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 depends on TRACE_IRQFLAGS_SUPPORT
67 depends on STACKTRACE_SUPPORT
74 config FUNCTION_TRACER
75 bool "Kernel Function Tracer"
76 depends on HAVE_FUNCTION_TRACER
80 select CONTEXT_SWITCH_TRACER
82 Enable the kernel to trace every kernel function. This is done
83 by using a compiler feature to insert a small, 5-byte No-Operation
84 instruction to the beginning of every kernel function, which NOP
85 sequence is then dynamically patched into a tracer call when
86 tracing is enabled by the administrator. If it's runtime disabled
87 (the bootup default), then the overhead of the instructions is very
88 small and not measurable even in micro-benchmarks.
90 config FUNCTION_GRAPH_TRACER
91 bool "Kernel Function Graph Tracer"
92 depends on HAVE_FUNCTION_GRAPH_TRACER
93 depends on FUNCTION_TRACER
96 Enable the kernel to trace a function at both its return
98 It's first purpose is to trace the duration of functions and
99 draw a call graph for each thread with some informations like
101 This is done by setting the current return address on the current
102 task structure into a stack of calls.
104 config IRQSOFF_TRACER
105 bool "Interrupts-off Latency Tracer"
107 depends on TRACE_IRQFLAGS_SUPPORT
108 depends on GENERIC_TIME
109 select TRACE_IRQFLAGS
111 select TRACER_MAX_TRACE
113 This option measures the time spent in irqs-off critical
114 sections, with microsecond accuracy.
116 The default measurement method is a maximum search, which is
117 disabled by default and can be runtime (re-)started
120 echo 0 > /debugfs/tracing/tracing_max_latency
122 (Note that kernel size and overhead increases with this option
123 enabled. This option and the preempt-off timing option can be
124 used together or separately.)
126 config PREEMPT_TRACER
127 bool "Preemption-off Latency Tracer"
129 depends on GENERIC_TIME
132 select TRACER_MAX_TRACE
134 This option measures the time spent in preemption off critical
135 sections, with microsecond accuracy.
137 The default measurement method is a maximum search, which is
138 disabled by default and can be runtime (re-)started
141 echo 0 > /debugfs/tracing/tracing_max_latency
143 (Note that kernel size and overhead increases with this option
144 enabled. This option and the irqs-off timing option can be
145 used together or separately.)
147 config SYSPROF_TRACER
148 bool "Sysprof Tracer"
151 select CONTEXT_SWITCH_TRACER
153 This tracer provides the trace needed by the 'Sysprof' userspace
157 bool "Scheduling Latency Tracer"
159 select CONTEXT_SWITCH_TRACER
160 select TRACER_MAX_TRACE
162 This tracer tracks the latency of the highest priority task
163 to be scheduled in, starting from the point it has woken up.
165 config CONTEXT_SWITCH_TRACER
166 bool "Trace process context switches"
170 This tracer gets called from the context switch and records
171 all switching of tasks.
174 bool "Trace various events in the kernel"
177 This tracer hooks to various trace points in the kernel
178 allowing the user to pick and choose which trace point they
181 config FTRACE_SYSCALLS
182 bool "Trace syscalls"
183 depends on HAVE_FTRACE_SYSCALLS
186 Basic tracer to catch the syscall entry and exit events.
189 bool "Trace boot initcalls"
191 select CONTEXT_SWITCH_TRACER
193 This tracer helps developers to optimize boot times: it records
194 the timings of the initcalls and traces key events and the identity
195 of tasks that can cause boot delays, such as context-switches.
197 Its aim is to be parsed by the /scripts/bootgraph.pl tool to
198 produce pretty graphics about boot inefficiencies, giving a visual
199 representation of the delays during initcalls - but the raw
200 /debug/tracing/trace text output is readable too.
202 You must pass in ftrace=initcall to the kernel command line
203 to enable this on bootup.
205 config TRACE_BRANCH_PROFILING
206 bool "Trace likely/unlikely profiler"
209 This tracer profiles all the the likely and unlikely macros
210 in the kernel. It will display the results in:
212 /debugfs/tracing/profile_annotated_branch
214 Note: this will add a significant overhead, only turn this
215 on if you need to profile the system's use of these macros.
219 config PROFILE_ALL_BRANCHES
220 bool "Profile all if conditionals"
221 depends on TRACE_BRANCH_PROFILING
223 This tracer profiles all branch conditions. Every if ()
224 taken in the kernel is recorded whether it hit or miss.
225 The results will be displayed in:
227 /debugfs/tracing/profile_branch
229 This configuration, when enabled, will impose a great overhead
230 on the system. This should only be enabled when the system
235 config TRACING_BRANCHES
238 Selected by tracers that will trace the likely and unlikely
239 conditions. This prevents the tracers themselves from being
240 profiled. Profiling the tracing infrastructure can only happen
241 when the likelys and unlikelys are not being traced.
