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 are 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.rst for more information. 38 39 40Implementation Details 41---------------------- 42 43See Documentation/trace/ftrace-design.rst 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_percent: 184 185 This is the watermark for how much the ring buffer needs to be filled 186 before a waiter is woken up. That is, if an application calls a 187 blocking read syscall on one of the per_cpu trace_pipe_raw files, it 188 will block until the given amount of data specified by buffer_percent 189 is in the ring buffer before it wakes the reader up. This also 190 controls how the splice system calls are blocked on this file:: 191 192 0 - means to wake up as soon as there is any data in the ring buffer. 193 50 - means to wake up when roughly half of the ring buffer sub-buffers 194 are full. 195 100 - means to block until the ring buffer is totally full and is 196 about to start overwriting the older data. 197 198 buffer_size_kb: 199 200 This sets or displays the number of kilobytes each CPU 201 buffer holds. By default, the trace buffers are the same size 202 for each CPU. The displayed number is the size of the 203 CPU buffer and not total size of all buffers. The 204 trace buffers are allocated in pages (blocks of memory 205 that the kernel uses for allocation, usually 4 KB in size). 206 A few extra pages may be allocated to accommodate buffer management 207 meta-data. If the last page allocated has room for more bytes 208 than requested, the rest of the page will be used, 209 making the actual allocation bigger than requested or shown. 210 ( Note, the size may not be a multiple of the page size 211 due to buffer management meta-data. ) 212 213 Buffer sizes for individual CPUs may vary 214 (see "per_cpu/cpu0/buffer_size_kb" below), and if they do 215 this file will show "X". 216 217 buffer_total_size_kb: 218 219 This displays the total combined size of all the trace buffers. 220 221 buffer_subbuf_size_kb: 222 223 This sets or displays the sub buffer size. The ring buffer is broken up 224 into several same size "sub buffers". An event can not be bigger than 225 the size of the sub buffer. Normally, the sub buffer is the size of the 226 architecture's page (4K on x86). The sub buffer also contains meta data 227 at the start which also limits the size of an event. That means when 228 the sub buffer is a page size, no event can be larger than the page 229 size minus the sub buffer meta data. 230 231 Note, the buffer_subbuf_size_kb is a way for the user to specify the 232 minimum size of the subbuffer. The kernel may make it bigger due to the 233 implementation details, or simply fail the operation if the kernel can 234 not handle the request. 235 236 Changing the sub buffer size allows for events to be larger than the 237 page size. 238 239 Note: When changing the sub-buffer size, tracing is stopped and any 240 data in the ring buffer and the snapshot buffer will be discarded. 241 242 free_buffer: 243 244 If a process is performing tracing, and the ring buffer should be 245 shrunk "freed" when the process is finished, even if it were to be 246 killed by a signal, this file can be used for that purpose. On close 247 of this file, the ring buffer will be resized to its minimum size. 248 Having a process that is tracing also open this file, when the process 249 exits its file descriptor for this file will be closed, and in doing so, 250 the ring buffer will be "freed". 251 252 It may also stop tracing if disable_on_free option is set. 253 254 tracing_cpumask: 255 256 This is a mask that lets the user only trace on specified CPUs. 257 The format is a hex string representing the CPUs. 258 259 set_ftrace_filter: 260 261 When dynamic ftrace is configured in (see the 262 section below "dynamic ftrace"), the code is dynamically 263 modified (code text rewrite) to disable calling of the 264 function profiler (mcount). This lets tracing be configured 265 in with practically no overhead in performance. This also 266 has a side effect of enabling or disabling specific functions 267 to be traced. Echoing names of functions into this file 268 will limit the trace to only those functions. 269 This influences the tracers "function" and "function_graph" 270 and thus also function profiling (see "function_profile_enabled"). 271 272 The functions listed in "available_filter_functions" are what 273 can be written into this file. 274 275 This interface also allows for commands to be used. See the 276 "Filter commands" section for more details. 277 278 As a speed up, since processing strings can be quite expensive 279 and requires a check of all functions registered to tracing, instead 280 an index can be written into this file. A number (starting with "1") 281 written will instead select the same corresponding at the line position 282 of the "available_filter_functions" file. 283 284 set_ftrace_notrace: 285 286 This has an effect opposite to that of 287 set_ftrace_filter. Any function that is added here will not 288 be traced. If a function exists in both set_ftrace_filter 289 and set_ftrace_notrace, the function will _not_ be traced. 290 291 set_ftrace_pid: 292 293 Have the function tracer only trace the threads whose PID are 294 listed in this file. 295 296 If the "function-fork" option is set, then when a task whose 297 PID is listed in this file forks, the child's PID will 298 automatically be added to this file, and the child will be 299 traced by the function tracer as well. This option will also 300 cause PIDs of tasks that exit to be removed from the file. 301 302 set_ftrace_notrace_pid: 303 304 Have the function tracer ignore threads whose PID are listed in 305 this file. 306 307 If the "function-fork" option is set, then when a task whose 308 PID is listed in this file forks, the child's PID will 309 automatically be added to this file, and the child will not be 310 traced by the function tracer as well. This option will also 311 cause PIDs of tasks that exit to be removed from the file. 312 313 If a PID is in both this file and "set_ftrace_pid", then this 314 file takes precedence, and the thread will not be traced. 315 316 set_event_pid: 317 318 Have the events only trace a task with a PID listed in this file. 319 Note, sched_switch and sched_wake_up will also trace events 320 listed in this file. 321 322 To have the PIDs of children of tasks with their PID in this file 323 added on fork, enable the "event-fork" option. That option will also 324 cause the PIDs of tasks to be removed from this file when the task 325 exits. 326 327 set_event_notrace_pid: 328 329 Have the events not trace a task with a PID listed in this file. 330 Note, sched_switch and sched_wakeup will trace threads not listed 331 in this file, even if a thread's PID is in the file if the 332 sched_switch or sched_wakeup events also trace a thread that should 333 be traced. 334 335 To have the PIDs of children of tasks with their PID in this file 336 added on fork, enable the "event-fork" option. That option will also 337 cause the PIDs of tasks to be removed from this file when the task 338 exits. 339 340 set_graph_function: 341 342 Functions listed in this file will cause the function graph 343 tracer to only trace these functions and the functions that 344 they call. (See the section "dynamic ftrace" for more details). 345 Note, set_ftrace_filter and set_ftrace_notrace still affects 346 what functions are being traced. 347 348 set_graph_notrace: 349 350 Similar to set_graph_function, but will disable function graph 351 tracing when the function is hit until it exits the function. 352 This makes it possible to ignore tracing functions that are called 353 by a specific function. 354 355 available_filter_functions: 356 357 This lists the functions that ftrace has processed and can trace. 358 These are the function names that you can pass to 359 "set_ftrace_filter", "set_ftrace_notrace", 360 "set_graph_function", or "set_graph_notrace". 361 (See the section "dynamic ftrace" below for more details.) 362 363 available_filter_functions_addrs: 364 365 Similar to available_filter_functions, but with address displayed 366 for each function. The displayed address is the patch-site address 367 and can differ from /proc/kallsyms address. 368 369 syscall_user_buf_size: 370 371 Some system call trace events will record the data from a user 372 space address that one of the parameters point to. The amount of 373 data per event is limited. This file holds the max number of bytes 374 that will be recorded into the ring buffer to hold this data. 375 The max value is currently 165. 376 377 dyn_ftrace_total_info: 378 379 This file is for debugging purposes. The number of functions that 380 have been converted to nops and are available to be traced. 381 382 enabled_functions: 383 384 This file is more for debugging ftrace, but can also be useful 385 in seeing if any function has a callback attached to it. 386 Not only does the trace infrastructure use ftrace function 387 trace utility, but other subsystems might too. This file 388 displays all functions that have a callback attached to them 389 as well as the number of callbacks that have been attached. 390 Note, a callback may also call multiple functions which will 391 not be listed in this count. 392 393 If the callback registered to be traced by a function with 394 the "save regs" attribute (thus even more overhead), an 'R' 395 will be displayed on the same line as the function that 396 is returning registers. 397 398 If the callback registered to be traced by a function with 399 the "ip modify" attribute (thus the regs->ip can be changed), 400 an 'I' will be displayed on the same line as the function that 401 can be overridden. 402 403 If a non-ftrace trampoline is attached (BPF) a 'D' will be displayed. 404 Note, normal ftrace trampolines can also be attached, but only one 405 "direct" trampoline can be attached to a given function at a time. 406 407 Some architectures can not call direct trampolines, but instead have 408 the ftrace ops function located above the function entry point. In 409 such cases an 'O' will be displayed. 410 411 If a function had either the "ip modify" or a "direct" call attached to 412 it in the past, a 'M' will be shown. This flag is never cleared. It is 413 used to know if a function was ever modified by the ftrace infrastructure, 414 and can be used for debugging. 415 416 If the architecture supports it, it will also show what callback 417 is being directly called by the function. If the count is greater 418 than 1 it most likely will be ftrace_ops_list_func(). 419 420 If the callback of a function jumps to a trampoline that is 421 specific to the callback and which is not the standard trampoline, 422 its address will be printed as well as the function that the 423 trampoline calls. 424 425 touched_functions: 426 427 This file contains all the functions that ever had a function callback 428 to it via the ftrace infrastructure. It has the same format as 429 enabled_functions but shows all functions that have ever been 430 traced. 431 432 To see any function that has every been modified by "ip modify" or a 433 direct trampoline, one can perform the following command: 434 435 grep ' M ' /sys/kernel/tracing/touched_functions 436 437 function_profile_enabled: 438 439 When set it will enable all functions with either the function 440 tracer, or if configured, the function graph tracer. It will 441 keep a histogram of the number of functions that were called 442 and if the function graph tracer was configured, it will also keep 443 track of the time spent in those functions. The histogram 444 content can be displayed in the files: 445 446 trace_stat/function<cpu> ( function0, function1, etc). 447 448 trace_stat: 449 450 A directory that holds different tracing stats. 451 452 kprobe_events: 453 454 Enable dynamic trace points. See kprobetrace.rst. 455 456 kprobe_profile: 457 458 Dynamic trace points stats. See kprobetrace.rst. 459 460 max_graph_depth: 461 462 Used with the function graph tracer. This is the max depth 463 it will trace into a function. Setting this to a value of 464 one will show only the first kernel function that is called 465 from user space. 466 467 printk_formats: 468 469 This is for tools that read the raw format files. If an event in 470 the ring buffer references a string, only a pointer to the string 471 is recorded into the buffer and not the string itself. This prevents 472 tools from knowing what that string was. This file displays the string 473 and address for the string allowing tools to map the pointers to what 474 the strings were. 475 476 saved_cmdlines: 477 478 Only the pid of the task is recorded in a trace event unless 479 the event specifically saves the task comm as well. Ftrace 480 makes a cache of pid mappings to comms to try to display 481 comms for events. If a pid for a comm is not listed, then 482 "<...>" is displayed in the output. 483 484 If the option "record-cmd" is set to "0", then comms of tasks 485 will not be saved during recording. By default, it is enabled. 486 487 saved_cmdlines_size: 488 489 By default, 128 comms are saved (see "saved_cmdlines" above). To 490 increase or decrease the amount of comms that are cached, echo 491 the number of comms to cache into this file. 492 493 saved_tgids: 494 495 If the option "record-tgid" is set, on each scheduling context switch 496 the Task Group ID of a task is saved in a table mapping the PID of 497 the thread to its TGID. By default, the "record-tgid" option is 498 disabled. 499 500 snapshot: 501 502 This displays the "snapshot" buffer and also lets the user 503 take a snapshot of the current running trace. 504 See the "Snapshot" section below for more details. 505 506 stack_max_size: 507 508 When the stack tracer is activated, this will display the 509 maximum stack size it has encountered. 510 See the "Stack Trace" section below. 511 512 stack_trace: 513 514 This displays the stack back trace of the largest stack 515 that was encountered when the stack tracer is activated. 516 See the "Stack Trace" section below. 517 518 stack_trace_filter: 519 520 This is similar to "set_ftrace_filter" but it limits what 521 functions the stack tracer will check. 522 523 trace_clock: 524 525 Whenever an event is recorded into the ring buffer, a 526 "timestamp" is added. This stamp comes from a specified 527 clock. By default, ftrace uses the "local" clock. This 528 clock is very fast and strictly per CPU, but on some 529 systems it may not be monotonic with respect to other 530 CPUs. In other words, the local clocks may not be in sync 531 with local clocks on other CPUs. 532 533 Usual clocks for tracing:: 534 535 # cat trace_clock 536 [local] global counter x86-tsc 537 538 The clock with the square brackets around it is the one in effect. 539 540 local: 541 Default clock, but may not be in sync across CPUs 542 543 global: 544 This clock is in sync with all CPUs but may 545 be a bit slower than the local clock. 546 547 counter: 548 This is not a clock at all, but literally an atomic 549 counter. It counts up one by one, but is in sync 550 with all CPUs. This is useful when you need to 551 know exactly the order events occurred with respect to 552 each other on different CPUs. 553 554 uptime: 555 This uses the jiffies counter and the time stamp 556 is relative to the time since boot up. 557 558 perf: 559 This makes ftrace use the same clock that perf uses. 560 Eventually perf will be able to read ftrace buffers 561 and this will help out in interleaving the data. 562 563 x86-tsc: 564 Architectures may define their own clocks. For 565 example, x86 uses its own TSC cycle clock here. 566 567 ppc-tb: 568 This uses the powerpc timebase register value. 569 This is in sync across CPUs and can also be used 570 to correlate events across hypervisor/guest if 571 tb_offset is known. 572 573 mono: 574 This uses the fast monotonic clock (CLOCK_MONOTONIC) 575 which is monotonic and is subject to NTP rate adjustments. 576 577 mono_raw: 578 This is the raw monotonic clock (CLOCK_MONOTONIC_RAW) 579 which is monotonic but is not subject to any rate adjustments 580 and ticks at the same rate as the hardware clocksource. 581 582 boot: 583 This is the boot clock (CLOCK_BOOTTIME) and is based on the 584 fast monotonic clock, but also accounts for time spent in 585 suspend. Since the clock access is designed for use in 586 tracing in the suspend path, some side effects are possible 587 if clock is accessed after the suspend time is accounted before 588 the fast mono clock is updated. In this case, the clock update 589 appears to happen slightly sooner than it normally would have. 590 Also on 32-bit systems, it's possible that the 64-bit boot offset 591 sees a partial update. These effects are rare and post 592 processing should be able to handle them. See comments in the 593 ktime_get_boot_fast_ns() function for more information. 