3 menu "printk and dmesg options"
6 bool "Show timing information on printks"
9 Selecting this option causes time stamps of the printk()
10 messages to be added to the output of the syslog() system
11 call and at the console.
13 The timestamp is always recorded internally, and exported
14 to /dev/kmsg. This flag just specifies if the timestamp should
15 be included, not that the timestamp is recorded.
17 The behavior is also controlled by the kernel command line
18 parameter printk.time=1. See Documentation/admin-guide/kernel-parameters.rst
21 bool "Show caller information on printks"
24 Selecting this option causes printk() to add a caller "thread id" (if
25 in task context) or a caller "processor id" (if not in task context)
28 This option is intended for environments where multiple threads
29 concurrently call printk() for many times, for it is difficult to
30 interpret without knowing where these lines (or sometimes individual
31 line which was divided into multiple lines due to race) came from.
33 Since toggling after boot makes the code racy, currently there is
34 no option to enable/disable at the kernel command line parameter or
37 config CONSOLE_LOGLEVEL_DEFAULT
38 int "Default console loglevel (1-15)"
42 Default loglevel to determine what will be printed on the console.
44 Setting a default here is equivalent to passing in loglevel=<x> in
45 the kernel bootargs. loglevel=<x> continues to override whatever
46 value is specified here as well.
48 Note: This does not affect the log level of un-prefixed printk()
49 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT
52 config CONSOLE_LOGLEVEL_QUIET
53 int "quiet console loglevel (1-15)"
57 loglevel to use when "quiet" is passed on the kernel commandline.
59 When "quiet" is passed on the kernel commandline this loglevel
60 will be used as the loglevel. IOW passing "quiet" will be the
61 equivalent of passing "loglevel=<CONSOLE_LOGLEVEL_QUIET>"
63 config MESSAGE_LOGLEVEL_DEFAULT
64 int "Default message log level (1-7)"
68 Default log level for printk statements with no specified priority.
70 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
71 that are auditing their logs closely may want to set it to a lower
74 Note: This does not affect what message level gets printed on the console
75 by default. To change that, use loglevel=<x> in the kernel bootargs,
76 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value.
78 config BOOT_PRINTK_DELAY
79 bool "Delay each boot printk message by N milliseconds"
80 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
82 This build option allows you to read kernel boot messages
83 by inserting a short delay after each one. The delay is
84 specified in milliseconds on the kernel command line,
87 It is likely that you would also need to use "lpj=M" to preset
88 the "loops per jiffie" value.
89 See a previous boot log for the "lpj" value to use for your
90 system, and then set "lpj=M" before setting "boot_delay=N".
91 NOTE: Using this option may adversely affect SMP systems.
92 I.e., processors other than the first one may not boot up.
93 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
94 what it believes to be lockup conditions.
97 bool "Enable dynamic printk() support"
103 Compiles debug level messages into the kernel, which would not
104 otherwise be available at runtime. These messages can then be
105 enabled/disabled based on various levels of scope - per source file,
106 function, module, format string, and line number. This mechanism
107 implicitly compiles in all pr_debug() and dev_dbg() calls, which
108 enlarges the kernel text size by about 2%.
110 If a source file is compiled with DEBUG flag set, any
111 pr_debug() calls in it are enabled by default, but can be
112 disabled at runtime as below. Note that DEBUG flag is
113 turned on by many CONFIG_*DEBUG* options.
117 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
118 which is contained in the 'debugfs' filesystem. Thus, the debugfs
119 filesystem must first be mounted before making use of this feature.
120 We refer the control file as: <debugfs>/dynamic_debug/control. This
121 file contains a list of the debug statements that can be enabled. The
122 format for each line of the file is:
124 filename:lineno [module]function flags format
126 filename : source file of the debug statement
127 lineno : line number of the debug statement
128 module : module that contains the debug statement
129 function : function that contains the debug statement
130 flags : '=p' means the line is turned 'on' for printing
131 format : the format used for the debug statement
135 nullarbor:~ # cat <debugfs>/dynamic_debug/control
136 # filename:lineno [module]function flags format
137 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012"
138 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012"
139 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012"
143 // enable the message at line 1603 of file svcsock.c
144 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
145 <debugfs>/dynamic_debug/control
147 // enable all the messages in file svcsock.c
148 nullarbor:~ # echo -n 'file svcsock.c +p' >
149 <debugfs>/dynamic_debug/control
151 // enable all the messages in the NFS server module
152 nullarbor:~ # echo -n 'module nfsd +p' >
153 <debugfs>/dynamic_debug/control
155 // enable all 12 messages in the function svc_process()
156 nullarbor:~ # echo -n 'func svc_process +p' >
157 <debugfs>/dynamic_debug/control
159 // disable all 12 messages in the function svc_process()
160 nullarbor:~ # echo -n 'func svc_process -p' >
161 <debugfs>/dynamic_debug/control
163 See Documentation/admin-guide/dynamic-debug-howto.rst for additional
166 endmenu # "printk and dmesg options"
168 menu "Compile-time checks and compiler options"
171 bool "Compile the kernel with debug info"
172 depends on DEBUG_KERNEL && !COMPILE_TEST
174 If you say Y here the resulting kernel image will include
175 debugging info resulting in a larger kernel image.
176 This adds debug symbols to the kernel and modules (gcc -g), and
177 is needed if you intend to use kernel crashdump or binary object
178 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
179 Say Y here only if you plan to debug the kernel.
183 config DEBUG_INFO_REDUCED
184 bool "Reduce debugging information"
185 depends on DEBUG_INFO
187 If you say Y here gcc is instructed to generate less debugging
188 information for structure types. This means that tools that
189 need full debugging information (like kgdb or systemtap) won't
190 be happy. But if you merely need debugging information to
191 resolve line numbers there is no loss. Advantage is that
192 build directory object sizes shrink dramatically over a full
193 DEBUG_INFO build and compile times are reduced too.
194 Only works with newer gcc versions.
196 config DEBUG_INFO_SPLIT
197 bool "Produce split debuginfo in .dwo files"
198 depends on DEBUG_INFO
199 depends on $(cc-option,-gsplit-dwarf)
201 Generate debug info into separate .dwo files. This significantly
202 reduces the build directory size for builds with DEBUG_INFO,
203 because it stores the information only once on disk in .dwo
204 files instead of multiple times in object files and executables.
205 In addition the debug information is also compressed.
207 Requires recent gcc (4.7+) and recent gdb/binutils.
208 Any tool that packages or reads debug information would need
209 to know about the .dwo files and include them.
210 Incompatible with older versions of ccache.
212 config DEBUG_INFO_DWARF4
213 bool "Generate dwarf4 debuginfo"
214 depends on DEBUG_INFO
215 depends on $(cc-option,-gdwarf-4)
217 Generate dwarf4 debug info. This requires recent versions
218 of gcc and gdb. It makes the debug information larger.
219 But it significantly improves the success of resolving
220 variables in gdb on optimized code.
222 config DEBUG_INFO_BTF
223 bool "Generate BTF typeinfo"
224 depends on DEBUG_INFO
226 Generate deduplicated BTF type information from DWARF debug info.
227 Turning this on expects presence of pahole tool, which will convert
228 DWARF type info into equivalent deduplicated BTF type info.
231 bool "Provide GDB scripts for kernel debugging"
232 depends on DEBUG_INFO
234 This creates the required links to GDB helper scripts in the
235 build directory. If you load vmlinux into gdb, the helper
236 scripts will be automatically imported by gdb as well, and
237 additional functions are available to analyze a Linux kernel
238 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst
241 config ENABLE_MUST_CHECK
242 bool "Enable __must_check logic"
245 Enable the __must_check logic in the kernel build. Disable this to
246 suppress the "warning: ignoring return value of 'foo', declared with
247 attribute warn_unused_result" messages.
