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_FUNCTION_GRAPH_FP_TEST
30 See Documentation/trace/ftrace-design.txt
32 config HAVE_FUNCTION_TRACE_MCOUNT_TEST
35 See Documentation/trace/ftrace-design.txt
37 config HAVE_DYNAMIC_FTRACE
40 See Documentation/trace/ftrace-design.txt
42 config HAVE_FTRACE_MCOUNT_RECORD
45 See Documentation/trace/ftrace-design.txt
47 config HAVE_SYSCALL_TRACEPOINTS
50 See Documentation/trace/ftrace-design.txt
55 Arch supports the gcc options -pg with -mfentry
57 config HAVE_C_RECORDMCOUNT
60 C version of recordmcount available?
62 config TRACER_MAX_TRACE
72 config FTRACE_NMI_ENTER
74 depends on HAVE_FTRACE_NMI_ENTER
78 select CONTEXT_SWITCH_TRACER
81 config EVENT_POWER_TRACING_DEPRECATED
82 depends on EVENT_TRACING
83 bool "Deprecated power event trace API, to be removed"
86 Provides old power event types:
87 C-state/idle accounting events:
90 and old cpufreq accounting event:
92 This is for userspace compatibility
93 and will vanish after 5 kernel iterations,
96 config CONTEXT_SWITCH_TRACER
99 config RING_BUFFER_ALLOW_SWAP
102 Allow the use of ring_buffer_swap_cpu.
103 Adds a very slight overhead to tracing when enabled.
105 # All tracer options should select GENERIC_TRACER. For those options that are
106 # enabled by all tracers (context switch and event tracer) they select TRACING.
107 # This allows those options to appear when no other tracer is selected. But the
108 # options do not appear when something else selects it. We need the two options
109 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
110 # hiding of the automatic options.
116 select STACKTRACE if STACKTRACE_SUPPORT
123 config GENERIC_TRACER
128 # Minimum requirements an architecture has to meet for us to
129 # be able to offer generic tracing facilities:
131 config TRACING_SUPPORT
133 # PPC32 has no irqflags tracing support, but it can use most of the
134 # tracers anyway, they were tested to build and work. Note that new
135 # exceptions to this list aren't welcomed, better implement the
136 # irqflags tracing for your architecture.
137 depends on TRACE_IRQFLAGS_SUPPORT || PPC32
138 depends on STACKTRACE_SUPPORT
145 default y if DEBUG_KERNEL
147 Enable the kernel tracing infrastructure.
151 config FUNCTION_TRACER
152 bool "Kernel Function Tracer"
153 depends on HAVE_FUNCTION_TRACER
155 select GENERIC_TRACER
156 select CONTEXT_SWITCH_TRACER
158 Enable the kernel to trace every kernel function. This is done
159 by using a compiler feature to insert a small, 5-byte No-Operation
160 instruction at the beginning of every kernel function, which NOP
161 sequence is then dynamically patched into a tracer call when
162 tracing is enabled by the administrator. If it's runtime disabled
163 (the bootup default), then the overhead of the instructions is very
164 small and not measurable even in micro-benchmarks.
166 config FUNCTION_GRAPH_TRACER
167 bool "Kernel Function Graph Tracer"
168 depends on HAVE_FUNCTION_GRAPH_TRACER
169 depends on FUNCTION_TRACER
170 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
173 Enable the kernel to trace a function at both its return
175 Its first purpose is to trace the duration of functions and
176 draw a call graph for each thread with some information like
177 the return value. This is done by setting the current return
178 address on the current task structure into a stack of calls.
181 config IRQSOFF_TRACER
182 bool "Interrupts-off Latency Tracer"
184 depends on TRACE_IRQFLAGS_SUPPORT
185 depends on !ARCH_USES_GETTIMEOFFSET
186 select TRACE_IRQFLAGS
187 select GENERIC_TRACER
188 select TRACER_MAX_TRACE
189 select RING_BUFFER_ALLOW_SWAP
191 This option measures the time spent in irqs-off critical
192 sections, with microsecond accuracy.
194 The default measurement method is a maximum search, which is
195 disabled by default and can be runtime (re-)started
198 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
200 (Note that kernel size and overhead increase with this option
201 enabled. This option and the preempt-off timing option can be
202 used together or separately.)
204 config PREEMPT_TRACER
205 bool "Preemption-off Latency Tracer"
207 depends on !ARCH_USES_GETTIMEOFFSET
209 select GENERIC_TRACER
210 select TRACER_MAX_TRACE
211 select RING_BUFFER_ALLOW_SWAP
213 This option measures the time spent in preemption-off critical
214 sections, with microsecond accuracy.