594 595 tai: 596 This is the tai clock (CLOCK_TAI) and is derived from the wall- 597 clock time. However, this clock does not experience 598 discontinuities and backwards jumps caused by NTP inserting leap 599 seconds. Since the clock access is designed for use in tracing, 600 side effects are possible. The clock access may yield wrong 601 readouts in case the internal TAI offset is updated e.g., caused 602 by setting the system time or using adjtimex() with an offset. 603 These effects are rare and post processing should be able to 604 handle them. See comments in the ktime_get_tai_fast_ns() 605 function for more information. 606 607 To set a clock, simply echo the clock name into this file:: 608 609 # echo global > trace_clock 610 611 Setting a clock clears the ring buffer content as well as the 612 "snapshot" buffer. 613 614 trace_marker: 615 616 This is a very useful file for synchronizing user space 617 with events happening in the kernel. Writing strings into 618 this file will be written into the ftrace buffer. 619 620 It is useful in applications to open this file at the start 621 of the application and just reference the file descriptor 622 for the file:: 623 624 void trace_write(const char *fmt, ...) 625 { 626 va_list ap; 627 char buf[256]; 628 int n; 629 630 if (trace_fd < 0) 631 return; 632 633 va_start(ap, fmt); 634 n = vsnprintf(buf, 256, fmt, ap); 635 va_end(ap); 636 637 write(trace_fd, buf, n); 638 } 639 640 start:: 641 642 trace_fd = open("trace_marker", O_WRONLY); 643 644 Note: Writing into the trace_marker file can also initiate triggers 645 that are written into /sys/kernel/tracing/events/ftrace/print/trigger 646 See "Event triggers" in Documentation/trace/events.rst and an 647 example in Documentation/trace/histogram.rst (Section 3.) 648 649 trace_marker_raw: 650 651 This is similar to trace_marker above, but is meant for binary data 652 to be written to it, where a tool can be used to parse the data 653 from trace_pipe_raw. 654 655 uprobe_events: 656 657 Add dynamic tracepoints in programs. 658 See uprobetracer.rst 659 660 uprobe_profile: 661 662 Uprobe statistics. See uprobetrace.txt 663 664 instances: 665 666 This is a way to make multiple trace buffers where different 667 events can be recorded in different buffers. 668 See "Instances" section below. 669 670 events: 671 672 This is the trace event directory. It holds event tracepoints 673 (also known as static tracepoints) that have been compiled 674 into the kernel. It shows what event tracepoints exist 675 and how they are grouped by system. There are "enable" 676 files at various levels that can enable the tracepoints 677 when a "1" is written to them. 678 679 See events.rst for more information. 680 681 set_event: 682 683 By echoing in the event into this file, will enable that event. 684 685 See events.rst for more information. 686 687 available_events: 688 689 A list of events that can be enabled in tracing. 690 691 See events.rst for more information. 692 693 timestamp_mode: 694 695 Certain tracers may change the timestamp mode used when 696 logging trace events into the event buffer. Events with 697 different modes can coexist within a buffer but the mode in 698 effect when an event is logged determines which timestamp mode 699 is used for that event. The default timestamp mode is 700 'delta'. 701 702 Usual timestamp modes for tracing: 703 704 # cat timestamp_mode 705 [delta] absolute 706 707 The timestamp mode with the square brackets around it is the 708 one in effect. 709 710 delta: Default timestamp mode - timestamp is a delta against 711 a per-buffer timestamp. 712 713 absolute: The timestamp is a full timestamp, not a delta 714 against some other value. As such it takes up more 715 space and is less efficient. 716 717 hwlat_detector: 718 719 Directory for the Hardware Latency Detector. 720 See "Hardware Latency Detector" section below. 721 722 per_cpu: 723 724 This is a directory that contains the trace per_cpu information. 725 726 per_cpu/cpu0/buffer_size_kb: 727 728 The ftrace buffer is defined per_cpu. That is, there's a separate 729 buffer for each CPU to allow writes to be done atomically, 730 and free from cache bouncing. These buffers may have different 731 size buffers. This file is similar to the buffer_size_kb 732 file, but it only displays or sets the buffer size for the 733 specific CPU. (here cpu0). 734 735 per_cpu/cpu0/trace: 736 737 This is similar to the "trace" file, but it will only display 738 the data specific for the CPU. If written to, it only clears 739 the specific CPU buffer. 740 741 per_cpu/cpu0/trace_pipe 742 743 This is similar to the "trace_pipe" file, and is a consuming 744 read, but it will only display (and consume) the data specific 745 for the CPU. 746 747 per_cpu/cpu0/trace_pipe_raw 748 749 For tools that can parse the ftrace ring buffer binary format, 750 the trace_pipe_raw file can be used to extract the data 751 from the ring buffer directly. With the use of the splice() 752 system call, the buffer data can be quickly transferred to 753 a file or to the network where a server is collecting the 754 data. 755 756 Like trace_pipe, this is a consuming reader, where multiple 757 reads will always produce different data. 758 759 per_cpu/cpu0/snapshot: 760 761 This is similar to the main "snapshot" file, but will only 762 snapshot the current CPU (if supported). It only displays 763 the content of the snapshot for a given CPU, and if 764 written to, only clears this CPU buffer. 765 766 per_cpu/cpu0/snapshot_raw: 767 768 Similar to the trace_pipe_raw, but will read the binary format 769 from the snapshot buffer for the given CPU. 770 771 per_cpu/cpu0/stats: 772 773 This displays certain stats about the ring buffer: 774 775 entries: 776 The number of events that are still in the buffer. 777 778 overrun: 779 The number of lost events due to overwriting when 780 the buffer was full. 781 782 commit overrun: 783 Should always be zero. 784 This gets set if so many events happened within a nested 785 event (ring buffer is re-entrant), that it fills the 786 buffer and starts dropping events. 787 788 bytes: 789 Bytes actually read (not overwritten). 790 791 oldest event ts: 792 The oldest timestamp in the buffer 793 794 now ts: 795 The current timestamp 796 797 dropped events: 798 Events lost due to overwrite option being off. 799 800 read events: 801 The number of events read. 802 803The Tracers 804----------- 805 806Here is the list of current tracers that may be configured. 807 808 "function" 809 810 Function call tracer to trace all kernel functions. 811 812 "function_graph" 813 814 Similar to the function tracer except that the 815 function tracer probes the functions on their entry 816 whereas the function graph tracer traces on both entry 817 and exit of the functions. It then provides the ability 818 to draw a graph of function calls similar to C code 819 source. 820 821 Note that the function graph calculates the timings of when the 822 function starts and returns internally and for each instance. If 823 there are two instances that run function graph tracer and traces 824 the same functions, the length of the timings may be slightly off as 825 each read the timestamp separately and not at the same time. 826 827 "blk" 828 829 The block tracer. The tracer used by the blktrace user 830 application. 831 832 "hwlat" 833 834 The Hardware Latency tracer is used to detect if the hardware 835 produces any latency. See "Hardware Latency Detector" section 836 below. 837 838 "irqsoff" 839 840 Traces the areas that disable interrupts and saves 841 the trace with the longest max latency. 842 See tracing_max_latency. When a new max is recorded, 843 it replaces the old trace. It is best to view this 844 trace with the latency-format option enabled, which 845 happens automatically when the tracer is selected. 846 847 "preemptoff" 848 849 Similar to irqsoff but traces and records the amount of 850 time for which preemption is disabled. 851 852 "preemptirqsoff" 853 854 Similar to irqsoff and preemptoff, but traces and 855 records the largest time for which irqs and/or preemption 856 is disabled. 857 858 "wakeup" 859 860 Traces and records the max latency that it takes for 861 the highest priority task to get scheduled after 862 it has been woken up. 863 Traces all tasks as an average developer would expect. 864 865 "wakeup_rt" 866 867 Traces and records the max latency that it takes for just 868 RT tasks (as the current "wakeup" does). This is useful 869 for those interested in wake up timings of RT tasks. 870 871 "wakeup_dl" 872 873 Traces and records the max latency that it takes for 874 a SCHED_DEADLINE task to be woken (as the "wakeup" and 875 "wakeup_rt" does). 876 877 "mmiotrace" 878 879 A special tracer that is used to trace binary modules. 880 It will trace all the calls that a module makes to the 881 hardware. Everything it writes and reads from the I/O 882 as well. 883 884 "branch" 885 886 This tracer can be configured when tracing likely/unlikely 887 calls within the kernel. It will trace when a likely and 888 unlikely branch is hit and if it was correct in its prediction 889 of being correct. 890 891 "nop" 892 893 This is the "trace nothing" tracer. To remove all 894 tracers from tracing simply echo "nop" into 895 current_tracer. 896 897Error conditions 898---------------- 899 900 For most ftrace commands, failure modes are obvious and communicated 901 using standard return codes. 902 903 For other more involved commands, extended error information may be 904 available via the tracing/error_log file. For the commands that 905 support it, reading the tracing/error_log file after an error will 906 display more detailed information about what went wrong, if 907 information is available. The tracing/error_log file is a circular 908 error log displaying a small number (currently, 8) of ftrace errors 909 for the last (8) failed commands. 910 911 The extended error information and usage takes the form shown in 912 this example:: 913 914 # echo xxx > /sys/kernel/tracing/events/sched/sched_wakeup/trigger 915 echo: write error: Invalid argument 916 917 # cat /sys/kernel/tracing/error_log 918 [ 5348.887237] location: error: Couldn't yyy: zzz 919 Command: xxx 920 ^ 921 [ 7517.023364] location: error: Bad rrr: sss 922 Command: ppp qqq 923 ^ 924 925 To clear the error log, echo the empty string into it:: 926 927 # echo > /sys/kernel/tracing/error_log 928 929Examples of using the tracer 930---------------------------- 931 932Here are typical examples of using the tracers when controlling 933them only with the tracefs interface (without using any 934user-land utilities). 935 936Output format: 937-------------- 938 939Here is an example of the output format of the file "trace":: 940 941 # tracer: function 942 # 943 # entries-in-buffer/entries-written: 140080/250280 #P:4 944 # 945 # _-----=> irqs-off 946 # / _----=> need-resched 947 # | / _---=> hardirq/softirq 948 # || / _--=> preempt-depth 949 # ||| / delay 950 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 951 # | | | |||| | | 952 bash-1977 [000] .... 17284.993652: sys_close <-system_call_fastpath 953 bash-1977 [000] .... 17284.993653: __close_fd <-sys_close 954 bash-1977 [000] .... 17284.993653: _raw_spin_lock <-__close_fd 955 sshd-1974 [003] .... 17284.993653: __srcu_read_unlock <-fsnotify 956 bash-1977 [000] .... 17284.993654: add_preempt_count <-_raw_spin_lock 957 bash-1977 [000] ...1 17284.993655: _raw_spin_unlock <-__close_fd 958 bash-1977 [000] ...1 17284.993656: sub_preempt_count <-_raw_spin_unlock 959 bash-1977 [000] .... 17284.993657: filp_close <-__close_fd 960 bash-1977 [000] .... 17284.993657: dnotify_flush <-filp_close 961 sshd-1974 [003] .... 17284.993658: sys_select <-system_call_fastpath 962 .... 963 964A header is printed with the tracer name that is represented by 965the trace. In this case the tracer is "function". Then it shows the 966number of events in the buffer as well as the total number of entries 967that were written. The difference is the number of entries that were 968lost due to the buffer filling up (250280 - 140080 = 110200 events 969lost). 970 971The header explains the content of the events. Task name "bash", the task 972PID "1977", the CPU that it was running on "000", the latency format 973(explained below), the timestamp in <secs>.<usecs> format, the 974function name that was traced "sys_close" and the parent function that 975called this function "system_call_fastpath". The timestamp is the time 976at which the function was entered. 977 978Latency trace format 979-------------------- 980 981When the latency-format option is enabled or when one of the latency 982tracers is set, the trace file gives somewhat more information to see 983why a latency happened. Here is a typical trace:: 984 985 # tracer: irqsoff 986 # 987 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 988 # -------------------------------------------------------------------- 989 # latency: 259 us, #4/4, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 990 # ----------------- 991 # | task: ps-6143 (uid:0 nice:0 policy:0 rt_prio:0) 992 # ----------------- 993 # => started at: __lock_task_sighand 994 # => ended at: _raw_spin_unlock_irqrestore 995 # 996 # 997 # _------=> CPU# 998 # / _-----=> irqs-off 999 # | / _----=> need-resched 1000 # || / _---=> hardirq/softirq 1001 # ||| / _--=> preempt-depth 1002 # |||| / delay 1003 # cmd pid ||||| time | caller 1004 # \ / ||||| \ | / 1005 ps-6143 2d... 0us!: trace_hardirqs_off <-__lock_task_sighand 1006 ps-6143 2d..1 259us+: trace_hardirqs_on <-_raw_spin_unlock_irqrestore 1007 ps-6143 2d..1 263us+: time_hardirqs_on <-_raw_spin_unlock_irqrestore 1008 ps-6143 2d..1 306us : <stack trace> 1009 => trace_hardirqs_on_caller 1010 => trace_hardirqs_on 1011 => _raw_spin_unlock_irqrestore 1012 => do_task_stat 1013 => proc_tgid_stat 1014 => proc_single_show 1015 => seq_read 1016 => vfs_read 1017 => sys_read 1018 => system_call_fastpath 1019 1020 1021This shows that the current tracer is "irqsoff" tracing the time 1022for which interrupts were disabled. It gives the trace version (which 1023never changes) and the version of the kernel upon which this was executed on 1024(3.8). Then it displays the max latency in microseconds (259 us). The number 1025of trace entries displayed and the total number (both are four: #4/4). 1026VP, KP, SP, and HP are always zero and are reserved for later use. 1027#P is the number of online CPUs (#P:4). 1028 1029The task is the process that was running when the latency 1030occurred. (ps pid: 6143). 1031 1032The start and stop (the functions in which the interrupts were 1033disabled and enabled respectively) that caused the latencies: 1034 1035 - __lock_task_sighand is where the interrupts were disabled. 1036 - _raw_spin_unlock_irqrestore is where they were enabled again. 1037 1038The next lines after the header are the trace itself. The header 1039explains which is which. 1040 1041 cmd: The name of the process in the trace. 1042 1043 pid: The PID of that process. 1044 1045 CPU#: The CPU which the process was running on. 1046 1047 irqs-off: 'd' interrupts are disabled. '.' otherwise. 1048 1049 need-resched: 1050 - 'B' all, TIF_NEED_RESCHED, PREEMPT_NEED_RESCHED and TIF_RESCHED_LAZY is set, 1051 - 'N' both TIF_NEED_RESCHED and PREEMPT_NEED_RESCHED is set, 1052 - 'n' only TIF_NEED_RESCHED is set, 1053 - 'p' only PREEMPT_NEED_RESCHED is set, 1054 - 'L' both PREEMPT_NEED_RESCHED and TIF_RESCHED_LAZY is set, 1055 - 'b' both TIF_NEED_RESCHED and TIF_RESCHED_LAZY is set, 1056 - 'l' only TIF_RESCHED_LAZY is set 1057 - '.' otherwise. 1058 1059 hardirq/softirq: 1060 - 'Z' - NMI occurred inside a hardirq 1061 - 'z' - NMI is running 1062 - 'H' - hard irq occurred inside a softirq. 1063 - 'h' - hard irq is running 1064 - 's' - soft irq is running 1065 - '.' - normal context. 1066 1067 preempt-depth: The level of preempt_disabled 1068 1069The above is mostly meaningful for kernel developers. 1070 1071 time: 1072 When the latency-format option is enabled, the trace file 1073 output includes a timestamp relative to the start of the 1074 trace. This differs from the output when latency-format 1075 is disabled, which includes an absolute timestamp. 1076 1077 delay: 1078 This is just to help catch your eye a bit better. And 1079 needs to be fixed to be only relative to the same CPU. 1080 The marks are determined by the difference between this 1081 current trace and the next trace. 1082 1083 - '$' - greater than 1 second 1084 - '@' - greater than 100 millisecond 1085 - '*' - greater than 10 millisecond 1086 - '#' - greater than 1000 microsecond 1087 - '!' - greater than 100 microsecond 1088 - '+' - greater than 10 microsecond 1089 - ' ' - less than or equal to 10 microsecond. 1090 1091 The rest is the same as the 'trace' file. 1092 1093 Note, the latency tracers will usually end with a back trace 1094 to easily find where the latency occurred. 