250 int "Warn for stack frames larger than (needs gcc 4.4)"
252 default 2048 if GCC_PLUGIN_LATENT_ENTROPY
253 default 1280 if (!64BIT && PARISC)
254 default 1024 if (!64BIT && !PARISC)
255 default 2048 if 64BIT
257 Tell gcc to warn at build time for stack frames larger than this.
258 Setting this too low will cause a lot of warnings.
259 Setting it to 0 disables the warning.
262 config STRIP_ASM_SYMS
263 bool "Strip assembler-generated symbols during link"
266 Strip internal assembler-generated symbols during a link (symbols
267 that look like '.Lxxx') so they don't pollute the output of
268 get_wchan() and suchlike.
271 bool "Generate readable assembler code"
272 depends on DEBUG_KERNEL
274 Disable some compiler optimizations that tend to generate human unreadable
275 assembler output. This may make the kernel slightly slower, but it helps
276 to keep kernel developers who have to stare a lot at assembler listings
279 config UNUSED_SYMBOLS
280 bool "Enable unused/obsolete exported symbols"
283 Unused but exported symbols make the kernel needlessly bigger. For
284 that reason most of these unused exports will soon be removed. This
285 option is provided temporarily to provide a transition period in case
286 some external kernel module needs one of these symbols anyway. If you
287 encounter such a case in your module, consider if you are actually
288 using the right API. (rationale: since nobody in the kernel is using
289 this in a module, there is a pretty good chance it's actually the
290 wrong interface to use). If you really need the symbol, please send a
291 mail to the linux kernel mailing list mentioning the symbol and why
292 you really need it, and what the merge plan to the mainline kernel for
296 bool "Debug Filesystem"
298 debugfs is a virtual file system that kernel developers use to put
299 debugging files into. Enable this option to be able to read and
300 write to these files.
302 For detailed documentation on the debugfs API, see
303 Documentation/filesystems/.
308 bool "Run 'make headers_check' when building vmlinux"
311 This option will extract the user-visible kernel headers whenever
312 building the kernel, and will run basic sanity checks on them to
313 ensure that exported files do not attempt to include files which
314 were not exported, etc.
316 If you're making modifications to header files which are
317 relevant for userspace, say 'Y', and check the headers
318 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
319 your build tree), to make sure they're suitable.
321 config OPTIMIZE_INLINING
322 bool "Allow compiler to uninline functions marked 'inline'"
324 This option determines if the kernel forces gcc to inline the functions
325 developers have marked 'inline'. Doing so takes away freedom from gcc to
326 do what it thinks is best, which is desirable for the gcc 3.x series of
327 compilers. The gcc 4.x series have a rewritten inlining algorithm and
328 enabling this option will generate a smaller kernel there. Hopefully
329 this algorithm is so good that allowing gcc 4.x and above to make the
330 decision will become the default in the future. Until then this option
331 is there to test gcc for this.
335 config DEBUG_SECTION_MISMATCH
336 bool "Enable full Section mismatch analysis"
338 The section mismatch analysis checks if there are illegal
339 references from one section to another section.
340 During linktime or runtime, some sections are dropped;
341 any use of code/data previously in these sections would
342 most likely result in an oops.
343 In the code, functions and variables are annotated with
344 __init,, etc. (see the full list in include/linux/init.h),
345 which results in the code/data being placed in specific sections.
346 The section mismatch analysis is always performed after a full
347 kernel build, and enabling this option causes the following
348 additional steps to occur:
349 - Add the option -fno-inline-functions-called-once to gcc commands.
350 When inlining a function annotated with __init in a non-init
351 function, we would lose the section information and thus
352 the analysis would not catch the illegal reference.
353 This option tells gcc to inline less (but it does result in
355 - Run the section mismatch analysis for each module/built-in.a file.
356 When we run the section mismatch analysis on vmlinux.o, we
357 lose valuable information about where the mismatch was
359 Running the analysis for each module/built-in.a file
360 tells where the mismatch happens much closer to the
361 source. The drawback is that the same mismatch is
362 reported at least twice.
363 - Enable verbose reporting from modpost in order to help resolve
364 the section mismatches that are reported.
366 config SECTION_MISMATCH_WARN_ONLY
367 bool "Make section mismatch errors non-fatal"
370 If you say N here, the build process will fail if there are any
371 section mismatch, instead of just throwing warnings.
376 # Select this config option from the architecture Kconfig, if it
377 # is preferred to always offer frame pointers as a config
378 # option on the architecture (regardless of KERNEL_DEBUG):
380 config ARCH_WANT_FRAME_POINTERS
384 bool "Compile the kernel with frame pointers"
385 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS
386 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
388 If you say Y here the resulting kernel image will be slightly
389 larger and slower, but it gives very useful debugging information
390 in case of kernel bugs. (precise oopses/stacktraces/warnings)
392 config STACK_VALIDATION
393 bool "Compile-time stack metadata validation"
394 depends on HAVE_STACK_VALIDATION
397 Add compile-time checks to validate stack metadata, including frame
398 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure
399 that runtime stack traces are more reliable.
401 This is also a prerequisite for generation of ORC unwind data, which
402 is needed for CONFIG_UNWINDER_ORC.
404 For more information, see
405 tools/objtool/Documentation/stack-validation.txt.
407 config DEBUG_FORCE_WEAK_PER_CPU
408 bool "Force weak per-cpu definitions"
409 depends on DEBUG_KERNEL
411 s390 and alpha require percpu variables in modules to be
412 defined weak to work around addressing range issue which
413 puts the following two restrictions on percpu variable
416 1. percpu symbols must be unique whether static or not
417 2. percpu variables can't be defined inside a function
419 To ensure that generic code follows the above rules, this
420 option forces all percpu variables to be defined as weak.
422 endmenu # "Compiler options"
425 bool "Magic SysRq key"
428 If you say Y here, you will have some control over the system even
429 if the system crashes for example during kernel debugging (e.g., you
430 will be able to flush the buffer cache to disk, reboot the system
431 immediately or dump some status information). This is accomplished
432 by pressing various keys while holding SysRq (Alt+PrintScreen). It
433 also works on a serial console (on PC hardware at least), if you
434 send a BREAK and then within 5 seconds a command keypress. The
435 keys are documented in <file:Documentation/admin-guide/sysrq.rst>.
436 Don't say Y unless you really know what this hack does.
438 config MAGIC_SYSRQ_DEFAULT_ENABLE
439 hex "Enable magic SysRq key functions by default"
440 depends on MAGIC_SYSRQ
443 Specifies which SysRq key functions are enabled by default.
444 This may be set to 1 or 0 to enable or disable them all, or
445 to a bitmask as described in Documentation/admin-guide/sysrq.rst.
447 config MAGIC_SYSRQ_SERIAL
448 bool "Enable magic SysRq key over serial"
449 depends on MAGIC_SYSRQ
452 Many embedded boards have a disconnected TTL level serial which can
453 generate some garbage that can lead to spurious false sysrq detects.
454 This option allows you to decide whether you want to enable the
458 bool "Kernel debugging"
460 Say Y here if you are developing drivers or trying to debug and
461 identify kernel problems.
464 bool "Miscellaneous debug code"
466 depends on DEBUG_KERNEL
468 Say Y here if you need to enable miscellaneous debug code that should
469 be under a more specific debug option but isn't.
