1menu "printk and dmesg options" 2 3config PRINTK_TIME 4 bool "Show timing information on printks" 5 depends on PRINTK 6 help 7 Selecting this option causes time stamps of the printk() 8 messages to be added to the output of the syslog() system 9 call and at the console. 10 11 The timestamp is always recorded internally, and exported 12 to /dev/kmsg. This flag just specifies if the timestamp should 13 be included, not that the timestamp is recorded. 14 15 The behavior is also controlled by the kernel command line 16 parameter printk.time=1. See Documentation/kernel-parameters.txt 17 18config MESSAGE_LOGLEVEL_DEFAULT 19 int "Default message log level (1-7)" 20 range 1 7 21 default "4" 22 help 23 Default log level for printk statements with no specified priority. 24 25 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks 26 that are auditing their logs closely may want to set it to a lower 27 priority. 28 29config BOOT_PRINTK_DELAY 30 bool "Delay each boot printk message by N milliseconds" 31 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY 32 help 33 This build option allows you to read kernel boot messages 34 by inserting a short delay after each one. The delay is 35 specified in milliseconds on the kernel command line, 36 using "boot_delay=N". 37 38 It is likely that you would also need to use "lpj=M" to preset 39 the "loops per jiffie" value. 40 See a previous boot log for the "lpj" value to use for your 41 system, and then set "lpj=M" before setting "boot_delay=N". 42 NOTE: Using this option may adversely affect SMP systems. 43 I.e., processors other than the first one may not boot up. 44 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect 45 what it believes to be lockup conditions. 46 47config DYNAMIC_DEBUG 48 bool "Enable dynamic printk() support" 49 default n 50 depends on PRINTK 51 depends on DEBUG_FS 52 help 53 54 Compiles debug level messages into the kernel, which would not 55 otherwise be available at runtime. These messages can then be 56 enabled/disabled based on various levels of scope - per source file, 57 function, module, format string, and line number. This mechanism 58 implicitly compiles in all pr_debug() and dev_dbg() calls, which 59 enlarges the kernel text size by about 2%. 60 61 If a source file is compiled with DEBUG flag set, any 62 pr_debug() calls in it are enabled by default, but can be 63 disabled at runtime as below. Note that DEBUG flag is 64 turned on by many CONFIG_*DEBUG* options. 65 66 Usage: 67 68 Dynamic debugging is controlled via the 'dynamic_debug/control' file, 69 which is contained in the 'debugfs' filesystem. Thus, the debugfs 70 filesystem must first be mounted before making use of this feature. 71 We refer the control file as: <debugfs>/dynamic_debug/control. This 72 file contains a list of the debug statements that can be enabled. The 73 format for each line of the file is: 74 75 filename:lineno [module]function flags format 76 77 filename : source file of the debug statement 78 lineno : line number of the debug statement 79 module : module that contains the debug statement 80 function : function that contains the debug statement 81 flags : '=p' means the line is turned 'on' for printing 82 format : the format used for the debug statement 83 84 From a live system: 85 86 nullarbor:~ # cat <debugfs>/dynamic_debug/control 87 # filename:lineno [module]function flags format 88 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012" 89 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012" 90 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012" 91 92 Example usage: 93 94 // enable the message at line 1603 of file svcsock.c 95 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' > 96 <debugfs>/dynamic_debug/control 97 98 // enable all the messages in file svcsock.c 99 nullarbor:~ # echo -n 'file svcsock.c +p' > 100 <debugfs>/dynamic_debug/control 101 102 // enable all the messages in the NFS server module 103 nullarbor:~ # echo -n 'module nfsd +p' > 104 <debugfs>/dynamic_debug/control 105 106 // enable all 12 messages in the function svc_process() 107 nullarbor:~ # echo -n 'func svc_process +p' > 108 <debugfs>/dynamic_debug/control 109 110 // disable all 12 messages in the function svc_process() 111 nullarbor:~ # echo -n 'func svc_process -p' > 112 <debugfs>/dynamic_debug/control 113 114 See Documentation/dynamic-debug-howto.txt for additional information. 115 116endmenu # "printk and dmesg options" 117 118menu "Compile-time checks and compiler options" 119 120config DEBUG_INFO 121 bool "Compile the kernel with debug info" 122 depends on DEBUG_KERNEL && !COMPILE_TEST 123 help 124 If you say Y here the resulting kernel image will include 125 debugging info resulting in a larger kernel image. 126 This adds debug symbols to the kernel and modules (gcc -g), and 127 is needed if you intend to use kernel crashdump or binary object 128 tools like crash, kgdb, LKCD, gdb, etc on the kernel. 129 Say Y here only if you plan to debug the kernel. 130 131 If unsure, say N. 132 133config DEBUG_INFO_REDUCED 134 bool "Reduce debugging information" 135 depends on DEBUG_INFO 136 help 137 If you say Y here gcc is instructed to generate less debugging 138 information for structure types. This means that tools that 139 need full debugging information (like kgdb or systemtap) won't 140 be happy. But if you merely need debugging information to 141 resolve line numbers there is no loss. Advantage is that 142 build directory object sizes shrink dramatically over a full 143 DEBUG_INFO build and compile times are reduced too. 144 Only works with newer gcc versions. 145 146config DEBUG_INFO_SPLIT 147 bool "Produce split debuginfo in .dwo files" 148 depends on DEBUG_INFO 149 help 150 Generate debug info into separate .dwo files. This significantly 151 reduces the build directory size for builds with DEBUG_INFO, 152 because it stores the information only once on disk in .dwo 153 files instead of multiple times in object files and executables. 154 In addition the debug information is also compressed. 155 156 Requires recent gcc (4.7+) and recent gdb/binutils. 157 Any tool that packages or reads debug information would need 158 to know about the .dwo files and include them. 159 Incompatible with older versions of ccache. 160 161config DEBUG_INFO_DWARF4 162 bool "Generate dwarf4 debuginfo" 163 depends on DEBUG_INFO 164 help 165 Generate dwarf4 debug info. This requires recent versions 166 of gcc and gdb. It makes the debug information larger. 167 But it significantly improves the success of resolving 168 variables in gdb on optimized code. 169 170config GDB_SCRIPTS 171 bool "Provide GDB scripts for kernel debugging" 172 depends on DEBUG_INFO 173 help 174 This creates the required links to GDB helper scripts in the 175 build directory. If you load vmlinux into gdb, the helper 176 scripts will be automatically imported by gdb as well, and 177 additional functions are available to analyze a Linux kernel 178 instance. See Documentation/gdb-kernel-debugging.txt for further 179 details. 180 181config ENABLE_WARN_DEPRECATED 182 bool "Enable __deprecated logic" 183 default y 184 help 185 Enable the __deprecated logic in the kernel build. 186 Disable this to suppress the "warning: 'foo' is deprecated 187 (declared at kernel/power/somefile.c:1234)" messages. 188 189config ENABLE_MUST_CHECK 190 bool "Enable __must_check logic" 191 default y 192 help 193 Enable the __must_check logic in the kernel build. Disable this to 194 suppress the "warning: ignoring return value of 'foo', declared with 195 attribute warn_unused_result" messages. 196 197config FRAME_WARN 198 int "Warn for stack frames larger than (needs gcc 4.4)" 199 range 0 8192 200 default 1024 if !64BIT 201 default 2048 if 64BIT 202 help 203 Tell gcc to warn at build time for stack frames larger than this. 204 Setting this too low will cause a lot of warnings. 205 Setting it to 0 disables the warning. 206 Requires gcc 4.4 207 208config STRIP_ASM_SYMS 209 bool "Strip assembler-generated symbols during link" 210 default n 211 help 212 Strip internal assembler-generated symbols during a link (symbols 213 that look like '.Lxxx') so they don't pollute the output of 214 get_wchan() and suchlike. 