| blob | 2b74f96d09d531aef703ff40af345ea48fabe476 |
1 ========================
2 ftrace - Function Tracer
3 ========================
5 Copyright 2008 Red Hat Inc.
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.
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>
18 Introduction
19 ------------
21 Ftrace is an internal tracer designed to help out developers and
22 designers of systems to find what is going on inside the kernel.
23 It can be used for debugging or analyzing latencies and
24 performance issues that take place outside of user-space.
26 Although ftrace is typically considered the function tracer, it
27 is really a framework of several assorted tracing utilities.
28 There's latency tracing to examine what occurs between interrupts
29 disabled and enabled, as well as for preemption and from a time
30 a task is woken to the task is actually scheduled in.
32 One of the most common uses of ftrace is the event tracing.
33 Throughout the kernel is hundreds of static event points that
34 can be enabled via the tracefs file system to see what is
35 going on in certain parts of the kernel.
37 See events.rst for more information.
40 Implementation Details
41 ----------------------
43 See Documentation/trace/ftrace-design.rst for details for arch porters and such.
46 The File System
47 ---------------
49 Ftrace uses the tracefs file system to hold the control files as
50 well as the files to display output.
52 When tracefs is configured into the kernel (which selecting any ftrace
53 option will do) the directory /sys/kernel/tracing will be created. To mount
54 this directory, you can add to your /etc/fstab file::
56 tracefs /sys/kernel/tracing tracefs defaults 0 0
58 Or you can mount it at run time with::
60 mount -t tracefs nodev /sys/kernel/tracing
62 For quicker access to that directory you may want to make a soft link to
63 it::
65 ln -s /sys/kernel/tracing /tracing
67 .. attention::
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:
74 /sys/kernel/debug/tracing
76 All files located in the tracefs file system will be located in that
77 debugfs file system directory as well.
79 .. attention::
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.
87 That's it! (assuming that you have ftrace configured into your kernel)
89 After mounting tracefs you will have access to the control and output files
90 of ftrace. Here is a list of some of the key files:
93 Note: all time values are in microseconds.
95 current_tracer:
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.
101 available_tracers:
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.
108 tracing_on:
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.
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.
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.
125 trace:
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.
136 trace_pipe:
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.
150 trace_options:
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).
157 options:
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.
164 tracing_max_latency:
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).
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.
176 tracing_thresh:
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)
183 buffer_percent:
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::
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.
198 buffer_size_kb:
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. )
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".
217 buffer_total_size_kb:
219 This displays the total combined size of all the trace buffers.
221 buffer_subbuf_size_kb:
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.
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.
236 Changing the sub buffer size allows for events to be larger than the
237 page size.
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.
242 free_buffer:
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".
252 It may also stop tracing if disable_on_free option is set.
254 tracing_cpumask:
256 This is a mask that lets the user only trace on specified CPUs.
257 The format is a hex string representing the CPUs.
259 set_ftrace_filter:
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").
272 The functions listed in "available_filter_functions" are what
273 can be written into this file.
275 This interface also allows for commands to be used. See the
276 "Filter commands" section for more details.
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.
284 set_ftrace_notrace:
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.
291 set_ftrace_pid:
293 Have the function tracer only trace the threads whose PID are
294 listed in this file.
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.
302 set_ftrace_notrace_pid:
304 Have the function tracer ignore threads whose PID are listed in
305 this file.
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.
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.
316 set_event_pid:
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.
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.
327 set_event_notrace_pid:
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.
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.
340 set_graph_function:
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.
348 set_graph_notrace:
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.
355 available_filter_functions:
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.)
363 available_filter_functions_addrs:
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.
369 dyn_ftrace_total_info:
371 This file is for debugging purposes. The number of functions that
372 have been converted to nops and are available to be traced.
374 enabled_functions:
376 This file is more for debugging ftrace, but can also be useful
377 in seeing if any function has a callback attached to it.
378 Not only does the trace infrastructure use ftrace function
379 trace utility, but other subsystems might too. This file
380 displays all functions that have a callback attached to them
381 as well as the number of callbacks that have been attached.
382 Note, a callback may also call multiple functions which will
383 not be listed in this count.
385 If the callback registered to be traced by a function with
386 the "save regs" attribute (thus even more overhead), a 'R'
387 will be displayed on the same line as the function that
388 is returning registers.
390 If the callback registered to be traced by a function with
391 the "ip modify" attribute (thus the regs->ip can be changed),
392 an 'I' will be displayed on the same line as the function that
393 can be overridden.
395 If a non ftrace trampoline is attached (BPF) a 'D' will be displayed.
396 Note, normal ftrace trampolines can also be attached, but only one
397 "direct" trampoline can be attached to a given function at a time.
399 Some architectures can not call direct trampolines, but instead have
400 the ftrace ops function located above the function entry point. In
401 such cases an 'O' will be displayed.
403 If a function had either the "ip modify" or a "direct" call attached to
404 it in the past, a 'M' will be shown. This flag is never cleared. It is
405 used to know if a function was every modified by the ftrace infrastructure,
406 and can be used for debugging.
408 If the architecture supports it, it will also show what callback
409 is being directly called by the function. If the count is greater
410 than 1 it most likely will be ftrace_ops_list_func().
412 If the callback of a function jumps to a trampoline that is
413 specific to the callback and which is not the standard trampoline,
414 its address will be printed as well as the function that the
415 trampoline calls.
417 touched_functions:
419 This file contains all the functions that ever had a function callback
420 to it via the ftrace infrastructure. It has the same format as
421 enabled_functions but shows all functions that have every been
422 traced.
424 To see any function that has every been modified by "ip modify" or a
425 direct trampoline, one can perform the following command:
427 grep ' M ' /sys/kernel/tracing/touched_functions
429 function_profile_enabled:
431 When set it will enable all functions with either the function
432 tracer, or if configured, the function graph tracer. It will
433 keep a histogram of the number of functions that were called
434 and if the function graph tracer was configured, it will also keep
435 track of the time spent in those functions. The histogram
436 content can be displayed in the files:
438 trace_stat/function<cpu> ( function0, function1, etc).
440 trace_stat:
442 A directory that holds different tracing stats.
444 kprobe_events:
446 Enable dynamic trace points. See kprobetrace.rst.
448 kprobe_profile:
450 Dynamic trace points stats. See kprobetrace.rst.
452 max_graph_depth:
454 Used with the function graph tracer. This is the max depth
455 it will trace into a function. Setting this to a value of
456 one will show only the first kernel function that is called
457 from user space.
459 printk_formats:
461 This is for tools that read the raw format files. If an event in
462 the ring buffer references a string, only a pointer to the string
463 is recorded into the buffer and not the string itself. This prevents
464 tools from knowing what that string was. This file displays the string
465 and address for the string allowing tools to map the pointers to what
466 the strings were.
468 saved_cmdlines:
470 Only the pid of the task is recorded in a trace event unless
471 the event specifically saves the task comm as well. Ftrace
472 makes a cache of pid mappings to comms to try to display
473 comms for events. If a pid for a comm is not listed, then
474 "<...>" is displayed in the output.
476 If the option "record-cmd" is set to "0", then comms of tasks
477 will not be saved during recording. By default, it is enabled.
479 saved_cmdlines_size:
481 By default, 128 comms are saved (see "saved_cmdlines" above). To
482 increase or decrease the amount of comms that are cached, echo
483 the number of comms to cache into this file.
485 saved_tgids:
487 If the option "record-tgid" is set, on each scheduling context switch
488 the Task Group ID of a task is saved in a table mapping the PID of
489 the thread to its TGID. By default, the "record-tgid" option is
490 disabled.
492 snapshot:
494 This displays the "snapshot" buffer and also lets the user
495 take a snapshot of the current running trace.
496 See the "Snapshot" section below for more details.
498 stack_max_size:
500 When the stack tracer is activated, this will display the
501 maximum stack size it has encountered.
502 See the "Stack Trace" section below.
504 stack_trace:
506 This displays the stack back trace of the largest stack
507 that was encountered when the stack tracer is activated.
508 See the "Stack Trace" section below.
510 stack_trace_filter:
512 This is similar to "set_ftrace_filter" but it limits what
513 functions the stack tracer will check.
515 trace_clock:
517 Whenever an event is recorded into the ring buffer, a
518 "timestamp" is added. This stamp comes from a specified
519 clock. By default, ftrace uses the "local" clock. This
520 clock is very fast and strictly per cpu, but on some
521 systems it may not be monotonic with respect to other
522 CPUs. In other words, the local clocks may not be in sync
523 with local clocks on other CPUs.
525 Usual clocks for tracing::
527 # cat trace_clock
528 [local] global counter x86-tsc
530 The clock with the square brackets around it is the one in effect.
532 local:
533 Default clock, but may not be in sync across CPUs
535 global:
536 This clock is in sync with all CPUs but may
537 be a bit slower than the local clock.
539 counter:
540 This is not a clock at all, but literally an atomic
541 counter. It counts up one by one, but is in sync
542 with all CPUs. This is useful when you need to
543 know exactly the order events occurred with respect to
544 each other on different CPUs.
546 uptime:
547 This uses the jiffies counter and the time stamp
548 is relative to the time since boot up.
550 perf:
551 This makes ftrace use the same clock that perf uses.
552 Eventually perf will be able to read ftrace buffers
553 and this will help out in interleaving the data.
555 x86-tsc:
556 Architectures may define their own clocks. For
557 example, x86 uses its own TSC cycle clock here.
559 ppc-tb:
560 This uses the powerpc timebase register value.
561 This is in sync across CPUs and can also be used
562 to correlate events across hypervisor/guest if
563 tb_offset is known.
565 mono:
566 This uses the fast monotonic clock (CLOCK_MONOTONIC)
567 which is monotonic and is subject to NTP rate adjustments.
569 mono_raw:
570 This is the raw monotonic clock (CLOCK_MONOTONIC_RAW)
571 which is monotonic but is not subject to any rate adjustments
572 and ticks at the same rate as the hardware clocksource.
574 boot:
575 This is the boot clock (CLOCK_BOOTTIME) and is based on the
576 fast monotonic clock, but also accounts for time spent in
577 suspend. Since the clock access is designed for use in
578 tracing in the suspend path, some side effects are possible
579 if clock is accessed after the suspend time is accounted before
580 the fast mono clock is updated. In this case, the clock update
581 appears to happen slightly sooner than it normally would have.
582 Also on 32-bit systems, it's possible that the 64-bit boot offset
583 sees a partial update. These effects are rare and post
584 processing should be able to handle them. See comments in the
585 ktime_get_boot_fast_ns() function for more information.
587 tai:
588 This is the tai clock (CLOCK_TAI) and is derived from the wall-
589 clock time. However, this clock does not experience
590 discontinuities and backwards jumps caused by NTP inserting leap
591 seconds. Since the clock access is designed for use in tracing,
592 side effects are possible. The clock access may yield wrong
593 readouts in case the internal TAI offset is updated e.g., caused
594 by setting the system time or using adjtimex() with an offset.
595 These effects are rare and post processing should be able to
596 handle them. See comments in the ktime_get_tai_fast_ns()
597 function for more information.
599 To set a clock, simply echo the clock name into this file::
601 # echo global > trace_clock
603 Setting a clock clears the ring buffer content as well as the
604 "snapshot" buffer.
606 trace_marker:
608 This is a very useful file for synchronizing user space
609 with events happening in the kernel. Writing strings into
610 this file will be written into the ftrace buffer.
612 It is useful in applications to open this file at the start
613 of the application and just reference the file descriptor
614 for the file::
616 void trace_write(const char *fmt, ...)
617 {
618 va_list ap;
619 char buf[256];
620 int n;
622 if (trace_fd < 0)
623 return;
625 va_start(ap, fmt);
626 n = vsnprintf(buf, 256, fmt, ap);
627 va_end(ap);
629 write(trace_fd, buf, n);
630 }
632 start::
634 trace_fd = open("trace_marker", O_WRONLY);
636 Note: Writing into the trace_marker file can also initiate triggers
637 that are written into /sys/kernel/tracing/events/ftrace/print/trigger
638 See "Event triggers" in Documentation/trace/events.rst and an
639 example in Documentation/trace/histogram.rst (Section 3.)
641 trace_marker_raw:
643 This is similar to trace_marker above, but is meant for binary data
644 to be written to it, where a tool can be used to parse the data
645 from trace_pipe_raw.
647 uprobe_events:
649 Add dynamic tracepoints in programs.
650 See uprobetracer.rst
652 uprobe_profile:
654 Uprobe statistics. See uprobetrace.txt
656 instances:
658 This is a way to make multiple trace buffers where different
659 events can be recorded in different buffers.
660 See "Instances" section below.
662 events:
664 This is the trace event directory. It holds event tracepoints
665 (also known as static tracepoints) that have been compiled
666 into the kernel. It shows what event tracepoints exist
667 and how they are grouped by system. There are "enable"
668 files at various levels that can enable the tracepoints
669 when a "1" is written to them.
671 See events.rst for more information.
673 set_event:
675 By echoing in the event into this file, will enable that event.
677 See events.rst for more information.
679 available_events:
681 A list of events that can be enabled in tracing.
683 See events.rst for more information.
685 timestamp_mode:
687 Certain tracers may change the timestamp mode used when
688 logging trace events into the event buffer. Events with
689 different modes can coexist within a buffer but the mode in
690 effect when an event is logged determines which timestamp mode
691 is used for that event. The default timestamp mode is
692 'delta'.
694 Usual timestamp modes for tracing:
696 # cat timestamp_mode
697 [delta] absolute
699 The timestamp mode with the square brackets around it is the
700 one in effect.
702 delta: Default timestamp mode - timestamp is a delta against
703 a per-buffer timestamp.
705 absolute: The timestamp is a full timestamp, not a delta
706 against some other value. As such it takes up more
707 space and is less efficient.
709 hwlat_detector:
711 Directory for the Hardware Latency Detector.
712 See "Hardware Latency Detector" section below.
714 per_cpu:
716 This is a directory that contains the trace per_cpu information.
718 per_cpu/cpu0/buffer_size_kb:
720 The ftrace buffer is defined per_cpu. That is, there's a separate
721 buffer for each CPU to allow writes to be done atomically,
722 and free from cache bouncing. These buffers may have different
723 size buffers. This file is similar to the buffer_size_kb
724 file, but it only displays or sets the buffer size for the
725 specific CPU. (here cpu0).
727 per_cpu/cpu0/trace:
729 This is similar to the "trace" file, but it will only display
730 the data specific for the CPU. If written to, it only clears
731 the specific CPU buffer.
733 per_cpu/cpu0/trace_pipe
735 This is similar to the "trace_pipe" file, and is a consuming
736 read, but it will only display (and consume) the data specific
737 for the CPU.
739 per_cpu/cpu0/trace_pipe_raw
741 For tools that can parse the ftrace ring buffer binary format,
742 the trace_pipe_raw file can be used to extract the data
743 from the ring buffer directly. With the use of the splice()
744 system call, the buffer data can be quickly transferred to
745 a file or to the network where a server is collecting the
746 data.
748 Like trace_pipe, this is a consuming reader, where multiple
749 reads will always produce different data.
751 per_cpu/cpu0/snapshot:
753 This is similar to the main "snapshot" file, but will only
754 snapshot the current CPU (if supported). It only displays
755 the content of the snapshot for a given CPU, and if
756 written to, only clears this CPU buffer.
758 per_cpu/cpu0/snapshot_raw:
760 Similar to the trace_pipe_raw, but will read the binary format
761 from the snapshot buffer for the given CPU.
763 per_cpu/cpu0/stats:
765 This displays certain stats about the ring buffer:
767 entries:
768 The number of events that are still in the buffer.
770 overrun:
771 The number of lost events due to overwriting when
772 the buffer was full.
774 commit overrun:
775 Should always be zero.
776 This gets set if so many events happened within a nested
777 event (ring buffer is re-entrant), that it fills the
778 buffer and starts dropping events.
780 bytes:
781 Bytes actually read (not overwritten).
783 oldest event ts:
784 The oldest timestamp in the buffer
786 now ts:
787 The current timestamp
789 dropped events:
790 Events lost due to overwrite option being off.
792 read events:
793 The number of events read.
795 The Tracers
796 -----------
798 Here is the list of current tracers that may be configured.
800 "function"
802 Function call tracer to trace all kernel functions.
804 "function_graph"
806 Similar to the function tracer except that the
807 function tracer probes the functions on their entry
808 whereas the function graph tracer traces on both entry
809 and exit of the functions. It then provides the ability
810 to draw a graph of function calls similar to C code
811 source.
813 Note that the function graph calculates the timings of when the
814 function starts and returns internally and for each instance. If
815 there are two instances that run function graph tracer and traces
816 the same functions, the length of the timings may be slightly off as
817 each read the timestamp separately and not at the same time.
819 "blk"
821 The block tracer. The tracer used by the blktrace user
822 application.
824 "hwlat"
826 The Hardware Latency tracer is used to detect if the hardware
827 produces any latency. See "Hardware Latency Detector" section
828 below.
830 "irqsoff"
832 Traces the areas that disable interrupts and saves
833 the trace with the longest max latency.
834 See tracing_max_latency. When a new max is recorded,
835 it replaces the old trace. It is best to view this
836 trace with the latency-format option enabled, which
837 happens automatically when the tracer is selected.
839 "preemptoff"
841 Similar to irqsoff but traces and records the amount of
842 time for which preemption is disabled.
