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
13 config HAVE_FUNCTION_TRACER
16 See Documentation/trace/ftrace-design.rst
18 config HAVE_FUNCTION_GRAPH_TRACER
21 See Documentation/trace/ftrace-design.rst
23 config HAVE_DYNAMIC_FTRACE
26 See Documentation/trace/ftrace-design.rst
28 config HAVE_DYNAMIC_FTRACE_WITH_REGS
31 config HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
34 config HAVE_DYNAMIC_FTRACE_WITH_ARGS
37 If this is set, then arguments and stack can be found from
38 the pt_regs passed into the function callback regs parameter
39 by default, even without setting the REGS flag in the ftrace_ops.
40 This allows for use of regs_get_kernel_argument() and
41 kernel_stack_pointer().
43 config HAVE_FTRACE_MCOUNT_RECORD
46 See Documentation/trace/ftrace-design.rst
48 config HAVE_SYSCALL_TRACEPOINTS
51 See Documentation/trace/ftrace-design.rst
56 Arch supports the gcc options -pg with -mfentry
58 config HAVE_NOP_MCOUNT
61 Arch supports the gcc options -pg with -mrecord-mcount and -nop-mcount
63 config HAVE_OBJTOOL_MCOUNT
66 Arch supports objtool --mcount
68 config HAVE_C_RECORDMCOUNT
71 C version of recordmcount available?
73 config HAVE_BUILDTIME_MCOUNT_SORT
76 An architecture selects this if it sorts the mcount_loc section
79 config BUILDTIME_MCOUNT_SORT
82 depends on HAVE_BUILDTIME_MCOUNT_SORT && DYNAMIC_FTRACE
84 Sort the mcount_loc section at build time.
86 config TRACER_MAX_TRACE
98 select CONTEXT_SWITCH_TRACER
102 config CONTEXT_SWITCH_TRACER
105 config RING_BUFFER_ALLOW_SWAP
108 Allow the use of ring_buffer_swap_cpu.
109 Adds a very slight overhead to tracing when enabled.
111 config PREEMPTIRQ_TRACEPOINTS
113 depends on TRACE_PREEMPT_TOGGLE || TRACE_IRQFLAGS
117 Create preempt/irq toggle tracepoints if needed, so that other parts
118 of the kernel can use them to generate or add hooks to them.
120 # All tracer options should select GENERIC_TRACER. For those options that are
121 # enabled by all tracers (context switch and event tracer) they select TRACING.
122 # This allows those options to appear when no other tracer is selected. But the
123 # options do not appear when something else selects it. We need the two options
124 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
125 # hiding of the automatic options.
130 select STACKTRACE if STACKTRACE_SUPPORT
137 config GENERIC_TRACER
142 # Minimum requirements an architecture has to meet for us to
143 # be able to offer generic tracing facilities:
145 config TRACING_SUPPORT
147 depends on TRACE_IRQFLAGS_SUPPORT
148 depends on STACKTRACE_SUPPORT
153 depends on TRACING_SUPPORT
154 default y if DEBUG_KERNEL
156 Enable the kernel tracing infrastructure.
160 config BOOTTIME_TRACING
161 bool "Boot-time Tracing support"
165 Enable developer to setup ftrace subsystem via supplemental
166 kernel cmdline at boot time for debugging (tracing) driver
167 initialization and boot process.
169 config FUNCTION_TRACER
170 bool "Kernel Function Tracer"
171 depends on HAVE_FUNCTION_TRACER
173 select GENERIC_TRACER
174 select CONTEXT_SWITCH_TRACER
176 select TASKS_RCU if PREEMPTION
177 select TASKS_RUDE_RCU
179 Enable the kernel to trace every kernel function. This is done
180 by using a compiler feature to insert a small, 5-byte No-Operation
181 instruction at the beginning of every kernel function, which NOP
182 sequence is then dynamically patched into a tracer call when
183 tracing is enabled by the administrator. If it's runtime disabled
184 (the bootup default), then the overhead of the instructions is very
185 small and not measurable even in micro-benchmarks.