215 216config READABLE_ASM 217 bool "Generate readable assembler code" 218 depends on DEBUG_KERNEL 219 help 220 Disable some compiler optimizations that tend to generate human unreadable 221 assembler output. This may make the kernel slightly slower, but it helps 222 to keep kernel developers who have to stare a lot at assembler listings 223 sane. 224 225config UNUSED_SYMBOLS 226 bool "Enable unused/obsolete exported symbols" 227 default y if X86 228 help 229 Unused but exported symbols make the kernel needlessly bigger. For 230 that reason most of these unused exports will soon be removed. This 231 option is provided temporarily to provide a transition period in case 232 some external kernel module needs one of these symbols anyway. If you 233 encounter such a case in your module, consider if you are actually 234 using the right API. (rationale: since nobody in the kernel is using 235 this in a module, there is a pretty good chance it's actually the 236 wrong interface to use). If you really need the symbol, please send a 237 mail to the linux kernel mailing list mentioning the symbol and why 238 you really need it, and what the merge plan to the mainline kernel for 239 your module is. 240 241config PAGE_OWNER 242 bool "Track page owner" 243 depends on DEBUG_KERNEL && STACKTRACE_SUPPORT 244 select DEBUG_FS 245 select STACKTRACE 246 select PAGE_EXTENSION 247 help 248 This keeps track of what call chain is the owner of a page, may 249 help to find bare alloc_page(s) leaks. Even if you include this 250 feature on your build, it is disabled in default. You should pass 251 "page_owner=on" to boot parameter in order to enable it. Eats 252 a fair amount of memory if enabled. See tools/vm/page_owner_sort.c 253 for user-space helper. 254 255 If unsure, say N. 256 257config DEBUG_FS 258 bool "Debug Filesystem" 259 help 260 debugfs is a virtual file system that kernel developers use to put 261 debugging files into. Enable this option to be able to read and 262 write to these files. 263 264 For detailed documentation on the debugfs API, see 265 Documentation/DocBook/filesystems. 266 267 If unsure, say N. 268 269config HEADERS_CHECK 270 bool "Run 'make headers_check' when building vmlinux" 271 depends on !UML 272 help 273 This option will extract the user-visible kernel headers whenever 274 building the kernel, and will run basic sanity checks on them to 275 ensure that exported files do not attempt to include files which 276 were not exported, etc. 277 278 If you're making modifications to header files which are 279 relevant for userspace, say 'Y', and check the headers 280 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in 281 your build tree), to make sure they're suitable. 282 283config DEBUG_SECTION_MISMATCH 284 bool "Enable full Section mismatch analysis" 285 help 286 The section mismatch analysis checks if there are illegal 287 references from one section to another section. 288 During linktime or runtime, some sections are dropped; 289 any use of code/data previously in these sections would 290 most likely result in an oops. 291 In the code, functions and variables are annotated with 292 __init,, etc. (see the full list in include/linux/init.h), 293 which results in the code/data being placed in specific sections. 294 The section mismatch analysis is always performed after a full 295 kernel build, and enabling this option causes the following 296 additional steps to occur: 297 - Add the option -fno-inline-functions-called-once to gcc commands. 298 When inlining a function annotated with __init in a non-init 299 function, we would lose the section information and thus 300 the analysis would not catch the illegal reference. 301 This option tells gcc to inline less (but it does result in 302 a larger kernel). 303 - Run the section mismatch analysis for each module/built-in.o file. 304 When we run the section mismatch analysis on vmlinux.o, we 305 lose valueble information about where the mismatch was 306 introduced. 307 Running the analysis for each module/built-in.o file 308 tells where the mismatch happens much closer to the 309 source. The drawback is that the same mismatch is 310 reported at least twice. 311 - Enable verbose reporting from modpost in order to help resolve 312 the section mismatches that are reported. 313 314# 315# Select this config option from the architecture Kconfig, if it 316# is preferred to always offer frame pointers as a config 317# option on the architecture (regardless of KERNEL_DEBUG): 318# 319config ARCH_WANT_FRAME_POINTERS 320 bool 321 help 322 323config FRAME_POINTER 324 bool "Compile the kernel with frame pointers" 325 depends on DEBUG_KERNEL && \ 326 (CRIS || M68K || FRV || UML || \ 327 AVR32 || SUPERH || BLACKFIN || MN10300 || METAG) || \ 328 ARCH_WANT_FRAME_POINTERS 329 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS 330 help 331 If you say Y here the resulting kernel image will be slightly 332 larger and slower, but it gives very useful debugging information 333 in case of kernel bugs. (precise oopses/stacktraces/warnings) 334 335config DEBUG_FORCE_WEAK_PER_CPU 336 bool "Force weak per-cpu definitions" 337 depends on DEBUG_KERNEL 338 help 339 s390 and alpha require percpu variables in modules to be 340 defined weak to work around addressing range issue which 341 puts the following two restrictions on percpu variable 342 definitions. 343 344 1. percpu symbols must be unique whether static or not 345 2. percpu variables can't be defined inside a function 346 347 To ensure that generic code follows the above rules, this 348 option forces all percpu variables to be defined as weak. 349 350endmenu # "Compiler options" 351 352config MAGIC_SYSRQ 353 bool "Magic SysRq key" 354 depends on !UML 355 help 356 If you say Y here, you will have some control over the system even 357 if the system crashes for example during kernel debugging (e.g., you 358 will be able to flush the buffer cache to disk, reboot the system 359 immediately or dump some status information). This is accomplished 360 by pressing various keys while holding SysRq (Alt+PrintScreen). It 361 also works on a serial console (on PC hardware at least), if you 362 send a BREAK and then within 5 seconds a command keypress. The 363 keys are documented in <file:Documentation/sysrq.txt>. Don't say Y 364 unless you really know what this hack does. 365 366config MAGIC_SYSRQ_DEFAULT_ENABLE 367 hex "Enable magic SysRq key functions by default" 368 depends on MAGIC_SYSRQ 369 default 0x1 370 help 371 Specifies which SysRq key functions are enabled by default. 372 This may be set to 1 or 0 to enable or disable them all, or 373 to a bitmask as described in Documentation/sysrq.txt. 374 375config DEBUG_KERNEL 376 bool "Kernel debugging" 377 help 378 Say Y here if you are developing drivers or trying to debug and 379 identify kernel problems. 380 381menu "Memory Debugging" 382 383source mm/Kconfig.debug 384 385config DEBUG_OBJECTS 386 bool "Debug object operations" 387 depends on DEBUG_KERNEL 388 help 389 If you say Y here, additional code will be inserted into the 390 kernel to track the life time of various objects and validate 391 the operations on those objects. 392 393config DEBUG_OBJECTS_SELFTEST 394 bool "Debug objects selftest" 395 depends on DEBUG_OBJECTS 396 help 397 This enables the selftest of the object debug code. 398 399config DEBUG_OBJECTS_FREE 400 bool "Debug objects in freed memory" 401 depends on DEBUG_OBJECTS 402 help 403 This enables checks whether a k/v free operation frees an area 404 which contains an object which has not been deactivated 405 properly. This can make kmalloc/kfree-intensive workloads 406 much slower. 407 408config DEBUG_OBJECTS_TIMERS 409 bool "Debug timer objects" 410 depends on DEBUG_OBJECTS 411 help 412 If you say Y here, additional code will be inserted into the 413 timer routines to track the life time of timer objects and 414 validate the timer operations. 415 416config DEBUG_OBJECTS_WORK 417 bool "Debug work objects" 418 depends on DEBUG_OBJECTS 419 help 420 If you say Y here, additional code will be inserted into the 421 work queue routines to track the life time of work objects and 422 validate the work operations. 423 424config DEBUG_OBJECTS_RCU_HEAD 425 bool "Debug RCU callbacks objects" 426 depends on DEBUG_OBJECTS 427 help 428 Enable this to turn on debugging of RCU list heads (call_rcu() usage). 429 430config DEBUG_OBJECTS_PERCPU_COUNTER 431 bool "Debug percpu counter objects" 432 depends on DEBUG_OBJECTS 433 help 434 If you say Y here, additional code will be inserted into the 435 percpu counter routines to track the life time of percpu counter 436 objects and validate the percpu counter operations. 