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
58 config FUNCTION_TRACER
59 bool "Kernel Function Tracer"
60 depends on HAVE_FUNCTION_TRACER
61 depends on DEBUG_KERNEL
65 select CONTEXT_SWITCH_TRACER
67 Enable the kernel to trace every kernel function. This is done
68 by using a compiler feature to insert a small, 5-byte No-Operation
69 instruction to the beginning of every kernel function, which NOP
70 sequence is then dynamically patched into a tracer call when
71 tracing is enabled by the administrator. If it's runtime disabled
72 (the bootup default), then the overhead of the instructions is very
73 small and not measurable even in micro-benchmarks.
75 config FUNCTION_GRAPH_TRACER
76 bool "Kernel Function Graph Tracer"
77 depends on HAVE_FUNCTION_GRAPH_TRACER
78 depends on FUNCTION_TRACER
81 Enable the kernel to trace a function at both its return
83 It's first purpose is to trace the duration of functions and
84 draw a call graph for each thread with some informations like
86 This is done by setting the current return address on the current
87 task structure into a stack of calls.
90 bool "Interrupts-off Latency Tracer"
92 depends on TRACE_IRQFLAGS_SUPPORT
93 depends on GENERIC_TIME
94 depends on DEBUG_KERNEL
97 select TRACER_MAX_TRACE
99 This option measures the time spent in irqs-off critical
100 sections, with microsecond accuracy.
102 The default measurement method is a maximum search, which is
103 disabled by default and can be runtime (re-)started
106 echo 0 > /debugfs/tracing/tracing_max_latency
108 (Note that kernel size and overhead increases with this option
109 enabled. This option and the preempt-off timing option can be
110 used together or separately.)
112 config PREEMPT_TRACER
113 bool "Preemption-off Latency Tracer"
115 depends on GENERIC_TIME
117 depends on DEBUG_KERNEL
119 select TRACER_MAX_TRACE
121 This option measures the time spent in preemption off critical
122 sections, with microsecond accuracy.
124 The default measurement method is a maximum search, which is
125 disabled by default and can be runtime (re-)started
128 echo 0 > /debugfs/tracing/tracing_max_latency
130 (Note that kernel size and overhead increases with this option
131 enabled. This option and the irqs-off timing option can be
132 used together or separately.)
134 config SYSPROF_TRACER
135 bool "Sysprof Tracer"
138 select CONTEXT_SWITCH_TRACER
140 This tracer provides the trace needed by the 'Sysprof' userspace
144 bool "Scheduling Latency Tracer"
145 depends on DEBUG_KERNEL
147 select CONTEXT_SWITCH_TRACER
148 select TRACER_MAX_TRACE
150 This tracer tracks the latency of the highest priority task
151 to be scheduled in, starting from the point it has woken up.
153 config CONTEXT_SWITCH_TRACER
154 bool "Trace process context switches"
155 depends on DEBUG_KERNEL
159 This tracer gets called from the context switch and records
160 all switching of tasks.
163 bool "Trace various events in the kernel"
164 depends on DEBUG_KERNEL
167 This tracer hooks to various trace points in the kernel
168 allowing the user to pick and choose which trace point they
172 bool "Trace boot initcalls"
173 depends on DEBUG_KERNEL
175 select CONTEXT_SWITCH_TRACER
177 This tracer helps developers to optimize boot times: it records
178 the timings of the initcalls and traces key events and the identity
179 of tasks that can cause boot delays, such as context-switches.
181 Its aim is to be parsed by the /scripts/bootgraph.pl tool to
182 produce pretty graphics about boot inefficiencies, giving a visual
183 representation of the delays during initcalls - but the raw
184 /debug/tracing/trace text output is readable too.
186 You must pass in ftrace=initcall to the kernel command line
187 to enable this on bootup.
189 config TRACE_BRANCH_PROFILING
190 bool "Trace likely/unlikely profiler"
191 depends on DEBUG_KERNEL
194 This tracer profiles all the the likely and unlikely macros
195 in the kernel. It will display the results in:
197 /debugfs/tracing/profile_annotated_branch
199 Note: this will add a significant overhead, only turn this
200 on if you need to profile the system's use of these macros.
204 config PROFILE_ALL_BRANCHES
205 bool "Profile all if conditionals"
206 depends on TRACE_BRANCH_PROFILING
208 This tracer profiles all branch conditions. Every if ()
209 taken in the kernel is recorded whether it hit or miss.
210 The results will be displayed in:
212 /debugfs/tracing/profile_branch
214 This configuration, when enabled, will impose a great overhead
215 on the system. This should only be enabled when the system
220 config TRACING_BRANCHES
223 Selected by tracers that will trace the likely and unlikely
224 conditions. This prevents the tracers themselves from being
225 profiled. Profiling the tracing infrastructure can only happen
226 when the likelys and unlikelys are not being traced.
229 bool "Trace likely/unlikely instances"
230 depends on TRACE_BRANCH_PROFILING
231 select TRACING_BRANCHES
233 This traces the events of likely and unlikely condition
234 calls in the kernel. The difference between this and the
235 "Trace likely/unlikely profiler" is that this is not a
236 histogram of the callers, but actually places the calling
237 events into a running trace buffer to see when and where the
238 events happened, as well as their results.