844 "preemptirqsoff"
846 Similar to irqsoff and preemptoff, but traces and
847 records the largest time for which irqs and/or preemption
848 is disabled.
850 "wakeup"
852 Traces and records the max latency that it takes for
853 the highest priority task to get scheduled after
854 it has been woken up.
855 Traces all tasks as an average developer would expect.
857 "wakeup_rt"
859 Traces and records the max latency that it takes for just
860 RT tasks (as the current "wakeup" does). This is useful
861 for those interested in wake up timings of RT tasks.
863 "wakeup_dl"
865 Traces and records the max latency that it takes for
866 a SCHED_DEADLINE task to be woken (as the "wakeup" and
867 "wakeup_rt" does).
869 "mmiotrace"
871 A special tracer that is used to trace binary module.
872 It will trace all the calls that a module makes to the
873 hardware. Everything it writes and reads from the I/O
874 as well.
876 "branch"
878 This tracer can be configured when tracing likely/unlikely
879 calls within the kernel. It will trace when a likely and
880 unlikely branch is hit and if it was correct in its prediction
881 of being correct.
883 "nop"
885 This is the "trace nothing" tracer. To remove all
886 tracers from tracing simply echo "nop" into
887 current_tracer.
889 Error conditions
890 ----------------
892 For most ftrace commands, failure modes are obvious and communicated
893 using standard return codes.
895 For other more involved commands, extended error information may be
896 available via the tracing/error_log file. For the commands that
897 support it, reading the tracing/error_log file after an error will
898 display more detailed information about what went wrong, if
899 information is available. The tracing/error_log file is a circular
900 error log displaying a small number (currently, 8) of ftrace errors
901 for the last (8) failed commands.
903 The extended error information and usage takes the form shown in
904 this example::
906 # echo xxx > /sys/kernel/tracing/events/sched/sched_wakeup/trigger
907 echo: write error: Invalid argument
909 # cat /sys/kernel/tracing/error_log
910 [ 5348.887237] location: error: Couldn't yyy: zzz
911 Command: xxx
912 ^
913 [ 7517.023364] location: error: Bad rrr: sss
914 Command: ppp qqq
915 ^
917 To clear the error log, echo the empty string into it::
919 # echo > /sys/kernel/tracing/error_log
921 Examples of using the tracer
922 ----------------------------
924 Here are typical examples of using the tracers when controlling
925 them only with the tracefs interface (without using any
926 user-land utilities).
928 Output format:
929 --------------
931 Here is an example of the output format of the file "trace"::
933 # tracer: function
934 #
935 # entries-in-buffer/entries-written: 140080/250280 #P:4
936 #
937 # _-----=> irqs-off
938 # / _----=> need-resched
939 # | / _---=> hardirq/softirq
940 # || / _--=> preempt-depth
941 # ||| / delay
942 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
943 # | | | |||| | |
944 bash-1977 [000] .... 17284.993652: sys_close <-system_call_fastpath
945 bash-1977 [000] .... 17284.993653: __close_fd <-sys_close
946 bash-1977 [000] .... 17284.993653: _raw_spin_lock <-__close_fd
947 sshd-1974 [003] .... 17284.993653: __srcu_read_unlock <-fsnotify
948 bash-1977 [000] .... 17284.993654: add_preempt_count <-_raw_spin_lock
949 bash-1977 [000] ...1 17284.993655: _raw_spin_unlock <-__close_fd
950 bash-1977 [000] ...1 17284.993656: sub_preempt_count <-_raw_spin_unlock
951 bash-1977 [000] .... 17284.993657: filp_close <-__close_fd
952 bash-1977 [000] .... 17284.993657: dnotify_flush <-filp_close
953 sshd-1974 [003] .... 17284.993658: sys_select <-system_call_fastpath
954 ....
956 A header is printed with the tracer name that is represented by
957 the trace. In this case the tracer is "function". Then it shows the
958 number of events in the buffer as well as the total number of entries
959 that were written. The difference is the number of entries that were
960 lost due to the buffer filling up (250280 - 140080 = 110200 events
961 lost).
963 The header explains the content of the events. Task name "bash", the task
964 PID "1977", the CPU that it was running on "000", the latency format
965 (explained below), the timestamp in <secs>.<usecs> format, the
966 function name that was traced "sys_close" and the parent function that
967 called this function "system_call_fastpath". The timestamp is the time
968 at which the function was entered.
970 Latency trace format
971 --------------------
973 When the latency-format option is enabled or when one of the latency
974 tracers is set, the trace file gives somewhat more information to see
975 why a latency happened. Here is a typical trace::
977 # tracer: irqsoff
978 #
979 # irqsoff latency trace v1.1.5 on 3.8.0-test+
980 # --------------------------------------------------------------------
981 # latency: 259 us, #4/4, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
982 # -----------------
983 # | task: ps-6143 (uid:0 nice:0 policy:0 rt_prio:0)
984 # -----------------
985 # => started at: __lock_task_sighand
986 # => ended at: _raw_spin_unlock_irqrestore
987 #
988 #
989 # _------=> CPU#
990 # / _-----=> irqs-off
991 # | / _----=> need-resched
992 # || / _---=> hardirq/softirq
993 # ||| / _--=> preempt-depth
994 # |||| / delay
995 # cmd pid ||||| time | caller
996 # \ / ||||| \ | /
997 ps-6143 2d... 0us!: trace_hardirqs_off <-__lock_task_sighand
998 ps-6143 2d..1 259us+: trace_hardirqs_on <-_raw_spin_unlock_irqrestore
999 ps-6143 2d..1 263us+: time_hardirqs_on <-_raw_spin_unlock_irqrestore
1000 ps-6143 2d..1 306us : <stack trace>
1001 => trace_hardirqs_on_caller
1002 => trace_hardirqs_on
1003 => _raw_spin_unlock_irqrestore
1004 => do_task_stat
1005 => proc_tgid_stat
1006 => proc_single_show
1007 => seq_read
1008 => vfs_read
1009 => sys_read
1010 => system_call_fastpath
1013 This shows that the current tracer is "irqsoff" tracing the time
1014 for which interrupts were disabled. It gives the trace version (which
1015 never changes) and the version of the kernel upon which this was executed on
1016 (3.8). Then it displays the max latency in microseconds (259 us). The number
1017 of trace entries displayed and the total number (both are four: #4/4).
1018 VP, KP, SP, and HP are always zero and are reserved for later use.
1019 #P is the number of online CPUs (#P:4).
1021 The task is the process that was running when the latency
1022 occurred. (ps pid: 6143).
1024 The start and stop (the functions in which the interrupts were
1025 disabled and enabled respectively) that caused the latencies:
1027 - __lock_task_sighand is where the interrupts were disabled.
1028 - _raw_spin_unlock_irqrestore is where they were enabled again.
1030 The next lines after the header are the trace itself. The header
1031 explains which is which.
1033 cmd: The name of the process in the trace.
1035 pid: The PID of that process.
1037 CPU#: The CPU which the process was running on.
1039 irqs-off: 'd' interrupts are disabled. '.' otherwise.
1041 need-resched:
1042 - 'B' all, TIF_NEED_RESCHED, PREEMPT_NEED_RESCHED and TIF_RESCHED_LAZY is set,
1043 - 'N' both TIF_NEED_RESCHED and PREEMPT_NEED_RESCHED is set,
1044 - 'n' only TIF_NEED_RESCHED is set,
1045 - 'p' only PREEMPT_NEED_RESCHED is set,
1046 - 'L' both PREEMPT_NEED_RESCHED and TIF_RESCHED_LAZY is set,
1047 - 'b' both TIF_NEED_RESCHED and TIF_RESCHED_LAZY is set,
1048 - 'l' only TIF_RESCHED_LAZY is set
1049 - '.' otherwise.
1051 hardirq/softirq:
1052 - 'Z' - NMI occurred inside a hardirq
1053 - 'z' - NMI is running
1054 - 'H' - hard irq occurred inside a softirq.
1055 - 'h' - hard irq is running
1056 - 's' - soft irq is running
1057 - '.' - normal context.
1059 preempt-depth: The level of preempt_disabled
1061 The above is mostly meaningful for kernel developers.
1063 time:
1064 When the latency-format option is enabled, the trace file
1065 output includes a timestamp relative to the start of the
1066 trace. This differs from the output when latency-format
1067 is disabled, which includes an absolute timestamp.
1069 delay:
1070 This is just to help catch your eye a bit better. And
1071 needs to be fixed to be only relative to the same CPU.
1072 The marks are determined by the difference between this
1073 current trace and the next trace.
1075 - '$' - greater than 1 second
1076 - '@' - greater than 100 millisecond
1077 - '*' - greater than 10 millisecond
1078 - '#' - greater than 1000 microsecond
1079 - '!' - greater than 100 microsecond
1080 - '+' - greater than 10 microsecond
1081 - ' ' - less than or equal to 10 microsecond.
1083 The rest is the same as the 'trace' file.
1085 Note, the latency tracers will usually end with a back trace
1086 to easily find where the latency occurred.
1088 trace_options
1089 -------------
1091 The trace_options file (or the options directory) is used to control
1092 what gets printed in the trace output, or manipulate the tracers.
1093 To see what is available, simply cat the file::
1095 cat trace_options
1096 print-parent
1097 nosym-offset
1098 nosym-addr
1099 noverbose
1100 noraw
1101 nohex
1102 nobin
1103 noblock
1104 nofields
1105 trace_printk
1106 annotate
1107 nouserstacktrace
1108 nosym-userobj
1109 noprintk-msg-only
1110 context-info
1111 nolatency-format
1112 record-cmd
1113 norecord-tgid
1114 overwrite
1115 nodisable_on_free
1116 irq-info
1117 markers
1118 noevent-fork
1119 function-trace
1120 nofunction-fork
1121 nodisplay-graph
1122 nostacktrace
1123 nobranch
1125 To disable one of the options, echo in the option prepended with
1126 "no"::
1128 echo noprint-parent > trace_options
1130 To enable an option, leave off the "no"::
1132 echo sym-offset > trace_options
1134 Here are the available options:
1136 print-parent
1137 On function traces, display the calling (parent)
1138 function as well as the function being traced.
1139 ::
1141 print-parent:
1142 bash-4000 [01] 1477.606694: simple_strtoul <-kstrtoul
1144 noprint-parent:
1145 bash-4000 [01] 1477.606694: simple_strtoul
1148 sym-offset
1149 Display not only the function name, but also the
1150 offset in the function. For example, instead of
1151 seeing just "ktime_get", you will see
1152 "ktime_get+0xb/0x20".
1153 ::
1155 sym-offset:
1156 bash-4000 [01] 1477.606694: simple_strtoul+0x6/0xa0
1158 sym-addr
1159 This will also display the function address as well
1160 as the function name.
1161 ::
1163 sym-addr:
1164 bash-4000 [01] 1477.606694: simple_strtoul <c0339346>
1166 verbose
1167 This deals with the trace file when the
1168 latency-format option is enabled.
1169 ::
1171 bash 4000 1 0 00000000 00010a95 [58127d26] 1720.415ms \
1172 (+0.000ms): simple_strtoul (kstrtoul)
1174 raw
1175 This will display raw numbers. This option is best for
1176 use with user applications that can translate the raw
1177 numbers better than having it done in the kernel.
1179 hex
1180 Similar to raw, but the numbers will be in a hexadecimal format.
1182 bin
1183 This will print out the formats in raw binary.
1185 block
1186 When set, reading trace_pipe will not block when polled.
1188 fields
1189 Print the fields as described by their types. This is a better
1190 option than using hex, bin or raw, as it gives a better parsing
1191 of the content of the event.
1193 trace_printk
1194 Can disable trace_printk() from writing into the buffer.
1196 trace_printk_dest
1197 Set to have trace_printk() and similar internal tracing functions
1198 write into this instance. Note, only one trace instance can have
1199 this set. By setting this flag, it clears the trace_printk_dest flag
1200 of the instance that had it set previously. By default, the top
1201 level trace has this set, and will get it set again if another
1202 instance has it set then clears it.
1204 This flag cannot be cleared by the top level instance, as it is the
1205 default instance. The only way the top level instance has this flag
1206 cleared, is by it being set in another instance.
1208 annotate
1209 It is sometimes confusing when the CPU buffers are full
1210 and one CPU buffer had a lot of events recently, thus
1211 a shorter time frame, were another CPU may have only had
1212 a few events, which lets it have older events. When
1213 the trace is reported, it shows the oldest events first,
1214 and it may look like only one CPU ran (the one with the
1215 oldest events). When the annotate option is set, it will
1216 display when a new CPU buffer started::
1218 <idle>-0 [001] dNs4 21169.031481: wake_up_idle_cpu <-add_timer_on
1219 <idle>-0 [001] dNs4 21169.031482: _raw_spin_unlock_irqrestore <-add_timer_on
1220 <idle>-0 [001] .Ns4 21169.031484: sub_preempt_count <-_raw_spin_unlock_irqrestore
1221 ##### CPU 2 buffer started ####
1222 <idle>-0 [002] .N.1 21169.031484: rcu_idle_exit <-cpu_idle
1223 <idle>-0 [001] .Ns3 21169.031484: _raw_spin_unlock <-clocksource_watchdog
1224 <idle>-0 [001] .Ns3 21169.031485: sub_preempt_count <-_raw_spin_unlock
1226 userstacktrace
1227 This option changes the trace. It records a
1228 stacktrace of the current user space thread after
1229 each trace event.
1231 sym-userobj
1232 when user stacktrace are enabled, look up which
1233 object the address belongs to, and print a
1234 relative address. This is especially useful when
1235 ASLR is on, otherwise you don't get a chance to
1236 resolve the address to object/file/line after
1237 the app is no longer running
1239 The lookup is performed when you read
1240 trace,trace_pipe. Example::
1242 a.out-1623 [000] 40874.465068: /root/a.out[+0x480] <-/root/a.out[+0
1243 x494] <- /root/a.out[+0x4a8] <- /lib/libc-2.7.so[+0x1e1a6]
1246 printk-msg-only
1247 When set, trace_printk()s will only show the format
1248 and not their parameters (if trace_bprintk() or
1249 trace_bputs() was used to save the trace_printk()).
1251 context-info
1252 Show only the event data. Hides the comm, PID,
1253 timestamp, CPU, and other useful data.
1255 latency-format
1256 This option changes the trace output. When it is enabled,
1257 the trace displays additional information about the
1258 latency, as described in "Latency trace format".
1260 pause-on-trace
1261 When set, opening the trace file for read, will pause
1262 writing to the ring buffer (as if tracing_on was set to zero).
1263 This simulates the original behavior of the trace file.
1264 When the file is closed, tracing will be enabled again.
1266 hash-ptr
1267 When set, "%p" in the event printk format displays the
1268 hashed pointer value instead of real address.
1269 This will be useful if you want to find out which hashed
1270 value is corresponding to the real value in trace log.
1272 record-cmd
1273 When any event or tracer is enabled, a hook is enabled
1274 in the sched_switch trace point to fill comm cache
1275 with mapped pids and comms. But this may cause some
1276 overhead, and if you only care about pids, and not the
1277 name of the task, disabling this option can lower the
1278 impact of tracing. See "saved_cmdlines".
1280 record-tgid
1281 When any event or tracer is enabled, a hook is enabled
1282 in the sched_switch trace point to fill the cache of
1283 mapped Thread Group IDs (TGID) mapping to pids. See
1284 "saved_tgids".
1286 overwrite
1287 This controls what happens when the trace buffer is
1288 full. If "1" (default), the oldest events are
1289 discarded and overwritten. If "0", then the newest
1290 events are discarded.
1291 (see per_cpu/cpu0/stats for overrun and dropped)
1293 disable_on_free
1294 When the free_buffer is closed, tracing will
1295 stop (tracing_on set to 0).
1297 irq-info
1298 Shows the interrupt, preempt count, need resched data.
1299 When disabled, the trace looks like::
1301 # tracer: function
1302 #
1303 # entries-in-buffer/entries-written: 144405/9452052 #P:4
1304 #
1305 # TASK-PID CPU# TIMESTAMP FUNCTION
1306 # | | | | |
1307 <idle>-0 [002] 23636.756054: ttwu_do_activate.constprop.89 <-try_to_wake_up
1308 <idle>-0 [002] 23636.756054: activate_task <-ttwu_do_activate.constprop.89
1309 <idle>-0 [002] 23636.756055: enqueue_task <-activate_task
1312 markers
1313 When set, the trace_marker is writable (only by root).
1314 When disabled, the trace_marker will error with EINVAL
1315 on write.
1317 event-fork
1318 When set, tasks with PIDs listed in set_event_pid will have
1319 the PIDs of their children added to set_event_pid when those
1320 tasks fork. Also, when tasks with PIDs in set_event_pid exit,
1321 their PIDs will be removed from the file.
1323 This affects PIDs listed in set_event_notrace_pid as well.
1325 function-trace
1326 The latency tracers will enable function tracing
1327 if this option is enabled (default it is). When
1328 it is disabled, the latency tracers do not trace
1329 functions. This keeps the overhead of the tracer down
1330 when performing latency tests.
1332 function-fork
1333 When set, tasks with PIDs listed in set_ftrace_pid will
1334 have the PIDs of their children added to set_ftrace_pid
1335 when those tasks fork. Also, when tasks with PIDs in
1336 set_ftrace_pid exit, their PIDs will be removed from the
1337 file.