437 438config DEBUG_OBJECTS_ENABLE_DEFAULT 439 int "debug_objects bootup default value (0-1)" 440 range 0 1 441 default "1" 442 depends on DEBUG_OBJECTS 443 help 444 Debug objects boot parameter default value 445 446config DEBUG_SLAB 447 bool "Debug slab memory allocations" 448 depends on DEBUG_KERNEL && SLAB && !KMEMCHECK 449 help 450 Say Y here to have the kernel do limited verification on memory 451 allocation as well as poisoning memory on free to catch use of freed 452 memory. This can make kmalloc/kfree-intensive workloads much slower. 453 454config DEBUG_SLAB_LEAK 455 bool "Memory leak debugging" 456 depends on DEBUG_SLAB 457 458config SLUB_DEBUG_ON 459 bool "SLUB debugging on by default" 460 depends on SLUB && SLUB_DEBUG && !KMEMCHECK 461 default n 462 help 463 Boot with debugging on by default. SLUB boots by default with 464 the runtime debug capabilities switched off. Enabling this is 465 equivalent to specifying the "slub_debug" parameter on boot. 466 There is no support for more fine grained debug control like 467 possible with slub_debug=xxx. SLUB debugging may be switched 468 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying 469 "slub_debug=-". 470 471config SLUB_STATS 472 default n 473 bool "Enable SLUB performance statistics" 474 depends on SLUB && SYSFS 475 help 476 SLUB statistics are useful to debug SLUBs allocation behavior in 477 order find ways to optimize the allocator. This should never be 478 enabled for production use since keeping statistics slows down 479 the allocator by a few percentage points. The slabinfo command 480 supports the determination of the most active slabs to figure 481 out which slabs are relevant to a particular load. 482 Try running: slabinfo -DA 483 484config HAVE_DEBUG_KMEMLEAK 485 bool 486 487config DEBUG_KMEMLEAK 488 bool "Kernel memory leak detector" 489 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK 490 select DEBUG_FS 491 select STACKTRACE if STACKTRACE_SUPPORT 492 select KALLSYMS 493 select CRC32 494 help 495 Say Y here if you want to enable the memory leak 496 detector. The memory allocation/freeing is traced in a way 497 similar to the Boehm's conservative garbage collector, the 498 difference being that the orphan objects are not freed but 499 only shown in /sys/kernel/debug/kmemleak. Enabling this 500 feature will introduce an overhead to memory 501 allocations. See Documentation/kmemleak.txt for more 502 details. 503 504 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances 505 of finding leaks due to the slab objects poisoning. 506 507 In order to access the kmemleak file, debugfs needs to be 508 mounted (usually at /sys/kernel/debug). 509 510config DEBUG_KMEMLEAK_EARLY_LOG_SIZE 511 int "Maximum kmemleak early log entries" 512 depends on DEBUG_KMEMLEAK 513 range 200 40000 514 default 400 515 help 516 Kmemleak must track all the memory allocations to avoid 517 reporting false positives. Since memory may be allocated or 518 freed before kmemleak is initialised, an early log buffer is 519 used to store these actions. If kmemleak reports "early log 520 buffer exceeded", please increase this value. 521 522config DEBUG_KMEMLEAK_TEST 523 tristate "Simple test for the kernel memory leak detector" 524 depends on DEBUG_KMEMLEAK && m 525 help 526 This option enables a module that explicitly leaks memory. 527 528 If unsure, say N. 529 530config DEBUG_KMEMLEAK_DEFAULT_OFF 531 bool "Default kmemleak to off" 532 depends on DEBUG_KMEMLEAK 533 help 534 Say Y here to disable kmemleak by default. It can then be enabled 535 on the command line via kmemleak=on. 536 537config DEBUG_STACK_USAGE 538 bool "Stack utilization instrumentation" 539 depends on DEBUG_KERNEL && !IA64 && !PARISC && !METAG 540 help 541 Enables the display of the minimum amount of free stack which each 542 task has ever had available in the sysrq-T and sysrq-P debug output. 543 544 This option will slow down process creation somewhat. 545 546config DEBUG_VM 547 bool "Debug VM" 548 depends on DEBUG_KERNEL 549 help 550 Enable this to turn on extended checks in the virtual-memory system 551 that may impact performance. 552 553 If unsure, say N. 554 555config DEBUG_VM_VMACACHE 556 bool "Debug VMA caching" 557 depends on DEBUG_VM 558 help 559 Enable this to turn on VMA caching debug information. Doing so 560 can cause significant overhead, so only enable it in non-production 561 environments. 562 563 If unsure, say N. 564 565config DEBUG_VM_RB 566 bool "Debug VM red-black trees" 567 depends on DEBUG_VM 568 help 569 Enable VM red-black tree debugging information and extra validations. 570 571 If unsure, say N. 572 573config DEBUG_VIRTUAL 574 bool "Debug VM translations" 575 depends on DEBUG_KERNEL && X86 576 help 577 Enable some costly sanity checks in virtual to page code. This can 578 catch mistakes with virt_to_page() and friends. 579 580 If unsure, say N. 581 582config DEBUG_NOMMU_REGIONS 583 bool "Debug the global anon/private NOMMU mapping region tree" 584 depends on DEBUG_KERNEL && !MMU 585 help 586 This option causes the global tree of anonymous and private mapping 587 regions to be regularly checked for invalid topology. 588 589config DEBUG_MEMORY_INIT 590 bool "Debug memory initialisation" if EXPERT 591 default !EXPERT 592 help 593 Enable this for additional checks during memory initialisation. 594 The sanity checks verify aspects of the VM such as the memory model 595 and other information provided by the architecture. Verbose 596 information will be printed at KERN_DEBUG loglevel depending 597 on the mminit_loglevel= command-line option. 598 599 If unsure, say Y 600 601config MEMORY_NOTIFIER_ERROR_INJECT 602 tristate "Memory hotplug notifier error injection module" 603 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION 604 help 605 This option provides the ability to inject artificial errors to 606 memory hotplug notifier chain callbacks. It is controlled through 607 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory 608 609 If the notifier call chain should be failed with some events 610 notified, write the error code to "actions/<notifier event>/error". 611 612 Example: Inject memory hotplug offline error (-12 == -ENOMEM) 613 614 # cd /sys/kernel/debug/notifier-error-inject/memory 615 # echo -12 > actions/MEM_GOING_OFFLINE/error 616 # echo offline > /sys/devices/system/memory/memoryXXX/state 617 bash: echo: write error: Cannot allocate memory 618 619 To compile this code as a module, choose M here: the module will 620 be called memory-notifier-error-inject. 621 622 If unsure, say N. 623 624config DEBUG_PER_CPU_MAPS 625 bool "Debug access to per_cpu maps" 626 depends on DEBUG_KERNEL 627 depends on SMP 628 help 629 Say Y to verify that the per_cpu map being accessed has 630 been set up. This adds a fair amount of code to kernel memory 631 and decreases performance. 632 633 Say N if unsure. 634 635config DEBUG_HIGHMEM 636 bool "Highmem debugging" 637 depends on DEBUG_KERNEL && HIGHMEM 638 help 639 This option enables additional error checking for high memory 640 systems. Disable for production systems. 641 642config HAVE_DEBUG_STACKOVERFLOW 643 bool 644 645config DEBUG_STACKOVERFLOW 646 bool "Check for stack overflows" 647 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW 648 ---help--- 649 Say Y here if you want to check for overflows of kernel, IRQ 650 and exception stacks (if your architecture uses them). This 651 option will show detailed messages if free stack space drops 652 below a certain limit. 653 654 These kinds of bugs usually occur when call-chains in the 655 kernel get too deep, especially when interrupts are 656 involved. 657 658 Use this in cases where you see apparently random memory 659 corruption, especially if it appears in 'struct thread_info' 660 661 If in doubt, say "N". 662 663source "lib/Kconfig.kmemcheck" 664 665source "lib/Kconfig.kasan" 666 667endmenu # "Memory Debugging" 668 669config DEBUG_SHIRQ 670 bool "Debug shared IRQ handlers" 671 depends on DEBUG_KERNEL 672 help 673 Enable this to generate a spurious interrupt as soon as a shared 674 interrupt handler is registered, and just before one is deregistered. 675 Drivers ought to be able to handle interrupts coming in at those 676 points; some don't and need to be caught. 677 678menu "Debug Lockups and Hangs" 679 680config LOCKUP_DETECTOR 681 bool "Detect Hard and Soft Lockups" 682 depends on DEBUG_KERNEL && !S390 683 help 684 Say Y here to enable the kernel to act as a watchdog to detect 685 hard and soft lockups. 