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 TRACER_MAX_TRACE
43 config FTRACE_NMI_ENTER
45 depends on HAVE_FTRACE_NMI_ENTER
52 select STACKTRACE if STACKTRACE_SUPPORT
57 # Minimum requirements an architecture has to meet for us to
58 # be able to offer generic tracing facilities:
60 config TRACING_SUPPORT
62 depends on TRACE_IRQFLAGS_SUPPORT
63 depends on STACKTRACE_SUPPORT
70 config FUNCTION_TRACER
71 bool "Kernel Function Tracer"
72 depends on HAVE_FUNCTION_TRACER
76 select CONTEXT_SWITCH_TRACER
78 Enable the kernel to trace every kernel function. This is done
79 by using a compiler feature to insert a small, 5-byte No-Operation
80 instruction to the beginning of every kernel function, which NOP
81 sequence is then dynamically patched into a tracer call when
82 tracing is enabled by the administrator. If it's runtime disabled
83 (the bootup default), then the overhead of the instructions is very
84 small and not measurable even in micro-benchmarks.
86 config FUNCTION_GRAPH_TRACER
87 bool "Kernel Function Graph Tracer"
88 depends on HAVE_FUNCTION_GRAPH_TRACER
89 depends on FUNCTION_TRACER
92 Enable the kernel to trace a function at both its return
94 It's first purpose is to trace the duration of functions and
95 draw a call graph for each thread with some informations like
97 This is done by setting the current return address on the current
98 task structure into a stack of calls.
101 bool "Binary printk for tracing"
106 config IRQSOFF_TRACER
107 bool "Interrupts-off Latency Tracer"
109 depends on TRACE_IRQFLAGS_SUPPORT
110 depends on GENERIC_TIME
111 select TRACE_IRQFLAGS
113 select TRACER_MAX_TRACE
115 This option measures the time spent in irqs-off critical
116 sections, with microsecond accuracy.
118 The default measurement method is a maximum search, which is
119 disabled by default and can be runtime (re-)started
122 echo 0 > /debugfs/tracing/tracing_max_latency
124 (Note that kernel size and overhead increases with this option
125 enabled. This option and the preempt-off timing option can be
126 used together or separately.)
128 config PREEMPT_TRACER
129 bool "Preemption-off Latency Tracer"
131 depends on GENERIC_TIME
134 select TRACER_MAX_TRACE
136 This option measures the time spent in preemption off critical
137 sections, with microsecond accuracy.
139 The default measurement method is a maximum search, which is
140 disabled by default and can be runtime (re-)started
143 echo 0 > /debugfs/tracing/tracing_max_latency
145 (Note that kernel size and overhead increases with this option
146 enabled. This option and the irqs-off timing option can be
147 used together or separately.)
149 config SYSPROF_TRACER
150 bool "Sysprof Tracer"
153 select CONTEXT_SWITCH_TRACER
155 This tracer provides the trace needed by the 'Sysprof' userspace
159 bool "Scheduling Latency Tracer"
161 select CONTEXT_SWITCH_TRACER
162 select TRACER_MAX_TRACE
164 This tracer tracks the latency of the highest priority task
165 to be scheduled in, starting from the point it has woken up.
167 config CONTEXT_SWITCH_TRACER
168 bool "Trace process context switches"
172 This tracer gets called from the context switch and records
173 all switching of tasks.
176 bool "Trace various events in the kernel"
179 This tracer hooks to various trace points in the kernel
180 allowing the user to pick and choose which trace point they
184 bool "Trace boot initcalls"
186 select CONTEXT_SWITCH_TRACER
188 This tracer helps developers to optimize boot times: it records
189 the timings of the initcalls and traces key events and the identity
190 of tasks that can cause boot delays, such as context-switches.
192 Its aim is to be parsed by the /scripts/bootgraph.pl tool to
193 produce pretty graphics about boot inefficiencies, giving a visual
194 representation of the delays during initcalls - but the raw
195 /debug/tracing/trace text output is readable too.
197 You must pass in ftrace=initcall to the kernel command line
198 to enable this on bootup.
200 config TRACE_BRANCH_PROFILING
201 bool "Trace likely/unlikely profiler"
204 This tracer profiles all the the likely and unlikely macros
205 in the kernel. It will display the results in:
207 /debugfs/tracing/profile_annotated_branch
209 Note: this will add a significant overhead, only turn this
210 on if you need to profile the system's use of these macros.
214 config PROFILE_ALL_BRANCHES
215 bool "Profile all if conditionals"
216 depends on TRACE_BRANCH_PROFILING
218 This tracer profiles all branch conditions. Every if ()
219 taken in the kernel is recorded whether it hit or miss.
220 The results will be displayed in:
222 /debugfs/tracing/profile_branch
224 This configuration, when enabled, will impose a great overhead
225 on the system. This should only be enabled when the system
230 config TRACING_BRANCHES
233 Selected by tracers that will trace the likely and unlikely
234 conditions. This prevents the tracers themselves from being
235 profiled. Profiling the tracing infrastructure can only happen
236 when the likelys and unlikelys are not being traced.
