1 menu "printk and dmesg options"
4 bool "Show timing information on printks"
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.
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.
15 The behavior is also controlled by the kernel command line
16 parameter printk.time=1. See Documentation/kernel-parameters.txt
18 config MESSAGE_LOGLEVEL_DEFAULT
19 int "Default message log level (1-7)"
23 Default log level for printk statements with no specified priority.
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
29 config BOOT_PRINTK_DELAY
30 bool "Delay each boot printk message by N milliseconds"
31 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
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,
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.
48 bool "Enable dynamic printk() support"
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%.
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.
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:
75 filename:lineno [module]function flags format
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
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"
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
98 // enable all the messages in file svcsock.c
99 nullarbor:~ # echo -n 'file svcsock.c +p' >
100 <debugfs>/dynamic_debug/control
102 // enable all the messages in the NFS server module
103 nullarbor:~ # echo -n 'module nfsd +p' >
104 <debugfs>/dynamic_debug/control
106 // enable all 12 messages in the function svc_process()
107 nullarbor:~ # echo -n 'func svc_process +p' >
108 <debugfs>/dynamic_debug/control
110 // disable all 12 messages in the function svc_process()
111 nullarbor:~ # echo -n 'func svc_process -p' >
112 <debugfs>/dynamic_debug/control
114 See Documentation/dynamic-debug-howto.txt for additional information.
116 endmenu # "printk and dmesg options"
118 menu "Compile-time checks and compiler options"
121 bool "Compile the kernel with debug info"
122 depends on DEBUG_KERNEL && !COMPILE_TEST
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.
133 config DEBUG_INFO_REDUCED
134 bool "Reduce debugging information"
135 depends on DEBUG_INFO
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.
146 config DEBUG_INFO_SPLIT
147 bool "Produce split debuginfo in .dwo files"
148 depends on DEBUG_INFO
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.
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.
161 config DEBUG_INFO_DWARF4
162 bool "Generate dwarf4 debuginfo"
163 depends on DEBUG_INFO
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.
171 bool "Provide GDB scripts for kernel debugging"
172 depends on DEBUG_INFO
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
181 config ENABLE_WARN_DEPRECATED
182 bool "Enable __deprecated logic"
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.
189 config ENABLE_MUST_CHECK
190 bool "Enable __must_check logic"
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.
198 int "Warn for stack frames larger than (needs gcc 4.4)"
201 default 2048 if GCC_PLUGIN_LATENT_ENTROPY
202 default 1024 if !64BIT
203 default 2048 if 64BIT
205 Tell gcc to warn at build time for stack frames larger than this.
206 Setting this too low will cause a lot of warnings.
207 Setting it to 0 disables the warning.
210 config STRIP_ASM_SYMS
211 bool "Strip assembler-generated symbols during link"
214 Strip internal assembler-generated symbols during a link (symbols
215 that look like '.Lxxx') so they don't pollute the output of
216 get_wchan() and suchlike.
219 bool "Generate readable assembler code"
220 depends on DEBUG_KERNEL
222 Disable some compiler optimizations that tend to generate human unreadable
223 assembler output. This may make the kernel slightly slower, but it helps
224 to keep kernel developers who have to stare a lot at assembler listings
227 config UNUSED_SYMBOLS
228 bool "Enable unused/obsolete exported symbols"
231 Unused but exported symbols make the kernel needlessly bigger. For
232 that reason most of these unused exports will soon be removed. This
233 option is provided temporarily to provide a transition period in case
234 some external kernel module needs one of these symbols anyway. If you
235 encounter such a case in your module, consider if you are actually
236 using the right API. (rationale: since nobody in the kernel is using
237 this in a module, there is a pretty good chance it's actually the
238 wrong interface to use). If you really need the symbol, please send a
239 mail to the linux kernel mailing list mentioning the symbol and why
240 you really need it, and what the merge plan to the mainline kernel for
244 bool "Track page owner"
245 depends on DEBUG_KERNEL && STACKTRACE_SUPPORT
249 select PAGE_EXTENSION
251 This keeps track of what call chain is the owner of a page, may
252 help to find bare alloc_page(s) leaks. Even if you include this
253 feature on your build, it is disabled in default. You should pass
254 "page_owner=on" to boot parameter in order to enable it. Eats
255 a fair amount of memory if enabled. See tools/vm/page_owner_sort.c
256 for user-space helper.
261 bool "Debug Filesystem"
264 debugfs is a virtual file system that kernel developers use to put
265 debugging files into. Enable this option to be able to read and
266 write to these files.
268 For detailed documentation on the debugfs API, see
269 Documentation/DocBook/filesystems.
274 bool "Run 'make headers_check' when building vmlinux"
277 This option will extract the user-visible kernel headers whenever
278 building the kernel, and will run basic sanity checks on them to
279 ensure that exported files do not attempt to include files which
280 were not exported, etc.
282 If you're making modifications to header files which are
283 relevant for userspace, say 'Y', and check the headers
284 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
285 your build tree), to make sure they're suitable.
287 config DEBUG_SECTION_MISMATCH
288 bool "Enable full Section mismatch analysis"
290 The section mismatch analysis checks if there are illegal
291 references from one section to another section.
292 During linktime or runtime, some sections are dropped;
293 any use of code/data previously in these sections would
294 most likely result in an oops.
295 In the code, functions and variables are annotated with
296 __init,, etc. (see the full list in include/linux/init.h),
297 which results in the code/data being placed in specific sections.
298 The section mismatch analysis is always performed after a full
299 kernel build, and enabling this option causes the following
300 additional steps to occur:
301 - Add the option -fno-inline-functions-called-once to gcc commands.
302 When inlining a function annotated with __init in a non-init
303 function, we would lose the section information and thus
304 the analysis would not catch the illegal reference.
305 This option tells gcc to inline less (but it does result in
307 - Run the section mismatch analysis for each module/built-in.o file.
308 When we run the section mismatch analysis on vmlinux.o, we
309 lose valuable information about where the mismatch was
311 Running the analysis for each module/built-in.o file
312 tells where the mismatch happens much closer to the
313 source. The drawback is that the same mismatch is
314 reported at least twice.
315 - Enable verbose reporting from modpost in order to help resolve
316 the section mismatches that are reported.
318 config SECTION_MISMATCH_WARN_ONLY
319 bool "Make section mismatch errors non-fatal"
322 If you say N here, the build process will fail if there are any
323 section mismatch, instead of just throwing warnings.
328 # Select this config option from the architecture Kconfig, if it
329 # is preferred to always offer frame pointers as a config
330 # option on the architecture (regardless of KERNEL_DEBUG):
332 config ARCH_WANT_FRAME_POINTERS
337 bool "Compile the kernel with frame pointers"
338 depends on DEBUG_KERNEL && \
339 (CRIS || M68K || FRV || UML || \
340 AVR32 || SUPERH || BLACKFIN || MN10300 || METAG) || \
341 ARCH_WANT_FRAME_POINTERS
342 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
344 If you say Y here the resulting kernel image will be slightly
345 larger and slower, but it gives very useful debugging information
346 in case of kernel bugs. (precise oopses/stacktraces/warnings)
348 config STACK_VALIDATION
349 bool "Compile-time stack metadata validation"
350 depends on HAVE_STACK_VALIDATION
353 Add compile-time checks to validate stack metadata, including frame
354 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure
355 that runtime stack traces are more reliable.
