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
148 config GENERIC_TRACER
153 # Minimum requirements an architecture has to meet for us to
154 # be able to offer generic tracing facilities:
156 config TRACING_SUPPORT
158 depends on TRACE_IRQFLAGS_SUPPORT
159 depends on STACKTRACE_SUPPORT
164 depends on TRACING_SUPPORT
165 default y if DEBUG_KERNEL
167 Enable the kernel tracing infrastructure.
171 config BOOTTIME_TRACING
172 bool "Boot-time Tracing support"
176 Enable developer to setup ftrace subsystem via supplemental
177 kernel cmdline at boot time for debugging (tracing) driver
178 initialization and boot process.
180 config FUNCTION_TRACER
181 bool "Kernel Function Tracer"
182 depends on HAVE_FUNCTION_TRACER
184 select GENERIC_TRACER
185 select CONTEXT_SWITCH_TRACER
187 select TASKS_RCU if PREEMPTION
188 select TASKS_RUDE_RCU
190 Enable the kernel to trace every kernel function. This is done
191 by using a compiler feature to insert a small, 5-byte No-Operation
192 instruction at the beginning of every kernel function, which NOP
193 sequence is then dynamically patched into a tracer call when
194 tracing is enabled by the administrator. If it's runtime disabled
195 (the bootup default), then the overhead of the instructions is very
196 small and not measurable even in micro-benchmarks.
198 config FUNCTION_GRAPH_TRACER
199 bool "Kernel Function Graph Tracer"
200 depends on HAVE_FUNCTION_GRAPH_TRACER
201 depends on FUNCTION_TRACER
202 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
205 Enable the kernel to trace a function at both its return
207 Its first purpose is to trace the duration of functions and
208 draw a call graph for each thread with some information like
209 the return value. This is done by setting the current return
210 address on the current task structure into a stack of calls.
212 config DYNAMIC_FTRACE
213 bool "enable/disable function tracing dynamically"
214 depends on FUNCTION_TRACER
215 depends on HAVE_DYNAMIC_FTRACE
218 This option will modify all the calls to function tracing
219 dynamically (will patch them out of the binary image and
220 replace them with a No-Op instruction) on boot up. During
221 compile time, a table is made of all the locations that ftrace
222 can function trace, and this table is linked into the kernel
223 image. When this is enabled, functions can be individually
224 enabled, and the functions not enabled will not affect
225 performance of the system.
227 See the files in /sys/kernel/debug/tracing:
228 available_filter_functions
232 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
233 otherwise has native performance as long as no tracing is active.
235 config DYNAMIC_FTRACE_WITH_REGS
237 depends on DYNAMIC_FTRACE
238 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS
240 config DYNAMIC_FTRACE_WITH_DIRECT_CALLS
242 depends on DYNAMIC_FTRACE_WITH_REGS
243 depends on HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
245 config DYNAMIC_FTRACE_WITH_ARGS
247 depends on DYNAMIC_FTRACE
248 depends on HAVE_DYNAMIC_FTRACE_WITH_ARGS
251 bool "Kernel Function Probe (fprobe)"
252 depends on FUNCTION_TRACER
253 depends on DYNAMIC_FTRACE_WITH_REGS
254 depends on HAVE_RETHOOK
258 This option enables kernel function probe (fprobe) based on ftrace.
259 The fprobe is similar to kprobes, but probes only for kernel function
260 entries and exits. This also can probe multiple functions by one
265 config FUNCTION_PROFILER
266 bool "Kernel function profiler"
267 depends on FUNCTION_TRACER
270 This option enables the kernel function profiler. A file is created
271 in debugfs called function_profile_enabled which defaults to zero.
272 When a 1 is echoed into this file profiling begins, and when a
273 zero is entered, profiling stops. A "functions" file is created in
274 the trace_stat directory; this file shows the list of functions that
275 have been hit and their counters.
280 bool "Trace max stack"
281 depends on HAVE_FUNCTION_TRACER
282 select FUNCTION_TRACER
286 This special tracer records the maximum stack footprint of the
287 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
289 This tracer works by hooking into every function call that the
290 kernel executes, and keeping a maximum stack depth value and
291 stack-trace saved. If this is configured with DYNAMIC_FTRACE
292 then it will not have any overhead while the stack tracer
295 To enable the stack tracer on bootup, pass in 'stacktrace'
296 on the kernel command line.
