Kconfig.debug 8.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302
  1. menu "Kernel hacking"
  2. config TRACE_IRQFLAGS_SUPPORT
  3. def_bool y
  4. source "lib/Kconfig.debug"
  5. config NONPROMISC_DEVMEM
  6. bool "Disable promiscuous /dev/mem"
  7. help
  8. The /dev/mem file by default only allows userspace access to PCI
  9. space and the BIOS code and data regions. This is sufficient for
  10. dosemu and X and all common users of /dev/mem. With this config
  11. option, you allow userspace access to all of memory, including
  12. kernel and userspace memory. Accidental access to this is
  13. obviously disasterous, but specific access can be used by people
  14. debugging the kernel.
  15. config EARLY_PRINTK
  16. bool "Early printk" if EMBEDDED
  17. default y
  18. help
  19. Write kernel log output directly into the VGA buffer or to a serial
  20. port.
  21. This is useful for kernel debugging when your machine crashes very
  22. early before the console code is initialized. For normal operation
  23. it is not recommended because it looks ugly and doesn't cooperate
  24. with klogd/syslogd or the X server. You should normally N here,
  25. unless you want to debug such a crash.
  26. config DEBUG_STACKOVERFLOW
  27. bool "Check for stack overflows"
  28. depends on DEBUG_KERNEL
  29. help
  30. This option will cause messages to be printed if free stack space
  31. drops below a certain limit.
  32. config DEBUG_STACK_USAGE
  33. bool "Stack utilization instrumentation"
  34. depends on DEBUG_KERNEL
  35. help
  36. Enables the display of the minimum amount of free stack which each
  37. task has ever had available in the sysrq-T and sysrq-P debug output.
  38. This option will slow down process creation somewhat.
  39. config DEBUG_PAGEALLOC
  40. bool "Debug page memory allocations"
  41. depends on DEBUG_KERNEL
  42. help
  43. Unmap pages from the kernel linear mapping after free_pages().
  44. This results in a large slowdown, but helps to find certain types
  45. of memory corruptions.
  46. config DEBUG_PER_CPU_MAPS
  47. bool "Debug access to per_cpu maps"
  48. depends on DEBUG_KERNEL
  49. depends on X86_64_SMP
  50. default n
  51. help
  52. Say Y to verify that the per_cpu map being accessed has
  53. been setup. Adds a fair amount of code to kernel memory
  54. and decreases performance.
  55. Say N if unsure.
  56. config X86_PTDUMP
  57. bool "Export kernel pagetable layout to userspace via debugfs"
  58. depends on DEBUG_KERNEL
  59. select DEBUG_FS
  60. help
  61. Say Y here if you want to show the kernel pagetable layout in a
  62. debugfs file. This information is only useful for kernel developers
  63. who are working in architecture specific areas of the kernel.
  64. It is probably not a good idea to enable this feature in a production
  65. kernel.
  66. If in doubt, say "N"
  67. config DEBUG_RODATA
  68. bool "Write protect kernel read-only data structures"
  69. default y
  70. depends on DEBUG_KERNEL
  71. help
  72. Mark the kernel read-only data as write-protected in the pagetables,
  73. in order to catch accidental (and incorrect) writes to such const
  74. data. This is recommended so that we can catch kernel bugs sooner.
  75. If in doubt, say "Y".
  76. config DIRECT_GBPAGES
  77. bool "Enable gbpages-mapped kernel pagetables"
  78. depends on DEBUG_KERNEL && EXPERIMENTAL && X86_64
  79. help
  80. Enable gigabyte pages support (if the CPU supports it). This can
  81. improve the kernel's performance a tiny bit by reducing TLB
  82. pressure.
  83. This is experimental code.
  84. If in doubt, say "N".
  85. config DEBUG_RODATA_TEST
  86. bool "Testcase for the DEBUG_RODATA feature"
  87. depends on DEBUG_RODATA
  88. help
  89. This option enables a testcase for the DEBUG_RODATA
  90. feature as well as for the change_page_attr() infrastructure.
  91. If in doubt, say "N"
  92. config DEBUG_NX_TEST
  93. tristate "Testcase for the NX non-executable stack feature"
  94. depends on DEBUG_KERNEL && m
  95. help
  96. This option enables a testcase for the CPU NX capability
  97. and the software setup of this feature.
  98. If in doubt, say "N"
  99. config 4KSTACKS
  100. bool "Use 4Kb for kernel stacks instead of 8Kb"
  101. depends on X86_32
  102. help
  103. If you say Y here the kernel will use a 4Kb stacksize for the
  104. kernel stack attached to each process/thread. This facilitates
  105. running more threads on a system and also reduces the pressure
  106. on the VM subsystem for higher order allocations. This option
  107. will also use IRQ stacks to compensate for the reduced stackspace.
  108. config X86_FIND_SMP_CONFIG
  109. def_bool y
  110. depends on X86_LOCAL_APIC || X86_VOYAGER
  111. depends on X86_32
  112. config X86_MPPARSE
  113. def_bool y
  114. depends on (X86_32 && (X86_LOCAL_APIC && !X86_VISWS)) || X86_64
  115. config DOUBLEFAULT
  116. default y
  117. bool "Enable doublefault exception handler" if EMBEDDED
  118. depends on X86_32
  119. help
  120. This option allows trapping of rare doublefault exceptions that
  121. would otherwise cause a system to silently reboot. Disabling this
  122. option saves about 4k and might cause you much additional grey
  123. hair.