472 menu "Memory Debugging"
474 source "mm/Kconfig.debug"
477 bool "Debug object operations"
478 depends on DEBUG_KERNEL
480 If you say Y here, additional code will be inserted into the
481 kernel to track the life time of various objects and validate
482 the operations on those objects.
484 config DEBUG_OBJECTS_SELFTEST
485 bool "Debug objects selftest"
486 depends on DEBUG_OBJECTS
488 This enables the selftest of the object debug code.
490 config DEBUG_OBJECTS_FREE
491 bool "Debug objects in freed memory"
492 depends on DEBUG_OBJECTS
494 This enables checks whether a k/v free operation frees an area
495 which contains an object which has not been deactivated
496 properly. This can make kmalloc/kfree-intensive workloads
499 config DEBUG_OBJECTS_TIMERS
500 bool "Debug timer objects"
501 depends on DEBUG_OBJECTS
503 If you say Y here, additional code will be inserted into the
504 timer routines to track the life time of timer objects and
505 validate the timer operations.
507 config DEBUG_OBJECTS_WORK
508 bool "Debug work objects"
509 depends on DEBUG_OBJECTS
511 If you say Y here, additional code will be inserted into the
512 work queue routines to track the life time of work objects and
513 validate the work operations.
515 config DEBUG_OBJECTS_RCU_HEAD
516 bool "Debug RCU callbacks objects"
517 depends on DEBUG_OBJECTS
519 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
521 config DEBUG_OBJECTS_PERCPU_COUNTER
522 bool "Debug percpu counter objects"
523 depends on DEBUG_OBJECTS
525 If you say Y here, additional code will be inserted into the
526 percpu counter routines to track the life time of percpu counter
527 objects and validate the percpu counter operations.
529 config DEBUG_OBJECTS_ENABLE_DEFAULT
530 int "debug_objects bootup default value (0-1)"
533 depends on DEBUG_OBJECTS
535 Debug objects boot parameter default value
538 bool "Debug slab memory allocations"
539 depends on DEBUG_KERNEL && SLAB
541 Say Y here to have the kernel do limited verification on memory
542 allocation as well as poisoning memory on free to catch use of freed
543 memory. This can make kmalloc/kfree-intensive workloads much slower.
546 bool "SLUB debugging on by default"
547 depends on SLUB && SLUB_DEBUG
550 Boot with debugging on by default. SLUB boots by default with
551 the runtime debug capabilities switched off. Enabling this is
552 equivalent to specifying the "slub_debug" parameter on boot.
553 There is no support for more fine grained debug control like
554 possible with slub_debug=xxx. SLUB debugging may be switched
555 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
560 bool "Enable SLUB performance statistics"
561 depends on SLUB && SYSFS
563 SLUB statistics are useful to debug SLUBs allocation behavior in
564 order find ways to optimize the allocator. This should never be
565 enabled for production use since keeping statistics slows down
566 the allocator by a few percentage points. The slabinfo command
567 supports the determination of the most active slabs to figure
568 out which slabs are relevant to a particular load.
569 Try running: slabinfo -DA
571 config HAVE_DEBUG_KMEMLEAK
574 config DEBUG_KMEMLEAK
575 bool "Kernel memory leak detector"
576 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK
578 select STACKTRACE if STACKTRACE_SUPPORT
582 Say Y here if you want to enable the memory leak
583 detector. The memory allocation/freeing is traced in a way
584 similar to the Boehm's conservative garbage collector, the
585 difference being that the orphan objects are not freed but
586 only shown in /sys/kernel/debug/kmemleak. Enabling this
587 feature will introduce an overhead to memory
588 allocations. See Documentation/dev-tools/kmemleak.rst for more
591 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
592 of finding leaks due to the slab objects poisoning.
594 In order to access the kmemleak file, debugfs needs to be
595 mounted (usually at /sys/kernel/debug).
597 config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
598 int "Maximum kmemleak early log entries"
599 depends on DEBUG_KMEMLEAK
603 Kmemleak must track all the memory allocations to avoid
604 reporting false positives. Since memory may be allocated or
605 freed before kmemleak is initialised, an early log buffer is
606 used to store these actions. If kmemleak reports "early log
607 buffer exceeded", please increase this value.
609 config DEBUG_KMEMLEAK_TEST
610 tristate "Simple test for the kernel memory leak detector"
611 depends on DEBUG_KMEMLEAK && m
613 This option enables a module that explicitly leaks memory.
617 config DEBUG_KMEMLEAK_DEFAULT_OFF
618 bool "Default kmemleak to off"
619 depends on DEBUG_KMEMLEAK
621 Say Y here to disable kmemleak by default. It can then be enabled
622 on the command line via kmemleak=on.
624 config DEBUG_KMEMLEAK_AUTO_SCAN
625 bool "Enable kmemleak auto scan thread on boot up"
627 depends on DEBUG_KMEMLEAK
629 Depending on the cpu, kmemleak scan may be cpu intensive and can
630 stall user tasks at times. This option enables/disables automatic
631 kmemleak scan at boot up.
633 Say N here to disable kmemleak auto scan thread to stop automatic
634 scanning. Disabling this option disables automatic reporting of
639 config DEBUG_STACK_USAGE
640 bool "Stack utilization instrumentation"
641 depends on DEBUG_KERNEL && !IA64
643 Enables the display of the minimum amount of free stack which each
644 task has ever had available in the sysrq-T and sysrq-P debug output.
646 This option will slow down process creation somewhat.
650 depends on DEBUG_KERNEL
652 Enable this to turn on extended checks in the virtual-memory system
653 that may impact performance.
657 config DEBUG_VM_VMACACHE
658 bool "Debug VMA caching"
661 Enable this to turn on VMA caching debug information. Doing so
662 can cause significant overhead, so only enable it in non-production
668 bool "Debug VM red-black trees"
671 Enable VM red-black tree debugging information and extra validations.
675 config DEBUG_VM_PGFLAGS
676 bool "Debug page-flags operations"
679 Enables extra validation on page flags operations.
683 config ARCH_HAS_DEBUG_VIRTUAL
687 bool "Debug VM translations"
688 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL
690 Enable some costly sanity checks in virtual to page code. This can
691 catch mistakes with virt_to_page() and friends.
695 config DEBUG_NOMMU_REGIONS
696 bool "Debug the global anon/private NOMMU mapping region tree"
697 depends on DEBUG_KERNEL && !MMU
699 This option causes the global tree of anonymous and private mapping
700 regions to be regularly checked for invalid topology.
702 config DEBUG_MEMORY_INIT
703 bool "Debug memory initialisation" if EXPERT
706 Enable this for additional checks during memory initialisation.
707 The sanity checks verify aspects of the VM such as the memory model
708 and other information provided by the architecture. Verbose
709 information will be printed at KERN_DEBUG loglevel depending
710 on the mminit_loglevel= command-line option.
714 config MEMORY_NOTIFIER_ERROR_INJECT
715 tristate "Memory hotplug notifier error injection module"
716 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION
718 This option provides the ability to inject artificial errors to
719 memory hotplug notifier chain callbacks. It is controlled through
720 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
722 If the notifier call chain should be failed with some events
723 notified, write the error code to "actions/<notifier event>/error".
725 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
727 # cd /sys/kernel/debug/notifier-error-inject/memory
728 # echo -12 > actions/MEM_GOING_OFFLINE/error
729 # echo offline > /sys/devices/system/memory/memoryXXX/state
730 bash: echo: write error: Cannot allocate memory
732 To compile this code as a module, choose M here: the module will
733 be called memory-notifier-error-inject.
