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 CONTEXT_SWITCH_TRACER
84 config RING_BUFFER_ALLOW_SWAP
87 Allow the use of ring_buffer_swap_cpu.
88 Adds a very slight overhead to tracing when enabled.
90 # All tracer options should select GENERIC_TRACER. For those options that are
91 # enabled by all tracers (context switch and event tracer) they select TRACING.
92 # This allows those options to appear when no other tracer is selected. But the
93 # options do not appear when something else selects it. We need the two options
94 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
95 # hiding of the automatic options.
101 select STACKTRACE if STACKTRACE_SUPPORT
109 config GENERIC_TRACER
114 # Minimum requirements an architecture has to meet for us to
115 # be able to offer generic tracing facilities:
117 config TRACING_SUPPORT
119 # PPC32 has no irqflags tracing support, but it can use most of the
120 # tracers anyway, they were tested to build and work. Note that new
121 # exceptions to this list aren't welcomed, better implement the
122 # irqflags tracing for your architecture.
123 depends on TRACE_IRQFLAGS_SUPPORT || PPC32
124 depends on STACKTRACE_SUPPORT
131 default y if DEBUG_KERNEL
133 Enable the kernel tracing infrastructure.
137 config FUNCTION_TRACER
138 bool "Kernel Function Tracer"
139 depends on HAVE_FUNCTION_TRACER
141 select GENERIC_TRACER
142 select CONTEXT_SWITCH_TRACER
144 Enable the kernel to trace every kernel function. This is done
145 by using a compiler feature to insert a small, 5-byte No-Operation
146 instruction at the beginning of every kernel function, which NOP
147 sequence is then dynamically patched into a tracer call when
148 tracing is enabled by the administrator. If it's runtime disabled
149 (the bootup default), then the overhead of the instructions is very
150 small and not measurable even in micro-benchmarks.
152 config FUNCTION_GRAPH_TRACER
153 bool "Kernel Function Graph Tracer"
154 depends on HAVE_FUNCTION_GRAPH_TRACER
155 depends on FUNCTION_TRACER
156 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
159 Enable the kernel to trace a function at both its return
161 Its first purpose is to trace the duration of functions and
162 draw a call graph for each thread with some information like
163 the return value. This is done by setting the current return
164 address on the current task structure into a stack of calls.
167 config IRQSOFF_TRACER
168 bool "Interrupts-off Latency Tracer"
170 depends on TRACE_IRQFLAGS_SUPPORT
171 depends on !ARCH_USES_GETTIMEOFFSET
172 select TRACE_IRQFLAGS
173 select GENERIC_TRACER
174 select TRACER_MAX_TRACE
175 select RING_BUFFER_ALLOW_SWAP
177 This option measures the time spent in irqs-off critical
178 sections, with microsecond accuracy.
180 The default measurement method is a maximum search, which is
181 disabled by default and can be runtime (re-)started
184 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
186 (Note that kernel size and overhead increase with this option
187 enabled. This option and the preempt-off timing option can be
188 used together or separately.)
190 config PREEMPT_TRACER
191 bool "Preemption-off Latency Tracer"
193 depends on !ARCH_USES_GETTIMEOFFSET
195 select GENERIC_TRACER
196 select TRACER_MAX_TRACE
197 select RING_BUFFER_ALLOW_SWAP
199 This option measures the time spent in preemption-off critical
200 sections, with microsecond accuracy.
202 The default measurement method is a maximum search, which is
203 disabled by default and can be runtime (re-)started
206 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
208 (Note that kernel size and overhead increase with this option
209 enabled. This option and the irqs-off timing option can be
210 used together or separately.)
213 bool "Scheduling Latency Tracer"
214 select GENERIC_TRACER
215 select CONTEXT_SWITCH_TRACER
216 select TRACER_MAX_TRACE
218 This tracer tracks the latency of the highest priority task
219 to be scheduled in, starting from the point it has woken up.
221 config ENABLE_DEFAULT_TRACERS
222 bool "Trace process context switches and events"
223 depends on !GENERIC_TRACER
226 This tracer hooks to various trace points in the kernel,
227 allowing the user to pick and choose which trace point they
228 want to trace. It also includes the sched_switch tracer plugin.