1339 This affects PIDs in set_ftrace_notrace_pid as well.
1341 display-graph
1342 When set, the latency tracers (irqsoff, wakeup, etc) will
1343 use function graph tracing instead of function tracing.
1345 stacktrace
1346 When set, a stack trace is recorded after any trace event
1347 is recorded.
1349 branch
1350 Enable branch tracing with the tracer. This enables branch
1351 tracer along with the currently set tracer. Enabling this
1352 with the "nop" tracer is the same as just enabling the
1353 "branch" tracer.
1355 .. tip:: Some tracers have their own options. They only appear in this
1356 file when the tracer is active. They always appear in the
1357 options directory.
1360 Here are the per tracer options:
1362 Options for function tracer:
1364 func_stack_trace
1365 When set, a stack trace is recorded after every
1366 function that is recorded. NOTE! Limit the functions
1367 that are recorded before enabling this, with
1368 "set_ftrace_filter" otherwise the system performance
1369 will be critically degraded. Remember to disable
1370 this option before clearing the function filter.
1372 Options for function_graph tracer:
1374 Since the function_graph tracer has a slightly different output
1375 it has its own options to control what is displayed.
1377 funcgraph-overrun
1378 When set, the "overrun" of the graph stack is
1379 displayed after each function traced. The
1380 overrun, is when the stack depth of the calls
1381 is greater than what is reserved for each task.
1382 Each task has a fixed array of functions to
1383 trace in the call graph. If the depth of the
1384 calls exceeds that, the function is not traced.
1385 The overrun is the number of functions missed
1386 due to exceeding this array.
1388 funcgraph-cpu
1389 When set, the CPU number of the CPU where the trace
1390 occurred is displayed.
1392 funcgraph-overhead
1393 When set, if the function takes longer than
1394 A certain amount, then a delay marker is
1395 displayed. See "delay" above, under the
1396 header description.
1398 funcgraph-proc
1399 Unlike other tracers, the process' command line
1400 is not displayed by default, but instead only
1401 when a task is traced in and out during a context
1402 switch. Enabling this options has the command
1403 of each process displayed at every line.
1405 funcgraph-duration
1406 At the end of each function (the return)
1407 the duration of the amount of time in the
1408 function is displayed in microseconds.
1410 funcgraph-abstime
1411 When set, the timestamp is displayed at each line.
1413 funcgraph-irqs
1414 When disabled, functions that happen inside an
1415 interrupt will not be traced.
1417 funcgraph-tail
1418 When set, the return event will include the function
1419 that it represents. By default this is off, and
1420 only a closing curly bracket "}" is displayed for
1421 the return of a function.
1423 funcgraph-retval
1424 When set, the return value of each traced function
1425 will be printed after an equal sign "=". By default
1426 this is off.
1428 funcgraph-retval-hex
1429 When set, the return value will always be printed
1430 in hexadecimal format. If the option is not set and
1431 the return value is an error code, it will be printed
1432 in signed decimal format; otherwise it will also be
1433 printed in hexadecimal format. By default, this option
1434 is off.
1436 sleep-time
1437 When running function graph tracer, to include
1438 the time a task schedules out in its function.
1439 When enabled, it will account time the task has been
1440 scheduled out as part of the function call.
1442 graph-time
1443 When running function profiler with function graph tracer,
1444 to include the time to call nested functions. When this is
1445 not set, the time reported for the function will only
1446 include the time the function itself executed for, not the
1447 time for functions that it called.
1449 Options for blk tracer:
1451 blk_classic
1452 Shows a more minimalistic output.
1455 irqsoff
1456 -------
1458 When interrupts are disabled, the CPU can not react to any other
1459 external event (besides NMIs and SMIs). This prevents the timer
1460 interrupt from triggering or the mouse interrupt from letting
1461 the kernel know of a new mouse event. The result is a latency
1462 with the reaction time.
1464 The irqsoff tracer tracks the time for which interrupts are
1465 disabled. When a new maximum latency is hit, the tracer saves
1466 the trace leading up to that latency point so that every time a
1467 new maximum is reached, the old saved trace is discarded and the
1468 new trace is saved.
1470 To reset the maximum, echo 0 into tracing_max_latency. Here is
1471 an example::
1473 # echo 0 > options/function-trace
1474 # echo irqsoff > current_tracer
1475 # echo 1 > tracing_on
1476 # echo 0 > tracing_max_latency
1477 # ls -ltr
1478 [...]
1479 # echo 0 > tracing_on
1480 # cat trace
1481 # tracer: irqsoff
1482 #
1483 # irqsoff latency trace v1.1.5 on 3.8.0-test+
1484 # --------------------------------------------------------------------
1485 # latency: 16 us, #4/4, CPU#0 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1486 # -----------------
1487 # | task: swapper/0-0 (uid:0 nice:0 policy:0 rt_prio:0)
1488 # -----------------
1489 # => started at: run_timer_softirq
1490 # => ended at: run_timer_softirq
1491 #
1492 #
1493 # _------=> CPU#
1494 # / _-----=> irqs-off
1495 # | / _----=> need-resched
1496 # || / _---=> hardirq/softirq
1497 # ||| / _--=> preempt-depth
1498 # |||| / delay
1499 # cmd pid ||||| time | caller
1500 # \ / ||||| \ | /
1501 <idle>-0 0d.s2 0us+: _raw_spin_lock_irq <-run_timer_softirq
1502 <idle>-0 0dNs3 17us : _raw_spin_unlock_irq <-run_timer_softirq
1503 <idle>-0 0dNs3 17us+: trace_hardirqs_on <-run_timer_softirq
1504 <idle>-0 0dNs3 25us : <stack trace>
1505 => _raw_spin_unlock_irq
1506 => run_timer_softirq
1507 => __do_softirq
1508 => call_softirq
1509 => do_softirq
1510 => irq_exit
1511 => smp_apic_timer_interrupt
1512 => apic_timer_interrupt
1513 => rcu_idle_exit
1514 => cpu_idle
1515 => rest_init
1516 => start_kernel
1517 => x86_64_start_reservations
1518 => x86_64_start_kernel
1520 Here we see that we had a latency of 16 microseconds (which is
1521 very good). The _raw_spin_lock_irq in run_timer_softirq disabled
1522 interrupts. The difference between the 16 and the displayed
1523 timestamp 25us occurred because the clock was incremented
1524 between the time of recording the max latency and the time of
1525 recording the function that had that latency.
1527 Note the above example had function-trace not set. If we set
1528 function-trace, we get a much larger output::
1530 with echo 1 > options/function-trace
1532 # tracer: irqsoff
1533 #
1534 # irqsoff latency trace v1.1.5 on 3.8.0-test+
1535 # --------------------------------------------------------------------
1536 # latency: 71 us, #168/168, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1537 # -----------------
1538 # | task: bash-2042 (uid:0 nice:0 policy:0 rt_prio:0)
1539 # -----------------
1540 # => started at: ata_scsi_queuecmd
1541 # => ended at: ata_scsi_queuecmd
1542 #
1543 #
1544 # _------=> CPU#
1545 # / _-----=> irqs-off
1546 # | / _----=> need-resched
1547 # || / _---=> hardirq/softirq
1548 # ||| / _--=> preempt-depth
1549 # |||| / delay
1550 # cmd pid ||||| time | caller
1551 # \ / ||||| \ | /
1552 bash-2042 3d... 0us : _raw_spin_lock_irqsave <-ata_scsi_queuecmd
1553 bash-2042 3d... 0us : add_preempt_count <-_raw_spin_lock_irqsave
1554 bash-2042 3d..1 1us : ata_scsi_find_dev <-ata_scsi_queuecmd
1555 bash-2042 3d..1 1us : __ata_scsi_find_dev <-ata_scsi_find_dev
1556 bash-2042 3d..1 2us : ata_find_dev.part.14 <-__ata_scsi_find_dev
1557 bash-2042 3d..1 2us : ata_qc_new_init <-__ata_scsi_queuecmd
1558 bash-2042 3d..1 3us : ata_sg_init <-__ata_scsi_queuecmd
1559 bash-2042 3d..1 4us : ata_scsi_rw_xlat <-__ata_scsi_queuecmd
1560 bash-2042 3d..1 4us : ata_build_rw_tf <-ata_scsi_rw_xlat
1561 [...]
1562 bash-2042 3d..1 67us : delay_tsc <-__delay
1563 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc
1564 bash-2042 3d..2 67us : sub_preempt_count <-delay_tsc
1565 bash-2042 3d..1 67us : add_preempt_count <-delay_tsc
1566 bash-2042 3d..2 68us : sub_preempt_count <-delay_tsc
1567 bash-2042 3d..1 68us+: ata_bmdma_start <-ata_bmdma_qc_issue
1568 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd
1569 bash-2042 3d..1 71us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd
1570 bash-2042 3d..1 72us+: trace_hardirqs_on <-ata_scsi_queuecmd
1571 bash-2042 3d..1 120us : <stack trace>
1572 => _raw_spin_unlock_irqrestore
1573 => ata_scsi_queuecmd
1574 => scsi_dispatch_cmd
1575 => scsi_request_fn
1576 => __blk_run_queue_uncond
1577 => __blk_run_queue
1578 => blk_queue_bio
1579 => submit_bio_noacct
1580 => submit_bio
1581 => submit_bh
1582 => __ext3_get_inode_loc
1583 => ext3_iget
1584 => ext3_lookup
1585 => lookup_real
1586 => __lookup_hash
1587 => walk_component
1588 => lookup_last
1589 => path_lookupat
1590 => filename_lookup
1591 => user_path_at_empty
1592 => user_path_at
1593 => vfs_fstatat
1594 => vfs_stat
1595 => sys_newstat
1596 => system_call_fastpath
1599 Here we traced a 71 microsecond latency. But we also see all the
1600 functions that were called during that time. Note that by
1601 enabling function tracing, we incur an added overhead. This
1602 overhead may extend the latency times. But nevertheless, this
1603 trace has provided some very helpful debugging information.
1605 If we prefer function graph output instead of function, we can set
1606 display-graph option::
1608 with echo 1 > options/display-graph
1610 # tracer: irqsoff
1611 #
1612 # irqsoff latency trace v1.1.5 on 4.20.0-rc6+
1613 # --------------------------------------------------------------------
1614 # latency: 3751 us, #274/274, CPU#0 | (M:desktop VP:0, KP:0, SP:0 HP:0 #P:4)
1615 # -----------------
1616 # | task: bash-1507 (uid:0 nice:0 policy:0 rt_prio:0)
1617 # -----------------
1618 # => started at: free_debug_processing
1619 # => ended at: return_to_handler
1620 #
1621 #
1622 # _-----=> irqs-off
1623 # / _----=> need-resched
1624 # | / _---=> hardirq/softirq
1625 # || / _--=> preempt-depth
1626 # ||| /
1627 # REL TIME CPU TASK/PID |||| DURATION FUNCTION CALLS
1628 # | | | | |||| | | | | | |
1629 0 us | 0) bash-1507 | d... | 0.000 us | _raw_spin_lock_irqsave();
1630 0 us | 0) bash-1507 | d..1 | 0.378 us | do_raw_spin_trylock();
1631 1 us | 0) bash-1507 | d..2 | | set_track() {
1632 2 us | 0) bash-1507 | d..2 | | save_stack_trace() {
1633 2 us | 0) bash-1507 | d..2 | | __save_stack_trace() {
1634 3 us | 0) bash-1507 | d..2 | | __unwind_start() {
1635 3 us | 0) bash-1507 | d..2 | | get_stack_info() {
1636 3 us | 0) bash-1507 | d..2 | 0.351 us | in_task_stack();
1637 4 us | 0) bash-1507 | d..2 | 1.107 us | }
1638 [...]
1639 3750 us | 0) bash-1507 | d..1 | 0.516 us | do_raw_spin_unlock();
1640 3750 us | 0) bash-1507 | d..1 | 0.000 us | _raw_spin_unlock_irqrestore();
1641 3764 us | 0) bash-1507 | d..1 | 0.000 us | tracer_hardirqs_on();
1642 bash-1507 0d..1 3792us : <stack trace>
1643 => free_debug_processing
1644 => __slab_free
1645 => kmem_cache_free
1646 => vm_area_free
1647 => remove_vma
1648 => exit_mmap
1649 => mmput
1650 => begin_new_exec
1651 => load_elf_binary
1652 => search_binary_handler
1653 => __do_execve_file.isra.32
1654 => __x64_sys_execve
1655 => do_syscall_64
1656 => entry_SYSCALL_64_after_hwframe
1658 preemptoff
1659 ----------
1661 When preemption is disabled, we may be able to receive
1662 interrupts but the task cannot be preempted and a higher
1663 priority task must wait for preemption to be enabled again
1664 before it can preempt a lower priority task.
1666 The preemptoff tracer traces the places that disable preemption.
1667 Like the irqsoff tracer, it records the maximum latency for
1668 which preemption was disabled. The control of preemptoff tracer
1669 is much like the irqsoff tracer.
1670 ::
1672 # echo 0 > options/function-trace
1673 # echo preemptoff > current_tracer
1674 # echo 1 > tracing_on
1675 # echo 0 > tracing_max_latency
1676 # ls -ltr
1677 [...]
1678 # echo 0 > tracing_on
1679 # cat trace
1680 # tracer: preemptoff
1681 #
1682 # preemptoff latency trace v1.1.5 on 3.8.0-test+
1683 # --------------------------------------------------------------------
1684 # latency: 46 us, #4/4, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1685 # -----------------
1686 # | task: sshd-1991 (uid:0 nice:0 policy:0 rt_prio:0)
1687 # -----------------
1688 # => started at: do_IRQ
1689 # => ended at: do_IRQ
1690 #
1691 #
1692 # _------=> CPU#
1693 # / _-----=> irqs-off
1694 # | / _----=> need-resched
1695 # || / _---=> hardirq/softirq
1696 # ||| / _--=> preempt-depth
1697 # |||| / delay
1698 # cmd pid ||||| time | caller
1699 # \ / ||||| \ | /
1700 sshd-1991 1d.h. 0us+: irq_enter <-do_IRQ
1701 sshd-1991 1d..1 46us : irq_exit <-do_IRQ
1702 sshd-1991 1d..1 47us+: trace_preempt_on <-do_IRQ
1703 sshd-1991 1d..1 52us : <stack trace>
1704 => sub_preempt_count
1705 => irq_exit
1706 => do_IRQ
1707 => ret_from_intr
1710 This has some more changes. Preemption was disabled when an
1711 interrupt came in (notice the 'h'), and was enabled on exit.
1712 But we also see that interrupts have been disabled when entering
1713 the preempt off section and leaving it (the 'd'). We do not know if
1714 interrupts were enabled in the mean time or shortly after this
1715 was over.
1716 ::
1718 # tracer: preemptoff
1719 #
1720 # preemptoff latency trace v1.1.5 on 3.8.0-test+
1721 # --------------------------------------------------------------------
1722 # latency: 83 us, #241/241, CPU#1 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1723 # -----------------
1724 # | task: bash-1994 (uid:0 nice:0 policy:0 rt_prio:0)
1725 # -----------------
1726 # => started at: wake_up_new_task
1727 # => ended at: task_rq_unlock
1728 #
1729 #
1730 # _------=> CPU#
1731 # / _-----=> irqs-off
1732 # | / _----=> need-resched
1733 # || / _---=> hardirq/softirq
1734 # ||| / _--=> preempt-depth
1735 # |||| / delay
1736 # cmd pid ||||| time | caller
1737 # \ / ||||| \ | /
1738 bash-1994 1d..1 0us : _raw_spin_lock_irqsave <-wake_up_new_task
1739 bash-1994 1d..1 0us : select_task_rq_fair <-select_task_rq
1740 bash-1994 1d..1 1us : __rcu_read_lock <-select_task_rq_fair
1741 bash-1994 1d..1 1us : source_load <-select_task_rq_fair
1742 bash-1994 1d..1 1us : source_load <-select_task_rq_fair
1743 [...]
1744 bash-1994 1d..1 12us : irq_enter <-smp_apic_timer_interrupt
1745 bash-1994 1d..1 12us : rcu_irq_enter <-irq_enter
1746 bash-1994 1d..1 13us : add_preempt_count <-irq_enter
1747 bash-1994 1d.h1 13us : exit_idle <-smp_apic_timer_interrupt
1748 bash-1994 1d.h1 13us : hrtimer_interrupt <-smp_apic_timer_interrupt
1749 bash-1994 1d.h1 13us : _raw_spin_lock <-hrtimer_interrupt
1750 bash-1994 1d.h1 14us : add_preempt_count <-_raw_spin_lock
1751 bash-1994 1d.h2 14us : ktime_get_update_offsets <-hrtimer_interrupt
1752 [...]
1753 bash-1994 1d.h1 35us : lapic_next_event <-clockevents_program_event
1754 bash-1994 1d.h1 35us : irq_exit <-smp_apic_timer_interrupt
1755 bash-1994 1d.h1 36us : sub_preempt_count <-irq_exit
1756 bash-1994 1d..2 36us : do_softirq <-irq_exit
1757 bash-1994 1d..2 36us : __do_softirq <-call_softirq
1758 bash-1994 1d..2 36us : __local_bh_disable <-__do_softirq
1759 bash-1994 1d.s2 37us : add_preempt_count <-_raw_spin_lock_irq
1760 bash-1994 1d.s3 38us : _raw_spin_unlock <-run_timer_softirq
1761 bash-1994 1d.s3 39us : sub_preempt_count <-_raw_spin_unlock
1762 bash-1994 1d.s2 39us : call_timer_fn <-run_timer_softirq
1763 [...]