216 The default measurement method is a maximum search, which is
217 disabled by default and can be runtime (re-)started
220 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
222 (Note that kernel size and overhead increase with this option
223 enabled. This option and the irqs-off timing option can be
224 used together or separately.)
227 bool "Scheduling Latency Tracer"
228 select GENERIC_TRACER
229 select CONTEXT_SWITCH_TRACER
230 select TRACER_MAX_TRACE
232 This tracer tracks the latency of the highest priority task
233 to be scheduled in, starting from the point it has woken up.
235 config ENABLE_DEFAULT_TRACERS
236 bool "Trace process context switches and events"
237 depends on !GENERIC_TRACER
240 This tracer hooks to various trace points in the kernel,
241 allowing the user to pick and choose which trace point they
242 want to trace. It also includes the sched_switch tracer plugin.
244 config FTRACE_SYSCALLS
245 bool "Trace syscalls"
246 depends on HAVE_SYSCALL_TRACEPOINTS
247 select GENERIC_TRACER
250 Basic tracer to catch the syscall entry and exit events.
252 config TRACE_BRANCH_PROFILING
254 select GENERIC_TRACER
257 prompt "Branch Profiling"
258 default BRANCH_PROFILE_NONE
260 The branch profiling is a software profiler. It will add hooks
261 into the C conditionals to test which path a branch takes.
263 The likely/unlikely profiler only looks at the conditions that
264 are annotated with a likely or unlikely macro.
266 The "all branch" profiler will profile every if-statement in the
267 kernel. This profiler will also enable the likely/unlikely
270 Either of the above profilers adds a bit of overhead to the system.
271 If unsure, choose "No branch profiling".
273 config BRANCH_PROFILE_NONE
274 bool "No branch profiling"
276 No branch profiling. Branch profiling adds a bit of overhead.
277 Only enable it if you want to analyse the branching behavior.
278 Otherwise keep it disabled.
280 config PROFILE_ANNOTATED_BRANCHES
281 bool "Trace likely/unlikely profiler"
282 select TRACE_BRANCH_PROFILING
284 This tracer profiles all likely and unlikely macros
285 in the kernel. It will display the results in:
287 /sys/kernel/debug/tracing/trace_stat/branch_annotated
289 Note: this will add a significant overhead; only turn this
290 on if you need to profile the system's use of these macros.
292 config PROFILE_ALL_BRANCHES
293 bool "Profile all if conditionals"
294 select TRACE_BRANCH_PROFILING
296 This tracer profiles all branch conditions. Every if ()
297 taken in the kernel is recorded whether it hit or miss.
298 The results will be displayed in:
300 /sys/kernel/debug/tracing/trace_stat/branch_all
302 This option also enables the likely/unlikely profiler.
304 This configuration, when enabled, will impose a great overhead
305 on the system. This should only be enabled when the system
306 is to be analyzed in much detail.
309 config TRACING_BRANCHES
312 Selected by tracers that will trace the likely and unlikely
313 conditions. This prevents the tracers themselves from being
314 profiled. Profiling the tracing infrastructure can only happen
315 when the likelys and unlikelys are not being traced.
318 bool "Trace likely/unlikely instances"
319 depends on TRACE_BRANCH_PROFILING
320 select TRACING_BRANCHES
322 This traces the events of likely and unlikely condition
323 calls in the kernel. The difference between this and the
324 "Trace likely/unlikely profiler" is that this is not a
325 histogram of the callers, but actually places the calling
326 events into a running trace buffer to see when and where the
327 events happened, as well as their results.
332 bool "Trace max stack"
333 depends on HAVE_FUNCTION_TRACER
334 select FUNCTION_TRACER
338 This special tracer records the maximum stack footprint of the
339 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
341 This tracer works by hooking into every function call that the
342 kernel executes, and keeping a maximum stack depth value and
343 stack-trace saved. If this is configured with DYNAMIC_FTRACE
344 then it will not have any overhead while the stack tracer
347 To enable the stack tracer on bootup, pass in 'stacktrace'
348 on the kernel command line.
350 The stack tracer can also be enabled or disabled via the
351 sysctl kernel.stack_tracer_enabled
355 config BLK_DEV_IO_TRACE
356 bool "Support for tracing block IO actions"
362 select GENERIC_TRACER
365 Say Y here if you want to be able to trace the block layer actions
366 on a given queue. Tracing allows you to see any traffic happening
367 on a block device queue. For more information (and the userspace
368 support tools needed), fetch the blktrace tools from:
370 git://git.kernel.dk/blktrace.git
372 Tracing also is possible using the ftrace interface, e.g.:
374 echo 1 > /sys/block/sda/sda1/trace/enable
375 echo blk > /sys/kernel/debug/tracing/current_tracer
376 cat /sys/kernel/debug/tracing/trace_pipe
382 depends on HAVE_REGS_AND_STACK_ACCESS_API
383 bool "Enable kprobes-based dynamic events"
388 This allows the user to add tracing events (similar to tracepoints)
389 on the fly via the ftrace interface. See
390 Documentation/trace/kprobetrace.txt for more details.
392 Those events can be inserted wherever kprobes can probe, and record
393 various register and memory values.