298 The stack tracer can also be enabled or disabled via the
299 sysctl kernel.stack_tracer_enabled
303 config TRACE_PREEMPT_TOGGLE
306 Enables hooks which will be called when preemption is first disabled,
309 config IRQSOFF_TRACER
310 bool "Interrupts-off Latency Tracer"
312 depends on TRACE_IRQFLAGS_SUPPORT
313 select TRACE_IRQFLAGS
314 select GENERIC_TRACER
315 select TRACER_MAX_TRACE
316 select RING_BUFFER_ALLOW_SWAP
317 select TRACER_SNAPSHOT
318 select TRACER_SNAPSHOT_PER_CPU_SWAP
320 This option measures the time spent in irqs-off critical
321 sections, with microsecond accuracy.
323 The default measurement method is a maximum search, which is
324 disabled by default and can be runtime (re-)started
327 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
329 (Note that kernel size and overhead increase with this option
330 enabled. This option and the preempt-off timing option can be
331 used together or separately.)
333 config PREEMPT_TRACER
334 bool "Preemption-off Latency Tracer"
336 depends on PREEMPTION
337 select GENERIC_TRACER
338 select TRACER_MAX_TRACE
339 select RING_BUFFER_ALLOW_SWAP
340 select TRACER_SNAPSHOT
341 select TRACER_SNAPSHOT_PER_CPU_SWAP
342 select TRACE_PREEMPT_TOGGLE
344 This option measures the time spent in preemption-off critical
345 sections, with microsecond accuracy.
347 The default measurement method is a maximum search, which is
348 disabled by default and can be runtime (re-)started
351 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
353 (Note that kernel size and overhead increase with this option
354 enabled. This option and the irqs-off timing option can be
355 used together or separately.)
358 bool "Scheduling Latency Tracer"
359 select GENERIC_TRACER
360 select CONTEXT_SWITCH_TRACER
361 select TRACER_MAX_TRACE
362 select TRACER_SNAPSHOT
364 This tracer tracks the latency of the highest priority task
365 to be scheduled in, starting from the point it has woken up.
368 bool "Tracer to detect hardware latencies (like SMIs)"
369 select GENERIC_TRACER
371 This tracer, when enabled will create one or more kernel threads,
372 depending on what the cpumask file is set to, which each thread
373 spinning in a loop looking for interruptions caused by
374 something other than the kernel. For example, if a
375 System Management Interrupt (SMI) takes a noticeable amount of
376 time, this tracer will detect it. This is useful for testing
377 if a system is reliable for Real Time tasks.
379 Some files are created in the tracing directory when this
382 hwlat_detector/width - time in usecs for how long to spin for
383 hwlat_detector/window - time in usecs between the start of each
386 A kernel thread is created that will spin with interrupts disabled
387 for "width" microseconds in every "window" cycle. It will not spin
388 for "window - width" microseconds, where the system can
391 The output will appear in the trace and trace_pipe files.
393 When the tracer is not running, it has no affect on the system,
394 but when it is running, it can cause the system to be
395 periodically non responsive. Do not run this tracer on a
398 To enable this tracer, echo in "hwlat" into the current_tracer
399 file. Every time a latency is greater than tracing_thresh, it will
400 be recorded into the ring buffer.
402 config OSNOISE_TRACER
403 bool "OS Noise tracer"
404 select GENERIC_TRACER
406 In the context of high-performance computing (HPC), the Operating
407 System Noise (osnoise) refers to the interference experienced by an
408 application due to activities inside the operating system. In the
409 context of Linux, NMIs, IRQs, SoftIRQs, and any other system thread
410 can cause noise to the system. Moreover, hardware-related jobs can
411 also cause noise, for example, via SMIs.
413 The osnoise tracer leverages the hwlat_detector by running a similar
414 loop with preemption, SoftIRQs and IRQs enabled, thus allowing all
415 the sources of osnoise during its execution. The osnoise tracer takes
416 note of the entry and exit point of any source of interferences,
417 increasing a per-cpu interference counter. It saves an interference
418 counter for each source of interference. The interference counter for
419 NMI, IRQs, SoftIRQs, and threads is increased anytime the tool
420 observes these interferences' entry events. When a noise happens
421 without any interference from the operating system level, the
422 hardware noise counter increases, pointing to a hardware-related
423 noise. In this way, osnoise can account for any source of
424 interference. At the end of the period, the osnoise tracer prints
425 the sum of all noise, the max single noise, the percentage of CPU
426 available for the thread, and the counters for the noise sources.