1764 bash-1994 1dNs2 81us : cpu_needs_another_gp <-rcu_process_callbacks
1765 bash-1994 1dNs2 82us : __local_bh_enable <-__do_softirq
1766 bash-1994 1dNs2 82us : sub_preempt_count <-__local_bh_enable
1767 bash-1994 1dN.2 82us : idle_cpu <-irq_exit
1768 bash-1994 1dN.2 83us : rcu_irq_exit <-irq_exit
1769 bash-1994 1dN.2 83us : sub_preempt_count <-irq_exit
1770 bash-1994 1.N.1 84us : _raw_spin_unlock_irqrestore <-task_rq_unlock
1771 bash-1994 1.N.1 84us+: trace_preempt_on <-task_rq_unlock
1772 bash-1994 1.N.1 104us : <stack trace>
1773 => sub_preempt_count
1774 => _raw_spin_unlock_irqrestore
1775 => task_rq_unlock
1776 => wake_up_new_task
1777 => do_fork
1778 => sys_clone
1779 => stub_clone
1782 The above is an example of the preemptoff trace with
1783 function-trace set. Here we see that interrupts were not disabled
1784 the entire time. The irq_enter code lets us know that we entered
1785 an interrupt 'h'. Before that, the functions being traced still
1786 show that it is not in an interrupt, but we can see from the
1787 functions themselves that this is not the case.
1789 preemptirqsoff
1790 --------------
1792 Knowing the locations that have interrupts disabled or
1793 preemption disabled for the longest times is helpful. But
1794 sometimes we would like to know when either preemption and/or
1795 interrupts are disabled.
1797 Consider the following code::
1799 local_irq_disable();
1800 call_function_with_irqs_off();
1801 preempt_disable();
1802 call_function_with_irqs_and_preemption_off();
1803 local_irq_enable();
1804 call_function_with_preemption_off();
1805 preempt_enable();
1807 The irqsoff tracer will record the total length of
1808 call_function_with_irqs_off() and
1809 call_function_with_irqs_and_preemption_off().
1811 The preemptoff tracer will record the total length of
1812 call_function_with_irqs_and_preemption_off() and
1813 call_function_with_preemption_off().
1815 But neither will trace the time that interrupts and/or
1816 preemption is disabled. This total time is the time that we can
1817 not schedule. To record this time, use the preemptirqsoff
1818 tracer.
1820 Again, using this trace is much like the irqsoff and preemptoff
1821 tracers.
1822 ::
1824 # echo 0 > options/function-trace
1825 # echo preemptirqsoff > current_tracer
1826 # echo 1 > tracing_on
1827 # echo 0 > tracing_max_latency
1828 # ls -ltr
1829 [...]
1830 # echo 0 > tracing_on
1831 # cat trace
1832 # tracer: preemptirqsoff
1833 #
1834 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+
1835 # --------------------------------------------------------------------
1836 # latency: 100 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1837 # -----------------
1838 # | task: ls-2230 (uid:0 nice:0 policy:0 rt_prio:0)
1839 # -----------------
1840 # => started at: ata_scsi_queuecmd
1841 # => ended at: ata_scsi_queuecmd
1842 #
1843 #
1844 # _------=> CPU#
1845 # / _-----=> irqs-off
1846 # | / _----=> need-resched
1847 # || / _---=> hardirq/softirq
1848 # ||| / _--=> preempt-depth
1849 # |||| / delay
1850 # cmd pid ||||| time | caller
1851 # \ / ||||| \ | /
1852 ls-2230 3d... 0us+: _raw_spin_lock_irqsave <-ata_scsi_queuecmd
1853 ls-2230 3...1 100us : _raw_spin_unlock_irqrestore <-ata_scsi_queuecmd
1854 ls-2230 3...1 101us+: trace_preempt_on <-ata_scsi_queuecmd
1855 ls-2230 3...1 111us : <stack trace>
1856 => sub_preempt_count
1857 => _raw_spin_unlock_irqrestore
1858 => ata_scsi_queuecmd
1859 => scsi_dispatch_cmd
1860 => scsi_request_fn
1861 => __blk_run_queue_uncond
1862 => __blk_run_queue
1863 => blk_queue_bio
1864 => submit_bio_noacct
1865 => submit_bio
1866 => submit_bh
1867 => ext3_bread
1868 => ext3_dir_bread
1869 => htree_dirblock_to_tree
1870 => ext3_htree_fill_tree
1871 => ext3_readdir
1872 => vfs_readdir
1873 => sys_getdents
1874 => system_call_fastpath
1877 The trace_hardirqs_off_thunk is called from assembly on x86 when
1878 interrupts are disabled in the assembly code. Without the
1879 function tracing, we do not know if interrupts were enabled
1880 within the preemption points. We do see that it started with
1881 preemption enabled.
1883 Here is a trace with function-trace set::
1885 # tracer: preemptirqsoff
1886 #
1887 # preemptirqsoff latency trace v1.1.5 on 3.8.0-test+
1888 # --------------------------------------------------------------------
1889 # latency: 161 us, #339/339, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
1890 # -----------------
1891 # | task: ls-2269 (uid:0 nice:0 policy:0 rt_prio:0)
1892 # -----------------
1893 # => started at: schedule
1894 # => ended at: mutex_unlock
1895 #
1896 #
1897 # _------=> CPU#
1898 # / _-----=> irqs-off
1899 # | / _----=> need-resched
1900 # || / _---=> hardirq/softirq
1901 # ||| / _--=> preempt-depth
1902 # |||| / delay
1903 # cmd pid ||||| time | caller
1904 # \ / ||||| \ | /
1905 kworker/-59 3...1 0us : __schedule <-schedule
1906 kworker/-59 3d..1 0us : rcu_preempt_qs <-rcu_note_context_switch
1907 kworker/-59 3d..1 1us : add_preempt_count <-_raw_spin_lock_irq
1908 kworker/-59 3d..2 1us : deactivate_task <-__schedule
1909 kworker/-59 3d..2 1us : dequeue_task <-deactivate_task
1910 kworker/-59 3d..2 2us : update_rq_clock <-dequeue_task
1911 kworker/-59 3d..2 2us : dequeue_task_fair <-dequeue_task
1912 kworker/-59 3d..2 2us : update_curr <-dequeue_task_fair
1913 kworker/-59 3d..2 2us : update_min_vruntime <-update_curr
1914 kworker/-59 3d..2 3us : cpuacct_charge <-update_curr
1915 kworker/-59 3d..2 3us : __rcu_read_lock <-cpuacct_charge
1916 kworker/-59 3d..2 3us : __rcu_read_unlock <-cpuacct_charge
1917 kworker/-59 3d..2 3us : update_cfs_rq_blocked_load <-dequeue_task_fair
1918 kworker/-59 3d..2 4us : clear_buddies <-dequeue_task_fair
1919 kworker/-59 3d..2 4us : account_entity_dequeue <-dequeue_task_fair
1920 kworker/-59 3d..2 4us : update_min_vruntime <-dequeue_task_fair
1921 kworker/-59 3d..2 4us : update_cfs_shares <-dequeue_task_fair
1922 kworker/-59 3d..2 5us : hrtick_update <-dequeue_task_fair
1923 kworker/-59 3d..2 5us : wq_worker_sleeping <-__schedule
1924 kworker/-59 3d..2 5us : kthread_data <-wq_worker_sleeping
1925 kworker/-59 3d..2 5us : put_prev_task_fair <-__schedule
1926 kworker/-59 3d..2 6us : pick_next_task_fair <-pick_next_task
1927 kworker/-59 3d..2 6us : clear_buddies <-pick_next_task_fair
1928 kworker/-59 3d..2 6us : set_next_entity <-pick_next_task_fair
1929 kworker/-59 3d..2 6us : update_stats_wait_end <-set_next_entity
1930 ls-2269 3d..2 7us : finish_task_switch <-__schedule
1931 ls-2269 3d..2 7us : _raw_spin_unlock_irq <-finish_task_switch
1932 ls-2269 3d..2 8us : do_IRQ <-ret_from_intr
1933 ls-2269 3d..2 8us : irq_enter <-do_IRQ
1934 ls-2269 3d..2 8us : rcu_irq_enter <-irq_enter
1935 ls-2269 3d..2 9us : add_preempt_count <-irq_enter
1936 ls-2269 3d.h2 9us : exit_idle <-do_IRQ
1937 [...]
1938 ls-2269 3d.h3 20us : sub_preempt_count <-_raw_spin_unlock
1939 ls-2269 3d.h2 20us : irq_exit <-do_IRQ
1940 ls-2269 3d.h2 21us : sub_preempt_count <-irq_exit
1941 ls-2269 3d..3 21us : do_softirq <-irq_exit
1942 ls-2269 3d..3 21us : __do_softirq <-call_softirq
1943 ls-2269 3d..3 21us+: __local_bh_disable <-__do_softirq
1944 ls-2269 3d.s4 29us : sub_preempt_count <-_local_bh_enable_ip
1945 ls-2269 3d.s5 29us : sub_preempt_count <-_local_bh_enable_ip
1946 ls-2269 3d.s5 31us : do_IRQ <-ret_from_intr
1947 ls-2269 3d.s5 31us : irq_enter <-do_IRQ
1948 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter
1949 [...]
1950 ls-2269 3d.s5 31us : rcu_irq_enter <-irq_enter
1951 ls-2269 3d.s5 32us : add_preempt_count <-irq_enter
1952 ls-2269 3d.H5 32us : exit_idle <-do_IRQ
1953 ls-2269 3d.H5 32us : handle_irq <-do_IRQ
1954 ls-2269 3d.H5 32us : irq_to_desc <-handle_irq
1955 ls-2269 3d.H5 33us : handle_fasteoi_irq <-handle_irq
1956 [...]
1957 ls-2269 3d.s5 158us : _raw_spin_unlock_irqrestore <-rtl8139_poll
1958 ls-2269 3d.s3 158us : net_rps_action_and_irq_enable.isra.65 <-net_rx_action
1959 ls-2269 3d.s3 159us : __local_bh_enable <-__do_softirq
1960 ls-2269 3d.s3 159us : sub_preempt_count <-__local_bh_enable
1961 ls-2269 3d..3 159us : idle_cpu <-irq_exit
1962 ls-2269 3d..3 159us : rcu_irq_exit <-irq_exit
1963 ls-2269 3d..3 160us : sub_preempt_count <-irq_exit
1964 ls-2269 3d... 161us : __mutex_unlock_slowpath <-mutex_unlock
1965 ls-2269 3d... 162us+: trace_hardirqs_on <-mutex_unlock
1966 ls-2269 3d... 186us : <stack trace>
1967 => __mutex_unlock_slowpath
1968 => mutex_unlock
1969 => process_output
1970 => n_tty_write
1971 => tty_write
1972 => vfs_write
1973 => sys_write
1974 => system_call_fastpath
1976 This is an interesting trace. It started with kworker running and
1977 scheduling out and ls taking over. But as soon as ls released the
1978 rq lock and enabled interrupts (but not preemption) an interrupt
1979 triggered. When the interrupt finished, it started running softirqs.
1980 But while the softirq was running, another interrupt triggered.
1981 When an interrupt is running inside a softirq, the annotation is 'H'.
1984 wakeup
1985 ------
1987 One common case that people are interested in tracing is the
1988 time it takes for a task that is woken to actually wake up.
1989 Now for non Real-Time tasks, this can be arbitrary. But tracing
1990 it nonetheless can be interesting.
1992 Without function tracing::
1994 # echo 0 > options/function-trace
1995 # echo wakeup > current_tracer
1996 # echo 1 > tracing_on
1997 # echo 0 > tracing_max_latency
1998 # chrt -f 5 sleep 1
1999 # echo 0 > tracing_on
2000 # cat trace
2001 # tracer: wakeup
2002 #
2003 # wakeup latency trace v1.1.5 on 3.8.0-test+
2004 # --------------------------------------------------------------------
2005 # latency: 15 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
2006 # -----------------
2007 # | task: kworker/3:1H-312 (uid:0 nice:-20 policy:0 rt_prio:0)
2008 # -----------------
2009 #
2010 # _------=> CPU#
2011 # / _-----=> irqs-off
2012 # | / _----=> need-resched
2013 # || / _---=> hardirq/softirq
2014 # ||| / _--=> preempt-depth
2015 # |||| / delay
2016 # cmd pid ||||| time | caller
2017 # \ / ||||| \ | /
2018 <idle>-0 3dNs7 0us : 0:120:R + [003] 312:100:R kworker/3:1H
2019 <idle>-0 3dNs7 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up
2020 <idle>-0 3d..3 15us : __schedule <-schedule
2021 <idle>-0 3d..3 15us : 0:120:R ==> [003] 312:100:R kworker/3:1H
2023 The tracer only traces the highest priority task in the system
2024 to avoid tracing the normal circumstances. Here we see that
2025 the kworker with a nice priority of -20 (not very nice), took
2026 just 15 microseconds from the time it woke up, to the time it
2027 ran.
2029 Non Real-Time tasks are not that interesting. A more interesting
2030 trace is to concentrate only on Real-Time tasks.
2032 wakeup_rt
2033 ---------
2035 In a Real-Time environment it is very important to know the
2036 wakeup time it takes for the highest priority task that is woken
2037 up to the time that it executes. This is also known as "schedule
2038 latency". I stress the point that this is about RT tasks. It is
2039 also important to know the scheduling latency of non-RT tasks,
2040 but the average schedule latency is better for non-RT tasks.
2041 Tools like LatencyTop are more appropriate for such
2042 measurements.
2044 Real-Time environments are interested in the worst case latency.
2045 That is the longest latency it takes for something to happen,
2046 and not the average. We can have a very fast scheduler that may
2047 only have a large latency once in a while, but that would not
2048 work well with Real-Time tasks. The wakeup_rt tracer was designed
2049 to record the worst case wakeups of RT tasks. Non-RT tasks are
2050 not recorded because the tracer only records one worst case and
2051 tracing non-RT tasks that are unpredictable will overwrite the
2052 worst case latency of RT tasks (just run the normal wakeup
2053 tracer for a while to see that effect).
2055 Since this tracer only deals with RT tasks, we will run this
2056 slightly differently than we did with the previous tracers.
2057 Instead of performing an 'ls', we will run 'sleep 1' under
2058 'chrt' which changes the priority of the task.
2059 ::
2061 # echo 0 > options/function-trace
2062 # echo wakeup_rt > current_tracer
2063 # echo 1 > tracing_on
2064 # echo 0 > tracing_max_latency
2065 # chrt -f 5 sleep 1
2066 # echo 0 > tracing_on
2067 # cat trace
2068 # tracer: wakeup
2069 #
2070 # tracer: wakeup_rt
2071 #
2072 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+
2073 # --------------------------------------------------------------------
2074 # latency: 5 us, #4/4, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
2075 # -----------------
2076 # | task: sleep-2389 (uid:0 nice:0 policy:1 rt_prio:5)
2077 # -----------------
2078 #
2079 # _------=> CPU#
2080 # / _-----=> irqs-off
2081 # | / _----=> need-resched
2082 # || / _---=> hardirq/softirq
2083 # ||| / _--=> preempt-depth
2084 # |||| / delay
2085 # cmd pid ||||| time | caller
2086 # \ / ||||| \ | /
2087 <idle>-0 3d.h4 0us : 0:120:R + [003] 2389: 94:R sleep
2088 <idle>-0 3d.h4 1us+: ttwu_do_activate.constprop.87 <-try_to_wake_up
2089 <idle>-0 3d..3 5us : __schedule <-schedule
2090 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep
2093 Running this on an idle system, we see that it only took 5 microseconds
2094 to perform the task switch. Note, since the trace point in the schedule
2095 is before the actual "switch", we stop the tracing when the recorded task
2096 is about to schedule in. This may change if we add a new marker at the
2097 end of the scheduler.
2099 Notice that the recorded task is 'sleep' with the PID of 2389
2100 and it has an rt_prio of 5. This priority is user-space priority
2101 and not the internal kernel priority. The policy is 1 for
2102 SCHED_FIFO and 2 for SCHED_RR.
2104 Note, that the trace data shows the internal priority (99 - rtprio).
2105 ::
2107 <idle>-0 3d..3 5us : 0:120:R ==> [003] 2389: 94:R sleep
2109 The 0:120:R means idle was running with a nice priority of 0 (120 - 120)
2110 and in the running state 'R'. The sleep task was scheduled in with
2111 2389: 94:R. That is the priority is the kernel rtprio (99 - 5 = 94)
2112 and it too is in the running state.
