Kconfig 7.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274
  1. #
  2. # General architecture dependent options
  3. #
  4. config OPROFILE
  5. tristate "OProfile system profiling"
  6. depends on PROFILING
  7. depends on HAVE_OPROFILE
  8. select RING_BUFFER
  9. select RING_BUFFER_ALLOW_SWAP
  10. help
  11. OProfile is a profiling system capable of profiling the
  12. whole system, include the kernel, kernel modules, libraries,
  13. and applications.
  14. If unsure, say N.
  15. config OPROFILE_EVENT_MULTIPLEX
  16. bool "OProfile multiplexing support (EXPERIMENTAL)"
  17. default n
  18. depends on OPROFILE && X86
  19. help
  20. The number of hardware counters is limited. The multiplexing
  21. feature enables OProfile to gather more events than counters
  22. are provided by the hardware. This is realized by switching
  23. between events at an user specified time interval.
  24. If unsure, say N.
  25. config HAVE_OPROFILE
  26. bool
  27. config OPROFILE_NMI_TIMER
  28. def_bool y
  29. depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI
  30. config KPROBES
  31. bool "Kprobes"
  32. depends on MODULES
  33. depends on HAVE_KPROBES
  34. select KALLSYMS
  35. help
  36. Kprobes allows you to trap at almost any kernel address and
  37. execute a callback function. register_kprobe() establishes
  38. a probepoint and specifies the callback. Kprobes is useful
  39. for kernel debugging, non-intrusive instrumentation and testing.
  40. If in doubt, say "N".
  41. config JUMP_LABEL
  42. bool "Optimize very unlikely/likely branches"
  43. depends on HAVE_ARCH_JUMP_LABEL
  44. help
  45. This option enables a transparent branch optimization that
  46. makes certain almost-always-true or almost-always-false branch
  47. conditions even cheaper to execute within the kernel.
  48. Certain performance-sensitive kernel code, such as trace points,
  49. scheduler functionality, networking code and KVM have such
  50. branches and include support for this optimization technique.
  51. If it is detected that the compiler has support for "asm goto",
  52. the kernel will compile such branches with just a nop
  53. instruction. When the condition flag is toggled to true, the
  54. nop will be converted to a jump instruction to execute the
  55. conditional block of instructions.
  56. This technique lowers overhead and stress on the branch prediction
  57. of the processor and generally makes the kernel faster. The update
  58. of the condition is slower, but those are always very rare.
  59. ( On 32-bit x86, the necessary options added to the compiler
  60. flags may increase the size of the kernel slightly. )
  61. config OPTPROBES
  62. def_bool y
  63. depends on KPROBES && HAVE_OPTPROBES
  64. depends on !PREEMPT
  65. config UPROBES
  66. bool "Transparent user-space probes (EXPERIMENTAL)"
  67. depends on UPROBE_EVENT && PERF_EVENTS
  68. default n
  69. help
  70. Uprobes is the user-space counterpart to kprobes: they
  71. enable instrumentation applications (such as 'perf probe')
  72. to establish unintrusive probes in user-space binaries and
  73. libraries, by executing handler functions when the probes
  74. are hit by user-space applications.
  75. ( These probes come in the form of single-byte breakpoints,
  76. managed by the kernel and kept transparent to the probed
  77. application. )
  78. If in doubt, say "N".
  79. config HAVE_EFFICIENT_UNALIGNED_ACCESS
  80. bool
  81. help
  82. Some architectures are unable to perform unaligned accesses
  83. without the use of get_unaligned/put_unaligned. Others are
  84. unable to perform such accesses efficiently (e.g. trap on
  85. unaligned access and require fixing it up in the exception
  86. handler.)
  87. This symbol should be selected by an architecture if it can
  88. perform unaligned accesses efficiently to allow different
  89. code paths to be selected for these cases. Some network
  90. drivers, for example, could opt to not fix up alignment
  91. problems with received packets if doing so would not help
  92. much.
  93. See Documentation/unaligned-memory-access.txt for more
  94. information on the topic of unaligned memory accesses.
  95. config HAVE_SYSCALL_WRAPPERS
  96. bool
  97. config KRETPROBES
  98. def_bool y
  99. depends on KPROBES && HAVE_KRETPROBES
  100. config USER_RETURN_NOTIFIER
  101. bool
  102. depends on HAVE_USER_RETURN_NOTIFIER
  103. help
  104. Provide a kernel-internal notification when a cpu is about to
  105. switch to user mode.
