2 # Architectures that offer an FUNCTION_TRACER implementation should
3 # select HAVE_FUNCTION_TRACER:
6 config USER_STACKTRACE_SUPPORT
12 config HAVE_FTRACE_NMI_ENTER
15 See Documentation/trace/ftrace-design.txt
17 config HAVE_FUNCTION_TRACER
20 See Documentation/trace/ftrace-design.txt
22 config HAVE_FUNCTION_GRAPH_TRACER
25 See Documentation/trace/ftrace-design.txt
27 config HAVE_DYNAMIC_FTRACE
30 See Documentation/trace/ftrace-design.txt
32 config HAVE_DYNAMIC_FTRACE_WITH_REGS
35 config HAVE_FTRACE_MCOUNT_RECORD
38 See Documentation/trace/ftrace-design.txt
40 config HAVE_SYSCALL_TRACEPOINTS
43 See Documentation/trace/ftrace-design.txt
48 Arch supports the gcc options -pg with -mfentry
50 config HAVE_C_RECORDMCOUNT
53 C version of recordmcount available?
55 config TRACER_MAX_TRACE
66 config FTRACE_NMI_ENTER
68 depends on HAVE_FTRACE_NMI_ENTER
72 select CONTEXT_SWITCH_TRACER
76 config CONTEXT_SWITCH_TRACER
79 config RING_BUFFER_ALLOW_SWAP
82 Allow the use of ring_buffer_swap_cpu.
83 Adds a very slight overhead to tracing when enabled.
85 # All tracer options should select GENERIC_TRACER. For those options that are
86 # enabled by all tracers (context switch and event tracer) they select TRACING.
87 # This allows those options to appear when no other tracer is selected. But the
88 # options do not appear when something else selects it. We need the two options
89 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
90 # hiding of the automatic options.
96 select STACKTRACE if STACKTRACE_SUPPORT
103 config GENERIC_TRACER
108 # Minimum requirements an architecture has to meet for us to
109 # be able to offer generic tracing facilities:
111 config TRACING_SUPPORT
113 # PPC32 has no irqflags tracing support, but it can use most of the
114 # tracers anyway, they were tested to build and work. Note that new
115 # exceptions to this list aren't welcomed, better implement the
116 # irqflags tracing for your architecture.
117 depends on TRACE_IRQFLAGS_SUPPORT || PPC32
118 depends on STACKTRACE_SUPPORT
125 default y if DEBUG_KERNEL
127 Enable the kernel tracing infrastructure.
131 config FUNCTION_TRACER
132 bool "Kernel Function Tracer"
133 depends on HAVE_FUNCTION_TRACER
135 select GENERIC_TRACER
136 select CONTEXT_SWITCH_TRACER
138 select TASKS_RCU if PREEMPT
140 Enable the kernel to trace every kernel function. This is done
141 by using a compiler feature to insert a small, 5-byte No-Operation
142 instruction at the beginning of every kernel function, which NOP
143 sequence is then dynamically patched into a tracer call when
144 tracing is enabled by the administrator. If it's runtime disabled
145 (the bootup default), then the overhead of the instructions is very
146 small and not measurable even in micro-benchmarks.
148 config FUNCTION_GRAPH_TRACER
149 bool "Kernel Function Graph Tracer"
150 depends on HAVE_FUNCTION_GRAPH_TRACER
151 depends on FUNCTION_TRACER
152 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
155 Enable the kernel to trace a function at both its return
157 Its first purpose is to trace the duration of functions and
158 draw a call graph for each thread with some information like
159 the return value. This is done by setting the current return
160 address on the current task structure into a stack of calls.
163 config PREEMPTIRQ_EVENTS
164 bool "Enable trace events for preempt and irq disable/enable"
165 select TRACE_IRQFLAGS
166 depends on DEBUG_PREEMPT || !PROVE_LOCKING
169 Enable tracing of disable and enable events for preemption and irqs.
170 For tracing preempt disable/enable events, DEBUG_PREEMPT must be
171 enabled. For tracing irq disable/enable events, PROVE_LOCKING must
174 config IRQSOFF_TRACER
175 bool "Interrupts-off Latency Tracer"
177 depends on TRACE_IRQFLAGS_SUPPORT
178 depends on !ARCH_USES_GETTIMEOFFSET
179 select TRACE_IRQFLAGS
180 select GENERIC_TRACER
181 select TRACER_MAX_TRACE
182 select RING_BUFFER_ALLOW_SWAP
183 select TRACER_SNAPSHOT
184 select TRACER_SNAPSHOT_PER_CPU_SWAP
186 This option measures the time spent in irqs-off critical
187 sections, with microsecond accuracy.