2114 Doing the same with chrt -r 5 and function-trace set.
2115 ::
2117 echo 1 > options/function-trace
2119 # tracer: wakeup_rt
2120 #
2121 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+
2122 # --------------------------------------------------------------------
2123 # latency: 29 us, #85/85, CPU#3 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
2124 # -----------------
2125 # | task: sleep-2448 (uid:0 nice:0 policy:1 rt_prio:5)
2126 # -----------------
2127 #
2128 # _------=> CPU#
2129 # / _-----=> irqs-off
2130 # | / _----=> need-resched
2131 # || / _---=> hardirq/softirq
2132 # ||| / _--=> preempt-depth
2133 # |||| / delay
2134 # cmd pid ||||| time | caller
2135 # \ / ||||| \ | /
2136 <idle>-0 3d.h4 1us+: 0:120:R + [003] 2448: 94:R sleep
2137 <idle>-0 3d.h4 2us : ttwu_do_activate.constprop.87 <-try_to_wake_up
2138 <idle>-0 3d.h3 3us : check_preempt_curr <-ttwu_do_wakeup
2139 <idle>-0 3d.h3 3us : resched_curr <-check_preempt_curr
2140 <idle>-0 3dNh3 4us : task_woken_rt <-ttwu_do_wakeup
2141 <idle>-0 3dNh3 4us : _raw_spin_unlock <-try_to_wake_up
2142 <idle>-0 3dNh3 4us : sub_preempt_count <-_raw_spin_unlock
2143 <idle>-0 3dNh2 5us : ttwu_stat <-try_to_wake_up
2144 <idle>-0 3dNh2 5us : _raw_spin_unlock_irqrestore <-try_to_wake_up
2145 <idle>-0 3dNh2 6us : sub_preempt_count <-_raw_spin_unlock_irqrestore
2146 <idle>-0 3dNh1 6us : _raw_spin_lock <-__run_hrtimer
2147 <idle>-0 3dNh1 6us : add_preempt_count <-_raw_spin_lock
2148 <idle>-0 3dNh2 7us : _raw_spin_unlock <-hrtimer_interrupt
2149 <idle>-0 3dNh2 7us : sub_preempt_count <-_raw_spin_unlock
2150 <idle>-0 3dNh1 7us : tick_program_event <-hrtimer_interrupt
2151 <idle>-0 3dNh1 7us : clockevents_program_event <-tick_program_event
2152 <idle>-0 3dNh1 8us : ktime_get <-clockevents_program_event
2153 <idle>-0 3dNh1 8us : lapic_next_event <-clockevents_program_event
2154 <idle>-0 3dNh1 8us : irq_exit <-smp_apic_timer_interrupt
2155 <idle>-0 3dNh1 9us : sub_preempt_count <-irq_exit
2156 <idle>-0 3dN.2 9us : idle_cpu <-irq_exit
2157 <idle>-0 3dN.2 9us : rcu_irq_exit <-irq_exit
2158 <idle>-0 3dN.2 10us : rcu_eqs_enter_common.isra.45 <-rcu_irq_exit
2159 <idle>-0 3dN.2 10us : sub_preempt_count <-irq_exit
2160 <idle>-0 3.N.1 11us : rcu_idle_exit <-cpu_idle
2161 <idle>-0 3dN.1 11us : rcu_eqs_exit_common.isra.43 <-rcu_idle_exit
2162 <idle>-0 3.N.1 11us : tick_nohz_idle_exit <-cpu_idle
2163 <idle>-0 3dN.1 12us : menu_hrtimer_cancel <-tick_nohz_idle_exit
2164 <idle>-0 3dN.1 12us : ktime_get <-tick_nohz_idle_exit
2165 <idle>-0 3dN.1 12us : tick_do_update_jiffies64 <-tick_nohz_idle_exit
2166 <idle>-0 3dN.1 13us : cpu_load_update_nohz <-tick_nohz_idle_exit
2167 <idle>-0 3dN.1 13us : _raw_spin_lock <-cpu_load_update_nohz
2168 <idle>-0 3dN.1 13us : add_preempt_count <-_raw_spin_lock
2169 <idle>-0 3dN.2 13us : __cpu_load_update <-cpu_load_update_nohz
2170 <idle>-0 3dN.2 14us : sched_avg_update <-__cpu_load_update
2171 <idle>-0 3dN.2 14us : _raw_spin_unlock <-cpu_load_update_nohz
2172 <idle>-0 3dN.2 14us : sub_preempt_count <-_raw_spin_unlock
2173 <idle>-0 3dN.1 15us : calc_load_nohz_stop <-tick_nohz_idle_exit
2174 <idle>-0 3dN.1 15us : touch_softlockup_watchdog <-tick_nohz_idle_exit
2175 <idle>-0 3dN.1 15us : hrtimer_cancel <-tick_nohz_idle_exit
2176 <idle>-0 3dN.1 15us : hrtimer_try_to_cancel <-hrtimer_cancel
2177 <idle>-0 3dN.1 16us : lock_hrtimer_base.isra.18 <-hrtimer_try_to_cancel
2178 <idle>-0 3dN.1 16us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18
2179 <idle>-0 3dN.1 16us : add_preempt_count <-_raw_spin_lock_irqsave
2180 <idle>-0 3dN.2 17us : __remove_hrtimer <-remove_hrtimer.part.16
2181 <idle>-0 3dN.2 17us : hrtimer_force_reprogram <-__remove_hrtimer
2182 <idle>-0 3dN.2 17us : tick_program_event <-hrtimer_force_reprogram
2183 <idle>-0 3dN.2 18us : clockevents_program_event <-tick_program_event
2184 <idle>-0 3dN.2 18us : ktime_get <-clockevents_program_event
2185 <idle>-0 3dN.2 18us : lapic_next_event <-clockevents_program_event
2186 <idle>-0 3dN.2 19us : _raw_spin_unlock_irqrestore <-hrtimer_try_to_cancel
2187 <idle>-0 3dN.2 19us : sub_preempt_count <-_raw_spin_unlock_irqrestore
2188 <idle>-0 3dN.1 19us : hrtimer_forward <-tick_nohz_idle_exit
2189 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward
2190 <idle>-0 3dN.1 20us : ktime_add_safe <-hrtimer_forward
2191 <idle>-0 3dN.1 20us : hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11
2192 <idle>-0 3dN.1 20us : __hrtimer_start_range_ns <-hrtimer_start_range_ns
2193 <idle>-0 3dN.1 21us : lock_hrtimer_base.isra.18 <-__hrtimer_start_range_ns
2194 <idle>-0 3dN.1 21us : _raw_spin_lock_irqsave <-lock_hrtimer_base.isra.18
2195 <idle>-0 3dN.1 21us : add_preempt_count <-_raw_spin_lock_irqsave
2196 <idle>-0 3dN.2 22us : ktime_add_safe <-__hrtimer_start_range_ns
2197 <idle>-0 3dN.2 22us : enqueue_hrtimer <-__hrtimer_start_range_ns
2198 <idle>-0 3dN.2 22us : tick_program_event <-__hrtimer_start_range_ns
2199 <idle>-0 3dN.2 23us : clockevents_program_event <-tick_program_event
2200 <idle>-0 3dN.2 23us : ktime_get <-clockevents_program_event
2201 <idle>-0 3dN.2 23us : lapic_next_event <-clockevents_program_event
2202 <idle>-0 3dN.2 24us : _raw_spin_unlock_irqrestore <-__hrtimer_start_range_ns
2203 <idle>-0 3dN.2 24us : sub_preempt_count <-_raw_spin_unlock_irqrestore
2204 <idle>-0 3dN.1 24us : account_idle_ticks <-tick_nohz_idle_exit
2205 <idle>-0 3dN.1 24us : account_idle_time <-account_idle_ticks
2206 <idle>-0 3.N.1 25us : sub_preempt_count <-cpu_idle
2207 <idle>-0 3.N.. 25us : schedule <-cpu_idle
2208 <idle>-0 3.N.. 25us : __schedule <-preempt_schedule
2209 <idle>-0 3.N.. 26us : add_preempt_count <-__schedule
2210 <idle>-0 3.N.1 26us : rcu_note_context_switch <-__schedule
2211 <idle>-0 3.N.1 26us : rcu_sched_qs <-rcu_note_context_switch
2212 <idle>-0 3dN.1 27us : rcu_preempt_qs <-rcu_note_context_switch
2213 <idle>-0 3.N.1 27us : _raw_spin_lock_irq <-__schedule
2214 <idle>-0 3dN.1 27us : add_preempt_count <-_raw_spin_lock_irq
2215 <idle>-0 3dN.2 28us : put_prev_task_idle <-__schedule
2216 <idle>-0 3dN.2 28us : pick_next_task_stop <-pick_next_task
2217 <idle>-0 3dN.2 28us : pick_next_task_rt <-pick_next_task
2218 <idle>-0 3dN.2 29us : dequeue_pushable_task <-pick_next_task_rt
2219 <idle>-0 3d..3 29us : __schedule <-preempt_schedule
2220 <idle>-0 3d..3 30us : 0:120:R ==> [003] 2448: 94:R sleep
2222 This isn't that big of a trace, even with function tracing enabled,
2223 so I included the entire trace.
2225 The interrupt went off while when the system was idle. Somewhere
2226 before task_woken_rt() was called, the NEED_RESCHED flag was set,
2227 this is indicated by the first occurrence of the 'N' flag.
2229 Latency tracing and events
2230 --------------------------
2231 As function tracing can induce a much larger latency, but without
2232 seeing what happens within the latency it is hard to know what
2233 caused it. There is a middle ground, and that is with enabling
2234 events.
2235 ::
2237 # echo 0 > options/function-trace
2238 # echo wakeup_rt > current_tracer
2239 # echo 1 > events/enable
2240 # echo 1 > tracing_on
2241 # echo 0 > tracing_max_latency
2242 # chrt -f 5 sleep 1
2243 # echo 0 > tracing_on
2244 # cat trace
2245 # tracer: wakeup_rt
2246 #
2247 # wakeup_rt latency trace v1.1.5 on 3.8.0-test+
2248 # --------------------------------------------------------------------
2249 # latency: 6 us, #12/12, CPU#2 | (M:preempt VP:0, KP:0, SP:0 HP:0 #P:4)
2250 # -----------------
2251 # | task: sleep-5882 (uid:0 nice:0 policy:1 rt_prio:5)
2252 # -----------------
2253 #
2254 # _------=> CPU#
2255 # / _-----=> irqs-off
2256 # | / _----=> need-resched
2257 # || / _---=> hardirq/softirq
2258 # ||| / _--=> preempt-depth
2259 # |||| / delay
2260 # cmd pid ||||| time | caller
2261 # \ / ||||| \ | /
2262 <idle>-0 2d.h4 0us : 0:120:R + [002] 5882: 94:R sleep
2263 <idle>-0 2d.h4 0us : ttwu_do_activate.constprop.87 <-try_to_wake_up
2264 <idle>-0 2d.h4 1us : sched_wakeup: comm=sleep pid=5882 prio=94 success=1 target_cpu=002
2265 <idle>-0 2dNh2 1us : hrtimer_expire_exit: hrtimer=ffff88007796feb8
2266 <idle>-0 2.N.2 2us : power_end: cpu_id=2
2267 <idle>-0 2.N.2 3us : cpu_idle: state=4294967295 cpu_id=2
2268 <idle>-0 2dN.3 4us : hrtimer_cancel: hrtimer=ffff88007d50d5e0
2269 <idle>-0 2dN.3 4us : hrtimer_start: hrtimer=ffff88007d50d5e0 function=tick_sched_timer expires=34311211000000 softexpires=34311211000000
2270 <idle>-0 2.N.2 5us : rcu_utilization: Start context switch
2271 <idle>-0 2.N.2 5us : rcu_utilization: End context switch
2272 <idle>-0 2d..3 6us : __schedule <-schedule
2273 <idle>-0 2d..3 6us : 0:120:R ==> [002] 5882: 94:R sleep
2276 Hardware Latency Detector
2277 -------------------------
2279 The hardware latency detector is executed by enabling the "hwlat" tracer.
2281 NOTE, this tracer will affect the performance of the system as it will
2282 periodically make a CPU constantly busy with interrupts disabled.
2283 ::
2285 # echo hwlat > current_tracer
2286 # sleep 100
2287 # cat trace
2288 # tracer: hwlat
2289 #
2290 # entries-in-buffer/entries-written: 13/13 #P:8
2291 #
2292 # _-----=> irqs-off
2293 # / _----=> need-resched
2294 # | / _---=> hardirq/softirq
2295 # || / _--=> preempt-depth
2296 # ||| / delay
2297 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
2298 # | | | |||| | |
2299 <...>-1729 [001] d... 678.473449: #1 inner/outer(us): 11/12 ts:1581527483.343962693 count:6
2300 <...>-1729 [004] d... 689.556542: #2 inner/outer(us): 16/9 ts:1581527494.889008092 count:1
2301 <...>-1729 [005] d... 714.756290: #3 inner/outer(us): 16/16 ts:1581527519.678961629 count:5
2302 <...>-1729 [001] d... 718.788247: #4 inner/outer(us): 9/17 ts:1581527523.889012713 count:1
2303 <...>-1729 [002] d... 719.796341: #5 inner/outer(us): 13/9 ts:1581527524.912872606 count:1
2304 <...>-1729 [006] d... 844.787091: #6 inner/outer(us): 9/12 ts:1581527649.889048502 count:2
2305 <...>-1729 [003] d... 849.827033: #7 inner/outer(us): 18/9 ts:1581527654.889013793 count:1
2306 <...>-1729 [007] d... 853.859002: #8 inner/outer(us): 9/12 ts:1581527658.889065736 count:1
2307 <...>-1729 [001] d... 855.874978: #9 inner/outer(us): 9/11 ts:1581527660.861991877 count:1
2308 <...>-1729 [001] d... 863.938932: #10 inner/outer(us): 9/11 ts:1581527668.970010500 count:1 nmi-total:7 nmi-count:1
2309 <...>-1729 [007] d... 878.050780: #11 inner/outer(us): 9/12 ts:1581527683.385002600 count:1 nmi-total:5 nmi-count:1
2310 <...>-1729 [007] d... 886.114702: #12 inner/outer(us): 9/12 ts:1581527691.385001600 count:1
2313 The above output is somewhat the same in the header. All events will have
2314 interrupts disabled 'd'. Under the FUNCTION title there is:
2316 #1
2317 This is the count of events recorded that were greater than the
2318 tracing_threshold (See below).
2320 inner/outer(us): 11/11
2322 This shows two numbers as "inner latency" and "outer latency". The test
2323 runs in a loop checking a timestamp twice. The latency detected within
2324 the two timestamps is the "inner latency" and the latency detected
2325 after the previous timestamp and the next timestamp in the loop is
2326 the "outer latency".
2328 ts:1581527483.343962693
2330 The absolute timestamp that the first latency was recorded in the window.
2332 count:6
2334 The number of times a latency was detected during the window.
2336 nmi-total:7 nmi-count:1
2338 On architectures that support it, if an NMI comes in during the
2339 test, the time spent in NMI is reported in "nmi-total" (in
2340 microseconds).
2342 All architectures that have NMIs will show the "nmi-count" if an
2343 NMI comes in during the test.
2345 hwlat files:
2347 tracing_threshold
2348 This gets automatically set to "10" to represent 10
2349 microseconds. This is the threshold of latency that
2350 needs to be detected before the trace will be recorded.
2352 Note, when hwlat tracer is finished (another tracer is
2353 written into "current_tracer"), the original value for
2354 tracing_threshold is placed back into this file.
2356 hwlat_detector/width
2357 The length of time the test runs with interrupts disabled.
2359 hwlat_detector/window
2360 The length of time of the window which the test
2361 runs. That is, the test will run for "width"
2362 microseconds per "window" microseconds
2364 tracing_cpumask
2365 When the test is started. A kernel thread is created that
2366 runs the test. This thread will alternate between CPUs
2367 listed in the tracing_cpumask between each period
2368 (one "window"). To limit the test to specific CPUs
2369 set the mask in this file to only the CPUs that the test
2370 should run on.
2372 function
2373 --------
2375 This tracer is the function tracer. Enabling the function tracer
2376 can be done from the debug file system. Make sure the
2377 ftrace_enabled is set; otherwise this tracer is a nop.
2378 See the "ftrace_enabled" section below.
2379 ::
2381 # sysctl kernel.ftrace_enabled=1
2382 # echo function > current_tracer
2383 # echo 1 > tracing_on
2384 # usleep 1
2385 # echo 0 > tracing_on
2386 # cat trace
2387 # tracer: function
2388 #
2389 # entries-in-buffer/entries-written: 24799/24799 #P:4
2390 #
2391 # _-----=> irqs-off
2392 # / _----=> need-resched
2393 # | / _---=> hardirq/softirq
2394 # || / _--=> preempt-depth
2395 # ||| / delay
2396 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
2397 # | | | |||| | |
2398 bash-1994 [002] .... 3082.063030: mutex_unlock <-rb_simple_write
2399 bash-1994 [002] .... 3082.063031: __mutex_unlock_slowpath <-mutex_unlock
2400 bash-1994 [002] .... 3082.063031: __fsnotify_parent <-fsnotify_modify
2401 bash-1994 [002] .... 3082.063032: fsnotify <-fsnotify_modify
2402 bash-1994 [002] .... 3082.063032: __srcu_read_lock <-fsnotify
2403 bash-1994 [002] .... 3082.063032: add_preempt_count <-__srcu_read_lock
2404 bash-1994 [002] ...1 3082.063032: sub_preempt_count <-__srcu_read_lock
2405 bash-1994 [002] .... 3082.063033: __srcu_read_unlock <-fsnotify
2406 [...]
2409 Note: function tracer uses ring buffers to store the above
2410 entries. The newest data may overwrite the oldest data.
