1 # SPDX-License-Identifier: GPL-2.0-only
3 # Architectures that offer an FUNCTION_TRACER implementation should
4 # select HAVE_FUNCTION_TRACER:
7 config USER_STACKTRACE_SUPPORT
18 depends on HAVE_RETHOOK
20 Enable generic return hooking feature. This is an internal
21 API, which will be used by other function-entry hooking
22 features like fprobe and kprobes.
24 config HAVE_FUNCTION_TRACER
27 See Documentation/trace/ftrace-design.rst
29 config HAVE_FUNCTION_GRAPH_TRACER
32 See Documentation/trace/ftrace-design.rst
34 config HAVE_DYNAMIC_FTRACE
37 See Documentation/trace/ftrace-design.rst
39 config HAVE_DYNAMIC_FTRACE_WITH_REGS
42 config HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
45 config HAVE_DYNAMIC_FTRACE_WITH_ARGS
48 If this is set, then arguments and stack can be found from
49 the pt_regs passed into the function callback regs parameter
50 by default, even without setting the REGS flag in the ftrace_ops.
51 This allows for use of regs_get_kernel_argument() and
52 kernel_stack_pointer().
54 config HAVE_FTRACE_MCOUNT_RECORD
57 See Documentation/trace/ftrace-design.rst
59 config HAVE_SYSCALL_TRACEPOINTS
62 See Documentation/trace/ftrace-design.rst
67 Arch supports the gcc options -pg with -mfentry
69 config HAVE_NOP_MCOUNT
72 Arch supports the gcc options -pg with -mrecord-mcount and -nop-mcount
74 config HAVE_OBJTOOL_MCOUNT
77 Arch supports objtool --mcount
79 config HAVE_C_RECORDMCOUNT
82 C version of recordmcount available?
84 config HAVE_BUILDTIME_MCOUNT_SORT
87 An architecture selects this if it sorts the mcount_loc section
90 config BUILDTIME_MCOUNT_SORT
93 depends on HAVE_BUILDTIME_MCOUNT_SORT && DYNAMIC_FTRACE
95 Sort the mcount_loc section at build time.
97 config TRACER_MAX_TRACE
109 select CONTEXT_SWITCH_TRACER
113 config CONTEXT_SWITCH_TRACER
116 config RING_BUFFER_ALLOW_SWAP
119 Allow the use of ring_buffer_swap_cpu.
120 Adds a very slight overhead to tracing when enabled.
122 config PREEMPTIRQ_TRACEPOINTS
124 depends on TRACE_PREEMPT_TOGGLE || TRACE_IRQFLAGS
128 Create preempt/irq toggle tracepoints if needed, so that other parts
129 of the kernel can use them to generate or add hooks to them.
131 # All tracer options should select GENERIC_TRACER. For those options that are
132 # enabled by all tracers (context switch and event tracer) they select TRACING.
133 # This allows those options to appear when no other tracer is selected. But the
134 # options do not appear when something else selects it. We need the two options
135 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
136 # hiding of the automatic options.
141 select STACKTRACE if STACKTRACE_SUPPORT
147 select TASKS_RCU if PREEMPTION
149 config GENERIC_TRACER
154 # Minimum requirements an architecture has to meet for us to
155 # be able to offer generic tracing facilities:
157 config TRACING_SUPPORT
159 depends on TRACE_IRQFLAGS_SUPPORT
160 depends on STACKTRACE_SUPPORT
165 depends on TRACING_SUPPORT
166 default y if DEBUG_KERNEL
168 Enable the kernel tracing infrastructure.
172 config BOOTTIME_TRACING
173 bool "Boot-time Tracing support"
177 Enable developer to setup ftrace subsystem via supplemental
178 kernel cmdline at boot time for debugging (tracing) driver
179 initialization and boot process.
181 config FUNCTION_TRACER
182 bool "Kernel Function Tracer"
183 depends on HAVE_FUNCTION_TRACER
185 select GENERIC_TRACER
186 select CONTEXT_SWITCH_TRACER
188 select TASKS_RCU if PREEMPTION
189 select TASKS_RUDE_RCU
191 Enable the kernel to trace every kernel function. This is done
192 by using a compiler feature to insert a small, 5-byte No-Operation
193 instruction at the beginning of every kernel function, which NOP
194 sequence is then dynamically patched into a tracer call when
195 tracing is enabled by the administrator. If it's runtime disabled
196 (the bootup default), then the overhead of the instructions is very
197 small and not measurable even in micro-benchmarks.