187 config FUNCTION_GRAPH_TRACER
188 bool "Kernel Function Graph Tracer"
189 depends on HAVE_FUNCTION_GRAPH_TRACER
190 depends on FUNCTION_TRACER
191 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
194 Enable the kernel to trace a function at both its return
196 Its first purpose is to trace the duration of functions and
197 draw a call graph for each thread with some information like
198 the return value. This is done by setting the current return
199 address on the current task structure into a stack of calls.
201 config DYNAMIC_FTRACE
202 bool "enable/disable function tracing dynamically"
203 depends on FUNCTION_TRACER
204 depends on HAVE_DYNAMIC_FTRACE
207 This option will modify all the calls to function tracing
208 dynamically (will patch them out of the binary image and
209 replace them with a No-Op instruction) on boot up. During
210 compile time, a table is made of all the locations that ftrace
211 can function trace, and this table is linked into the kernel
212 image. When this is enabled, functions can be individually
213 enabled, and the functions not enabled will not affect
214 performance of the system.
216 See the files in /sys/kernel/debug/tracing:
217 available_filter_functions
221 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
222 otherwise has native performance as long as no tracing is active.
224 config DYNAMIC_FTRACE_WITH_REGS
226 depends on DYNAMIC_FTRACE
227 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS
229 config DYNAMIC_FTRACE_WITH_DIRECT_CALLS
231 depends on DYNAMIC_FTRACE_WITH_REGS
232 depends on HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
234 config DYNAMIC_FTRACE_WITH_ARGS
236 depends on DYNAMIC_FTRACE
237 depends on HAVE_DYNAMIC_FTRACE_WITH_ARGS
239 config FUNCTION_PROFILER
240 bool "Kernel function profiler"
241 depends on FUNCTION_TRACER
244 This option enables the kernel function profiler. A file is created
245 in debugfs called function_profile_enabled which defaults to zero.
246 When a 1 is echoed into this file profiling begins, and when a
247 zero is entered, profiling stops. A "functions" file is created in
248 the trace_stat directory; this file shows the list of functions that
249 have been hit and their counters.
254 bool "Trace max stack"
255 depends on HAVE_FUNCTION_TRACER
256 select FUNCTION_TRACER
260 This special tracer records the maximum stack footprint of the
261 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
263 This tracer works by hooking into every function call that the
264 kernel executes, and keeping a maximum stack depth value and
265 stack-trace saved. If this is configured with DYNAMIC_FTRACE
266 then it will not have any overhead while the stack tracer
269 To enable the stack tracer on bootup, pass in 'stacktrace'
270 on the kernel command line.
272 The stack tracer can also be enabled or disabled via the
273 sysctl kernel.stack_tracer_enabled
277 config TRACE_PREEMPT_TOGGLE
280 Enables hooks which will be called when preemption is first disabled,
283 config IRQSOFF_TRACER
284 bool "Interrupts-off Latency Tracer"
286 depends on TRACE_IRQFLAGS_SUPPORT
287 select TRACE_IRQFLAGS
288 select GENERIC_TRACER
289 select TRACER_MAX_TRACE
290 select RING_BUFFER_ALLOW_SWAP
291 select TRACER_SNAPSHOT
292 select TRACER_SNAPSHOT_PER_CPU_SWAP
294 This option measures the time spent in irqs-off critical
295 sections, with microsecond accuracy.
297 The default measurement method is a maximum search, which is
298 disabled by default and can be runtime (re-)started
301 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
303 (Note that kernel size and overhead increase with this option
304 enabled. This option and the preempt-off timing option can be
305 used together or separately.)
307 config PREEMPT_TRACER
308 bool "Preemption-off Latency Tracer"
310 depends on PREEMPTION
311 select GENERIC_TRACER
312 select TRACER_MAX_TRACE
313 select RING_BUFFER_ALLOW_SWAP
314 select TRACER_SNAPSHOT
315 select TRACER_SNAPSHOT_PER_CPU_SWAP
316 select TRACE_PREEMPT_TOGGLE
318 This option measures the time spent in preemption-off critical
319 sections, with microsecond accuracy.
