txn 2966 2967Dynamic ftrace with the function graph tracer 2968--------------------------------------------- 2969 2970Although what has been explained above concerns both the 2971function tracer and the function-graph-tracer, there are some 2972special features only available in the function-graph tracer. 2973 2974If you want to trace only one function and all of its children, 2975you just have to echo its name into set_graph_function:: 2976 2977 echo __do_fault > set_graph_function 2978 2979will produce the following "expanded" trace of the __do_fault() 2980function:: 2981 2982 0) | __do_fault() { 2983 0) | filemap_fault() { 2984 0) | find_lock_page() { 2985 0) 0.804 us | find_get_page(); 2986 0) | __might_sleep() { 2987 0) 1.329 us | } 2988 0) 3.904 us | } 2989 0) 4.979 us | } 2990 0) 0.653 us | _spin_lock(); 2991 0) 0.578 us | page_add_file_rmap(); 2992 0) 0.525 us | native_set_pte_at(); 2993 0) 0.585 us | _spin_unlock(); 2994 0) | unlock_page() { 2995 0) 0.541 us | page_waitqueue(); 2996 0) 0.639 us | __wake_up_bit(); 2997 0) 2.786 us | } 2998 0) + 14.237 us | } 2999 0) | __do_fault() { 3000 0) | filemap_fault() { 3001 0) | find_lock_page() { 3002 0) 0.698 us | find_get_page(); 3003 0) | __might_sleep() { 3004 0) 1.412 us | } 3005 0) 3.950 us | } 3006 0) 5.098 us | } 3007 0) 0.631 us | _spin_lock(); 3008 0) 0.571 us | page_add_file_rmap(); 3009 0) 0.526 us | native_set_pte_at(); 3010 0) 0.586 us | _spin_unlock(); 3011 0) | unlock_page() { 3012 0) 0.533 us | page_waitqueue(); 3013 0) 0.638 us | __wake_up_bit(); 3014 0) 2.793 us | } 3015 0) + 14.012 us | } 3016 3017You can also expand several functions at once:: 3018 3019 echo sys_open > set_graph_function 3020 echo sys_close >> set_graph_function 3021 3022Now if you want to go back to trace all functions you can clear 3023this special filter via:: 3024 3025 echo > set_graph_function 3026 3027 3028ftrace_enabled 3029-------------- 3030 3031Note, the proc sysctl ftrace_enable is a big on/off switch for the 3032function tracer. By default it is enabled (when function tracing is 3033enabled in the kernel). If it is disabled, all function tracing is 3034disabled. This includes not only the function tracers for ftrace, but 3035also for any other uses (perf, kprobes, stack tracing, profiling, etc). It 3036cannot be disabled if there is a callback with FTRACE_OPS_FL_PERMANENT set 3037registered. 3038 3039Please disable this with care. 3040 3041This can be disable (and enabled) with:: 3042 3043 sysctl kernel.ftrace_enabled=0 3044 sysctl kernel.ftrace_enabled=1 3045 3046 or 3047 3048 echo 0 > /proc/sys/kernel/ftrace_enabled 3049 echo 1 > /proc/sys/kernel/ftrace_enabled 3050 3051 3052Filter commands 3053--------------- 3054 3055A few commands are supported by the set_ftrace_filter interface. 3056Trace commands have the following format:: 3057 3058 <function>:<command>:<parameter> 3059 3060The following commands are supported: 3061 3062- mod: 3063 This command enables function filtering per module. The 3064 parameter defines the module. For example, if only the write* 3065 functions in the ext3 module are desired, run: 3066 3067 echo 'write*:mod:ext3' > set_ftrace_filter 3068 3069 This command interacts with the filter in the same way as 3070 filtering based on function names. Thus, adding more functions 3071 in a different module is accomplished by appending (>>) to the 3072 filter file. Remove specific module functions by prepending 3073 '!':: 3074 3075 echo '!writeback*:mod:ext3' >> set_ftrace_filter 3076 3077 Mod command supports module globbing. Disable tracing for all 3078 functions except a specific module:: 3079 3080 echo '!*:mod:!ext3' >> set_ftrace_filter 3081 3082 Disable tracing for all modules, but still trace kernel:: 3083 3084 echo '!*:mod:*' >> set_ftrace_filter 3085 3086 Enable filter only for kernel:: 3087 3088 echo '*write*:mod:!