395 This option is also required by perf-probe subcommand of perf tools.
396 If you want to use perf tools, this option is strongly recommended.
399 bool "Enable uprobes-based dynamic events"
400 depends on ARCH_SUPPORTS_UPROBES
407 This allows the user to add tracing events on top of userspace
408 dynamic events (similar to tracepoints) on the fly via the trace
409 events interface. Those events can be inserted wherever uprobes
410 can probe, and record various registers.
411 This option is required if you plan to use perf-probe subcommand
412 of perf tools on user space applications.
417 config DYNAMIC_FTRACE
418 bool "enable/disable ftrace tracepoints dynamically"
419 depends on FUNCTION_TRACER
420 depends on HAVE_DYNAMIC_FTRACE
423 This option will modify all the calls to ftrace dynamically
424 (will patch them out of the binary image and replace them
425 with a No-Op instruction) as they are called. A table is
426 created to dynamically enable them again.
428 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
429 otherwise has native performance as long as no tracing is active.
431 The changes to the code are done by a kernel thread that
432 wakes up once a second and checks to see if any ftrace calls
433 were made. If so, it runs stop_machine (stops all CPUS)
434 and modifies the code to jump over the call to ftrace.
436 config FUNCTION_PROFILER
437 bool "Kernel function profiler"
438 depends on FUNCTION_TRACER
441 This option enables the kernel function profiler. A file is created
442 in debugfs called function_profile_enabled which defaults to zero.
443 When a 1 is echoed into this file profiling begins, and when a
444 zero is entered, profiling stops. A "functions" file is created in
445 the trace_stats directory; this file shows the list of functions that
446 have been hit and their counters.
450 config FTRACE_MCOUNT_RECORD
452 depends on DYNAMIC_FTRACE
453 depends on HAVE_FTRACE_MCOUNT_RECORD
455 config FTRACE_SELFTEST
458 config FTRACE_STARTUP_TEST
459 bool "Perform a startup test on ftrace"
460 depends on GENERIC_TRACER
461 select FTRACE_SELFTEST
463 This option performs a series of startup tests on ftrace. On bootup
464 a series of tests are made to verify that the tracer is
465 functioning properly. It will do tests on all the configured
468 config EVENT_TRACE_TEST_SYSCALLS
469 bool "Run selftest on syscall events"
470 depends on FTRACE_STARTUP_TEST
472 This option will also enable testing every syscall event.
473 It only enables the event and disables it and runs various loads
474 with the event enabled. This adds a bit more time for kernel boot
475 up since it runs this on every system call defined.
477 TBD - enable a way to actually call the syscalls as we test their
481 bool "Memory mapped IO tracing"
482 depends on HAVE_MMIOTRACE_SUPPORT && PCI
483 select GENERIC_TRACER
485 Mmiotrace traces Memory Mapped I/O access and is meant for
486 debugging and reverse engineering. It is called from the ioremap
487 implementation and works via page faults. Tracing is disabled by
488 default and can be enabled at run-time.
490 See Documentation/trace/mmiotrace.txt.
491 If you are not helping to develop drivers, say N.
493 config MMIOTRACE_TEST
494 tristate "Test module for mmiotrace"
495 depends on MMIOTRACE && m
497 This is a dumb module for testing mmiotrace. It is very dangerous
498 as it will write garbage to IO memory starting at a given address.
499 However, it should be safe to use on e.g. unused portion of VRAM.
501 Say N, unless you absolutely know what you are doing.
503 config RING_BUFFER_BENCHMARK
504 tristate "Ring buffer benchmark stress tester"
505 depends on RING_BUFFER
507 This option creates a test to stress the ring buffer and benchmark it.
508 It creates its own ring buffer such that it will not interfere with
509 any other users of the ring buffer (such as ftrace). It then creates
510 a producer and consumer that will run for 10 seconds and sleep for
511 10 seconds. Each interval it will print out the number of events
512 it recorded and give a rough estimate of how long each iteration took.
514 It does not disable interrupts or raise its priority, so it may be
515 affected by processes that are running.
521 endif # TRACING_SUPPORT