199 config FUNCTION_GRAPH_TRACER
200 bool "Kernel Function Graph Tracer"
201 depends on HAVE_FUNCTION_GRAPH_TRACER
202 depends on FUNCTION_TRACER
203 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
206 Enable the kernel to trace a function at both its return
208 Its first purpose is to trace the duration of functions and
209 draw a call graph for each thread with some information like
210 the return value. This is done by setting the current return
211 address on the current task structure into a stack of calls.
213 config DYNAMIC_FTRACE
214 bool "enable/disable function tracing dynamically"
215 depends on FUNCTION_TRACER
216 depends on HAVE_DYNAMIC_FTRACE
219 This option will modify all the calls to function tracing
220 dynamically (will patch them out of the binary image and
221 replace them with a No-Op instruction) on boot up. During
222 compile time, a table is made of all the locations that ftrace
223 can function trace, and this table is linked into the kernel
224 image. When this is enabled, functions can be individually
225 enabled, and the functions not enabled will not affect
226 performance of the system.
228 See the files in /sys/kernel/debug/tracing:
229 available_filter_functions
233 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
234 otherwise has native performance as long as no tracing is active.
236 config DYNAMIC_FTRACE_WITH_REGS
238 depends on DYNAMIC_FTRACE
239 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS
241 config DYNAMIC_FTRACE_WITH_DIRECT_CALLS
243 depends on DYNAMIC_FTRACE_WITH_REGS
244 depends on HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
246 config DYNAMIC_FTRACE_WITH_ARGS
248 depends on DYNAMIC_FTRACE
249 depends on HAVE_DYNAMIC_FTRACE_WITH_ARGS
252 bool "Kernel Function Probe (fprobe)"
253 depends on FUNCTION_TRACER
254 depends on DYNAMIC_FTRACE_WITH_REGS
255 depends on HAVE_RETHOOK
259 This option enables kernel function probe (fprobe) based on ftrace.
260 The fprobe is similar to kprobes, but probes only for kernel function
261 entries and exits. This also can probe multiple functions by one
266 config FUNCTION_PROFILER
267 bool "Kernel function profiler"
268 depends on FUNCTION_TRACER
271 This option enables the kernel function profiler. A file is created
272 in debugfs called function_profile_enabled which defaults to zero.
273 When a 1 is echoed into this file profiling begins, and when a
274 zero is entered, profiling stops. A "functions" file is created in
275 the trace_stat directory; this file shows the list of functions that
276 have been hit and their counters.
281 bool "Trace max stack"
282 depends on HAVE_FUNCTION_TRACER
283 select FUNCTION_TRACER
287 This special tracer records the maximum stack footprint of the
288 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
290 This tracer works by hooking into every function call that the
291 kernel executes, and keeping a maximum stack depth value and
292 stack-trace saved. If this is configured with DYNAMIC_FTRACE
293 then it will not have any overhead while the stack tracer
296 To enable the stack tracer on bootup, pass in 'stacktrace'
297 on the kernel command line.
299 The stack tracer can also be enabled or disabled via the
300 sysctl kernel.stack_tracer_enabled
304 config TRACE_PREEMPT_TOGGLE
307 Enables hooks which will be called when preemption is first disabled,
310 config IRQSOFF_TRACER
311 bool "Interrupts-off Latency Tracer"
313 depends on TRACE_IRQFLAGS_SUPPORT
314 select TRACE_IRQFLAGS
315 select GENERIC_TRACER
316 select TRACER_MAX_TRACE
317 select RING_BUFFER_ALLOW_SWAP
318 select TRACER_SNAPSHOT
319 select TRACER_SNAPSHOT_PER_CPU_SWAP
321 This option measures the time spent in irqs-off critical
322 sections, with microsecond accuracy.
