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_FUNCTION_GRAPH_RETVAL
37 config HAVE_DYNAMIC_FTRACE
40 See Documentation/trace/ftrace-design.rst
42 config HAVE_DYNAMIC_FTRACE_WITH_REGS
45 config HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
48 config HAVE_DYNAMIC_FTRACE_WITH_CALL_OPS
51 config HAVE_DYNAMIC_FTRACE_WITH_ARGS
54 If this is set, then arguments and stack can be found from
55 the ftrace_regs passed into the function callback regs parameter
56 by default, even without setting the REGS flag in the ftrace_ops.
57 This allows for use of ftrace_regs_get_argument() and
58 ftrace_regs_get_stack_pointer().
60 config HAVE_DYNAMIC_FTRACE_NO_PATCHABLE
63 If the architecture generates __patchable_function_entries sections
64 but does not want them included in the ftrace locations.
66 config HAVE_FTRACE_MCOUNT_RECORD
69 See Documentation/trace/ftrace-design.rst
71 config HAVE_SYSCALL_TRACEPOINTS
74 See Documentation/trace/ftrace-design.rst
79 Arch supports the gcc options -pg with -mfentry
81 config HAVE_NOP_MCOUNT
84 Arch supports the gcc options -pg with -mrecord-mcount and -nop-mcount
86 config HAVE_OBJTOOL_MCOUNT
89 Arch supports objtool --mcount
91 config HAVE_OBJTOOL_NOP_MCOUNT
94 Arch supports the objtool options --mcount with --mnop.
95 An architecture can select this if it wants to enable nop'ing
98 config HAVE_C_RECORDMCOUNT
101 C version of recordmcount available?
103 config HAVE_BUILDTIME_MCOUNT_SORT
106 An architecture selects this if it sorts the mcount_loc section
109 config BUILDTIME_MCOUNT_SORT
112 depends on HAVE_BUILDTIME_MCOUNT_SORT && DYNAMIC_FTRACE
114 Sort the mcount_loc section at build time.
116 config TRACER_MAX_TRACE
128 select CONTEXT_SWITCH_TRACER
132 config CONTEXT_SWITCH_TRACER
135 config RING_BUFFER_ALLOW_SWAP
138 Allow the use of ring_buffer_swap_cpu.
139 Adds a very slight overhead to tracing when enabled.
141 config PREEMPTIRQ_TRACEPOINTS
143 depends on TRACE_PREEMPT_TOGGLE || TRACE_IRQFLAGS
147 Create preempt/irq toggle tracepoints if needed, so that other parts
148 of the kernel can use them to generate or add hooks to them.
150 # All tracer options should select GENERIC_TRACER. For those options that are
151 # enabled by all tracers (context switch and event tracer) they select TRACING.
152 # This allows those options to appear when no other tracer is selected. But the
153 # options do not appear when something else selects it. We need the two options
154 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
155 # hiding of the automatic options.
160 select STACKTRACE if STACKTRACE_SUPPORT
166 select TASKS_RCU if PREEMPTION
168 config GENERIC_TRACER
173 # Minimum requirements an architecture has to meet for us to
174 # be able to offer generic tracing facilities:
176 config TRACING_SUPPORT
178 depends on TRACE_IRQFLAGS_SUPPORT
179 depends on STACKTRACE_SUPPORT
184 depends on TRACING_SUPPORT
185 default y if DEBUG_KERNEL
187 Enable the kernel tracing infrastructure.
191 config BOOTTIME_TRACING
192 bool "Boot-time Tracing support"
196 Enable developer to setup ftrace subsystem via supplemental
197 kernel cmdline at boot time for debugging (tracing) driver
198 initialization and boot process.
200 config FUNCTION_TRACER
201 bool "Kernel Function Tracer"
202 depends on HAVE_FUNCTION_TRACER
204 select GENERIC_TRACER
205 select CONTEXT_SWITCH_TRACER
207 select TASKS_RCU if PREEMPTION
208 select TASKS_RUDE_RCU
210 Enable the kernel to trace every kernel function. This is done
211 by using a compiler feature to insert a small, 5-byte No-Operation
212 instruction at the beginning of every kernel function, which NOP
213 sequence is then dynamically patched into a tracer call when
214 tracing is enabled by the administrator. If it's runtime disabled
215 (the bootup default), then the overhead of the instructions is very
216 small and not measurable even in micro-benchmarks (at least on
217 x86, but may have impact on other architectures).