2411 Sometimes using echo to stop the trace is not sufficient because
2412 the tracing could have overwritten the data that you wanted to
2413 record. For this reason, it is sometimes better to disable
2414 tracing directly from a program. This allows you to stop the
2415 tracing at the point that you hit the part that you are
2416 interested in. To disable the tracing directly from a C program,
2417 something like following code snippet can be used::
2419 int trace_fd;
2420 [...]
2421 int main(int argc, char *argv[]) {
2422 [...]
2423 trace_fd = open(tracing_file("tracing_on"), O_WRONLY);
2424 [...]
2425 if (condition_hit()) {
2426 write(trace_fd, "0", 1);
2427 }
2428 [...]
2429 }
2432 Single thread tracing
2433 ---------------------
2435 By writing into set_ftrace_pid you can trace a
2436 single thread. For example::
2438 # cat set_ftrace_pid
2439 no pid
2440 # echo 3111 > set_ftrace_pid
2441 # cat set_ftrace_pid
2442 3111
2443 # echo function > current_tracer
2444 # cat trace | head
2445 # tracer: function
2446 #
2447 # TASK-PID CPU# TIMESTAMP FUNCTION
2448 # | | | | |
2449 yum-updatesd-3111 [003] 1637.254676: finish_task_switch <-thread_return
2450 yum-updatesd-3111 [003] 1637.254681: hrtimer_cancel <-schedule_hrtimeout_range
2451 yum-updatesd-3111 [003] 1637.254682: hrtimer_try_to_cancel <-hrtimer_cancel
2452 yum-updatesd-3111 [003] 1637.254683: lock_hrtimer_base <-hrtimer_try_to_cancel
2453 yum-updatesd-3111 [003] 1637.254685: fget_light <-do_sys_poll
2454 yum-updatesd-3111 [003] 1637.254686: pipe_poll <-do_sys_poll
2455 # echo > set_ftrace_pid
2456 # cat trace |head
2457 # tracer: function
2458 #
2459 # TASK-PID CPU# TIMESTAMP FUNCTION
2460 # | | | | |
2461 ##### CPU 3 buffer started ####
2462 yum-updatesd-3111 [003] 1701.957688: free_poll_entry <-poll_freewait
2463 yum-updatesd-3111 [003] 1701.957689: remove_wait_queue <-free_poll_entry
2464 yum-updatesd-3111 [003] 1701.957691: fput <-free_poll_entry
2465 yum-updatesd-3111 [003] 1701.957692: audit_syscall_exit <-sysret_audit
2466 yum-updatesd-3111 [003] 1701.957693: path_put <-audit_syscall_exit
2468 If you want to trace a function when executing, you could use
2469 something like this simple program.
2470 ::
2472 #include <stdio.h>
2473 #include <stdlib.h>
2474 #include <sys/types.h>
2475 #include <sys/stat.h>
2476 #include <fcntl.h>
2477 #include <unistd.h>
2478 #include <string.h>
2480 #define _STR(x) #x
2481 #define STR(x) _STR(x)
2482 #define MAX_PATH 256
2484 const char *find_tracefs(void)
2485 {
2486 static char tracefs[MAX_PATH+1];
2487 static int tracefs_found;
2488 char type[100];
2489 FILE *fp;
2491 if (tracefs_found)
2492 return tracefs;
2494 if ((fp = fopen("/proc/mounts","r")) == NULL) {
2495 perror("/proc/mounts");
2496 return NULL;
2497 }
2499 while (fscanf(fp, "%*s %"
2500 STR(MAX_PATH)
2501 "s %99s %*s %*d %*d\n",
2502 tracefs, type) == 2) {
2503 if (strcmp(type, "tracefs") == 0)
2504 break;
2505 }
2506 fclose(fp);
2508 if (strcmp(type, "tracefs") != 0) {
2509 fprintf(stderr, "tracefs not mounted");
2510 return NULL;
2511 }
2513 strcat(tracefs, "/tracing/");
2514 tracefs_found = 1;
2516 return tracefs;
2517 }
2519 const char *tracing_file(const char *file_name)
2520 {
2521 static char trace_file[MAX_PATH+1];
2522 snprintf(trace_file, MAX_PATH, "%s/%s", find_tracefs(), file_name);
2523 return trace_file;
2524 }
2526 int main (int argc, char **argv)
2527 {
2528 if (argc < 1)
2529 exit(-1);
2531 if (fork() > 0) {
2532 int fd, ffd;
2533 char line[64];
2534 int s;
2536 ffd = open(tracing_file("current_tracer"), O_WRONLY);
2537 if (ffd < 0)
2538 exit(-1);
2539 write(ffd, "nop", 3);
2541 fd = open(tracing_file("set_ftrace_pid"), O_WRONLY);
2542 s = sprintf(line, "%d\n", getpid());
2543 write(fd, line, s);
2545 write(ffd, "function", 8);
2547 close(fd);
2548 close(ffd);
2550 execvp(argv[1], argv+1);
2551 }
2553 return 0;
2554 }
2556 Or this simple script!
2557 ::
2559 #!/bin/bash
2561 tracefs=`sed -ne 's/^tracefs \(.*\) tracefs.*/\1/p' /proc/mounts`
2562 echo 0 > $tracefs/tracing_on
2563 echo $$ > $tracefs/set_ftrace_pid
2564 echo function > $tracefs/current_tracer
2565 echo 1 > $tracefs/tracing_on
2566 exec "$@"
2569 function graph tracer
2570 ---------------------------
2572 This tracer is similar to the function tracer except that it
2573 probes a function on its entry and its exit. This is done by
2574 using a dynamically allocated stack of return addresses in each
2575 task_struct. On function entry the tracer overwrites the return
2576 address of each function traced to set a custom probe. Thus the
2577 original return address is stored on the stack of return address
2578 in the task_struct.
2580 Probing on both ends of a function leads to special features
2581 such as:
2583 - measure of a function's time execution
2584 - having a reliable call stack to draw function calls graph
2586 This tracer is useful in several situations:
2588 - you want to find the reason of a strange kernel behavior and
2589 need to see what happens in detail on any areas (or specific
2590 ones).
2592 - you are experiencing weird latencies but it's difficult to
2593 find its origin.
2595 - you want to find quickly which path is taken by a specific
2596 function
2598 - you just want to peek inside a working kernel and want to see
2599 what happens there.
2601 ::
2603 # tracer: function_graph
2604 #
2605 # CPU DURATION FUNCTION CALLS
2606 # | | | | | | |
2608 0) | sys_open() {
2609 0) | do_sys_open() {
2610 0) | getname() {
2611 0) | kmem_cache_alloc() {
2612 0) 1.382 us | __might_sleep();
2613 0) 2.478 us | }
2614 0) | strncpy_from_user() {
2615 0) | might_fault() {
2616 0) 1.389 us | __might_sleep();
2617 0) 2.553 us | }
2618 0) 3.807 us | }
2619 0) 7.876 us | }
2620 0) | alloc_fd() {
2621 0) 0.668 us | _spin_lock();
2622 0) 0.570 us | expand_files();
2623 0) 0.586 us | _spin_unlock();
2626 There are several columns that can be dynamically
2627 enabled/disabled. You can use every combination of options you
2628 want, depending on your needs.
2630 - The cpu number on which the function executed is default
2631 enabled. It is sometimes better to only trace one cpu (see
2632 tracing_cpumask file) or you might sometimes see unordered
2633 function calls while cpu tracing switch.
2635 - hide: echo nofuncgraph-cpu > trace_options
2636 - show: echo funcgraph-cpu > trace_options
2638 - The duration (function's time of execution) is displayed on
2639 the closing bracket line of a function or on the same line
2640 than the current function in case of a leaf one. It is default
2641 enabled.
2643 - hide: echo nofuncgraph-duration > trace_options
2644 - show: echo funcgraph-duration > trace_options
2646 - The overhead field precedes the duration field in case of
2647 reached duration thresholds.
2649 - hide: echo nofuncgraph-overhead > trace_options
2650 - show: echo funcgraph-overhead > trace_options
2651 - depends on: funcgraph-duration
2653 ie::
2655 3) # 1837.709 us | } /* __switch_to */
2656 3) | finish_task_switch() {
2657 3) 0.313 us | _raw_spin_unlock_irq();
2658 3) 3.177 us | }
2659 3) # 1889.063 us | } /* __schedule */
2660 3) ! 140.417 us | } /* __schedule */
2661 3) # 2034.948 us | } /* schedule */
2662 3) * 33998.59 us | } /* schedule_preempt_disabled */
2664 [...]
2666 1) 0.260 us | msecs_to_jiffies();
2667 1) 0.313 us | __rcu_read_unlock();
2668 1) + 61.770 us | }
2669 1) + 64.479 us | }
2670 1) 0.313 us | rcu_bh_qs();
2671 1) 0.313 us | __local_bh_enable();
2672 1) ! 217.240 us | }
2673 1) 0.365 us | idle_cpu();
2674 1) | rcu_irq_exit() {
2675 1) 0.417 us | rcu_eqs_enter_common.isra.47();
2676 1) 3.125 us | }
2677 1) ! 227.812 us | }
2678 1) ! 457.395 us | }
2679 1) @ 119760.2 us | }
2681 [...]
2683 2) | handle_IPI() {
2684 1) 6.979 us | }
2685 2) 0.417 us | scheduler_ipi();
2686 1) 9.791 us | }
2687 1) + 12.917 us | }
2688 2) 3.490 us | }
2689 1) + 15.729 us | }
2690 1) + 18.542 us | }
2691 2) $ 3594274 us | }
2693 Flags::
2695 + means that the function exceeded 10 usecs.
2696 ! means that the function exceeded 100 usecs.
2697 # means that the function exceeded 1000 usecs.
2698 * means that the function exceeded 10 msecs.
2699 @ means that the function exceeded 100 msecs.
2700 $ means that the function exceeded 1 sec.
2703 - The task/pid field displays the thread cmdline and pid which
2704 executed the function. It is default disabled.
2706 - hide: echo nofuncgraph-proc > trace_options
2707 - show: echo funcgraph-proc > trace_options
2709 ie::
2711 # tracer: function_graph
2712 #
2713 # CPU TASK/PID DURATION FUNCTION CALLS
2714 # | | | | | | | | |
2715 0) sh-4802 | | d_free() {
2716 0) sh-4802 | | call_rcu() {
2717 0) sh-4802 | | __call_rcu() {
2718 0) sh-4802 | 0.616 us | rcu_process_gp_end();
2719 0) sh-4802 | 0.586 us | check_for_new_grace_period();
2720 0) sh-4802 | 2.899 us | }
2721 0) sh-4802 | 4.040 us | }
2722 0) sh-4802 | 5.151 us | }
2723 0) sh-4802 | + 49.370 us | }
2726 - The absolute time field is an absolute timestamp given by the
2727 system clock since it started. A snapshot of this time is
2728 given on each entry/exit of functions
2730 - hide: echo nofuncgraph-abstime > trace_options
2731 - show: echo funcgraph-abstime > trace_options
2733 ie::
2735 #
2736 # TIME CPU DURATION FUNCTION CALLS
2737 # | | | | | | | |
2738 360.774522 | 1) 0.541 us | }
2739 360.774522 | 1) 4.663 us | }
2740 360.774523 | 1) 0.541 us | __wake_up_bit();
2741 360.774524 | 1) 6.796 us | }
2742 360.774524 | 1) 7.952 us | }
2743 360.774525 | 1) 9.063 us | }
2744 360.774525 | 1) 0.615 us | journal_mark_dirty();
2745 360.774527 | 1) 0.578 us | __brelse();
2746 360.774528 | 1) | reiserfs_prepare_for_journal() {
2747 360.774528 | 1) | unlock_buffer() {
2748 360.774529 | 1) | wake_up_bit() {
2749 360.774529 | 1) | bit_waitqueue() {
2750 360.774530 | 1) 0.594 us | __phys_addr();
2753 The function name is always displayed after the closing bracket
2754 for a function if the start of that function is not in the
2755 trace buffer.
2757 Display of the function name after the closing bracket may be
2758 enabled for functions whose start is in the trace buffer,
2759 allowing easier searching with grep for function durations.
2760 It is default disabled.
2762 - hide: echo nofuncgraph-tail > trace_options
2763 - show: echo funcgraph-tail > trace_options
2765 Example with nofuncgraph-tail (default)::
2767 0) | putname() {
2768 0) | kmem_cache_free() {
2769 0) 0.518 us | __phys_addr();
2770 0) 1.757 us | }
2771 0) 2.861 us | }
2773 Example with funcgraph-tail::
2775 0) | putname() {
2776 0) | kmem_cache_free() {
2777 0) 0.518 us | __phys_addr();
2778 0) 1.757 us | } /* kmem_cache_free() */
2779 0) 2.861 us | } /* putname() */
2781 The return value of each traced function can be displayed after
2782 an equal sign "=". When encountering system call failures, it
2783 can be very helpful to quickly locate the function that first
2784 returns an error code.
2786 - hide: echo nofuncgraph-retval > trace_options
2787 - show: echo funcgraph-retval > trace_options
2789 Example with funcgraph-retval::
2791 1) | cgroup_migrate() {
2792 1) 0.651 us | cgroup_migrate_add_task(); /* = 0xffff93fcfd346c00 */
2793 1) | cgroup_migrate_execute() {
2794 1) | cpu_cgroup_can_attach() {
2795 1) | cgroup_taskset_first() {
2796 1) 0.732 us | cgroup_taskset_next(); /* = 0xffff93fc8fb20000 */
2797 1) 1.232 us | } /* cgroup_taskset_first = 0xffff93fc8fb20000 */
2798 1) 0.380 us | sched_rt_can_attach(); /* = 0x0 */
2799 1) 2.335 us | } /* cpu_cgroup_can_attach = -22 */
2800 1) 4.369 us | } /* cgroup_migrate_execute = -22 */
2801 1) 7.143 us | } /* cgroup_migrate = -22 */
2803 The above example shows that the function cpu_cgroup_can_attach
2804 returned the error code -22 firstly, then we can read the code
2805 of this function to get the root cause.
2807 When the option funcgraph-retval-hex is not set, the return value can
2808 be displayed in a smart way. Specifically, if it is an error code,
2809 it will be printed in signed decimal format, otherwise it will
2810 printed in hexadecimal format.
2812 - smart: echo nofuncgraph-retval-hex > trace_options
2813 - hexadecimal: echo funcgraph-retval-hex > trace_options
2815 Example with funcgraph-retval-hex::
2817 1) | cgroup_migrate() {
2818 1) 0.651 us | cgroup_migrate_add_task(); /* = 0xffff93fcfd346c00 */
2819 1) | cgroup_migrate_execute() {
2820 1) | cpu_cgroup_can_attach() {
2821 1) | cgroup_taskset_first() {
2822 1) 0.732 us | cgroup_taskset_next(); /* = 0xffff93fc8fb20000 */
2823 1) 1.232 us | } /* cgroup_taskset_first = 0xffff93fc8fb20000 */
2824 1) 0.380 us | sched_rt_can_attach(); /* = 0x0 */
2825 1) 2.335 us | } /* cpu_cgroup_can_attach = 0xffffffea */
2826 1) 4.369 us | } /* cgroup_migrate_execute = 0xffffffea */
2827 1) 7.143 us | } /* cgroup_migrate = 0xffffffea */
2829 At present, there are some limitations when using the funcgraph-retval
2830 option, and these limitations will be eliminated in the future:
2832 - Even if the function return type is void, a return value will still
2833 be printed, and you can just ignore it.
2835 - Even if return values are stored in multiple registers, only the
2836 value contained in the first register will be recorded and printed.
2837 To illustrate, in the x86 architecture, eax and edx are used to store
2838 a 64-bit return value, with the lower 32 bits saved in eax and the
2839 upper 32 bits saved in edx. However, only the value stored in eax
2840 will be recorded and printed.
2842 - In certain procedure call standards, such as arm64's AAPCS64, when a
2843 type is smaller than a GPR, it is the responsibility of the consumer
2844 to perform the narrowing, and the upper bits may contain UNKNOWN values.
2845 Therefore, it is advisable to check the code for such cases. For instance,
2846 when using a u8 in a 64-bit GPR, bits [63:8] may contain arbitrary values,
2847 especially when larger types are truncated, whether explicitly or implicitly.
2848 Here are some specific cases to illustrate this point:
2850 **Case One**:
2852 The function narrow_to_u8 is defined as follows::
2854 u8 narrow_to_u8(u64 val)
2855 {
2856 // implicitly truncated
2857 return val;
2858 }
2860 It may be compiled to::
2862 narrow_to_u8:
2863 < ... ftrace instrumentation ... >
2864 RET
2866 If you pass 0x123456789abcdef to this function and want to narrow it,
2867 it may be recorded as 0x123456789abcdef instead of 0xef.
2869 **Case Two**:
2871 The function error_if_not_4g_aligned is defined as follows::
2873 int error_if_not_4g_aligned(u64 val)
2874 {
2875 if (val & GENMASK(31, 0))
2876 return -EINVAL;
2878 return 0;
2879 }
2881 It could be compiled to::
2883 error_if_not_4g_aligned:
2884 CBNZ w0, .Lnot_aligned
2885 RET // bits [31:0] are zero, bits
2886 // [63:32] are UNKNOWN
2887 .Lnot_aligned:
2888 MOV x0, #-EINVAL
2889 RET
2891 When passing 0x2_0000_0000 to it, the return value may be recorded as
2892 0x2_0000_0000 instead of 0.