737 config DEBUG_PER_CPU_MAPS
738 bool "Debug access to per_cpu maps"
739 depends on DEBUG_KERNEL
742 Say Y to verify that the per_cpu map being accessed has
743 been set up. This adds a fair amount of code to kernel memory
744 and decreases performance.
749 bool "Highmem debugging"
750 depends on DEBUG_KERNEL && HIGHMEM
752 This option enables additional error checking for high memory
753 systems. Disable for production systems.
755 config HAVE_DEBUG_STACKOVERFLOW
758 config DEBUG_STACKOVERFLOW
759 bool "Check for stack overflows"
760 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW
762 Say Y here if you want to check for overflows of kernel, IRQ
763 and exception stacks (if your architecture uses them). This
764 option will show detailed messages if free stack space drops
765 below a certain limit.
767 These kinds of bugs usually occur when call-chains in the
768 kernel get too deep, especially when interrupts are
771 Use this in cases where you see apparently random memory
772 corruption, especially if it appears in 'struct thread_info'
774 If in doubt, say "N".
776 source "lib/Kconfig.kasan"
778 endmenu # "Memory Debugging"
783 An architecture should select this when it can successfully
784 build and run with CONFIG_KCOV. This typically requires
785 disabling instrumentation for some early boot code.
787 config CC_HAS_SANCOV_TRACE_PC
788 def_bool $(cc-option,-fsanitize-coverage=trace-pc)
791 bool "Code coverage for fuzzing"
792 depends on ARCH_HAS_KCOV
793 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS
795 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC
797 KCOV exposes kernel code coverage information in a form suitable
798 for coverage-guided fuzzing (randomized testing).
800 If RANDOMIZE_BASE is enabled, PC values will not be stable across
801 different machines and across reboots. If you need stable PC values,
802 disable RANDOMIZE_BASE.
804 For more details, see Documentation/dev-tools/kcov.rst.
806 config KCOV_ENABLE_COMPARISONS
807 bool "Enable comparison operands collection by KCOV"
809 depends on $(cc-option,-fsanitize-coverage=trace-cmp)
811 KCOV also exposes operands of every comparison in the instrumented
812 code along with operand sizes and PCs of the comparison instructions.
813 These operands can be used by fuzzing engines to improve the quality
816 config KCOV_INSTRUMENT_ALL
817 bool "Instrument all code by default"
821 If you are doing generic system call fuzzing (like e.g. syzkaller),
822 then you will want to instrument the whole kernel and you should
823 say y here. If you are doing more targeted fuzzing (like e.g.
824 filesystem fuzzing with AFL) then you will want to enable coverage
825 for more specific subsets of files, and should say n here.
828 bool "Debug shared IRQ handlers"
829 depends on DEBUG_KERNEL
831 Enable this to generate a spurious interrupt as soon as a shared
832 interrupt handler is registered, and just before one is deregistered.
833 Drivers ought to be able to handle interrupts coming in at those
834 points; some don't and need to be caught.
836 menu "Debug Lockups and Hangs"
838 config LOCKUP_DETECTOR
841 config SOFTLOCKUP_DETECTOR
842 bool "Detect Soft Lockups"
843 depends on DEBUG_KERNEL && !S390
844 select LOCKUP_DETECTOR
846 Say Y here to enable the kernel to act as a watchdog to detect
849 Softlockups are bugs that cause the kernel to loop in kernel
850 mode for more than 20 seconds, without giving other tasks a
851 chance to run. The current stack trace is displayed upon
852 detection and the system will stay locked up.
854 config BOOTPARAM_SOFTLOCKUP_PANIC
855 bool "Panic (Reboot) On Soft Lockups"
856 depends on SOFTLOCKUP_DETECTOR
858 Say Y here to enable the kernel to panic on "soft lockups",
859 which are bugs that cause the kernel to loop in kernel
860 mode for more than 20 seconds (configurable using the watchdog_thresh
861 sysctl), without giving other tasks a chance to run.
863 The panic can be used in combination with panic_timeout,
864 to cause the system to reboot automatically after a
865 lockup has been detected. This feature is useful for
866 high-availability systems that have uptime guarantees and
867 where a lockup must be resolved ASAP.
871 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
873 depends on SOFTLOCKUP_DETECTOR
875 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
876 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
878 config HARDLOCKUP_DETECTOR_PERF
880 select SOFTLOCKUP_DETECTOR
883 # Enables a timestamp based low pass filter to compensate for perf based
884 # hard lockup detection which runs too fast due to turbo modes.
886 config HARDLOCKUP_CHECK_TIMESTAMP
890 # arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard
891 # lockup detector rather than the perf based detector.
893 config HARDLOCKUP_DETECTOR
894 bool "Detect Hard Lockups"
895 depends on DEBUG_KERNEL && !S390
896 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_ARCH
897 select LOCKUP_DETECTOR
898 select HARDLOCKUP_DETECTOR_PERF if HAVE_HARDLOCKUP_DETECTOR_PERF
899 select HARDLOCKUP_DETECTOR_ARCH if HAVE_HARDLOCKUP_DETECTOR_ARCH
901 Say Y here to enable the kernel to act as a watchdog to detect
904 Hardlockups are bugs that cause the CPU to loop in kernel mode
905 for more than 10 seconds, without letting other interrupts have a
906 chance to run. The current stack trace is displayed upon detection
907 and the system will stay locked up.
909 config BOOTPARAM_HARDLOCKUP_PANIC
910 bool "Panic (Reboot) On Hard Lockups"
911 depends on HARDLOCKUP_DETECTOR
913 Say Y here to enable the kernel to panic on "hard lockups",
914 which are bugs that cause the kernel to loop in kernel
915 mode with interrupts disabled for more than 10 seconds (configurable
916 using the watchdog_thresh sysctl).
920 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
922 depends on HARDLOCKUP_DETECTOR
924 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
925 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
927 config DETECT_HUNG_TASK
928 bool "Detect Hung Tasks"
929 depends on DEBUG_KERNEL
930 default SOFTLOCKUP_DETECTOR
932 Say Y here to enable the kernel to detect "hung tasks",
933 which are bugs that cause the task to be stuck in
934 uninterruptible "D" state indefinitely.
936 When a hung task is detected, the kernel will print the
937 current stack trace (which you should report), but the
938 task will stay in uninterruptible state. If lockdep is
939 enabled then all held locks will also be reported. This
940 feature has negligible overhead.
942 config DEFAULT_HUNG_TASK_TIMEOUT
943 int "Default timeout for hung task detection (in seconds)"
944 depends on DETECT_HUNG_TASK
947 This option controls the default timeout (in seconds) used
948 to determine when a task has become non-responsive and should
951 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
952 sysctl or by writing a value to
953 /proc/sys/kernel/hung_task_timeout_secs.
955 A timeout of 0 disables the check. The default is two minutes.
956 Keeping the default should be fine in most cases.
958 config BOOTPARAM_HUNG_TASK_PANIC
959 bool "Panic (Reboot) On Hung Tasks"
960 depends on DETECT_HUNG_TASK
962 Say Y here to enable the kernel to panic on "hung tasks",
963 which are bugs that cause the kernel to leave a task stuck
964 in uninterruptible "D" state.
966 The panic can be used in combination with panic_timeout,
967 to cause the system to reboot automatically after a
968 hung task has been detected. This feature is useful for
969 high-availability systems that have uptime guarantees and
970 where a hung tasks must be resolved ASAP.