428 In addition to the tracer, a set of tracepoints were added to
429 facilitate the identification of the osnoise source.
431 The output will appear in the trace and trace_pipe files.
433 To enable this tracer, echo in "osnoise" into the current_tracer
436 config TIMERLAT_TRACER
437 bool "Timerlat tracer"
438 select OSNOISE_TRACER
439 select GENERIC_TRACER
441 The timerlat tracer aims to help the preemptive kernel developers
442 to find sources of wakeup latencies of real-time threads.
444 The tracer creates a per-cpu kernel thread with real-time priority.
445 The tracer thread sets a periodic timer to wakeup itself, and goes
446 to sleep waiting for the timer to fire. At the wakeup, the thread
447 then computes a wakeup latency value as the difference between
448 the current time and the absolute time that the timer was set
451 The tracer prints two lines at every activation. The first is the
452 timer latency observed at the hardirq context before the
453 activation of the thread. The second is the timer latency observed
454 by the thread, which is the same level that cyclictest reports. The
455 ACTIVATION ID field serves to relate the irq execution to its
456 respective thread execution.
458 The tracer is build on top of osnoise tracer, and the osnoise:
459 events can be used to trace the source of interference from NMI,
460 IRQs and other threads. It also enables the capture of the
461 stacktrace at the IRQ context, which helps to identify the code
462 path that can cause thread delay.
465 bool "Memory mapped IO tracing"
466 depends on HAVE_MMIOTRACE_SUPPORT && PCI
467 select GENERIC_TRACER
469 Mmiotrace traces Memory Mapped I/O access and is meant for
470 debugging and reverse engineering. It is called from the ioremap
471 implementation and works via page faults. Tracing is disabled by
472 default and can be enabled at run-time.
474 See Documentation/trace/mmiotrace.rst.
475 If you are not helping to develop drivers, say N.
477 config ENABLE_DEFAULT_TRACERS
478 bool "Trace process context switches and events"
479 depends on !GENERIC_TRACER
482 This tracer hooks to various trace points in the kernel,
483 allowing the user to pick and choose which trace point they
484 want to trace. It also includes the sched_switch tracer plugin.
486 config FTRACE_SYSCALLS
487 bool "Trace syscalls"
488 depends on HAVE_SYSCALL_TRACEPOINTS
489 select GENERIC_TRACER
492 Basic tracer to catch the syscall entry and exit events.
494 config TRACER_SNAPSHOT
495 bool "Create a snapshot trace buffer"
496 select TRACER_MAX_TRACE
498 Allow tracing users to take snapshot of the current buffer using the
499 ftrace interface, e.g.:
501 echo 1 > /sys/kernel/debug/tracing/snapshot
504 config TRACER_SNAPSHOT_PER_CPU_SWAP
505 bool "Allow snapshot to swap per CPU"
506 depends on TRACER_SNAPSHOT
507 select RING_BUFFER_ALLOW_SWAP
509 Allow doing a snapshot of a single CPU buffer instead of a
510 full swap (all buffers). If this is set, then the following is
513 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot
515 After which, only the tracing buffer for CPU 2 was swapped with
516 the main tracing buffer, and the other CPU buffers remain the same.
518 When this is enabled, this adds a little more overhead to the
519 trace recording, as it needs to add some checks to synchronize
520 recording with swaps. But this does not affect the performance
521 of the overall system. This is enabled by default when the preempt
522 or irq latency tracers are enabled, as those need to swap as well
523 and already adds the overhead (plus a lot more).
525 config TRACE_BRANCH_PROFILING
527 select GENERIC_TRACER
530 prompt "Branch Profiling"
531 default BRANCH_PROFILE_NONE
533 The branch profiling is a software profiler. It will add hooks
534 into the C conditionals to test which path a branch takes.
536 The likely/unlikely profiler only looks at the conditions that
537 are annotated with a likely or unlikely macro.