357 For more information, see
358 tools/objtool/Documentation/stack-validation.txt.
360 config DEBUG_FORCE_WEAK_PER_CPU
361 bool "Force weak per-cpu definitions"
362 depends on DEBUG_KERNEL
364 s390 and alpha require percpu variables in modules to be
365 defined weak to work around addressing range issue which
366 puts the following two restrictions on percpu variable
369 1. percpu symbols must be unique whether static or not
370 2. percpu variables can't be defined inside a function
372 To ensure that generic code follows the above rules, this
373 option forces all percpu variables to be defined as weak.
375 endmenu # "Compiler options"
378 bool "Magic SysRq key"
381 If you say Y here, you will have some control over the system even
382 if the system crashes for example during kernel debugging (e.g., you
383 will be able to flush the buffer cache to disk, reboot the system
384 immediately or dump some status information). This is accomplished
385 by pressing various keys while holding SysRq (Alt+PrintScreen). It
386 also works on a serial console (on PC hardware at least), if you
387 send a BREAK and then within 5 seconds a command keypress. The
388 keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
389 unless you really know what this hack does.
391 config MAGIC_SYSRQ_DEFAULT_ENABLE
392 hex "Enable magic SysRq key functions by default"
393 depends on MAGIC_SYSRQ
396 Specifies which SysRq key functions are enabled by default.
397 This may be set to 1 or 0 to enable or disable them all, or
398 to a bitmask as described in Documentation/sysrq.txt.
401 bool "Kernel debugging"
403 Say Y here if you are developing drivers or trying to debug and
404 identify kernel problems.
406 menu "Memory Debugging"
408 source mm/Kconfig.debug
411 bool "Debug object operations"
412 depends on DEBUG_KERNEL
414 If you say Y here, additional code will be inserted into the
415 kernel to track the life time of various objects and validate
416 the operations on those objects.
418 config DEBUG_OBJECTS_SELFTEST
419 bool "Debug objects selftest"
420 depends on DEBUG_OBJECTS
422 This enables the selftest of the object debug code.
424 config DEBUG_OBJECTS_FREE
425 bool "Debug objects in freed memory"
426 depends on DEBUG_OBJECTS
428 This enables checks whether a k/v free operation frees an area
429 which contains an object which has not been deactivated
430 properly. This can make kmalloc/kfree-intensive workloads
433 config DEBUG_OBJECTS_TIMERS
434 bool "Debug timer objects"
435 depends on DEBUG_OBJECTS
437 If you say Y here, additional code will be inserted into the
438 timer routines to track the life time of timer objects and
439 validate the timer operations.
441 config DEBUG_OBJECTS_WORK
442 bool "Debug work objects"
443 depends on DEBUG_OBJECTS
445 If you say Y here, additional code will be inserted into the
446 work queue routines to track the life time of work objects and
447 validate the work operations.
449 config DEBUG_OBJECTS_RCU_HEAD
450 bool "Debug RCU callbacks objects"
451 depends on DEBUG_OBJECTS
453 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
455 config DEBUG_OBJECTS_PERCPU_COUNTER
456 bool "Debug percpu counter objects"
457 depends on DEBUG_OBJECTS
459 If you say Y here, additional code will be inserted into the
460 percpu counter routines to track the life time of percpu counter
461 objects and validate the percpu counter operations.
463 config DEBUG_OBJECTS_ENABLE_DEFAULT
464 int "debug_objects bootup default value (0-1)"
467 depends on DEBUG_OBJECTS
469 Debug objects boot parameter default value
472 bool "Debug slab memory allocations"
473 depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
475 Say Y here to have the kernel do limited verification on memory
476 allocation as well as poisoning memory on free to catch use of freed
477 memory. This can make kmalloc/kfree-intensive workloads much slower.
479 config DEBUG_SLAB_LEAK
480 bool "Memory leak debugging"
481 depends on DEBUG_SLAB
484 bool "SLUB debugging on by default"
485 depends on SLUB && SLUB_DEBUG && !KMEMCHECK
488 Boot with debugging on by default. SLUB boots by default with
489 the runtime debug capabilities switched off. Enabling this is
490 equivalent to specifying the "slub_debug" parameter on boot.
491 There is no support for more fine grained debug control like
492 possible with slub_debug=xxx. SLUB debugging may be switched
493 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
498 bool "Enable SLUB performance statistics"
499 depends on SLUB && SYSFS
501 SLUB statistics are useful to debug SLUBs allocation behavior in
502 order find ways to optimize the allocator. This should never be
503 enabled for production use since keeping statistics slows down
504 the allocator by a few percentage points. The slabinfo command
505 supports the determination of the most active slabs to figure
506 out which slabs are relevant to a particular load.
507 Try running: slabinfo -DA
509 config HAVE_DEBUG_KMEMLEAK
512 config DEBUG_KMEMLEAK
513 bool "Kernel memory leak detector"
514 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK
516 select STACKTRACE if STACKTRACE_SUPPORT
520 Say Y here if you want to enable the memory leak
521 detector. The memory allocation/freeing is traced in a way
522 similar to the Boehm's conservative garbage collector, the
523 difference being that the orphan objects are not freed but
524 only shown in /sys/kernel/debug/kmemleak. Enabling this
525 feature will introduce an overhead to memory
526 allocations. See Documentation/kmemleak.txt for more
529 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
530 of finding leaks due to the slab objects poisoning.
532 In order to access the kmemleak file, debugfs needs to be
533 mounted (usually at /sys/kernel/debug).
535 config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
536 int "Maximum kmemleak early log entries"
537 depends on DEBUG_KMEMLEAK
541 Kmemleak must track all the memory allocations to avoid
542 reporting false positives. Since memory may be allocated or
543 freed before kmemleak is initialised, an early log buffer is
544 used to store these actions. If kmemleak reports "early log
545 buffer exceeded", please increase this value.
547 config DEBUG_KMEMLEAK_TEST
548 tristate "Simple test for the kernel memory leak detector"
549 depends on DEBUG_KMEMLEAK && m
551 This option enables a module that explicitly leaks memory.
555 config DEBUG_KMEMLEAK_DEFAULT_OFF
556 bool "Default kmemleak to off"
557 depends on DEBUG_KMEMLEAK
559 Say Y here to disable kmemleak by default. It can then be enabled
560 on the command line via kmemleak=on.
562 config DEBUG_STACK_USAGE
563 bool "Stack utilization instrumentation"
564 depends on DEBUG_KERNEL && !IA64
566 Enables the display of the minimum amount of free stack which each
567 task has ever had available in the sysrq-T and sysrq-P debug output.
569 This option will slow down process creation somewhat.
573 depends on DEBUG_KERNEL
575 Enable this to turn on extended checks in the virtual-memory system
576 that may impact performance.
580 config DEBUG_VM_VMACACHE
581 bool "Debug VMA caching"
584 Enable this to turn on VMA caching debug information. Doing so
585 can cause significant overhead, so only enable it in non-production
591 bool "Debug VM red-black trees"
594 Enable VM red-black tree debugging information and extra validations.
598 config DEBUG_VM_PGFLAGS
599 bool "Debug page-flags operations"
602 Enables extra validation on page flags operations.
607 bool "Debug VM translations"
608 depends on DEBUG_KERNEL && X86
610 Enable some costly sanity checks in virtual to page code. This can
611 catch mistakes with virt_to_page() and friends.