686 687 Softlockups are bugs that cause the kernel to loop in kernel 688 mode for more than 20 seconds, without giving other tasks a 689 chance to run. The current stack trace is displayed upon 690 detection and the system will stay locked up. 691 692 Hardlockups are bugs that cause the CPU to loop in kernel mode 693 for more than 10 seconds, without letting other interrupts have a 694 chance to run. The current stack trace is displayed upon detection 695 and the system will stay locked up. 696 697 The overhead should be minimal. A periodic hrtimer runs to 698 generate interrupts and kick the watchdog task every 4 seconds. 699 An NMI is generated every 10 seconds or so to check for hardlockups. 700 701 The frequency of hrtimer and NMI events and the soft and hard lockup 702 thresholds can be controlled through the sysctl watchdog_thresh. 703 704config HARDLOCKUP_DETECTOR 705 def_bool y 706 depends on LOCKUP_DETECTOR && !HAVE_NMI_WATCHDOG 707 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI 708 709config BOOTPARAM_HARDLOCKUP_PANIC 710 bool "Panic (Reboot) On Hard Lockups" 711 depends on HARDLOCKUP_DETECTOR 712 help 713 Say Y here to enable the kernel to panic on "hard lockups", 714 which are bugs that cause the kernel to loop in kernel 715 mode with interrupts disabled for more than 10 seconds (configurable 716 using the watchdog_thresh sysctl). 717 718 Say N if unsure. 719 720config BOOTPARAM_HARDLOCKUP_PANIC_VALUE 721 int 722 depends on HARDLOCKUP_DETECTOR 723 range 0 1 724 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC 725 default 1 if BOOTPARAM_HARDLOCKUP_PANIC 726 727config BOOTPARAM_SOFTLOCKUP_PANIC 728 bool "Panic (Reboot) On Soft Lockups" 729 depends on LOCKUP_DETECTOR 730 help 731 Say Y here to enable the kernel to panic on "soft lockups", 732 which are bugs that cause the kernel to loop in kernel 733 mode for more than 20 seconds (configurable using the watchdog_thresh 734 sysctl), without giving other tasks a chance to run. 735 736 The panic can be used in combination with panic_timeout, 737 to cause the system to reboot automatically after a 738 lockup has been detected. This feature is useful for 739 high-availability systems that have uptime guarantees and 740 where a lockup must be resolved ASAP. 741 742 Say N if unsure. 743 744config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE 745 int 746 depends on LOCKUP_DETECTOR 747 range 0 1 748 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC 749 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC 750 751config DETECT_HUNG_TASK 752 bool "Detect Hung Tasks" 753 depends on DEBUG_KERNEL 754 default LOCKUP_DETECTOR 755 help 756 Say Y here to enable the kernel to detect "hung tasks", 757 which are bugs that cause the task to be stuck in 758 uninterruptible "D" state indefinitiley. 759 760 When a hung task is detected, the kernel will print the 761 current stack trace (which you should report), but the 762 task will stay in uninterruptible state. If lockdep is 763 enabled then all held locks will also be reported. This 764 feature has negligible overhead. 765 766config DEFAULT_HUNG_TASK_TIMEOUT 767 int "Default timeout for hung task detection (in seconds)" 768 depends on DETECT_HUNG_TASK 769 default 120 770 help 771 This option controls the default timeout (in seconds) used 772 to determine when a task has become non-responsive and should 773 be considered hung. 774 775 It can be adjusted at runtime via the kernel.hung_task_timeout_secs 776 sysctl or by writing a value to 777 /proc/sys/kernel/hung_task_timeout_secs. 778 779 A timeout of 0 disables the check. The default is two minutes. 780 Keeping the default should be fine in most cases. 781 782config BOOTPARAM_HUNG_TASK_PANIC 783 bool "Panic (Reboot) On Hung Tasks" 784 depends on DETECT_HUNG_TASK 785 help 786 Say Y here to enable the kernel to panic on "hung tasks", 787 which are bugs that cause the kernel to leave a task stuck 788 in uninterruptible "D" state. 789 790 The panic can be used in combination with panic_timeout, 791 to cause the system to reboot automatically after a 792 hung task has been detected. This feature is useful for 793 high-availability systems that have uptime guarantees and 794 where a hung tasks must be resolved ASAP. 795 796 Say N if unsure. 797 798config BOOTPARAM_HUNG_TASK_PANIC_VALUE 799 int 800 depends on DETECT_HUNG_TASK 801 range 0 1 802 default 0 if !BOOTPARAM_HUNG_TASK_PANIC 803 default 1 if BOOTPARAM_HUNG_TASK_PANIC 804 805endmenu # "Debug lockups and hangs" 806 807config PANIC_ON_OOPS 808 bool "Panic on Oops" 809 help 810 Say Y here to enable the kernel to panic when it oopses. This 811 has the same effect as setting oops=panic on the kernel command 812 line. 813 814 This feature is useful to ensure that the kernel does not do 815 anything erroneous after an oops which could result in data 816 corruption or other issues. 817 818 Say N if unsure. 819 820config PANIC_ON_OOPS_VALUE 821 int 822 range 0 1 823 default 0 if !PANIC_ON_OOPS 824 default 1 if PANIC_ON_OOPS 825 826config PANIC_TIMEOUT 827 int "panic timeout" 828 default 0 829 help 830 Set the timeout value (in seconds) until a reboot occurs when the 831 the kernel panics. If n = 0, then we wait forever. A timeout 832 value n > 0 will wait n seconds before rebooting, while a timeout 833 value n < 0 will reboot immediately. 834 835config SCHED_DEBUG 836 bool "Collect scheduler debugging info" 837 depends on DEBUG_KERNEL && PROC_FS 838 default y 839 help 840 If you say Y here, the /proc/sched_debug file will be provided 841 that can help debug the scheduler. The runtime overhead of this 842 option is minimal. 843 844config SCHED_INFO 845 bool 846 default n 847 848config SCHEDSTATS 849 bool "Collect scheduler statistics" 850 depends on DEBUG_KERNEL && PROC_FS 851 select SCHED_INFO 852 help 853 If you say Y here, additional code will be inserted into the 854 scheduler and related routines to collect statistics about 855 scheduler behavior and provide them in /proc/schedstat. These 856 stats may be useful for both tuning and debugging the scheduler 857 If you aren't debugging the scheduler or trying to tune a specific 858 application, you can say N to avoid the very slight overhead 859 this adds. 860 861config SCHED_STACK_END_CHECK 862 bool "Detect stack corruption on calls to schedule()" 863 depends on DEBUG_KERNEL 864 default n 865 help 866 This option checks for a stack overrun on calls to schedule(). 867 If the stack end location is found to be over written always panic as 868 the content of the corrupted region can no longer be trusted. 869 This is to ensure no erroneous behaviour occurs which could result in 870 data corruption or a sporadic crash at a later stage once the region 871 is examined. The runtime overhead introduced is minimal. 872 873config DEBUG_TIMEKEEPING 874 bool "Enable extra timekeeping sanity checking" 875 help 876 This option will enable additional timekeeping sanity checks 877 which may be helpful when diagnosing issues where timekeeping 878 problems are suspected. 879 880 This may include checks in the timekeeping hotpaths, so this 881 option may have a (very small) performance impact to some 882 workloads. 883 884 If unsure, say N. 885 886config TIMER_STATS 887 bool "Collect kernel timers statistics" 888 depends on DEBUG_KERNEL && PROC_FS 889 help 890 If you say Y here, additional code will be inserted into the 891 timer routines to collect statistics about kernel timers being 892 reprogrammed. The statistics can be read from /proc/timer_stats. 893 The statistics collection is started by writing 1 to /proc/timer_stats, 894 writing 0 stops it. This feature is useful to collect information 895 about timer usage patterns in kernel and userspace. This feature 896 is lightweight if enabled in the kernel config but not activated 897 (it defaults to deactivated on bootup and will only be activated 898 if some application like powertop activates it explicitly). 899 900config DEBUG_PREEMPT 901 bool "Debug preemptible kernel" 902 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT 903 default y 904 help 905 If you say Y here then the kernel will use a debug variant of the 906 commonly used smp_processor_id() function and will print warnings 907 if kernel code uses it in a preemption-unsafe way. Also, the kernel 908 will detect preemption count underflows. 