219 config FUNCTION_GRAPH_TRACER
220 bool "Kernel Function Graph Tracer"
221 depends on HAVE_FUNCTION_GRAPH_TRACER
222 depends on FUNCTION_TRACER
223 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
226 Enable the kernel to trace a function at both its return
228 Its first purpose is to trace the duration of functions and
229 draw a call graph for each thread with some information like
230 the return value. This is done by setting the current return
231 address on the current task structure into a stack of calls.
233 config FUNCTION_GRAPH_RETVAL
234 bool "Kernel Function Graph Return Value"
235 depends on HAVE_FUNCTION_GRAPH_RETVAL
236 depends on FUNCTION_GRAPH_TRACER
239 Support recording and printing the function return value when
240 using function graph tracer. It can be helpful to locate functions
241 that return errors. This feature is off by default, and you can
242 enable it via the trace option funcgraph-retval.
243 See Documentation/trace/ftrace.rst
245 config DYNAMIC_FTRACE
246 bool "enable/disable function tracing dynamically"
247 depends on FUNCTION_TRACER
248 depends on HAVE_DYNAMIC_FTRACE
251 This option will modify all the calls to function tracing
252 dynamically (will patch them out of the binary image and
253 replace them with a No-Op instruction) on boot up. During
254 compile time, a table is made of all the locations that ftrace
255 can function trace, and this table is linked into the kernel
256 image. When this is enabled, functions can be individually
257 enabled, and the functions not enabled will not affect
258 performance of the system.
260 See the files in /sys/kernel/tracing:
261 available_filter_functions
265 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
266 otherwise has native performance as long as no tracing is active.
268 config DYNAMIC_FTRACE_WITH_REGS
270 depends on DYNAMIC_FTRACE
271 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS
273 config DYNAMIC_FTRACE_WITH_DIRECT_CALLS
275 depends on DYNAMIC_FTRACE_WITH_REGS || DYNAMIC_FTRACE_WITH_ARGS
276 depends on HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
278 config DYNAMIC_FTRACE_WITH_CALL_OPS
280 depends on HAVE_DYNAMIC_FTRACE_WITH_CALL_OPS
282 config DYNAMIC_FTRACE_WITH_ARGS
284 depends on DYNAMIC_FTRACE
285 depends on HAVE_DYNAMIC_FTRACE_WITH_ARGS
288 bool "Kernel Function Probe (fprobe)"
289 depends on FUNCTION_TRACER
290 depends on DYNAMIC_FTRACE_WITH_REGS
291 depends on HAVE_RETHOOK
295 This option enables kernel function probe (fprobe) based on ftrace.
296 The fprobe is similar to kprobes, but probes only for kernel function
297 entries and exits. This also can probe multiple functions by one
302 config FUNCTION_PROFILER
303 bool "Kernel function profiler"
304 depends on FUNCTION_TRACER
307 This option enables the kernel function profiler. A file is created
308 in debugfs called function_profile_enabled which defaults to zero.
309 When a 1 is echoed into this file profiling begins, and when a
310 zero is entered, profiling stops. A "functions" file is created in
311 the trace_stat directory; this file shows the list of functions that
312 have been hit and their counters.
317 bool "Trace max stack"
318 depends on HAVE_FUNCTION_TRACER
319 select FUNCTION_TRACER
323 This special tracer records the maximum stack footprint of the
324 kernel and displays it in /sys/kernel/tracing/stack_trace.
326 This tracer works by hooking into every function call that the
327 kernel executes, and keeping a maximum stack depth value and
328 stack-trace saved. If this is configured with DYNAMIC_FTRACE
329 then it will not have any overhead while the stack tracer
332 To enable the stack tracer on bootup, pass in 'stacktrace'
333 on the kernel command line.
335 The stack tracer can also be enabled or disabled via the
336 sysctl kernel.stack_tracer_enabled
340 config TRACE_PREEMPT_TOGGLE
343 Enables hooks which will be called when preemption is first disabled,
346 config IRQSOFF_TRACER
347 bool "Interrupts-off Latency Tracer"
349 depends on TRACE_IRQFLAGS_SUPPORT
350 select TRACE_IRQFLAGS
351 select GENERIC_TRACER
352 select TRACER_MAX_TRACE
353 select RING_BUFFER_ALLOW_SWAP
354 select TRACER_SNAPSHOT
355 select TRACER_SNAPSHOT_PER_CPU_SWAP
357 This option measures the time spent in irqs-off critical
358 sections, with microsecond accuracy.