539 The "all branch" profiler will profile every if-statement in the
540 kernel. This profiler will also enable the likely/unlikely
543 Either of the above profilers adds a bit of overhead to the system.
544 If unsure, choose "No branch profiling".
546 config BRANCH_PROFILE_NONE
547 bool "No branch profiling"
549 No branch profiling. Branch profiling adds a bit of overhead.
550 Only enable it if you want to analyse the branching behavior.
551 Otherwise keep it disabled.
553 config PROFILE_ANNOTATED_BRANCHES
554 bool "Trace likely/unlikely profiler"
555 select TRACE_BRANCH_PROFILING
557 This tracer profiles all likely and unlikely macros
558 in the kernel. It will display the results in:
560 /sys/kernel/debug/tracing/trace_stat/branch_annotated
562 Note: this will add a significant overhead; only turn this
563 on if you need to profile the system's use of these macros.
565 config PROFILE_ALL_BRANCHES
566 bool "Profile all if conditionals" if !FORTIFY_SOURCE
567 select TRACE_BRANCH_PROFILING
569 This tracer profiles all branch conditions. Every if ()
570 taken in the kernel is recorded whether it hit or miss.
571 The results will be displayed in:
573 /sys/kernel/debug/tracing/trace_stat/branch_all
575 This option also enables the likely/unlikely profiler.
577 This configuration, when enabled, will impose a great overhead
578 on the system. This should only be enabled when the system
579 is to be analyzed in much detail.
582 config TRACING_BRANCHES
585 Selected by tracers that will trace the likely and unlikely
586 conditions. This prevents the tracers themselves from being
587 profiled. Profiling the tracing infrastructure can only happen
588 when the likelys and unlikelys are not being traced.
591 bool "Trace likely/unlikely instances"
592 depends on TRACE_BRANCH_PROFILING
593 select TRACING_BRANCHES
595 This traces the events of likely and unlikely condition
596 calls in the kernel. The difference between this and the
597 "Trace likely/unlikely profiler" is that this is not a
598 histogram of the callers, but actually places the calling
599 events into a running trace buffer to see when and where the
600 events happened, as well as their results.
604 config BLK_DEV_IO_TRACE
605 bool "Support for tracing block IO actions"
611 select GENERIC_TRACER
614 Say Y here if you want to be able to trace the block layer actions
615 on a given queue. Tracing allows you to see any traffic happening
616 on a block device queue. For more information (and the userspace
617 support tools needed), fetch the blktrace tools from:
619 git://git.kernel.dk/blktrace.git
621 Tracing also is possible using the ftrace interface, e.g.:
623 echo 1 > /sys/block/sda/sda1/trace/enable
624 echo blk > /sys/kernel/debug/tracing/current_tracer
625 cat /sys/kernel/debug/tracing/trace_pipe
631 depends on HAVE_REGS_AND_STACK_ACCESS_API
632 bool "Enable kprobes-based dynamic events"
635 select DYNAMIC_EVENTS
638 This allows the user to add tracing events (similar to tracepoints)
639 on the fly via the ftrace interface. See
640 Documentation/trace/kprobetrace.rst for more details.
642 Those events can be inserted wherever kprobes can probe, and record
643 various register and memory values.
645 This option is also required by perf-probe subcommand of perf tools.
646 If you want to use perf tools, this option is strongly recommended.
648 config KPROBE_EVENTS_ON_NOTRACE
649 bool "Do NOT protect notrace function from kprobe events"
650 depends on KPROBE_EVENTS
651 depends on DYNAMIC_FTRACE
654 This is only for the developers who want to debug ftrace itself
657 If kprobes can use ftrace instead of breakpoint, ftrace related
658 functions are protected from kprobe-events to prevent an infinite
659 recursion or any unexpected execution path which leads to a kernel
662 This option disables such protection and allows you to put kprobe
663 events on ftrace functions for debugging ftrace by itself.
664 Note that this might let you shoot yourself in the foot.
669 bool "Enable uprobes-based dynamic events"
670 depends on ARCH_SUPPORTS_UPROBES
672 depends on PERF_EVENTS
675 select DYNAMIC_EVENTS
679 This allows the user to add tracing events on top of userspace
680 dynamic events (similar to tracepoints) on the fly via the trace
681 events interface. Those events can be inserted wherever uprobes
682 can probe, and record various registers.
683 This option is required if you plan to use perf-probe subcommand
684 of perf tools on user space applications.