615 config DEBUG_NOMMU_REGIONS
616 bool "Debug the global anon/private NOMMU mapping region tree"
617 depends on DEBUG_KERNEL && !MMU
619 This option causes the global tree of anonymous and private mapping
620 regions to be regularly checked for invalid topology.
622 config DEBUG_MEMORY_INIT
623 bool "Debug memory initialisation" if EXPERT
626 Enable this for additional checks during memory initialisation.
627 The sanity checks verify aspects of the VM such as the memory model
628 and other information provided by the architecture. Verbose
629 information will be printed at KERN_DEBUG loglevel depending
630 on the mminit_loglevel= command-line option.
634 config MEMORY_NOTIFIER_ERROR_INJECT
635 tristate "Memory hotplug notifier error injection module"
636 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION
638 This option provides the ability to inject artificial errors to
639 memory hotplug notifier chain callbacks. It is controlled through
640 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
642 If the notifier call chain should be failed with some events
643 notified, write the error code to "actions/<notifier event>/error".
645 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
647 # cd /sys/kernel/debug/notifier-error-inject/memory
648 # echo -12 > actions/MEM_GOING_OFFLINE/error
649 # echo offline > /sys/devices/system/memory/memoryXXX/state
650 bash: echo: write error: Cannot allocate memory
652 To compile this code as a module, choose M here: the module will
653 be called memory-notifier-error-inject.
657 config DEBUG_PER_CPU_MAPS
658 bool "Debug access to per_cpu maps"
659 depends on DEBUG_KERNEL
662 Say Y to verify that the per_cpu map being accessed has
663 been set up. This adds a fair amount of code to kernel memory
664 and decreases performance.
669 bool "Highmem debugging"
670 depends on DEBUG_KERNEL && HIGHMEM
672 This option enables additional error checking for high memory
673 systems. Disable for production systems.
675 config HAVE_DEBUG_STACKOVERFLOW
678 config DEBUG_STACKOVERFLOW
679 bool "Check for stack overflows"
680 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW
682 Say Y here if you want to check for overflows of kernel, IRQ
683 and exception stacks (if your architecture uses them). This
684 option will show detailed messages if free stack space drops
685 below a certain limit.
687 These kinds of bugs usually occur when call-chains in the
688 kernel get too deep, especially when interrupts are
691 Use this in cases where you see apparently random memory
692 corruption, especially if it appears in 'struct thread_info'
694 If in doubt, say "N".
696 source "lib/Kconfig.kmemcheck"
698 source "lib/Kconfig.kasan"
700 endmenu # "Memory Debugging"
705 KCOV does not have any arch-specific code, but currently it is enabled
706 only for x86_64. KCOV requires testing on other archs, and most likely
707 disabling of instrumentation for some early boot code.
710 bool "Code coverage for fuzzing"
711 depends on ARCH_HAS_KCOV
713 select GCC_PLUGINS if !COMPILE_TEST
714 select GCC_PLUGIN_SANCOV if !COMPILE_TEST
716 KCOV exposes kernel code coverage information in a form suitable
717 for coverage-guided fuzzing (randomized testing).
719 If RANDOMIZE_BASE is enabled, PC values will not be stable across
720 different machines and across reboots. If you need stable PC values,
721 disable RANDOMIZE_BASE.
723 For more details, see Documentation/kcov.txt.
725 config KCOV_INSTRUMENT_ALL
726 bool "Instrument all code by default"
730 If you are doing generic system call fuzzing (like e.g. syzkaller),
731 then you will want to instrument the whole kernel and you should
732 say y here. If you are doing more targeted fuzzing (like e.g.
733 filesystem fuzzing with AFL) then you will want to enable coverage
734 for more specific subsets of files, and should say n here.
737 bool "Debug shared IRQ handlers"
738 depends on DEBUG_KERNEL
740 Enable this to generate a spurious interrupt as soon as a shared
741 interrupt handler is registered, and just before one is deregistered.
742 Drivers ought to be able to handle interrupts coming in at those
743 points; some don't and need to be caught.
745 menu "Debug Lockups and Hangs"
747 config LOCKUP_DETECTOR
748 bool "Detect Hard and Soft Lockups"
749 depends on DEBUG_KERNEL && !S390
751 Say Y here to enable the kernel to act as a watchdog to detect
752 hard and soft lockups.
754 Softlockups are bugs that cause the kernel to loop in kernel
755 mode for more than 20 seconds, without giving other tasks a
756 chance to run. The current stack trace is displayed upon
757 detection and the system will stay locked up.
759 Hardlockups are bugs that cause the CPU to loop in kernel mode
760 for more than 10 seconds, without letting other interrupts have a
761 chance to run. The current stack trace is displayed upon detection
762 and the system will stay locked up.
764 The overhead should be minimal. A periodic hrtimer runs to
765 generate interrupts and kick the watchdog task every 4 seconds.
766 An NMI is generated every 10 seconds or so to check for hardlockups.
768 The frequency of hrtimer and NMI events and the soft and hard lockup
769 thresholds can be controlled through the sysctl watchdog_thresh.
771 config HARDLOCKUP_DETECTOR
773 depends on LOCKUP_DETECTOR && !HAVE_NMI_WATCHDOG
774 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI
776 config BOOTPARAM_HARDLOCKUP_PANIC
777 bool "Panic (Reboot) On Hard Lockups"
778 depends on HARDLOCKUP_DETECTOR
780 Say Y here to enable the kernel to panic on "hard lockups",
781 which are bugs that cause the kernel to loop in kernel
782 mode with interrupts disabled for more than 10 seconds (configurable
783 using the watchdog_thresh sysctl).
787 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
789 depends on HARDLOCKUP_DETECTOR
791 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
792 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
794 config BOOTPARAM_SOFTLOCKUP_PANIC
795 bool "Panic (Reboot) On Soft Lockups"
796 depends on LOCKUP_DETECTOR
798 Say Y here to enable the kernel to panic on "soft lockups",
799 which are bugs that cause the kernel to loop in kernel
800 mode for more than 20 seconds (configurable using the watchdog_thresh
801 sysctl), without giving other tasks a chance to run.
803 The panic can be used in combination with panic_timeout,
804 to cause the system to reboot automatically after a
805 lockup has been detected. This feature is useful for
806 high-availability systems that have uptime guarantees and
807 where a lockup must be resolved ASAP.
811 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
813 depends on LOCKUP_DETECTOR
815 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
816 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
818 config DETECT_HUNG_TASK
819 bool "Detect Hung Tasks"
820 depends on DEBUG_KERNEL
821 default LOCKUP_DETECTOR
823 Say Y here to enable the kernel to detect "hung tasks",
824 which are bugs that cause the task to be stuck in
825 uninterruptible "D" state indefinitely.
827 When a hung task is detected, the kernel will print the
828 current stack trace (which you should report), but the
829 task will stay in uninterruptible state. If lockdep is
830 enabled then all held locks will also be reported. This
831 feature has negligible overhead.
833 config DEFAULT_HUNG_TASK_TIMEOUT
834 int "Default timeout for hung task detection (in seconds)"
835 depends on DETECT_HUNG_TASK
838 This option controls the default timeout (in seconds) used
839 to determine when a task has become non-responsive and should
842 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
843 sysctl or by writing a value to
844 /proc/sys/kernel/hung_task_timeout_secs.