360 The default measurement method is a maximum search, which is
361 disabled by default and can be runtime (re-)started
364 echo 0 > /sys/kernel/tracing/tracing_max_latency
366 (Note that kernel size and overhead increase with this option
367 enabled. This option and the preempt-off timing option can be
368 used together or separately.)
370 config PREEMPT_TRACER
371 bool "Preemption-off Latency Tracer"
373 depends on PREEMPTION
374 select GENERIC_TRACER
375 select TRACER_MAX_TRACE
376 select RING_BUFFER_ALLOW_SWAP
377 select TRACER_SNAPSHOT
378 select TRACER_SNAPSHOT_PER_CPU_SWAP
379 select TRACE_PREEMPT_TOGGLE
381 This option measures the time spent in preemption-off critical
382 sections, with microsecond accuracy.
384 The default measurement method is a maximum search, which is
385 disabled by default and can be runtime (re-)started
388 echo 0 > /sys/kernel/tracing/tracing_max_latency
390 (Note that kernel size and overhead increase with this option
391 enabled. This option and the irqs-off timing option can be
392 used together or separately.)
395 bool "Scheduling Latency Tracer"
396 select GENERIC_TRACER
397 select CONTEXT_SWITCH_TRACER
398 select TRACER_MAX_TRACE
399 select TRACER_SNAPSHOT
401 This tracer tracks the latency of the highest priority task
402 to be scheduled in, starting from the point it has woken up.
405 bool "Tracer to detect hardware latencies (like SMIs)"
406 select GENERIC_TRACER
407 select TRACER_MAX_TRACE
409 This tracer, when enabled will create one or more kernel threads,
410 depending on what the cpumask file is set to, which each thread
411 spinning in a loop looking for interruptions caused by
412 something other than the kernel. For example, if a
413 System Management Interrupt (SMI) takes a noticeable amount of
414 time, this tracer will detect it. This is useful for testing
415 if a system is reliable for Real Time tasks.
417 Some files are created in the tracing directory when this
420 hwlat_detector/width - time in usecs for how long to spin for
421 hwlat_detector/window - time in usecs between the start of each
424 A kernel thread is created that will spin with interrupts disabled
425 for "width" microseconds in every "window" cycle. It will not spin
426 for "window - width" microseconds, where the system can
429 The output will appear in the trace and trace_pipe files.
431 When the tracer is not running, it has no affect on the system,
432 but when it is running, it can cause the system to be
433 periodically non responsive. Do not run this tracer on a
436 To enable this tracer, echo in "hwlat" into the current_tracer
437 file. Every time a latency is greater than tracing_thresh, it will
438 be recorded into the ring buffer.
440 config OSNOISE_TRACER
441 bool "OS Noise tracer"
442 select GENERIC_TRACER
443 select TRACER_MAX_TRACE
445 In the context of high-performance computing (HPC), the Operating
446 System Noise (osnoise) refers to the interference experienced by an
447 application due to activities inside the operating system. In the
448 context of Linux, NMIs, IRQs, SoftIRQs, and any other system thread
449 can cause noise to the system. Moreover, hardware-related jobs can
450 also cause noise, for example, via SMIs.
452 The osnoise tracer leverages the hwlat_detector by running a similar
453 loop with preemption, SoftIRQs and IRQs enabled, thus allowing all
454 the sources of osnoise during its execution. The osnoise tracer takes
455 note of the entry and exit point of any source of interferences,
456 increasing a per-cpu interference counter. It saves an interference
457 counter for each source of interference. The interference counter for
458 NMI, IRQs, SoftIRQs, and threads is increased anytime the tool
459 observes these interferences' entry events. When a noise happens
460 without any interference from the operating system level, the
461 hardware noise counter increases, pointing to a hardware-related
462 noise. In this way, osnoise can account for any source of
463 interference. At the end of the period, the osnoise tracer prints
464 the sum of all noise, the max single noise, the percentage of CPU
465 available for the thread, and the counters for the noise sources.
467 In addition to the tracer, a set of tracepoints were added to
468 facilitate the identification of the osnoise source.
470 The output will appear in the trace and trace_pipe files.
472 To enable this tracer, echo in "osnoise" into the current_tracer
475 config TIMERLAT_TRACER
476 bool "Timerlat tracer"
477 select OSNOISE_TRACER
478 select GENERIC_TRACER
480 The timerlat tracer aims to help the preemptive kernel developers
481 to find sources of wakeup latencies of real-time threads.