687 depends on BPF_SYSCALL
688 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS
692 This allows the user to attach BPF programs to kprobe, uprobe, and
695 config DYNAMIC_EVENTS
701 config BPF_KPROBE_OVERRIDE
702 bool "Enable BPF programs to override a kprobed function"
703 depends on BPF_EVENTS
704 depends on FUNCTION_ERROR_INJECTION
707 Allows BPF to override the execution of a probed function and
708 set a different return value. This is used for error injection.
710 config FTRACE_MCOUNT_RECORD
712 depends on DYNAMIC_FTRACE
713 depends on HAVE_FTRACE_MCOUNT_RECORD
715 config FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
717 depends on FTRACE_MCOUNT_RECORD
719 config FTRACE_MCOUNT_USE_CC
721 depends on $(cc-option,-mrecord-mcount)
722 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
723 depends on FTRACE_MCOUNT_RECORD
725 config FTRACE_MCOUNT_USE_OBJTOOL
727 depends on HAVE_OBJTOOL_MCOUNT
728 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
729 depends on !FTRACE_MCOUNT_USE_CC
730 depends on FTRACE_MCOUNT_RECORD
732 config FTRACE_MCOUNT_USE_RECORDMCOUNT
734 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
735 depends on !FTRACE_MCOUNT_USE_CC
736 depends on !FTRACE_MCOUNT_USE_OBJTOOL
737 depends on FTRACE_MCOUNT_RECORD
741 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
743 tracing_map is a special-purpose lock-free map for tracing,
744 separated out as a stand-alone facility in order to allow it
745 to be shared between multiple tracers. It isn't meant to be
746 generally used outside of that context, and is normally
747 selected by tracers that use it.
750 bool "Synthetic trace events"
752 select DYNAMIC_EVENTS
755 Synthetic events are user-defined trace events that can be
756 used to combine data from other trace events or in fact any
757 data source. Synthetic events can be generated indirectly
758 via the trace() action of histogram triggers or directly
759 by way of an in-kernel API.
761 See Documentation/trace/events.rst or
762 Documentation/trace/histogram.rst for details and examples.
767 bool "User trace events"
769 select DYNAMIC_EVENTS
771 User trace events are user-defined trace events that
772 can be used like an existing kernel trace event. User trace
773 events are generated by writing to a tracefs file. User
774 processes can determine if their tracing events should be
775 generated by memory mapping a tracefs file and checking for
776 an associated byte being non-zero.
781 bool "Histogram triggers"
782 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
785 select DYNAMIC_EVENTS
789 Hist triggers allow one or more arbitrary trace event fields
790 to be aggregated into hash tables and dumped to stdout by
791 reading a debugfs/tracefs file. They're useful for
792 gathering quick and dirty (though precise) summaries of
793 event activity as an initial guide for further investigation
794 using more advanced tools.
796 Inter-event tracing of quantities such as latencies is also
797 supported using hist triggers under this option.
799 See Documentation/trace/histogram.rst.
802 config TRACE_EVENT_INJECT
803 bool "Trace event injection"
806 Allow user-space to inject a specific trace event into the ring
807 buffer. This is mainly used for testing purpose.
811 config TRACEPOINT_BENCHMARK
812 bool "Add tracepoint that benchmarks tracepoints"
814 This option creates the tracepoint "benchmark:benchmark_event".
815 When the tracepoint is enabled, it kicks off a kernel thread that
816 goes into an infinite loop (calling cond_resched() to let other tasks
817 run), and calls the tracepoint. Each iteration will record the time
818 it took to write to the tracepoint and the next iteration that
819 data will be passed to the tracepoint itself. That is, the tracepoint
820 will report the time it took to do the previous tracepoint.
821 The string written to the tracepoint is a static string of 128 bytes
822 to keep the time the same. The initial string is simply a write of
823 "START". The second string records the cold cache time of the first
824 write which is not added to the rest of the calculations.
826 As it is a tight loop, it benchmarks as hot cache. That's fine because
827 we care most about hot paths that are probably in cache already.
829 An example of the output:
832 first=3672 [COLD CACHED]
833 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712
834 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337
835 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064
836 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411
837 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389
838 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666
841 config RING_BUFFER_BENCHMARK
842 tristate "Ring buffer benchmark stress tester"
843 depends on RING_BUFFER
845 This option creates a test to stress the ring buffer and benchmark it.
846 It creates its own ring buffer such that it will not interfere with
847 any other users of the ring buffer (such as ftrace). It then creates
848 a producer and consumer that will run for 10 seconds and sleep for
849 10 seconds. Each interval it will print out the number of events
850 it recorded and give a rough estimate of how long each iteration took.
