Kconfig 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359
  1. #
  2. # Architectures that offer an FUNCTION_TRACER implementation should
  3. # select HAVE_FUNCTION_TRACER:
  4. #
  5. config USER_STACKTRACE_SUPPORT
  6. bool
  7. config NOP_TRACER
  8. bool
  9. config HAVE_FUNCTION_TRACER
  10. bool
  11. config HAVE_FUNCTION_GRAPH_TRACER
  12. bool
  13. config HAVE_FUNCTION_TRACE_MCOUNT_TEST
  14. bool
  15. help
  16. This gets selected when the arch tests the function_trace_stop
  17. variable at the mcount call site. Otherwise, this variable
  18. is tested by the called function.
  19. config HAVE_DYNAMIC_FTRACE
  20. bool
  21. config HAVE_FTRACE_MCOUNT_RECORD
  22. bool
  23. config HAVE_HW_BRANCH_TRACER
  24. bool
  25. config TRACER_MAX_TRACE
  26. bool
  27. config RING_BUFFER
  28. bool
  29. config TRACING
  30. bool
  31. select DEBUG_FS
  32. select RING_BUFFER
  33. select STACKTRACE if STACKTRACE_SUPPORT
  34. select TRACEPOINTS
  35. select NOP_TRACER
  36. menu "Tracers"
  37. config FUNCTION_TRACER
  38. bool "Kernel Function Tracer"
  39. depends on HAVE_FUNCTION_TRACER
  40. depends on DEBUG_KERNEL
  41. select FRAME_POINTER
  42. select TRACING
  43. select CONTEXT_SWITCH_TRACER
  44. help
  45. Enable the kernel to trace every kernel function. This is done
  46. by using a compiler feature to insert a small, 5-byte No-Operation
  47. instruction to the beginning of every kernel function, which NOP
  48. sequence is then dynamically patched into a tracer call when
  49. tracing is enabled by the administrator. If it's runtime disabled
  50. (the bootup default), then the overhead of the instructions is very
  51. small and not measurable even in micro-benchmarks.
  52. config FUNCTION_GRAPH_TRACER
  53. bool "Kernel Function Graph Tracer"
  54. depends on HAVE_FUNCTION_GRAPH_TRACER
  55. depends on FUNCTION_TRACER
  56. default y
  57. help
  58. Enable the kernel to trace a function at both its return
  59. and its entry.
  60. It's first purpose is to trace the duration of functions and
  61. draw a call graph for each thread with some informations like
  62. the return value.
  63. This is done by setting the current return address on the current
  64. task structure into a stack of calls.
  65. config IRQSOFF_TRACER
  66. bool "Interrupts-off Latency Tracer"
  67. default n
  68. depends on TRACE_IRQFLAGS_SUPPORT
  69. depends on GENERIC_TIME
  70. depends on DEBUG_KERNEL
  71. select TRACE_IRQFLAGS
  72. select TRACING
  73. select TRACER_MAX_TRACE
  74. help
  75. This option measures the time spent in irqs-off critical
  76. sections, with microsecond accuracy.
  77. The default measurement method is a maximum search, which is
  78. disabled by default and can be runtime (re-)started
  79. via:
  80. echo 0 > /debugfs/tracing/tracing_max_latency
  81. (Note that kernel size and overhead increases with this option
  82. enabled. This option and the preempt-off timing option can be
  83. used together or separately.)
  84. config PREEMPT_TRACER
  85. bool "Preemption-off Latency Tracer"
  86. default n
  87. depends on GENERIC_TIME
  88. depends on PREEMPT
  89. depends on DEBUG_KERNEL
  90. select TRACING
  91. select TRACER_MAX_TRACE
  92. help
  93. This option measures the time spent in preemption off critical
  94. sections, with microsecond accuracy.
  95. The default measurement method is a maximum search, which is
  96. disabled by default and can be runtime (re-)started
  97. via:
  98. echo 0 > /debugfs/tracing/tracing_max_latency
  99. (Note that kernel size and overhead increases with this option
  100. enabled. This option and the irqs-off timing option can be
  101. used together or separately.)
  102. config SYSPROF_TRACER
  103. bool "Sysprof Tracer"
  104. depends on X86
  105. select TRACING
  106. help
  107. This tracer provides the trace needed by the 'Sysprof' userspace
  108. tool.