189 The default measurement method is a maximum search, which is
190 disabled by default and can be runtime (re-)started
193 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
195 (Note that kernel size and overhead increase with this option
196 enabled. This option and the preempt-off timing option can be
197 used together or separately.)
199 config PREEMPT_TRACER
200 bool "Preemption-off Latency Tracer"
202 depends on !ARCH_USES_GETTIMEOFFSET
204 select GENERIC_TRACER
205 select TRACER_MAX_TRACE
206 select RING_BUFFER_ALLOW_SWAP
207 select TRACER_SNAPSHOT
208 select TRACER_SNAPSHOT_PER_CPU_SWAP
210 This option measures the time spent in preemption-off critical
211 sections, with microsecond accuracy.
213 The default measurement method is a maximum search, which is
214 disabled by default and can be runtime (re-)started
217 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
219 (Note that kernel size and overhead increase with this option
220 enabled. This option and the irqs-off timing option can be
221 used together or separately.)
224 bool "Scheduling Latency Tracer"
225 select GENERIC_TRACER
226 select CONTEXT_SWITCH_TRACER
227 select TRACER_MAX_TRACE
228 select TRACER_SNAPSHOT
230 This tracer tracks the latency of the highest priority task
231 to be scheduled in, starting from the point it has woken up.
234 bool "Tracer to detect hardware latencies (like SMIs)"
235 select GENERIC_TRACER
237 This tracer, when enabled will create one or more kernel threads,
238 depending on what the cpumask file is set to, which each thread
239 spinning in a loop looking for interruptions caused by
240 something other than the kernel. For example, if a
241 System Management Interrupt (SMI) takes a noticeable amount of
242 time, this tracer will detect it. This is useful for testing
243 if a system is reliable for Real Time tasks.
245 Some files are created in the tracing directory when this
248 hwlat_detector/width - time in usecs for how long to spin for
249 hwlat_detector/window - time in usecs between the start of each
252 A kernel thread is created that will spin with interrupts disabled
253 for "width" microseconds in every "window" cycle. It will not spin
254 for "window - width" microseconds, where the system can
257 The output will appear in the trace and trace_pipe files.
259 When the tracer is not running, it has no affect on the system,
260 but when it is running, it can cause the system to be
261 periodically non responsive. Do not run this tracer on a
264 To enable this tracer, echo in "hwlat" into the current_tracer
265 file. Every time a latency is greater than tracing_thresh, it will
266 be recorded into the ring buffer.
268 config ENABLE_DEFAULT_TRACERS
269 bool "Trace process context switches and events"
270 depends on !GENERIC_TRACER
273 This tracer hooks to various trace points in the kernel,
274 allowing the user to pick and choose which trace point they
275 want to trace. It also includes the sched_switch tracer plugin.
277 config FTRACE_SYSCALLS
278 bool "Trace syscalls"
279 depends on HAVE_SYSCALL_TRACEPOINTS
280 select GENERIC_TRACER
283 Basic tracer to catch the syscall entry and exit events.
285 config TRACER_SNAPSHOT
286 bool "Create a snapshot trace buffer"
287 select TRACER_MAX_TRACE
289 Allow tracing users to take snapshot of the current buffer using the
290 ftrace interface, e.g.:
292 echo 1 > /sys/kernel/debug/tracing/snapshot
295 config TRACER_SNAPSHOT_PER_CPU_SWAP
296 bool "Allow snapshot to swap per CPU"
297 depends on TRACER_SNAPSHOT
298 select RING_BUFFER_ALLOW_SWAP
300 Allow doing a snapshot of a single CPU buffer instead of a
301 full swap (all buffers). If this is set, then the following is
304 echo 1 > /sys/kernel/debug/tracing/per_cpu/cpu2/snapshot
306 After which, only the tracing buffer for CPU 2 was swapped with
307 the main tracing buffer, and the other CPU buffers remain the same.