321 The default measurement method is a maximum search, which is
322 disabled by default and can be runtime (re-)started
325 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
327 (Note that kernel size and overhead increase with this option
328 enabled. This option and the irqs-off timing option can be
329 used together or separately.)
332 bool "Scheduling Latency Tracer"
333 select GENERIC_TRACER
334 select CONTEXT_SWITCH_TRACER
335 select TRACER_MAX_TRACE
336 select TRACER_SNAPSHOT
338 This tracer tracks the latency of the highest priority task
339 to be scheduled in, starting from the point it has woken up.
342 bool "Tracer to detect hardware latencies (like SMIs)"
343 select GENERIC_TRACER
345 This tracer, when enabled will create one or more kernel threads,
346 depending on what the cpumask file is set to, which each thread
347 spinning in a loop looking for interruptions caused by
348 something other than the kernel. For example, if a
349 System Management Interrupt (SMI) takes a noticeable amount of
350 time, this tracer will detect it. This is useful for testing
351 if a system is reliable for Real Time tasks.
353 Some files are created in the tracing directory when this
356 hwlat_detector/width - time in usecs for how long to spin for
357 hwlat_detector/window - time in usecs between the start of each
360 A kernel thread is created that will spin with interrupts disabled
361 for "width" microseconds in every "window" cycle. It will not spin
362 for "window - width" microseconds, where the system can
365 The output will appear in the trace and trace_pipe files.
367 When the tracer is not running, it has no affect on the system,
368 but when it is running, it can cause the system to be
369 periodically non responsive. Do not run this tracer on a
372 To enable this tracer, echo in "hwlat" into the current_tracer
373 file. Every time a latency is greater than tracing_thresh, it will
374 be recorded into the ring buffer.
376 config OSNOISE_TRACER
377 bool "OS Noise tracer"
378 select GENERIC_TRACER
380 In the context of high-performance computing (HPC), the Operating
381 System Noise (osnoise) refers to the interference experienced by an
382 application due to activities inside the operating system. In the
383 context of Linux, NMIs, IRQs, SoftIRQs, and any other system thread
384 can cause noise to the system. Moreover, hardware-related jobs can
385 also cause noise, for example, via SMIs.
387 The osnoise tracer leverages the hwlat_detector by running a similar
388 loop with preemption, SoftIRQs and IRQs enabled, thus allowing all
389 the sources of osnoise during its execution. The osnoise tracer takes
390 note of the entry and exit point of any source of interferences,
391 increasing a per-cpu interference counter. It saves an interference
392 counter for each source of interference. The interference counter for
393 NMI, IRQs, SoftIRQs, and threads is increased anytime the tool
394 observes these interferences' entry events. When a noise happens
395 without any interference from the operating system level, the
396 hardware noise counter increases, pointing to a hardware-related
397 noise. In this way, osnoise can account for any source of
398 interference. At the end of the period, the osnoise tracer prints
399 the sum of all noise, the max single noise, the percentage of CPU
400 available for the thread, and the counters for the noise sources.
402 In addition to the tracer, a set of tracepoints were added to
403 facilitate the identification of the osnoise source.
405 The output will appear in the trace and trace_pipe files.
407 To enable this tracer, echo in "osnoise" into the current_tracer
410 config TIMERLAT_TRACER
411 bool "Timerlat tracer"
412 select OSNOISE_TRACER
413 select GENERIC_TRACER
415 The timerlat tracer aims to help the preemptive kernel developers
416 to find sources of wakeup latencies of real-time threads.
418 The tracer creates a per-cpu kernel thread with real-time priority.
419 The tracer thread sets a periodic timer to wakeup itself, and goes
420 to sleep waiting for the timer to fire. At the wakeup, the thread
421 then computes a wakeup latency value as the difference between
422 the current time and the absolute time that the timer was set
425 The tracer prints two lines at every activation. The first is the
426 timer latency observed at the hardirq context before the
427 activation of the thread. The second is the timer latency observed
428 by the thread, which is the same level that cyclictest reports. The
429 ACTIVATION ID field serves to relate the irq execution to its
430 respective thread execution.