324 The default measurement method is a maximum search, which is
325 disabled by default and can be runtime (re-)started
328 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
330 (Note that kernel size and overhead increase with this option
331 enabled. This option and the preempt-off timing option can be
332 used together or separately.)
334 config PREEMPT_TRACER
335 bool "Preemption-off Latency Tracer"
337 depends on PREEMPTION
338 select GENERIC_TRACER
339 select TRACER_MAX_TRACE
340 select RING_BUFFER_ALLOW_SWAP
341 select TRACER_SNAPSHOT
342 select TRACER_SNAPSHOT_PER_CPU_SWAP
343 select TRACE_PREEMPT_TOGGLE
345 This option measures the time spent in preemption-off critical
346 sections, with microsecond accuracy.
348 The default measurement method is a maximum search, which is
349 disabled by default and can be runtime (re-)started
352 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
354 (Note that kernel size and overhead increase with this option
355 enabled. This option and the irqs-off timing option can be
356 used together or separately.)
359 bool "Scheduling Latency Tracer"
360 select GENERIC_TRACER
361 select CONTEXT_SWITCH_TRACER
362 select TRACER_MAX_TRACE
363 select TRACER_SNAPSHOT
365 This tracer tracks the latency of the highest priority task
366 to be scheduled in, starting from the point it has woken up.
369 bool "Tracer to detect hardware latencies (like SMIs)"
370 select GENERIC_TRACER
372 This tracer, when enabled will create one or more kernel threads,
373 depending on what the cpumask file is set to, which each thread
374 spinning in a loop looking for interruptions caused by
375 something other than the kernel. For example, if a
376 System Management Interrupt (SMI) takes a noticeable amount of
377 time, this tracer will detect it. This is useful for testing
378 if a system is reliable for Real Time tasks.
380 Some files are created in the tracing directory when this
383 hwlat_detector/width - time in usecs for how long to spin for
384 hwlat_detector/window - time in usecs between the start of each
387 A kernel thread is created that will spin with interrupts disabled
388 for "width" microseconds in every "window" cycle. It will not spin
389 for "window - width" microseconds, where the system can
392 The output will appear in the trace and trace_pipe files.
394 When the tracer is not running, it has no affect on the system,
395 but when it is running, it can cause the system to be
396 periodically non responsive. Do not run this tracer on a
399 To enable this tracer, echo in "hwlat" into the current_tracer
400 file. Every time a latency is greater than tracing_thresh, it will
401 be recorded into the ring buffer.
403 config OSNOISE_TRACER
404 bool "OS Noise tracer"
405 select GENERIC_TRACER
407 In the context of high-performance computing (HPC), the Operating
408 System Noise (osnoise) refers to the interference experienced by an
409 application due to activities inside the operating system. In the
410 context of Linux, NMIs, IRQs, SoftIRQs, and any other system thread
411 can cause noise to the system. Moreover, hardware-related jobs can
412 also cause noise, for example, via SMIs.
414 The osnoise tracer leverages the hwlat_detector by running a similar
415 loop with preemption, SoftIRQs and IRQs enabled, thus allowing all
416 the sources of osnoise during its execution. The osnoise tracer takes
417 note of the entry and exit point of any source of interferences,
418 increasing a per-cpu interference counter. It saves an interference
419 counter for each source of interference. The interference counter for
420 NMI, IRQs, SoftIRQs, and threads is increased anytime the tool
421 observes these interferences' entry events. When a noise happens
422 without any interference from the operating system level, the
423 hardware noise counter increases, pointing to a hardware-related
424 noise. In this way, osnoise can account for any source of
425 interference. At the end of the period, the osnoise tracer prints
426 the sum of all noise, the max single noise, the percentage of CPU
427 available for the thread, and the counters for the noise sources.
429 In addition to the tracer, a set of tracepoints were added to
430 facilitate the identification of the osnoise source.
432 The output will appear in the trace and trace_pipe files.
434 To enable this tracer, echo in "osnoise" into the current_tracer
437 config TIMERLAT_TRACER
438 bool "Timerlat tracer"
439 select OSNOISE_TRACER
440 select GENERIC_TRACER
442 The timerlat tracer aims to help the preemptive kernel developers
443 to find sources of wakeup latencies of real-time threads.