483 The tracer creates a per-cpu kernel thread with real-time priority.
484 The tracer thread sets a periodic timer to wakeup itself, and goes
485 to sleep waiting for the timer to fire. At the wakeup, the thread
486 then computes a wakeup latency value as the difference between
487 the current time and the absolute time that the timer was set
490 The tracer prints two lines at every activation. The first is the
491 timer latency observed at the hardirq context before the
492 activation of the thread. The second is the timer latency observed
493 by the thread, which is the same level that cyclictest reports. The
494 ACTIVATION ID field serves to relate the irq execution to its
495 respective thread execution.
497 The tracer is build on top of osnoise tracer, and the osnoise:
498 events can be used to trace the source of interference from NMI,
499 IRQs and other threads. It also enables the capture of the
500 stacktrace at the IRQ context, which helps to identify the code
501 path that can cause thread delay.
504 bool "Memory mapped IO tracing"
505 depends on HAVE_MMIOTRACE_SUPPORT && PCI
506 select GENERIC_TRACER
508 Mmiotrace traces Memory Mapped I/O access and is meant for
509 debugging and reverse engineering. It is called from the ioremap
510 implementation and works via page faults. Tracing is disabled by
511 default and can be enabled at run-time.
513 See Documentation/trace/mmiotrace.rst.
514 If you are not helping to develop drivers, say N.
516 config ENABLE_DEFAULT_TRACERS
517 bool "Trace process context switches and events"
518 depends on !GENERIC_TRACER
521 This tracer hooks to various trace points in the kernel,
522 allowing the user to pick and choose which trace point they
523 want to trace. It also includes the sched_switch tracer plugin.
525 config FTRACE_SYSCALLS
526 bool "Trace syscalls"
527 depends on HAVE_SYSCALL_TRACEPOINTS
528 select GENERIC_TRACER
531 Basic tracer to catch the syscall entry and exit events.
533 config TRACER_SNAPSHOT
534 bool "Create a snapshot trace buffer"
535 select TRACER_MAX_TRACE
537 Allow tracing users to take snapshot of the current buffer using the
538 ftrace interface, e.g.:
540 echo 1 > /sys/kernel/tracing/snapshot
543 config TRACER_SNAPSHOT_PER_CPU_SWAP
544 bool "Allow snapshot to swap per CPU"
545 depends on TRACER_SNAPSHOT
546 select RING_BUFFER_ALLOW_SWAP
548 Allow doing a snapshot of a single CPU buffer instead of a
549 full swap (all buffers). If this is set, then the following is
552 echo 1 > /sys/kernel/tracing/per_cpu/cpu2/snapshot
554 After which, only the tracing buffer for CPU 2 was swapped with
555 the main tracing buffer, and the other CPU buffers remain the same.
557 When this is enabled, this adds a little more overhead to the
558 trace recording, as it needs to add some checks to synchronize
559 recording with swaps. But this does not affect the performance
560 of the overall system. This is enabled by default when the preempt
561 or irq latency tracers are enabled, as those need to swap as well
562 and already adds the overhead (plus a lot more).
564 config TRACE_BRANCH_PROFILING
566 select GENERIC_TRACER
569 prompt "Branch Profiling"
570 default BRANCH_PROFILE_NONE
572 The branch profiling is a software profiler. It will add hooks
573 into the C conditionals to test which path a branch takes.
575 The likely/unlikely profiler only looks at the conditions that
576 are annotated with a likely or unlikely macro.
578 The "all branch" profiler will profile every if-statement in the
579 kernel. This profiler will also enable the likely/unlikely
582 Either of the above profilers adds a bit of overhead to the system.
583 If unsure, choose "No branch profiling".
585 config BRANCH_PROFILE_NONE
586 bool "No branch profiling"
588 No branch profiling. Branch profiling adds a bit of overhead.
589 Only enable it if you want to analyse the branching behavior.
590 Otherwise keep it disabled.
592 config PROFILE_ANNOTATED_BRANCHES
593 bool "Trace likely/unlikely profiler"
594 select TRACE_BRANCH_PROFILING
596 This tracer profiles all likely and unlikely macros
597 in the kernel. It will display the results in:
599 /sys/kernel/tracing/trace_stat/branch_annotated
601 Note: this will add a significant overhead; only turn this
602 on if you need to profile the system's use of these macros.