909 910menu "Lock Debugging (spinlocks, mutexes, etc...)" 911 912config DEBUG_RT_MUTEXES 913 bool "RT Mutex debugging, deadlock detection" 914 depends on DEBUG_KERNEL && RT_MUTEXES 915 help 916 This allows rt mutex semantics violations and rt mutex related 917 deadlocks (lockups) to be detected and reported automatically. 918 919config RT_MUTEX_TESTER 920 bool "Built-in scriptable tester for rt-mutexes" 921 depends on DEBUG_KERNEL && RT_MUTEXES && BROKEN 922 help 923 This option enables a rt-mutex tester. 924 925config DEBUG_SPINLOCK 926 bool "Spinlock and rw-lock debugging: basic checks" 927 depends on DEBUG_KERNEL 928 select UNINLINE_SPIN_UNLOCK 929 help 930 Say Y here and build SMP to catch missing spinlock initialization 931 and certain other kinds of spinlock errors commonly made. This is 932 best used in conjunction with the NMI watchdog so that spinlock 933 deadlocks are also debuggable. 934 935config DEBUG_MUTEXES 936 bool "Mutex debugging: basic checks" 937 depends on DEBUG_KERNEL 938 help 939 This feature allows mutex semantics violations to be detected and 940 reported. 941 942config DEBUG_WW_MUTEX_SLOWPATH 943 bool "Wait/wound mutex debugging: Slowpath testing" 944 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 945 select DEBUG_LOCK_ALLOC 946 select DEBUG_SPINLOCK 947 select DEBUG_MUTEXES 948 help 949 This feature enables slowpath testing for w/w mutex users by 950 injecting additional -EDEADLK wound/backoff cases. Together with 951 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this 952 will test all possible w/w mutex interface abuse with the 953 exception of simply not acquiring all the required locks. 954 Note that this feature can introduce significant overhead, so 955 it really should not be enabled in a production or distro kernel, 956 even a debug kernel. If you are a driver writer, enable it. If 957 you are a distro, do not. 958 959config DEBUG_LOCK_ALLOC 960 bool "Lock debugging: detect incorrect freeing of live locks" 961 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 962 select DEBUG_SPINLOCK 963 select DEBUG_MUTEXES 964 select LOCKDEP 965 help 966 This feature will check whether any held lock (spinlock, rwlock, 967 mutex or rwsem) is incorrectly freed by the kernel, via any of the 968 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(), 969 vfree(), etc.), whether a live lock is incorrectly reinitialized via 970 spin_lock_init()/mutex_init()/etc., or whether there is any lock 971 held during task exit. 972 973config PROVE_LOCKING 974 bool "Lock debugging: prove locking correctness" 975 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 976 select LOCKDEP 977 select DEBUG_SPINLOCK 978 select DEBUG_MUTEXES 979 select DEBUG_LOCK_ALLOC 980 select TRACE_IRQFLAGS 981 default n 982 help 983 This feature enables the kernel to prove that all locking 984 that occurs in the kernel runtime is mathematically 985 correct: that under no circumstance could an arbitrary (and 986 not yet triggered) combination of observed locking 987 sequences (on an arbitrary number of CPUs, running an 988 arbitrary number of tasks and interrupt contexts) cause a 989 deadlock. 990 991 In short, this feature enables the kernel to report locking 992 related deadlocks before they actually occur. 993 994 The proof does not depend on how hard and complex a 995 deadlock scenario would be to trigger: how many 996 participant CPUs, tasks and irq-contexts would be needed 997 for it to trigger. The proof also does not depend on 998 timing: if a race and a resulting deadlock is possible 999 theoretically (no matter how unlikely the race scenario 1000 is), it will be proven so and will immediately be 1001 reported by the kernel (once the event is observed that 1002 makes the deadlock theoretically possible). 1003 1004 If a deadlock is impossible (i.e. the locking rules, as 1005 observed by the kernel, are mathematically correct), the 1006 kernel reports nothing. 1007 1008 NOTE: this feature can also be enabled for rwlocks, mutexes 1009 and rwsems - in which case all dependencies between these 1010 different locking variants are observed and mapped too, and 1011 the proof of observed correctness is also maintained for an 1012 arbitrary combination of these separate locking variants. 1013 1014 For more details, see Documentation/locking/lockdep-design.txt. 1015 1016config LOCKDEP 1017 bool 1018 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 1019 select STACKTRACE 1020 select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE && !ARC && !SCORE 1021 select KALLSYMS 1022 select KALLSYMS_ALL 1023 1024config LOCK_STAT 1025 bool "Lock usage statistics" 1026 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 1027 select LOCKDEP 1028 select DEBUG_SPINLOCK 1029 select DEBUG_MUTEXES 1030 select DEBUG_LOCK_ALLOC 1031 default n 1032 help 1033 This feature enables tracking lock contention points 1034 1035 For more details, see Documentation/locking/lockstat.txt 1036 1037 This also enables lock events required by "perf lock", 1038 subcommand of perf. 1039 If you want to use "perf lock", you also need to turn on 1040 CONFIG_EVENT_TRACING. 1041 1042 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events. 1043 (CONFIG_LOCKDEP defines "acquire" and "release" events.) 1044 1045config DEBUG_LOCKDEP 1046 bool "Lock dependency engine debugging" 1047 depends on DEBUG_KERNEL && LOCKDEP 1048 help 1049 If you say Y here, the lock dependency engine will do 1050 additional runtime checks to debug itself, at the price 1051 of more runtime overhead. 1052 1053config DEBUG_ATOMIC_SLEEP 1054 bool "Sleep inside atomic section checking" 1055 select PREEMPT_COUNT 1056 depends on DEBUG_KERNEL 1057 help 1058 If you say Y here, various routines which may sleep will become very 1059 noisy if they are called inside atomic sections: when a spinlock is 1060 held, inside an rcu read side critical section, inside preempt disabled 1061 sections, inside an interrupt, etc... 1062 1063config DEBUG_LOCKING_API_SELFTESTS 1064 bool "Locking API boot-time self-tests" 1065 depends on DEBUG_KERNEL 1066 help 1067 Say Y here if you want the kernel to run a short self-test during 1068 bootup. The self-test checks whether common types of locking bugs 1069 are detected by debugging mechanisms or not. (if you disable 1070 lock debugging then those bugs wont be detected of course.) 1071 The following locking APIs are covered: spinlocks, rwlocks, 1072 mutexes and rwsems. 1073 1074config LOCK_TORTURE_TEST 1075 tristate "torture tests for locking" 1076 depends on DEBUG_KERNEL 1077 select TORTURE_TEST 1078 default n 1079 help 1080 This option provides a kernel module that runs torture tests 1081 on kernel locking primitives. The kernel module may be built 1082 after the fact on the running kernel to be tested, if desired. 1083 1084 Say Y here if you want kernel locking-primitive torture tests 1085 to be built into the kernel. 1086 Say M if you want these torture tests to build as a module. 1087 Say N if you are unsure. 1088 1089endmenu # lock debugging 1090 1091config TRACE_IRQFLAGS 1092 bool 1093 help 1094 Enables hooks to interrupt enabling and disabling for 1095 either tracing or lock debugging. 1096 1097config STACKTRACE 1098 bool "Stack backtrace support" 1099 depends on STACKTRACE_SUPPORT 1100 help 1101 This option causes the kernel to create a /proc/pid/stack for 1102 every process, showing its current stack trace. 1103 It is also used by various kernel debugging features that require 1104 stack trace generation. 1105 1106config DEBUG_KOBJECT 1107 bool "kobject debugging" 1108 depends on DEBUG_KERNEL 1109 help 1110 If you say Y here, some extra kobject debugging messages will be sent 1111 to the syslog. 1112 1113config DEBUG_KOBJECT_RELEASE 1114 bool "kobject release debugging" 1115 depends on DEBUG_OBJECTS_TIMERS 1116 help 1117 kobjects are reference counted objects. This means that their 1118 last reference count put is not predictable, and the kobject can 1119 live on past the point at which a driver decides to drop it's 1120 initial reference to the kobject gained on allocation. An 1121 example of this would be a struct device which has just been 1122 unregistered. 1123 1124 However, some buggy drivers assume that after such an operation, 1125 the memory backing the kobject can be immediately freed. This 1126 goes completely against the principles of a refcounted object. 