1095 1096trace_options 1097------------- 1098 1099The trace_options file (or the options directory) is used to control 1100what gets printed in the trace output, or manipulate the tracers. 1101To see what is available, simply cat the file:: 1102 1103 cat trace_options 1104 print-parent 1105 nosym-offset 1106 nosym-addr 1107 noverbose 1108 noraw 1109 nohex 1110 nobin 1111 noblock 1112 nofields 1113 trace_printk 1114 annotate 1115 nouserstacktrace 1116 nosym-userobj 1117 noprintk-msg-only 1118 context-info 1119 nolatency-format 1120 record-cmd 1121 norecord-tgid 1122 overwrite 1123 nodisable_on_free 1124 irq-info 1125 markers 1126 noevent-fork 1127 function-trace 1128 nofunction-fork 1129 nodisplay-graph 1130 nostacktrace 1131 nobranch 1132 1133To disable one of the options, echo in the option prepended with 1134"no":: 1135 1136 echo noprint-parent > trace_options 1137 1138To enable an option, leave off the "no":: 1139 1140 echo sym-offset > trace_options 1141 1142Here are the available options: 1143 1144 print-parent 1145 On function traces, display the calling (parent) 1146 function as well as the function being traced. 1147 :: 1148 1149 print-parent: 1150 bash-4000 [01] 1477.606694: simple_strtoul <-kstrtoul 1151 1152 noprint-parent: 1153 bash-4000 [01] 1477.606694: simple_strtoul 1154 1155 1156 sym-offset 1157 Display not only the function name, but also the 1158 offset in the function. For example, instead of 1159 seeing just "ktime_get", you will see 1160 "ktime_get+0xb/0x20". 1161 :: 1162 1163 sym-offset: 1164 bash-4000 [01] 1477.606694: simple_strtoul+0x6/0xa0 1165 1166 sym-addr 1167 This will also display the function address as well 1168 as the function name. 1169 :: 1170 1171 sym-addr: 1172 bash-4000 [01] 1477.606694: simple_strtoul <c0339346> 1173 1174 verbose 1175 This deals with the trace file when the 1176 latency-format option is enabled. 1177 :: 1178 1179 bash 4000 1 0 00000000 00010a95 [58127d26] 1720.415ms \ 1180 (+0.000ms): simple_strtoul (kstrtoul) 1181 1182 raw 1183 This will display raw numbers. This option is best for 1184 use with user applications that can translate the raw 1185 numbers better than having it done in the kernel. 1186 1187 hex 1188 Similar to raw, but the numbers will be in a hexadecimal format. 1189 1190 bin 1191 This will print out the formats in raw binary. 1192 1193 block 1194 When set, reading trace_pipe will not block when polled. 1195 1196 fields 1197 Print the fields as described by their types. This is a better 1198 option than using hex, bin or raw, as it gives a better parsing 1199 of the content of the event. 1200 1201 trace_printk 1202 Can disable trace_printk() from writing into the buffer. 1203 1204 trace_printk_dest 1205 Set to have trace_printk() and similar internal tracing functions 1206 write into this instance. Note, only one trace instance can have 1207 this set. By setting this flag, it clears the trace_printk_dest flag 1208 of the instance that had it set previously. By default, the top 1209 level trace has this set, and will get it set again if another 1210 instance has it set then clears it. 1211 1212 This flag cannot be cleared by the top level instance, as it is the 1213 default instance. The only way the top level instance has this flag 1214 cleared, is by it being set in another instance. 1215 1216 copy_trace_marker 1217 If there are applications that hard code writing into the top level 1218 trace_marker file (/sys/kernel/tracing/trace_marker or trace_marker_raw), 1219 and the tooling would like it to go into an instance, this option can 1220 be used. Create an instance and set this option, and then all writes 1221 into the top level trace_marker file will also be redirected into this 1222 instance. 1223 1224 Note, by default this option is set for the top level instance. If it 1225 is disabled, then writes to the trace_marker or trace_marker_raw files 1226 will not be written into the top level file. If no instance has this 1227 option set, then a write will error with the errno of ENODEV. 1228 1229 annotate 1230 It is sometimes confusing when the CPU buffers are full 1231 and one CPU buffer had a lot of events recently, thus 1232 a shorter time frame, were another CPU may have only had 1233 a few events, which lets it have older events. When 1234 the trace is reported, it shows the oldest events first, 1235 and it may look like only one CPU ran (the one with the 1236 oldest events). When the annotate option is set, it will 1237 display when a new CPU buffer started:: 1238 1239 <idle>-0 [001] dNs4 21169.031481: wake_up_idle_cpu <-add_timer_on 1240 <idle>-0 [001] dNs4 21169.031482: _raw_spin_unlock_irqrestore <-add_timer_on 1241 <idle>-0 [001] .Ns4 21169.031484: sub_preempt_count <-_raw_spin_unlock_irqrestore 1242 ##### CPU 2 buffer started #### 1243 <idle>-0 [002] .N.1 21169.031484: rcu_idle_exit <-cpu_idle 1244 <idle>-0 [001] .Ns3 21169.031484: _raw_spin_unlock <-clocksource_watchdog 1245 <idle>-0 [001] .Ns3 21169.031485: sub_preempt_count <-_raw_spin_unlock 1246 1247 userstacktrace 1248 This option changes the trace. It records a 1249 stacktrace of the current user space thread after 1250 each trace event. 1251 1252 sym-userobj 1253 when user stacktrace are enabled, look up which 1254 object the address belongs to, and print a 1255 relative address. This is especially useful when 1256 ASLR is on, otherwise you don't get a chance to 1257 resolve the address to object/file/line after 1258 the app is no longer running 1259 1260 The lookup is performed when you read 1261 trace,trace_pipe. Example:: 1262 1263 a.out-1623 [000] 40874.465068: /root/a.out[+0x480] <-/root/a.out[+0 1264 x494] <- /root/a.out[+0x4a8] <- /lib/libc-2.7.so[+0x1e1a6] 1265 1266 1267 printk-msg-only 1268 When set, trace_printk()s will only show the format 1269 and not their parameters (if trace_bprintk() or 1270 trace_bputs() was used to save the trace_printk()). 1271 1272 context-info 1273 Show only the event data. Hides the comm, PID, 1274 timestamp, CPU, and other useful data. 1275 1276 latency-format 1277 This option changes the trace output. When it is enabled, 1278 the trace displays additional information about the 1279 latency, as described in "Latency trace format". 1280 1281 pause-on-trace 1282 When set, opening the trace file for read, will pause 1283 writing to the ring buffer (as if tracing_on was set to zero). 1284 This simulates the original behavior of the trace file. 1285 When the file is closed, tracing will be enabled again. 1286 1287 hash-ptr 1288 When set, "%p" in the event printk format displays the 1289 hashed pointer value instead of real address. 1290 This will be useful if you want to find out which hashed 1291 value is corresponding to the real value in trace log. 1292 1293 record-cmd 1294 When any event or tracer is enabled, a hook is enabled 1295 in the sched_switch trace point to fill comm cache 1296 with mapped pids and comms. But this may cause some 1297 overhead, and if you only care about pids, and not the 1298 name of the task, disabling this option can lower the 1299 impact of tracing. See "saved_cmdlines". 1300 1301 record-tgid 1302 When any event or tracer is enabled, a hook is enabled 1303 in the sched_switch trace point to fill the cache of 1304 mapped Thread Group IDs (TGID) mapping to pids. See 1305 "saved_tgids". 1306 1307 overwrite 1308 This controls what happens when the trace buffer is 1309 full. If "1" (default), the oldest events are 1310 discarded and overwritten. If "0", then the newest 1311 events are discarded. 1312 (see per_cpu/cpu0/stats for overrun and dropped) 1313 1314 disable_on_free 1315 When the free_buffer is closed, tracing will 1316 stop (tracing_on set to 0). 1317 1318 irq-info 1319 Shows the interrupt, preempt count, need resched data. 1320 When disabled, the trace looks like:: 1321 1322 # tracer: function 1323 # 1324 # entries-in-buffer/entries-written: 144405/9452052 #P:4 1325 # 1326 # TASK-PID CPU# TIMESTAMP FUNCTION 1327 # | | | | | 1328 <idle>-0 [002] 23636.756054: ttwu_do_activate.constprop.89 <-try_to_wake_up 1329 <idle>-0 [002] 23636.756054: activate_task <-ttwu_do_activate.constprop.89 1330 <idle>-0 [002] 23636.756055: enqueue_task <-activate_task 1331 1332 1333 markers 1334 When set, the trace_marker is writable (only by root). 1335 When disabled, the trace_marker will error with EINVAL 1336 on write. 1337 1338 event-fork 1339 When set, tasks with PIDs listed in set_event_pid will have 1340 the PIDs of their children added to set_event_pid when those 1341 tasks fork. Also, when tasks with PIDs in set_event_pid exit, 1342 their PIDs will be removed from the file. 1343 1344 This affects PIDs listed in set_event_notrace_pid as well. 1345 1346 function-trace 1347 The latency tracers will enable function tracing 1348 if this option is enabled (default it is). When 1349 it is disabled, the latency tracers do not trace 1350 functions. This keeps the overhead of the tracer down 1351 when performing latency tests. 1352 1353 function-fork 1354 When set, tasks with PIDs listed in set_ftrace_pid will 1355 have the PIDs of their children added to set_ftrace_pid 1356 when those tasks fork. Also, when tasks with PIDs in 1357 set_ftrace_pid exit, their PIDs will be removed from the 1358 file. 1359 1360 This affects PIDs in set_ftrace_notrace_pid as well. 1361 1362 display-graph 1363 When set, the latency tracers (irqsoff, wakeup, etc) will 1364 use function graph tracing instead of function tracing. 1365 1366 stacktrace 1367 When set, a stack trace is recorded after any trace event 1368 is recorded. 1369 1370 branch 1371 Enable branch tracing with the tracer. This enables branch 1372 tracer along with the currently set tracer. Enabling this 1373 with the "nop" tracer is the same as just enabling the 1374 "branch" tracer. 1375 1376.. tip:: Some tracers have their own options. They only appear in this 1377 file when the tracer is active. They always appear in the 1378 options directory. 1379 1380 1381Here are the per tracer options: 1382 1383Options for function tracer: 1384 1385 func_stack_trace 1386 When set, a stack trace is recorded after every 1387 function that is recorded. NOTE! Limit the functions 1388 that are recorded before enabling this, with 1389 "set_ftrace_filter" otherwise the system performance 1390 will be critically degraded. Remember to disable 1391 this option before clearing the function filter. 1392 1393Options for function_graph tracer: 1394 1395 Since the function_graph tracer has a slightly different output 1396 it has its own options to control what is displayed. 1397 1398 funcgraph-overrun 1399 When set, the "overrun" of the graph stack is 1400 displayed after each function traced. The 1401 overrun, is when the stack depth of the calls 1402 is greater than what is reserved for each task. 1403 Each task has a fixed array of functions to 1404 trace in the call graph. If the depth of the 1405 calls exceeds that, the function is not traced. 1406 The overrun is the number of functions missed 1407 due to exceeding this array. 1408 1409 funcgraph-cpu 1410 When set, the CPU number of the CPU where the trace 1411 occurred is displayed. 1412 1413 funcgraph-overhead 1414 When set, if the function takes longer than 1415 A certain amount, then a delay marker is 1416 displayed. See "delay" above, under the 1417 header description. 1418 1419 funcgraph-proc 1420 Unlike other tracers, the process' command line 1421 is not displayed by default, but instead only 1422 when a task is traced in and out during a context 1423 switch. Enabling this options has the command 1424 of each process displayed at every line. 1425 1426 funcgraph-duration 1427 At the end of each function (the return) 1428 the duration of the amount of time in the 1429 function is displayed in microseconds. 1430 1431 funcgraph-abstime 1432 When set, the timestamp is displayed at each line. 1433 1434 funcgraph-irqs 1435 When disabled, functions that happen inside an 1436 interrupt will not be traced. 1437 1438 funcgraph-tail 1439 When set, the return event will include the function 1440 that it represents. By default this is off, and 1441 only a closing curly bracket "}" is displayed for 1442 the return of a function. 1443 1444 funcgraph-retval 1445 When set, the return value of each traced function 1446 will be printed after an equal sign "=". By default 1447 this is off. 1448 1449 funcgraph-retval-hex 1450 When set, the return value will always be printed 1451 in hexadecimal format. If the option is not set and 1452 the return value is an error code, it will be printed 1453 in signed decimal format; otherwise it will also be 1454 printed in hexadecimal format. By default, this option 1455 is off. 1456 1457 sleep-time 1458 When running function graph tracer, to include 1459 the time a task schedules out in its function. 1460 When enabled, it will account time the task has been 1461 scheduled out as part of the function call. 1462 1463 graph-time 1464 When running function profiler with function graph tracer, 1465 to include the time to call nested functions. When this is 1466 not set, the time reported for the function will only 1467 include the time the function itself executed for, not the 1468 time for functions that it called. 1469 1470Options for blk tracer: 1471 1472 blk_classic 1473 Shows a more minimalistic output. 1474 1475 1476irqsoff 1477------- 1478 1479When interrupts are disabled, the CPU can not react to any other 1480external event (besides NMIs and SMIs). This prevents the timer 1481interrupt from triggering or the mouse interrupt from letting 1482the kernel know of a new mouse event. The result is a latency 1483with the reaction time. 1484 1485The irqsoff tracer tracks the time for which interrupts are 1486disabled. When a new maximum latency is hit, the tracer saves 1487the trace leading up to that latency point so that every time a 1488new maximum is reached, the old saved trace is discarded and the 1489new trace is saved. 1490 1491To reset the maximum, echo 0 into tracing_max_latency. Here is 1492an example:: 1493 1494 # echo 0 > options/function-trace 1495 # echo irqsoff > current_tracer 1496 # echo 1 > tracing_on 1497 # echo 0 > tracing_max_latency 1498 # ls -ltr 1499 [...] 1500 # echo 0 > tracing_on 1501 # cat trace 1502 # tracer: irqsoff 1503 # 1504 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 1505 # -------------------------------------------------------------------- 1506 # latency: 16 us, #4/4, CPU#0 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1507 # ----------------- 1508 # | task: swapper/0-0 (uid:0 nice:0 policy:0 rt_prio:0) 1509 # ----------------- 1510 # => started at: run_timer_softirq 1511 # => ended at: run_timer_softirq 1512 # 1513 # 1514 # _------=> CPU# 1515 # / _-----=> irqs-off 1516 # | / _----=> need-resched 1517 # || / _---=> hardirq/softirq 1518 # ||| / _--=> preempt-depth 1519 # |||| / delay 1520 # cmd pid ||||| time | caller 1521 # \ / ||||| \ | / 1522 <idle>-0 0d.s2 0us+: _raw_spin_lock_irq <-run_timer_softirq 1523 <idle>-0 0dNs3 17us : _raw_spin_unlock_irq <-run_timer_softirq 1524 <idle>-0 0dNs3 17us+: trace_hardirqs_on <-run_timer_softirq 1525 <idle>-0 0dNs3 25us : <stack trace> 1526 => _raw_spin_unlock_irq 1527 => run_timer_softirq 1528 => __do_softirq 1529 => call_softirq 1530 => do_softirq 1531 => irq_exit 1532 => smp_apic_timer_interrupt 1533 => apic_timer_interrupt 1534 => rcu_idle_exit 1535 => cpu_idle 1536 => rest_init 1537 => start_kernel 1538 => x86_64_start_reservations 1539 => x86_64_start_kernel 1540 1541Here we see that we had a latency of 16 microseconds (which is 1542very good). The _raw_spin_lock_irq in run_timer_softirq disabled 1543interrupts. The difference between the 16 and the displayed 1544timestamp 25us occurred because the clock was incremented 1545between the time of recording the max latency and the time of 1546recording the function that had that latency. 1547 1548Note the above example had function-trace not set. If we set 1549function-trace, we get a much larger output:: 1550 1551 with echo 1 > options/function-trace 1552 1553 # tracer: irqsoff 1554 # 1555 # irqsoff latency trace v1.1.5 on 3.8.0-test+ 1556 # -------------------------------------------------------------------- 1557 # latency: 71 us, #168/168, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1558 # ----------------- 1559 # | task: bash-2042 (uid:0 nice:0 policy:0 rt_prio:0) 1560 # ----------------- 1561 # => started at: ata_scsi_queuecmd 1562 # => ended at: ata_scsi_queuecmd 1563 # 1564 # 1565 # _------=> CPU# 1566 # / _-----=> irqs-off 1567 # | / _----=> need-resched 1568 # || / _---=> hardirq/softirq 1569 # ||| / _--=> preempt-depth 1570 # |||| / delay 1571 # cmd pid ||||| time | caller 1572 # \ / ||||| \ | / 1573 bash-2042 3d... 0us : _raw_spin_lock_irqsave <-ata_scsi_queuecmd 1574 bash-2042 3d... 0us : add_preempt_count <-_raw_spin_lock_irqsave 1575 bash-2042 3d..1 1us : ata_scsi_find_dev <-ata_scsi_queuecmd 1576 bash-2042 3d..1 1us : __ata_scsi_find_dev <-ata_scsi_find_dev 1577 bash-2042 3d..1 2us : ata_find_dev.part.14 <-__ata_scsi_find_dev 1578 bash-2042 3d..1 2us : ata_qc_new_init <-__ata_scsi_queuecmd 1579 bash-2042 3d..1 3us : ata_sg_init <-__ata_scsi_queuecmd 1580 bash-2042 3d..1 4us : ata_scsi_rw_xlat <-__ata_scsi_queuecmd 1581 bash-2042 3d..1 4us : ata_build_rw_tf <-ata_scsi_rw_xlat 1582 [...] 