445 The tracer creates a per-cpu kernel thread with real-time priority.
446 The tracer thread sets a periodic timer to wakeup itself, and goes
447 to sleep waiting for the timer to fire. At the wakeup, the thread
448 then computes a wakeup latency value as the difference between
449 the current time and the absolute time that the timer was set
452 The tracer prints two lines at every activation. The first is the
453 timer latency observed at the hardirq context before the
454 activation of the thread. The second is the timer latency observed
455 by the thread, which is the same level that cyclictest reports. The
456 ACTIVATION ID field serves to relate the irq execution to its
457 respective thread execution.
459 The tracer is build on top of osnoise tracer, and the osnoise:
460 events can be used to trace the source of interference from NMI,
461 IRQs and other threads. It also enables the capture of the
462 stacktrace at the IRQ context, which helps to identify the code
463 path that can cause thread delay.
466 bool "Memory mapped IO tracing"
467 depends on HAVE_MMIOTRACE_SUPPORT && PCI
468 select GENERIC_TRACER
470 Mmiotrace traces Memory Mapped I/O access and is meant for
471 debugging and reverse engineering. It is called from the ioremap
472 implementation and works via page faults. Tracing is disabled by
473 default and can be enabled at run-time.
475 See Documentation/trace/mmiotrace.rst.
476 If you are not helping to develop drivers, say N.
478 config ENABLE_DEFAULT_TRACERS
479 bool "Trace process context switches and events"
480 depends on !GENERIC_TRACER
483 This tracer hooks to various trace points in the kernel,
484 allowing the user to pick and choose which trace point they
485 want to trace. It also includes the sched_switch tracer plugin.
487 config FTRACE_SYSCALLS
488 bool "Trace syscalls"
489 depends on HAVE_SYSCALL_TRACEPOINTS
490 select GENERIC_TRACER
493 Basic tracer to catch the syscall entry and exit events.
495 config TRACER_SNAPSHOT
496 bool "Create a snapshot trace buffer"
497 select TRACER_MAX_TRACE
499 Allow tracing users to take snapshot of the current buffer using the
500 ftrace interface, e.g.:
502 echo 1 > /sys/kernel/debug/tracing/snapshot
505 config TRACER_SNAPSHOT_PER_CPU_SWAP
506 bool "Allow snapshot to swap per CPU"
507 depends on TRACER_SNAPSHOT
508 select RING_BUFFER_ALLOW_SWAP
510 Allow doing a snapshot of a single CPU buffer instead of a
511 full swap (all buffers). If this is set, then the following is
514 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot
516 After which, only the tracing buffer for CPU 2 was swapped with
517 the main tracing buffer, and the other CPU buffers remain the same.
519 When this is enabled, this adds a little more overhead to the
520 trace recording, as it needs to add some checks to synchronize
521 recording with swaps. But this does not affect the performance
522 of the overall system. This is enabled by default when the preempt
523 or irq latency tracers are enabled, as those need to swap as well
524 and already adds the overhead (plus a lot more).
526 config TRACE_BRANCH_PROFILING
528 select GENERIC_TRACER
531 prompt "Branch Profiling"
532 default BRANCH_PROFILE_NONE
534 The branch profiling is a software profiler. It will add hooks
535 into the C conditionals to test which path a branch takes.
537 The likely/unlikely profiler only looks at the conditions that
538 are annotated with a likely or unlikely macro.
540 The "all branch" profiler will profile every if-statement in the
541 kernel. This profiler will also enable the likely/unlikely
544 Either of the above profilers adds a bit of overhead to the system.
545 If unsure, choose "No branch profiling".
547 config BRANCH_PROFILE_NONE
548 bool "No branch profiling"
550 No branch profiling. Branch profiling adds a bit of overhead.
551 Only enable it if you want to analyse the branching behavior.
552 Otherwise keep it disabled.
554 config PROFILE_ANNOTATED_BRANCHES
555 bool "Trace likely/unlikely profiler"
556 select TRACE_BRANCH_PROFILING
558 This tracer profiles all likely and unlikely macros
559 in the kernel. It will display the results in:
561 /sys/kernel/debug/tracing/trace_stat/branch_annotated
563 Note: this will add a significant overhead; only turn this
564 on if you need to profile the system's use of these macros.