*' >> set_ftrace_filter 3089 3090 Enable filter for module globbing:: 3091 3092 echo '*write*:mod:*snd*' >> set_ftrace_filter 3093 3094- traceon/traceoff: 3095 These commands turn tracing on and off when the specified 3096 functions are hit. The parameter determines how many times the 3097 tracing system is turned on and off. If unspecified, there is 3098 no limit. For example, to disable tracing when a schedule bug 3099 is hit the first 5 times, run:: 3100 3101 echo '__schedule_bug:traceoff:5' > set_ftrace_filter 3102 3103 To always disable tracing when __schedule_bug is hit:: 3104 3105 echo '__schedule_bug:traceoff' > set_ftrace_filter 3106 3107 These commands are cumulative whether or not they are appended 3108 to set_ftrace_filter. To remove a command, prepend it by '!' 3109 and drop the parameter:: 3110 3111 echo '!__schedule_bug:traceoff:0' > set_ftrace_filter 3112 3113 The above removes the traceoff command for __schedule_bug 3114 that have a counter. To remove commands without counters:: 3115 3116 echo '!__schedule_bug:traceoff' > set_ftrace_filter 3117 3118- snapshot: 3119 Will cause a snapshot to be triggered when the function is hit. 3120 :: 3121 3122 echo 'native_flush_tlb_others:snapshot' > set_ftrace_filter 3123 3124 To only snapshot once: 3125 :: 3126 3127 echo 'native_flush_tlb_others:snapshot:1' > set_ftrace_filter 3128 3129 To remove the above commands:: 3130 3131 echo '!native_flush_tlb_others:snapshot' > set_ftrace_filter 3132 echo '!native_flush_tlb_others:snapshot:0' > set_ftrace_filter 3133 3134- enable_event/disable_event: 3135 These commands can enable or disable a trace event. Note, because 3136 function tracing callbacks are very sensitive, when these commands 3137 are registered, the trace point is activated, but disabled in 3138 a "soft" mode. That is, the tracepoint will be called, but 3139 just will not be traced. The event tracepoint stays in this mode 3140 as long as there's a command that triggers it. 3141 :: 3142 3143 echo 'try_to_wake_up:enable_event:sched:sched_switch:2' > \ 3144 set_ftrace_filter 3145 3146 The format is:: 3147 3148 <function>:enable_event:<system>:<event>[:count] 3149 <function>:disable_event:<system>:<event>[:count] 3150 3151 To remove the events commands:: 3152 3153 echo '!try_to_wake_up:enable_event:sched:sched_switch:0' > \ 3154 set_ftrace_filter 3155 echo '!schedule:disable_event:sched:sched_switch' > \ 3156 set_ftrace_filter 3157 3158- dump: 3159 When the function is hit, it will dump the contents of the ftrace 3160 ring buffer to the console. This is useful if you need to debug 3161 something, and want to dump the trace when a certain function 3162 is hit. Perhaps it's a function that is called before a triple 3163 fault happens and does not allow you to get a regular dump. 3164 3165- cpudump: 3166 When the function is hit, it will dump the contents of the ftrace 3167 ring buffer for the current CPU to the console. Unlike the "dump" 3168 command, it only prints out the contents of the ring buffer for the 3169 CPU that executed the function that triggered the dump. 3170 3171- stacktrace: 3172 When the function is hit, a stack trace is recorded. 3173 3174trace_pipe 3175---------- 3176 3177The trace_pipe outputs the same content as the trace file, but 3178the effect on the tracing is different. Every read from 3179trace_pipe is consumed. This means that subsequent reads will be 3180different. The trace is live. 3181:: 3182 3183 # echo function > current_tracer 3184 # cat trace_pipe > /tmp/trace.out & 3185 [1] 4153 3186 # echo 1 > tracing_on 3187 # usleep 1 3188 # echo 0 > tracing_on 3189 # cat trace 3190 # tracer: function 3191 # 3192 # entries-in-buffer/entries-written: 0/0 #P:4 3193 # 3194 # _-----=> irqs-off 3195 # / _----=> need-resched 3196 # | / _---=> hardirq/softirq 3197 # || / _--=> preempt-depth 3198 # ||| / delay 3199 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3200 # | | | |||| | | 3201 3202 # 3203 # cat /tmp/trace.out 3204 bash-1994 [000] .... 