1583 bash-2042 3d..1 67us : delay_tsc <-__delay 1584 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc 1585 bash-2042 3d..2 67us : sub_preempt_count <-delay_tsc 1586 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc 1587 bash-2042 3d..2 68us : sub_preempt_count <-delay_tsc 1588 bash-2042 3d..1 68us+: ata_bmdma_start <-ata_bmdma_qc_issue 1589 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1590 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1591 bash-2042 3d..1 72us+: trace_hardirqs_on <-ata_scsi_queuecmd 1592 bash-2042 3d..1 120us : <stack trace> 1593 => _raw_spin_unlock_irqrestore 1594 => ata_scsi_queuecmd 1595 => scsi_dispatch_cmd 1596 => scsi_request_fn 1597 => __blk_run_queue_uncond 1598 => __blk_run_queue 1599 => blk_queue_bio 1600 => submit_bio_noacct 1601 => submit_bio 1602 => submit_bh 1603 => __ext3_get_inode_loc 1604 => ext3_iget 1605 => ext3_lookup 1606 => lookup_real 1607 => __lookup_hash 1608 => walk_component 1609 => lookup_last 1610 => path_lookupat 1611 => filename_lookup 1612 => user_path_at_empty 1613 => user_path_at 1614 => vfs_fstatat 1615 => vfs_stat 1616 => sys_newstat 1617 => system_call_fastpath 1618 1619 1620Here we traced a 71 microsecond latency. But we also see all the 1621functions that were called during that time. Note that by 1622enabling function tracing, we incur an added overhead. This 1623overhead may extend the latency times. But nevertheless, this 1624trace has provided some very helpful debugging information. 1625 1626If we prefer function graph output instead of function, we can set 1627display-graph option:: 1628 1629 with echo 1 > options/display-graph 1630 1631 # tracer: irqsoff 1632 # 1633 # irqsoff latency trace v1.1.5 on 4.20.0-rc6+ 1634 # -------------------------------------------------------------------- 1635 # latency: 3751 us, #274/274, CPU#0 | (M:desktop VP:0, KP:0, SP:0 HP:0 #P:4) 1636 # ----------------- 1637 # | task: bash-1507 (uid:0 nice:0 policy:0 rt_prio:0) 1638 # ----------------- 1639 # => started at: free_debug_processing 1640 # => ended at: return_to_handler 1641 # 1642 # 1643 # _-----=> irqs-off 1644 # / _----=> need-resched 1645 # | / _---=> hardirq/softirq 1646 # || / _--=> preempt-depth 1647 # ||| / 1648 # REL TIME CPU TASK/PID |||| DURATION FUNCTION CALLS 1649 # | | | | |||| | | | | | | 1650 0 us | 0) bash-1507 | d... | 0.000 us | _raw_spin_lock_irqsave(); 1651 0 us | 0) bash-1507 | d..1 | 0.378 us | do_raw_spin_trylock(); 1652 1 us | 0) bash-1507 | d..2 | | set_track() { 1653 2 us | 0) bash-1507 | d..2 | | save_stack_trace() { 1654 2 us | 0) bash-1507 | d..2 | | __save_stack_trace() { 1655 3 us | 0) bash-1507 | d..2 | | __unwind_start() { 1656 3 us | 0) bash-1507 | d..2 | | get_stack_info() { 1657 3 us | 0) bash-1507 | d..2 | 0.351 us | in_task_stack(); 1658 4 us | 0) bash-1507 | d..2 | 1.107 us | } 1659 [...] 1660 3750 us | 0) bash-1507 | d..1 | 0.516 us | do_raw_spin_unlock(); 1661 3750 us | 0) bash-1507 | d..1 | 0.000 us | _raw_spin_unlock_irqrestore(); 1662 3764 us | 0) bash-1507 | d..1 | 0.000 us | tracer_hardirqs_on(); 1663 bash-1507 0d..1 3792us : <stack trace> 1664 => free_debug_processing 1665 => __slab_free 1666 => kmem_cache_free 1667 => vm_area_free 1668 => remove_vma 1669 => exit_mmap 1670 => mmput 1671 => begin_new_exec 1672 => load_elf_binary 1673 => search_binary_handler 1674 => __do_execve_file.isra.32 1675 => __x64_sys_execve 1676 => do_syscall_64 1677 => entry_SYSCALL_64_after_hwframe 1678 1679preemptoff 1680---------- 1681 1682When preemption is disabled, we may be able to receive 1683interrupts but the task cannot be preempted and a higher 1684priority task must wait for preemption to be enabled again 1685before it can preempt a lower priority task. 1686 1687The preemptoff tracer traces the places that disable preemption. 1688Like the irqsoff tracer, it records the maximum latency for 1689which preemption was disabled. The control of preemptoff tracer 1690is much like the irqsoff tracer. 1691:: 1692 1693 # echo 0 > options/function-trace 1694 # echo preemptoff > current_tracer 1695 # echo 1 > tracing_on 1696 # echo 0 > tracing_max_latency 1697 # ls -ltr 1698 [...] 1699 # echo 0 > tracing_on 1700 # cat trace 1701 # tracer: preemptoff 1702 # 1703 # preemptoff latency trace v1.1.5 on 3.8.0-test+ 1704 # -------------------------------------------------------------------- 1705 # latency: 46 us, #4/4, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1706 # ----------------- 1707 # | task: sshd-1991 (uid:0 nice:0 policy:0 rt_prio:0) 1708 # ----------------- 1709 # => started at: do_IRQ 1710 # => ended at: do_IRQ 1711 # 1712 # 1713 # _------=> CPU# 1714 # / _-----=> irqs-off 1715 # | / _----=> need-resched 1716 # || / _---=> hardirq/softirq 1717 # ||| / _--=> preempt-depth 1718 # |||| / delay 1719 # cmd pid ||||| time | caller 1720 # \ / ||||| \ | / 1721 sshd-1991 1d.h. 0us+: irq_enter <-do_IRQ 1722 sshd-1991 1d..1 46us : irq_exit <-do_IRQ 1723 sshd-1991 1d..1 47us+: trace_preempt_on <-do_IRQ 1724 sshd-1991 1d..1 52us : <stack trace> 1725 => sub_preempt_count 1726 => irq_exit 1727 => do_IRQ 1728 => ret_from_intr 1729 1730 1731This has some more changes. Preemption was disabled when an 1732interrupt came in (notice the 'h'), and was enabled on exit. 1733But we also see that interrupts have been disabled when entering 1734the preempt off section and leaving it (the 'd'). We do not know if 1735interrupts were enabled in the mean time or shortly after this 1736was over. 1737:: 1738 1739 # tracer: preemptoff 1740 # 1741 # preemptoff latency trace v1.1.5 on 3.8.0-test+ 1742 # -------------------------------------------------------------------- 1743 # latency: 83 us, #241/241, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1744 # ----------------- 1745 # | task: bash-1994 (uid:0 nice:0 policy:0 rt_prio:0) 1746 # ----------------- 1747 # => started at: wake_up_new_task 1748 # => ended at: task_rq_unlock 1749 # 1750 # 1751 # _------=> CPU# 1752 # / _-----=> irqs-off 1753 # | / _----=> need-resched 1754 # || / _---=> hardirq/softirq 1755 # ||| / _--=> preempt-depth 1756 # |||| / delay 1757 # cmd pid ||||| time | caller 1758 # \ / ||||| \ | / 1759 bash-1994 1d..1 0us : _raw_spin_lock_irqsave <-wake_up_new_task 1760 bash-1994 1d..1 0us : select_task_rq_fair <-select_task_rq 1761 bash-1994 1d..1 1us : __rcu_read_lock <-select_task_rq_fair 1762 bash-1994 1d..1 1us : source_load <-select_task_rq_fair 1763 bash-1994 1d..1 1us : source_load <-select_task_rq_fair 1764 [...] 1765 bash-1994 1d..1 12us : irq_enter <-smp_apic_timer_interrupt 1766 bash-1994 1d..1 12us : rcu_irq_enter <-irq_enter 1767 bash-1994 1d..1 13us : add_preempt_count <-irq_enter 1768 bash-1994 1d.h1 13us : exit_idle <-smp_apic_timer_interrupt 1769 bash-1994 1d.h1 13us : hrtimer_interrupt <-smp_apic_timer_interrupt 1770 bash-1994 1d.h1 13us : _raw_spin_lock <-hrtimer_interrupt 1771 bash-1994 1d.h1 14us : add_preempt_count <-_raw_spin_lock 1772 bash-1994 1d.h2 14us : ktime_get_update_offsets <-hrtimer_interrupt 1773 [...] 1774 bash-1994 1d.h1 35us : lapic_next_event <-clockevents_program_event 1775 bash-1994 1d.h1 35us : irq_exit <-smp_apic_timer_interrupt 1776 bash-1994 1d.h1 36us : sub_preempt_count <-irq_exit 1777 bash-1994 1d..2 36us : do_softirq <-irq_exit 1778 bash-1994 1d..2 36us : __do_softirq <-call_softirq 1779 bash-1994 1d..2 36us : __local_bh_disable <-__do_softirq 1780 bash-1994 1d.s2 37us : add_preempt_count <-_raw_spin_lock_irq 1781 bash-1994 1d.s3 38us : _raw_spin_unlock <-run_timer_softirq 1782 bash-1994 1d.s3 39us : sub_preempt_count <-_raw_spin_unlock 1783 bash-1994 1d.s2 39us : call_timer_fn <-run_timer_softirq 1784 [...] 1785 bash-1994 1dNs2 81us : cpu_needs_another_gp <-rcu_process_callbacks 1786 bash-1994 1dNs2 82us : __local_bh_enable <-__do_softirq 1787 bash-1994 1dNs2 82us : sub_preempt_count <-__local_bh_enable 1788 bash-1994 1dN.2 82us : idle_cpu <-irq_exit 1789 bash-1994 1dN.2 83us : rcu_irq_exit <-irq_exit 1790 bash-1994 1dN.2 83us : sub_preempt_count <-irq_exit 1791 bash-1994 1.N.1 84us : _raw_spin_unlock_irqrestore <-task_rq_unlock 1792 bash-1994 1.N.1 84us+: trace_preempt_on <-task_rq_unlock 1793 bash-1994 1.N.1 104us : <stack trace> 1794 => sub_preempt_count 1795 => _raw_spin_unlock_irqrestore 1796 => task_rq_unlock 1797 => wake_up_new_task 1798 => do_fork 1799 => sys_clone 1800 => stub_clone 1801 1802 1803The above is an example of the preemptoff trace with 1804function-trace set. Here we see that interrupts were not disabled 1805the entire time. The irq_enter code lets us know that we entered 1806an interrupt 'h'. Before that, the functions being traced still 1807show that it is not in an interrupt, but we can see from the 1808functions themselves that this is not the case. 1809 1810preemptirqsoff 1811-------------- 1812 1813Knowing the locations that have interrupts disabled or 1814preemption disabled for the longest times is helpful. But 1815sometimes we would like to know when either preemption and/or 1816interrupts are disabled. 1817 1818Consider the following code:: 1819 1820 local_irq_disable(); 1821 call_function_with_irqs_off(); 1822 preempt_disable(); 1823 call_function_with_irqs_and_preemption_off(); 1824 local_irq_enable(); 1825 call_function_with_preemption_off(); 1826 preempt_enable(); 1827 1828The irqsoff tracer will record the total length of 1829call_function_with_irqs_off() and 1830call_function_with_irqs_and_preemption_off(). 1831 1832The preemptoff tracer will record the total length of 1833call_function_with_irqs_and_preemption_off() and 1834call_function_with_preemption_off(). 1835 1836But neither will trace the time that interrupts and/or 1837preemption is disabled. This total time is the time that we can 1838not schedule. To record this time, use the preemptirqsoff 1839tracer. 1840 1841Again, using this trace is much like the irqsoff and preemptoff 1842tracers. 1843:: 1844 1845 # echo 0 > options/function-trace 1846 # echo preemptirqsoff > current_tracer 1847 # echo 1 > tracing_on 1848 # echo 0 > tracing_max_latency 1849 # ls -ltr 1850 [...] 1851 # echo 0 > tracing_on 1852 # cat trace 1853 # tracer: preemptirqsoff 1854 # 1855 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+ 1856 # -------------------------------------------------------------------- 1857 # latency: 100 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1858 # ----------------- 1859 # | task: ls-2230 (uid:0 nice:0 policy:0 rt_prio:0) 1860 # ----------------- 1861 # => started at: ata_scsi_queuecmd 1862 # => ended at: ata_scsi_queuecmd 1863 # 1864 # 1865 # _------=> CPU# 1866 # / _-----=> irqs-off 1867 # | / _----=> need-resched 1868 # || / _---=> hardirq/softirq 1869 # ||| / _--=> preempt-depth 1870 # |||| / delay 1871 # cmd pid ||||| time | caller 1872 # \ / ||||| \ | / 1873 ls-2230 3d... 0us+: _raw_spin_lock_irqsave <-ata_scsi_queuecmd 1874 ls-2230 3...1 100us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd 1875 ls-2230 3...1 101us+: trace_preempt_on <-ata_scsi_queuecmd 1876 ls-2230 3...1 111us : <stack trace> 1877 => sub_preempt_count 1878 => _raw_spin_unlock_irqrestore 1879 => ata_scsi_queuecmd 1880 => scsi_dispatch_cmd 1881 => scsi_request_fn 1882 => __blk_run_queue_uncond 1883 => __blk_run_queue 1884 => blk_queue_bio 1885 => submit_bio_noacct 1886 => submit_bio 1887 => submit_bh 1888 => ext3_bread 1889 => ext3_dir_bread 1890 => htree_dirblock_to_tree 1891 => ext3_htree_fill_tree 1892 => ext3_readdir 1893 => vfs_readdir 1894 => sys_getdents 1895 => system_call_fastpath 1896 1897 1898The trace_hardirqs_off_thunk is called from assembly on x86 when 1899interrupts are disabled in the assembly code. Without the 1900function tracing, we do not know if interrupts were enabled 1901within the preemption points. We do see that it started with 1902preemption enabled. 1903 1904Here is a trace with function-trace set:: 1905 1906 # tracer: preemptirqsoff 1907 # 1908 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+ 1909 # -------------------------------------------------------------------- 1910 # latency: 161 us, #339/339, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 1911 # ----------------- 1912 # | task: ls-2269 (uid:0 nice:0 policy:0 rt_prio:0) 1913 # ----------------- 1914 # => started at: schedule 1915 # => ended at: mutex_unlock 1916 # 1917 # 1918 # _------=> CPU# 1919 # / _-----=> irqs-off 1920 # | / _----=> need-resched 1921 # || / _---=> hardirq/softirq 1922 # ||| / _--=> preempt-depth 1923 # |||| / delay 1924 # cmd pid ||||| time | caller 1925 # \ / ||||| \ | / 1926 kworker/-59 3...1 0us : __schedule <-schedule 1927 kworker/-59 3d..1 0us : rcu_preempt_qs <-rcu_note_context_switch 1928 kworker/-59 3d..1 1us : add_preempt_count <-_raw_spin_lock_irq 1929 kworker/-59 3d..2 1us : deactivate_task <-__schedule 1930 kworker/-59 3d..2 1us : dequeue_task <-deactivate_task 1931 kworker/-59 3d..2 2us : update_rq_clock <-dequeue_task 1932 kworker/-59 3d..2 2us : dequeue_task_fair <-dequeue_task 1933 kworker/-59 3d..2 2us : update_curr <-dequeue_task_fair 1934 kworker/-59 3d..2 2us : update_min_vruntime <-update_curr 1935 kworker/-59 3d..2 3us : cpuacct_charge <-update_curr 1936 kworker/-59 3d..2 3us : __rcu_read_lock <-cpuacct_charge 1937 kworker/-59 3d..2 3us : __rcu_read_unlock <-cpuacct_charge 1938 kworker/-59 3d..2 3us : update_cfs_rq_blocked_load <-dequeue_task_fair 1939 kworker/-59 3d..2 4us : clear_buddies <-dequeue_task_fair 1940 kworker/-59 3d..2 4us : account_entity_dequeue <-dequeue_task_fair 1941 kworker/-59 3d..2 4us : update_min_vruntime <-dequeue_task_fair 1942 kworker/-59 3d..2 4us : update_cfs_shares <-dequeue_task_fair 1943 kworker/-59 3d..2 5us : hrtick_update <-dequeue_task_fair 1944 kworker/-59 3d..2 5us : wq_worker_sleeping <-__schedule 1945 kworker/-59 3d..2 5us : kthread_data <-wq_worker_sleeping 1946 kworker/-59 3d..2 5us : put_prev_task_fair <-__schedule 1947 kworker/-59 3d..2 6us : pick_next_task_fair <-pick_next_task 1948 kworker/-59 3d..2 6us : clear_buddies <-pick_next_task_fair 1949 kworker/-59 3d..2 6us : set_next_entity <-pick_next_task_fair 1950 kworker/-59 3d..2 6us : update_stats_wait_end <-set_next_entity 1951 ls-2269 3d..2 7us : finish_task_switch <-__schedule 1952 ls-2269 3d..2 7us : _raw_spin_unlock_irq <-finish_task_switch 1953 ls-2269 3d..2 8us : do_IRQ <-ret_from_intr 1954 ls-2269 3d..2 8us : irq_enter <-do_IRQ 1955 ls-2269 3d..2 8us : rcu_irq_enter <-irq_enter 1956 ls-2269 3d..2 9us : add_preempt_count <-irq_enter 1957 ls-2269 3d.h2 9us : exit_idle <-do_IRQ 1958 [...] 1959 ls-2269 3d.h3 20us : sub_preempt_count <-_raw_spin_unlock 1960 ls-2269 3d.h2 20us : irq_exit <-do_IRQ 1961 ls-2269 3d.h2 21us : sub_preempt_count <-irq_exit 1962 ls-2269 3d..3 21us : do_softirq <-irq_exit 1963 ls-2269 3d..3 21us : __do_softirq <-call_softirq 1964 ls-2269 3d..3 21us+: __local_bh_disable <-__do_softirq 1965 ls-2269 3d.s4 29us : sub_preempt_count <-_local_bh_enable_ip 1966 ls-2269 3d.s5 29us : sub_preempt_count <-_local_bh_enable_ip 1967 ls-2269 3d.s5 31us : do_IRQ <-ret_from_intr 1968 ls-2269 3d.s5 31us : irq_enter <-do_IRQ 1969 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter 1970 [...] 1971 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter 1972 ls-2269 3d.s5 32us : add_preempt_count <-irq_enter 1973 ls-2269 3d.H5 32us : exit_idle <-do_IRQ 1974 ls-2269 3d.H5 32us : handle_irq <-do_IRQ 1975 ls-2269 3d.H5 32us : irq_to_desc <-handle_irq 1976 ls-2269 3d.H5 33us : handle_fasteoi_irq <-handle_irq 1977 [...] 1978 ls-2269 3d.s5 158us : _raw_spin_unlock_irqrestore <-rtl8139_poll 1979 ls-2269 3d.s3 158us : net_rps_action_and_irq_enable.isra.65 <-net_rx_action 1980 ls-2269 3d.s3 159us : __local_bh_enable <-__do_softirq 1981 ls-2269 3d.s3 159us : sub_preempt_count <-__local_bh_enable 1982 ls-2269 3d..3 159us : idle_cpu <-irq_exit 1983 ls-2269 3d..3 159us : rcu_irq_exit <-irq_exit 1984 ls-2269 3d..3 160us : sub_preempt_count <-irq_exit 1985 ls-2269 3d... 161us : __mutex_unlock_slowpath <-mutex_unlock 1986 ls-2269 3d... 162us+: trace_hardirqs_on <-mutex_unlock 1987 ls-2269 3d... 186us : <stack trace> 1988 => __mutex_unlock_slowpath 1989 => mutex_unlock 1990 => process_output 1991 => n_tty_write 1992 => tty_write 1993 => vfs_write 1994 => sys_write 1995 => system_call_fastpath 1996 1997This is an interesting trace. It started with kworker running and 1998scheduling out and ls taking over. But as soon as ls released the 1999rq lock and enabled interrupts (but not preemption) an interrupt 2000triggered. When the interrupt finished, it started running softirqs. 2001But while the softirq was running, another interrupt triggered. 