604 config PROFILE_ALL_BRANCHES
605 bool "Profile all if conditionals" if !FORTIFY_SOURCE
606 select TRACE_BRANCH_PROFILING
608 This tracer profiles all branch conditions. Every if ()
609 taken in the kernel is recorded whether it hit or miss.
610 The results will be displayed in:
612 /sys/kernel/tracing/trace_stat/branch_all
614 This option also enables the likely/unlikely profiler.
616 This configuration, when enabled, will impose a great overhead
617 on the system. This should only be enabled when the system
618 is to be analyzed in much detail.
621 config TRACING_BRANCHES
624 Selected by tracers that will trace the likely and unlikely
625 conditions. This prevents the tracers themselves from being
626 profiled. Profiling the tracing infrastructure can only happen
627 when the likelys and unlikelys are not being traced.
630 bool "Trace likely/unlikely instances"
631 depends on TRACE_BRANCH_PROFILING
632 select TRACING_BRANCHES
634 This traces the events of likely and unlikely condition
635 calls in the kernel. The difference between this and the
636 "Trace likely/unlikely profiler" is that this is not a
637 histogram of the callers, but actually places the calling
638 events into a running trace buffer to see when and where the
639 events happened, as well as their results.
643 config BLK_DEV_IO_TRACE
644 bool "Support for tracing block IO actions"
650 select GENERIC_TRACER
653 Say Y here if you want to be able to trace the block layer actions
654 on a given queue. Tracing allows you to see any traffic happening
655 on a block device queue. For more information (and the userspace
656 support tools needed), fetch the blktrace tools from:
658 git://git.kernel.dk/blktrace.git
660 Tracing also is possible using the ftrace interface, e.g.:
662 echo 1 > /sys/block/sda/sda1/trace/enable
663 echo blk > /sys/kernel/tracing/current_tracer
664 cat /sys/kernel/tracing/trace_pipe
670 depends on HAVE_REGS_AND_STACK_ACCESS_API
671 bool "Enable fprobe-based dynamic events"
674 select DYNAMIC_EVENTS
677 This allows user to add tracing events on the function entry and
678 exit via ftrace interface. The syntax is same as the kprobe events
679 and the kprobe events on function entry and exit will be
680 transparently converted to this fprobe events.
682 config PROBE_EVENTS_BTF_ARGS
683 depends on HAVE_FUNCTION_ARG_ACCESS_API
684 depends on FPROBE_EVENTS || KPROBE_EVENTS
685 depends on DEBUG_INFO_BTF && BPF_SYSCALL
686 bool "Support BTF function arguments for probe events"
689 The user can specify the arguments of the probe event using the names
690 of the arguments of the probed function, when the probe location is a
691 kernel function entry or a tracepoint.
692 This is available only if BTF (BPF Type Format) support is enabled.
696 depends on HAVE_REGS_AND_STACK_ACCESS_API
697 bool "Enable kprobes-based dynamic events"
700 select DYNAMIC_EVENTS
703 This allows the user to add tracing events (similar to tracepoints)
704 on the fly via the ftrace interface. See
705 Documentation/trace/kprobetrace.rst for more details.
707 Those events can be inserted wherever kprobes can probe, and record
708 various register and memory values.
710 This option is also required by perf-probe subcommand of perf tools.
711 If you want to use perf tools, this option is strongly recommended.
713 config KPROBE_EVENTS_ON_NOTRACE
714 bool "Do NOT protect notrace function from kprobe events"
715 depends on KPROBE_EVENTS
716 depends on DYNAMIC_FTRACE
719 This is only for the developers who want to debug ftrace itself
722 If kprobes can use ftrace instead of breakpoint, ftrace related
723 functions are protected from kprobe-events to prevent an infinite
724 recursion or any unexpected execution path which leads to a kernel
727 This option disables such protection and allows you to put kprobe
728 events on ftrace functions for debugging ftrace by itself.
729 Note that this might let you shoot yourself in the foot.
734 bool "Enable uprobes-based dynamic events"
735 depends on ARCH_SUPPORTS_UPROBES
737 depends on PERF_EVENTS
740 select DYNAMIC_EVENTS
744 This allows the user to add tracing events on top of userspace
745 dynamic events (similar to tracepoints) on the fly via the trace
746 events interface. Those events can be inserted wherever uprobes
747 can probe, and record various registers.
748 This option is required if you plan to use perf-probe subcommand
749 of perf tools on user space applications.