  109. config SCHED_TRACER
  110. bool "Scheduling Latency Tracer"
  111. depends on DEBUG_KERNEL
  112. select TRACING
  113. select CONTEXT_SWITCH_TRACER
  114. select TRACER_MAX_TRACE
  115. help
  116. This tracer tracks the latency of the highest priority task
  117. to be scheduled in, starting from the point it has woken up.
  118. config CONTEXT_SWITCH_TRACER
  119. bool "Trace process context switches"
  120. depends on DEBUG_KERNEL
  121. select TRACING
  122. select MARKERS
  123. help
  124. This tracer gets called from the context switch and records
  125. all switching of tasks.
  126. config BOOT_TRACER
  127. bool "Trace boot initcalls"
  128. depends on DEBUG_KERNEL
  129. select TRACING
  130. select CONTEXT_SWITCH_TRACER
  131. help
  132. This tracer helps developers to optimize boot times: it records
  133. the timings of the initcalls and traces key events and the identity
  134. of tasks that can cause boot delays, such as context-switches.
  135. Its aim is to be parsed by the /scripts/bootgraph.pl tool to
  136. produce pretty graphics about boot inefficiencies, giving a visual
  137. representation of the delays during initcalls - but the raw
  138. /debug/tracing/trace text output is readable too.
  139. You must pass in ftrace=initcall to the kernel command line
  140. to enable this on bootup.
  141. config TRACE_BRANCH_PROFILING
  142. bool "Trace likely/unlikely profiler"
  143. depends on DEBUG_KERNEL
  144. select TRACING
  145. help
  146. This tracer profiles all the the likely and unlikely macros
  147. in the kernel. It will display the results in:
  148. /debugfs/tracing/profile_annotated_branch
  149. Note: this will add a significant overhead, only turn this
  150. on if you need to profile the system's use of these macros.
  151. Say N if unsure.
  152. config PROFILE_ALL_BRANCHES
  153. bool "Profile all if conditionals"
  154. depends on TRACE_BRANCH_PROFILING
  155. help
  156. This tracer profiles all branch conditions. Every if ()
  157. taken in the kernel is recorded whether it hit or miss.
  158. The results will be displayed in:
  159. /debugfs/tracing/profile_branch
  160. This configuration, when enabled, will impose a great overhead
  161. on the system. This should only be enabled when the system
  162. is to be analyzed
  163. Say N if unsure.
  164. config TRACING_BRANCHES
  165. bool
  166. help
  167. Selected by tracers that will trace the likely and unlikely
  168. conditions. This prevents the tracers themselves from being
  169. profiled. Profiling the tracing infrastructure can only happen
  170. when the likelys and unlikelys are not being traced.
  171. config BRANCH_TRACER
  172. bool "Trace likely/unlikely instances"
  173. depends on TRACE_BRANCH_PROFILING
  174. select TRACING_BRANCHES
  175. help
  176. This traces the events of likely and unlikely condition
  177. calls in the kernel. The difference between this and the
  178. "Trace likely/unlikely profiler" is that this is not a
  179. histogram of the callers, but actually places the calling
  180. events into a running trace buffer to see when and where the
  181. events happened, as well as their results.
  182. Say N if unsure.
  183. config POWER_TRACER
  184. bool "Trace power consumption behavior"
  185. depends on DEBUG_KERNEL
  186. depends on X86
  187. select TRACING
  188. help
  189. This tracer helps developers to analyze and optimize the kernels
  190. power management decisions, specifically the C-state and P-state
  191. behavior.
  192. config STACK_TRACER
  193. bool "Trace max stack"
  194. depends on HAVE_FUNCTION_TRACER
  195. depends on DEBUG_KERNEL
  196. select FUNCTION_TRACER
  197. select STACKTRACE
  198. help
  199. This special tracer records the maximum stack footprint of the
  200. kernel and displays it in debugfs/tracing/stack_trace.
  201. This tracer works by hooking into every function call that the
  202. kernel executes, and keeping a maximum stack depth value and
  203. stack-trace saved. If this is configured with DYNAMIC_FTRACE
  204. then it will not have any overhead while the stack tracer
  205. is disabled.