2002When an interrupt is running inside a softirq, the annotation is 'H'. 2003 2004 2005wakeup 2006------ 2007 2008One common case that people are interested in tracing is the 2009time it takes for a task that is woken to actually wake up. 2010Now for non Real-Time tasks, this can be arbitrary. But tracing 2011it nonetheless can be interesting. 2012 2013Without function tracing:: 2014 2015 # echo 0 > options/function-trace 2016 # echo wakeup > current_tracer 2017 # echo 1 > tracing_on 2018 # echo 0 > tracing_max_latency 2019 # chrt -f 5 sleep 1 2020 # echo 0 > tracing_on 2021 # cat trace 2022 # tracer: wakeup 2023 # 2024 # wakeup latency trace v1.1.5 on 3.8.0-test+ 2025 # -------------------------------------------------------------------- 2026 # latency: 15 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2027 # ----------------- 2028 # | task: kworker/3:1H-312 (uid:0 nice:-20 policy:0 rt_prio:0) 2029 # ----------------- 2030 # 2031 # _------=> CPU# 2032 # / _-----=> irqs-off 2033 # | / _----=> need-resched 2034 # || / _---=> hardirq/softirq 2035 # ||| / _--=> preempt-depth 2036 # |||| / delay 2037 # cmd pid ||||| time | caller 2038 # \ / ||||| \ | / 2039 <idle>-0 3dNs7 0us : 0:120:R + [003] 312:100:R kworker/3:1H 2040 <idle>-0 3dNs7 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up 2041 <idle>-0 3d..3 15us : __schedule <-schedule 2042 <idle>-0 3d..3 15us : 0:120:R ==> [003] 312:100:R kworker/3:1H 2043 2044The tracer only traces the highest priority task in the system 2045to avoid tracing the normal circumstances. Here we see that 2046the kworker with a nice priority of -20 (not very nice), took 2047just 15 microseconds from the time it woke up, to the time it 2048ran. 2049 2050Non Real-Time tasks are not that interesting. A more interesting 2051trace is to concentrate only on Real-Time tasks. 2052 2053wakeup_rt 2054--------- 2055 2056In a Real-Time environment it is very important to know the 2057wakeup time it takes for the highest priority task that is woken 2058up to the time that it executes. This is also known as "schedule 2059latency". I stress the point that this is about RT tasks. It is 2060also important to know the scheduling latency of non-RT tasks, 2061but the average schedule latency is better for non-RT tasks. 2062Tools like LatencyTop are more appropriate for such 2063measurements. 2064 2065Real-Time environments are interested in the worst case latency. 2066That is the longest latency it takes for something to happen, 2067and not the average. We can have a very fast scheduler that may 2068only have a large latency once in a while, but that would not 2069work well with Real-Time tasks. The wakeup_rt tracer was designed 2070to record the worst case wakeups of RT tasks. Non-RT tasks are 2071not recorded because the tracer only records one worst case and 2072tracing non-RT tasks that are unpredictable will overwrite the 2073worst case latency of RT tasks (just run the normal wakeup 2074tracer for a while to see that effect). 2075 2076Since this tracer only deals with RT tasks, we will run this 2077slightly differently than we did with the previous tracers. 2078Instead of performing an 'ls', we will run 'sleep 1' under 2079'chrt' which changes the priority of the task. 2080:: 2081 2082 # echo 0 > options/function-trace 2083 # echo wakeup_rt > current_tracer 2084 # echo 1 > tracing_on 2085 # echo 0 > tracing_max_latency 2086 # chrt -f 5 sleep 1 2087 # echo 0 > tracing_on 2088 # cat trace 2089 # tracer: wakeup 2090 # 2091 # tracer: wakeup_rt 2092 # 2093 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 2094 # -------------------------------------------------------------------- 2095 # latency: 5 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2096 # ----------------- 2097 # | task: sleep-2389 (uid:0 nice:0 policy:1 rt_prio:5) 2098 # ----------------- 2099 # 2100 # _------=> CPU# 2101 # / _-----=> irqs-off 2102 # | / _----=> need-resched 2103 # || / _---=> hardirq/softirq 2104 # ||| / _--=> preempt-depth 2105 # |||| / delay 2106 # cmd pid ||||| time | caller 2107 # \ / ||||| \ | / 2108 <idle>-0 3d.h4 0us : 0:120:R + [003] 2389: 94:R sleep 2109 <idle>-0 3d.h4 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up 2110 <idle>-0 3d..3 5us : __schedule <-schedule 2111 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep 2112 2113 2114Running this on an idle system, we see that it only took 5 microseconds 2115to perform the task switch. Note, since the trace point in the schedule 2116is before the actual "switch", we stop the tracing when the recorded task 2117is about to schedule in. This may change if we add a new marker at the 2118end of the scheduler. 2119 2120Notice that the recorded task is 'sleep' with the PID of 2389 2121and it has an rt_prio of 5. This priority is user-space priority 2122and not the internal kernel priority. The policy is 1 for 2123SCHED_FIFO and 2 for SCHED_RR. 2124 2125Note, that the trace data shows the internal priority (99 - rtprio). 2126:: 2127 2128 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep 2129 2130The 0:120:R means idle was running with a nice priority of 0 (120 - 120) 2131and in the running state 'R'. The sleep task was scheduled in with 21322389: 94:R. That is the priority is the kernel rtprio (99 - 5 = 94) 2133and it too is in the running state. 2134 2135Doing the same with chrt -r 5 and function-trace set. 2136:: 2137 2138 echo 1 > options/function-trace 2139 2140 # tracer: wakeup_rt 2141 # 2142 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 2143 # -------------------------------------------------------------------- 2144 # latency: 29 us, #85/85, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2145 # ----------------- 2146 # | task: sleep-2448 (uid:0 nice:0 policy:1 rt_prio:5) 2147 # ----------------- 2148 # 2149 # _------=> CPU# 2150 # / _-----=> irqs-off 2151 # | / _----=> need-resched 2152 # || / _---=> hardirq/softirq 2153 # ||| / _--=> preempt-depth 2154 # |||| / delay 2155 # cmd pid ||||| time | caller 2156 # \ / ||||| \ | / 2157 <idle>-0 3d.h4 1us+: 0:120:R + [003] 2448: 94:R sleep 2158 <idle>-0 3d.h4 2us : ttwu_do_activate.constprop.87 <-try_to_wake_up 2159 <idle>-0 3d.h3 3us : check_preempt_curr <-ttwu_do_wakeup 2160 <idle>-0 3d.h3 3us : resched_curr <-check_preempt_curr 2161 <idle>-0 3dNh3 4us : task_woken_rt <-ttwu_do_wakeup 2162 <idle>-0 3dNh3 4us : _raw_spin_unlock <-try_to_wake_up 2163 <idle>-0 3dNh3 4us : sub_preempt_count <-_raw_spin_unlock 2164 <idle>-0 3dNh2 5us : ttwu_stat <-try_to_wake_up 2165 <idle>-0 3dNh2 5us : _raw_spin_unlock_irqrestore <-try_to_wake_up 2166 <idle>-0 3dNh2 6us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2167 <idle>-0 3dNh1 6us : _raw_spin_lock <-__run_hrtimer 2168 <idle>-0 3dNh1 6us : add_preempt_count <-_raw_spin_lock 2169 <idle>-0 3dNh2 7us : _raw_spin_unlock <-hrtimer_interrupt 2170 <idle>-0 3dNh2 7us : sub_preempt_count <-_raw_spin_unlock 2171 <idle>-0 3dNh1 7us : tick_program_event <-hrtimer_interrupt 2172 <idle>-0 3dNh1 7us : clockevents_program_event <-tick_program_event 2173 <idle>-0 3dNh1 8us : ktime_get <-clockevents_program_event 2174 <idle>-0 3dNh1 8us : lapic_next_event <-clockevents_program_event 2175 <idle>-0 3dNh1 8us : irq_exit <-smp_apic_timer_interrupt 2176 <idle>-0 3dNh1 9us : sub_preempt_count <-irq_exit 2177 <idle>-0 3dN.2 9us : idle_cpu <-irq_exit 2178 <idle>-0 3dN.2 9us : rcu_irq_exit <-irq_exit 2179 <idle>-0 3dN.2 10us : rcu_eqs_enter_common.isra.45 <-rcu_irq_exit 2180 <idle>-0 3dN.2 10us : sub_preempt_count <-irq_exit 2181 <idle>-0 3.N.1 11us : rcu_idle_exit <-cpu_idle 2182 <idle>-0 3dN.1 11us : rcu_eqs_exit_common.isra.43 <-rcu_idle_exit 2183 <idle>-0 3.N.1 11us : tick_nohz_idle_exit <-cpu_idle 2184 <idle>-0 3dN.1 12us : menu_hrtimer_cancel <-tick_nohz_idle_exit 2185 <idle>-0 3dN.1 12us : ktime_get <-tick_nohz_idle_exit 2186 <idle>-0 3dN.1 12us : tick_do_update_jiffies64 <-tick_nohz_idle_exit 2187 <idle>-0 3dN.1 13us : cpu_load_update_nohz <-tick_nohz_idle_exit 2188 <idle>-0 3dN.1 13us : _raw_spin_lock <-cpu_load_update_nohz 2189 <idle>-0 3dN.1 13us : add_preempt_count <-_raw_spin_lock 2190 <idle>-0 3dN.2 13us : __cpu_load_update <-cpu_load_update_nohz 2191 <idle>-0 3dN.2 14us : sched_avg_update <-__cpu_load_update 2192 <idle>-0 3dN.2 14us : _raw_spin_unlock <-cpu_load_update_nohz 2193 <idle>-0 3dN.2 14us : sub_preempt_count <-_raw_spin_unlock 2194 <idle>-0 3dN.1 15us : calc_load_nohz_stop <-tick_nohz_idle_exit 2195 <idle>-0 3dN.1 15us : touch_softlockup_watchdog <-tick_nohz_idle_exit 2196 <idle>-0 3dN.1 15us : hrtimer_cancel <-tick_nohz_idle_exit 2197 <idle>-0 3dN.1 15us : hrtimer_try_to_cancel <-hrtimer_cancel 2198 <idle>-0 3dN.1 16us : lock_hrtimer_base.isra.18 <-hrtimer_try_to_cancel 2199 <idle>-0 3dN.1 16us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18 2200 <idle>-0 3dN.1 16us : add_preempt_count <-_raw_spin_lock_irqsave 2201 <idle>-0 3dN.2 17us : __remove_hrtimer <-remove_hrtimer.part.16 2202 <idle>-0 3dN.2 17us : hrtimer_force_reprogram <-__remove_hrtimer 2203 <idle>-0 3dN.2 17us : tick_program_event <-hrtimer_force_reprogram 2204 <idle>-0 3dN.2 18us : clockevents_program_event <-tick_program_event 2205 <idle>-0 3dN.2 18us : ktime_get <-clockevents_program_event 2206 <idle>-0 3dN.2 18us : lapic_next_event <-clockevents_program_event 2207 <idle>-0 3dN.2 19us : _raw_spin_unlock_irqrestore <-hrtimer_try_to_cancel 2208 <idle>-0 3dN.2 19us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2209 <idle>-0 3dN.1 19us : hrtimer_forward <-tick_nohz_idle_exit 2210 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward 2211 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward 2212 <idle>-0 3dN.1 20us : hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11 2213 <idle>-0 3dN.1 20us : __hrtimer_start_range_ns <-hrtimer_start_range_ns 2214 <idle>-0 3dN.1 21us : lock_hrtimer_base.isra.18 <-__hrtimer_start_range_ns 2215 <idle>-0 3dN.1 21us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18 2216 <idle>-0 3dN.1 21us : add_preempt_count <-_raw_spin_lock_irqsave 2217 <idle>-0 3dN.2 22us : ktime_add_safe <-__hrtimer_start_range_ns 2218 <idle>-0 3dN.2 22us : enqueue_hrtimer <-__hrtimer_start_range_ns 2219 <idle>-0 3dN.2 22us : tick_program_event <-__hrtimer_start_range_ns 2220 <idle>-0 3dN.2 23us : clockevents_program_event <-tick_program_event 2221 <idle>-0 3dN.2 23us : ktime_get <-clockevents_program_event 2222 <idle>-0 3dN.2 23us : lapic_next_event <-clockevents_program_event 2223 <idle>-0 3dN.2 24us : _raw_spin_unlock_irqrestore <-__hrtimer_start_range_ns 2224 <idle>-0 3dN.2 24us : sub_preempt_count <-_raw_spin_unlock_irqrestore 2225 <idle>-0 3dN.1 24us : account_idle_ticks <-tick_nohz_idle_exit 2226 <idle>-0 3dN.1 24us : account_idle_time <-account_idle_ticks 2227 <idle>-0 3.N.1 25us : sub_preempt_count <-cpu_idle 2228 <idle>-0 3.N.. 25us : schedule <-cpu_idle 2229 <idle>-0 3.N.. 25us : __schedule <-preempt_schedule 2230 <idle>-0 3.N.. 26us : add_preempt_count <-__schedule 2231 <idle>-0 3.N.1 26us : rcu_note_context_switch <-__schedule 2232 <idle>-0 3.N.1 26us : rcu_sched_qs <-rcu_note_context_switch 2233 <idle>-0 3dN.1 27us : rcu_preempt_qs <-rcu_note_context_switch 2234 <idle>-0 3.N.1 27us : _raw_spin_lock_irq <-__schedule 2235 <idle>-0 3dN.1 27us : add_preempt_count <-_raw_spin_lock_irq 2236 <idle>-0 3dN.2 28us : put_prev_task_idle <-__schedule 2237 <idle>-0 3dN.2 28us : pick_next_task_stop <-pick_next_task 2238 <idle>-0 3dN.2 28us : pick_next_task_rt <-pick_next_task 2239 <idle>-0 3dN.2 29us : dequeue_pushable_task <-pick_next_task_rt 2240 <idle>-0 3d..3 29us : __schedule <-preempt_schedule 2241 <idle>-0 3d..3 30us : 0:120:R ==> [003] 2448: 94:R sleep 2242 2243This isn't that big of a trace, even with function tracing enabled, 2244so I included the entire trace. 2245 2246The interrupt went off while when the system was idle. Somewhere 2247before task_woken_rt() was called, the NEED_RESCHED flag was set, 2248this is indicated by the first occurrence of the 'N' flag. 2249 2250Latency tracing and events 2251-------------------------- 2252As function tracing can induce a much larger latency, but without 2253seeing what happens within the latency it is hard to know what 2254caused it. There is a middle ground, and that is with enabling 2255events. 2256:: 2257 2258 # echo 0 > options/function-trace 2259 # echo wakeup_rt > current_tracer 2260 # echo 1 > events/enable 2261 # echo 1 > tracing_on 2262 # echo 0 > tracing_max_latency 2263 # chrt -f 5 sleep 1 2264 # echo 0 > tracing_on 2265 # cat trace 2266 # tracer: wakeup_rt 2267 # 2268 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+ 2269 # -------------------------------------------------------------------- 2270 # latency: 6 us, #12/12, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4) 2271 # ----------------- 2272 # | task: sleep-5882 (uid:0 nice:0 policy:1 rt_prio:5) 2273 # ----------------- 2274 # 2275 # _------=> CPU# 2276 # / _-----=> irqs-off 2277 # | / _----=> need-resched 2278 # || / _---=> hardirq/softirq 2279 # ||| / _--=> preempt-depth 2280 # |||| / delay 2281 # cmd pid ||||| time | caller 2282 # \ / ||||| \ | / 2283 <idle>-0 2d.h4 0us : 0:120:R + [002] 5882: 94:R sleep 2284 <idle>-0 2d.h4 0us : ttwu_do_activate.constprop.87 <-try_to_wake_up 2285 <idle>-0 2d.h4 1us : sched_wakeup: comm=sleep pid=5882 prio=94 success=1 target_cpu=002 2286 <idle>-0 2dNh2 1us : hrtimer_expire_exit: hrtimer=ffff88007796feb8 2287 <idle>-0 2.N.2 2us : power_end: cpu_id=2 2288 <idle>-0 2.N.2 3us : cpu_idle: state=4294967295 cpu_id=2 2289 <idle>-0 2dN.3 4us : hrtimer_cancel: hrtimer=ffff88007d50d5e0 2290 <idle>-0 2dN.3 4us : hrtimer_start: hrtimer=ffff88007d50d5e0 function=tick_sched_timer expires=34311211000000 softexpires=34311211000000 2291 <idle>-0 2.N.2 5us : rcu_utilization: Start context switch 2292 <idle>-0 2.N.2 5us : rcu_utilization: End context switch 2293 <idle>-0 2d..3 6us : __schedule <-schedule 2294 <idle>-0 2d..3 6us : 0:120:R ==> [002] 5882: 94:R sleep 2295 2296 2297Hardware Latency Detector 2298------------------------- 2299 2300The hardware latency detector is executed by enabling the "hwlat" tracer. 2301 2302NOTE, this tracer will affect the performance of the system as it will 2303periodically make a CPU constantly busy with interrupts disabled. 2304:: 2305 2306 # echo hwlat > current_tracer 2307 # sleep 100 2308 # cat trace 2309 # tracer: hwlat 2310 # 2311 # entries-in-buffer/entries-written: 13/13 #P:8 2312 # 2313 # _-----=> irqs-off 2314 # / _----=> need-resched 2315 # | / _---=> hardirq/softirq 2316 # || / _--=> preempt-depth 2317 # ||| / delay 2318 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2319 # | | | |||| | | 2320 <...>-1729 [001] d... 678.473449: #1 inner/outer(us): 11/12 ts:1581527483.343962693 count:6 2321 <...>-1729 [004] d... 689.556542: #2 inner/outer(us): 16/9 ts:1581527494.889008092 count:1 2322 <...>-1729 [005] d... 714.756290: #3 inner/outer(us): 16/16 ts:1581527519.678961629 count:5 2323 <...>-1729 [001] d... 718.788247: #4 inner/outer(us): 9/17 ts:1581527523.889012713 count:1 2324 <...>-1729 [002] d... 719.796341: #5 inner/outer(us): 13/9 ts:1581527524.912872606 count:1 2325 <...>-1729 [006] d... 844.787091: #6 inner/outer(us): 9/12 ts:1581527649.889048502 count:2 2326 <...>-1729 [003] d... 849.827033: #7 inner/outer(us): 18/9 ts:1581527654.889013793 count:1 2327 <...>-1729 [007] d... 853.859002: #8 inner/outer(us): 9/12 ts:1581527658.889065736 count:1 2328 <...>-1729 [001] d... 855.874978: #9 inner/outer(us): 9/11 ts:1581527660.861991877 count:1 2329 <...>-1729 [001] d... 863.938932: #10 inner/outer(us): 9/11 ts:1581527668.970010500 count:1 nmi-total:7 nmi-count:1 2330 <...>-1729 [007] d... 878.050780: #11 inner/outer(us): 9/12 ts:1581527683.385002600 count:1 nmi-total:5 nmi-count:1 2331 <...>-1729 [007] d... 886.114702: #12 inner/outer(us): 9/12 ts:1581527691.385001600 count:1 2332 2333 2334The above output is somewhat the same in the header. All events will have 2335interrupts disabled 'd'. Under the FUNCTION title there is: 2336 2337 #1 2338 This is the count of events recorded that were greater than the 2339 tracing_threshold (See below). 2340 2341 inner/outer(us): 11/11 2342 2343 This shows two numbers as "inner latency" and "outer latency". The test 2344 runs in a loop checking a timestamp twice. The latency detected within 2345 the two timestamps is the "inner latency" and the latency detected 2346 after the previous timestamp and the next timestamp in the loop is 2347 the "outer latency". 2348 2349 ts:1581527483.343962693 2350 2351 The absolute timestamp that the first latency was recorded in the window. 2352 2353 count:6 2354 2355 The number of times a latency was detected during the window. 2356 2357 nmi-total:7 nmi-count:1 2358 2359 On architectures that support it, if an NMI comes in during the 2360 test, the time spent in NMI is reported in "nmi-total" (in 2361 microseconds). 