566 config PROFILE_ALL_BRANCHES
567 bool "Profile all if conditionals" if !FORTIFY_SOURCE
568 select TRACE_BRANCH_PROFILING
570 This tracer profiles all branch conditions. Every if ()
571 taken in the kernel is recorded whether it hit or miss.
572 The results will be displayed in:
574 /sys/kernel/debug/tracing/trace_stat/branch_all
576 This option also enables the likely/unlikely profiler.
578 This configuration, when enabled, will impose a great overhead
579 on the system. This should only be enabled when the system
580 is to be analyzed in much detail.
583 config TRACING_BRANCHES
586 Selected by tracers that will trace the likely and unlikely
587 conditions. This prevents the tracers themselves from being
588 profiled. Profiling the tracing infrastructure can only happen
589 when the likelys and unlikelys are not being traced.
592 bool "Trace likely/unlikely instances"
593 depends on TRACE_BRANCH_PROFILING
594 select TRACING_BRANCHES
596 This traces the events of likely and unlikely condition
597 calls in the kernel. The difference between this and the
598 "Trace likely/unlikely profiler" is that this is not a
599 histogram of the callers, but actually places the calling
600 events into a running trace buffer to see when and where the
601 events happened, as well as their results.
605 config BLK_DEV_IO_TRACE
606 bool "Support for tracing block IO actions"
612 select GENERIC_TRACER
615 Say Y here if you want to be able to trace the block layer actions
616 on a given queue. Tracing allows you to see any traffic happening
617 on a block device queue. For more information (and the userspace
618 support tools needed), fetch the blktrace tools from:
620 git://git.kernel.dk/blktrace.git
622 Tracing also is possible using the ftrace interface, e.g.:
624 echo 1 > /sys/block/sda/sda1/trace/enable
625 echo blk > /sys/kernel/debug/tracing/current_tracer
626 cat /sys/kernel/debug/tracing/trace_pipe
632 depends on HAVE_REGS_AND_STACK_ACCESS_API
633 bool "Enable kprobes-based dynamic events"
636 select DYNAMIC_EVENTS
639 This allows the user to add tracing events (similar to tracepoints)
640 on the fly via the ftrace interface. See
641 Documentation/trace/kprobetrace.rst for more details.
643 Those events can be inserted wherever kprobes can probe, and record
644 various register and memory values.
646 This option is also required by perf-probe subcommand of perf tools.
647 If you want to use perf tools, this option is strongly recommended.
649 config KPROBE_EVENTS_ON_NOTRACE
650 bool "Do NOT protect notrace function from kprobe events"
651 depends on KPROBE_EVENTS
652 depends on DYNAMIC_FTRACE
655 This is only for the developers who want to debug ftrace itself
658 If kprobes can use ftrace instead of breakpoint, ftrace related
659 functions are protected from kprobe-events to prevent an infinite
660 recursion or any unexpected execution path which leads to a kernel
663 This option disables such protection and allows you to put kprobe
664 events on ftrace functions for debugging ftrace by itself.
665 Note that this might let you shoot yourself in the foot.
670 bool "Enable uprobes-based dynamic events"
671 depends on ARCH_SUPPORTS_UPROBES
673 depends on PERF_EVENTS
676 select DYNAMIC_EVENTS
680 This allows the user to add tracing events on top of userspace
681 dynamic events (similar to tracepoints) on the fly via the trace
682 events interface. Those events can be inserted wherever uprobes
683 can probe, and record various registers.
684 This option is required if you plan to use perf-probe subcommand
685 of perf tools on user space applications.
688 depends on BPF_SYSCALL
689 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS
693 This allows the user to attach BPF programs to kprobe, uprobe, and
696 config DYNAMIC_EVENTS
702 config BPF_KPROBE_OVERRIDE
703 bool "Enable BPF programs to override a kprobed function"
704 depends on BPF_EVENTS
705 depends on FUNCTION_ERROR_INJECTION
708 Allows BPF to override the execution of a probed function and
709 set a different return value. This is used for error injection.