  106. config HAVE_IOREMAP_PROT
  107. bool
  108. config HAVE_KPROBES
  109. bool
  110. config HAVE_KRETPROBES
  111. bool
  112. config HAVE_OPTPROBES
  113. bool
  114. config HAVE_NMI_WATCHDOG
  115. bool
  116. #
  117. # An arch should select this if it provides all these things:
  118. #
  119. # task_pt_regs() in asm/processor.h or asm/ptrace.h
  120. # arch_has_single_step() if there is hardware single-step support
  121. # arch_has_block_step() if there is hardware block-step support
  122. # asm/syscall.h supplying asm-generic/syscall.h interface
  123. # linux/regset.h user_regset interfaces
  124. # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
  125. # TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit}
  126. # TIF_NOTIFY_RESUME calls tracehook_notify_resume()
  127. # signal delivery calls tracehook_signal_handler()
  128. #
  129. config HAVE_ARCH_TRACEHOOK
  130. bool
  131. config HAVE_DMA_ATTRS
  132. bool
  133. config USE_GENERIC_SMP_HELPERS
  134. bool
  135. config GENERIC_SMP_IDLE_THREAD
  136. bool
  137. # Select if arch init_task initializer is different to init/init_task.c
  138. config ARCH_INIT_TASK
  139. bool
  140. # Select if arch has its private alloc_task_struct() function
  141. config ARCH_TASK_STRUCT_ALLOCATOR
  142. bool
  143. # Select if arch has its private alloc_thread_info() function
  144. config ARCH_THREAD_INFO_ALLOCATOR
  145. bool
  146. config HAVE_REGS_AND_STACK_ACCESS_API
  147. bool
  148. help
  149. This symbol should be selected by an architecure if it supports
  150. the API needed to access registers and stack entries from pt_regs,
  151. declared in asm/ptrace.h
  152. For example the kprobes-based event tracer needs this API.
  153. config HAVE_CLK
  154. bool
  155. help
  156. The <linux/clk.h> calls support software clock gating and
  157. thus are a key power management tool on many systems.
  158. config HAVE_DMA_API_DEBUG
  159. bool
  160. config HAVE_HW_BREAKPOINT
  161. bool
  162. depends on PERF_EVENTS
  163. config HAVE_MIXED_BREAKPOINTS_REGS
  164. bool
  165. depends on HAVE_HW_BREAKPOINT
  166. help
  167. Depending on the arch implementation of hardware breakpoints,
  168. some of them have separate registers for data and instruction
  169. breakpoints addresses, others have mixed registers to store
  170. them but define the access type in a control register.
  171. Select this option if your arch implements breakpoints under the
  172. latter fashion.
  173. config HAVE_USER_RETURN_NOTIFIER
  174. bool
  175. config HAVE_PERF_EVENTS_NMI
  176. bool
  177. help
  178. System hardware can generate an NMI using the perf event
  179. subsystem. Also has support for calculating CPU cycle events
  180. to determine how many clock cycles in a given period.
  181. config HAVE_ARCH_JUMP_LABEL
  182. bool
  183. config HAVE_ARCH_MUTEX_CPU_RELAX
  184. bool
  185. config HAVE_RCU_TABLE_FREE
  186. bool
  187. config ARCH_HAVE_NMI_SAFE_CMPXCHG
  188. bool
  189. config HAVE_ALIGNED_STRUCT_PAGE
  190. bool
  191. help
  192. This makes sure that struct pages are double word aligned and that
  193. e.g. the SLUB allocator can perform double word atomic operations
  194. on a struct page for better performance. However selecting this
  195. might increase the size of a struct page by a word.
  196. config HAVE_CMPXCHG_LOCAL
  197. bool
  198. config HAVE_CMPXCHG_DOUBLE
  199. bool
  200. config ARCH_WANT_OLD_COMPAT_IPC
  201. bool
  202. config HAVE_ARCH_SECCOMP_FILTER
  203. bool
  204. help
  205. An arch should select this symbol if it provides all of these things:
  206. - syscall_get_arch()
  207. - syscall_get_arguments()
  208. - syscall_rollback()
  209. - syscall_set_return_value()
  210. - SIGSYS siginfo_t support
  211. - secure_computing is called from a ptrace_event()-safe context
  212. - secure_computing return value is checked and a return value of -1
  213. results in the system call being skipped immediately.
  214. config SECCOMP_FILTER
  215. def_bool y
  216. depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
  217. help
  218. Enable tasks to build secure computing environments defined
  219. in terms of Berkeley Packet Filter programs which implement
  220. task-defined system call filtering polices.
  221. See Documentation/prctl/seccomp_filter.txt for details.
  222. source "kernel/gcov/Kconfig"