2362 2363 All architectures that have NMIs will show the "nmi-count" if an 2364 NMI comes in during the test. 2365 2366hwlat files: 2367 2368 tracing_threshold 2369 This gets automatically set to "10" to represent 10 2370 microseconds. This is the threshold of latency that 2371 needs to be detected before the trace will be recorded. 2372 2373 Note, when hwlat tracer is finished (another tracer is 2374 written into "current_tracer"), the original value for 2375 tracing_threshold is placed back into this file. 2376 2377 hwlat_detector/width 2378 The length of time the test runs with interrupts disabled. 2379 2380 hwlat_detector/window 2381 The length of time of the window which the test 2382 runs. That is, the test will run for "width" 2383 microseconds per "window" microseconds 2384 2385 tracing_cpumask 2386 When the test is started. A kernel thread is created that 2387 runs the test. This thread will alternate between CPUs 2388 listed in the tracing_cpumask between each period 2389 (one "window"). To limit the test to specific CPUs 2390 set the mask in this file to only the CPUs that the test 2391 should run on. 2392 2393function 2394-------- 2395 2396This tracer is the function tracer. Enabling the function tracer 2397can be done from the debug file system. Make sure the 2398ftrace_enabled is set; otherwise this tracer is a nop. 2399See the "ftrace_enabled" section below. 2400:: 2401 2402 # sysctl kernel.ftrace_enabled=1 2403 # echo function > current_tracer 2404 # echo 1 > tracing_on 2405 # usleep 1 2406 # echo 0 > tracing_on 2407 # cat trace 2408 # tracer: function 2409 # 2410 # entries-in-buffer/entries-written: 24799/24799 #P:4 2411 # 2412 # _-----=> irqs-off 2413 # / _----=> need-resched 2414 # | / _---=> hardirq/softirq 2415 # || / _--=> preempt-depth 2416 # ||| / delay 2417 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 2418 # | | | |||| | | 2419 bash-1994 [002] .... 3082.063030: mutex_unlock <-rb_simple_write 2420 bash-1994 [002] .... 3082.063031: __mutex_unlock_slowpath <-mutex_unlock 2421 bash-1994 [002] .... 3082.063031: __fsnotify_parent <-fsnotify_modify 2422 bash-1994 [002] .... 3082.063032: fsnotify <-fsnotify_modify 2423 bash-1994 [002] .... 3082.063032: __srcu_read_lock <-fsnotify 2424 bash-1994 [002] .... 3082.063032: add_preempt_count <-__srcu_read_lock 2425 bash-1994 [002] ...1 3082.063032: sub_preempt_count <-__srcu_read_lock 2426 bash-1994 [002] .... 3082.063033: __srcu_read_unlock <-fsnotify 2427 [...] 2428 2429 2430Note: function tracer uses ring buffers to store the above 2431entries. The newest data may overwrite the oldest data. 2432Sometimes using echo to stop the trace is not sufficient because 2433the tracing could have overwritten the data that you wanted to 2434record. For this reason, it is sometimes better to disable 2435tracing directly from a program. This allows you to stop the 2436tracing at the point that you hit the part that you are 2437interested in. To disable the tracing directly from a C program, 2438something like following code snippet can be used:: 2439 2440 int trace_fd; 2441 [...] 2442 int main(int argc, char *argv[]) { 2443 [...] 2444 trace_fd = open(tracing_file("tracing_on"), O_WRONLY); 2445 [...] 2446 if (condition_hit()) { 2447 write(trace_fd, "0", 1); 2448 } 2449 [...] 2450 } 2451 2452 2453Single thread tracing 2454--------------------- 2455 2456By writing into set_ftrace_pid you can trace a 2457single thread. For example:: 2458 2459 # cat set_ftrace_pid 2460 no pid 2461 # echo 3111 > set_ftrace_pid 2462 # cat set_ftrace_pid 2463 3111 2464 # echo function > current_tracer 2465 # cat trace | head 2466 # tracer: function 2467 # 2468 # TASK-PID CPU# TIMESTAMP FUNCTION 2469 # | | | | | 2470 yum-updatesd-3111 [003] 1637.254676: finish_task_switch <-thread_return 2471 yum-updatesd-3111 [003] 1637.254681: hrtimer_cancel <-schedule_hrtimeout_range 2472 yum-updatesd-3111 [003] 1637.254682: hrtimer_try_to_cancel <-hrtimer_cancel 2473 yum-updatesd-3111 [003] 1637.254683: lock_hrtimer_base <-hrtimer_try_to_cancel 2474 yum-updatesd-3111 [003] 1637.254685: fget_light <-do_sys_poll 2475 yum-updatesd-3111 [003] 1637.254686: pipe_poll <-do_sys_poll 2476 # echo > set_ftrace_pid 2477 # cat trace |head 2478 # tracer: function 2479 # 2480 # TASK-PID CPU# TIMESTAMP FUNCTION 2481 # | | | | | 2482 ##### CPU 3 buffer started #### 2483 yum-updatesd-3111 [003] 1701.957688: free_poll_entry <-poll_freewait 2484 yum-updatesd-3111 [003] 1701.957689: remove_wait_queue <-free_poll_entry 2485 yum-updatesd-3111 [003] 1701.957691: fput <-free_poll_entry 2486 yum-updatesd-3111 [003] 1701.957692: audit_syscall_exit <-sysret_audit 2487 yum-updatesd-3111 [003] 1701.957693: path_put <-audit_syscall_exit 2488 2489If you want to trace a function when executing, you could use 2490something like this simple program. 2491:: 2492 2493 #include <stdio.h> 2494 #include <stdlib.h> 2495 #include <sys/types.h> 2496 #include <sys/stat.h> 2497 #include <fcntl.h> 2498 #include <unistd.h> 2499 #include <string.h> 2500 2501 #define _STR(x) #x 2502 #define STR(x) _STR(x) 2503 #define MAX_PATH 256 2504 2505 const char *find_tracefs(void) 2506 { 2507 static char tracefs[MAX_PATH+1]; 2508 static int tracefs_found; 2509 char type[100]; 2510 FILE *fp; 2511 2512 if (tracefs_found) 2513 return tracefs; 2514 2515 if ((fp = fopen("/proc/mounts","r")) == NULL) { 2516 perror("/proc/mounts"); 2517 return NULL; 2518 } 2519 2520 while (fscanf(fp, "%*s %" 2521 STR(MAX_PATH) 2522 "s %99s %*s %*d %*d\n", 2523 tracefs, type) == 2) { 2524 if (strcmp(type, "tracefs") == 0) 2525 break; 2526 } 2527 fclose(fp); 2528 2529 if (strcmp(type, "tracefs") != 0) { 2530 fprintf(stderr, "tracefs not mounted"); 2531 return NULL; 2532 } 2533 2534 strcat(tracefs, "/tracing/"); 2535 tracefs_found = 1; 2536 2537 return tracefs; 2538 } 2539 2540 const char *tracing_file(const char *file_name) 2541 { 2542 static char trace_file[MAX_PATH+1]; 2543 snprintf(trace_file, MAX_PATH, "%s/%s", find_tracefs(), file_name); 2544 return trace_file; 2545 } 2546 2547 int main (int argc, char **argv) 2548 { 2549 if (argc < 1) 2550 exit(-1); 2551 2552 if (fork() > 0) { 2553 int fd, ffd; 2554 char line[64]; 2555 int s; 2556 2557 ffd = open(tracing_file("current_tracer"), O_WRONLY); 2558 if (ffd < 0) 2559 exit(-1); 2560 write(ffd, "nop", 3); 2561 2562 fd = open(tracing_file("set_ftrace_pid"), O_WRONLY); 2563 s = sprintf(line, "%d\n", getpid()); 2564 write(fd, line, s); 2565 2566 write(ffd, "function", 8); 2567 2568 close(fd); 2569 close(ffd); 2570 2571 execvp(argv[1], argv+1); 2572 } 2573 2574 return 0; 2575 } 2576 2577Or this simple script! 2578:: 2579 2580 #!/bin/bash 2581 2582 tracefs=`sed -ne 's/^tracefs \(.*\) tracefs.*/\1/p' /proc/mounts` 2583 echo 0 > $tracefs/tracing_on 2584 echo $$ > $tracefs/set_ftrace_pid 2585 echo function > $tracefs/current_tracer 2586 echo 1 > $tracefs/tracing_on 2587 exec "$@" 2588 2589 2590function graph tracer 2591--------------------------- 2592 2593This tracer is similar to the function tracer except that it 2594probes a function on its entry and its exit. This is done by 2595using a dynamically allocated stack of return addresses in each 2596task_struct. On function entry the tracer overwrites the return 2597address of each function traced to set a custom probe. Thus the 2598original return address is stored on the stack of return address 2599in the task_struct. 2600 2601Probing on both ends of a function leads to special features 2602such as: 2603 2604- measure of a function's time execution 2605- having a reliable call stack to draw function calls graph 2606 2607This tracer is useful in several situations: 2608 2609- you want to find the reason of a strange kernel behavior and 2610 need to see what happens in detail on any areas (or specific 2611 ones). 2612 2613- you are experiencing weird latencies but it's difficult to 2614 find its origin. 2615 2616- you want to find quickly which path is taken by a specific 2617 function 2618 2619- you just want to peek inside a working kernel and want to see 2620 what happens there. 2621 2622:: 2623 2624 # tracer: function_graph 2625 # 2626 # CPU DURATION FUNCTION CALLS 2627 # | | | | | | | 2628 2629 0) | sys_open() { 2630 0) | do_sys_open() { 2631 0) | getname() { 2632 0) | kmem_cache_alloc() { 2633 0) 1.382 us | __might_sleep(); 2634 0) 2.478 us | } 2635 0) | strncpy_from_user() { 2636 0) | might_fault() { 2637 0) 1.389 us | __might_sleep(); 2638 0) 2.553 us | } 2639 0) 3.807 us | } 2640 0) 7.876 us | } 2641 0) | alloc_fd() { 2642 0) 0.668 us | _spin_lock(); 2643 0) 0.570 us | expand_files(); 2644 0) 0.586 us | _spin_unlock(); 2645 2646 2647There are several columns that can be dynamically 2648enabled/disabled. You can use every combination of options you 2649want, depending on your needs. 2650 2651- The cpu number on which the function executed is default 2652 enabled. It is sometimes better to only trace one cpu (see 2653 tracing_cpumask file) or you might sometimes see unordered 2654 function calls while cpu tracing switch. 2655 2656 - hide: echo nofuncgraph-cpu > trace_options 2657 - show: echo funcgraph-cpu > trace_options 2658 2659- The duration (function's time of execution) is displayed on 2660 the closing bracket line of a function or on the same line 2661 than the current function in case of a leaf one. It is default 2662 enabled. 2663 2664 - hide: echo nofuncgraph-duration > trace_options 2665 - show: echo funcgraph-duration > trace_options 2666 2667- The overhead field precedes the duration field in case of 2668 reached duration thresholds. 2669 2670 - hide: echo nofuncgraph-overhead > trace_options 2671 - show: echo funcgraph-overhead > trace_options 2672 - depends on: funcgraph-duration 2673 2674 ie:: 2675 2676 3) # 1837.709 us | } /* __switch_to */ 2677 3) | finish_task_switch() { 2678 3) 0.313 us | _raw_spin_unlock_irq(); 2679 3) 3.177 us | } 2680 3) # 1889.063 us | } /* __schedule */ 2681 3) ! 140.417 us | } /* __schedule */ 2682 3) # 2034.948 us | } /* schedule */ 2683 3) * 33998.59 us | } /* schedule_preempt_disabled */ 2684 2685 [...] 2686 2687 1) 0.260 us | msecs_to_jiffies(); 2688 1) 0.313 us | __rcu_read_unlock(); 2689 1) + 61.770 us | } 2690 1) + 64.479 us | } 2691 1) 0.313 us | rcu_bh_qs(); 2692 1) 0.313 us | __local_bh_enable(); 2693 1) ! 217.240 us | } 2694 1) 0.365 us | idle_cpu(); 2695 1) | rcu_irq_exit() { 2696 1) 0.417 us | rcu_eqs_enter_common.isra.47(); 2697 1) 3.125 us | } 2698 1) ! 227.812 us | } 2699 1) ! 457.395 us | } 2700 1) @ 119760.2 us | } 2701 2702 [...] 2703 2704 2) | handle_IPI() { 2705 1) 6.979 us | } 2706 2) 0.417 us | scheduler_ipi(); 2707 1) 9.791 us | } 2708 1) + 12.917 us | } 2709 2) 3.490 us | } 2710 1) + 15.729 us | } 2711 1) + 18.542 us | } 2712 2) $ 3594274 us | } 2713 2714Flags:: 2715 2716 + means that the function exceeded 10 usecs. 2717 ! means that the function exceeded 100 usecs. 2718 # means that the function exceeded 1000 usecs. 2719 * means that the function exceeded 10 msecs. 2720 @ means that the function exceeded 100 msecs. 2721 $ means that the function exceeded 1 sec. 2722 2723 2724- The task/pid field displays the thread cmdline and pid which 2725 executed the function. It is default disabled. 2726 2727 - hide: echo nofuncgraph-proc > trace_options 2728 - show: echo funcgraph-proc > trace_options 2729 2730 ie:: 2731 2732 # tracer: function_graph 2733 # 2734 # CPU TASK/PID DURATION FUNCTION CALLS 2735 # | | | | | | | | | 2736 0) sh-4802 | | d_free() { 2737 0) sh-4802 | | call_rcu() { 2738 0) sh-4802 | | __call_rcu() { 2739 0) sh-4802 | 0.616 us | rcu_process_gp_end(); 2740 0) sh-4802 | 0.586 us | check_for_new_grace_period(); 2741 0) sh-4802 | 2.899 us | } 2742 0) sh-4802 | 4.040 us | } 2743 0) sh-4802 | 5.151 us | } 2744 0) sh-4802 | + 49.370 us | } 2745 2746 2747- The absolute time field is an absolute timestamp given by the 2748 system clock since it started. A snapshot of this time is 2749 given on each entry/exit of functions 2750 2751 - hide: echo nofuncgraph-abstime > trace_options 2752 - show: echo funcgraph-abstime > trace_options 2753 2754 ie:: 2755 2756 # 2757 # TIME CPU DURATION FUNCTION CALLS 2758 # | | | | | | | | 2759 360.774522 | 1) 0.541 us | } 2760 360.774522 | 1) 4.663 us | } 2761 360.774523 | 1) 0.541 us | __wake_up_bit(); 2762 360.774524 | 1) 6.796 us | } 2763 360.774524 | 1) 7.952 us | } 2764 360.774525 | 1) 9.063 us | } 2765 360.774525 | 1) 0.615 us | journal_mark_dirty(); 2766 360.774527 | 1) 0.578 us | __brelse(); 2767 360.774528 | 1) | reiserfs_prepare_for_journal() { 2768 360.774528 | 1) | unlock_buffer() { 2769 360.774529 | 1) | wake_up_bit() { 2770 360.774529 | 1) | bit_waitqueue() { 2771 360.774530 | 1) 0.594 us | __phys_addr(); 2772 2773 2774The function name is always displayed after the closing bracket 2775for a function if the start of that function is not in the 2776trace buffer. 2777 2778Display of the function name after the closing bracket may be 2779enabled for functions whose start is in the trace buffer, 2780allowing easier searching with grep for function durations. 2781It is default disabled. 2782 2783 - hide: echo nofuncgraph-tail > trace_options 2784 - show: echo funcgraph-tail > trace_options 2785 2786 Example with nofuncgraph-tail (default):: 2787 2788 0) | putname() { 2789 0) | kmem_cache_free() { 2790 0) 0.518 us | __phys_addr(); 2791 0) 1.757 us | } 2792 0) 2.861 us | } 2793 2794 Example with funcgraph-tail:: 2795 2796 0) | putname() { 2797 0) | kmem_cache_free() { 2798 0) 0.518 us | __phys_addr(); 2799 0) 1.757 us | } /* kmem_cache_free() */ 2800 0) 2.861 us | } /* putname() */ 2801 2802The return value of each traced function can be displayed after 2803an equal sign "=". When encountering system call failures, it 2804can be very helpful to quickly locate the function that first 2805returns an error code. 2806 2807 - hide: echo nofuncgraph-retval > trace_options 2808 - show: echo funcgraph-retval > trace_options 2809 2810 Example with funcgraph-retval:: 2811 2812 1) | cgroup_migrate() { 2813 1) 0.651 us | cgroup_migrate_add_task(); /* = 0xffff93fcfd346c00 */ 2814 1) | cgroup_migrate_execute() { 2815 1) | cpu_cgroup_can_attach() { 2816 1) | cgroup_taskset_first() { 2817 1) 0.732 us | cgroup_taskset_next(); /* = 0xffff93fc8fb20000 */ 2818 1) 1.232 us | } /* cgroup_taskset_first = 0xffff93fc8fb20000 */ 2819 1) 0.380 us | sched_rt_can_attach(); /* = 0x0 */ 2820 1) 2.335 us | } /* cpu_cgroup_can_attach = -22 */ 2821 1) 4.369 us | } /* cgroup_migrate_execute = -22 */ 2822 1) 7.143 us | } /* cgroup_migrate = -22 */ 2823 2824The above example shows that the function cpu_cgroup_can_attach 2825returned the error code -22 firstly, then we can read the code 2826of this function to get the root cause. 2827 2828When the option funcgraph-retval-hex is not set, the return value can 2829be displayed in a smart way. Specifically, if it is an error code, 2830it will be printed in signed decimal format, otherwise it will 2831printed in hexadecimal format. 2832 2833 - smart: echo nofuncgraph-retval-hex > trace_options 2834 - hexadecimal: echo funcgraph-retval-hex > trace_options 2835 2836 Example with funcgraph-retval-hex:: 2837 2838 1) | cgroup_migrate() { 2839 1) 0.651 us | cgroup_migrate_add_task(); /* = 0xffff93fcfd346c00 */ 2840 1) | cgroup_migrate_execute() { 2841 1) | cpu_cgroup_can_attach() { 2842 1) | cgroup_taskset_first() { 2843 1) 0.732 us | cgroup_taskset_next(); /* = 0xffff93fc8fb20000 */ 2844 1) 1.232 us | } /* cgroup_taskset_first = 0xffff93fc8fb20000 */ 2845 1) 0.380 us | sched_rt_can_attach(); /* = 0x0 */ 2846 1) 2.335 us | } /* cpu_cgroup_can_attach = 0xffffffea */ 2847 1) 4.369 us | } /* cgroup_migrate_execute = 0xffffffea */ 2848 1) 7.143 us | } /* cgroup_migrate = 0xffffffea */ 2849 2850At present, there are some limitations when using the funcgraph-retval 2851option, and these limitations will be eliminated in the future: 2852 2853- Even if the function return type is void, a return value will still 2854 be printed, and you can just ignore it. 2855 2856- Even if return values are stored in multiple registers, only the 2857 value contained in the first register will be recorded and printed. 2858 To illustrate, in the x86 architecture, eax and edx are used to store 2859 a 64-bit return value, with the lower 32 bits saved in eax and the 2860 upper 32 bits saved in edx. However, only the value stored in eax 2861 will be recorded and printed. 2862 2863- In certain procedure call standards, such as arm64's AAPCS64, when a 2864 type is smaller than a GPR, it is the responsibility of the consumer 2865 to perform the narrowing, and the upper bits may contain UNKNOWN values. 2866 Therefore, it is advisable to check the code for such cases. For instance, 2867 when using a u8 in a 64-bit GPR, bits [63:8] may contain arbitrary values, 2868 especially when larger types are truncated, whether explicitly or implicitly. 