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 An arch may pass in a unique value (frame pointer) to both the
31 entering and exiting of a function. On exit, the value is compared
32 and if it does not match, then it will panic the kernel.
34 config HAVE_FUNCTION_TRACE_MCOUNT_TEST
37 See Documentation/trace/ftrace-design.txt
39 config HAVE_DYNAMIC_FTRACE
42 See Documentation/trace/ftrace-design.txt
44 config HAVE_FTRACE_MCOUNT_RECORD
47 See Documentation/trace/ftrace-design.txt
49 config HAVE_HW_BRANCH_TRACER
52 config HAVE_SYSCALL_TRACEPOINTS
55 See Documentation/trace/ftrace-design.txt
57 config TRACER_MAX_TRACE
63 config FTRACE_NMI_ENTER
65 depends on HAVE_FTRACE_NMI_ENTER
69 select CONTEXT_SWITCH_TRACER
72 config CONTEXT_SWITCH_TRACER
75 config RING_BUFFER_ALLOW_SWAP
78 Allow the use of ring_buffer_swap_cpu.
79 Adds a very slight overhead to tracing when enabled.
81 # All tracer options should select GENERIC_TRACER. For those options that are
82 # enabled by all tracers (context switch and event tracer) they select TRACING.
83 # This allows those options to appear when no other tracer is selected. But the
84 # options do not appear when something else selects it. We need the two options
85 # GENERIC_TRACER and TRACING to avoid circular dependencies to accomplish the
86 # hiding of the automatic options.
92 select STACKTRACE if STACKTRACE_SUPPORT
103 # Minimum requirements an architecture has to meet for us to
104 # be able to offer generic tracing facilities:
106 config TRACING_SUPPORT
108 # PPC32 has no irqflags tracing support, but it can use most of the
109 # tracers anyway, they were tested to build and work. Note that new
110 # exceptions to this list aren't welcomed, better implement the
111 # irqflags tracing for your architecture.
112 depends on TRACE_IRQFLAGS_SUPPORT || PPC32
113 depends on STACKTRACE_SUPPORT
120 default y if DEBUG_KERNEL
122 Enable the kernel tracing infrastructure.
126 config FUNCTION_TRACER
127 bool "Kernel Function Tracer"
128 depends on HAVE_FUNCTION_TRACER
131 select GENERIC_TRACER
132 select CONTEXT_SWITCH_TRACER
134 Enable the kernel to trace every kernel function. This is done
135 by using a compiler feature to insert a small, 5-byte No-Operation
136 instruction at the beginning of every kernel function, which NOP
137 sequence is then dynamically patched into a tracer call when
138 tracing is enabled by the administrator. If it's runtime disabled
139 (the bootup default), then the overhead of the instructions is very
140 small and not measurable even in micro-benchmarks.
142 config FUNCTION_GRAPH_TRACER
143 bool "Kernel Function Graph Tracer"
144 depends on HAVE_FUNCTION_GRAPH_TRACER
145 depends on FUNCTION_TRACER
146 depends on !X86_32 || !CC_OPTIMIZE_FOR_SIZE
149 Enable the kernel to trace a function at both its return
151 Its first purpose is to trace the duration of functions and
152 draw a call graph for each thread with some information like
153 the return value. This is done by setting the current return
154 address on the current task structure into a stack of calls.
157 config IRQSOFF_TRACER
158 bool "Interrupts-off Latency Tracer"
160 depends on TRACE_IRQFLAGS_SUPPORT
161 depends on GENERIC_TIME
162 select TRACE_IRQFLAGS
163 select GENERIC_TRACER
164 select TRACER_MAX_TRACE
165 select RING_BUFFER_ALLOW_SWAP
167 This option measures the time spent in irqs-off critical
168 sections, with microsecond accuracy.
170 The default measurement method is a maximum search, which is
171 disabled by default and can be runtime (re-)started
174 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
176 (Note that kernel size and overhead increase with this option
177 enabled. This option and the preempt-off timing option can be
178 used together or separately.)