  206. To enable the stack tracer on bootup, pass in 'stacktrace'
  207. on the kernel command line.
  208. The stack tracer can also be enabled or disabled via the
  209. sysctl kernel.stack_tracer_enabled
  210. Say N if unsure.
  211. config HW_BRANCH_TRACER
  212. depends on HAVE_HW_BRANCH_TRACER
  213. bool "Trace hw branches"
  214. select TRACING
  215. help
  216. This tracer records all branches on the system in a circular
  217. buffer giving access to the last N branches for each cpu.
  218. config KMEMTRACE
  219. bool "Trace SLAB allocations"
  220. select TRACING
  221. help
  222. kmemtrace provides tracing for slab allocator functions, such as
  223. kmalloc, kfree, kmem_cache_alloc, kmem_cache_free etc.. Collected
  224. data is then fed to the userspace application in order to analyse
  225. allocation hotspots, internal fragmentation and so on, making it
  226. possible to see how well an allocator performs, as well as debug
  227. and profile kernel code.
  228. This requires an userspace application to use. See
  229. Documentation/vm/kmemtrace.txt for more information.
  230. Saying Y will make the kernel somewhat larger and slower. However,
  231. if you disable kmemtrace at run-time or boot-time, the performance
  232. impact is minimal (depending on the arch the kernel is built for).
  233. If unsure, say N.
  234. config WORKQUEUE_TRACER
  235. bool "Trace workqueues"
  236. select TRACING
  237. help
  238. The workqueue tracer provides some statistical informations
  239. about each cpu workqueue thread such as the number of the
  240. works inserted and executed since their creation. It can help
  241. to evaluate the amount of work each of them have to perform.
  242. For example it can help a developer to decide whether he should
  243. choose a per cpu workqueue instead of a singlethreaded one.
  244. config DYNAMIC_FTRACE
  245. bool "enable/disable ftrace tracepoints dynamically"
  246. depends on FUNCTION_TRACER
  247. depends on HAVE_DYNAMIC_FTRACE
  248. depends on DEBUG_KERNEL
  249. default y
  250. help
  251. This option will modify all the calls to ftrace dynamically
  252. (will patch them out of the binary image and replaces them
  253. with a No-Op instruction) as they are called. A table is
  254. created to dynamically enable them again.
  255. This way a CONFIG_FUNCTION_TRACER kernel is slightly larger, but otherwise
  256. has native performance as long as no tracing is active.
  257. The changes to the code are done by a kernel thread that
  258. wakes up once a second and checks to see if any ftrace calls
  259. were made. If so, it runs stop_machine (stops all CPUS)
  260. and modifies the code to jump over the call to ftrace.
  261. config FTRACE_MCOUNT_RECORD
  262. def_bool y
  263. depends on DYNAMIC_FTRACE
  264. depends on HAVE_FTRACE_MCOUNT_RECORD
  265. config FTRACE_SELFTEST
  266. bool
  267. config FTRACE_STARTUP_TEST
  268. bool "Perform a startup test on ftrace"
  269. depends on TRACING && DEBUG_KERNEL
  270. select FTRACE_SELFTEST
  271. help
  272. This option performs a series of startup tests on ftrace. On bootup
  273. a series of tests are made to verify that the tracer is
  274. functioning properly. It will do tests on all the configured
  275. tracers of ftrace.
  276. config MMIOTRACE
  277. bool "Memory mapped IO tracing"
  278. depends on HAVE_MMIOTRACE_SUPPORT && DEBUG_KERNEL && PCI
  279. select TRACING
  280. help
  281. Mmiotrace traces Memory Mapped I/O access and is meant for
  282. debugging and reverse engineering. It is called from the ioremap
  283. implementation and works via page faults. Tracing is disabled by
  284. default and can be enabled at run-time.
  285. See Documentation/tracers/mmiotrace.txt.
  286. If you are not helping to develop drivers, say N.
  287. config MMIOTRACE_TEST
  288. tristate "Test module for mmiotrace"
  289. depends on MMIOTRACE && m
  290. help
  291. This is a dumb module for testing mmiotrace. It is very dangerous
  292. as it will write garbage to IO memory starting at a given address.
  293. However, it should be safe to use on e.g. unused portion of VRAM.
  294. Say N, unless you absolutely know what you are doing.
  295. endmenu