432 The tracer is build on top of osnoise tracer, and the osnoise:
433 events can be used to trace the source of interference from NMI,
434 IRQs and other threads. It also enables the capture of the
435 stacktrace at the IRQ context, which helps to identify the code
436 path that can cause thread delay.
439 bool "Memory mapped IO tracing"
440 depends on HAVE_MMIOTRACE_SUPPORT && PCI
441 select GENERIC_TRACER
443 Mmiotrace traces Memory Mapped I/O access and is meant for
444 debugging and reverse engineering. It is called from the ioremap
445 implementation and works via page faults. Tracing is disabled by
446 default and can be enabled at run-time.
448 See Documentation/trace/mmiotrace.rst.
449 If you are not helping to develop drivers, say N.
451 config ENABLE_DEFAULT_TRACERS
452 bool "Trace process context switches and events"
453 depends on !GENERIC_TRACER
456 This tracer hooks to various trace points in the kernel,
457 allowing the user to pick and choose which trace point they
458 want to trace. It also includes the sched_switch tracer plugin.
460 config FTRACE_SYSCALLS
461 bool "Trace syscalls"
462 depends on HAVE_SYSCALL_TRACEPOINTS
463 select GENERIC_TRACER
466 Basic tracer to catch the syscall entry and exit events.
468 config TRACER_SNAPSHOT
469 bool "Create a snapshot trace buffer"
470 select TRACER_MAX_TRACE
472 Allow tracing users to take snapshot of the current buffer using the
473 ftrace interface, e.g.:
475 echo 1 > /sys/kernel/debug/tracing/snapshot
478 config TRACER_SNAPSHOT_PER_CPU_SWAP
479 bool "Allow snapshot to swap per CPU"
480 depends on TRACER_SNAPSHOT
481 select RING_BUFFER_ALLOW_SWAP
483 Allow doing a snapshot of a single CPU buffer instead of a
484 full swap (all buffers). If this is set, then the following is
487 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot
489 After which, only the tracing buffer for CPU 2 was swapped with
490 the main tracing buffer, and the other CPU buffers remain the same.
492 When this is enabled, this adds a little more overhead to the
493 trace recording, as it needs to add some checks to synchronize
494 recording with swaps. But this does not affect the performance
495 of the overall system. This is enabled by default when the preempt
496 or irq latency tracers are enabled, as those need to swap as well
497 and already adds the overhead (plus a lot more).
499 config TRACE_BRANCH_PROFILING
501 select GENERIC_TRACER
504 prompt "Branch Profiling"
505 default BRANCH_PROFILE_NONE
507 The branch profiling is a software profiler. It will add hooks
508 into the C conditionals to test which path a branch takes.
510 The likely/unlikely profiler only looks at the conditions that
511 are annotated with a likely or unlikely macro.
513 The "all branch" profiler will profile every if-statement in the
514 kernel. This profiler will also enable the likely/unlikely
517 Either of the above profilers adds a bit of overhead to the system.
518 If unsure, choose "No branch profiling".
520 config BRANCH_PROFILE_NONE
521 bool "No branch profiling"
523 No branch profiling. Branch profiling adds a bit of overhead.
524 Only enable it if you want to analyse the branching behavior.
525 Otherwise keep it disabled.
527 config PROFILE_ANNOTATED_BRANCHES
528 bool "Trace likely/unlikely profiler"
529 select TRACE_BRANCH_PROFILING
531 This tracer profiles all likely and unlikely macros
532 in the kernel. It will display the results in:
534 /sys/kernel/debug/tracing/trace_stat/branch_annotated
536 Note: this will add a significant overhead; only turn this
537 on if you need to profile the system's use of these macros.
539 config PROFILE_ALL_BRANCHES
540 bool "Profile all if conditionals" if !FORTIFY_SOURCE
541 select TRACE_BRANCH_PROFILING
543 This tracer profiles all branch conditions. Every if ()
544 taken in the kernel is recorded whether it hit or miss.