974 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
976 depends on DETECT_HUNG_TASK
978 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
979 default 1 if BOOTPARAM_HUNG_TASK_PANIC
982 bool "Detect Workqueue Stalls"
983 depends on DEBUG_KERNEL
985 Say Y here to enable stall detection on workqueues. If a
986 worker pool doesn't make forward progress on a pending work
987 item for over a given amount of time, 30s by default, a
988 warning message is printed along with dump of workqueue
989 state. This can be configured through kernel parameter
990 "workqueue.watchdog_thresh" and its sysfs counterpart.
992 endmenu # "Debug lockups and hangs"
997 Say Y here to enable the kernel to panic when it oopses. This
998 has the same effect as setting oops=panic on the kernel command
1001 This feature is useful to ensure that the kernel does not do
1002 anything erroneous after an oops which could result in data
1003 corruption or other issues.
1007 config PANIC_ON_OOPS_VALUE
1010 default 0 if !PANIC_ON_OOPS
1011 default 1 if PANIC_ON_OOPS
1013 config PANIC_TIMEOUT
1017 Set the timeout value (in seconds) until a reboot occurs when the
1018 the kernel panics. If n = 0, then we wait forever. A timeout
1019 value n > 0 will wait n seconds before rebooting, while a timeout
1020 value n < 0 will reboot immediately.
1023 bool "Collect scheduler debugging info"
1024 depends on DEBUG_KERNEL && PROC_FS
1027 If you say Y here, the /proc/sched_debug file will be provided
1028 that can help debug the scheduler. The runtime overhead of this
1036 bool "Collect scheduler statistics"
1037 depends on DEBUG_KERNEL && PROC_FS
1040 If you say Y here, additional code will be inserted into the
1041 scheduler and related routines to collect statistics about
1042 scheduler behavior and provide them in /proc/schedstat. These
1043 stats may be useful for both tuning and debugging the scheduler
1044 If you aren't debugging the scheduler or trying to tune a specific
1045 application, you can say N to avoid the very slight overhead
1048 config SCHED_STACK_END_CHECK
1049 bool "Detect stack corruption on calls to schedule()"
1050 depends on DEBUG_KERNEL
1053 This option checks for a stack overrun on calls to schedule().
1054 If the stack end location is found to be over written always panic as
1055 the content of the corrupted region can no longer be trusted.
1056 This is to ensure no erroneous behaviour occurs which could result in
1057 data corruption or a sporadic crash at a later stage once the region
1058 is examined. The runtime overhead introduced is minimal.
1060 config DEBUG_TIMEKEEPING
1061 bool "Enable extra timekeeping sanity checking"
1063 This option will enable additional timekeeping sanity checks
1064 which may be helpful when diagnosing issues where timekeeping
1065 problems are suspected.
1067 This may include checks in the timekeeping hotpaths, so this
1068 option may have a (very small) performance impact to some
1073 config DEBUG_PREEMPT
1074 bool "Debug preemptible kernel"
1075 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
1078 If you say Y here then the kernel will use a debug variant of the
1079 commonly used smp_processor_id() function and will print warnings
1080 if kernel code uses it in a preemption-unsafe way. Also, the kernel
1081 will detect preemption count underflows.
1083 menu "Lock Debugging (spinlocks, mutexes, etc...)"
1085 config LOCK_DEBUGGING_SUPPORT
1087 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1090 config PROVE_LOCKING
1091 bool "Lock debugging: prove locking correctness"
1092 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1094 select DEBUG_SPINLOCK
1095 select DEBUG_MUTEXES
1096 select DEBUG_RT_MUTEXES if RT_MUTEXES
1097 select DEBUG_RWSEMS if RWSEM_SPIN_ON_OWNER
1098 select DEBUG_WW_MUTEX_SLOWPATH
1099 select DEBUG_LOCK_ALLOC
1100 select TRACE_IRQFLAGS
1103 This feature enables the kernel to prove that all locking
1104 that occurs in the kernel runtime is mathematically
1105 correct: that under no circumstance could an arbitrary (and
1106 not yet triggered) combination of observed locking
1107 sequences (on an arbitrary number of CPUs, running an
1108 arbitrary number of tasks and interrupt contexts) cause a
1111 In short, this feature enables the kernel to report locking
1112 related deadlocks before they actually occur.
1114 The proof does not depend on how hard and complex a
1115 deadlock scenario would be to trigger: how many
1116 participant CPUs, tasks and irq-contexts would be needed
1117 for it to trigger. The proof also does not depend on
1118 timing: if a race and a resulting deadlock is possible
1119 theoretically (no matter how unlikely the race scenario
1120 is), it will be proven so and will immediately be
1121 reported by the kernel (once the event is observed that
1122 makes the deadlock theoretically possible).
1124 If a deadlock is impossible (i.e. the locking rules, as
1125 observed by the kernel, are mathematically correct), the
1126 kernel reports nothing.
1128 NOTE: this feature can also be enabled for rwlocks, mutexes
1129 and rwsems - in which case all dependencies between these
1130 different locking variants are observed and mapped too, and
1131 the proof of observed correctness is also maintained for an
1132 arbitrary combination of these separate locking variants.
1134 For more details, see Documentation/locking/lockdep-design.txt.
1137 bool "Lock usage statistics"
1138 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1140 select DEBUG_SPINLOCK
1141 select DEBUG_MUTEXES
1142 select DEBUG_RT_MUTEXES if RT_MUTEXES
1143 select DEBUG_LOCK_ALLOC
1146 This feature enables tracking lock contention points
1148 For more details, see Documentation/locking/lockstat.txt
1150 This also enables lock events required by "perf lock",
1152 If you want to use "perf lock", you also need to turn on
1153 CONFIG_EVENT_TRACING.
1155 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
1156 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
1158 config DEBUG_RT_MUTEXES
1159 bool "RT Mutex debugging, deadlock detection"
1160 depends on DEBUG_KERNEL && RT_MUTEXES
1162 This allows rt mutex semantics violations and rt mutex related
1163 deadlocks (lockups) to be detected and reported automatically.
1165 config DEBUG_SPINLOCK
1166 bool "Spinlock and rw-lock debugging: basic checks"
1167 depends on DEBUG_KERNEL
1168 select UNINLINE_SPIN_UNLOCK
1170 Say Y here and build SMP to catch missing spinlock initialization
1171 and certain other kinds of spinlock errors commonly made. This is
1172 best used in conjunction with the NMI watchdog so that spinlock
1173 deadlocks are also debuggable.
1175 config DEBUG_MUTEXES
1176 bool "Mutex debugging: basic checks"
1177 depends on DEBUG_KERNEL
1179 This feature allows mutex semantics violations to be detected and
1182 config DEBUG_WW_MUTEX_SLOWPATH
1183 bool "Wait/wound mutex debugging: Slowpath testing"
1184 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1185 select DEBUG_LOCK_ALLOC
1186 select DEBUG_SPINLOCK
1187 select DEBUG_MUTEXES
1189 This feature enables slowpath testing for w/w mutex users by
1190 injecting additional -EDEADLK wound/backoff cases. Together with
1191 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this
1192 will test all possible w/w mutex interface abuse with the
1193 exception of simply not acquiring all the required locks.
1194 Note that this feature can introduce significant overhead, so
1195 it really should not be enabled in a production or distro kernel,
1196 even a debug kernel. If you are a driver writer, enable it. If
1197 you are a distro, do not.
1200 bool "RW Semaphore debugging: basic checks"
1201 depends on DEBUG_KERNEL && RWSEM_SPIN_ON_OWNER
1203 This debugging feature allows mismatched rw semaphore locks and unlocks
1204 to be detected and reported.
