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 config HAVE_FUNCTION_TRACER
18 config HAVE_FUNCTION_GRAPH_TRACER
21 config HAVE_FUNCTION_TRACE_MCOUNT_TEST
24 This gets selected when the arch tests the function_trace_stop
25 variable at the mcount call site. Otherwise, this variable
26 is tested by the called function.
28 config HAVE_DYNAMIC_FTRACE
31 config HAVE_FTRACE_MCOUNT_RECORD
34 config HAVE_HW_BRANCH_TRACER
37 config HAVE_FTRACE_SYSCALLS
40 config TRACER_MAX_TRACE
46 config FTRACE_NMI_ENTER
48 depends on HAVE_FTRACE_NMI_ENTER
55 select STACKTRACE if STACKTRACE_SUPPORT
61 # Minimum requirements an architecture has to meet for us to
62 # be able to offer generic tracing facilities:
64 config TRACING_SUPPORT
66 # PPC32 has no irqflags tracing support, but it can use most of the
67 # tracers anyway, they were tested to build and work. Note that new
68 # exceptions to this list aren't welcomed, better implement the
69 # irqflags tracing for your architecture.
70 depends on TRACE_IRQFLAGS_SUPPORT || PPC32
71 depends on STACKTRACE_SUPPORT
78 config FUNCTION_TRACER
79 bool "Kernel Function Tracer"
80 depends on HAVE_FUNCTION_TRACER
84 select CONTEXT_SWITCH_TRACER
86 Enable the kernel to trace every kernel function. This is done
87 by using a compiler feature to insert a small, 5-byte No-Operation
88 instruction to the beginning of every kernel function, which NOP
89 sequence is then dynamically patched into a tracer call when
90 tracing is enabled by the administrator. If it's runtime disabled
91 (the bootup default), then the overhead of the instructions is very
92 small and not measurable even in micro-benchmarks.
94 config FUNCTION_GRAPH_TRACER
95 bool "Kernel Function Graph Tracer"
96 depends on HAVE_FUNCTION_GRAPH_TRACER
97 depends on FUNCTION_TRACER
100 Enable the kernel to trace a function at both its return
102 It's first purpose is to trace the duration of functions and
103 draw a call graph for each thread with some informations like
105 This is done by setting the current return address on the current
106 task structure into a stack of calls.
109 config IRQSOFF_TRACER
110 bool "Interrupts-off Latency Tracer"
112 depends on TRACE_IRQFLAGS_SUPPORT
113 depends on GENERIC_TIME
114 select TRACE_IRQFLAGS
116 select TRACER_MAX_TRACE
118 This option measures the time spent in irqs-off critical
119 sections, with microsecond accuracy.
121 The default measurement method is a maximum search, which is
122 disabled by default and can be runtime (re-)started
125 echo 0 > /debugfs/tracing/tracing_max_latency
127 (Note that kernel size and overhead increases with this option
128 enabled. This option and the preempt-off timing option can be
129 used together or separately.)
131 config PREEMPT_TRACER
132 bool "Preemption-off Latency Tracer"
134 depends on GENERIC_TIME
137 select TRACER_MAX_TRACE
139 This option measures the time spent in preemption off critical
140 sections, with microsecond accuracy.
142 The default measurement method is a maximum search, which is
143 disabled by default and can be runtime (re-)started
146 echo 0 > /debugfs/tracing/tracing_max_latency
148 (Note that kernel size and overhead increases with this option
149 enabled. This option and the irqs-off timing option can be
150 used together or separately.)
152 config SYSPROF_TRACER
153 bool "Sysprof Tracer"
156 select CONTEXT_SWITCH_TRACER
158 This tracer provides the trace needed by the 'Sysprof' userspace
162 bool "Scheduling Latency Tracer"
164 select CONTEXT_SWITCH_TRACER
165 select TRACER_MAX_TRACE
167 This tracer tracks the latency of the highest priority task
168 to be scheduled in, starting from the point it has woken up.