846 A timeout of 0 disables the check. The default is two minutes.
847 Keeping the default should be fine in most cases.
849 config BOOTPARAM_HUNG_TASK_PANIC
850 bool "Panic (Reboot) On Hung Tasks"
851 depends on DETECT_HUNG_TASK
853 Say Y here to enable the kernel to panic on "hung tasks",
854 which are bugs that cause the kernel to leave a task stuck
855 in uninterruptible "D" state.
857 The panic can be used in combination with panic_timeout,
858 to cause the system to reboot automatically after a
859 hung task has been detected. This feature is useful for
860 high-availability systems that have uptime guarantees and
861 where a hung tasks must be resolved ASAP.
865 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
867 depends on DETECT_HUNG_TASK
869 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
870 default 1 if BOOTPARAM_HUNG_TASK_PANIC
873 bool "Detect Workqueue Stalls"
874 depends on DEBUG_KERNEL
876 Say Y here to enable stall detection on workqueues. If a
877 worker pool doesn't make forward progress on a pending work
878 item for over a given amount of time, 30s by default, a
879 warning message is printed along with dump of workqueue
880 state. This can be configured through kernel parameter
881 "workqueue.watchdog_thresh" and its sysfs counterpart.
883 endmenu # "Debug lockups and hangs"
888 Say Y here to enable the kernel to panic when it oopses. This
889 has the same effect as setting oops=panic on the kernel command
892 This feature is useful to ensure that the kernel does not do
893 anything erroneous after an oops which could result in data
894 corruption or other issues.
898 config PANIC_ON_OOPS_VALUE
901 default 0 if !PANIC_ON_OOPS
902 default 1 if PANIC_ON_OOPS
908 Set the timeout value (in seconds) until a reboot occurs when the
909 the kernel panics. If n = 0, then we wait forever. A timeout
910 value n > 0 will wait n seconds before rebooting, while a timeout
911 value n < 0 will reboot immediately.
914 bool "Collect scheduler debugging info"
915 depends on DEBUG_KERNEL && PROC_FS
918 If you say Y here, the /proc/sched_debug file will be provided
919 that can help debug the scheduler. The runtime overhead of this
927 bool "Collect scheduler statistics"
928 depends on DEBUG_KERNEL && PROC_FS
931 If you say Y here, additional code will be inserted into the
932 scheduler and related routines to collect statistics about
933 scheduler behavior and provide them in /proc/schedstat. These
934 stats may be useful for both tuning and debugging the scheduler
935 If you aren't debugging the scheduler or trying to tune a specific
936 application, you can say N to avoid the very slight overhead
939 config SCHED_STACK_END_CHECK
940 bool "Detect stack corruption on calls to schedule()"
941 depends on DEBUG_KERNEL
944 This option checks for a stack overrun on calls to schedule().
945 If the stack end location is found to be over written always panic as
946 the content of the corrupted region can no longer be trusted.
947 This is to ensure no erroneous behaviour occurs which could result in
948 data corruption or a sporadic crash at a later stage once the region
949 is examined. The runtime overhead introduced is minimal.
951 config DEBUG_TIMEKEEPING
952 bool "Enable extra timekeeping sanity checking"
954 This option will enable additional timekeeping sanity checks
955 which may be helpful when diagnosing issues where timekeeping
956 problems are suspected.
958 This may include checks in the timekeeping hotpaths, so this
959 option may have a (very small) performance impact to some
965 bool "Collect kernel timers statistics"
966 depends on DEBUG_KERNEL && PROC_FS
968 If you say Y here, additional code will be inserted into the
969 timer routines to collect statistics about kernel timers being
970 reprogrammed. The statistics can be read from /proc/timer_stats.
971 The statistics collection is started by writing 1 to /proc/timer_stats,
972 writing 0 stops it. This feature is useful to collect information
973 about timer usage patterns in kernel and userspace. This feature
974 is lightweight if enabled in the kernel config but not activated
975 (it defaults to deactivated on bootup and will only be activated
976 if some application like powertop activates it explicitly).
979 bool "Debug preemptible kernel"
980 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
983 If you say Y here then the kernel will use a debug variant of the
984 commonly used smp_processor_id() function and will print warnings
985 if kernel code uses it in a preemption-unsafe way. Also, the kernel
986 will detect preemption count underflows.
988 menu "Lock Debugging (spinlocks, mutexes, etc...)"
990 config DEBUG_RT_MUTEXES
991 bool "RT Mutex debugging, deadlock detection"
992 depends on DEBUG_KERNEL && RT_MUTEXES
994 This allows rt mutex semantics violations and rt mutex related
995 deadlocks (lockups) to be detected and reported automatically.
997 config DEBUG_SPINLOCK
998 bool "Spinlock and rw-lock debugging: basic checks"
999 depends on DEBUG_KERNEL
1000 select UNINLINE_SPIN_UNLOCK
1002 Say Y here and build SMP to catch missing spinlock initialization
1003 and certain other kinds of spinlock errors commonly made. This is
1004 best used in conjunction with the NMI watchdog so that spinlock
1005 deadlocks are also debuggable.
1007 config DEBUG_MUTEXES
1008 bool "Mutex debugging: basic checks"
1009 depends on DEBUG_KERNEL
1011 This feature allows mutex semantics violations to be detected and
1014 config DEBUG_WW_MUTEX_SLOWPATH
1015 bool "Wait/wound mutex debugging: Slowpath testing"
1016 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1017 select DEBUG_LOCK_ALLOC
1018 select DEBUG_SPINLOCK
1019 select DEBUG_MUTEXES
1021 This feature enables slowpath testing for w/w mutex users by
1022 injecting additional -EDEADLK wound/backoff cases. Together with
1023 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this
1024 will test all possible w/w mutex interface abuse with the
1025 exception of simply not acquiring all the required locks.
1026 Note that this feature can introduce significant overhead, so
1027 it really should not be enabled in a production or distro kernel,
1028 even a debug kernel. If you are a driver writer, enable it. If
1029 you are a distro, do not.
1031 config DEBUG_LOCK_ALLOC
1032 bool "Lock debugging: detect incorrect freeing of live locks"
1033 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1034 select DEBUG_SPINLOCK
1035 select DEBUG_MUTEXES
1038 This feature will check whether any held lock (spinlock, rwlock,
1039 mutex or rwsem) is incorrectly freed by the kernel, via any of the
1040 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
1041 vfree(), etc.), whether a live lock is incorrectly reinitialized via
1042 spin_lock_init()/mutex_init()/etc., or whether there is any lock
1043 held during task exit.
1045 config PROVE_LOCKING
1046 bool "Lock debugging: prove locking correctness"
1047 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1049 select DEBUG_SPINLOCK
1050 select DEBUG_MUTEXES
1051 select DEBUG_LOCK_ALLOC
1052 select TRACE_IRQFLAGS
1055 This feature enables the kernel to prove that all locking
1056 that occurs in the kernel runtime is mathematically
1057 correct: that under no circumstance could an arbitrary (and
1058 not yet triggered) combination of observed locking
1059 sequences (on an arbitrary number of CPUs, running an
1060 arbitrary number of tasks and interrupt contexts) cause a
1063 In short, this feature enables the kernel to report locking
1064 related deadlocks before they actually occur.