180 config PREEMPT_TRACER
181 bool "Preemption-off Latency Tracer"
183 depends on GENERIC_TIME
185 select GENERIC_TRACER
186 select TRACER_MAX_TRACE
187 select RING_BUFFER_ALLOW_SWAP
189 This option measures the time spent in preemption-off critical
190 sections, with microsecond accuracy.
192 The default measurement method is a maximum search, which is
193 disabled by default and can be runtime (re-)started
196 echo 0 > /sys/kernel/debug/tracing/tracing_max_latency
198 (Note that kernel size and overhead increase with this option
199 enabled. This option and the irqs-off timing option can be
200 used together or separately.)
202 config SYSPROF_TRACER
203 bool "Sysprof Tracer"
205 select GENERIC_TRACER
206 select CONTEXT_SWITCH_TRACER
208 This tracer provides the trace needed by the 'Sysprof' userspace
212 bool "Scheduling Latency Tracer"
213 select GENERIC_TRACER
214 select CONTEXT_SWITCH_TRACER
215 select TRACER_MAX_TRACE
217 This tracer tracks the latency of the highest priority task
218 to be scheduled in, starting from the point it has woken up.
220 config ENABLE_DEFAULT_TRACERS
221 bool "Trace process context switches and events"
222 depends on !GENERIC_TRACER
225 This tracer hooks to various trace points in the kernel,
226 allowing the user to pick and choose which trace point they
227 want to trace. It also includes the sched_switch tracer plugin.
229 config FTRACE_SYSCALLS
230 bool "Trace syscalls"
231 depends on HAVE_SYSCALL_TRACEPOINTS
232 select GENERIC_TRACER
235 Basic tracer to catch the syscall entry and exit events.
238 bool "Trace boot initcalls"
239 select GENERIC_TRACER
240 select CONTEXT_SWITCH_TRACER
242 This tracer helps developers to optimize boot times: it records
243 the timings of the initcalls and traces key events and the identity
244 of tasks that can cause boot delays, such as context-switches.
246 Its aim is to be parsed by the scripts/bootgraph.pl tool to
247 produce pretty graphics about boot inefficiencies, giving a visual
248 representation of the delays during initcalls - but the raw
249 /debug/tracing/trace text output is readable too.
251 You must pass in initcall_debug and ftrace=initcall to the kernel
252 command line to enable this on bootup.
254 config TRACE_BRANCH_PROFILING
256 select GENERIC_TRACER
259 prompt "Branch Profiling"
260 default BRANCH_PROFILE_NONE
262 The branch profiling is a software profiler. It will add hooks
263 into the C conditionals to test which path a branch takes.
265 The likely/unlikely profiler only looks at the conditions that
266 are annotated with a likely or unlikely macro.
268 The "all branch" profiler will profile every if-statement in the
269 kernel. This profiler will also enable the likely/unlikely
272 Either of the above profilers adds a bit of overhead to the system.
273 If unsure, choose "No branch profiling".
275 config BRANCH_PROFILE_NONE
276 bool "No branch profiling"
278 No branch profiling. Branch profiling adds a bit of overhead.
279 Only enable it if you want to analyse the branching behavior.
280 Otherwise keep it disabled.
282 config PROFILE_ANNOTATED_BRANCHES
283 bool "Trace likely/unlikely profiler"
284 select TRACE_BRANCH_PROFILING
286 This tracer profiles all the the likely and unlikely macros
287 in the kernel. It will display the results in:
289 /sys/kernel/debug/tracing/profile_annotated_branch
291 Note: this will add a significant overhead; only turn this
292 on if you need to profile the system's use of these macros.
294 config PROFILE_ALL_BRANCHES
295 bool "Profile all if conditionals"
296 select TRACE_BRANCH_PROFILING
298 This tracer profiles all branch conditions. Every if ()
299 taken in the kernel is recorded whether it hit or miss.
300 The results will be displayed in:
302 /sys/kernel/debug/tracing/profile_branch
304 This option also enables the likely/unlikely profiler.
306 This configuration, when enabled, will impose a great overhead
307 on the system. This should only be enabled when the system
308 is to be analyzed in much detail.