309 When this is enabled, this adds a little more overhead to the
310 trace recording, as it needs to add some checks to synchronize
311 recording with swaps. But this does not affect the performance
312 of the overall system. This is enabled by default when the preempt
313 or irq latency tracers are enabled, as those need to swap as well
314 and already adds the overhead (plus a lot more).
316 config TRACE_BRANCH_PROFILING
318 select GENERIC_TRACER
321 prompt "Branch Profiling"
322 default BRANCH_PROFILE_NONE
324 The branch profiling is a software profiler. It will add hooks
325 into the C conditionals to test which path a branch takes.
327 The likely/unlikely profiler only looks at the conditions that
328 are annotated with a likely or unlikely macro.
330 The "all branch" profiler will profile every if-statement in the
331 kernel. This profiler will also enable the likely/unlikely
334 Either of the above profilers adds a bit of overhead to the system.
335 If unsure, choose "No branch profiling".
337 config BRANCH_PROFILE_NONE
338 bool "No branch profiling"
340 No branch profiling. Branch profiling adds a bit of overhead.
341 Only enable it if you want to analyse the branching behavior.
342 Otherwise keep it disabled.
344 config PROFILE_ANNOTATED_BRANCHES
345 bool "Trace likely/unlikely profiler"
346 select TRACE_BRANCH_PROFILING
348 This tracer profiles all likely and unlikely macros
349 in the kernel. It will display the results in:
351 /sys/kernel/debug/tracing/trace_stat/branch_annotated
353 Note: this will add a significant overhead; only turn this
354 on if you need to profile the system's use of these macros.
356 config PROFILE_ALL_BRANCHES
357 bool "Profile all if conditionals"
358 select TRACE_BRANCH_PROFILING
360 This tracer profiles all branch conditions. Every if ()
361 taken in the kernel is recorded whether it hit or miss.
362 The results will be displayed in:
364 /sys/kernel/debug/tracing/trace_stat/branch_all
366 This option also enables the likely/unlikely profiler.
368 This configuration, when enabled, will impose a great overhead
369 on the system. This should only be enabled when the system
370 is to be analyzed in much detail.
373 config TRACING_BRANCHES
376 Selected by tracers that will trace the likely and unlikely
377 conditions. This prevents the tracers themselves from being
378 profiled. Profiling the tracing infrastructure can only happen
379 when the likelys and unlikelys are not being traced.
382 bool "Trace likely/unlikely instances"
383 depends on TRACE_BRANCH_PROFILING
384 select TRACING_BRANCHES
386 This traces the events of likely and unlikely condition
387 calls in the kernel. The difference between this and the
388 "Trace likely/unlikely profiler" is that this is not a
389 histogram of the callers, but actually places the calling
390 events into a running trace buffer to see when and where the
391 events happened, as well as their results.
396 bool "Trace max stack"
397 depends on HAVE_FUNCTION_TRACER
398 select FUNCTION_TRACER
402 This special tracer records the maximum stack footprint of the
403 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
405 This tracer works by hooking into every function call that the
406 kernel executes, and keeping a maximum stack depth value and
407 stack-trace saved. If this is configured with DYNAMIC_FTRACE
408 then it will not have any overhead while the stack tracer
411 To enable the stack tracer on bootup, pass in 'stacktrace'
412 on the kernel command line.
414 The stack tracer can also be enabled or disabled via the
415 sysctl kernel.stack_tracer_enabled
419 config BLK_DEV_IO_TRACE
420 bool "Support for tracing block IO actions"
426 select GENERIC_TRACER
429 Say Y here if you want to be able to trace the block layer actions
430 on a given queue. Tracing allows you to see any traffic happening
431 on a block device queue. For more information (and the userspace
432 support tools needed), fetch the blktrace tools from:
434 git://git.kernel.dk/blktrace.git
436 Tracing also is possible using the ftrace interface, e.g.:
438 echo 1 > /sys/block/sda/sda1/trace/enable
439 echo blk > /sys/kernel/debug/tracing/current_tracer
440 cat /sys/kernel/debug/tracing/trace_pipe
446 depends on HAVE_REGS_AND_STACK_ACCESS_API
447 bool "Enable kprobes-based dynamic events"
452 This allows the user to add tracing events (similar to tracepoints)
453 on the fly via the ftrace interface. See
454 Documentation/trace/kprobetrace.txt for more details.
456 Those events can be inserted wherever kprobes can probe, and record
457 various register and memory values.