545 The results will be displayed in:
547 /sys/kernel/debug/tracing/trace_stat/branch_all
549 This option also enables the likely/unlikely profiler.
551 This configuration, when enabled, will impose a great overhead
552 on the system. This should only be enabled when the system
553 is to be analyzed in much detail.
556 config TRACING_BRANCHES
559 Selected by tracers that will trace the likely and unlikely
560 conditions. This prevents the tracers themselves from being
561 profiled. Profiling the tracing infrastructure can only happen
562 when the likelys and unlikelys are not being traced.
565 bool "Trace likely/unlikely instances"
566 depends on TRACE_BRANCH_PROFILING
567 select TRACING_BRANCHES
569 This traces the events of likely and unlikely condition
570 calls in the kernel. The difference between this and the
571 "Trace likely/unlikely profiler" is that this is not a
572 histogram of the callers, but actually places the calling
573 events into a running trace buffer to see when and where the
574 events happened, as well as their results.
578 config BLK_DEV_IO_TRACE
579 bool "Support for tracing block IO actions"
585 select GENERIC_TRACER
588 Say Y here if you want to be able to trace the block layer actions
589 on a given queue. Tracing allows you to see any traffic happening
590 on a block device queue. For more information (and the userspace
591 support tools needed), fetch the blktrace tools from:
593 git://git.kernel.dk/blktrace.git
595 Tracing also is possible using the ftrace interface, e.g.:
597 echo 1 > /sys/block/sda/sda1/trace/enable
598 echo blk > /sys/kernel/debug/tracing/current_tracer
599 cat /sys/kernel/debug/tracing/trace_pipe
605 depends on HAVE_REGS_AND_STACK_ACCESS_API
606 bool "Enable kprobes-based dynamic events"
609 select DYNAMIC_EVENTS
612 This allows the user to add tracing events (similar to tracepoints)
613 on the fly via the ftrace interface. See
614 Documentation/trace/kprobetrace.rst for more details.
616 Those events can be inserted wherever kprobes can probe, and record
617 various register and memory values.
619 This option is also required by perf-probe subcommand of perf tools.
620 If you want to use perf tools, this option is strongly recommended.
622 config KPROBE_EVENTS_ON_NOTRACE
623 bool "Do NOT protect notrace function from kprobe events"
624 depends on KPROBE_EVENTS
625 depends on DYNAMIC_FTRACE
628 This is only for the developers who want to debug ftrace itself
631 If kprobes can use ftrace instead of breakpoint, ftrace related
632 functions are protected from kprobe-events to prevent an infinite
633 recursion or any unexpected execution path which leads to a kernel
636 This option disables such protection and allows you to put kprobe
637 events on ftrace functions for debugging ftrace by itself.
638 Note that this might let you shoot yourself in the foot.
643 bool "Enable uprobes-based dynamic events"
644 depends on ARCH_SUPPORTS_UPROBES
646 depends on PERF_EVENTS
649 select DYNAMIC_EVENTS
653 This allows the user to add tracing events on top of userspace
654 dynamic events (similar to tracepoints) on the fly via the trace
655 events interface. Those events can be inserted wherever uprobes
656 can probe, and record various registers.
657 This option is required if you plan to use perf-probe subcommand
658 of perf tools on user space applications.
661 depends on BPF_SYSCALL
662 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS
666 This allows the user to attach BPF programs to kprobe, uprobe, and
669 config DYNAMIC_EVENTS
675 config BPF_KPROBE_OVERRIDE
676 bool "Enable BPF programs to override a kprobed function"
677 depends on BPF_EVENTS
678 depends on FUNCTION_ERROR_INJECTION
681 Allows BPF to override the execution of a probed function and
682 set a different return value. This is used for error injection.