311 config TRACING_BRANCHES
314 Selected by tracers that will trace the likely and unlikely
315 conditions. This prevents the tracers themselves from being
316 profiled. Profiling the tracing infrastructure can only happen
317 when the likelys and unlikelys are not being traced.
320 bool "Trace likely/unlikely instances"
321 depends on TRACE_BRANCH_PROFILING
322 select TRACING_BRANCHES
324 This traces the events of likely and unlikely condition
325 calls in the kernel. The difference between this and the
326 "Trace likely/unlikely profiler" is that this is not a
327 histogram of the callers, but actually places the calling
328 events into a running trace buffer to see when and where the
329 events happened, as well as their results.
334 bool "Trace power consumption behavior"
336 select GENERIC_TRACER
338 This tracer helps developers to analyze and optimize the kernel's
339 power management decisions, specifically the C-state and P-state
343 bool "Trace read and write access on kernel memory locations"
344 depends on HAVE_HW_BREAKPOINT
347 This tracer helps find read and write operations on any given kernel
348 symbol i.e. /proc/kallsyms.
350 config PROFILE_KSYM_TRACER
351 bool "Profile all kernel memory accesses on 'watched' variables"
352 depends on KSYM_TRACER
354 This tracer profiles kernel accesses on variables watched through the
355 ksym tracer ftrace plugin. Depending upon the hardware, all read
356 and write operations on kernel variables can be monitored for
359 The results will be displayed in:
360 /debugfs/tracing/profile_ksym
365 bool "Trace max stack"
366 depends on HAVE_FUNCTION_TRACER
367 select FUNCTION_TRACER
371 This special tracer records the maximum stack footprint of the
372 kernel and displays it in /sys/kernel/debug/tracing/stack_trace.
374 This tracer works by hooking into every function call that the
375 kernel executes, and keeping a maximum stack depth value and
376 stack-trace saved. If this is configured with DYNAMIC_FTRACE
377 then it will not have any overhead while the stack tracer
380 To enable the stack tracer on bootup, pass in 'stacktrace'
381 on the kernel command line.
383 The stack tracer can also be enabled or disabled via the
384 sysctl kernel.stack_tracer_enabled
388 config HW_BRANCH_TRACER
389 depends on HAVE_HW_BRANCH_TRACER
390 bool "Trace hw branches"
391 select GENERIC_TRACER
393 This tracer records all branches on the system in a circular
394 buffer, giving access to the last N branches for each cpu.
397 bool "Trace SLAB allocations"
398 select GENERIC_TRACER
400 kmemtrace provides tracing for slab allocator functions, such as
401 kmalloc, kfree, kmem_cache_alloc, kmem_cache_free, etc. Collected
402 data is then fed to the userspace application in order to analyse
403 allocation hotspots, internal fragmentation and so on, making it
404 possible to see how well an allocator performs, as well as debug
405 and profile kernel code.
407 This requires an userspace application to use. See
408 Documentation/trace/kmemtrace.txt for more information.
410 Saying Y will make the kernel somewhat larger and slower. However,
411 if you disable kmemtrace at run-time or boot-time, the performance
412 impact is minimal (depending on the arch the kernel is built for).
416 config WORKQUEUE_TRACER
417 bool "Trace workqueues"
418 select GENERIC_TRACER
420 The workqueue tracer provides some statistical information
421 about each cpu workqueue thread such as the number of the
422 works inserted and executed since their creation. It can help
423 to evaluate the amount of work each of them has to perform.
424 For example it can help a developer to decide whether he should
425 choose a per-cpu workqueue instead of a singlethreaded one.
427 config BLK_DEV_IO_TRACE
428 bool "Support for tracing block IO actions"
434 select GENERIC_TRACER
437 Say Y here if you want to be able to trace the block layer actions
438 on a given queue. Tracing allows you to see any traffic happening
439 on a block device queue. For more information (and the userspace
440 support tools needed), fetch the blktrace tools from:
442 git://git.kernel.dk/blktrace.git
444 Tracing also is possible using the ftrace interface, e.g.:
446 echo 1 > /sys/block/sda/sda1/trace/enable
447 echo blk > /sys/kernel/debug/tracing/current_tracer
448 cat /sys/kernel/debug/tracing/trace_pipe
455 bool "Enable kprobes-based dynamic events"
459 This allows the user to add tracing events (similar to tracepoints)
460 on the fly via the ftrace interface. See
461 Documentation/trace/kprobetrace.txt for more details.
463 Those events can be inserted wherever kprobes can probe, and record
464 various register and memory values.