5281.568961: mutex_unlock <-rb_simple_write 3205 bash-1994 [000] .... 5281.568963: __mutex_unlock_slowpath <-mutex_unlock 3206 bash-1994 [000] .... 5281.568963: __fsnotify_parent <-fsnotify_modify 3207 bash-1994 [000] .... 5281.568964: fsnotify <-fsnotify_modify 3208 bash-1994 [000] .... 5281.568964: __srcu_read_lock <-fsnotify 3209 bash-1994 [000] .... 5281.568964: add_preempt_count <-__srcu_read_lock 3210 bash-1994 [000] ...1 5281.568965: sub_preempt_count <-__srcu_read_lock 3211 bash-1994 [000] .... 5281.568965: __srcu_read_unlock <-fsnotify 3212 bash-1994 [000] .... 5281.568967: sys_dup2 <-system_call_fastpath 3213 3214 3215Note, reading the trace_pipe file will block until more input is 3216added. This is contrary to the trace file. If any process opened 3217the trace file for reading, it will actually disable tracing and 3218prevent new entries from being added. The trace_pipe file does 3219not have this limitation. 3220 3221trace entries 3222------------- 3223 3224Having too much or not enough data can be troublesome in 3225diagnosing an issue in the kernel. The file buffer_size_kb is 3226used to modify the size of the internal trace buffers. The 3227number listed is the number of entries that can be recorded per 3228CPU. To know the full size, multiply the number of possible CPUs 3229with the number of entries. 3230:: 3231 3232 # cat buffer_size_kb 3233 1408 (units kilobytes) 3234 3235Or simply read buffer_total_size_kb 3236:: 3237 3238 # cat buffer_total_size_kb 3239 5632 3240 3241To modify the buffer, simple echo in a number (in 1024 byte segments). 3242:: 3243 3244 # echo 10000 > buffer_size_kb 3245 # cat buffer_size_kb 3246 10000 (units kilobytes) 3247 3248It will try to allocate as much as possible. If you allocate too 3249much, it can cause Out-Of-Memory to trigger. 3250:: 3251 3252 # echo 1000000000000 > buffer_size_kb 3253 -bash: echo: write error: Cannot allocate memory 3254 # cat buffer_size_kb 3255 85 3256 3257The per_cpu buffers can be changed individually as well: 3258:: 3259 3260 # echo 10000 > per_cpu/cpu0/buffer_size_kb 3261 # echo 100 > per_cpu/cpu1/buffer_size_kb 3262 3263When the per_cpu buffers are not the same, the buffer_size_kb 3264at the top level will just show an X 3265:: 3266 3267 # cat buffer_size_kb 3268 X 3269 3270This is where the buffer_total_size_kb is useful: 3271:: 3272 3273 # cat buffer_total_size_kb 3274 12916 3275 3276Writing to the top level buffer_size_kb will reset all the buffers 3277to be the same again. 3278 3279Snapshot 3280-------- 3281CONFIG_TRACER_SNAPSHOT makes a generic snapshot feature 3282available to all non latency tracers. (Latency tracers which 3283record max latency, such as "irqsoff" or "wakeup", can't use 3284this feature, since those are already using the snapshot 3285mechanism internally.) 3286 3287Snapshot preserves a current trace buffer at a particular point 3288in time without stopping tracing. Ftrace swaps the current 3289buffer with a spare buffer, and tracing continues in the new 3290current (=previous spare) buffer. 3291 3292The following tracefs files in "tracing" are related to this 3293feature: 3294 3295 snapshot: 3296 3297 This is used to take a snapshot and to read the output 3298 of the snapshot. Echo 1 into this file to allocate a 3299 spare buffer and to take a snapshot (swap), then read 3300 the snapshot from this file in the same format as 3301 "trace" (described above in the section "The File 3302 System"). Both reads snapshot and tracing are executable 3303 in parallel. When the spare buffer is allocated, echoing 3304 0 frees it, and echoing else (positive) values clear the 3305 snapshot contents. 3306 More details are shown in the table below. 