459 This option is also required by perf-probe subcommand of perf tools.
460 If you want to use perf tools, this option is strongly recommended.
463 bool "Enable uprobes-based dynamic events"
464 depends on ARCH_SUPPORTS_UPROBES
466 depends on PERF_EVENTS
472 This allows the user to add tracing events on top of userspace
473 dynamic events (similar to tracepoints) on the fly via the trace
474 events interface. Those events can be inserted wherever uprobes
475 can probe, and record various registers.
476 This option is required if you plan to use perf-probe subcommand
477 of perf tools on user space applications.
480 depends on BPF_SYSCALL
481 depends on (KPROBE_EVENTS || UPROBE_EVENTS) && PERF_EVENTS
485 This allows the user to attach BPF programs to kprobe events.
490 config DYNAMIC_FTRACE
491 bool "enable/disable function tracing dynamically"
492 depends on FUNCTION_TRACER
493 depends on HAVE_DYNAMIC_FTRACE
496 This option will modify all the calls to function tracing
497 dynamically (will patch them out of the binary image and
498 replace them with a No-Op instruction) on boot up. During
499 compile time, a table is made of all the locations that ftrace
500 can function trace, and this table is linked into the kernel
501 image. When this is enabled, functions can be individually
502 enabled, and the functions not enabled will not affect
503 performance of the system.
505 See the files in /sys/kernel/debug/tracing:
506 available_filter_functions
510 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
511 otherwise has native performance as long as no tracing is active.
513 config DYNAMIC_FTRACE_WITH_REGS
515 depends on DYNAMIC_FTRACE
516 depends on HAVE_DYNAMIC_FTRACE_WITH_REGS
518 config FUNCTION_PROFILER
519 bool "Kernel function profiler"
520 depends on FUNCTION_TRACER
523 This option enables the kernel function profiler. A file is created
524 in debugfs called function_profile_enabled which defaults to zero.
525 When a 1 is echoed into this file profiling begins, and when a
526 zero is entered, profiling stops. A "functions" file is created in
527 the trace_stats directory; this file shows the list of functions that
528 have been hit and their counters.
532 config FTRACE_MCOUNT_RECORD
534 depends on DYNAMIC_FTRACE
535 depends on HAVE_FTRACE_MCOUNT_RECORD
537 config FTRACE_SELFTEST
540 config FTRACE_STARTUP_TEST
541 bool "Perform a startup test on ftrace"
542 depends on GENERIC_TRACER
543 select FTRACE_SELFTEST
545 This option performs a series of startup tests on ftrace. On bootup
546 a series of tests are made to verify that the tracer is
547 functioning properly. It will do tests on all the configured
550 config EVENT_TRACE_TEST_SYSCALLS
551 bool "Run selftest on syscall events"
552 depends on FTRACE_STARTUP_TEST
554 This option will also enable testing every syscall event.
555 It only enables the event and disables it and runs various loads
556 with the event enabled. This adds a bit more time for kernel boot
557 up since it runs this on every system call defined.
559 TBD - enable a way to actually call the syscalls as we test their
563 bool "Memory mapped IO tracing"
564 depends on HAVE_MMIOTRACE_SUPPORT && PCI
565 select GENERIC_TRACER
567 Mmiotrace traces Memory Mapped I/O access and is meant for
568 debugging and reverse engineering. It is called from the ioremap
569 implementation and works via page faults. Tracing is disabled by
570 default and can be enabled at run-time.
572 See Documentation/trace/mmiotrace.txt.
573 If you are not helping to develop drivers, say N.
577 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
579 tracing_map is a special-purpose lock-free map for tracing,
580 separated out as a stand-alone facility in order to allow it
581 to be shared between multiple tracers. It isn't meant to be
582 generally used outside of that context, and is normally
583 selected by tracers that use it.
586 bool "Histogram triggers"
587 depends on ARCH_HAVE_NMI_SAFE_CMPXCHG
592 Hist triggers allow one or more arbitrary trace event fields
593 to be aggregated into hash tables and dumped to stdout by
594 reading a debugfs/tracefs file. They're useful for
595 gathering quick and dirty (though precise) summaries of
596 event activity as an initial guide for further investigation
597 using more advanced tools.