244 bool "Trace likely/unlikely instances"
245 depends on TRACE_BRANCH_PROFILING
246 select TRACING_BRANCHES
248 This traces the events of likely and unlikely condition
249 calls in the kernel. The difference between this and the
250 "Trace likely/unlikely profiler" is that this is not a
251 histogram of the callers, but actually places the calling
252 events into a running trace buffer to see when and where the
253 events happened, as well as their results.
258 bool "Trace power consumption behavior"
262 This tracer helps developers to analyze and optimize the kernels
263 power management decisions, specifically the C-state and P-state
268 bool "Trace max stack"
269 depends on HAVE_FUNCTION_TRACER
270 select FUNCTION_TRACER
274 This special tracer records the maximum stack footprint of the
275 kernel and displays it in debugfs/tracing/stack_trace.
277 This tracer works by hooking into every function call that the
278 kernel executes, and keeping a maximum stack depth value and
279 stack-trace saved. If this is configured with DYNAMIC_FTRACE
280 then it will not have any overhead while the stack tracer
283 To enable the stack tracer on bootup, pass in 'stacktrace'
284 on the kernel command line.
286 The stack tracer can also be enabled or disabled via the
287 sysctl kernel.stack_tracer_enabled
291 config HW_BRANCH_TRACER
292 depends on HAVE_HW_BRANCH_TRACER
293 bool "Trace hw branches"
296 This tracer records all branches on the system in a circular
297 buffer giving access to the last N branches for each cpu.
300 bool "Trace SLAB allocations"
303 kmemtrace provides tracing for slab allocator functions, such as
304 kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected
305 data is then fed to the userspace application in order to analyse
306 allocation hotspots, internal fragmentation and so on, making it
307 possible to see how well an allocator performs, as well as debug
308 and profile kernel code.
310 This requires an userspace application to use. See
311 Documentation/vm/kmemtrace.txt for more information.
313 Saying Y will make the kernel somewhat larger and slower. However,
314 if you disable kmemtrace at run-time or boot-time, the performance
315 impact is minimal (depending on the arch the kernel is built for).
319 config WORKQUEUE_TRACER
320 bool "Trace workqueues"
323 The workqueue tracer provides some statistical informations
324 about each cpu workqueue thread such as the number of the
325 works inserted and executed since their creation. It can help
326 to evaluate the amount of work each of them have to perform.
327 For example it can help a developer to decide whether he should
328 choose a per cpu workqueue instead of a singlethreaded one.
330 config BLK_DEV_IO_TRACE
331 bool "Support for tracing block io actions"
340 Say Y here if you want to be able to trace the block layer actions
341 on a given queue. Tracing allows you to see any traffic happening
342 on a block device queue. For more information (and the userspace
343 support tools needed), fetch the blktrace tools from:
345 git://git.kernel.dk/blktrace.git
347 Tracing also is possible using the ftrace interface, e.g.:
349 echo 1 > /sys/block/sda/sda1/trace/enable
350 echo blk > /sys/kernel/debug/tracing/current_tracer
351 cat /sys/kernel/debug/tracing/trace_pipe
355 config DYNAMIC_FTRACE
356 bool "enable/disable ftrace tracepoints dynamically"
357 depends on FUNCTION_TRACER
358 depends on HAVE_DYNAMIC_FTRACE
361 This option will modify all the calls to ftrace dynamically
362 (will patch them out of the binary image and replaces them
363 with a No-Op instruction) as they are called. A table is
364 created to dynamically enable them again.
366 This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise
367 has native performance as long as no tracing is active.
369 The changes to the code are done by a kernel thread that
370 wakes up once a second and checks to see if any ftrace calls
371 were made. If so, it runs stop_machine (stops all CPUS)
372 and modifies the code to jump over the call to ftrace.
374 config FTRACE_MCOUNT_RECORD
376 depends on DYNAMIC_FTRACE
377 depends on HAVE_FTRACE_MCOUNT_RECORD
379 config FTRACE_SELFTEST
382 config FTRACE_STARTUP_TEST
383 bool "Perform a startup test on ftrace"
385 select FTRACE_SELFTEST
387 This option performs a series of startup tests on ftrace. On bootup
388 a series of tests are made to verify that the tracer is
389 functioning properly. It will do tests on all the configured
393 bool "Memory mapped IO tracing"
394 depends on HAVE_MMIOTRACE_SUPPORT && PCI
397 Mmiotrace traces Memory Mapped I/O access and is meant for
398 debugging and reverse engineering. It is called from the ioremap
399 implementation and works via page faults. Tracing is disabled by
400 default and can be enabled at run-time.
402 See Documentation/tracers/mmiotrace.txt.
403 If you are not helping to develop drivers, say N.
405 config MMIOTRACE_TEST
406 tristate "Test module for mmiotrace"
407 depends on MMIOTRACE && m
409 This is a dumb module for testing mmiotrace. It is very dangerous
410 as it will write garbage to IO memory starting at a given address.
411 However, it should be safe to use on e.g. unused portion of VRAM.
413 Say N, unless you absolutely know what you are doing.
417 endif # TRACING_SUPPORT