1206 config DEBUG_LOCK_ALLOC
1207 bool "Lock debugging: detect incorrect freeing of live locks"
1208 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1209 select DEBUG_SPINLOCK
1210 select DEBUG_MUTEXES
1211 select DEBUG_RT_MUTEXES if RT_MUTEXES
1214 This feature will check whether any held lock (spinlock, rwlock,
1215 mutex or rwsem) is incorrectly freed by the kernel, via any of the
1216 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
1217 vfree(), etc.), whether a live lock is incorrectly reinitialized via
1218 spin_lock_init()/mutex_init()/etc., or whether there is any lock
1219 held during task exit.
1223 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1225 select FRAME_POINTER if !MIPS && !PPC && !ARM && !S390 && !MICROBLAZE && !ARC && !X86
1229 config LOCKDEP_SMALL
1232 config DEBUG_LOCKDEP
1233 bool "Lock dependency engine debugging"
1234 depends on DEBUG_KERNEL && LOCKDEP
1236 If you say Y here, the lock dependency engine will do
1237 additional runtime checks to debug itself, at the price
1238 of more runtime overhead.
1240 config DEBUG_ATOMIC_SLEEP
1241 bool "Sleep inside atomic section checking"
1242 select PREEMPT_COUNT
1243 depends on DEBUG_KERNEL
1244 depends on !ARCH_NO_PREEMPT
1246 If you say Y here, various routines which may sleep will become very
1247 noisy if they are called inside atomic sections: when a spinlock is
1248 held, inside an rcu read side critical section, inside preempt disabled
1249 sections, inside an interrupt, etc...
1251 config DEBUG_LOCKING_API_SELFTESTS
1252 bool "Locking API boot-time self-tests"
1253 depends on DEBUG_KERNEL
1255 Say Y here if you want the kernel to run a short self-test during
1256 bootup. The self-test checks whether common types of locking bugs
1257 are detected by debugging mechanisms or not. (if you disable
1258 lock debugging then those bugs wont be detected of course.)
1259 The following locking APIs are covered: spinlocks, rwlocks,
1262 config LOCK_TORTURE_TEST
1263 tristate "torture tests for locking"
1264 depends on DEBUG_KERNEL
1267 This option provides a kernel module that runs torture tests
1268 on kernel locking primitives. The kernel module may be built
1269 after the fact on the running kernel to be tested, if desired.
1271 Say Y here if you want kernel locking-primitive torture tests
1272 to be built into the kernel.
1273 Say M if you want these torture tests to build as a module.
1274 Say N if you are unsure.
1276 config WW_MUTEX_SELFTEST
1277 tristate "Wait/wound mutex selftests"
1279 This option provides a kernel module that runs tests on the
1280 on the struct ww_mutex locking API.
1282 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction
1283 with this test harness.
1285 Say M if you want these self tests to build as a module.
1286 Say N if you are unsure.
1288 endmenu # lock debugging
1290 config TRACE_IRQFLAGS
1293 Enables hooks to interrupt enabling and disabling for
1294 either tracing or lock debugging.
1297 bool "Stack backtrace support"
1298 depends on STACKTRACE_SUPPORT
1300 This option causes the kernel to create a /proc/pid/stack for
1301 every process, showing its current stack trace.
1302 It is also used by various kernel debugging features that require
1303 stack trace generation.
1305 config WARN_ALL_UNSEEDED_RANDOM
1306 bool "Warn for all uses of unseeded randomness"
1309 Some parts of the kernel contain bugs relating to their use of
1310 cryptographically secure random numbers before it's actually possible
1311 to generate those numbers securely. This setting ensures that these
1312 flaws don't go unnoticed, by enabling a message, should this ever
1313 occur. This will allow people with obscure setups to know when things
1314 are going wrong, so that they might contact developers about fixing
1317 Unfortunately, on some models of some architectures getting
1318 a fully seeded CRNG is extremely difficult, and so this can
1319 result in dmesg getting spammed for a surprisingly long
1320 time. This is really bad from a security perspective, and
1321 so architecture maintainers really need to do what they can
1322 to get the CRNG seeded sooner after the system is booted.
1323 However, since users cannot do anything actionable to
1324 address this, by default the kernel will issue only a single
1325 warning for the first use of unseeded randomness.
1327 Say Y here if you want to receive warnings for all uses of
1328 unseeded randomness. This will be of use primarily for
1329 those developers interested in improving the security of
1330 Linux kernels running on their architecture (or
1333 config DEBUG_KOBJECT
1334 bool "kobject debugging"
1335 depends on DEBUG_KERNEL
1337 If you say Y here, some extra kobject debugging messages will be sent
1340 config DEBUG_KOBJECT_RELEASE
1341 bool "kobject release debugging"
1342 depends on DEBUG_OBJECTS_TIMERS
1344 kobjects are reference counted objects. This means that their
1345 last reference count put is not predictable, and the kobject can
1346 live on past the point at which a driver decides to drop it's
1347 initial reference to the kobject gained on allocation. An
1348 example of this would be a struct device which has just been
1351 However, some buggy drivers assume that after such an operation,
1352 the memory backing the kobject can be immediately freed. This
1353 goes completely against the principles of a refcounted object.
1355 If you say Y here, the kernel will delay the release of kobjects
1356 on the last reference count to improve the visibility of this
1357 kind of kobject release bug.
1359 config HAVE_DEBUG_BUGVERBOSE
1362 config DEBUG_BUGVERBOSE
1363 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
1364 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
1367 Say Y here to make BUG() panics output the file name and line number
1368 of the BUG call as well as the EIP and oops trace. This aids
1369 debugging but costs about 70-100K of memory.
1372 bool "Debug linked list manipulation"
1373 depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION
1375 Enable this to turn on extended checks in the linked-list
1381 bool "Debug priority linked list manipulation"
1382 depends on DEBUG_KERNEL
1384 Enable this to turn on extended checks in the priority-ordered
1385 linked-list (plist) walking routines. This checks the entire
1386 list multiple times during each manipulation.
1391 bool "Debug SG table operations"
1392 depends on DEBUG_KERNEL
1394 Enable this to turn on checks on scatter-gather tables. This can
1395 help find problems with drivers that do not properly initialize
1400 config DEBUG_NOTIFIERS
1401 bool "Debug notifier call chains"
1402 depends on DEBUG_KERNEL
1404 Enable this to turn on sanity checking for notifier call chains.
1405 This is most useful for kernel developers to make sure that
1406 modules properly unregister themselves from notifier chains.
1407 This is a relatively cheap check but if you care about maximum
1410 config DEBUG_CREDENTIALS
1411 bool "Debug credential management"
1412 depends on DEBUG_KERNEL
1414 Enable this to turn on some debug checking for credential
1415 management. The additional code keeps track of the number of
1416 pointers from task_structs to any given cred struct, and checks to
1417 see that this number never exceeds the usage count of the cred
1420 Furthermore, if SELinux is enabled, this also checks that the
1421 security pointer in the cred struct is never seen to be invalid.
1425 source "kernel/rcu/Kconfig.debug"
1427 config DEBUG_WQ_FORCE_RR_CPU
1428 bool "Force round-robin CPU selection for unbound work items"
1429 depends on DEBUG_KERNEL
1432 Workqueue used to implicitly guarantee that work items queued
1433 without explicit CPU specified are put on the local CPU. This
1434 guarantee is no longer true and while local CPU is still
1435 preferred work items may be put on foreign CPUs. Kernel
1436 parameter "workqueue.debug_force_rr_cpu" is added to force
1437 round-robin CPU selection to flush out usages which depend on the
1438 now broken guarantee. This config option enables the debug
1439 feature by default. When enabled, memory and cache locality will
1442 config DEBUG_BLOCK_EXT_DEVT
1443 bool "Force extended block device numbers and spread them"
1444 depends on DEBUG_KERNEL
1448 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1449 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1450 YOU ARE DOING. Distros, please enable this and fix whatever
1453 Conventionally, block device numbers are allocated from
1454 predetermined contiguous area. However, extended block area
1455 may introduce non-contiguous block device numbers. This
1456 option forces most block device numbers to be allocated from
1457 the extended space and spreads them to discover kernel or
1458 userland code paths which assume predetermined contiguous
1459 device number allocation.