230 config FTRACE_SYSCALLS
231 bool "Trace syscalls"
232 depends on HAVE_SYSCALL_TRACEPOINTS
233 select GENERIC_TRACER
236 Basic tracer to catch the syscall entry and exit events.
238 config TRACE_BRANCH_PROFILING
240 select GENERIC_TRACER
243 prompt "Branch Profiling"
244 default BRANCH_PROFILE_NONE
246 The branch profiling is a software profiler. It will add hooks
247 into the C conditionals to test which path a branch takes.
249 The likely/unlikely profiler only looks at the conditions that
250 are annotated with a likely or unlikely macro.
252 The "all branch" profiler will profile every if-statement in the
253 kernel. This profiler will also enable the likely/unlikely
256 Either of the above profilers adds a bit of overhead to the system.
257 If unsure, choose "No branch profiling".
259 config BRANCH_PROFILE_NONE
260 bool "No branch profiling"
262 No branch profiling. Branch profiling adds a bit of overhead.
263 Only enable it if you want to analyse the branching behavior.
264 Otherwise keep it disabled.
266 config PROFILE_ANNOTATED_BRANCHES
267 bool "Trace likely/unlikely profiler"
268 select TRACE_BRANCH_PROFILING
270 This tracer profiles all likely and unlikely macros
271 in the kernel. It will display the results in:
273 /sys/kernel/debug/tracing/trace_stat/branch_annotated
275 Note: this will add a significant overhead; only turn this
276 on if you need to profile the system's use of these macros.
278 config PROFILE_ALL_BRANCHES
279 bool "Profile all if conditionals"
280 select TRACE_BRANCH_PROFILING
282 This tracer profiles all branch conditions. Every if ()
283 taken in the kernel is recorded whether it hit or miss.
284 The results will be displayed in:
286 /sys/kernel/debug/tracing/trace_stat/branch_all
288 This option also enables the likely/unlikely profiler.
290 This configuration, when enabled, will impose a great overhead
291 on the system. This should only be enabled when the system
292 is to be analyzed in much detail.
295 config TRACING_BRANCHES
298 Selected by tracers that will trace the likely and unlikely
299 conditions. This prevents the tracers themselves from being
300 profiled. Profiling the tracing infrastructure can only happen
301 when the likelys and unlikelys are not being traced.
304 bool "Trace likely/unlikely instances"
305 depends on TRACE_BRANCH_PROFILING
306 select TRACING_BRANCHES
308 This traces the events of likely and unlikely condition
309 calls in the kernel. The difference between this and the
310 "Trace likely/unlikely profiler" is that this is not a
311 histogram of the callers, but actually places the calling
312 events into a running trace buffer to see when and where the
313 events happened, as well as their results.
318 bool "Trace max stack"
319 depends on HAVE_FUNCTION_TRACER
320 select FUNCTION_TRACER
324 This special tracer records the maximum stack footprint of the
325 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
327 This tracer works by hooking into every function call that the
328 kernel executes, and keeping a maximum stack depth value and
329 stack-trace saved. If this is configured with DYNAMIC_FTRACE
330 then it will not have any overhead while the stack tracer
333 To enable the stack tracer on bootup, pass in 'stacktrace'
334 on the kernel command line.
336 The stack tracer can also be enabled or disabled via the
337 sysctl kernel.stack_tracer_enabled
341 config BLK_DEV_IO_TRACE
342 bool "Support for tracing block IO actions"
348 select GENERIC_TRACER
351 Say Y here if you want to be able to trace the block layer actions
352 on a given queue. Tracing allows you to see any traffic happening
353 on a block device queue. For more information (and the userspace
354 support tools needed), fetch the blktrace tools from:
356 git://git.kernel.dk/blktrace.git
358 Tracing also is possible using the ftrace interface, e.g.:
360 echo 1 > /sys/block/sda/sda1/trace/enable
361 echo blk > /sys/kernel/debug/tracing/current_tracer
362 cat /sys/kernel/debug/tracing/trace_pipe
368 depends on HAVE_REGS_AND_STACK_ACCESS_API
369 bool "Enable kprobes-based dynamic events"
374 This allows the user to add tracing events (similar to tracepoints)
375 on the fly via the ftrace interface. See
376 Documentation/trace/kprobetrace.txt for more details.
378 Those events can be inserted wherever kprobes can probe, and record
379 various register and memory values.