711 config FTRACE_MCOUNT_RECORD
713 depends on DYNAMIC_FTRACE
714 depends on HAVE_FTRACE_MCOUNT_RECORD
716 config FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
718 depends on FTRACE_MCOUNT_RECORD
720 config FTRACE_MCOUNT_USE_CC
722 depends on $(cc-option,-mrecord-mcount)
723 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
724 depends on FTRACE_MCOUNT_RECORD
726 config FTRACE_MCOUNT_USE_OBJTOOL
728 depends on HAVE_OBJTOOL_MCOUNT
729 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
730 depends on !FTRACE_MCOUNT_USE_CC
731 depends on FTRACE_MCOUNT_RECORD
734 config FTRACE_MCOUNT_USE_RECORDMCOUNT
736 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
737 depends on !FTRACE_MCOUNT_USE_CC
738 depends on !FTRACE_MCOUNT_USE_OBJTOOL
739 depends on FTRACE_MCOUNT_RECORD
743 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
745 tracing_map is a special-purpose lock-free map for tracing,
746 separated out as a stand-alone facility in order to allow it
747 to be shared between multiple tracers. It isn't meant to be
748 generally used outside of that context, and is normally
749 selected by tracers that use it.
752 bool "Synthetic trace events"
754 select DYNAMIC_EVENTS
757 Synthetic events are user-defined trace events that can be
758 used to combine data from other trace events or in fact any
759 data source. Synthetic events can be generated indirectly
760 via the trace() action of histogram triggers or directly
761 by way of an in-kernel API.
763 See Documentation/trace/events.rst or
764 Documentation/trace/histogram.rst for details and examples.
769 bool "User trace events"
771 select DYNAMIC_EVENTS
772 depends on BROKEN || COMPILE_TEST # API needs to be straighten out
774 User trace events are user-defined trace events that
775 can be used like an existing kernel trace event. User trace
776 events are generated by writing to a tracefs file. User
777 processes can determine if their tracing events should be
778 generated by memory mapping a tracefs file and checking for
779 an associated byte being non-zero.
784 bool "Histogram triggers"
785 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
788 select DYNAMIC_EVENTS
792 Hist triggers allow one or more arbitrary trace event fields
793 to be aggregated into hash tables and dumped to stdout by
794 reading a debugfs/tracefs file. They're useful for
795 gathering quick and dirty (though precise) summaries of
796 event activity as an initial guide for further investigation
797 using more advanced tools.
799 Inter-event tracing of quantities such as latencies is also
800 supported using hist triggers under this option.
802 See Documentation/trace/histogram.rst.
805 config TRACE_EVENT_INJECT
806 bool "Trace event injection"
809 Allow user-space to inject a specific trace event into the ring
810 buffer. This is mainly used for testing purpose.
814 config TRACEPOINT_BENCHMARK
815 bool "Add tracepoint that benchmarks tracepoints"
817 This option creates the tracepoint "benchmark:benchmark_event".
818 When the tracepoint is enabled, it kicks off a kernel thread that
819 goes into an infinite loop (calling cond_resched() to let other tasks
820 run), and calls the tracepoint. Each iteration will record the time
821 it took to write to the tracepoint and the next iteration that
822 data will be passed to the tracepoint itself. That is, the tracepoint
823 will report the time it took to do the previous tracepoint.
824 The string written to the tracepoint is a static string of 128 bytes
825 to keep the time the same. The initial string is simply a write of
826 "START". The second string records the cold cache time of the first
827 write which is not added to the rest of the calculations.
829 As it is a tight loop, it benchmarks as hot cache. That's fine because
830 we care most about hot paths that are probably in cache already.
832 An example of the output:
835 first=3672 [COLD CACHED]
836 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712
837 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337
838 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064
839 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411
840 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389
841 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666
844 config RING_BUFFER_BENCHMARK
845 tristate "Ring buffer benchmark stress tester"
846 depends on RING_BUFFER
848 This option creates a test to stress the ring buffer and benchmark it.
849 It creates its own ring buffer such that it will not interfere with
850 any other users of the ring buffer (such as ftrace). It then creates
851 a producer and consumer that will run for 10 seconds and sleep for
852 10 seconds. Each interval it will print out the number of events
853 it recorded and give a rough estimate of how long each iteration took.