684 config FTRACE_MCOUNT_RECORD
686 depends on DYNAMIC_FTRACE
687 depends on HAVE_FTRACE_MCOUNT_RECORD
689 config FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
691 depends on FTRACE_MCOUNT_RECORD
693 config FTRACE_MCOUNT_USE_CC
695 depends on $(cc-option,-mrecord-mcount)
696 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
697 depends on FTRACE_MCOUNT_RECORD
699 config FTRACE_MCOUNT_USE_OBJTOOL
701 depends on HAVE_OBJTOOL_MCOUNT
702 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
703 depends on !FTRACE_MCOUNT_USE_CC
704 depends on FTRACE_MCOUNT_RECORD
706 config FTRACE_MCOUNT_USE_RECORDMCOUNT
708 depends on !FTRACE_MCOUNT_USE_PATCHABLE_FUNCTION_ENTRY
709 depends on !FTRACE_MCOUNT_USE_CC
710 depends on !FTRACE_MCOUNT_USE_OBJTOOL
711 depends on FTRACE_MCOUNT_RECORD
715 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
717 tracing_map is a special-purpose lock-free map for tracing,
718 separated out as a stand-alone facility in order to allow it
719 to be shared between multiple tracers. It isn't meant to be
720 generally used outside of that context, and is normally
721 selected by tracers that use it.
724 bool "Synthetic trace events"
726 select DYNAMIC_EVENTS
729 Synthetic events are user-defined trace events that can be
730 used to combine data from other trace events or in fact any
731 data source. Synthetic events can be generated indirectly
732 via the trace() action of histogram triggers or directly
733 by way of an in-kernel API.
735 See Documentation/trace/events.rst or
736 Documentation/trace/histogram.rst for details and examples.
741 bool "User trace events"
743 select DYNAMIC_EVENTS
744 depends on BROKEN || COMPILE_TEST # API needs to be straighten out
746 User trace events are user-defined trace events that
747 can be used like an existing kernel trace event. User trace
748 events are generated by writing to a tracefs file. User
749 processes can determine if their tracing events should be
750 generated by memory mapping a tracefs file and checking for
751 an associated byte being non-zero.
756 bool "Histogram triggers"
757 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
760 select DYNAMIC_EVENTS
764 Hist triggers allow one or more arbitrary trace event fields
765 to be aggregated into hash tables and dumped to stdout by
766 reading a debugfs/tracefs file. They're useful for
767 gathering quick and dirty (though precise) summaries of
768 event activity as an initial guide for further investigation
769 using more advanced tools.
771 Inter-event tracing of quantities such as latencies is also
772 supported using hist triggers under this option.
774 See Documentation/trace/histogram.rst.
777 config TRACE_EVENT_INJECT
778 bool "Trace event injection"
781 Allow user-space to inject a specific trace event into the ring
782 buffer. This is mainly used for testing purpose.
786 config TRACEPOINT_BENCHMARK
787 bool "Add tracepoint that benchmarks tracepoints"
789 This option creates the tracepoint "benchmark:benchmark_event".
790 When the tracepoint is enabled, it kicks off a kernel thread that
791 goes into an infinite loop (calling cond_resched() to let other tasks
792 run), and calls the tracepoint. Each iteration will record the time
793 it took to write to the tracepoint and the next iteration that
794 data will be passed to the tracepoint itself. That is, the tracepoint
795 will report the time it took to do the previous tracepoint.
796 The string written to the tracepoint is a static string of 128 bytes
797 to keep the time the same. The initial string is simply a write of
798 "START". The second string records the cold cache time of the first
799 write which is not added to the rest of the calculations.
801 As it is a tight loop, it benchmarks as hot cache. That's fine because
802 we care most about hot paths that are probably in cache already.
804 An example of the output:
807 first=3672 [COLD CACHED]
808 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712
809 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337
810 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064
811 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411
812 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389
813 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666
816 config RING_BUFFER_BENCHMARK
817 tristate "Ring buffer benchmark stress tester"
818 depends on RING_BUFFER
820 This option creates a test to stress the ring buffer and benchmark it.
821 It creates its own ring buffer such that it will not interfere with
822 any other users of the ring buffer (such as ftrace). It then creates
823 a producer and consumer that will run for 10 seconds and sleep for
824 10 seconds. Each interval it will print out the number of events
825 it recorded and give a rough estimate of how long each iteration took.
827 It does not disable interrupts or raise its priority, so it may be
828 affected by processes that are running.