170 config CONTEXT_SWITCH_TRACER
171 bool "Trace process context switches"
175 This tracer gets called from the context switch and records
176 all switching of tasks.
179 bool "Trace various events in the kernel"
182 This tracer hooks to various trace points in the kernel
183 allowing the user to pick and choose which trace point they
186 config FTRACE_SYSCALLS
187 bool "Trace syscalls"
188 depends on HAVE_FTRACE_SYSCALLS
192 Basic tracer to catch the syscall entry and exit events.
195 bool "Trace boot initcalls"
197 select CONTEXT_SWITCH_TRACER
199 This tracer helps developers to optimize boot times: it records
200 the timings of the initcalls and traces key events and the identity
201 of tasks that can cause boot delays, such as context-switches.
203 Its aim is to be parsed by the /scripts/bootgraph.pl tool to
204 produce pretty graphics about boot inefficiencies, giving a visual
205 representation of the delays during initcalls - but the raw
206 /debug/tracing/trace text output is readable too.
208 You must pass in ftrace=initcall to the kernel command line
209 to enable this on bootup.
211 config TRACE_BRANCH_PROFILING
212 bool "Trace likely/unlikely profiler"
215 This tracer profiles all the the likely and unlikely macros
216 in the kernel. It will display the results in:
218 /debugfs/tracing/profile_annotated_branch
220 Note: this will add a significant overhead, only turn this
221 on if you need to profile the system's use of these macros.
225 config PROFILE_ALL_BRANCHES
226 bool "Profile all if conditionals"
227 depends on TRACE_BRANCH_PROFILING
229 This tracer profiles all branch conditions. Every if ()
230 taken in the kernel is recorded whether it hit or miss.
231 The results will be displayed in:
233 /debugfs/tracing/profile_branch
235 This configuration, when enabled, will impose a great overhead
236 on the system. This should only be enabled when the system
241 config TRACING_BRANCHES
244 Selected by tracers that will trace the likely and unlikely
245 conditions. This prevents the tracers themselves from being
246 profiled. Profiling the tracing infrastructure can only happen
247 when the likelys and unlikelys are not being traced.
250 bool "Trace likely/unlikely instances"
251 depends on TRACE_BRANCH_PROFILING
252 select TRACING_BRANCHES
254 This traces the events of likely and unlikely condition
255 calls in the kernel. The difference between this and the
256 "Trace likely/unlikely profiler" is that this is not a
257 histogram of the callers, but actually places the calling
258 events into a running trace buffer to see when and where the
259 events happened, as well as their results.
264 bool "Trace power consumption behavior"
268 This tracer helps developers to analyze and optimize the kernels
269 power management decisions, specifically the C-state and P-state
274 bool "Trace max stack"
275 depends on HAVE_FUNCTION_TRACER
276 select FUNCTION_TRACER
280 This special tracer records the maximum stack footprint of the
281 kernel and displays it in debugfs/tracing/stack_trace.
283 This tracer works by hooking into every function call that the
284 kernel executes, and keeping a maximum stack depth value and
285 stack-trace saved. If this is configured with DYNAMIC_FTRACE
286 then it will not have any overhead while the stack tracer
289 To enable the stack tracer on bootup, pass in 'stacktrace'
290 on the kernel command line.
292 The stack tracer can also be enabled or disabled via the
293 sysctl kernel.stack_tracer_enabled
297 config HW_BRANCH_TRACER
298 depends on HAVE_HW_BRANCH_TRACER
299 bool "Trace hw branches"
302 This tracer records all branches on the system in a circular
303 buffer giving access to the last N branches for each cpu.
306 bool "Trace SLAB allocations"
309 kmemtrace provides tracing for slab allocator functions, such as
310 kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected
311 data is then fed to the userspace application in order to analyse
312 allocation hotspots, internal fragmentation and so on, making it
313 possible to see how well an allocator performs, as well as debug
314 and profile kernel code.