855 It does not disable interrupts or raise its priority, so it may be
856 affected by processes that are running.
860 config TRACE_EVAL_MAP_FILE
861 bool "Show eval mappings for trace events"
864 The "print fmt" of the trace events will show the enum/sizeof names
865 instead of their values. This can cause problems for user space tools
866 that use this string to parse the raw data as user space does not know
867 how to convert the string to its value.
869 To fix this, there's a special macro in the kernel that can be used
870 to convert an enum/sizeof into its value. If this macro is used, then
871 the print fmt strings will be converted to their values.
873 If something does not get converted properly, this option can be
874 used to show what enums/sizeof the kernel tried to convert.
876 This option is for debugging the conversions. A file is created
877 in the tracing directory called "eval_map" that will show the
878 names matched with their values and what trace event system they
881 Normally, the mapping of the strings to values will be freed after
882 boot up or module load. With this option, they will not be freed, as
883 they are needed for the "eval_map" file. Enabling this option will
884 increase the memory footprint of the running kernel.
888 config FTRACE_RECORD_RECURSION
889 bool "Record functions that recurse in function tracing"
890 depends on FUNCTION_TRACER
892 All callbacks that attach to the function tracing have some sort
893 of protection against recursion. Even though the protection exists,
894 it adds overhead. This option will create a file in the tracefs
895 file system called "recursed_functions" that will list the functions
896 that triggered a recursion.
898 This will add more overhead to cases that have recursion.
902 config FTRACE_RECORD_RECURSION_SIZE
903 int "Max number of recursed functions to record"
905 depends on FTRACE_RECORD_RECURSION
907 This defines the limit of number of functions that can be
908 listed in the "recursed_functions" file, that lists all
909 the functions that caused a recursion to happen.
910 This file can be reset, but the limit can not change in
913 config RING_BUFFER_RECORD_RECURSION
914 bool "Record functions that recurse in the ring buffer"
915 depends on FTRACE_RECORD_RECURSION
916 # default y, because it is coupled with FTRACE_RECORD_RECURSION
919 The ring buffer has its own internal recursion. Although when
920 recursion happens it wont cause harm because of the protection,
921 but it does cause an unwanted overhead. Enabling this option will
922 place where recursion was detected into the ftrace "recursed_functions"
925 This will add more overhead to cases that have recursion.
927 config GCOV_PROFILE_FTRACE
928 bool "Enable GCOV profiling on ftrace subsystem"
929 depends on GCOV_KERNEL
931 Enable GCOV profiling on ftrace subsystem for checking
932 which functions/lines are tested.
936 Note that on a kernel compiled with this config, ftrace will
937 run significantly slower.
939 config FTRACE_SELFTEST
942 config FTRACE_STARTUP_TEST
943 bool "Perform a startup test on ftrace"
944 depends on GENERIC_TRACER
945 select FTRACE_SELFTEST
947 This option performs a series of startup tests on ftrace. On bootup
948 a series of tests are made to verify that the tracer is
949 functioning properly. It will do tests on all the configured
952 config EVENT_TRACE_STARTUP_TEST
953 bool "Run selftest on trace events"
954 depends on FTRACE_STARTUP_TEST
957 This option performs a test on all trace events in the system.
958 It basically just enables each event and runs some code that
959 will trigger events (not necessarily the event it enables)
960 This may take some time run as there are a lot of events.
962 config EVENT_TRACE_TEST_SYSCALLS
963 bool "Run selftest on syscall events"
964 depends on EVENT_TRACE_STARTUP_TEST
966 This option will also enable testing every syscall event.
967 It only enables the event and disables it and runs various loads
968 with the event enabled. This adds a bit more time for kernel boot
969 up since it runs this on every system call defined.
971 TBD - enable a way to actually call the syscalls as we test their
974 config FTRACE_SORT_STARTUP_TEST
975 bool "Verify compile time sorting of ftrace functions"
976 depends on DYNAMIC_FTRACE
977 depends on BUILDTIME_MCOUNT_SORT
979 Sorting of the mcount_loc sections that is used to find the
980 where the ftrace knows where to patch functions for tracing
981 and other callbacks is done at compile time. But if the sort
982 is not done correctly, it will cause non-deterministic failures.