2869 Here are some specific cases to illustrate this point: 2870 2871 **Case One**: 2872 2873 The function narrow_to_u8 is defined as follows:: 2874 2875 u8 narrow_to_u8(u64 val) 2876 { 2877 // implicitly truncated 2878 return val; 2879 } 2880 2881 It may be compiled to:: 2882 2883 narrow_to_u8: 2884 < ... ftrace instrumentation ... > 2885 RET 2886 2887 If you pass 0x123456789abcdef to this function and want to narrow it, 2888 it may be recorded as 0x123456789abcdef instead of 0xef. 2889 2890 **Case Two**: 2891 2892 The function error_if_not_4g_aligned is defined as follows:: 2893 2894 int error_if_not_4g_aligned(u64 val) 2895 { 2896 if (val & GENMASK(31, 0)) 2897 return -EINVAL; 2898 2899 return 0; 2900 } 2901 2902 It could be compiled to:: 2903 2904 error_if_not_4g_aligned: 2905 CBNZ w0, .Lnot_aligned 2906 RET // bits [31:0] are zero, bits 2907 // [63:32] are UNKNOWN 2908 .Lnot_aligned: 2909 MOV x0, #-EINVAL 2910 RET 2911 2912 When passing 0x2_0000_0000 to it, the return value may be recorded as 2913 0x2_0000_0000 instead of 0. 2914 2915You can put some comments on specific functions by using 2916trace_printk() For example, if you want to put a comment inside 2917the __might_sleep() function, you just have to include 2918<linux/ftrace.h> and call trace_printk() inside __might_sleep():: 2919 2920 trace_printk("I'm a comment!\n") 2921 2922will produce:: 2923 2924 1) | __might_sleep() { 2925 1) | /* I'm a comment! */ 2926 1) 1.449 us | } 2927 2928 2929You might find other useful features for this tracer in the 2930following "dynamic ftrace" section such as tracing only specific 2931functions or tasks. 2932 2933dynamic ftrace 2934-------------- 2935 2936If CONFIG_DYNAMIC_FTRACE is set, the system will run with 2937virtually no overhead when function tracing is disabled. The way 2938this works is the mcount function call (placed at the start of 2939every kernel function, produced by the -pg switch in gcc), 2940starts of pointing to a simple return. (Enabling FTRACE will 2941include the -pg switch in the compiling of the kernel.) 2942 2943At compile time every C file object is run through the 2944recordmcount program (located in the scripts directory). This 2945program will parse the ELF headers in the C object to find all 2946the locations in the .text section that call mcount. Starting 2947with gcc version 4.6, the -mfentry has been added for x86, which 2948calls "__fentry__" instead of "mcount". Which is called before 2949the creation of the stack frame. 2950 2951Note, not all sections are traced. They may be prevented by either 2952a notrace, or blocked another way and all inline functions are not 2953traced. Check the "available_filter_functions" file to see what functions 2954can be traced. 2955 2956A section called "__mcount_loc" is created that holds 2957references to all the mcount/fentry call sites in the .text section. 2958The recordmcount program re-links this section back into the 2959original object. The final linking stage of the kernel will add all these 2960references into a single table. 2961 2962On boot up, before SMP is initialized, the dynamic ftrace code 2963scans this table and updates all the locations into nops. It 2964also records the locations, which are added to the 2965available_filter_functions list. Modules are processed as they 2966are loaded and before they are executed. When a module is 2967unloaded, it also removes its functions from the ftrace function 2968list. This is automatic in the module unload code, and the 2969module author does not need to worry about it. 2970 2971When tracing is enabled, the process of modifying the function 2972tracepoints is dependent on architecture. The old method is to use 2973kstop_machine to prevent races with the CPUs executing code being 2974modified (which can cause the CPU to do undesirable things, especially 2975if the modified code crosses cache (or page) boundaries), and the nops are 2976patched back to calls. But this time, they do not call mcount 2977(which is just a function stub). They now call into the ftrace 2978infrastructure. 2979 2980The new method of modifying the function tracepoints is to place 2981a breakpoint at the location to be modified, sync all CPUs, modify 2982the rest of the instruction not covered by the breakpoint. Sync 2983all CPUs again, and then remove the breakpoint with the finished 2984version to the ftrace call site. 2985 2986Some archs do not even need to monkey around with the synchronization, 2987and can just slap the new code on top of the old without any 2988problems with other CPUs executing it at the same time. 2989 2990One special side-effect to the recording of the functions being 2991traced is that we can now selectively choose which functions we 2992wish to trace and which ones we want the mcount calls to remain 2993as nops. 2994 2995Two files are used, one for enabling and one for disabling the 2996tracing of specified functions. They are: 2997 2998 set_ftrace_filter 2999 3000and 3001 3002 set_ftrace_notrace 3003 3004A list of available functions that you can add to these files is 3005listed in: 3006 3007 available_filter_functions 3008 3009:: 3010 3011 # cat available_filter_functions 3012 put_prev_task_idle 3013 kmem_cache_create 3014 pick_next_task_rt 3015 cpus_read_lock 3016 pick_next_task_fair 3017 mutex_lock 3018 [...] 3019 3020If I am only interested in sys_nanosleep and hrtimer_interrupt:: 3021 3022 # echo sys_nanosleep hrtimer_interrupt > set_ftrace_filter 3023 # echo function > current_tracer 3024 # echo 1 > tracing_on 3025 # usleep 1 3026 # echo 0 > tracing_on 3027 # cat trace 3028 # tracer: function 3029 # 3030 # entries-in-buffer/entries-written: 5/5 #P:4 3031 # 3032 # _-----=> irqs-off 3033 # / _----=> need-resched 3034 # | / _---=> hardirq/softirq 3035 # || / _--=> preempt-depth 3036 # ||| / delay 3037 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3038 # | | | |||| | | 3039 usleep-2665 [001] .... 4186.475355: sys_nanosleep <-system_call_fastpath 3040 <idle>-0 [001] d.h1 4186.475409: hrtimer_interrupt <-smp_apic_timer_interrupt 3041 usleep-2665 [001] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt 3042 <idle>-0 [003] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt 3043 <idle>-0 [002] d.h1 4186.475427: hrtimer_interrupt <-smp_apic_timer_interrupt 3044 3045To see which functions are being traced, you can cat the file: 3046:: 3047 3048 # cat set_ftrace_filter 3049 hrtimer_interrupt 3050 sys_nanosleep 3051 3052 3053Perhaps this is not enough. The filters also allow glob(7) matching. 3054 3055 ``<match>*`` 3056 will match functions that begin with <match> 3057 ``*<match>`` 3058 will match functions that end with <match> 3059 ``*<match>*`` 3060 will match functions that have <match> in it 3061 ``<match1>*<match2>`` 3062 will match functions that begin with <match1> and end with <match2> 3063 3064.. note:: 3065 It is better to use quotes to enclose the wild cards, 3066 otherwise the shell may expand the parameters into names 3067 of files in the local directory. 3068 3069:: 3070 3071 # echo 'hrtimer_*' > set_ftrace_filter 3072 3073Produces:: 3074 3075 # tracer: function 3076 # 3077 # entries-in-buffer/entries-written: 897/897 #P:4 3078 # 3079 # _-----=> irqs-off 3080 # / _----=> need-resched 3081 # | / _---=> hardirq/softirq 3082 # || / _--=> preempt-depth 3083 # ||| / delay 3084 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3085 # | | | |||| | | 3086 <idle>-0 [003] dN.1 4228.547803: hrtimer_cancel <-tick_nohz_idle_exit 3087 <idle>-0 [003] dN.1 4228.547804: hrtimer_try_to_cancel <-hrtimer_cancel 3088 <idle>-0 [003] dN.2 4228.547805: hrtimer_force_reprogram <-__remove_hrtimer 3089 <idle>-0 [003] dN.1 4228.547805: hrtimer_forward <-tick_nohz_idle_exit 3090 <idle>-0 [003] dN.1 4228.547805: hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11 3091 <idle>-0 [003] d..1 4228.547858: hrtimer_get_next_event <-get_next_timer_interrupt 3092 <idle>-0 [003] d..1 4228.547859: hrtimer_start <-__tick_nohz_idle_enter 3093 <idle>-0 [003] d..2 4228.547860: hrtimer_force_reprogram <-__rem 3094 3095Notice that we lost the sys_nanosleep. 3096:: 3097 3098 # cat set_ftrace_filter 3099 hrtimer_run_queues 3100 hrtimer_run_pending 3101 hrtimer_setup 3102 hrtimer_cancel 3103 hrtimer_try_to_cancel 3104 hrtimer_forward 3105 hrtimer_start 3106 hrtimer_reprogram 3107 hrtimer_force_reprogram 3108 hrtimer_get_next_event 3109 hrtimer_interrupt 3110 hrtimer_nanosleep 3111 hrtimer_wakeup 3112 hrtimer_get_remaining 3113 hrtimer_get_res 3114 hrtimer_init_sleeper 3115 3116 3117This is because the '>' and '>>' act just like they do in bash. 3118To rewrite the filters, use '>' 3119To append to the filters, use '>>' 3120 3121To clear out a filter so that all functions will be recorded 3122again:: 3123 3124 # echo > set_ftrace_filter 3125 # cat set_ftrace_filter 3126 # 3127 3128Again, now we want to append. 3129 3130:: 3131 3132 # echo sys_nanosleep > set_ftrace_filter 3133 # cat set_ftrace_filter 3134 sys_nanosleep 3135 # echo 'hrtimer_*' >> set_ftrace_filter 3136 # cat set_ftrace_filter 3137 hrtimer_run_queues 3138 hrtimer_run_pending 3139 hrtimer_setup 3140 hrtimer_cancel 3141 hrtimer_try_to_cancel 3142 hrtimer_forward 3143 hrtimer_start 3144 hrtimer_reprogram 3145 hrtimer_force_reprogram 3146 hrtimer_get_next_event 3147 hrtimer_interrupt 3148 sys_nanosleep 3149 hrtimer_nanosleep 3150 hrtimer_wakeup 3151 hrtimer_get_remaining 3152 hrtimer_get_res 3153 hrtimer_init_sleeper 3154 3155 3156The set_ftrace_notrace prevents those functions from being 3157traced. 3158:: 3159 3160 # echo '*preempt*' '*lock*' > set_ftrace_notrace 3161 3162Produces:: 3163 3164 # tracer: function 3165 # 3166 # entries-in-buffer/entries-written: 39608/39608 #P:4 3167 # 3168 # _-----=> irqs-off 3169 # / _----=> need-resched 3170 # | / _---=> hardirq/softirq 3171 # || / _--=> preempt-depth 3172 # ||| / delay 3173 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3174 # | | | |||| | | 3175 bash-1994 [000] .... 4342.324896: file_ra_state_init <-do_dentry_open 3176 bash-1994 [000] .... 4342.324897: open_check_o_direct <-do_last 3177 bash-1994 [000] .... 4342.324897: ima_file_check <-do_last 3178 bash-1994 [000] .... 4342.324898: process_measurement <-ima_file_check 3179 bash-1994 [000] .... 4342.324898: ima_get_action <-process_measurement 3180 bash-1994 [000] .... 4342.324898: ima_match_policy <-ima_get_action 3181 bash-1994 [000] .... 4342.324899: do_truncate <-do_last 3182 bash-1994 [000] .... 4342.324899: setattr_should_drop_suidgid <-do_truncate 3183 bash-1994 [000] .... 4342.324899: notify_change <-do_truncate 3184 bash-1994 [000] .... 4342.324900: current_fs_time <-notify_change 3185 bash-1994 [000] .... 4342.324900: current_kernel_time <-current_fs_time 3186 bash-1994 [000] .... 4342.324900: timespec_trunc <-current_fs_time 3187 3188We can see that there's no more lock or preempt tracing. 3189 3190Selecting function filters via index 3191------------------------------------ 3192 3193Because processing of strings is expensive (the address of the function 3194needs to be looked up before comparing to the string being passed in), 3195an index can be used as well to enable functions. This is useful in the 3196case of setting thousands of specific functions at a time. By passing 3197in a list of numbers, no string processing will occur. Instead, the function 3198at the specific location in the internal array (which corresponds to the 3199functions in the "available_filter_functions" file), is selected. 3200 3201:: 3202 3203 # echo 1 > set_ftrace_filter 3204 3205Will select the first function listed in "available_filter_functions" 3206 3207:: 3208 3209 # head -1 available_filter_functions 3210 trace_initcall_finish_cb 3211 3212 # cat set_ftrace_filter 3213 trace_initcall_finish_cb 3214 3215 # head -50 available_filter_functions | tail -1 3216 x86_pmu_commit_txn 3217 3218 # echo 1 50 > set_ftrace_filter 3219 # cat set_ftrace_filter 3220 trace_initcall_finish_cb 3221 x86_pmu_commit_txn 3222 3223Dynamic ftrace with the function graph tracer 3224--------------------------------------------- 3225 3226Although what has been explained above concerns both the 3227function tracer and the function-graph-tracer, there are some 3228special features only available in the function-graph tracer. 3229 3230If you want to trace only one function and all of its children, 3231you just have to echo its name into set_graph_function:: 3232 3233 echo __do_fault > set_graph_function 3234 3235will produce the following "expanded" trace of the __do_fault() 3236function:: 3237 3238 0) | __do_fault() { 3239 0) | filemap_fault() { 3240 0) | find_lock_page() { 3241 0) 0.804 us | find_get_page(); 3242 0) | __might_sleep() { 3243 0) 1.329 us | } 3244 0) 3.904 us | } 3245 0) 4.979 us | } 3246 0) 0.653 us | _spin_lock(); 3247 0) 0.578 us | page_add_file_rmap(); 3248 0) 0.525 us | native_set_pte_at(); 3249 0) 0.585 us | _spin_unlock(); 3250 0) | unlock_page() { 3251 0) 0.541 us | page_waitqueue(); 3252 0) 0.639 us | __wake_up_bit(); 3253 0) 2.786 us | } 3254 0) + 14.237 us | } 3255 0) | __do_fault() { 3256 0) | filemap_fault() { 3257 0) | find_lock_page() { 3258 0) 0.698 us | find_get_page(); 3259 0) | __might_sleep() { 3260 0) 1.412 us | } 3261 0) 3.950 us | } 3262 0) 5.098 us | } 3263 0) 0.631 us | _spin_lock(); 3264 0) 0.571 us | page_add_file_rmap(); 3265 0) 0.526 us | native_set_pte_at(); 3266 0) 0.586 us | _spin_unlock(); 3267 0) | unlock_page() { 3268 0) 0.533 us | page_waitqueue(); 3269 0) 0.638 us | __wake_up_bit(); 3270 0) 2.793 us | } 3271 0) + 14.012 us | } 3272 3273You can also expand several functions at once:: 3274 3275 echo sys_open > set_graph_function 3276 echo sys_close >> set_graph_function 3277 3278Now if you want to go back to trace all functions you can clear 3279this special filter via:: 3280 3281 echo > set_graph_function 3282 3283 3284ftrace_enabled 3285-------------- 3286 3287Note, the proc sysctl ftrace_enable is a big on/off switch for the 3288function tracer. By default it is enabled (when function tracing is 3289enabled in the kernel). If it is disabled, all function tracing is 3290disabled. This includes not only the function tracers for ftrace, but 3291also for any other uses (perf, kprobes, stack tracing, profiling, etc). It 3292cannot be disabled if there is a callback with FTRACE_OPS_FL_PERMANENT set 3293registered. 3294 3295Please disable this with care. 3296 3297This can be disable (and enabled) with:: 3298 3299 sysctl kernel.ftrace_enabled=0 3300 sysctl kernel.ftrace_enabled=1 3301 3302 or 3303 3304 echo 0 > /proc/sys/kernel/ftrace_enabled 3305 echo 1 > /proc/sys/kernel/ftrace_enabled 3306 3307 3308Filter commands 3309--------------- 3310 3311A few commands are supported by the set_ftrace_filter interface. 3312Trace commands have the following format:: 3313 3314 <function>:<command>:<parameter> 3315 3316The following commands are supported: 3317 3318- mod: 3319 This command enables function filtering per module. The 3320 parameter defines the module. For example, if only the write* 3321 functions in the ext3 module are desired, run: 3322 3323 echo 'write*:mod:ext3' > set_ftrace_filter 3324 3325 This command interacts with the filter in the same way as 3326 filtering based on function names. Thus, adding more functions 3327 in a different module is accomplished by appending (>>) to the 3328 filter file. Remove specific module functions by prepending 3329 '!':: 3330 3331 echo '!writeback*:mod:ext3' >> set_ftrace_filter 3332 3333 Mod command supports module globbing. Disable tracing for all 3334 functions except a specific module:: 3335 3336 echo '!*:mod:!ext3' >> set_ftrace_filter 3337 3338 Disable tracing for all modules, but still trace kernel:: 3339 3340 echo '!*:mod:*' >> set_ftrace_filter 3341 3342 Enable filter only for kernel:: 3343 3344 echo '*write*:mod:!*' >> set_ftrace_filter 3345 3346 Enable filter for module globbing:: 3347 3348 echo '*write*:mod:*snd*' >> set_ftrace_filter 3349 3350- traceon/traceoff: 3351 These commands turn tracing on and off when the specified 3352 functions are hit. The parameter determines how many times the 3353 tracing system is turned on and off. If unspecified, there is 3354 no limit. For example, to disable tracing when a schedule bug 3355 is hit the first 5 times, run:: 3356 3357 echo '__schedule_bug:traceoff:5' > set_ftrace_filter 3358 3359 To always disable tracing when __schedule_bug is hit:: 3360 3361 echo '__schedule_bug:traceoff' > set_ftrace_filter 3362 3363 These commands are cumulative whether or not they are appended 3364 to set_ftrace_filter. To remove a command, prepend it by '!' 3365 and drop the parameter:: 3366 3367 echo '!__schedule_bug:traceoff:0' > set_ftrace_filter 3368 3369 The above removes the traceoff command for __schedule_bug 3370 that have a counter. To remove commands without counters:: 3371 3372 echo '!