1066 The proof does not depend on how hard and complex a
1067 deadlock scenario would be to trigger: how many
1068 participant CPUs, tasks and irq-contexts would be needed
1069 for it to trigger. The proof also does not depend on
1070 timing: if a race and a resulting deadlock is possible
1071 theoretically (no matter how unlikely the race scenario
1072 is), it will be proven so and will immediately be
1073 reported by the kernel (once the event is observed that
1074 makes the deadlock theoretically possible).
1076 If a deadlock is impossible (i.e. the locking rules, as
1077 observed by the kernel, are mathematically correct), the
1078 kernel reports nothing.
1080 NOTE: this feature can also be enabled for rwlocks, mutexes
1081 and rwsems - in which case all dependencies between these
1082 different locking variants are observed and mapped too, and
1083 the proof of observed correctness is also maintained for an
1084 arbitrary combination of these separate locking variants.
1086 For more details, see Documentation/locking/lockdep-design.txt.
1090 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1092 select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE && !ARC && !SCORE
1097 bool "Lock usage statistics"
1098 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1100 select DEBUG_SPINLOCK
1101 select DEBUG_MUTEXES
1102 select DEBUG_LOCK_ALLOC
1105 This feature enables tracking lock contention points
1107 For more details, see Documentation/locking/lockstat.txt
1109 This also enables lock events required by "perf lock",
1111 If you want to use "perf lock", you also need to turn on
1112 CONFIG_EVENT_TRACING.
1114 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
1115 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
1117 config DEBUG_LOCKDEP
1118 bool "Lock dependency engine debugging"
1119 depends on DEBUG_KERNEL && LOCKDEP
1121 If you say Y here, the lock dependency engine will do
1122 additional runtime checks to debug itself, at the price
1123 of more runtime overhead.
1125 config DEBUG_ATOMIC_SLEEP
1126 bool "Sleep inside atomic section checking"
1127 select PREEMPT_COUNT
1128 depends on DEBUG_KERNEL
1130 If you say Y here, various routines which may sleep will become very
1131 noisy if they are called inside atomic sections: when a spinlock is
1132 held, inside an rcu read side critical section, inside preempt disabled
1133 sections, inside an interrupt, etc...
1135 config DEBUG_LOCKING_API_SELFTESTS
1136 bool "Locking API boot-time self-tests"
1137 depends on DEBUG_KERNEL
1139 Say Y here if you want the kernel to run a short self-test during
1140 bootup. The self-test checks whether common types of locking bugs
1141 are detected by debugging mechanisms or not. (if you disable
1142 lock debugging then those bugs wont be detected of course.)
1143 The following locking APIs are covered: spinlocks, rwlocks,
1146 config LOCK_TORTURE_TEST
1147 tristate "torture tests for locking"
1148 depends on DEBUG_KERNEL
1152 This option provides a kernel module that runs torture tests
1153 on kernel locking primitives. The kernel module may be built
1154 after the fact on the running kernel to be tested, if desired.
1156 Say Y here if you want kernel locking-primitive torture tests
1157 to be built into the kernel.
1158 Say M if you want these torture tests to build as a module.
1159 Say N if you are unsure.
1161 endmenu # lock debugging
1163 config TRACE_IRQFLAGS
1166 Enables hooks to interrupt enabling and disabling for
1167 either tracing or lock debugging.
1170 bool "Stack backtrace support"
1171 depends on STACKTRACE_SUPPORT
1173 This option causes the kernel to create a /proc/pid/stack for
1174 every process, showing its current stack trace.
1175 It is also used by various kernel debugging features that require
1176 stack trace generation.
1178 config DEBUG_KOBJECT
1179 bool "kobject debugging"
1180 depends on DEBUG_KERNEL
1182 If you say Y here, some extra kobject debugging messages will be sent
1185 config DEBUG_KOBJECT_RELEASE
1186 bool "kobject release debugging"
1187 depends on DEBUG_OBJECTS_TIMERS
1189 kobjects are reference counted objects. This means that their
1190 last reference count put is not predictable, and the kobject can
1191 live on past the point at which a driver decides to drop it's
1192 initial reference to the kobject gained on allocation. An
1193 example of this would be a struct device which has just been
1196 However, some buggy drivers assume that after such an operation,
1197 the memory backing the kobject can be immediately freed. This
1198 goes completely against the principles of a refcounted object.
1200 If you say Y here, the kernel will delay the release of kobjects
1201 on the last reference count to improve the visibility of this
1202 kind of kobject release bug.
1204 config HAVE_DEBUG_BUGVERBOSE
1207 config DEBUG_BUGVERBOSE
1208 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
1209 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
1212 Say Y here to make BUG() panics output the file name and line number
1213 of the BUG call as well as the EIP and oops trace. This aids
1214 debugging but costs about 70-100K of memory.
1217 bool "Debug linked list manipulation"
1218 depends on DEBUG_KERNEL
1220 Enable this to turn on extended checks in the linked-list
1225 config DEBUG_PI_LIST
1226 bool "Debug priority linked list manipulation"
1227 depends on DEBUG_KERNEL
1229 Enable this to turn on extended checks in the priority-ordered
1230 linked-list (plist) walking routines. This checks the entire
1231 list multiple times during each manipulation.
1236 bool "Debug SG table operations"
1237 depends on DEBUG_KERNEL
1239 Enable this to turn on checks on scatter-gather tables. This can
1240 help find problems with drivers that do not properly initialize
1245 config DEBUG_NOTIFIERS
1246 bool "Debug notifier call chains"
1247 depends on DEBUG_KERNEL
1249 Enable this to turn on sanity checking for notifier call chains.
1250 This is most useful for kernel developers to make sure that
1251 modules properly unregister themselves from notifier chains.
1252 This is a relatively cheap check but if you care about maximum
1255 config DEBUG_CREDENTIALS
1256 bool "Debug credential management"
1257 depends on DEBUG_KERNEL
1259 Enable this to turn on some debug checking for credential
1260 management. The additional code keeps track of the number of
1261 pointers from task_structs to any given cred struct, and checks to
1262 see that this number never exceeds the usage count of the cred
1265 Furthermore, if SELinux is enabled, this also checks that the
1266 security pointer in the cred struct is never seen to be invalid.
1270 menu "RCU Debugging"
1273 def_bool PROVE_LOCKING
1275 config PROVE_RCU_REPEATEDLY
1276 bool "RCU debugging: don't disable PROVE_RCU on first splat"
1277 depends on PROVE_RCU
1280 By itself, PROVE_RCU will disable checking upon issuing the
1281 first warning (or "splat"). This feature prevents such
1282 disabling, allowing multiple RCU-lockdep warnings to be printed
1285 Say Y to allow multiple RCU-lockdep warnings per boot.
1287 Say N if you are unsure.
1289 config SPARSE_RCU_POINTER
1290 bool "RCU debugging: sparse-based checks for pointer usage"
1293 This feature enables the __rcu sparse annotation for
1294 RCU-protected pointers. This annotation will cause sparse
1295 to flag any non-RCU used of annotated pointers. This can be
1296 helpful when debugging RCU usage. Please note that this feature
1297 is not intended to enforce code cleanliness; it is instead merely
1300 Say Y to make sparse flag questionable use of RCU-protected pointers
1302 Say N if you are unsure.
1308 config RCU_PERF_TEST
1309 tristate "performance tests for RCU"
1310 depends on DEBUG_KERNEL
1316 This option provides a kernel module that runs performance
1317 tests on the RCU infrastructure. The kernel module may be built
1318 after the fact on the running kernel to be tested, if desired.