3307 3308 +--------------+------------+------------+------------+ 3309 |status\\input | 0 | 1 | else | 3310 +==============+============+============+============+ 3311 |not allocated |(do nothing)| alloc+swap |(do nothing)| 3312 +--------------+------------+------------+------------+ 3313 |allocated | free | swap | clear | 3314 +--------------+------------+------------+------------+ 3315 3316Here is an example of using the snapshot feature. 3317:: 3318 3319 # echo 1 > events/sched/enable 3320 # echo 1 > snapshot 3321 # cat snapshot 3322 # tracer: nop 3323 # 3324 # entries-in-buffer/entries-written: 71/71 #P:8 3325 # 3326 # _-----=> irqs-off 3327 # / _----=> need-resched 3328 # | / _---=> hardirq/softirq 3329 # || / _--=> preempt-depth 3330 # ||| / delay 3331 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3332 # | | | |||| | | 3333 <idle>-0 [005] d... 2440.603828: sched_switch: prev_comm=swapper/5 prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=snapshot-test-2 next_pid=2242 next_prio=120 3334 sleep-2242 [005] d... 2440.603846: sched_switch: prev_comm=snapshot-test-2 prev_pid=2242 prev_prio=120 prev_state=R ==> next_comm=kworker/5:1 next_pid=60 next_prio=120 3335 [...] 3336 <idle>-0 [002] d... 2440.707230: sched_switch: prev_comm=swapper/2 prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=snapshot-test-2 next_pid=2229 next_prio=120 3337 3338 # cat trace 3339 # tracer: nop 3340 # 3341 # entries-in-buffer/entries-written: 77/77 #P:8 3342 # 3343 # _-----=> irqs-off 3344 # / _----=> need-resched 3345 # | / _---=> hardirq/softirq 3346 # || / _--=> preempt-depth 3347 # ||| / delay 3348 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3349 # | | | |||| | | 3350 <idle>-0 [007] d... 2440.707395: sched_switch: prev_comm=swapper/7 prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=snapshot-test-2 next_pid=2243 next_prio=120 3351 snapshot-test-2-2229 [002] d... 2440.707438: sched_switch: prev_comm=snapshot-test-2 prev_pid=2229 prev_prio=120 prev_state=S ==> next_comm=swapper/2 next_pid=0 next_prio=120 3352 [...] 3353 3354 3355If you try to use this snapshot feature when current tracer is 3356one of the latency tracers, you will get the following results. 3357:: 3358 3359 # echo wakeup > current_tracer 3360 # echo 1 > snapshot 3361 bash: echo: write error: Device or resource busy 3362 # cat snapshot 3363 cat: snapshot: Device or resource busy 3364 3365 3366Instances 3367--------- 3368In the tracefs tracing directory is a directory called "instances". 3369This directory can have new directories created inside of it using 3370mkdir, and removing directories with rmdir. The directory created 3371with mkdir in this directory will already contain files and other 3372directories after it is created. 3373:: 3374 3375 # mkdir instances/foo 3376 # ls instances/foo 3377 buffer_size_kb buffer_total_size_kb events free_buffer per_cpu 3378 set_event snapshot trace trace_clock trace_marker trace_options 3379 trace_pipe tracing_on 3380 3381As you can see, the new directory looks similar to the tracing directory 3382itself. In fact, it is very similar, except that the buffer and 3383events are agnostic from the main directory, or from any other 3384instances that are created. 3385 3386The files in the new directory work just like the files with the 3387same name in the tracing directory except the buffer that is used 3388is a separate and new buffer. The files affect that buffer but do not 3389affect the main buffer with the exception of trace_options. Currently, 3390the trace_options affect all instances and the top level buffer 3391the same, but this may change in future releases. That is, options 3392may become specific to the instance they reside in. 3393 3394Notice that none of the function tracer files are there, nor is 3395current_tracer and available_tracers. This is because the buffers 3396can currently only have events enabled for them. 3397:: 3398 3399 # mkdir instances/foo 3400 # mkdir instances/bar 3401 # mkdir instances/zoot 3402 # echo 100000 > buffer_size_kb 3403 # echo 1000 > instances/foo/buffer_size_kb 3404 # echo 5000 > instances/bar/per_cpu/cpu1/buffer_size_kb 3405 # echo function > current_trace 3406 # echo 1 > instances/foo/events/sched/sched_wakeup/enable 3407 # echo 1 > instances/foo/events/sched/sched_wakeup_new/enable 3408 # echo 1 > instances/foo/events/sched/sched_switch/enable 3409 # echo 1 > instances/bar/events/irq/enable 3410 # echo 1 > instances/zoot/events/syscalls/enable 3411 # cat trace_pipe 3412 CPU:2 [LOST 11745 EVENTS] 3413 bash-2044 [002] .... 10594.481032: _raw_spin_lock_irqsave <-get_page_from_freelist 3414 bash-2044 [002] d... 10594.481032: add_preempt_count <-_raw_spin_lock_irqsave 3415 bash-2044 [002] d..1 10594.481032: __rmqueue <-get_page_from_freelist 3416 bash-2044 [002] d..1 10594.481033: _raw_spin_unlock <-get_page_from_freelist 3417 bash-2044 [002] d..1 10594.481033: sub_preempt_count <-_raw_spin_unlock 3418 bash-2044 [002] d... 10594.481033: get_pageblock_flags_group <-get_pageblock_migratetype 3419 bash-2044 [002] d... 10594.481034: __mod_zone_page_state <-get_page_from_freelist 3420 bash-2044 [002] d... 10594.481034: zone_statistics <-get_page_from_freelist 3421 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics 3422 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics 3423 bash-2044 [002] .... 10594.481035: arch_dup_task_struct <-copy_process 3424 [...] 3425 3426 # cat instances/foo/trace_pipe 3427 bash-1998 [000] d..4 136.676759: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000 3428 bash-1998 [000] dN.4 136.676760: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000 3429 <idle>-0 [003] d.h3 136.676906: sched_wakeup: comm=rcu_preempt pid=9 prio=120 success=1 target_cpu=003 3430 <idle>-0 [003] d..3 136.676909: sched_switch: prev_comm=swapper/3 prev_pid=0 prev_prio=120 prev_state=R ==> next_comm=rcu_preempt next_pid=9 next_prio=120 3431 rcu_preempt-9 [003] d..3 136.676916: sched_switch: prev_comm=rcu_preempt prev_pid=9 prev_prio=120 prev_state=S ==> next_comm=swapper/3 next_pid=0 next_prio=120 3432 bash-1998 [000] d..4 136.677014: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000 3433 bash-1998 [000] dN.4 136.677016: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000 3434 bash-1998 [000] d..3 136.677018: sched_switch: prev_comm=bash prev_pid=1998 prev_prio=120 prev_state=R+ ==> next_comm=kworker/0:1 next_pid=59 next_prio=120 3435 kworker/0:1-59 [000] d..4 136.677022: sched_wakeup: comm=sshd pid=1995 prio=120 success=1 target_cpu=001 3436 kworker/0:1-59 [000] d..3 136.677025: sched_switch: prev_comm=kworker/0:1 prev_pid=59 prev_prio=120 prev_state=S ==> next_comm=bash next_pid=1998 next_prio=120 3437 [...] 3438 3439 # cat instances/bar/trace_pipe 3440 migration/1-14 [001] d.h3 138.732674: softirq_raise: vec=3 [action=NET_RX] 3441 <idle>-0 [001] dNh3 138.732725: softirq_raise: vec=3 [action=NET_RX] 3442 bash-1998 [000] d.h1 138.733101: softirq_raise: vec=1 [action=TIMER] 3443 bash-1998 [000] d.h1 138.733102: softirq_raise: vec=9 [action=RCU] 3444 bash-1998 [000] ..s2 138.733105: softirq_entry: vec=1 [action=TIMER] 3445 bash-1998 [000] ..s2 138.733106: softirq_exit: vec=1 [action=TIMER] 3446 bash-1998 [000] ..s2 138.733106: softirq_entry: vec=9 [action=RCU] 3447 bash-1998 [000] ..s2 138.733109: softirq_exit: vec=9 [action=RCU] 3448 sshd-1995 [001] d.h1 138.733278: irq_handler_entry: irq=21 name=uhci_hcd:usb4 3449 sshd-1995 [001] d.h1 138.733280: irq_handler_exit: irq=21 ret=unhandled 3450 sshd-1995 [001] d.h1 138.733281: irq_handler_entry: irq=21 name=eth0 3451 sshd-1995 [001] d.h1 138.733283: irq_handler_exit: irq=21 ret=handled 3452 [...] 