316 This requires an userspace application to use. See
317 Documentation/vm/kmemtrace.txt for more information.
319 Saying Y will make the kernel somewhat larger and slower. However,
320 if you disable kmemtrace at run-time or boot-time, the performance
321 impact is minimal (depending on the arch the kernel is built for).
325 config WORKQUEUE_TRACER
326 bool "Trace workqueues"
329 The workqueue tracer provides some statistical informations
330 about each cpu workqueue thread such as the number of the
331 works inserted and executed since their creation. It can help
332 to evaluate the amount of work each of them have to perform.
333 For example it can help a developer to decide whether he should
334 choose a per cpu workqueue instead of a singlethreaded one.
336 config BLK_DEV_IO_TRACE
337 bool "Support for tracing block io actions"
346 Say Y here if you want to be able to trace the block layer actions
347 on a given queue. Tracing allows you to see any traffic happening
348 on a block device queue. For more information (and the userspace
349 support tools needed), fetch the blktrace tools from:
351 git://git.kernel.dk/blktrace.git
353 Tracing also is possible using the ftrace interface, e.g.:
355 echo 1 > /sys/block/sda/sda1/trace/enable
356 echo blk > /sys/kernel/debug/tracing/current_tracer
357 cat /sys/kernel/debug/tracing/trace_pipe
361 config DYNAMIC_FTRACE
362 bool "enable/disable ftrace tracepoints dynamically"
363 depends on FUNCTION_TRACER
364 depends on HAVE_DYNAMIC_FTRACE
367 This option will modify all the calls to ftrace dynamically
368 (will patch them out of the binary image and replaces them
369 with a No-Op instruction) as they are called. A table is
370 created to dynamically enable them again.
372 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise
373 has native performance as long as no tracing is active.
375 The changes to the code are done by a kernel thread that
376 wakes up once a second and checks to see if any ftrace calls
377 were made. If so, it runs stop_machine (stops all CPUS)
378 and modifies the code to jump over the call to ftrace.
380 config FUNCTION_PROFILER
381 bool "Kernel function profiler"
382 depends on FUNCTION_TRACER
385 This option enables the kernel function profiler. A file is created
386 in debugfs called function_profile_enabled which defaults to zero.
387 When a 1 is echoed into this file profiling begins, and when a
388 zero is entered, profiling stops. A file in the trace_stats
389 directory called functions, that show the list of functions that
390 have been hit and their counters.
394 config FTRACE_MCOUNT_RECORD
396 depends on DYNAMIC_FTRACE
397 depends on HAVE_FTRACE_MCOUNT_RECORD
399 config FTRACE_SELFTEST
402 config FTRACE_STARTUP_TEST
403 bool "Perform a startup test on ftrace"
405 select FTRACE_SELFTEST
407 This option performs a series of startup tests on ftrace. On bootup
408 a series of tests are made to verify that the tracer is
409 functioning properly. It will do tests on all the configured
413 bool "Memory mapped IO tracing"
414 depends on HAVE_MMIOTRACE_SUPPORT && PCI
417 Mmiotrace traces Memory Mapped I/O access and is meant for
418 debugging and reverse engineering. It is called from the ioremap
419 implementation and works via page faults. Tracing is disabled by
420 default and can be enabled at run-time.
422 See Documentation/tracers/mmiotrace.txt.
423 If you are not helping to develop drivers, say N.
425 config MMIOTRACE_TEST
426 tristate "Test module for mmiotrace"
427 depends on MMIOTRACE && m
429 This is a dumb module for testing mmiotrace. It is very dangerous
430 as it will write garbage to IO memory starting at a given address.
431 However, it should be safe to use on e.g. unused portion of VRAM.
433 Say N, unless you absolutely know what you are doing.
437 endif # TRACING_SUPPORT