752 depends on BPF_SYSCALL
753 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS
757 This allows the user to attach BPF programs to kprobe, uprobe, and
760 config DYNAMIC_EVENTS
766 config BPF_KPROBE_OVERRIDE
767 bool "Enable BPF programs to override a kprobed function"
768 depends on BPF_EVENTS
769 depends on FUNCTION_ERROR_INJECTION
772 Allows BPF to override the execution of a probed function and
773 set a different return value. This is used for error injection.
775 config FTRACE_MCOUNT_RECORD
777 depends on DYNAMIC_FTRACE
778 depends on HAVE_FTRACE_MCOUNT_RECORD
780 config FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
782 depends on FTRACE_MCOUNT_RECORD
784 config FTRACE_MCOUNT_USE_CC
786 depends on $(cc-option,-mrecord-mcount)
787 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
788 depends on FTRACE_MCOUNT_RECORD
790 config FTRACE_MCOUNT_USE_OBJTOOL
792 depends on HAVE_OBJTOOL_MCOUNT
793 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
794 depends on !FTRACE_MCOUNT_USE_CC
795 depends on FTRACE_MCOUNT_RECORD
798 config FTRACE_MCOUNT_USE_RECORDMCOUNT
800 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
801 depends on !FTRACE_MCOUNT_USE_CC
802 depends on !FTRACE_MCOUNT_USE_OBJTOOL
803 depends on FTRACE_MCOUNT_RECORD
807 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
809 tracing_map is a special-purpose lock-free map for tracing,
810 separated out as a stand-alone facility in order to allow it
811 to be shared between multiple tracers. It isn't meant to be
812 generally used outside of that context, and is normally
813 selected by tracers that use it.
816 bool "Synthetic trace events"
818 select DYNAMIC_EVENTS
821 Synthetic events are user-defined trace events that can be
822 used to combine data from other trace events or in fact any
823 data source. Synthetic events can be generated indirectly
824 via the trace() action of histogram triggers or directly
825 by way of an in-kernel API.
827 See Documentation/trace/events.rst or
828 Documentation/trace/histogram.rst for details and examples.
833 bool "User trace events"
835 select DYNAMIC_EVENTS
837 User trace events are user-defined trace events that
838 can be used like an existing kernel trace event. User trace
839 events are generated by writing to a tracefs file. User
840 processes can determine if their tracing events should be
841 generated by registering a value and bit with the kernel
842 that reflects when it is enabled or not.
844 See Documentation/trace/user_events.rst.
848 bool "Histogram triggers"
849 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
852 select DYNAMIC_EVENTS
856 Hist triggers allow one or more arbitrary trace event fields
857 to be aggregated into hash tables and dumped to stdout by
858 reading a debugfs/tracefs file. They're useful for
859 gathering quick and dirty (though precise) summaries of
860 event activity as an initial guide for further investigation
861 using more advanced tools.
863 Inter-event tracing of quantities such as latencies is also
864 supported using hist triggers under this option.
866 See Documentation/trace/histogram.rst.
869 config TRACE_EVENT_INJECT
870 bool "Trace event injection"
873 Allow user-space to inject a specific trace event into the ring
874 buffer. This is mainly used for testing purpose.
878 config TRACEPOINT_BENCHMARK
879 bool "Add tracepoint that benchmarks tracepoints"
881 This option creates the tracepoint "benchmark:benchmark_event".
882 When the tracepoint is enabled, it kicks off a kernel thread that
883 goes into an infinite loop (calling cond_resched() to let other tasks
884 run), and calls the tracepoint. Each iteration will record the time
885 it took to write to the tracepoint and the next iteration that
886 data will be passed to the tracepoint itself. That is, the tracepoint
887 will report the time it took to do the previous tracepoint.
888 The string written to the tracepoint is a static string of 128 bytes
889 to keep the time the same. The initial string is simply a write of
890 "START". The second string records the cold cache time of the first
891 write which is not added to the rest of the calculations.
893 As it is a tight loop, it benchmarks as hot cache. That's fine because
894 we care most about hot paths that are probably in cache already.
896 An example of the output:
899 first=3672 [COLD CACHED]
900 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712
901 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337
902 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064
903 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411
904 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389
905 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666
908 config RING_BUFFER_BENCHMARK
909 tristate "Ring buffer benchmark stress tester"
910 depends on RING_BUFFER
912 This option creates a test to stress the ring buffer and benchmark it.
913 It creates its own ring buffer such that it will not interfere with
914 any other users of the ring buffer (such as ftrace). It then creates
915 a producer and consumer that will run for 10 seconds and sleep for
916 10 seconds. Each interval it will print out the number of events
917 it recorded and give a rough estimate of how long each iteration took.