983 When this is set, the sorted sections will be verified that they
984 are in deed sorted and will warn if they are not.
988 config RING_BUFFER_STARTUP_TEST
989 bool "Ring buffer startup self test"
990 depends on RING_BUFFER
992 Run a simple self test on the ring buffer on boot up. Late in the
993 kernel boot sequence, the test will start that kicks off
994 a thread per cpu. Each thread will write various size events
995 into the ring buffer. Another thread is created to send IPIs
996 to each of the threads, where the IPI handler will also write
997 to the ring buffer, to test/stress the nesting ability.
998 If any anomalies are discovered, a warning will be displayed
999 and all ring buffers will be disabled.
1001 The test runs for 10 seconds. This will slow your boot time
1002 by at least 10 more seconds.
1004 At the end of the test, statics and more checks are done.
1005 It will output the stats of each per cpu buffer. What
1006 was written, the sizes, what was read, what was lost, and
1007 other similar details.
1011 config RING_BUFFER_VALIDATE_TIME_DELTAS
1012 bool "Verify ring buffer time stamp deltas"
1013 depends on RING_BUFFER
1015 This will audit the time stamps on the ring buffer sub
1016 buffer to make sure that all the time deltas for the
1017 events on a sub buffer matches the current time stamp.
1018 This audit is performed for every event that is not
1019 interrupted, or interrupting another event. A check
1020 is also made when traversing sub buffers to make sure
1021 that all the deltas on the previous sub buffer do not
1022 add up to be greater than the current time stamp.
1024 NOTE: This adds significant overhead to recording of events,
1025 and should only be used to test the logic of the ring buffer.
1026 Do not use it on production systems.
1028 Only say Y if you understand what this does, and you
1029 still want it enabled. Otherwise say N
1031 config MMIOTRACE_TEST
1032 tristate "Test module for mmiotrace"
1033 depends on MMIOTRACE && m
1035 This is a dumb module for testing mmiotrace. It is very dangerous
1036 as it will write garbage to IO memory starting at a given address.
1037 However, it should be safe to use on e.g. unused portion of VRAM.
1039 Say N, unless you absolutely know what you are doing.
1041 config PREEMPTIRQ_DELAY_TEST
1042 tristate "Test module to create a preempt / IRQ disable delay thread to test latency tracers"
1045 Select this option to build a test module that can help test latency
1046 tracers by executing a preempt or irq disable section with a user
1047 configurable delay. The module busy waits for the duration of the
1050 For example, the following invocation generates a burst of three
1051 irq-disabled critical sections for 500us:
1052 modprobe preemptirq_delay_test test_mode=irq delay=500 burst_size=3
1054 What's more, if you want to attach the test on the cpu which the latency
1055 tracer is running on, specify cpu_affinity=cpu_num at the end of the
1060 config SYNTH_EVENT_GEN_TEST
1061 tristate "Test module for in-kernel synthetic event generation"
1062 depends on SYNTH_EVENTS
1064 This option creates a test module to check the base
1065 functionality of in-kernel synthetic event definition and
1068 To test, insert the module, and then check the trace buffer
1069 for the generated sample events.
1073 config KPROBE_EVENT_GEN_TEST
1074 tristate "Test module for in-kernel kprobe event generation"
1075 depends on KPROBE_EVENTS
1077 This option creates a test module to check the base
1078 functionality of in-kernel kprobe event definition.
1080 To test, insert the module, and then check the trace buffer
1081 for the generated kprobe events.
1085 config HIST_TRIGGERS_DEBUG
1086 bool "Hist trigger debug support"
1087 depends on HIST_TRIGGERS
1089 Add "hist_debug" file for each event, which when read will
1090 dump out a bunch of internal details about the hist triggers
1091 defined on that event.
1093 The hist_debug file serves a couple of purposes:
1095 - Helps developers verify that nothing is broken.
1097 - Provides educational information to support the details
1098 of the hist trigger internals as described by
1099 Documentation/trace/histogram-design.rst.
1101 The hist_debug output only covers the data structures
1102 related to the histogram definitions themselves and doesn't
1103 display the internals of map buckets or variable values of