1127 1128 If you say Y here, the kernel will delay the release of kobjects 1129 on the last reference count to improve the visibility of this 1130 kind of kobject release bug. 1131 1132config HAVE_DEBUG_BUGVERBOSE 1133 bool 1134 1135config DEBUG_BUGVERBOSE 1136 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT 1137 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE) 1138 default y 1139 help 1140 Say Y here to make BUG() panics output the file name and line number 1141 of the BUG call as well as the EIP and oops trace. This aids 1142 debugging but costs about 70-100K of memory. 1143 1144config DEBUG_LIST 1145 bool "Debug linked list manipulation" 1146 depends on DEBUG_KERNEL 1147 help 1148 Enable this to turn on extended checks in the linked-list 1149 walking routines. 1150 1151 If unsure, say N. 1152 1153config DEBUG_PI_LIST 1154 bool "Debug priority linked list manipulation" 1155 depends on DEBUG_KERNEL 1156 help 1157 Enable this to turn on extended checks in the priority-ordered 1158 linked-list (plist) walking routines. This checks the entire 1159 list multiple times during each manipulation. 1160 1161 If unsure, say N. 1162 1163config DEBUG_SG 1164 bool "Debug SG table operations" 1165 depends on DEBUG_KERNEL 1166 help 1167 Enable this to turn on checks on scatter-gather tables. This can 1168 help find problems with drivers that do not properly initialize 1169 their sg tables. 1170 1171 If unsure, say N. 1172 1173config DEBUG_NOTIFIERS 1174 bool "Debug notifier call chains" 1175 depends on DEBUG_KERNEL 1176 help 1177 Enable this to turn on sanity checking for notifier call chains. 1178 This is most useful for kernel developers to make sure that 1179 modules properly unregister themselves from notifier chains. 1180 This is a relatively cheap check but if you care about maximum 1181 performance, say N. 1182 1183config DEBUG_CREDENTIALS 1184 bool "Debug credential management" 1185 depends on DEBUG_KERNEL 1186 help 1187 Enable this to turn on some debug checking for credential 1188 management. The additional code keeps track of the number of 1189 pointers from task_structs to any given cred struct, and checks to 1190 see that this number never exceeds the usage count of the cred 1191 struct. 1192 1193 Furthermore, if SELinux is enabled, this also checks that the 1194 security pointer in the cred struct is never seen to be invalid. 1195 1196 If unsure, say N. 1197 1198menu "RCU Debugging" 1199 1200config PROVE_RCU 1201 def_bool PROVE_LOCKING 1202 1203config PROVE_RCU_REPEATEDLY 1204 bool "RCU debugging: don't disable PROVE_RCU on first splat" 1205 depends on PROVE_RCU 1206 default n 1207 help 1208 By itself, PROVE_RCU will disable checking upon issuing the 1209 first warning (or "splat"). This feature prevents such 1210 disabling, allowing multiple RCU-lockdep warnings to be printed 1211 on a single reboot. 1212 1213 Say Y to allow multiple RCU-lockdep warnings per boot. 1214 1215 Say N if you are unsure. 1216 1217config SPARSE_RCU_POINTER 1218 bool "RCU debugging: sparse-based checks for pointer usage" 1219 default n 1220 help 1221 This feature enables the __rcu sparse annotation for 1222 RCU-protected pointers. This annotation will cause sparse 1223 to flag any non-RCU used of annotated pointers. This can be 1224 helpful when debugging RCU usage. Please note that this feature 1225 is not intended to enforce code cleanliness; it is instead merely 1226 a debugging aid. 1227 1228 Say Y to make sparse flag questionable use of RCU-protected pointers 1229 1230 Say N if you are unsure. 1231 1232config TORTURE_TEST 1233 tristate 1234 default n 1235 1236config RCU_TORTURE_TEST 1237 tristate "torture tests for RCU" 1238 depends on DEBUG_KERNEL 1239 select TORTURE_TEST 1240 select SRCU 1241 select TASKS_RCU 1242 default n 1243 help 1244 This option provides a kernel module that runs torture tests 1245 on the RCU infrastructure. The kernel module may be built 1246 after the fact on the running kernel to be tested, if desired. 1247 1248 Say Y here if you want RCU torture tests to be built into 1249 the kernel. 1250 Say M if you want the RCU torture tests to build as a module. 1251 Say N if you are unsure. 1252 1253config RCU_TORTURE_TEST_RUNNABLE 1254 bool "torture tests for RCU runnable by default" 1255 depends on RCU_TORTURE_TEST = y 1256 default n 1257 help 1258 This option provides a way to build the RCU torture tests 1259 directly into the kernel without them starting up at boot 1260 time. You can use /proc/sys/kernel/rcutorture_runnable 1261 to manually override this setting. This /proc file is 1262 available only when the RCU torture tests have been built 1263 into the kernel. 1264 1265 Say Y here if you want the RCU torture tests to start during 1266 boot (you probably don't). 1267 Say N here if you want the RCU torture tests to start only 1268 after being manually enabled via /proc. 1269 1270config RCU_TORTURE_TEST_SLOW_PREINIT 1271 bool "Slow down RCU grace-period pre-initialization to expose races" 1272 depends on RCU_TORTURE_TEST 1273 help 1274 This option delays grace-period pre-initialization (the 1275 propagation of CPU-hotplug changes up the rcu_node combining 1276 tree) for a few jiffies between initializing each pair of 1277 consecutive rcu_node structures. This helps to expose races 1278 involving grace-period pre-initialization, in other words, it 1279 makes your kernel less stable. It can also greatly increase 1280 grace-period latency, especially on systems with large numbers 1281 of CPUs. This is useful when torture-testing RCU, but in 1282 almost no other circumstance. 1283 1284 Say Y here if you want your system to crash and hang more often. 1285 Say N if you want a sane system. 1286 1287config RCU_TORTURE_TEST_SLOW_PREINIT_DELAY 1288 int "How much to slow down RCU grace-period pre-initialization" 1289 range 0 5 1290 default 3 1291 depends on RCU_TORTURE_TEST_SLOW_PREINIT 1292 help 1293 This option specifies the number of jiffies to wait between 1294 each rcu_node structure pre-initialization step. 1295 1296config RCU_TORTURE_TEST_SLOW_INIT 1297 bool "Slow down RCU grace-period initialization to expose races" 1298 depends on RCU_TORTURE_TEST 1299 help 1300 This option delays grace-period initialization for a few 1301 jiffies between initializing each pair of consecutive 1302 rcu_node structures. This helps to expose races involving 1303 grace-period initialization, in other words, it makes your 1304 kernel less stable. It can also greatly increase grace-period 1305 latency, especially on systems with large numbers of CPUs. 1306 This is useful when torture-testing RCU, but in almost no 1307 other circumstance. 1308 1309 Say Y here if you want your system to crash and hang more often. 1310 Say N if you want a sane system. 1311 1312config RCU_TORTURE_TEST_SLOW_INIT_DELAY 1313 int "How much to slow down RCU grace-period initialization" 1314 range 0 5 1315 default 3 1316 depends on RCU_TORTURE_TEST_SLOW_INIT 1317 help 1318 This option specifies the number of jiffies to wait between 1319 each rcu_node structure initialization. 1320 1321config RCU_TORTURE_TEST_SLOW_CLEANUP 1322 bool "Slow down RCU grace-period cleanup to expose races" 1323 depends on RCU_TORTURE_TEST 1324 help 1325 This option delays grace-period cleanup for a few jiffies 1326 between cleaning up each pair of consecutive rcu_node 1327 structures. This helps to expose races involving grace-period 1328 cleanup, in other words, it makes your kernel less stable. 1329 It can also greatly increase grace-period latency, especially 1330 on systems with large numbers of CPUs. This is useful when 1331 torture-testing RCU, but in almost no other circumstance. 1332 1333 Say Y here if you want your system to crash and hang more often. 1334 Say N if you want a sane system. 1335 1336config RCU_TORTURE_TEST_SLOW_CLEANUP_DELAY 1337 int "How much to slow down RCU grace-period cleanup" 1338 range 0 5 1339 default 3 1340 depends on RCU_TORTURE_TEST_SLOW_CLEANUP 1341 help 1342 This option specifies the number of jiffies to wait between 1343 each rcu_node structure cleanup operation. 1344 1345config RCU_CPU_STALL_TIMEOUT 1346 int "RCU CPU stall timeout in seconds" 1347 depends on RCU_STALL_COMMON 1348 range 3 300 1349 default 21 1350 help 1351 If a given RCU grace period extends more than the specified 1352 number of seconds, a CPU stall warning is printed. If the 1353 RCU grace period persists, additional CPU stall warnings are 1354 printed at more widely spaced intervals. 