1320 Say Y here if you want RCU performance tests to be built into
1322 Say M if you want the RCU performance tests to build as a module.
1323 Say N if you are unsure.
1325 config RCU_TORTURE_TEST
1326 tristate "torture tests for RCU"
1327 depends on DEBUG_KERNEL
1333 This option provides a kernel module that runs torture tests
1334 on the RCU infrastructure. The kernel module may be built
1335 after the fact on the running kernel to be tested, if desired.
1337 Say Y here if you want RCU torture tests to be built into
1339 Say M if you want the RCU torture tests to build as a module.
1340 Say N if you are unsure.
1342 config RCU_TORTURE_TEST_SLOW_PREINIT
1343 bool "Slow down RCU grace-period pre-initialization to expose races"
1344 depends on RCU_TORTURE_TEST
1346 This option delays grace-period pre-initialization (the
1347 propagation of CPU-hotplug changes up the rcu_node combining
1348 tree) for a few jiffies between initializing each pair of
1349 consecutive rcu_node structures. This helps to expose races
1350 involving grace-period pre-initialization, in other words, it
1351 makes your kernel less stable. It can also greatly increase
1352 grace-period latency, especially on systems with large numbers
1353 of CPUs. This is useful when torture-testing RCU, but in
1354 almost no other circumstance.
1356 Say Y here if you want your system to crash and hang more often.
1357 Say N if you want a sane system.
1359 config RCU_TORTURE_TEST_SLOW_PREINIT_DELAY
1360 int "How much to slow down RCU grace-period pre-initialization"
1363 depends on RCU_TORTURE_TEST_SLOW_PREINIT
1365 This option specifies the number of jiffies to wait between
1366 each rcu_node structure pre-initialization step.
1368 config RCU_TORTURE_TEST_SLOW_INIT
1369 bool "Slow down RCU grace-period initialization to expose races"
1370 depends on RCU_TORTURE_TEST
1372 This option delays grace-period initialization for a few
1373 jiffies between initializing each pair of consecutive
1374 rcu_node structures. This helps to expose races involving
1375 grace-period initialization, in other words, it makes your
1376 kernel less stable. It can also greatly increase grace-period
1377 latency, especially on systems with large numbers of CPUs.
1378 This is useful when torture-testing RCU, but in almost no
1381 Say Y here if you want your system to crash and hang more often.
1382 Say N if you want a sane system.
1384 config RCU_TORTURE_TEST_SLOW_INIT_DELAY
1385 int "How much to slow down RCU grace-period initialization"
1388 depends on RCU_TORTURE_TEST_SLOW_INIT
1390 This option specifies the number of jiffies to wait between
1391 each rcu_node structure initialization.
1393 config RCU_TORTURE_TEST_SLOW_CLEANUP
1394 bool "Slow down RCU grace-period cleanup to expose races"
1395 depends on RCU_TORTURE_TEST
1397 This option delays grace-period cleanup for a few jiffies
1398 between cleaning up each pair of consecutive rcu_node
1399 structures. This helps to expose races involving grace-period
1400 cleanup, in other words, it makes your kernel less stable.
1401 It can also greatly increase grace-period latency, especially
1402 on systems with large numbers of CPUs. This is useful when
1403 torture-testing RCU, but in almost no other circumstance.
1405 Say Y here if you want your system to crash and hang more often.
1406 Say N if you want a sane system.
1408 config RCU_TORTURE_TEST_SLOW_CLEANUP_DELAY
1409 int "How much to slow down RCU grace-period cleanup"
1412 depends on RCU_TORTURE_TEST_SLOW_CLEANUP
1414 This option specifies the number of jiffies to wait between
1415 each rcu_node structure cleanup operation.
1417 config RCU_CPU_STALL_TIMEOUT
1418 int "RCU CPU stall timeout in seconds"
1419 depends on RCU_STALL_COMMON
1423 If a given RCU grace period extends more than the specified
1424 number of seconds, a CPU stall warning is printed. If the
1425 RCU grace period persists, additional CPU stall warnings are
1426 printed at more widely spaced intervals.
1429 bool "Enable tracing for RCU"
1430 depends on DEBUG_KERNEL
1433 This option provides tracing in RCU which presents stats
1434 in debugfs for debugging RCU implementation.
1436 Say Y here if you want to enable RCU tracing
1437 Say N if you are unsure.
1439 config RCU_EQS_DEBUG
1440 bool "Provide debugging asserts for adding NO_HZ support to an arch"
1441 depends on DEBUG_KERNEL
1443 This option provides consistency checks in RCU's handling of
1444 NO_HZ. These checks have proven quite helpful in detecting
1445 bugs in arch-specific NO_HZ code.
1447 Say N here if you need ultimate kernel/user switch latencies
1448 Say Y if you are unsure
1450 endmenu # "RCU Debugging"
1452 config DEBUG_WQ_FORCE_RR_CPU
1453 bool "Force round-robin CPU selection for unbound work items"
1454 depends on DEBUG_KERNEL
1457 Workqueue used to implicitly guarantee that work items queued
1458 without explicit CPU specified are put on the local CPU. This
1459 guarantee is no longer true and while local CPU is still
1460 preferred work items may be put on foreign CPUs. Kernel
1461 parameter "workqueue.debug_force_rr_cpu" is added to force
1462 round-robin CPU selection to flush out usages which depend on the
1463 now broken guarantee. This config option enables the debug
1464 feature by default. When enabled, memory and cache locality will
1467 config DEBUG_BLOCK_EXT_DEVT
1468 bool "Force extended block device numbers and spread them"
1469 depends on DEBUG_KERNEL
1473 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1474 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1475 YOU ARE DOING. Distros, please enable this and fix whatever
1478 Conventionally, block device numbers are allocated from
1479 predetermined contiguous area. However, extended block area
1480 may introduce non-contiguous block device numbers. This
1481 option forces most block device numbers to be allocated from
1482 the extended space and spreads them to discover kernel or
1483 userland code paths which assume predetermined contiguous
1484 device number allocation.
1486 Note that turning on this debug option shuffles all the
1487 device numbers for all IDE and SCSI devices including libata
1488 ones, so root partition specified using device number
1489 directly (via rdev or root=MAJ:MIN) won't work anymore.
1490 Textual device names (root=/dev/sdXn) will continue to work.
1492 Say N if you are unsure.
1494 config CPU_HOTPLUG_STATE_CONTROL
1495 bool "Enable CPU hotplug state control"
1496 depends on DEBUG_KERNEL
1497 depends on HOTPLUG_CPU
1500 Allows to write steps between "offline" and "online" to the CPUs
1501 sysfs target file so states can be stepped granular. This is a debug
1502 option for now as the hotplug machinery cannot be stopped and
1503 restarted at arbitrary points yet.
1505 Say N if your are unsure.
1507 config NOTIFIER_ERROR_INJECTION
1508 tristate "Notifier error injection"
1509 depends on DEBUG_KERNEL
1512 This option provides the ability to inject artificial errors to
1513 specified notifier chain callbacks. It is useful to test the error
1514 handling of notifier call chain failures.