919 It does not disable interrupts or raise its priority, so it may be
920 affected by processes that are running.
924 config TRACE_EVAL_MAP_FILE
925 bool "Show eval mappings for trace events"
928 The "print fmt" of the trace events will show the enum/sizeof names
929 instead of their values. This can cause problems for user space tools
930 that use this string to parse the raw data as user space does not know
931 how to convert the string to its value.
933 To fix this, there's a special macro in the kernel that can be used
934 to convert an enum/sizeof into its value. If this macro is used, then
935 the print fmt strings will be converted to their values.
937 If something does not get converted properly, this option can be
938 used to show what enums/sizeof the kernel tried to convert.
940 This option is for debugging the conversions. A file is created
941 in the tracing directory called "eval_map" that will show the
942 names matched with their values and what trace event system they
945 Normally, the mapping of the strings to values will be freed after
946 boot up or module load. With this option, they will not be freed, as
947 they are needed for the "eval_map" file. Enabling this option will
948 increase the memory footprint of the running kernel.
952 config FTRACE_RECORD_RECURSION
953 bool "Record functions that recurse in function tracing"
954 depends on FUNCTION_TRACER
956 All callbacks that attach to the function tracing have some sort
957 of protection against recursion. Even though the protection exists,
958 it adds overhead. This option will create a file in the tracefs
959 file system called "recursed_functions" that will list the functions
960 that triggered a recursion.
962 This will add more overhead to cases that have recursion.
966 config FTRACE_RECORD_RECURSION_SIZE
967 int "Max number of recursed functions to record"
969 depends on FTRACE_RECORD_RECURSION
971 This defines the limit of number of functions that can be
972 listed in the "recursed_functions" file, that lists all
973 the functions that caused a recursion to happen.
974 This file can be reset, but the limit can not change in
977 config RING_BUFFER_RECORD_RECURSION
978 bool "Record functions that recurse in the ring buffer"
979 depends on FTRACE_RECORD_RECURSION
980 # default y, because it is coupled with FTRACE_RECORD_RECURSION
983 The ring buffer has its own internal recursion. Although when
984 recursion happens it won't cause harm because of the protection,
985 but it does cause unwanted overhead. Enabling this option will
986 place where recursion was detected into the ftrace "recursed_functions"
989 This will add more overhead to cases that have recursion.
991 config GCOV_PROFILE_FTRACE
992 bool "Enable GCOV profiling on ftrace subsystem"
993 depends on GCOV_KERNEL
995 Enable GCOV profiling on ftrace subsystem for checking
996 which functions/lines are tested.
1000 Note that on a kernel compiled with this config, ftrace will
1001 run significantly slower.
1003 config FTRACE_SELFTEST
1006 config FTRACE_STARTUP_TEST
1007 bool "Perform a startup test on ftrace"
1008 depends on GENERIC_TRACER
1009 select FTRACE_SELFTEST
1011 This option performs a series of startup tests on ftrace. On bootup
1012 a series of tests are made to verify that the tracer is
1013 functioning properly. It will do tests on all the configured
1016 config EVENT_TRACE_STARTUP_TEST
1017 bool "Run selftest on trace events"
1018 depends on FTRACE_STARTUP_TEST
1021 This option performs a test on all trace events in the system.
1022 It basically just enables each event and runs some code that
1023 will trigger events (not necessarily the event it enables)
1024 This may take some time run as there are a lot of events.
1026 config EVENT_TRACE_TEST_SYSCALLS
1027 bool "Run selftest on syscall events"
1028 depends on EVENT_TRACE_STARTUP_TEST
1030 This option will also enable testing every syscall event.
1031 It only enables the event and disables it and runs various loads
1032 with the event enabled. This adds a bit more time for kernel boot
1033 up since it runs this on every system call defined.
1035 TBD - enable a way to actually call the syscalls as we test their
1038 config FTRACE_SORT_STARTUP_TEST
1039 bool "Verify compile time sorting of ftrace functions"
1040 depends on DYNAMIC_FTRACE
1041 depends on BUILDTIME_MCOUNT_SORT
1043 Sorting of the mcount_loc sections that is used to find the
1044 where the ftrace knows where to patch functions for tracing
1045 and other callbacks is done at compile time. But if the sort
1046 is not done correctly, it will cause non-deterministic failures.