1355 1356config RCU_TRACE 1357 bool "Enable tracing for RCU" 1358 depends on DEBUG_KERNEL 1359 select TRACE_CLOCK 1360 help 1361 This option provides tracing in RCU which presents stats 1362 in debugfs for debugging RCU implementation. 1363 1364 Say Y here if you want to enable RCU tracing 1365 Say N if you are unsure. 1366 1367config RCU_EQS_DEBUG 1368 bool "Provide debugging asserts for adding NO_HZ support to an arch" 1369 depends on DEBUG_KERNEL 1370 help 1371 This option provides consistency checks in RCU's handling of 1372 NO_HZ. These checks have proven quite helpful in detecting 1373 bugs in arch-specific NO_HZ code. 1374 1375 Say N here if you need ultimate kernel/user switch latencies 1376 Say Y if you are unsure 1377 1378endmenu # "RCU Debugging" 1379 1380config DEBUG_BLOCK_EXT_DEVT 1381 bool "Force extended block device numbers and spread them" 1382 depends on DEBUG_KERNEL 1383 depends on BLOCK 1384 default n 1385 help 1386 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON 1387 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT 1388 YOU ARE DOING. Distros, please enable this and fix whatever 1389 is broken. 1390 1391 Conventionally, block device numbers are allocated from 1392 predetermined contiguous area. However, extended block area 1393 may introduce non-contiguous block device numbers. This 1394 option forces most block device numbers to be allocated from 1395 the extended space and spreads them to discover kernel or 1396 userland code paths which assume predetermined contiguous 1397 device number allocation. 1398 1399 Note that turning on this debug option shuffles all the 1400 device numbers for all IDE and SCSI devices including libata 1401 ones, so root partition specified using device number 1402 directly (via rdev or root=MAJ:MIN) won't work anymore. 1403 Textual device names (root=/dev/sdXn) will continue to work. 1404 1405 Say N if you are unsure. 1406 1407config NOTIFIER_ERROR_INJECTION 1408 tristate "Notifier error injection" 1409 depends on DEBUG_KERNEL 1410 select DEBUG_FS 1411 help 1412 This option provides the ability to inject artificial errors to 1413 specified notifier chain callbacks. It is useful to test the error 1414 handling of notifier call chain failures. 1415 1416 Say N if unsure. 1417 1418config CPU_NOTIFIER_ERROR_INJECT 1419 tristate "CPU notifier error injection module" 1420 depends on HOTPLUG_CPU && NOTIFIER_ERROR_INJECTION 1421 help 1422 This option provides a kernel module that can be used to test 1423 the error handling of the cpu notifiers by injecting artificial 1424 errors to CPU notifier chain callbacks. It is controlled through 1425 debugfs interface under /sys/kernel/debug/notifier-error-inject/cpu 1426 1427 If the notifier call chain should be failed with some events 1428 notified, write the error code to "actions/<notifier event>/error". 1429 1430 Example: Inject CPU offline error (-1 == -EPERM) 1431 1432 # cd /sys/kernel/debug/notifier-error-inject/cpu 1433 # echo -1 > actions/CPU_DOWN_PREPARE/error 1434 # echo 0 > /sys/devices/system/cpu/cpu1/online 1435 bash: echo: write error: Operation not permitted 1436 1437 To compile this code as a module, choose M here: the module will 1438 be called cpu-notifier-error-inject. 1439 1440 If unsure, say N. 1441 1442config PM_NOTIFIER_ERROR_INJECT 1443 tristate "PM notifier error injection module" 1444 depends on PM && NOTIFIER_ERROR_INJECTION 1445 default m if PM_DEBUG 1446 help 1447 This option provides the ability to inject artificial errors to 1448 PM notifier chain callbacks. It is controlled through debugfs 1449 interface /sys/kernel/debug/notifier-error-inject/pm 1450 1451 If the notifier call chain should be failed with some events 1452 notified, write the error code to "actions/<notifier event>/error". 1453 1454 Example: Inject PM suspend error (-12 = -ENOMEM) 1455 1456 # cd /sys/kernel/debug/notifier-error-inject/pm/ 1457 # echo -12 > actions/PM_SUSPEND_PREPARE/error 1458 # echo mem > /sys/power/state 1459 bash: echo: write error: Cannot allocate memory 1460 1461 To compile this code as a module, choose M here: the module will 1462 be called pm-notifier-error-inject. 1463 1464 If unsure, say N. 1465 1466config OF_RECONFIG_NOTIFIER_ERROR_INJECT 1467 tristate "OF reconfig notifier error injection module" 1468 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION 1469 help 1470 This option provides the ability to inject artificial errors to 1471 OF reconfig notifier chain callbacks. It is controlled 1472 through debugfs interface under 1473 /sys/kernel/debug/notifier-error-inject/OF-reconfig/ 1474 1475 If the notifier call chain should be failed with some events 1476 notified, write the error code to "actions/<notifier event>/error". 1477 1478 To compile this code as a module, choose M here: the module will 1479 be called of-reconfig-notifier-error-inject. 1480 1481 If unsure, say N. 1482 1483config FAULT_INJECTION 1484 bool "Fault-injection framework" 1485 depends on DEBUG_KERNEL 1486 help 1487 Provide fault-injection framework. 1488 For more details, see Documentation/fault-injection/. 1489 1490config FAILSLAB 1491 bool "Fault-injection capability for kmalloc" 1492 depends on FAULT_INJECTION 1493 depends on SLAB || SLUB 1494 help 1495 Provide fault-injection capability for kmalloc. 1496 1497config FAIL_PAGE_ALLOC 1498 bool "Fault-injection capabilitiy for alloc_pages()" 1499 depends on FAULT_INJECTION 1500 help 1501 Provide fault-injection capability for alloc_pages(). 1502 1503config FAIL_MAKE_REQUEST 1504 bool "Fault-injection capability for disk IO" 1505 depends on FAULT_INJECTION && BLOCK 1506 help 1507 Provide fault-injection capability for disk IO. 1508 1509config FAIL_IO_TIMEOUT 1510 bool "Fault-injection capability for faking disk interrupts" 1511 depends on FAULT_INJECTION && BLOCK 1512 help 1513 Provide fault-injection capability on end IO handling. This 1514 will make the block layer "forget" an interrupt as configured, 1515 thus exercising the error handling. 1516 1517 Only works with drivers that use the generic timeout handling, 1518 for others it wont do anything. 1519 1520config FAIL_MMC_REQUEST 1521 bool "Fault-injection capability for MMC IO" 1522 select DEBUG_FS 1523 depends on FAULT_INJECTION && MMC 1524 help 1525 Provide fault-injection capability for MMC IO. 1526 This will make the mmc core return data errors. This is 1527 useful to test the error handling in the mmc block device 1528 and to test how the mmc host driver handles retries from 1529 the block device. 1530 1531config FAULT_INJECTION_DEBUG_FS 1532 bool "Debugfs entries for fault-injection capabilities" 1533 depends on FAULT_INJECTION && SYSFS && DEBUG_FS 1534 help 1535 Enable configuration of fault-injection capabilities via debugfs. 1536 1537config FAULT_INJECTION_STACKTRACE_FILTER 1538 bool "stacktrace filter for fault-injection capabilities" 1539 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT 1540 depends on !X86_64 1541 select STACKTRACE 1542 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !SCORE 1543 help 1544 Provide stacktrace filter for fault-injection capabilities 1545 1546config LATENCYTOP 1547 bool "Latency measuring infrastructure" 1548 depends on HAVE_LATENCYTOP_SUPPORT 1549 depends on DEBUG_KERNEL 1550 depends on STACKTRACE_SUPPORT 1551 depends on PROC_FS 1552 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC 1553 select KALLSYMS 1554 select KALLSYMS_ALL 1555 select STACKTRACE 1556 select SCHEDSTATS 1557 select SCHED_DEBUG 1558 help 1559 Enable this option if you want to use the LatencyTOP tool 1560 to find out which userspace is blocking on what kernel operations. 1561 1562config ARCH_HAS_DEBUG_STRICT_USER_COPY_CHECKS 1563 bool 1564 1565config DEBUG_STRICT_USER_COPY_CHECKS 1566 bool "Strict user copy size checks" 1567 depends on ARCH_HAS_DEBUG_STRICT_USER_COPY_CHECKS 1568 depends on DEBUG_KERNEL && !TRACE_BRANCH_PROFILING 1569 help 1570 Enabling this option turns a certain set of sanity checks for user 1571 copy operations into compile time failures. 1572 1573 The copy_from_user() etc checks are there to help test if there 1574 are sufficient security checks on the length argument of 1575 the copy operation, by having gcc prove that the argument is 1576 within bounds. 1577 1578 If unsure, say N. 1579 1580source kernel/trace/Kconfig 1581 1582menu "Runtime Testing" 1583 1584config LKDTM 1585 tristate "Linux Kernel Dump Test Tool Module" 1586 depends on DEBUG_FS 1587 depends on BLOCK 1588 default n 1589 help 1590 This module enables testing of the different dumping mechanisms by 1591 inducing system failures at predefined crash points. 