239 bool "Trace likely/unlikely instances"
240 depends on TRACE_BRANCH_PROFILING
241 select TRACING_BRANCHES
243 This traces the events of likely and unlikely condition
244 calls in the kernel. The difference between this and the
245 "Trace likely/unlikely profiler" is that this is not a
246 histogram of the callers, but actually places the calling
247 events into a running trace buffer to see when and where the
248 events happened, as well as their results.
253 bool "Trace power consumption behavior"
257 This tracer helps developers to analyze and optimize the kernels
258 power management decisions, specifically the C-state and P-state
263 bool "Trace max stack"
264 depends on HAVE_FUNCTION_TRACER
265 select FUNCTION_TRACER
269 This special tracer records the maximum stack footprint of the
270 kernel and displays it in debugfs/tracing/stack_trace.
272 This tracer works by hooking into every function call that the
273 kernel executes, and keeping a maximum stack depth value and
274 stack-trace saved. If this is configured with DYNAMIC_FTRACE
275 then it will not have any overhead while the stack tracer
278 To enable the stack tracer on bootup, pass in 'stacktrace'
279 on the kernel command line.
281 The stack tracer can also be enabled or disabled via the
282 sysctl kernel.stack_tracer_enabled
286 config HW_BRANCH_TRACER
287 depends on HAVE_HW_BRANCH_TRACER
288 bool "Trace hw branches"
291 This tracer records all branches on the system in a circular
292 buffer giving access to the last N branches for each cpu.
295 bool "Trace SLAB allocations"
298 kmemtrace provides tracing for slab allocator functions, such as
299 kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected
300 data is then fed to the userspace application in order to analyse
301 allocation hotspots, internal fragmentation and so on, making it
302 possible to see how well an allocator performs, as well as debug
303 and profile kernel code.
305 This requires an userspace application to use. See
306 Documentation/vm/kmemtrace.txt for more information.
308 Saying Y will make the kernel somewhat larger and slower. However,
309 if you disable kmemtrace at run-time or boot-time, the performance
310 impact is minimal (depending on the arch the kernel is built for).
314 config WORKQUEUE_TRACER
315 bool "Trace workqueues"
318 The workqueue tracer provides some statistical informations
319 about each cpu workqueue thread such as the number of the
320 works inserted and executed since their creation. It can help
321 to evaluate the amount of work each of them have to perform.
322 For example it can help a developer to decide whether he should
323 choose a per cpu workqueue instead of a singlethreaded one.
325 config BLK_DEV_IO_TRACE
326 bool "Support for tracing block io actions"
335 Say Y here if you want to be able to trace the block layer actions
336 on a given queue. Tracing allows you to see any traffic happening
337 on a block device queue. For more information (and the userspace
338 support tools needed), fetch the blktrace tools from:
340 git://git.kernel.dk/blktrace.git
342 Tracing also is possible using the ftrace interface, e.g.:
344 echo 1 > /sys/block/sda/sda1/trace/enable
345 echo blk > /sys/kernel/debug/tracing/current_tracer
346 cat /sys/kernel/debug/tracing/trace_pipe
350 config DYNAMIC_FTRACE
351 bool "enable/disable ftrace tracepoints dynamically"
352 depends on FUNCTION_TRACER
353 depends on HAVE_DYNAMIC_FTRACE
356 This option will modify all the calls to ftrace dynamically
357 (will patch them out of the binary image and replaces them
358 with a No-Op instruction) as they are called. A table is
359 created to dynamically enable them again.
361 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise
362 has native performance as long as no tracing is active.
364 The changes to the code are done by a kernel thread that
365 wakes up once a second and checks to see if any ftrace calls
366 were made. If so, it runs stop_machine (stops all CPUS)
367 and modifies the code to jump over the call to ftrace.
369 config FTRACE_MCOUNT_RECORD
371 depends on DYNAMIC_FTRACE
372 depends on HAVE_FTRACE_MCOUNT_RECORD
374 config FTRACE_SELFTEST
377 config FTRACE_STARTUP_TEST
378 bool "Perform a startup test on ftrace"
380 select FTRACE_SELFTEST
382 This option performs a series of startup tests on ftrace. On bootup
383 a series of tests are made to verify that the tracer is
384 functioning properly. It will do tests on all the configured
388 bool "Memory mapped IO tracing"
389 depends on HAVE_MMIOTRACE_SUPPORT && PCI
392 Mmiotrace traces Memory Mapped I/O access and is meant for
393 debugging and reverse engineering. It is called from the ioremap
394 implementation and works via page faults. Tracing is disabled by
395 default and can be enabled at run-time.
397 See Documentation/tracers/mmiotrace.txt.
398 If you are not helping to develop drivers, say N.
400 config MMIOTRACE_TEST
401 tristate "Test module for mmiotrace"
402 depends on MMIOTRACE && m
404 This is a dumb module for testing mmiotrace. It is very dangerous
405 as it will write garbage to IO memory starting at a given address.
406 However, it should be safe to use on e.g. unused portion of VRAM.
408 Say N, unless you absolutely know what you are doing.
412 endif # TRACING_SUPPORT