852 It does not disable interrupts or raise its priority, so it may be
853 affected by processes that are running.
857 config TRACE_EVAL_MAP_FILE
858 bool "Show eval mappings for trace events"
861 The "print fmt" of the trace events will show the enum/sizeof names
862 instead of their values. This can cause problems for user space tools
863 that use this string to parse the raw data as user space does not know
864 how to convert the string to its value.
866 To fix this, there's a special macro in the kernel that can be used
867 to convert an enum/sizeof into its value. If this macro is used, then
868 the print fmt strings will be converted to their values.
870 If something does not get converted properly, this option can be
871 used to show what enums/sizeof the kernel tried to convert.
873 This option is for debugging the conversions. A file is created
874 in the tracing directory called "eval_map" that will show the
875 names matched with their values and what trace event system they
878 Normally, the mapping of the strings to values will be freed after
879 boot up or module load. With this option, they will not be freed, as
880 they are needed for the "eval_map" file. Enabling this option will
881 increase the memory footprint of the running kernel.
885 config FTRACE_RECORD_RECURSION
886 bool "Record functions that recurse in function tracing"
887 depends on FUNCTION_TRACER
889 All callbacks that attach to the function tracing have some sort
890 of protection against recursion. Even though the protection exists,
891 it adds overhead. This option will create a file in the tracefs
892 file system called "recursed_functions" that will list the functions
893 that triggered a recursion.
895 This will add more overhead to cases that have recursion.
899 config FTRACE_RECORD_RECURSION_SIZE
900 int "Max number of recursed functions to record"
902 depends on FTRACE_RECORD_RECURSION
904 This defines the limit of number of functions that can be
905 listed in the "recursed_functions" file, that lists all
906 the functions that caused a recursion to happen.
907 This file can be reset, but the limit can not change in
910 config RING_BUFFER_RECORD_RECURSION
911 bool "Record functions that recurse in the ring buffer"
912 depends on FTRACE_RECORD_RECURSION
913 # default y, because it is coupled with FTRACE_RECORD_RECURSION
916 The ring buffer has its own internal recursion. Although when
917 recursion happens it wont cause harm because of the protection,
918 but it does cause an unwanted overhead. Enabling this option will
919 place where recursion was detected into the ftrace "recursed_functions"
922 This will add more overhead to cases that have recursion.
924 config GCOV_PROFILE_FTRACE
925 bool "Enable GCOV profiling on ftrace subsystem"
926 depends on GCOV_KERNEL
928 Enable GCOV profiling on ftrace subsystem for checking
929 which functions/lines are tested.
933 Note that on a kernel compiled with this config, ftrace will
934 run significantly slower.
936 config FTRACE_SELFTEST
939 config FTRACE_STARTUP_TEST
940 bool "Perform a startup test on ftrace"
941 depends on GENERIC_TRACER
942 select FTRACE_SELFTEST
944 This option performs a series of startup tests on ftrace. On bootup
945 a series of tests are made to verify that the tracer is
946 functioning properly. It will do tests on all the configured
949 config EVENT_TRACE_STARTUP_TEST
950 bool "Run selftest on trace events"
951 depends on FTRACE_STARTUP_TEST
954 This option performs a test on all trace events in the system.
955 It basically just enables each event and runs some code that
956 will trigger events (not necessarily the event it enables)
957 This may take some time run as there are a lot of events.
959 config EVENT_TRACE_TEST_SYSCALLS
960 bool "Run selftest on syscall events"
961 depends on EVENT_TRACE_STARTUP_TEST
963 This option will also enable testing every syscall event.
964 It only enables the event and disables it and runs various loads
965 with the event enabled. This adds a bit more time for kernel boot
966 up since it runs this on every system call defined.
968 TBD - enable a way to actually call the syscalls as we test their
971 config FTRACE_SORT_STARTUP_TEST
972 bool "Verify compile time sorting of ftrace functions"
973 depends on DYNAMIC_FTRACE
974 depends on BUILDTIME_MCOUNT_SORT
976 Sorting of the mcount_loc sections that is used to find the
977 where the ftrace knows where to patch functions for tracing
978 and other callbacks is done at compile time. But if the sort
979 is not done correctly, it will cause non-deterministic failures.