2894 You can put some comments on specific functions by using
2895 trace_printk() For example, if you want to put a comment inside
2896 the __might_sleep() function, you just have to include
2897 <linux/ftrace.h> and call trace_printk() inside __might_sleep()::
2899 trace_printk("I'm a comment!\n")
2901 will produce::
2903 1) | __might_sleep() {
2904 1) | /* I'm a comment! */
2905 1) 1.449 us | }
2908 You might find other useful features for this tracer in the
2909 following "dynamic ftrace" section such as tracing only specific
2910 functions or tasks.
2912 dynamic ftrace
2913 --------------
2915 If CONFIG_DYNAMIC_FTRACE is set, the system will run with
2916 virtually no overhead when function tracing is disabled. The way
2917 this works is the mcount function call (placed at the start of
2918 every kernel function, produced by the -pg switch in gcc),
2919 starts of pointing to a simple return. (Enabling FTRACE will
2920 include the -pg switch in the compiling of the kernel.)
2922 At compile time every C file object is run through the
2923 recordmcount program (located in the scripts directory). This
2924 program will parse the ELF headers in the C object to find all
2925 the locations in the .text section that call mcount. Starting
2926 with gcc version 4.6, the -mfentry has been added for x86, which
2927 calls "__fentry__" instead of "mcount". Which is called before
2928 the creation of the stack frame.
2930 Note, not all sections are traced. They may be prevented by either
2931 a notrace, or blocked another way and all inline functions are not
2932 traced. Check the "available_filter_functions" file to see what functions
2933 can be traced.
2935 A section called "__mcount_loc" is created that holds
2936 references to all the mcount/fentry call sites in the .text section.
2937 The recordmcount program re-links this section back into the
2938 original object. The final linking stage of the kernel will add all these
2939 references into a single table.
2941 On boot up, before SMP is initialized, the dynamic ftrace code
2942 scans this table and updates all the locations into nops. It
2943 also records the locations, which are added to the
2944 available_filter_functions list. Modules are processed as they
2945 are loaded and before they are executed. When a module is
2946 unloaded, it also removes its functions from the ftrace function
2947 list. This is automatic in the module unload code, and the
2948 module author does not need to worry about it.
2950 When tracing is enabled, the process of modifying the function
2951 tracepoints is dependent on architecture. The old method is to use
2952 kstop_machine to prevent races with the CPUs executing code being
2953 modified (which can cause the CPU to do undesirable things, especially
2954 if the modified code crosses cache (or page) boundaries), and the nops are
2955 patched back to calls. But this time, they do not call mcount
2956 (which is just a function stub). They now call into the ftrace
2957 infrastructure.
2959 The new method of modifying the function tracepoints is to place
2960 a breakpoint at the location to be modified, sync all CPUs, modify
2961 the rest of the instruction not covered by the breakpoint. Sync
2962 all CPUs again, and then remove the breakpoint with the finished
2963 version to the ftrace call site.
2965 Some archs do not even need to monkey around with the synchronization,
2966 and can just slap the new code on top of the old without any
2967 problems with other CPUs executing it at the same time.
2969 One special side-effect to the recording of the functions being
2970 traced is that we can now selectively choose which functions we
2971 wish to trace and which ones we want the mcount calls to remain
2972 as nops.
2974 Two files are used, one for enabling and one for disabling the
2975 tracing of specified functions. They are:
2977 set_ftrace_filter
2979 and
2981 set_ftrace_notrace
2983 A list of available functions that you can add to these files is
2984 listed in:
2986 available_filter_functions
2988 ::
2990 # cat available_filter_functions
2991 put_prev_task_idle
2992 kmem_cache_create
2993 pick_next_task_rt
2994 cpus_read_lock
2995 pick_next_task_fair
2996 mutex_lock
2997 [...]
2999 If I am only interested in sys_nanosleep and hrtimer_interrupt::
3001 # echo sys_nanosleep hrtimer_interrupt > set_ftrace_filter
3002 # echo function > current_tracer
3003 # echo 1 > tracing_on
3004 # usleep 1
3005 # echo 0 > tracing_on
3006 # cat trace
3007 # tracer: function
3008 #
3009 # entries-in-buffer/entries-written: 5/5 #P:4
3010 #
3011 # _-----=> irqs-off
3012 # / _----=> need-resched
3013 # | / _---=> hardirq/softirq
3014 # || / _--=> preempt-depth
3015 # ||| / delay
3016 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3017 # | | | |||| | |
3018 usleep-2665 [001] .... 4186.475355: sys_nanosleep <-system_call_fastpath
3019 <idle>-0 [001] d.h1 4186.475409: hrtimer_interrupt <-smp_apic_timer_interrupt
3020 usleep-2665 [001] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt
3021 <idle>-0 [003] d.h1 4186.475426: hrtimer_interrupt <-smp_apic_timer_interrupt
3022 <idle>-0 [002] d.h1 4186.475427: hrtimer_interrupt <-smp_apic_timer_interrupt
3024 To see which functions are being traced, you can cat the file:
3025 ::
3027 # cat set_ftrace_filter
3028 hrtimer_interrupt
3029 sys_nanosleep
3032 Perhaps this is not enough. The filters also allow glob(7) matching.
3034 ``<match>*``
3035 will match functions that begin with <match>
3036 ``*<match>``
3037 will match functions that end with <match>
3038 ``*<match>*``
3039 will match functions that have <match> in it
3040 ``<match1>*<match2>``
3041 will match functions that begin with <match1> and end with <match2>
3043 .. note::
3044 It is better to use quotes to enclose the wild cards,
3045 otherwise the shell may expand the parameters into names
3046 of files in the local directory.
3048 ::
3050 # echo 'hrtimer_*' > set_ftrace_filter
3052 Produces::
3054 # tracer: function
3055 #
3056 # entries-in-buffer/entries-written: 897/897 #P:4
3057 #
3058 # _-----=> irqs-off
3059 # / _----=> need-resched
3060 # | / _---=> hardirq/softirq
3061 # || / _--=> preempt-depth
3062 # ||| / delay
3063 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3064 # | | | |||| | |
3065 <idle>-0 [003] dN.1 4228.547803: hrtimer_cancel <-tick_nohz_idle_exit
3066 <idle>-0 [003] dN.1 4228.547804: hrtimer_try_to_cancel <-hrtimer_cancel
3067 <idle>-0 [003] dN.2 4228.547805: hrtimer_force_reprogram <-__remove_hrtimer
3068 <idle>-0 [003] dN.1 4228.547805: hrtimer_forward <-tick_nohz_idle_exit
3069 <idle>-0 [003] dN.1 4228.547805: hrtimer_start_range_ns <-hrtimer_start_expires.constprop.11
3070 <idle>-0 [003] d..1 4228.547858: hrtimer_get_next_event <-get_next_timer_interrupt
3071 <idle>-0 [003] d..1 4228.547859: hrtimer_start <-__tick_nohz_idle_enter
3072 <idle>-0 [003] d..2 4228.547860: hrtimer_force_reprogram <-__rem
3074 Notice that we lost the sys_nanosleep.
3075 ::
3077 # cat set_ftrace_filter
3078 hrtimer_run_queues
3079 hrtimer_run_pending
3080 hrtimer_init
3081 hrtimer_cancel
3082 hrtimer_try_to_cancel
3083 hrtimer_forward
3084 hrtimer_start
3085 hrtimer_reprogram
3086 hrtimer_force_reprogram
3087 hrtimer_get_next_event
3088 hrtimer_interrupt
3089 hrtimer_nanosleep
3090 hrtimer_wakeup
3091 hrtimer_get_remaining
3092 hrtimer_get_res
3093 hrtimer_init_sleeper
3096 This is because the '>' and '>>' act just like they do in bash.
3097 To rewrite the filters, use '>'
3098 To append to the filters, use '>>'
3100 To clear out a filter so that all functions will be recorded
3101 again::
3103 # echo > set_ftrace_filter
3104 # cat set_ftrace_filter
3105 #
3107 Again, now we want to append.
3109 ::
3111 # echo sys_nanosleep > set_ftrace_filter
3112 # cat set_ftrace_filter
3113 sys_nanosleep
3114 # echo 'hrtimer_*' >> set_ftrace_filter
3115 # cat set_ftrace_filter
3116 hrtimer_run_queues
3117 hrtimer_run_pending
3118 hrtimer_init
3119 hrtimer_cancel
3120 hrtimer_try_to_cancel
3121 hrtimer_forward
3122 hrtimer_start
3123 hrtimer_reprogram
3124 hrtimer_force_reprogram
3125 hrtimer_get_next_event
3126 hrtimer_interrupt
3127 sys_nanosleep
3128 hrtimer_nanosleep
3129 hrtimer_wakeup
3130 hrtimer_get_remaining
3131 hrtimer_get_res
3132 hrtimer_init_sleeper
3135 The set_ftrace_notrace prevents those functions from being
3136 traced.
3137 ::
3139 # echo '*preempt*' '*lock*' > set_ftrace_notrace
3141 Produces::
3143 # tracer: function
3144 #
3145 # entries-in-buffer/entries-written: 39608/39608 #P:4
3146 #
3147 # _-----=> irqs-off
3148 # / _----=> need-resched
3149 # | / _---=> hardirq/softirq
3150 # || / _--=> preempt-depth
3151 # ||| / delay
3152 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3153 # | | | |||| | |
3154 bash-1994 [000] .... 4342.324896: file_ra_state_init <-do_dentry_open
3155 bash-1994 [000] .... 4342.324897: open_check_o_direct <-do_last
3156 bash-1994 [000] .... 4342.324897: ima_file_check <-do_last
3157 bash-1994 [000] .... 4342.324898: process_measurement <-ima_file_check
3158 bash-1994 [000] .... 4342.324898: ima_get_action <-process_measurement
3159 bash-1994 [000] .... 4342.324898: ima_match_policy <-ima_get_action
3160 bash-1994 [000] .... 4342.324899: do_truncate <-do_last
3161 bash-1994 [000] .... 4342.324899: setattr_should_drop_suidgid <-do_truncate
3162 bash-1994 [000] .... 4342.324899: notify_change <-do_truncate
3163 bash-1994 [000] .... 4342.324900: current_fs_time <-notify_change
3164 bash-1994 [000] .... 4342.324900: current_kernel_time <-current_fs_time
3165 bash-1994 [000] .... 4342.324900: timespec_trunc <-current_fs_time
3167 We can see that there's no more lock or preempt tracing.
3169 Selecting function filters via index
3170 ------------------------------------
3172 Because processing of strings is expensive (the address of the function
3173 needs to be looked up before comparing to the string being passed in),
3174 an index can be used as well to enable functions. This is useful in the
3175 case of setting thousands of specific functions at a time. By passing
3176 in a list of numbers, no string processing will occur. Instead, the function
3177 at the specific location in the internal array (which corresponds to the
3178 functions in the "available_filter_functions" file), is selected.
3180 ::
3182 # echo 1 > set_ftrace_filter
3184 Will select the first function listed in "available_filter_functions"
3186 ::
3188 # head -1 available_filter_functions
3189 trace_initcall_finish_cb
3191 # cat set_ftrace_filter
3192 trace_initcall_finish_cb
3194 # head -50 available_filter_functions | tail -1
3195 x86_pmu_commit_txn
3197 # echo 1 50 > set_ftrace_filter
3198 # cat set_ftrace_filter
3199 trace_initcall_finish_cb
3200 x86_pmu_commit_txn
3202 Dynamic ftrace with the function graph tracer
3203 ---------------------------------------------
3205 Although what has been explained above concerns both the
3206 function tracer and the function-graph-tracer, there are some
3207 special features only available in the function-graph tracer.
3209 If you want to trace only one function and all of its children,
3210 you just have to echo its name into set_graph_function::
3212 echo __do_fault > set_graph_function
3214 will produce the following "expanded" trace of the __do_fault()
3215 function::
3217 0) | __do_fault() {
3218 0) | filemap_fault() {
3219 0) | find_lock_page() {
3220 0) 0.804 us | find_get_page();
3221 0) | __might_sleep() {
3222 0) 1.329 us | }
3223 0) 3.904 us | }
3224 0) 4.979 us | }
3225 0) 0.653 us | _spin_lock();
3226 0) 0.578 us | page_add_file_rmap();
3227 0) 0.525 us | native_set_pte_at();
3228 0) 0.585 us | _spin_unlock();
3229 0) | unlock_page() {
3230 0) 0.541 us | page_waitqueue();
3231 0) 0.639 us | __wake_up_bit();
3232 0) 2.786 us | }
3233 0) + 14.237 us | }
3234 0) | __do_fault() {
3235 0) | filemap_fault() {
3236 0) | find_lock_page() {
3237 0) 0.698 us | find_get_page();
3238 0) | __might_sleep() {
3239 0) 1.412 us | }
3240 0) 3.950 us | }
3241 0) 5.098 us | }
3242 0) 0.631 us | _spin_lock();
3243 0) 0.571 us | page_add_file_rmap();
3244 0) 0.526 us | native_set_pte_at();
3245 0) 0.586 us | _spin_unlock();
3246 0) | unlock_page() {
3247 0) 0.533 us | page_waitqueue();
3248 0) 0.638 us | __wake_up_bit();
3249 0) 2.793 us | }
3250 0) + 14.012 us | }
3252 You can also expand several functions at once::
3254 echo sys_open > set_graph_function
3255 echo sys_close >> set_graph_function
3257 Now if you want to go back to trace all functions you can clear
3258 this special filter via::
3260 echo > set_graph_function
3263 ftrace_enabled
3264 --------------
3266 Note, the proc sysctl ftrace_enable is a big on/off switch for the
3267 function tracer. By default it is enabled (when function tracing is
3268 enabled in the kernel). If it is disabled, all function tracing is
3269 disabled. This includes not only the function tracers for ftrace, but
3270 also for any other uses (perf, kprobes, stack tracing, profiling, etc). It
3271 cannot be disabled if there is a callback with FTRACE_OPS_FL_PERMANENT set
3272 registered.
3274 Please disable this with care.
3276 This can be disable (and enabled) with::
3278 sysctl kernel.ftrace_enabled=0
3279 sysctl kernel.ftrace_enabled=1
3281 or
3283 echo 0 > /proc/sys/kernel/ftrace_enabled
3284 echo 1 > /proc/sys/kernel/ftrace_enabled
3287 Filter commands
3288 ---------------
3290 A few commands are supported by the set_ftrace_filter interface.
3291 Trace commands have the following format::
3293 <function>:<command>:<parameter>
3295 The following commands are supported:
3297 - mod:
3298 This command enables function filtering per module. The
3299 parameter defines the module. For example, if only the write*
3300 functions in the ext3 module are desired, run:
3302 echo 'write*:mod:ext3' > set_ftrace_filter
3304 This command interacts with the filter in the same way as
3305 filtering based on function names. Thus, adding more functions
3306 in a different module is accomplished by appending (>>) to the
3307 filter file. Remove specific module functions by prepending
3308 '!'::
3310 echo '!writeback*:mod:ext3' >> set_ftrace_filter
3312 Mod command supports module globbing. Disable tracing for all
3313 functions except a specific module::
3315 echo '!*:mod:!ext3' >> set_ftrace_filter
3317 Disable tracing for all modules, but still trace kernel::
3319 echo '!*:mod:*' >> set_ftrace_filter
3321 Enable filter only for kernel::
3323 echo '*write*:mod:!*' >> set_ftrace_filter
3325 Enable filter for module globbing::
3327 echo '*write*:mod:*snd*' >> set_ftrace_filter
3329 - traceon/traceoff:
3330 These commands turn tracing on and off when the specified
3331 functions are hit. The parameter determines how many times the
3332 tracing system is turned on and off. If unspecified, there is
3333 no limit. For example, to disable tracing when a schedule bug
3334 is hit the first 5 times, run::
3336 echo '__schedule_bug:traceoff:5' > set_ftrace_filter
3338 To always disable tracing when __schedule_bug is hit::
3340 echo '__schedule_bug:traceoff' > set_ftrace_filter
3342 These commands are cumulative whether or not they are appended
3343 to set_ftrace_filter. To remove a command, prepend it by '!'
3344 and drop the parameter::
3346 echo '!__schedule_bug:traceoff:0' > set_ftrace_filter
3348 The above removes the traceoff command for __schedule_bug
3349 that have a counter. To remove commands without counters::
3351 echo '!__schedule_bug:traceoff' > set_ftrace_filter
3353 - snapshot:
3354 Will cause a snapshot to be triggered when the function is hit.
3355 ::
3357 echo 'native_flush_tlb_others:snapshot' > set_ftrace_filter
3359 To only snapshot once:
3360 ::
3362 echo 'native_flush_tlb_others:snapshot:1' > set_ftrace_filter
3364 To remove the above commands::
3366 echo '!native_flush_tlb_others:snapshot' > set_ftrace_filter
3367 echo '!native_flush_tlb_others:snapshot:0' > set_ftrace_filter
3369 - enable_event/disable_event:
3370 These commands can enable or disable a trace event. Note, because
3371 function tracing callbacks are very sensitive, when these commands
3372 are registered, the trace point is activated, but disabled in
3373 a "soft" mode. That is, the tracepoint will be called, but
3374 just will not be traced. The event tracepoint stays in this mode
3375 as long as there's a command that triggers it.