832 config TRACE_EVAL_MAP_FILE
833 bool "Show eval mappings for trace events"
836 The "print fmt" of the trace events will show the enum/sizeof names
837 instead of their values. This can cause problems for user space tools
838 that use this string to parse the raw data as user space does not know
839 how to convert the string to its value.
841 To fix this, there's a special macro in the kernel that can be used
842 to convert an enum/sizeof into its value. If this macro is used, then
843 the print fmt strings will be converted to their values.
845 If something does not get converted properly, this option can be
846 used to show what enums/sizeof the kernel tried to convert.
848 This option is for debugging the conversions. A file is created
849 in the tracing directory called "eval_map" that will show the
850 names matched with their values and what trace event system they
853 Normally, the mapping of the strings to values will be freed after
854 boot up or module load. With this option, they will not be freed, as
855 they are needed for the "eval_map" file. Enabling this option will
856 increase the memory footprint of the running kernel.
860 config FTRACE_RECORD_RECURSION
861 bool "Record functions that recurse in function tracing"
862 depends on FUNCTION_TRACER
864 All callbacks that attach to the function tracing have some sort
865 of protection against recursion. Even though the protection exists,
866 it adds overhead. This option will create a file in the tracefs
867 file system called "recursed_functions" that will list the functions
868 that triggered a recursion.
870 This will add more overhead to cases that have recursion.
874 config FTRACE_RECORD_RECURSION_SIZE
875 int "Max number of recursed functions to record"
877 depends on FTRACE_RECORD_RECURSION
879 This defines the limit of number of functions that can be
880 listed in the "recursed_functions" file, that lists all
881 the functions that caused a recursion to happen.
882 This file can be reset, but the limit can not change in
885 config RING_BUFFER_RECORD_RECURSION
886 bool "Record functions that recurse in the ring buffer"
887 depends on FTRACE_RECORD_RECURSION
888 # default y, because it is coupled with FTRACE_RECORD_RECURSION
891 The ring buffer has its own internal recursion. Although when
892 recursion happens it wont cause harm because of the protection,
893 but it does cause an unwanted overhead. Enabling this option will
894 place where recursion was detected into the ftrace "recursed_functions"
897 This will add more overhead to cases that have recursion.
899 config GCOV_PROFILE_FTRACE
900 bool "Enable GCOV profiling on ftrace subsystem"
901 depends on GCOV_KERNEL
903 Enable GCOV profiling on ftrace subsystem for checking
904 which functions/lines are tested.
908 Note that on a kernel compiled with this config, ftrace will
909 run significantly slower.
911 config FTRACE_SELFTEST
914 config FTRACE_STARTUP_TEST
915 bool "Perform a startup test on ftrace"
916 depends on GENERIC_TRACER
917 select FTRACE_SELFTEST
919 This option performs a series of startup tests on ftrace. On bootup
920 a series of tests are made to verify that the tracer is
921 functioning properly. It will do tests on all the configured
924 config EVENT_TRACE_STARTUP_TEST
925 bool "Run selftest on trace events"
926 depends on FTRACE_STARTUP_TEST
929 This option performs a test on all trace events in the system.
930 It basically just enables each event and runs some code that
931 will trigger events (not necessarily the event it enables)
932 This may take some time run as there are a lot of events.
934 config EVENT_TRACE_TEST_SYSCALLS
935 bool "Run selftest on syscall events"
936 depends on EVENT_TRACE_STARTUP_TEST
938 This option will also enable testing every syscall event.
939 It only enables the event and disables it and runs various loads
940 with the event enabled. This adds a bit more time for kernel boot
941 up since it runs this on every system call defined.
943 TBD - enable a way to actually call the syscalls as we test their
946 config FTRACE_SORT_STARTUP_TEST
947 bool "Verify compile time sorting of ftrace functions"
948 depends on DYNAMIC_FTRACE
949 depends on BUILDTIME_MCOUNT_SORT
951 Sorting of the mcount_loc sections that is used to find the
952 where the ftrace knows where to patch functions for tracing
953 and other callbacks is done at compile time. But if the sort
954 is not done correctly, it will cause non-deterministic failures.