599 See Documentation/trace/events.txt.
602 config MMIOTRACE_TEST
603 tristate "Test module for mmiotrace"
604 depends on MMIOTRACE && m
606 This is a dumb module for testing mmiotrace. It is very dangerous
607 as it will write garbage to IO memory starting at a given address.
608 However, it should be safe to use on e.g. unused portion of VRAM.
610 Say N, unless you absolutely know what you are doing.
612 config TRACEPOINT_BENCHMARK
613 bool "Add tracepoint that benchmarks tracepoints"
615 This option creates the tracepoint "benchmark:benchmark_event".
616 When the tracepoint is enabled, it kicks off a kernel thread that
617 goes into an infinite loop (calling cond_sched() to let other tasks
618 run), and calls the tracepoint. Each iteration will record the time
619 it took to write to the tracepoint and the next iteration that
620 data will be passed to the tracepoint itself. That is, the tracepoint
621 will report the time it took to do the previous tracepoint.
622 The string written to the tracepoint is a static string of 128 bytes
623 to keep the time the same. The initial string is simply a write of
624 "START". The second string records the cold cache time of the first
625 write which is not added to the rest of the calculations.
627 As it is a tight loop, it benchmarks as hot cache. That's fine because
628 we care most about hot paths that are probably in cache already.
630 An example of the output:
633 first=3672 [COLD CACHED]
634 last=632 first=3672 max=632 min=632 avg=316 std=446 std^2=199712
635 last=278 first=3672 max=632 min=278 avg=303 std=316 std^2=100337
636 last=277 first=3672 max=632 min=277 avg=296 std=258 std^2=67064
637 last=273 first=3672 max=632 min=273 avg=292 std=224 std^2=50411
638 last=273 first=3672 max=632 min=273 avg=288 std=200 std^2=40389
639 last=281 first=3672 max=632 min=273 avg=287 std=183 std^2=33666
642 config RING_BUFFER_BENCHMARK
643 tristate "Ring buffer benchmark stress tester"
644 depends on RING_BUFFER
646 This option creates a test to stress the ring buffer and benchmark it.
647 It creates its own ring buffer such that it will not interfere with
648 any other users of the ring buffer (such as ftrace). It then creates
649 a producer and consumer that will run for 10 seconds and sleep for
650 10 seconds. Each interval it will print out the number of events
651 it recorded and give a rough estimate of how long each iteration took.
653 It does not disable interrupts or raise its priority, so it may be
654 affected by processes that are running.
658 config RING_BUFFER_STARTUP_TEST
659 bool "Ring buffer startup self test"
660 depends on RING_BUFFER
662 Run a simple self test on the ring buffer on boot up. Late in the
663 kernel boot sequence, the test will start that kicks off
664 a thread per cpu. Each thread will write various size events
665 into the ring buffer. Another thread is created to send IPIs
666 to each of the threads, where the IPI handler will also write
667 to the ring buffer, to test/stress the nesting ability.
668 If any anomalies are discovered, a warning will be displayed
669 and all ring buffers will be disabled.
671 The test runs for 10 seconds. This will slow your boot time
672 by at least 10 more seconds.
674 At the end of the test, statics and more checks are done.
675 It will output the stats of each per cpu buffer. What
676 was written, the sizes, what was read, what was lost, and
677 other similar details.
681 config TRACE_EVAL_MAP_FILE
682 bool "Show eval mappings for trace events"
685 The "print fmt" of the trace events will show the enum/sizeof names
686 instead of their values. This can cause problems for user space tools
687 that use this string to parse the raw data as user space does not know
688 how to convert the string to its value.
690 To fix this, there's a special macro in the kernel that can be used
691 to convert an enum/sizeof into its value. If this macro is used, then
692 the print fmt strings will be converted to their values.
694 If something does not get converted properly, this option can be
695 used to show what enums/sizeof the kernel tried to convert.
697 This option is for debugging the conversions. A file is created
698 in the tracing directory called "eval_map" that will show the
699 names matched with their values and what trace event system they
702 Normally, the mapping of the strings to values will be freed after
703 boot up or module load. With this option, they will not be freed, as
704 they are needed for the "eval_map" file. Enabling this option will
705 increase the memory footprint of the running kernel.
709 config TRACING_EVENTS_GPIO
710 bool "Trace gpio events"
714 Enable tracing events for gpio subsystem
718 endif # TRACING_SUPPORT