1461 Note that turning on this debug option shuffles all the
1462 device numbers for all IDE and SCSI devices including libata
1463 ones, so root partition specified using device number
1464 directly (via rdev or root=MAJ:MIN) won't work anymore.
1465 Textual device names (root=/dev/sdXn) will continue to work.
1467 Say N if you are unsure.
1469 config CPU_HOTPLUG_STATE_CONTROL
1470 bool "Enable CPU hotplug state control"
1471 depends on DEBUG_KERNEL
1472 depends on HOTPLUG_CPU
1475 Allows to write steps between "offline" and "online" to the CPUs
1476 sysfs target file so states can be stepped granular. This is a debug
1477 option for now as the hotplug machinery cannot be stopped and
1478 restarted at arbitrary points yet.
1480 Say N if your are unsure.
1482 config NOTIFIER_ERROR_INJECTION
1483 tristate "Notifier error injection"
1484 depends on DEBUG_KERNEL
1487 This option provides the ability to inject artificial errors to
1488 specified notifier chain callbacks. It is useful to test the error
1489 handling of notifier call chain failures.
1493 config PM_NOTIFIER_ERROR_INJECT
1494 tristate "PM notifier error injection module"
1495 depends on PM && NOTIFIER_ERROR_INJECTION
1496 default m if PM_DEBUG
1498 This option provides the ability to inject artificial errors to
1499 PM notifier chain callbacks. It is controlled through debugfs
1500 interface /sys/kernel/debug/notifier-error-inject/pm
1502 If the notifier call chain should be failed with some events
1503 notified, write the error code to "actions/<notifier event>/error".
1505 Example: Inject PM suspend error (-12 = -ENOMEM)
1507 # cd /sys/kernel/debug/notifier-error-inject/pm/
1508 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1509 # echo mem > /sys/power/state
1510 bash: echo: write error: Cannot allocate memory
1512 To compile this code as a module, choose M here: the module will
1513 be called pm-notifier-error-inject.
1517 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1518 tristate "OF reconfig notifier error injection module"
1519 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1521 This option provides the ability to inject artificial errors to
1522 OF reconfig notifier chain callbacks. It is controlled
1523 through debugfs interface under
1524 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1526 If the notifier call chain should be failed with some events
1527 notified, write the error code to "actions/<notifier event>/error".
1529 To compile this code as a module, choose M here: the module will
1530 be called of-reconfig-notifier-error-inject.
1534 config NETDEV_NOTIFIER_ERROR_INJECT
1535 tristate "Netdev notifier error injection module"
1536 depends on NET && NOTIFIER_ERROR_INJECTION
1538 This option provides the ability to inject artificial errors to
1539 netdevice notifier chain callbacks. It is controlled through debugfs
1540 interface /sys/kernel/debug/notifier-error-inject/netdev
1542 If the notifier call chain should be failed with some events
1543 notified, write the error code to "actions/<notifier event>/error".
1545 Example: Inject netdevice mtu change error (-22 = -EINVAL)
1547 # cd /sys/kernel/debug/notifier-error-inject/netdev
1548 # echo -22 > actions/NETDEV_CHANGEMTU/error
1549 # ip link set eth0 mtu 1024
1550 RTNETLINK answers: Invalid argument
1552 To compile this code as a module, choose M here: the module will
1553 be called netdev-notifier-error-inject.
1557 config FUNCTION_ERROR_INJECTION
1559 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES
1561 config FAULT_INJECTION
1562 bool "Fault-injection framework"
1563 depends on DEBUG_KERNEL
1565 Provide fault-injection framework.
1566 For more details, see Documentation/fault-injection/.
1569 bool "Fault-injection capability for kmalloc"
1570 depends on FAULT_INJECTION
1571 depends on SLAB || SLUB
1573 Provide fault-injection capability for kmalloc.
1575 config FAIL_PAGE_ALLOC
1576 bool "Fault-injection capabilitiy for alloc_pages()"
1577 depends on FAULT_INJECTION
1579 Provide fault-injection capability for alloc_pages().
1581 config FAIL_MAKE_REQUEST
1582 bool "Fault-injection capability for disk IO"
1583 depends on FAULT_INJECTION && BLOCK
1585 Provide fault-injection capability for disk IO.
1587 config FAIL_IO_TIMEOUT
1588 bool "Fault-injection capability for faking disk interrupts"
1589 depends on FAULT_INJECTION && BLOCK
1591 Provide fault-injection capability on end IO handling. This
1592 will make the block layer "forget" an interrupt as configured,
1593 thus exercising the error handling.
1595 Only works with drivers that use the generic timeout handling,
1596 for others it wont do anything.
1599 bool "Fault-injection capability for futexes"
1601 depends on FAULT_INJECTION && FUTEX
1603 Provide fault-injection capability for futexes.
1605 config FAULT_INJECTION_DEBUG_FS
1606 bool "Debugfs entries for fault-injection capabilities"
1607 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1609 Enable configuration of fault-injection capabilities via debugfs.
1611 config FAIL_FUNCTION
1612 bool "Fault-injection capability for functions"
1613 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION
1615 Provide function-based fault-injection capability.
1616 This will allow you to override a specific function with a return
1617 with given return value. As a result, function caller will see
1618 an error value and have to handle it. This is useful to test the
1619 error handling in various subsystems.
1621 config FAIL_MMC_REQUEST
1622 bool "Fault-injection capability for MMC IO"
1623 depends on FAULT_INJECTION_DEBUG_FS && MMC
1625 Provide fault-injection capability for MMC IO.
1626 This will make the mmc core return data errors. This is
1627 useful to test the error handling in the mmc block device
1628 and to test how the mmc host driver handles retries from
1631 config FAULT_INJECTION_STACKTRACE_FILTER
1632 bool "stacktrace filter for fault-injection capabilities"
1633 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1636 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1638 Provide stacktrace filter for fault-injection capabilities
1641 bool "Latency measuring infrastructure"
1642 depends on DEBUG_KERNEL
1643 depends on STACKTRACE_SUPPORT
1645 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1652 Enable this option if you want to use the LatencyTOP tool
1653 to find out which userspace is blocking on what kernel operations.
1655 source "kernel/trace/Kconfig"
1657 config PROVIDE_OHCI1394_DMA_INIT
1658 bool "Remote debugging over FireWire early on boot"
1659 depends on PCI && X86
1661 If you want to debug problems which hang or crash the kernel early
1662 on boot and the crashing machine has a FireWire port, you can use
1663 this feature to remotely access the memory of the crashed machine
1664 over FireWire. This employs remote DMA as part of the OHCI1394
1665 specification which is now the standard for FireWire controllers.
1667 With remote DMA, you can monitor the printk buffer remotely using
1668 firescope and access all memory below 4GB using fireproxy from gdb.
1669 Even controlling a kernel debugger is possible using remote DMA.
1673 If ohci1394_dma=early is used as boot parameter, it will initialize
1674 all OHCI1394 controllers which are found in the PCI config space.