  124. config IOMMU_DEBUG
  125. bool "Enable IOMMU debugging"
  126. depends on GART_IOMMU && DEBUG_KERNEL
  127. depends on X86_64
  128. help
  129. Force the IOMMU to on even when you have less than 4GB of
  130. memory and add debugging code. On overflow always panic. And
  131. allow to enable IOMMU leak tracing. Can be disabled at boot
  132. time with iommu=noforce. This will also enable scatter gather
  133. list merging. Currently not recommended for production
  134. code. When you use it make sure you have a big enough
  135. IOMMU/AGP aperture. Most of the options enabled by this can
  136. be set more finegrained using the iommu= command line
  137. options. See Documentation/x86_64/boot-options.txt for more
  138. details.
  139. config IOMMU_LEAK
  140. bool "IOMMU leak tracing"
  141. depends on DEBUG_KERNEL
  142. depends on IOMMU_DEBUG
  143. help
  144. Add a simple leak tracer to the IOMMU code. This is useful when you
  145. are debugging a buggy device driver that leaks IOMMU mappings.
  146. config MMIOTRACE_HOOKS
  147. bool
  148. config MMIOTRACE
  149. bool "Memory mapped IO tracing"
  150. depends on DEBUG_KERNEL
  151. select TRACING
  152. select MMIOTRACE_HOOKS
  153. default y
  154. help
  155. Mmiotrace traces Memory Mapped I/O access and is meant for
  156. debugging and reverse engineering. It is called from the ioremap
  157. implementation and works via page faults. A user space program is
  158. required to collect the MMIO data from debugfs files.
  159. Tracing is disabled by default and can be enabled from a debugfs
  160. file.
  161. See http://nouveau.freedesktop.org/wiki/MmioTrace
  162. If you are not helping to develop drivers, say N.
  163. config MMIOTRACE_TEST
  164. tristate "Test module for mmiotrace"
  165. depends on MMIOTRACE && m
  166. help
  167. This is a dumb module for testing mmiotrace. It is very dangerous
  168. as it will write garbage to IO memory starting at a given address.
  169. However, it should be safe to use on e.g. unused portion of VRAM.
  170. Say N, unless you absolutely know what you are doing.
  171. #
  172. # IO delay types:
  173. #
  174. config IO_DELAY_TYPE_0X80
  175. int
  176. default "0"
  177. config IO_DELAY_TYPE_0XED
  178. int
  179. default "1"
  180. config IO_DELAY_TYPE_UDELAY
  181. int
  182. default "2"
  183. config IO_DELAY_TYPE_NONE
  184. int
  185. default "3"
  186. choice
  187. prompt "IO delay type"
  188. default IO_DELAY_0X80
  189. config IO_DELAY_0X80
  190. bool "port 0x80 based port-IO delay [recommended]"
  191. help
  192. This is the traditional Linux IO delay used for in/out_p.
  193. It is the most tested hence safest selection here.
  194. config IO_DELAY_0XED
  195. bool "port 0xed based port-IO delay"
  196. help
  197. Use port 0xed as the IO delay. This frees up port 0x80 which is
  198. often used as a hardware-debug port.
  199. config IO_DELAY_UDELAY
  200. bool "udelay based port-IO delay"
  201. help
  202. Use udelay(2) as the IO delay method. This provides the delay
  203. while not having any side-effect on the IO port space.
  204. config IO_DELAY_NONE
  205. bool "no port-IO delay"
  206. help
  207. No port-IO delay. Will break on old boxes that require port-IO
  208. delay for certain operations. Should work on most new machines.
  209. endchoice
  210. if IO_DELAY_0X80
  211. config DEFAULT_IO_DELAY_TYPE
  212. int
  213. default IO_DELAY_TYPE_0X80
  214. endif
  215. if IO_DELAY_0XED
  216. config DEFAULT_IO_DELAY_TYPE
  217. int
  218. default IO_DELAY_TYPE_0XED
  219. endif
  220. if IO_DELAY_UDELAY
  221. config DEFAULT_IO_DELAY_TYPE
  222. int
  223. default IO_DELAY_TYPE_UDELAY
  224. endif
  225. if IO_DELAY_NONE
  226. config DEFAULT_IO_DELAY_TYPE
  227. int
  228. default IO_DELAY_TYPE_NONE
  229. endif
  230. config DEBUG_BOOT_PARAMS
  231. bool "Debug boot parameters"
  232. depends on DEBUG_KERNEL
  233. depends on DEBUG_FS
  234. help
  235. This option will cause struct boot_params to be exported via debugfs.
  236. config CPA_DEBUG
  237. bool "CPA self-test code"
  238. depends on DEBUG_KERNEL
  239. help
  240. Do change_page_attr() self-tests every 30 seconds.
  241. config OPTIMIZE_INLINING
  242. bool "Allow gcc to uninline functions marked 'inline'"
  243. depends on BROKEN
  244. help
  245. This option determines if the kernel forces gcc to inline the functions
  246. developers have marked 'inline'. Doing so takes away freedom from gcc to
  247. do what it thinks is best, which is desirable for the gcc 3.x series of
  248. compilers. The gcc 4.x series have a rewritten inlining algorithm and
  249. disabling this option will generate a smaller kernel there. Hopefully
  250. this algorithm is so good that allowing gcc4 to make the decision can
  251. become the default in the future, until then this option is there to
  252. test gcc for this.
  253. endmenu