955 When this is set, the sorted sections will be verified that they
956 are in deed sorted and will warn if they are not.
960 config RING_BUFFER_STARTUP_TEST
961 bool "Ring buffer startup self test"
962 depends on RING_BUFFER
964 Run a simple self test on the ring buffer on boot up. Late in the
965 kernel boot sequence, the test will start that kicks off
966 a thread per cpu. Each thread will write various size events
967 into the ring buffer. Another thread is created to send IPIs
968 to each of the threads, where the IPI handler will also write
969 to the ring buffer, to test/stress the nesting ability.
970 If any anomalies are discovered, a warning will be displayed
971 and all ring buffers will be disabled.
973 The test runs for 10 seconds. This will slow your boot time
974 by at least 10 more seconds.
976 At the end of the test, statics and more checks are done.
977 It will output the stats of each per cpu buffer. What
978 was written, the sizes, what was read, what was lost, and
979 other similar details.
983 config RING_BUFFER_VALIDATE_TIME_DELTAS
984 bool "Verify ring buffer time stamp deltas"
985 depends on RING_BUFFER
987 This will audit the time stamps on the ring buffer sub
988 buffer to make sure that all the time deltas for the
989 events on a sub buffer matches the current time stamp.
990 This audit is performed for every event that is not
991 interrupted, or interrupting another event. A check
992 is also made when traversing sub buffers to make sure
993 that all the deltas on the previous sub buffer do not
994 add up to be greater than the current time stamp.
996 NOTE: This adds significant overhead to recording of events,
997 and should only be used to test the logic of the ring buffer.
998 Do not use it on production systems.
1000 Only say Y if you understand what this does, and you
1001 still want it enabled. Otherwise say N
1003 config MMIOTRACE_TEST
1004 tristate "Test module for mmiotrace"
1005 depends on MMIOTRACE && m
1007 This is a dumb module for testing mmiotrace. It is very dangerous
1008 as it will write garbage to IO memory starting at a given address.
1009 However, it should be safe to use on e.g. unused portion of VRAM.
1011 Say N, unless you absolutely know what you are doing.
1013 config PREEMPTIRQ_DELAY_TEST
1014 tristate "Test module to create a preempt / IRQ disable delay thread to test latency tracers"
1017 Select this option to build a test module that can help test latency
1018 tracers by executing a preempt or irq disable section with a user
1019 configurable delay. The module busy waits for the duration of the
1022 For example, the following invocation generates a burst of three
1023 irq-disabled critical sections for 500us:
1024 modprobe preemptirq_delay_test test_mode=irq delay=500 burst_size=3
1026 What's more, if you want to attach the test on the cpu which the latency
1027 tracer is running on, specify cpu_affinity=cpu_num at the end of the
1032 config SYNTH_EVENT_GEN_TEST
1033 tristate "Test module for in-kernel synthetic event generation"
1034 depends on SYNTH_EVENTS
1036 This option creates a test module to check the base
1037 functionality of in-kernel synthetic event definition and
1040 To test, insert the module, and then check the trace buffer
1041 for the generated sample events.
1045 config KPROBE_EVENT_GEN_TEST
1046 tristate "Test module for in-kernel kprobe event generation"
1047 depends on KPROBE_EVENTS
1049 This option creates a test module to check the base
1050 functionality of in-kernel kprobe event definition.
1052 To test, insert the module, and then check the trace buffer
1053 for the generated kprobe events.
1057 config HIST_TRIGGERS_DEBUG
1058 bool "Hist trigger debug support"
1059 depends on HIST_TRIGGERS
1061 Add "hist_debug" file for each event, which when read will
1062 dump out a bunch of internal details about the hist triggers
1063 defined on that event.
1065 The hist_debug file serves a couple of purposes:
1067 - Helps developers verify that nothing is broken.
1069 - Provides educational information to support the details
1070 of the hist trigger internals as described by
1071 Documentation/trace/histogram-design.rst.
1073 The hist_debug output only covers the data structures
1074 related to the histogram definitions themselves and doesn't
1075 display the internals of map buckets or variable values of