466 This option is also required by perf-probe subcommand of perf tools.
467 If you want to use perf tools, this option is strongly recommended.
469 config DYNAMIC_FTRACE
470 bool "enable/disable ftrace tracepoints dynamically"
471 depends on FUNCTION_TRACER
472 depends on HAVE_DYNAMIC_FTRACE
475 This option will modify all the calls to ftrace dynamically
476 (will patch them out of the binary image and replace them
477 with a No-Op instruction) as they are called. A table is
478 created to dynamically enable them again.
480 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but
481 otherwise has native performance as long as no tracing is active.
483 The changes to the code are done by a kernel thread that
484 wakes up once a second and checks to see if any ftrace calls
485 were made. If so, it runs stop_machine (stops all CPUS)
486 and modifies the code to jump over the call to ftrace.
488 config FUNCTION_PROFILER
489 bool "Kernel function profiler"
490 depends on FUNCTION_TRACER
493 This option enables the kernel function profiler. A file is created
494 in debugfs called function_profile_enabled which defaults to zero.
495 When a 1 is echoed into this file profiling begins, and when a
496 zero is entered, profiling stops. A "functions" file is created in
497 the trace_stats directory; this file shows the list of functions that
498 have been hit and their counters.
502 config FTRACE_MCOUNT_RECORD
504 depends on DYNAMIC_FTRACE
505 depends on HAVE_FTRACE_MCOUNT_RECORD
507 config FTRACE_SELFTEST
510 config FTRACE_STARTUP_TEST
511 bool "Perform a startup test on ftrace"
512 depends on GENERIC_TRACER
513 select FTRACE_SELFTEST
515 This option performs a series of startup tests on ftrace. On bootup
516 a series of tests are made to verify that the tracer is
517 functioning properly. It will do tests on all the configured
520 config EVENT_TRACE_TEST_SYSCALLS
521 bool "Run selftest on syscall events"
522 depends on FTRACE_STARTUP_TEST
524 This option will also enable testing every syscall event.
525 It only enables the event and disables it and runs various loads
526 with the event enabled. This adds a bit more time for kernel boot
527 up since it runs this on every system call defined.
529 TBD - enable a way to actually call the syscalls as we test their
533 bool "Memory mapped IO tracing"
534 depends on HAVE_MMIOTRACE_SUPPORT && PCI
535 select GENERIC_TRACER
537 Mmiotrace traces Memory Mapped I/O access and is meant for
538 debugging and reverse engineering. It is called from the ioremap
539 implementation and works via page faults. Tracing is disabled by
540 default and can be enabled at run-time.
542 See Documentation/trace/mmiotrace.txt.
543 If you are not helping to develop drivers, say N.
545 config MMIOTRACE_TEST
546 tristate "Test module for mmiotrace"
547 depends on MMIOTRACE && m
549 This is a dumb module for testing mmiotrace. It is very dangerous
550 as it will write garbage to IO memory starting at a given address.
551 However, it should be safe to use on e.g. unused portion of VRAM.
553 Say N, unless you absolutely know what you are doing.
555 config RING_BUFFER_BENCHMARK
556 tristate "Ring buffer benchmark stress tester"
557 depends on RING_BUFFER
559 This option creates a test to stress the ring buffer and benchmark it.
560 It creates its own ring buffer such that it will not interfere with
561 any other users of the ring buffer (such as ftrace). It then creates
562 a producer and consumer that will run for 10 seconds and sleep for
563 10 seconds. Each interval it will print out the number of events
564 it recorded and give a rough estimate of how long each iteration took.
566 It does not disable interrupts or raise its priority, so it may be
567 affected by processes that are running.
573 endif # TRACING_SUPPORT