3376 ::
3378 echo 'try_to_wake_up:enable_event:sched:sched_switch:2' > \
3379 set_ftrace_filter
3381 The format is::
3383 <function>:enable_event:<system>:<event>[:count]
3384 <function>:disable_event:<system>:<event>[:count]
3386 To remove the events commands::
3388 echo '!try_to_wake_up:enable_event:sched:sched_switch:0' > \
3389 set_ftrace_filter
3390 echo '!schedule:disable_event:sched:sched_switch' > \
3391 set_ftrace_filter
3393 - dump:
3394 When the function is hit, it will dump the contents of the ftrace
3395 ring buffer to the console. This is useful if you need to debug
3396 something, and want to dump the trace when a certain function
3397 is hit. Perhaps it's a function that is called before a triple
3398 fault happens and does not allow you to get a regular dump.
3400 - cpudump:
3401 When the function is hit, it will dump the contents of the ftrace
3402 ring buffer for the current CPU to the console. Unlike the "dump"
3403 command, it only prints out the contents of the ring buffer for the
3404 CPU that executed the function that triggered the dump.
3406 - stacktrace:
3407 When the function is hit, a stack trace is recorded.
3409 trace_pipe
3410 ----------
3412 The trace_pipe outputs the same content as the trace file, but
3413 the effect on the tracing is different. Every read from
3414 trace_pipe is consumed. This means that subsequent reads will be
3415 different. The trace is live.
3416 ::
3418 # echo function > current_tracer
3419 # cat trace_pipe > /tmp/trace.out &
3420 [1] 4153
3421 # echo 1 > tracing_on
3422 # usleep 1
3423 # echo 0 > tracing_on
3424 # cat trace
3425 # tracer: function
3426 #
3427 # entries-in-buffer/entries-written: 0/0 #P:4
3428 #
3429 # _-----=> irqs-off
3430 # / _----=> need-resched
3431 # | / _---=> hardirq/softirq
3432 # || / _--=> preempt-depth
3433 # ||| / delay
3434 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3435 # | | | |||| | |
3437 #
3438 # cat /tmp/trace.out
3439 bash-1994 [000] .... 5281.568961: mutex_unlock <-rb_simple_write
3440 bash-1994 [000] .... 5281.568963: __mutex_unlock_slowpath <-mutex_unlock
3441 bash-1994 [000] .... 5281.568963: __fsnotify_parent <-fsnotify_modify
3442 bash-1994 [000] .... 5281.568964: fsnotify <-fsnotify_modify
3443 bash-1994 [000] .... 5281.568964: __srcu_read_lock <-fsnotify
3444 bash-1994 [000] .... 5281.568964: add_preempt_count <-__srcu_read_lock
3445 bash-1994 [000] ...1 5281.568965: sub_preempt_count <-__srcu_read_lock
3446 bash-1994 [000] .... 5281.568965: __srcu_read_unlock <-fsnotify
3447 bash-1994 [000] .... 5281.568967: sys_dup2 <-system_call_fastpath
3450 Note, reading the trace_pipe file will block until more input is
3451 added. This is contrary to the trace file. If any process opened
3452 the trace file for reading, it will actually disable tracing and
3453 prevent new entries from being added. The trace_pipe file does
3454 not have this limitation.
3456 trace entries
3457 -------------
3459 Having too much or not enough data can be troublesome in
3460 diagnosing an issue in the kernel. The file buffer_size_kb is
3461 used to modify the size of the internal trace buffers. The
3462 number listed is the number of entries that can be recorded per
3463 CPU. To know the full size, multiply the number of possible CPUs
3464 with the number of entries.
3465 ::
3467 # cat buffer_size_kb
3468 1408 (units kilobytes)
3470 Or simply read buffer_total_size_kb
3471 ::
3473 # cat buffer_total_size_kb
3474 5632
3476 To modify the buffer, simple echo in a number (in 1024 byte segments).
3477 ::
3479 # echo 10000 > buffer_size_kb
3480 # cat buffer_size_kb
3481 10000 (units kilobytes)
3483 It will try to allocate as much as possible. If you allocate too
3484 much, it can cause Out-Of-Memory to trigger.
3485 ::
3487 # echo 1000000000000 > buffer_size_kb
3488 -bash: echo: write error: Cannot allocate memory
3489 # cat buffer_size_kb
3490 85
3492 The per_cpu buffers can be changed individually as well:
3493 ::
3495 # echo 10000 > per_cpu/cpu0/buffer_size_kb
3496 # echo 100 > per_cpu/cpu1/buffer_size_kb
3498 When the per_cpu buffers are not the same, the buffer_size_kb
3499 at the top level will just show an X
3500 ::
3502 # cat buffer_size_kb
3503 X
3505 This is where the buffer_total_size_kb is useful:
3506 ::
3508 # cat buffer_total_size_kb
3509 12916
3511 Writing to the top level buffer_size_kb will reset all the buffers
3512 to be the same again.
3514 Snapshot
3515 --------
3516 CONFIG_TRACER_SNAPSHOT makes a generic snapshot feature
3517 available to all non latency tracers. (Latency tracers which
3518 record max latency, such as "irqsoff" or "wakeup", can't use
3519 this feature, since those are already using the snapshot
3520 mechanism internally.)
3522 Snapshot preserves a current trace buffer at a particular point
3523 in time without stopping tracing. Ftrace swaps the current
3524 buffer with a spare buffer, and tracing continues in the new
3525 current (=previous spare) buffer.
3527 The following tracefs files in "tracing" are related to this
3528 feature:
3530 snapshot:
3532 This is used to take a snapshot and to read the output
3533 of the snapshot. Echo 1 into this file to allocate a
3534 spare buffer and to take a snapshot (swap), then read
3535 the snapshot from this file in the same format as
3536 "trace" (described above in the section "The File
3537 System"). Both reads snapshot and tracing are executable
3538 in parallel. When the spare buffer is allocated, echoing
3539 0 frees it, and echoing else (positive) values clear the
3540 snapshot contents.
3541 More details are shown in the table below.
3543 +--------------+------------+------------+------------+
3544 |status\\input | 0 | 1 | else |
3545 +==============+============+============+============+
3546 |not allocated |(do nothing)| alloc+swap |(do nothing)|
3547 +--------------+------------+------------+------------+
3548 |allocated | free | swap | clear |
3549 +--------------+------------+------------+------------+
3551 Here is an example of using the snapshot feature.
3552 ::
3554 # echo 1 > events/sched/enable
3555 # echo 1 > snapshot
3556 # cat snapshot
3557 # tracer: nop
3558 #
3559 # entries-in-buffer/entries-written: 71/71 #P:8
3560 #
3561 # _-----=> irqs-off
3562 # / _----=> need-resched
3563 # | / _---=> hardirq/softirq
3564 # || / _--=> preempt-depth
3565 # ||| / delay
3566 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3567 # | | | |||| | |
3568 <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
3569 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
3570 [...]
3571 <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
3573 # cat trace
3574 # tracer: nop
3575 #
3576 # entries-in-buffer/entries-written: 77/77 #P:8
3577 #
3578 # _-----=> irqs-off
3579 # / _----=> need-resched
3580 # | / _---=> hardirq/softirq
3581 # || / _--=> preempt-depth
3582 # ||| / delay
3583 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3584 # | | | |||| | |
3585 <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
3586 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
3587 [...]
3590 If you try to use this snapshot feature when current tracer is
3591 one of the latency tracers, you will get the following results.
3592 ::
3594 # echo wakeup > current_tracer
3595 # echo 1 > snapshot
3596 bash: echo: write error: Device or resource busy
3597 # cat snapshot
3598 cat: snapshot: Device or resource busy
3601 Instances
3602 ---------
3603 In the tracefs tracing directory, there is a directory called "instances".
3604 This directory can have new directories created inside of it using
3605 mkdir, and removing directories with rmdir. The directory created
3606 with mkdir in this directory will already contain files and other
3607 directories after it is created.
3608 ::
3610 # mkdir instances/foo
3611 # ls instances/foo
3612 buffer_size_kb buffer_total_size_kb events free_buffer per_cpu
3613 set_event snapshot trace trace_clock trace_marker trace_options
3614 trace_pipe tracing_on
3616 As you can see, the new directory looks similar to the tracing directory
3617 itself. In fact, it is very similar, except that the buffer and
3618 events are agnostic from the main directory, or from any other
3619 instances that are created.
3621 The files in the new directory work just like the files with the
3622 same name in the tracing directory except the buffer that is used
3623 is a separate and new buffer. The files affect that buffer but do not
3624 affect the main buffer with the exception of trace_options. Currently,
3625 the trace_options affect all instances and the top level buffer
3626 the same, but this may change in future releases. That is, options
3627 may become specific to the instance they reside in.
3629 Notice that none of the function tracer files are there, nor is
3630 current_tracer and available_tracers. This is because the buffers
3631 can currently only have events enabled for them.
3632 ::
3634 # mkdir instances/foo
3635 # mkdir instances/bar
3636 # mkdir instances/zoot
3637 # echo 100000 > buffer_size_kb
3638 # echo 1000 > instances/foo/buffer_size_kb
3639 # echo 5000 > instances/bar/per_cpu/cpu1/buffer_size_kb
3640 # echo function > current_trace
3641 # echo 1 > instances/foo/events/sched/sched_wakeup/enable
3642 # echo 1 > instances/foo/events/sched/sched_wakeup_new/enable
3643 # echo 1 > instances/foo/events/sched/sched_switch/enable
3644 # echo 1 > instances/bar/events/irq/enable
3645 # echo 1 > instances/zoot/events/syscalls/enable
3646 # cat trace_pipe
3647 CPU:2 [LOST 11745 EVENTS]
3648 bash-2044 [002] .... 10594.481032: _raw_spin_lock_irqsave <-get_page_from_freelist
3649 bash-2044 [002] d... 10594.481032: add_preempt_count <-_raw_spin_lock_irqsave
3650 bash-2044 [002] d..1 10594.481032: __rmqueue <-get_page_from_freelist
3651 bash-2044 [002] d..1 10594.481033: _raw_spin_unlock <-get_page_from_freelist
3652 bash-2044 [002] d..1 10594.481033: sub_preempt_count <-_raw_spin_unlock
3653 bash-2044 [002] d... 10594.481033: get_pageblock_flags_group <-get_pageblock_migratetype
3654 bash-2044 [002] d... 10594.481034: __mod_zone_page_state <-get_page_from_freelist
3655 bash-2044 [002] d... 10594.481034: zone_statistics <-get_page_from_freelist
3656 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics
3657 bash-2044 [002] d... 10594.481034: __inc_zone_state <-zone_statistics
3658 bash-2044 [002] .... 10594.481035: arch_dup_task_struct <-copy_process
3659 [...]
3661 # cat instances/foo/trace_pipe
3662 bash-1998 [000] d..4 136.676759: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000
3663 bash-1998 [000] dN.4 136.676760: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000
3664 <idle>-0 [003] d.h3 136.676906: sched_wakeup: comm=rcu_preempt pid=9 prio=120 success=1 target_cpu=003
3665 <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
3666 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
3667 bash-1998 [000] d..4 136.677014: sched_wakeup: comm=kworker/0:1 pid=59 prio=120 success=1 target_cpu=000
3668 bash-1998 [000] dN.4 136.677016: sched_wakeup: comm=bash pid=1998 prio=120 success=1 target_cpu=000
3669 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
3670 kworker/0:1-59 [000] d..4 136.677022: sched_wakeup: comm=sshd pid=1995 prio=120 success=1 target_cpu=001
3671 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
3672 [...]
3674 # cat instances/bar/trace_pipe
3675 migration/1-14 [001] d.h3 138.732674: softirq_raise: vec=3 [action=NET_RX]
3676 <idle>-0 [001] dNh3 138.732725: softirq_raise: vec=3 [action=NET_RX]
3677 bash-1998 [000] d.h1 138.733101: softirq_raise: vec=1 [action=TIMER]
3678 bash-1998 [000] d.h1 138.733102: softirq_raise: vec=9 [action=RCU]
3679 bash-1998 [000] ..s2 138.733105: softirq_entry: vec=1 [action=TIMER]
3680 bash-1998 [000] ..s2 138.733106: softirq_exit: vec=1 [action=TIMER]
3681 bash-1998 [000] ..s2 138.733106: softirq_entry: vec=9 [action=RCU]
3682 bash-1998 [000] ..s2 138.733109: softirq_exit: vec=9 [action=RCU]
3683 sshd-1995 [001] d.h1 138.733278: irq_handler_entry: irq=21 name=uhci_hcd:usb4
3684 sshd-1995 [001] d.h1 138.733280: irq_handler_exit: irq=21 ret=unhandled
3685 sshd-1995 [001] d.h1 138.733281: irq_handler_entry: irq=21 name=eth0
3686 sshd-1995 [001] d.h1 138.733283: irq_handler_exit: irq=21 ret=handled
3687 [...]
3689 # cat instances/zoot/trace
3690 # tracer: nop
3691 #
3692 # entries-in-buffer/entries-written: 18996/18996 #P:4
3693 #
3694 # _-----=> irqs-off
3695 # / _----=> need-resched
3696 # | / _---=> hardirq/softirq
3697 # || / _--=> preempt-depth
3698 # ||| / delay
3699 # TASK-PID CPU# |||| TIMESTAMP FUNCTION
3700 # | | | |||| | |
3701 bash-1998 [000] d... 140.733501: sys_write -> 0x2
3702 bash-1998 [000] d... 140.733504: sys_dup2(oldfd: a, newfd: 1)
3703 bash-1998 [000] d... 140.733506: sys_dup2 -> 0x1
3704 bash-1998 [000] d... 140.733508: sys_fcntl(fd: a, cmd: 1, arg: 0)
3705 bash-1998 [000] d... 140.733509: sys_fcntl -> 0x1
3706 bash-1998 [000] d... 140.733510: sys_close(fd: a)
3707 bash-1998 [000] d... 140.733510: sys_close -> 0x0
3708 bash-1998 [000] d... 140.733514: sys_rt_sigprocmask(how: 0, nset: 0, oset: 6e2768, sigsetsize: 8)
3709 bash-1998 [000] d... 140.733515: sys_rt_sigprocmask -> 0x0
3710 bash-1998 [000] d... 140.733516: sys_rt_sigaction(sig: 2, act: 7fff718846f0, oact: 7fff71884650, sigsetsize: 8)
3711 bash-1998 [000] d... 140.733516: sys_rt_sigaction -> 0x0
3713 You can see that the trace of the top most trace buffer shows only
3714 the function tracing. The foo instance displays wakeups and task
3715 switches.
3717 To remove the instances, simply delete their directories:
3718 ::
3720 # rmdir instances/foo
3721 # rmdir instances/bar
3722 # rmdir instances/zoot
3724 Note, if a process has a trace file open in one of the instance
3725 directories, the rmdir will fail with EBUSY.
3728 Stack trace
3729 -----------
3730 Since the kernel has a fixed sized stack, it is important not to
3731 waste it in functions. A kernel developer must be conscious of
3732 what they allocate on the stack. If they add too much, the system
3733 can be in danger of a stack overflow, and corruption will occur,
3734 usually leading to a system panic.
3736 There are some tools that check this, usually with interrupts
3737 periodically checking usage. But if you can perform a check
3738 at every function call that will become very useful. As ftrace provides
3739 a function tracer, it makes it convenient to check the stack size
3740 at every function call. This is enabled via the stack tracer.
3742 CONFIG_STACK_TRACER enables the ftrace stack tracing functionality.
3743 To enable it, write a '1' into /proc/sys/kernel/stack_tracer_enabled.
3744 ::
3746 # echo 1 > /proc/sys/kernel/stack_tracer_enabled
3748 You can also enable it from the kernel command line to trace
3749 the stack size of the kernel during boot up, by adding "stacktrace"
3750 to the kernel command line parameter.
3752 After running it for a few minutes, the output looks like:
3753 ::
3755 # cat stack_max_size
3756 2928
3758 # cat stack_trace
3759 Depth Size Location (18 entries)
3760 ----- ---- --------
3761 0) 2928 224 update_sd_lb_stats+0xbc/0x4ac
3762 1) 2704 160 find_busiest_group+0x31/0x1f1
3763 2) 2544 256 load_balance+0xd9/0x662
3764 3) 2288 80 idle_balance+0xbb/0x130
3765 4) 2208 128 __schedule+0x26e/0x5b9
3766 5) 2080 16 schedule+0x64/0x66
3767 6) 2064 128 schedule_timeout+0x34/0xe0
3768 7) 1936 112 wait_for_common+0x97/0xf1
3769 8) 1824 16 wait_for_completion+0x1d/0x1f
3770 9) 1808 128 flush_work+0xfe/0x119
3771 10) 1680 16 tty_flush_to_ldisc+0x1e/0x20
3772 11) 1664 48 input_available_p+0x1d/0x5c
3773 12) 1616 48 n_tty_poll+0x6d/0x134
3774 13) 1568 64 tty_poll+0x64/0x7f
3775 14) 1504 880 do_select+0x31e/0x511
3776 15) 624 400 core_sys_select+0x177/0x216
3777 16) 224 96 sys_select+0x91/0xb9
3778 17) 128 128 system_call_fastpath+0x16/0x1b
3780 Note, if -mfentry is being used by gcc, functions get traced before
3781 they set up the stack frame. This means that leaf level functions
3782 are not tested by the stack tracer when -mfentry is used.
3784 Currently, -mfentry is used by gcc 4.6.0 and above on x86 only.
3786 More
3787 ----
3788 More details can be found in the source code, in the `kernel/trace/*.c` files.