3453 3454 # cat instances/zoot/trace 3455 # tracer: nop 3456 # 3457 # entries-in-buffer/entries-written: 18996/18996 #P:4 3458 # 3459 # _-----=> irqs-off 3460 # / _----=> need-resched 3461 # | / _---=> hardirq/softirq 3462 # || / _--=> preempt-depth 3463 # ||| / delay 3464 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3465 # | | | |||| | | 3466 bash-1998 [000] d... 140.733501: sys_write -> 0x2 3467 bash-1998 [000] d... 140.733504: sys_dup2(oldfd: a, newfd: 1) 3468 bash-1998 [000] d... 140.733506: sys_dup2 -> 0x1 3469 bash-1998 [000] d... 140.733508: sys_fcntl(fd: a, cmd: 1, arg: 0) 3470 bash-1998 [000] d... 140.733509: sys_fcntl -> 0x1 3471 bash-1998 [000] d... 140.733510: sys_close(fd: a) 3472 bash-1998 [000] d... 140.733510: sys_close -> 0x0 3473 bash-1998 [000] d... 140.733514: sys_rt_sigprocmask(how: 0, nset: 0, oset: 6e2768, sigsetsize: 8) 3474 bash-1998 [000] d... 140.733515: sys_rt_sigprocmask -> 0x0 3475 bash-1998 [000] d... 140.733516: sys_rt_sigaction(sig: 2, act: 7fff718846f0, oact: 7fff71884650, sigsetsize: 8) 3476 bash-1998 [000] d... 140.733516: sys_rt_sigaction -> 0x0 3477 3478You can see that the trace of the top most trace buffer shows only 3479the function tracing. The foo instance displays wakeups and task 3480switches. 3481 3482To remove the instances, simply delete their directories: 3483:: 3484 3485 # rmdir instances/foo 3486 # rmdir instances/bar 3487 # rmdir instances/zoot 3488 3489Note, if a process has a trace file open in one of the instance 3490directories, the rmdir will fail with EBUSY. 3491 3492 3493Stack trace 3494----------- 3495Since the kernel has a fixed sized stack, it is important not to 3496waste it in functions. A kernel developer must be conscience of 3497what they allocate on the stack. If they add too much, the system 3498can be in danger of a stack overflow, and corruption will occur, 3499usually leading to a system panic. 3500 3501There are some tools that check this, usually with interrupts 3502periodically checking usage. But if you can perform a check 3503at every function call that will become very useful. As ftrace provides 3504a function tracer, it makes it convenient to check the stack size 3505at every function call. This is enabled via the stack tracer. 3506 3507CONFIG_STACK_TRACER enables the ftrace stack tracing functionality. 3508To enable it, write a '1' into /proc/sys/kernel/stack_tracer_enabled. 3509:: 3510 3511 # echo 1 > /proc/sys/kernel/stack_tracer_enabled 3512 3513You can also enable it from the kernel command line to trace 3514the stack size of the kernel during boot up, by adding "stacktrace" 3515to the kernel command line parameter. 3516 3517After running it for a few minutes, the output looks like: 3518:: 3519 3520 # cat stack_max_size 3521 2928 3522 3523 # cat stack_trace 3524 Depth Size Location (18 entries) 3525 ----- ---- -------- 3526 0) 2928 224 update_sd_lb_stats+0xbc/0x4ac 3527 1) 2704 160 find_busiest_group+0x31/0x1f1 3528 2) 2544 256 load_balance+0xd9/0x662 3529 3) 2288 80 idle_balance+0xbb/0x130 3530 4) 2208 128 __schedule+0x26e/0x5b9 3531 5) 2080 16 schedule+0x64/0x66 3532 6) 2064 128 schedule_timeout+0x34/0xe0 3533 7) 1936 112 wait_for_common+0x97/0xf1 3534 8) 1824 16 wait_for_completion+0x1d/0x1f 3535 9) 1808 128 flush_work+0xfe/0x119 3536 10) 1680 16 tty_flush_to_ldisc+0x1e/0x20 3537 11) 1664 48 input_available_p+0x1d/0x5c 3538 12) 1616 48 n_tty_poll+0x6d/0x134 3539 13) 1568 64 tty_poll+0x64/0x7f 3540 14) 1504 880 do_select+0x31e/0x511 3541 15) 624 400 core_sys_select+0x177/0x216 3542 16) 224 96 sys_select+0x91/0xb9 3543 17) 128 128 system_call_fastpath+0x16/0x1b 3544 3545Note, if -mfentry is being used by gcc, functions get traced before 3546they set up the stack frame. This means that leaf level functions 3547are not tested by the stack tracer when -mfentry is used. 3548 3549Currently, -mfentry is used by gcc 4.6.0 and above on x86 only. 3550 3551More 3552---- 3553More details can be found in the source code, in the `kernel/trace/*.c` files. 3554