1518 config CPU_NOTIFIER_ERROR_INJECT
1519 tristate "CPU notifier error injection module"
1520 depends on HOTPLUG_CPU && NOTIFIER_ERROR_INJECTION
1522 This option provides a kernel module that can be used to test
1523 the error handling of the cpu notifiers by injecting artificial
1524 errors to CPU notifier chain callbacks. It is controlled through
1525 debugfs interface under /sys/kernel/debug/notifier-error-inject/cpu
1527 If the notifier call chain should be failed with some events
1528 notified, write the error code to "actions/<notifier event>/error".
1530 Example: Inject CPU offline error (-1 == -EPERM)
1532 # cd /sys/kernel/debug/notifier-error-inject/cpu
1533 # echo -1 > actions/CPU_DOWN_PREPARE/error
1534 # echo 0 > /sys/devices/system/cpu/cpu1/online
1535 bash: echo: write error: Operation not permitted
1537 To compile this code as a module, choose M here: the module will
1538 be called cpu-notifier-error-inject.
1542 config PM_NOTIFIER_ERROR_INJECT
1543 tristate "PM notifier error injection module"
1544 depends on PM && NOTIFIER_ERROR_INJECTION
1545 default m if PM_DEBUG
1547 This option provides the ability to inject artificial errors to
1548 PM notifier chain callbacks. It is controlled through debugfs
1549 interface /sys/kernel/debug/notifier-error-inject/pm
1551 If the notifier call chain should be failed with some events
1552 notified, write the error code to "actions/<notifier event>/error".
1554 Example: Inject PM suspend error (-12 = -ENOMEM)
1556 # cd /sys/kernel/debug/notifier-error-inject/pm/
1557 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1558 # echo mem > /sys/power/state
1559 bash: echo: write error: Cannot allocate memory
1561 To compile this code as a module, choose M here: the module will
1562 be called pm-notifier-error-inject.
1566 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1567 tristate "OF reconfig notifier error injection module"
1568 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1570 This option provides the ability to inject artificial errors to
1571 OF reconfig notifier chain callbacks. It is controlled
1572 through debugfs interface under
1573 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1575 If the notifier call chain should be failed with some events
1576 notified, write the error code to "actions/<notifier event>/error".
1578 To compile this code as a module, choose M here: the module will
1579 be called of-reconfig-notifier-error-inject.
1583 config NETDEV_NOTIFIER_ERROR_INJECT
1584 tristate "Netdev notifier error injection module"
1585 depends on NET && NOTIFIER_ERROR_INJECTION
1587 This option provides the ability to inject artificial errors to
1588 netdevice notifier chain callbacks. It is controlled through debugfs
1589 interface /sys/kernel/debug/notifier-error-inject/netdev
1591 If the notifier call chain should be failed with some events
1592 notified, write the error code to "actions/<notifier event>/error".
1594 Example: Inject netdevice mtu change error (-22 = -EINVAL)
1596 # cd /sys/kernel/debug/notifier-error-inject/netdev
1597 # echo -22 > actions/NETDEV_CHANGEMTU/error
1598 # ip link set eth0 mtu 1024
1599 RTNETLINK answers: Invalid argument
1601 To compile this code as a module, choose M here: the module will
1602 be called netdev-notifier-error-inject.
1606 config FAULT_INJECTION
1607 bool "Fault-injection framework"
1608 depends on DEBUG_KERNEL
1610 Provide fault-injection framework.
1611 For more details, see Documentation/fault-injection/.
1614 bool "Fault-injection capability for kmalloc"
1615 depends on FAULT_INJECTION
1616 depends on SLAB || SLUB
1618 Provide fault-injection capability for kmalloc.
1620 config FAIL_PAGE_ALLOC
1621 bool "Fault-injection capabilitiy for alloc_pages()"
1622 depends on FAULT_INJECTION
1624 Provide fault-injection capability for alloc_pages().
1626 config FAIL_MAKE_REQUEST
1627 bool "Fault-injection capability for disk IO"
1628 depends on FAULT_INJECTION && BLOCK
1630 Provide fault-injection capability for disk IO.
1632 config FAIL_IO_TIMEOUT
1633 bool "Fault-injection capability for faking disk interrupts"
1634 depends on FAULT_INJECTION && BLOCK
1636 Provide fault-injection capability on end IO handling. This
1637 will make the block layer "forget" an interrupt as configured,
1638 thus exercising the error handling.
1640 Only works with drivers that use the generic timeout handling,
1641 for others it wont do anything.
1643 config FAIL_MMC_REQUEST
1644 bool "Fault-injection capability for MMC IO"
1645 depends on FAULT_INJECTION_DEBUG_FS && MMC
1647 Provide fault-injection capability for MMC IO.
1648 This will make the mmc core return data errors. This is
1649 useful to test the error handling in the mmc block device
1650 and to test how the mmc host driver handles retries from
1654 bool "Fault-injection capability for futexes"
1656 depends on FAULT_INJECTION && FUTEX
1658 Provide fault-injection capability for futexes.
1660 config FAULT_INJECTION_DEBUG_FS
1661 bool "Debugfs entries for fault-injection capabilities"
1662 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1664 Enable configuration of fault-injection capabilities via debugfs.
1666 config FAULT_INJECTION_STACKTRACE_FILTER
1667 bool "stacktrace filter for fault-injection capabilities"
1668 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1671 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !SCORE
1673 Provide stacktrace filter for fault-injection capabilities
1676 bool "Latency measuring infrastructure"
1677 depends on DEBUG_KERNEL
1678 depends on STACKTRACE_SUPPORT
1680 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC
1687 Enable this option if you want to use the LatencyTOP tool
1688 to find out which userspace is blocking on what kernel operations.
1690 source kernel/trace/Kconfig
1692 menu "Runtime Testing"
1695 tristate "Linux Kernel Dump Test Tool Module"
1700 This module enables testing of the different dumping mechanisms by
1701 inducing system failures at predefined crash points.
1702 If you don't need it: say N
1703 Choose M here to compile this code as a module. The module will be
1706 Documentation on how to use the module can be found in
1707 Documentation/fault-injection/provoke-crashes.txt
1709 config TEST_LIST_SORT
1710 bool "Linked list sorting test"
1711 depends on DEBUG_KERNEL
1713 Enable this to turn on 'list_sort()' function test. This test is
1714 executed only once during system boot, so affects only boot time.
1718 config KPROBES_SANITY_TEST
1719 bool "Kprobes sanity tests"
1720 depends on DEBUG_KERNEL
1724 This option provides for testing basic kprobes functionality on
1725 boot. A sample kprobe, jprobe and kretprobe are inserted and
1726 verified for functionality.
1728 Say N if you are unsure.
1730 config BACKTRACE_SELF_TEST
1731 tristate "Self test for the backtrace code"
1732 depends on DEBUG_KERNEL
1735 This option provides a kernel module that can be used to test
1736 the kernel stack backtrace code. This option is not useful
1737 for distributions or general kernels, but only for kernel
1738 developers working on architecture code.
1740 Note that if you want to also test saved backtraces, you will
1741 have to enable STACKTRACE as well.