381 This option is also required by perf-probe subcommand of perf tools.
382 If you want to use perf tools, this option is strongly recommended.
385 bool "Enable uprobes-based dynamic events"
386 depends on ARCH_SUPPORTS_UPROBES
393 This allows the user to add tracing events on top of userspace
394 dynamic events (similar to tracepoints) on the fly via the trace
395 events interface. Those events can be inserted wherever uprobes
396 can probe, and record various registers.
397 This option is required if you plan to use perf-probe subcommand
398 of perf tools on user space applications.
403 config DYNAMIC_FTRACE
404 bool "enable/disable ftrace tracepoints dynamically"
405 depends on FUNCTION_TRACER
406 depends on HAVE_DYNAMIC_FTRACE
409 This option will modify all the calls to ftrace dynamically
410 (will patch them out of the binary image and replace them
411 with a No-Op instruction) as they are called. A table is
412 created to dynamically enable them again.
414 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
415 otherwise has native performance as long as no tracing is active.
417 The changes to the code are done by a kernel thread that
418 wakes up once a second and checks to see if any ftrace calls
419 were made. If so, it runs stop_machine (stops all CPUS)
420 and modifies the code to jump over the call to ftrace.
422 config FUNCTION_PROFILER
423 bool "Kernel function profiler"
424 depends on FUNCTION_TRACER
427 This option enables the kernel function profiler. A file is created
428 in debugfs called function_profile_enabled which defaults to zero.
429 When a 1 is echoed into this file profiling begins, and when a
430 zero is entered, profiling stops. A "functions" file is created in
431 the trace_stats directory; this file shows the list of functions that
432 have been hit and their counters.
436 config FTRACE_MCOUNT_RECORD
438 depends on DYNAMIC_FTRACE
439 depends on HAVE_FTRACE_MCOUNT_RECORD
441 config FTRACE_SELFTEST
444 config FTRACE_STARTUP_TEST
445 bool "Perform a startup test on ftrace"
446 depends on GENERIC_TRACER
447 select FTRACE_SELFTEST
449 This option performs a series of startup tests on ftrace. On bootup
450 a series of tests are made to verify that the tracer is
451 functioning properly. It will do tests on all the configured
454 config EVENT_TRACE_TEST_SYSCALLS
455 bool "Run selftest on syscall events"
456 depends on FTRACE_STARTUP_TEST
458 This option will also enable testing every syscall event.
459 It only enables the event and disables it and runs various loads
460 with the event enabled. This adds a bit more time for kernel boot
461 up since it runs this on every system call defined.
463 TBD - enable a way to actually call the syscalls as we test their
467 bool "Memory mapped IO tracing"
468 depends on HAVE_MMIOTRACE_SUPPORT && PCI
469 select GENERIC_TRACER
471 Mmiotrace traces Memory Mapped I/O access and is meant for
472 debugging and reverse engineering. It is called from the ioremap
473 implementation and works via page faults. Tracing is disabled by
474 default and can be enabled at run-time.
476 See Documentation/trace/mmiotrace.txt.
477 If you are not helping to develop drivers, say N.
479 config MMIOTRACE_TEST
480 tristate "Test module for mmiotrace"
481 depends on MMIOTRACE && m
483 This is a dumb module for testing mmiotrace. It is very dangerous
484 as it will write garbage to IO memory starting at a given address.
485 However, it should be safe to use on e.g. unused portion of VRAM.
487 Say N, unless you absolutely know what you are doing.
489 config RING_BUFFER_BENCHMARK
490 tristate "Ring buffer benchmark stress tester"
491 depends on RING_BUFFER
493 This option creates a test to stress the ring buffer and benchmark it.
494 It creates its own ring buffer such that it will not interfere with
495 any other users of the ring buffer (such as ftrace). It then creates
496 a producer and consumer that will run for 10 seconds and sleep for
497 10 seconds. Each interval it will print out the number of events
498 it recorded and give a rough estimate of how long each iteration took.
500 It does not disable interrupts or raise its priority, so it may be
501 affected by processes that are running.
507 endif # TRACING_SUPPORT