1592 If you don't need it: say N 1593 Choose M here to compile this code as a module. The module will be 1594 called lkdtm. 1595 1596 Documentation on how to use the module can be found in 1597 Documentation/fault-injection/provoke-crashes.txt 1598 1599config TEST_LIST_SORT 1600 bool "Linked list sorting test" 1601 depends on DEBUG_KERNEL 1602 help 1603 Enable this to turn on 'list_sort()' function test. This test is 1604 executed only once during system boot, so affects only boot time. 1605 1606 If unsure, say N. 1607 1608config KPROBES_SANITY_TEST 1609 bool "Kprobes sanity tests" 1610 depends on DEBUG_KERNEL 1611 depends on KPROBES 1612 default n 1613 help 1614 This option provides for testing basic kprobes functionality on 1615 boot. A sample kprobe, jprobe and kretprobe are inserted and 1616 verified for functionality. 1617 1618 Say N if you are unsure. 1619 1620config BACKTRACE_SELF_TEST 1621 tristate "Self test for the backtrace code" 1622 depends on DEBUG_KERNEL 1623 default n 1624 help 1625 This option provides a kernel module that can be used to test 1626 the kernel stack backtrace code. This option is not useful 1627 for distributions or general kernels, but only for kernel 1628 developers working on architecture code. 1629 1630 Note that if you want to also test saved backtraces, you will 1631 have to enable STACKTRACE as well. 1632 1633 Say N if you are unsure. 1634 1635config RBTREE_TEST 1636 tristate "Red-Black tree test" 1637 depends on DEBUG_KERNEL 1638 help 1639 A benchmark measuring the performance of the rbtree library. 1640 Also includes rbtree invariant checks. 1641 1642config INTERVAL_TREE_TEST 1643 tristate "Interval tree test" 1644 depends on m && DEBUG_KERNEL 1645 select INTERVAL_TREE 1646 help 1647 A benchmark measuring the performance of the interval tree library 1648 1649config PERCPU_TEST 1650 tristate "Per cpu operations test" 1651 depends on m && DEBUG_KERNEL 1652 help 1653 Enable this option to build test module which validates per-cpu 1654 operations. 1655 1656 If unsure, say N. 1657 1658config ATOMIC64_SELFTEST 1659 bool "Perform an atomic64_t self-test at boot" 1660 help 1661 Enable this option to test the atomic64_t functions at boot. 1662 1663 If unsure, say N. 1664 1665config ASYNC_RAID6_TEST 1666 tristate "Self test for hardware accelerated raid6 recovery" 1667 depends on ASYNC_RAID6_RECOV 1668 select ASYNC_MEMCPY 1669 ---help--- 1670 This is a one-shot self test that permutes through the 1671 recovery of all the possible two disk failure scenarios for a 1672 N-disk array. Recovery is performed with the asynchronous 1673 raid6 recovery routines, and will optionally use an offload 1674 engine if one is available. 1675 1676 If unsure, say N. 1677 1678config TEST_HEXDUMP 1679 tristate "Test functions located in the hexdump module at runtime" 1680 1681config TEST_STRING_HELPERS 1682 tristate "Test functions located in the string_helpers module at runtime" 1683 1684config TEST_KSTRTOX 1685 tristate "Test kstrto*() family of functions at runtime" 1686 1687config TEST_RHASHTABLE 1688 tristate "Perform selftest on resizable hash table" 1689 default n 1690 help 1691 Enable this option to test the rhashtable functions at boot. 1692 1693 If unsure, say N. 1694 1695endmenu # runtime tests 1696 1697config PROVIDE_OHCI1394_DMA_INIT 1698 bool "Remote debugging over FireWire early on boot" 1699 depends on PCI && X86 1700 help 1701 If you want to debug problems which hang or crash the kernel early 1702 on boot and the crashing machine has a FireWire port, you can use 1703 this feature to remotely access the memory of the crashed machine 1704 over FireWire. This employs remote DMA as part of the OHCI1394 1705 specification which is now the standard for FireWire controllers. 1706 1707 With remote DMA, you can monitor the printk buffer remotely using 1708 firescope and access all memory below 4GB using fireproxy from gdb. 1709 Even controlling a kernel debugger is possible using remote DMA. 1710 1711 Usage: 1712 1713 If ohci1394_dma=early is used as boot parameter, it will initialize 1714 all OHCI1394 controllers which are found in the PCI config space. 1715 1716 As all changes to the FireWire bus such as enabling and disabling 1717 devices cause a bus reset and thereby disable remote DMA for all 1718 devices, be sure to have the cable plugged and FireWire enabled on 1719 the debugging host before booting the debug target for debugging. 1720 1721 This code (~1k) is freed after boot. By then, the firewire stack 1722 in charge of the OHCI-1394 controllers should be used instead. 1723 1724 See Documentation/debugging-via-ohci1394.txt for more information. 1725 1726config BUILD_DOCSRC 1727 bool "Build targets in Documentation/ tree" 1728 depends on HEADERS_CHECK 1729 help 1730 This option attempts to build objects from the source files in the 1731 kernel Documentation/ tree. 1732 1733 Say N if you are unsure. 1734 1735config DMA_API_DEBUG 1736 bool "Enable debugging of DMA-API usage" 1737 depends on HAVE_DMA_API_DEBUG 1738 help 1739 Enable this option to debug the use of the DMA API by device drivers. 1740 With this option you will be able to detect common bugs in device 1741 drivers like double-freeing of DMA mappings or freeing mappings that 1742 were never allocated. 1743 1744 This also attempts to catch cases where a page owned by DMA is 1745 accessed by the cpu in a way that could cause data corruption. For 1746 example, this enables cow_user_page() to check that the source page is 1747 not undergoing DMA. 1748 1749 This option causes a performance degradation. Use only if you want to 1750 debug device drivers and dma interactions. 1751 1752 If unsure, say N. 1753 1754config TEST_LKM 1755 tristate "Test module loading with 'hello world' module" 1756 default n 1757 depends on m 1758 help 1759 This builds the "test_module" module that emits "Hello, world" 1760 on printk when loaded. It is designed to be used for basic 1761 evaluation of the module loading subsystem (for example when 1762 validating module verification). It lacks any extra dependencies, 1763 and will not normally be loaded by the system unless explicitly 1764 requested by name. 1765 1766 If unsure, say N. 1767 1768config TEST_USER_COPY 1769 tristate "Test user/kernel boundary protections" 1770 default n 1771 depends on m 1772 help 1773 This builds the "test_user_copy" module that runs sanity checks 1774 on the copy_to/from_user infrastructure, making sure basic 1775 user/kernel boundary testing is working. If it fails to load, 1776 a regression has been detected in the user/kernel memory boundary 1777 protections. 1778 1779 If unsure, say N. 1780 1781config TEST_BPF 1782 tristate "Test BPF filter functionality" 1783 default n 1784 depends on m && NET 1785 help 1786 This builds the "test_bpf" module that runs various test vectors 1787 against the BPF interpreter or BPF JIT compiler depending on the 1788 current setting. This is in particular useful for BPF JIT compiler 1789 development, but also to run regression tests against changes in 1790 the interpreter code. It also enables test stubs for eBPF maps and 1791 verifier used by user space verifier testsuite. 1792 1793 If unsure, say N. 1794 1795config TEST_FIRMWARE 1796 tristate "Test firmware loading via userspace interface" 1797 default n 1798 depends on FW_LOADER 1799 help 1800 This builds the "test_firmware" module that creates a userspace 1801 interface for testing firmware loading. This can be used to 1802 control the triggering of firmware loading without needing an 1803 actual firmware-using device. The contents can be rechecked by 1804 userspace. 1805 1806 If unsure, say N. 1807 1808config TEST_UDELAY 1809 tristate "udelay test driver" 1810 default n 1811 help 1812 This builds the "udelay_test" module that helps to make sure 1813 that udelay() is working properly. 1814 1815 If unsure, say N. 1816 1817config MEMTEST 1818 bool "Memtest" 1819 depends on HAVE_MEMBLOCK 1820 ---help--- 1821 This option adds a kernel parameter 'memtest', which allows memtest 1822 to be set. 1823 memtest=0, mean disabled; -- default 1824 memtest=1, mean do 1 test pattern; 1825 ... 1826 memtest=17, mean do 17 test patterns. 1827 If you are unsure how to answer this question, answer N. 1828 1829source "samples/Kconfig" 1830 1831source "lib/Kconfig.kgdb" 1832 1833