1743 Say N if you are unsure.
1746 tristate "Red-Black tree test"
1747 depends on DEBUG_KERNEL
1749 A benchmark measuring the performance of the rbtree library.
1750 Also includes rbtree invariant checks.
1752 config INTERVAL_TREE_TEST
1753 tristate "Interval tree test"
1754 depends on m && DEBUG_KERNEL
1755 select INTERVAL_TREE
1757 A benchmark measuring the performance of the interval tree library
1760 tristate "Per cpu operations test"
1761 depends on m && DEBUG_KERNEL
1763 Enable this option to build test module which validates per-cpu
1768 config ATOMIC64_SELFTEST
1769 bool "Perform an atomic64_t self-test at boot"
1771 Enable this option to test the atomic64_t functions at boot.
1775 config ASYNC_RAID6_TEST
1776 tristate "Self test for hardware accelerated raid6 recovery"
1777 depends on ASYNC_RAID6_RECOV
1780 This is a one-shot self test that permutes through the
1781 recovery of all the possible two disk failure scenarios for a
1782 N-disk array. Recovery is performed with the asynchronous
1783 raid6 recovery routines, and will optionally use an offload
1784 engine if one is available.
1789 tristate "Test functions located in the hexdump module at runtime"
1791 config TEST_STRING_HELPERS
1792 tristate "Test functions located in the string_helpers module at runtime"
1795 tristate "Test kstrto*() family of functions at runtime"
1798 tristate "Test printf() family of functions at runtime"
1801 tristate "Test bitmap_*() family of functions at runtime"
1804 Enable this option to test the bitmap functions at boot.
1809 tristate "Test functions located in the uuid module at runtime"
1811 config TEST_RHASHTABLE
1812 tristate "Perform selftest on resizable hash table"
1815 Enable this option to test the rhashtable functions at boot.
1820 tristate "Perform selftest on hash functions"
1823 Enable this option to test the kernel's integer (<linux/hash,h>)
1824 and string (<linux/stringhash.h>) hash functions on boot
1827 This is intended to help people writing architecture-specific
1828 optimized versions. If unsure, say N.
1830 endmenu # runtime tests
1832 config PROVIDE_OHCI1394_DMA_INIT
1833 bool "Remote debugging over FireWire early on boot"
1834 depends on PCI && X86
1836 If you want to debug problems which hang or crash the kernel early
1837 on boot and the crashing machine has a FireWire port, you can use
1838 this feature to remotely access the memory of the crashed machine
1839 over FireWire. This employs remote DMA as part of the OHCI1394
1840 specification which is now the standard for FireWire controllers.
1842 With remote DMA, you can monitor the printk buffer remotely using
1843 firescope and access all memory below 4GB using fireproxy from gdb.
1844 Even controlling a kernel debugger is possible using remote DMA.
1848 If ohci1394_dma=early is used as boot parameter, it will initialize
1849 all OHCI1394 controllers which are found in the PCI config space.
1851 As all changes to the FireWire bus such as enabling and disabling
1852 devices cause a bus reset and thereby disable remote DMA for all
1853 devices, be sure to have the cable plugged and FireWire enabled on
1854 the debugging host before booting the debug target for debugging.
1856 This code (~1k) is freed after boot. By then, the firewire stack
1857 in charge of the OHCI-1394 controllers should be used instead.
1859 See Documentation/debugging-via-ohci1394.txt for more information.
1861 config DMA_API_DEBUG
1862 bool "Enable debugging of DMA-API usage"
1863 depends on HAVE_DMA_API_DEBUG
1865 Enable this option to debug the use of the DMA API by device drivers.
1866 With this option you will be able to detect common bugs in device
1867 drivers like double-freeing of DMA mappings or freeing mappings that
1868 were never allocated.
1870 This also attempts to catch cases where a page owned by DMA is
1871 accessed by the cpu in a way that could cause data corruption. For
1872 example, this enables cow_user_page() to check that the source page is
1875 This option causes a performance degradation. Use only if you want to
1876 debug device drivers and dma interactions.
1881 tristate "Test module loading with 'hello world' module"
1885 This builds the "test_module" module that emits "Hello, world"
1886 on printk when loaded. It is designed to be used for basic
1887 evaluation of the module loading subsystem (for example when
1888 validating module verification). It lacks any extra dependencies,
1889 and will not normally be loaded by the system unless explicitly
1894 config TEST_USER_COPY
1895 tristate "Test user/kernel boundary protections"
1899 This builds the "test_user_copy" module that runs sanity checks
1900 on the copy_to/from_user infrastructure, making sure basic
1901 user/kernel boundary testing is working. If it fails to load,
1902 a regression has been detected in the user/kernel memory boundary
1908 tristate "Test BPF filter functionality"
1912 This builds the "test_bpf" module that runs various test vectors
1913 against the BPF interpreter or BPF JIT compiler depending on the
1914 current setting. This is in particular useful for BPF JIT compiler
1915 development, but also to run regression tests against changes in
1916 the interpreter code. It also enables test stubs for eBPF maps and
1917 verifier used by user space verifier testsuite.
1921 config TEST_FIRMWARE
1922 tristate "Test firmware loading via userspace interface"
1924 depends on FW_LOADER
1926 This builds the "test_firmware" module that creates a userspace
1927 interface for testing firmware loading. This can be used to
1928 control the triggering of firmware loading without needing an
1929 actual firmware-using device. The contents can be rechecked by
1935 tristate "udelay test driver"
1938 This builds the "udelay_test" module that helps to make sure
1939 that udelay() is working properly.
1945 depends on HAVE_MEMBLOCK
1947 This option adds a kernel parameter 'memtest', which allows memtest
1949 memtest=0, mean disabled; -- default
1950 memtest=1, mean do 1 test pattern;
1952 memtest=17, mean do 17 test patterns.
1953 If you are unsure how to answer this question, answer N.
1955 config TEST_STATIC_KEYS
1956 tristate "Test static keys"
1960 Test the static key interfaces.
1964 source "samples/Kconfig"
1966 source "lib/Kconfig.kgdb"
1968 source "lib/Kconfig.ubsan"
1970 config ARCH_HAS_DEVMEM_IS_ALLOWED
1973 config STRICT_DEVMEM
1974 bool "Filter access to /dev/mem"
1976 depends on ARCH_HAS_DEVMEM_IS_ALLOWED
1977 default y if TILE || PPC
1979 If this option is disabled, you allow userspace (root) access to all
1980 of memory, including kernel and userspace memory. Accidental
1981 access to this is obviously disastrous, but specific access can
1982 be used by people debugging the kernel. Note that with PAT support
1983 enabled, even in this case there are restrictions on /dev/mem
1984 use due to the cache aliasing requirements.
1986 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem
1987 file only allows userspace access to PCI space and the BIOS code and
1988 data regions. This is sufficient for dosemu and X and all common
1993 config IO_STRICT_DEVMEM
1994 bool "Filter I/O access to /dev/mem"
1995 depends on STRICT_DEVMEM
1997 If this option is disabled, you allow userspace (root) access to all
1998 io-memory regardless of whether a driver is actively using that
1999 range. Accidental access to this is obviously disastrous, but
2000 specific access can be used by people debugging kernel drivers.
2002 If this option is switched on, the /dev/mem file only allows
2003 userspace access to *idle* io-memory ranges (see /proc/iomem) This
2004 may break traditional users of /dev/mem (dosemu, legacy X, etc...)
2005 if the driver using a given range cannot be disabled.