1047 When this is set, the sorted sections will be verified that they
1048 are in deed sorted and will warn if they are not.
1052 config RING_BUFFER_STARTUP_TEST
1053 bool "Ring buffer startup self test"
1054 depends on RING_BUFFER
1056 Run a simple self test on the ring buffer on boot up. Late in the
1057 kernel boot sequence, the test will start that kicks off
1058 a thread per cpu. Each thread will write various size events
1059 into the ring buffer. Another thread is created to send IPIs
1060 to each of the threads, where the IPI handler will also write
1061 to the ring buffer, to test/stress the nesting ability.
1062 If any anomalies are discovered, a warning will be displayed
1063 and all ring buffers will be disabled.
1065 The test runs for 10 seconds. This will slow your boot time
1066 by at least 10 more seconds.
1068 At the end of the test, statistics and more checks are done.
1069 It will output the stats of each per cpu buffer: What
1070 was written, the sizes, what was read, what was lost, and
1071 other similar details.
1075 config RING_BUFFER_VALIDATE_TIME_DELTAS
1076 bool "Verify ring buffer time stamp deltas"
1077 depends on RING_BUFFER
1079 This will audit the time stamps on the ring buffer sub
1080 buffer to make sure that all the time deltas for the
1081 events on a sub buffer matches the current time stamp.
1082 This audit is performed for every event that is not
1083 interrupted, or interrupting another event. A check
1084 is also made when traversing sub buffers to make sure
1085 that all the deltas on the previous sub buffer do not
1086 add up to be greater than the current time stamp.
1088 NOTE: This adds significant overhead to recording of events,
1089 and should only be used to test the logic of the ring buffer.
1090 Do not use it on production systems.
1092 Only say Y if you understand what this does, and you
1093 still want it enabled. Otherwise say N
1095 config MMIOTRACE_TEST
1096 tristate "Test module for mmiotrace"
1097 depends on MMIOTRACE && m
1099 This is a dumb module for testing mmiotrace. It is very dangerous
1100 as it will write garbage to IO memory starting at a given address.
1101 However, it should be safe to use on e.g. unused portion of VRAM.
1103 Say N, unless you absolutely know what you are doing.
1105 config PREEMPTIRQ_DELAY_TEST
1106 tristate "Test module to create a preempt / IRQ disable delay thread to test latency tracers"
1109 Select this option to build a test module that can help test latency
1110 tracers by executing a preempt or irq disable section with a user
1111 configurable delay. The module busy waits for the duration of the
1114 For example, the following invocation generates a burst of three
1115 irq-disabled critical sections for 500us:
1116 modprobe preemptirq_delay_test test_mode=irq delay=500 burst_size=3
1118 What's more, if you want to attach the test on the cpu which the latency
1119 tracer is running on, specify cpu_affinity=cpu_num at the end of the
1124 config SYNTH_EVENT_GEN_TEST
1125 tristate "Test module for in-kernel synthetic event generation"
1126 depends on SYNTH_EVENTS
1128 This option creates a test module to check the base
1129 functionality of in-kernel synthetic event definition and
1132 To test, insert the module, and then check the trace buffer
1133 for the generated sample events.
1137 config KPROBE_EVENT_GEN_TEST
1138 tristate "Test module for in-kernel kprobe event generation"
1139 depends on KPROBE_EVENTS
1141 This option creates a test module to check the base
1142 functionality of in-kernel kprobe event definition.
1144 To test, insert the module, and then check the trace buffer
1145 for the generated kprobe events.
1149 config HIST_TRIGGERS_DEBUG
1150 bool "Hist trigger debug support"
1151 depends on HIST_TRIGGERS
1153 Add "hist_debug" file for each event, which when read will
1154 dump out a bunch of internal details about the hist triggers
1155 defined on that event.
1157 The hist_debug file serves a couple of purposes:
1159 - Helps developers verify that nothing is broken.
1161 - Provides educational information to support the details
1162 of the hist trigger internals as described by
1163 Documentation/trace/histogram-design.rst.
1165 The hist_debug output only covers the data structures
1166 related to the histogram definitions themselves and doesn't
1167 display the internals of map buckets or variable values of
1172 source "kernel/trace/rv/Kconfig"
git.samba.org / sfrench / cifs-2.6.git / blob