243 bool "Trace power consumption behavior"
244 depends on DEBUG_KERNEL
248 This tracer helps developers to analyze and optimize the kernels
249 power management decisions, specifically the C-state and P-state
254 bool "Trace max stack"
255 depends on HAVE_FUNCTION_TRACER
256 depends on DEBUG_KERNEL
257 select FUNCTION_TRACER
261 This special tracer records the maximum stack footprint of the
262 kernel and displays it in debugfs/tracing/stack_trace.
264 This tracer works by hooking into every function call that the
265 kernel executes, and keeping a maximum stack depth value and
266 stack-trace saved. If this is configured with DYNAMIC_FTRACE
267 then it will not have any overhead while the stack tracer
270 To enable the stack tracer on bootup, pass in 'stacktrace'
271 on the kernel command line.
273 The stack tracer can also be enabled or disabled via the
274 sysctl kernel.stack_tracer_enabled
278 config HW_BRANCH_TRACER
279 depends on HAVE_HW_BRANCH_TRACER
280 bool "Trace hw branches"
283 This tracer records all branches on the system in a circular
284 buffer giving access to the last N branches for each cpu.
287 bool "Trace SLAB allocations"
290 kmemtrace provides tracing for slab allocator functions, such as
291 kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected
292 data is then fed to the userspace application in order to analyse
293 allocation hotspots, internal fragmentation and so on, making it
294 possible to see how well an allocator performs, as well as debug
295 and profile kernel code.
297 This requires an userspace application to use. See
298 Documentation/vm/kmemtrace.txt for more information.
300 Saying Y will make the kernel somewhat larger and slower. However,
301 if you disable kmemtrace at run-time or boot-time, the performance
302 impact is minimal (depending on the arch the kernel is built for).
306 config WORKQUEUE_TRACER
307 bool "Trace workqueues"
310 The workqueue tracer provides some statistical informations
311 about each cpu workqueue thread such as the number of the
312 works inserted and executed since their creation. It can help
313 to evaluate the amount of work each of them have to perform.
314 For example it can help a developer to decide whether he should
315 choose a per cpu workqueue instead of a singlethreaded one.
317 config BLK_DEV_IO_TRACE
318 bool "Support for tracing block io actions"
327 Say Y here if you want to be able to trace the block layer actions
328 on a given queue. Tracing allows you to see any traffic happening
329 on a block device queue. For more information (and the userspace
330 support tools needed), fetch the blktrace tools from:
332 git://git.kernel.dk/blktrace.git
334 Tracing also is possible using the ftrace interface, e.g.:
336 echo 1 > /sys/block/sda/sda1/trace/enable
337 echo blk > /sys/kernel/debug/tracing/current_tracer
338 cat /sys/kernel/debug/tracing/trace_pipe
342 config DYNAMIC_FTRACE
343 bool "enable/disable ftrace tracepoints dynamically"
344 depends on FUNCTION_TRACER
345 depends on HAVE_DYNAMIC_FTRACE
346 depends on DEBUG_KERNEL
349 This option will modify all the calls to ftrace dynamically
350 (will patch them out of the binary image and replaces them
351 with a No-Op instruction) as they are called. A table is
352 created to dynamically enable them again.
354 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise
355 has native performance as long as no tracing is active.
357 The changes to the code are done by a kernel thread that
358 wakes up once a second and checks to see if any ftrace calls
359 were made. If so, it runs stop_machine (stops all CPUS)
360 and modifies the code to jump over the call to ftrace.
362 config FTRACE_MCOUNT_RECORD
364 depends on DYNAMIC_FTRACE
365 depends on HAVE_FTRACE_MCOUNT_RECORD
367 config FTRACE_SELFTEST
370 config FTRACE_STARTUP_TEST
371 bool "Perform a startup test on ftrace"
372 depends on TRACING && DEBUG_KERNEL
373 select FTRACE_SELFTEST
375 This option performs a series of startup tests on ftrace. On bootup
376 a series of tests are made to verify that the tracer is
377 functioning properly. It will do tests on all the configured
381 bool "Memory mapped IO tracing"
382 depends on HAVE_MMIOTRACE_SUPPORT && DEBUG_KERNEL && PCI
385 Mmiotrace traces Memory Mapped I/O access and is meant for
386 debugging and reverse engineering. It is called from the ioremap
387 implementation and works via page faults. Tracing is disabled by
388 default and can be enabled at run-time.
390 See Documentation/tracers/mmiotrace.txt.
391 If you are not helping to develop drivers, say N.
393 config MMIOTRACE_TEST
394 tristate "Test module for mmiotrace"
395 depends on MMIOTRACE && m
397 This is a dumb module for testing mmiotrace. It is very dangerous
398 as it will write garbage to IO memory starting at a given address.
399 However, it should be safe to use on e.g. unused portion of VRAM.
401 Say N, unless you absolutely know what you are doing.