__schedule_bug:traceoff' > set_ftrace_filter 3373 3374- snapshot: 3375 Will cause a snapshot to be triggered when the function is hit. 3376 :: 3377 3378 echo 'native_flush_tlb_others:snapshot' > set_ftrace_filter 3379 3380 To only snapshot once: 3381 :: 3382 3383 echo 'native_flush_tlb_others:snapshot:1' > set_ftrace_filter 3384 3385 To remove the above commands:: 3386 3387 echo '!native_flush_tlb_others:snapshot' > set_ftrace_filter 3388 echo '!native_flush_tlb_others:snapshot:0' > set_ftrace_filter 3389 3390- enable_event/disable_event: 3391 These commands can enable or disable a trace event. Note, because 3392 function tracing callbacks are very sensitive, when these commands 3393 are registered, the trace point is activated, but disabled in 3394 a "soft" mode. That is, the tracepoint will be called, but 3395 just will not be traced. The event tracepoint stays in this mode 3396 as long as there's a command that triggers it. 3397 :: 3398 3399 echo 'try_to_wake_up:enable_event:sched:sched_switch:2' > \ 3400 set_ftrace_filter 3401 3402 The format is:: 3403 3404 <function>:enable_event:<system>:<event>[:count] 3405 <function>:disable_event:<system>:<event>[:count] 3406 3407 To remove the events commands:: 3408 3409 echo '!try_to_wake_up:enable_event:sched:sched_switch:0' > \ 3410 set_ftrace_filter 3411 echo '!schedule:disable_event:sched:sched_switch' > \ 3412 set_ftrace_filter 3413 3414- dump: 3415 When the function is hit, it will dump the contents of the ftrace 3416 ring buffer to the console. This is useful if you need to debug 3417 something, and want to dump the trace when a certain function 3418 is hit. Perhaps it's a function that is called before a triple 3419 fault happens and does not allow you to get a regular dump. 3420 3421- cpudump: 3422 When the function is hit, it will dump the contents of the ftrace 3423 ring buffer for the current CPU to the console. Unlike the "dump" 3424 command, it only prints out the contents of the ring buffer for the 3425 CPU that executed the function that triggered the dump. 3426 3427- stacktrace: 3428 When the function is hit, a stack trace is recorded. 3429 3430trace_pipe 3431---------- 3432 3433The trace_pipe outputs the same content as the trace file, but 3434the effect on the tracing is different. Every read from 3435trace_pipe is consumed. This means that subsequent reads will be 3436different. The trace is live. 3437:: 3438 3439 # echo function > current_tracer 3440 # cat trace_pipe > /tmp/trace.out & 3441 [1] 4153 3442 # echo 1 > tracing_on 3443 # usleep 1 3444 # echo 0 > tracing_on 3445 # cat trace 3446 # tracer: function 3447 # 3448 # entries-in-buffer/entries-written: 0/0 #P:4 3449 # 3450 # _-----=> irqs-off 3451 # / _----=> need-resched 3452 # | / _---=> hardirq/softirq 3453 # || / _--=> preempt-depth 3454 # ||| / delay 3455 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3456 # | | | |||| | | 3457 3458 # 3459 # cat /tmp/trace.out 3460 bash-1994 [000] .... 5281.568961: mutex_unlock <-rb_simple_write 3461 bash-1994 [000] .... 5281.568963: __mutex_unlock_slowpath <-mutex_unlock 3462 bash-1994 [000] .... 5281.568963: __fsnotify_parent <-fsnotify_modify 3463 bash-1994 [000] .... 5281.568964: fsnotify <-fsnotify_modify 3464 bash-1994 [000] .... 5281.568964: __srcu_read_lock <-fsnotify 3465 bash-1994 [000] .... 5281.568964: add_preempt_count <-__srcu_read_lock 3466 bash-1994 [000] ...1 5281.568965: sub_preempt_count <-__srcu_read_lock 3467 bash-1994 [000] .... 5281.568965: __srcu_read_unlock <-fsnotify 3468 bash-1994 [000] .... 5281.568967: sys_dup2 <-system_call_fastpath 3469 3470 3471Note, reading the trace_pipe file will block until more input is 3472added. This is contrary to the trace file. If any process opened 3473the trace file for reading, it will actually disable tracing and 3474prevent new entries from being added. The trace_pipe file does 3475not have this limitation. 3476 3477trace entries 3478------------- 3479 3480Having too much or not enough data can be troublesome in 3481diagnosing an issue in the kernel. The file buffer_size_kb is 3482used to modify the size of the internal trace buffers. The 3483number listed is the number of entries that can be recorded per 3484CPU. To know the full size, multiply the number of possible CPUs 3485with the number of entries. 3486:: 3487 3488 # cat buffer_size_kb 3489 1408 (units kilobytes) 3490 3491Or simply read buffer_total_size_kb 3492:: 3493 3494 # cat buffer_total_size_kb 3495 5632 3496 3497To modify the buffer, simple echo in a number (in 1024 byte segments). 3498:: 3499 3500 # echo 10000 > buffer_size_kb 3501 # cat buffer_size_kb 3502 10000 (units kilobytes) 3503 3504It will try to allocate as much as possible. If you allocate too 3505much, it can cause Out-Of-Memory to trigger. 3506:: 3507 3508 # echo 1000000000000 > buffer_size_kb 3509 -bash: echo: write error: Cannot allocate memory 3510 # cat buffer_size_kb 3511 85 3512 3513The per_cpu buffers can be changed individually as well: 3514:: 3515 3516 # echo 10000 > per_cpu/cpu0/buffer_size_kb 3517 # echo 100 > per_cpu/cpu1/buffer_size_kb 3518 3519When the per_cpu buffers are not the same, the buffer_size_kb 3520at the top level will just show an X 3521:: 3522 3523 # cat buffer_size_kb 3524 X 3525 3526This is where the buffer_total_size_kb is useful: 3527:: 3528 3529 # cat buffer_total_size_kb 3530 12916 3531 3532Writing to the top level buffer_size_kb will reset all the buffers 3533to be the same again. 3534 3535Snapshot 3536-------- 3537CONFIG_TRACER_SNAPSHOT makes a generic snapshot feature 3538available to all non latency tracers. (Latency tracers which 3539record max latency, such as "irqsoff" or "wakeup", can't use 3540this feature, since those are already using the snapshot 3541mechanism internally.) 3542 3543Snapshot preserves a current trace buffer at a particular point 3544in time without stopping tracing. Ftrace swaps the current 3545buffer with a spare buffer, and tracing continues in the new 3546current (=previous spare) buffer. 3547 3548The following tracefs files in "tracing" are related to this 3549feature: 3550 3551 snapshot: 3552 3553 This is used to take a snapshot and to read the output 3554 of the snapshot. Echo 1 into this file to allocate a 3555 spare buffer and to take a snapshot (swap), then read 3556 the snapshot from this file in the same format as 3557 "trace" (described above in the section "The File 3558 System"). Both reads snapshot and tracing are executable 3559 in parallel. When the spare buffer is allocated, echoing 3560 0 frees it, and echoing else (positive) values clear the 3561 snapshot contents. 3562 More details are shown in the table below. 3563 3564 +--------------+------------+------------+------------+ 3565 |status\\input | 0 | 1 | else | 3566 +==============+============+============+============+ 3567 |not allocated |(do nothing)| alloc+swap |(do nothing)| 3568 +--------------+------------+------------+------------+ 3569 |allocated | free | swap | clear | 3570 +--------------+------------+------------+------------+ 3571 3572Here is an example of using the snapshot feature. 3573:: 3574 3575 # echo 1 > events/sched/enable 3576 # echo 1 > snapshot 3577 # cat snapshot 3578 # tracer: nop 3579 # 3580 # entries-in-buffer/entries-written: 71/71 #P:8 3581 # 3582 # _-----=> irqs-off 3583 # / _----=> need-resched 3584 # | / _---=> hardirq/softirq 3585 # || / _--=> preempt-depth 3586 # ||| / delay 3587 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3588 # | | | |||| | | 3589 <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 3590 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 3591 [...] 3592 <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 3593 3594 # cat trace 3595 # tracer: nop 3596 # 3597 # entries-in-buffer/entries-written: 77/77 #P:8 3598 # 3599 # _-----=> irqs-off 3600 # / _----=> need-resched 3601 # | / _---=> hardirq/softirq 3602 # || / _--=> preempt-depth 3603 # ||| / delay 3604 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3605 # | | | |||| | | 3606 <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 3607 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 3608 [...] 3609 3610 3611If you try to use this snapshot feature when current tracer is 3612one of the latency tracers, you will get the following results. 3613:: 3614 3615 # echo wakeup > current_tracer 3616 # echo 1 > snapshot 3617 bash: echo: write error: Device or resource busy 3618 # cat snapshot 3619 cat: snapshot: Device or resource busy 3620 3621 3622Instances 3623--------- 3624In the tracefs tracing directory, there is a directory called "instances". 3625This directory can have new directories created inside of it using 3626mkdir, and removing directories with rmdir. The directory created 3627with mkdir in this directory will already contain files and other 3628directories after it is created. 3629:: 3630 3631 # mkdir instances/foo 3632 # ls instances/foo 3633 buffer_size_kb buffer_total_size_kb events free_buffer per_cpu 3634 set_event snapshot trace trace_clock trace_marker trace_options 3635 trace_pipe tracing_on 3636 3637As you can see, the new directory looks similar to the tracing directory 3638itself. In fact, it is very similar, except that the buffer and 3639events are agnostic from the main directory, or from any other 3640instances that are created. 3641 3642The files in the new directory work just like the files with the 3643same name in the tracing directory except the buffer that is used 3644is a separate and new buffer. The files affect that buffer but do not 3645affect the main buffer with the exception of trace_options. Currently, 3646the trace_options affect all instances and the top level buffer 3647the same, but this may change in future releases. That is, options 3648may become specific to the instance they reside in. 3649 3650Notice that none of the function tracer files are there, nor is 3651current_tracer and available_tracers. This is because the buffers 3652can currently only have events enabled for them. 3653:: 3654 3655 # mkdir instances/foo 3656 # mkdir instances/bar 3657 # mkdir instances/zoot 3658 # echo 100000 > buffer_size_kb 3659 # echo 1000 > instances/foo/buffer_size_kb 3660 # echo 5000 > instances/bar/per_cpu/cpu1/buffer_size_kb 3661 # echo function > current_trace 3662 # echo 1 > instances/foo/events/sched/sched_wakeup/enable 3663 # echo 1 > instances/foo/events/sched/sched_wakeup_new/enable 3664 # echo 1 > instances/foo/events/sched/sched_switch/enable 3665 # echo 1 > instances/bar/events/irq/enable 3666 # echo 1 > instances/zoot/events/syscalls/enable 3667 # cat trace_pipe 3668 CPU:2 [LOST 11745 EVENTS] 3669 bash-2044 [002] .... 10594.481032: _raw_spin_lock_irqsave <-get_page_from_freelist 3670 bash-2044 [002] d... 10594.481032: add_preempt_count <-_raw_spin_lock_irqsave 3671 bash-2044 [002] d..1 10594.481032: __rmqueue <-get_page_from_freelist 3672 bash-2044 [002] d..1 10594.481033: _raw_spin_unlock <-get_page_from_freelist 3673 bash-2044 [002] d..1 10594.481033: sub_preempt_count <-_raw_spin_unlock 3674 bash-2044 [002] d... 10594.481033: get_pageblock_flags_group <-get_pageblock_migratetype 3675 bash-2044 [002] d... 10594.481034: __mod_zone_page_state <-get_page_from_freelist 3676 bash-2044 [002] d... 10594.481034: zone_statistics <-get_page_from_freelist 3677 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics 3678 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics 3679 bash-2044 [002] .... 10594.481035: arch_dup_task_struct <-copy_process 3680 [...] 3681 3682 # cat instances/foo/trace_pipe 3683 bash-1998 [000] d..4 136.676759: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000 3684 bash-1998 [000] dN.4 136.676760: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000 3685 <idle>-0 [003] d.h3 136.676906: sched_wakeup: comm=rcu_preempt pid=9 prio=120 success=1 target_cpu=003 3686 <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 3687 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 3688 bash-1998 [000] d..4 136.677014: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000 3689 bash-1998 [000] dN.4 136.677016: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000 3690 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 3691 kworker/0:1-59 [000] d..4 136.677022: sched_wakeup: comm=sshd pid=1995 prio=120 success=1 target_cpu=001 3692 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 3693 [...] 3694 3695 # cat instances/bar/trace_pipe 3696 migration/1-14 [001] d.h3 138.732674: softirq_raise: vec=3 [action=NET_RX] 3697 <idle>-0 [001] dNh3 138.732725: softirq_raise: vec=3 [action=NET_RX] 3698 bash-1998 [000] d.h1 138.733101: softirq_raise: vec=1 [action=TIMER] 3699 bash-1998 [000] d.h1 138.733102: softirq_raise: vec=9 [action=RCU] 3700 bash-1998 [000] ..s2 138.733105: softirq_entry: vec=1 [action=TIMER] 3701 bash-1998 [000] ..s2 138.733106: softirq_exit: vec=1 [action=TIMER] 3702 bash-1998 [000] ..s2 138.733106: softirq_entry: vec=9 [action=RCU] 3703 bash-1998 [000] ..s2 138.733109: softirq_exit: vec=9 [action=RCU] 3704 sshd-1995 [001] d.h1 138.733278: irq_handler_entry: irq=21 name=uhci_hcd:usb4 3705 sshd-1995 [001] d.h1 138.733280: irq_handler_exit: irq=21 ret=unhandled 3706 sshd-1995 [001] d.h1 138.733281: irq_handler_entry: irq=21 name=eth0 3707 sshd-1995 [001] d.h1 138.733283: irq_handler_exit: irq=21 ret=handled 3708 [...] 3709 3710 # cat instances/zoot/trace 3711 # tracer: nop 3712 # 3713 # entries-in-buffer/entries-written: 18996/18996 #P:4 3714 # 3715 # _-----=> irqs-off 3716 # / _----=> need-resched 3717 # | / _---=> hardirq/softirq 3718 # || / _--=> preempt-depth 3719 # ||| / delay 3720 # TASK-PID CPU# |||| TIMESTAMP FUNCTION 3721 # | | | |||| | | 3722 bash-1998 [000] d... 140.733501: sys_write -> 0x2 3723 bash-1998 [000] d... 140.733504: sys_dup2(oldfd: a, newfd: 1) 3724 bash-1998 [000] d... 140.733506: sys_dup2 -> 0x1 3725 bash-1998 [000] d... 140.733508: sys_fcntl(fd: a, cmd: 1, arg: 0) 3726 bash-1998 [000] d... 140.733509: sys_fcntl -> 0x1 3727 bash-1998 [000] d... 140.733510: sys_close(fd: a) 3728 bash-1998 [000] d... 140.733510: sys_close -> 0x0 3729 bash-1998 [000] d... 140.733514: sys_rt_sigprocmask(how: 0, nset: 0, oset: 6e2768, sigsetsize: 8) 3730 bash-1998 [000] d... 140.733515: sys_rt_sigprocmask -> 0x0 3731 bash-1998 [000] d... 140.733516: sys_rt_sigaction(sig: 2, act: 7fff718846f0, oact: 7fff71884650, sigsetsize: 8) 3732 bash-1998 [000] d... 140.733516: sys_rt_sigaction -> 0x0 3733 3734You can see that the trace of the top most trace buffer shows only 3735the function tracing. The foo instance displays wakeups and task 3736switches. 3737 3738To remove the instances, simply delete their directories: 3739:: 3740 3741 # rmdir instances/foo 3742 # rmdir instances/bar 3743 # rmdir instances/zoot 3744 3745Note, if a process has a trace file open in one of the instance 3746directories, the rmdir will fail with EBUSY. 3747 3748 3749Stack trace 3750----------- 3751Since the kernel has a fixed sized stack, it is important not to 3752waste it in functions. A kernel developer must be conscious of 3753what they allocate on the stack. If they add too much, the system 3754can be in danger of a stack overflow, and corruption will occur, 3755usually leading to a system panic. 3756 3757There are some tools that check this, usually with interrupts 3758periodically checking usage. But if you can perform a check 3759at every function call that will become very useful. As ftrace provides 3760a function tracer, it makes it convenient to check the stack size 3761at every function call. This is enabled via the stack tracer. 3762 3763CONFIG_STACK_TRACER enables the ftrace stack tracing functionality. 3764To enable it, write a '1' into /proc/sys/kernel/stack_tracer_enabled. 3765:: 3766 3767 # echo 1 > /proc/sys/kernel/stack_tracer_enabled 3768 3769You can also enable it from the kernel command line to trace 3770the stack size of the kernel during boot up, by adding "stacktrace" 3771to the kernel command line parameter. 3772 3773After running it for a few minutes, the output looks like: 3774:: 3775 3776 # cat stack_max_size 3777 2928 3778 3779 # cat stack_trace 3780 Depth Size Location (18 entries) 3781 ----- ---- -------- 3782 0) 2928 224 update_sd_lb_stats+0xbc/0x4ac 3783 1) 2704 160 find_busiest_group+0x31/0x1f1 3784 2) 2544 256 load_balance+0xd9/0x662 3785 3) 2288 80 idle_balance+0xbb/0x130 3786 4) 2208 128 __schedule+0x26e/0x5b9 3787 5) 2080 16 schedule+0x64/0x66 3788 6) 2064 128 schedule_timeout+0x34/0xe0 3789 7) 1936 112 wait_for_common+0x97/0xf1 3790 8) 1824 16 wait_for_completion+0x1d/0x1f 3791 9) 1808 128 flush_work+0xfe/0x119 3792 10) 1680 16 tty_flush_to_ldisc+0x1e/0x20 3793 11) 1664 48 input_available_p+0x1d/0x5c 3794 12) 1616 48 n_tty_poll+0x6d/0x134 3795 13) 1568 64 tty_poll+0x64/0x7f 3796 14) 1504 880 do_select+0x31e/0x511 3797 15) 624 400 core_sys_select+0x177/0x216 3798 16) 224 96 sys_select+0x91/0xb9 3799 17) 128 128 system_call_fastpath+0x16/0x1b 3800 3801Note, if -mfentry is being used by gcc, functions get traced before 3802they set up the stack frame. This means that leaf level functions 3803are not tested by the stack tracer when -mfentry is used. 3804 3805Currently, -mfentry is used by gcc 4.6.0 and above on x86 only. 3806 3807More 3808---- 3809More details can be found in the source code, in the `kernel/trace/*.c` files. 3810