980 When this is set, the sorted sections will be verified that they
981 are in deed sorted and will warn if they are not.
985 config RING_BUFFER_STARTUP_TEST
986 bool "Ring buffer startup self test"
987 depends on RING_BUFFER
989 Run a simple self test on the ring buffer on boot up. Late in the
990 kernel boot sequence, the test will start that kicks off
991 a thread per cpu. Each thread will write various size events
992 into the ring buffer. Another thread is created to send IPIs
993 to each of the threads, where the IPI handler will also write
994 to the ring buffer, to test/stress the nesting ability.
995 If any anomalies are discovered, a warning will be displayed
996 and all ring buffers will be disabled.
998 The test runs for 10 seconds. This will slow your boot time
999 by at least 10 more seconds.
1001 At the end of the test, statics and more checks are done.
1002 It will output the stats of each per cpu buffer. What
1003 was written, the sizes, what was read, what was lost, and
1004 other similar details.
1008 config RING_BUFFER_VALIDATE_TIME_DELTAS
1009 bool "Verify ring buffer time stamp deltas"
1010 depends on RING_BUFFER
1012 This will audit the time stamps on the ring buffer sub
1013 buffer to make sure that all the time deltas for the
1014 events on a sub buffer matches the current time stamp.
1015 This audit is performed for every event that is not
1016 interrupted, or interrupting another event. A check
1017 is also made when traversing sub buffers to make sure
1018 that all the deltas on the previous sub buffer do not
1019 add up to be greater than the current time stamp.
1021 NOTE: This adds significant overhead to recording of events,
1022 and should only be used to test the logic of the ring buffer.
1023 Do not use it on production systems.
1025 Only say Y if you understand what this does, and you
1026 still want it enabled. Otherwise say N
1028 config MMIOTRACE_TEST
1029 tristate "Test module for mmiotrace"
1030 depends on MMIOTRACE && m
1032 This is a dumb module for testing mmiotrace. It is very dangerous
1033 as it will write garbage to IO memory starting at a given address.
1034 However, it should be safe to use on e.g. unused portion of VRAM.
1036 Say N, unless you absolutely know what you are doing.
1038 config PREEMPTIRQ_DELAY_TEST
1039 tristate "Test module to create a preempt / IRQ disable delay thread to test latency tracers"
1042 Select this option to build a test module that can help test latency
1043 tracers by executing a preempt or irq disable section with a user
1044 configurable delay. The module busy waits for the duration of the
1047 For example, the following invocation generates a burst of three
1048 irq-disabled critical sections for 500us:
1049 modprobe preemptirq_delay_test test_mode=irq delay=500 burst_size=3
1051 What's more, if you want to attach the test on the cpu which the latency
1052 tracer is running on, specify cpu_affinity=cpu_num at the end of the
1057 config SYNTH_EVENT_GEN_TEST
1058 tristate "Test module for in-kernel synthetic event generation"
1059 depends on SYNTH_EVENTS
1061 This option creates a test module to check the base
1062 functionality of in-kernel synthetic event definition and
1065 To test, insert the module, and then check the trace buffer
1066 for the generated sample events.
1070 config KPROBE_EVENT_GEN_TEST
1071 tristate "Test module for in-kernel kprobe event generation"
1072 depends on KPROBE_EVENTS
1074 This option creates a test module to check the base
1075 functionality of in-kernel kprobe event definition.
1077 To test, insert the module, and then check the trace buffer
1078 for the generated kprobe events.
1082 config HIST_TRIGGERS_DEBUG
1083 bool "Hist trigger debug support"
1084 depends on HIST_TRIGGERS
1086 Add "hist_debug" file for each event, which when read will
1087 dump out a bunch of internal details about the hist triggers
1088 defined on that event.
1090 The hist_debug file serves a couple of purposes:
1092 - Helps developers verify that nothing is broken.
1094 - Provides educational information to support the details
1095 of the hist trigger internals as described by
1096 Documentation/trace/histogram-design.rst.
1098 The hist_debug output only covers the data structures
1099 related to the histogram definitions themselves and doesn't
1100 display the internals of map buckets or variable values of