1676 As all changes to the FireWire bus such as enabling and disabling
1677 devices cause a bus reset and thereby disable remote DMA for all
1678 devices, be sure to have the cable plugged and FireWire enabled on
1679 the debugging host before booting the debug target for debugging.
1681 This code (~1k) is freed after boot. By then, the firewire stack
1682 in charge of the OHCI-1394 controllers should be used instead.
1684 See Documentation/debugging-via-ohci1394.txt for more information.
1686 menuconfig RUNTIME_TESTING_MENU
1687 bool "Runtime Testing"
1690 if RUNTIME_TESTING_MENU
1693 tristate "Linux Kernel Dump Test Tool Module"
1696 This module enables testing of the different dumping mechanisms by
1697 inducing system failures at predefined crash points.
1698 If you don't need it: say N
1699 Choose M here to compile this code as a module. The module will be
1702 Documentation on how to use the module can be found in
1703 Documentation/fault-injection/provoke-crashes.txt
1705 config TEST_LIST_SORT
1706 tristate "Linked list sorting test"
1707 depends on DEBUG_KERNEL || m
1709 Enable this to turn on 'list_sort()' function test. This test is
1710 executed only once during system boot (so affects only boot time),
1711 or at module load time.
1716 tristate "Array-based sort test"
1717 depends on DEBUG_KERNEL || m
1719 This option enables the self-test function of 'sort()' at boot,
1720 or at module load time.
1724 config KPROBES_SANITY_TEST
1725 bool "Kprobes sanity tests"
1726 depends on DEBUG_KERNEL
1729 This option provides for testing basic kprobes functionality on
1730 boot. Samples of kprobe and kretprobe are inserted and
1731 verified for functionality.
1733 Say N if you are unsure.
1735 config BACKTRACE_SELF_TEST
1736 tristate "Self test for the backtrace code"
1737 depends on DEBUG_KERNEL
1739 This option provides a kernel module that can be used to test
1740 the kernel stack backtrace code. This option is not useful
1741 for distributions or general kernels, but only for kernel
1742 developers working on architecture code.
1744 Note that if you want to also test saved backtraces, you will
1745 have to enable STACKTRACE as well.
1747 Say N if you are unsure.
1750 tristate "Red-Black tree test"
1751 depends on DEBUG_KERNEL
1753 A benchmark measuring the performance of the rbtree library.
1754 Also includes rbtree invariant checks.
1756 config INTERVAL_TREE_TEST
1757 tristate "Interval tree test"
1758 depends on DEBUG_KERNEL
1759 select INTERVAL_TREE
1761 A benchmark measuring the performance of the interval tree library
1764 tristate "Per cpu operations test"
1765 depends on m && DEBUG_KERNEL
1767 Enable this option to build test module which validates per-cpu
1772 config ATOMIC64_SELFTEST
1773 tristate "Perform an atomic64_t self-test"
1775 Enable this option to test the atomic64_t functions at boot or
1776 at module load time.
1780 config ASYNC_RAID6_TEST
1781 tristate "Self test for hardware accelerated raid6 recovery"
1782 depends on ASYNC_RAID6_RECOV
1785 This is a one-shot self test that permutes through the
1786 recovery of all the possible two disk failure scenarios for a
1787 N-disk array. Recovery is performed with the asynchronous
1788 raid6 recovery routines, and will optionally use an offload
1789 engine if one is available.
1794 tristate "Test functions located in the hexdump module at runtime"
1796 config TEST_STRING_HELPERS
1797 tristate "Test functions located in the string_helpers module at runtime"
1800 tristate "Test strscpy*() family of functions at runtime"
1803 tristate "Test kstrto*() family of functions at runtime"
1806 tristate "Test printf() family of functions at runtime"
1809 tristate "Test bitmap_*() family of functions at runtime"
1811 Enable this option to test the bitmap functions at boot.
1815 config TEST_BITFIELD
1816 tristate "Test bitfield functions at runtime"
1818 Enable this option to test the bitfield functions at boot.
1823 tristate "Test functions located in the uuid module at runtime"
1826 tristate "Test the XArray code at runtime"
1828 config TEST_OVERFLOW
1829 tristate "Test check_*_overflow() functions at runtime"
1831 config TEST_RHASHTABLE
1832 tristate "Perform selftest on resizable hash table"
1834 Enable this option to test the rhashtable functions at boot.
1839 tristate "Perform selftest on hash functions"
1841 Enable this option to test the kernel's integer (<linux/hash.h>),
1842 string (<linux/stringhash.h>), and siphash (<linux/siphash.h>)
1843 hash functions on boot (or module load).
1845 This is intended to help people writing architecture-specific
1846 optimized versions. If unsure, say N.
1849 tristate "Perform selftest on IDA functions"
1852 tristate "Perform selftest on priority array manager"
1855 Enable this option to test priority array manager on boot
1861 tristate "Test module loading with 'hello world' module"
1864 This builds the "test_module" module that emits "Hello, world"
1865 on printk when loaded. It is designed to be used for basic
1866 evaluation of the module loading subsystem (for example when
1867 validating module verification). It lacks any extra dependencies,
1868 and will not normally be loaded by the system unless explicitly
1874 tristate "Test module for stress/performance analysis of vmalloc allocator"
1879 This builds the "test_vmalloc" module that should be used for
1880 stress and performance analysis. So, any new change for vmalloc
1881 subsystem can be evaluated from performance and stability point
1886 config TEST_USER_COPY
1887 tristate "Test user/kernel boundary protections"
1890 This builds the "test_user_copy" module that runs sanity checks
1891 on the copy_to/from_user infrastructure, making sure basic
1892 user/kernel boundary testing is working. If it fails to load,
1893 a regression has been detected in the user/kernel memory boundary
1899 tristate "Test BPF filter functionality"
1902 This builds the "test_bpf" module that runs various test vectors
1903 against the BPF interpreter or BPF JIT compiler depending on the
1904 current setting. This is in particular useful for BPF JIT compiler
1905 development, but also to run regression tests against changes in
1906 the interpreter code. It also enables test stubs for eBPF maps and
1907 verifier used by user space verifier testsuite.
1911 config FIND_BIT_BENCHMARK
1912 tristate "Test find_bit functions"
1914 This builds the "test_find_bit" module that measure find_*_bit()
1915 functions performance.
1919 config TEST_FIRMWARE
1920 tristate "Test firmware loading via userspace interface"
1921 depends on FW_LOADER
1923 This builds the "test_firmware" module that creates a userspace
1924 interface for testing firmware loading. This can be used to
1925 control the triggering of firmware loading without needing an
1926 actual firmware-using device. The contents can be rechecked by
1932 tristate "sysctl test driver"
1933 depends on PROC_SYSCTL
1935 This builds the "test_sysctl" module. This driver enables to test the
1936 proc sysctl interfaces available to drivers safely without affecting
1937 production knobs which might alter system functionality.
1942 tristate "udelay test driver"
1944 This builds the "udelay_test" module that helps to make sure
1945 that udelay() is working properly.
1949 config TEST_STATIC_KEYS
1950 tristate "Test static keys"
1953 Test the static key interfaces.
1958 tristate "kmod stress tester"
1960 depends on NETDEVICES && NET_CORE && INET # for TUN
1967 Test the kernel's module loading mechanism: kmod. kmod implements
1968 support to load modules using the Linux kernel's usermode helper.
1969 This test provides a series of tests against kmod.
1971 Although technically you can either build test_kmod as a module or
1972 into the kernel we disallow building it into the kernel since
1973 it stress tests request_module() and this will very likely cause
1974 some issues by taking over precious threads available from other
1975 module load requests, ultimately this could be fatal.
1979 tools/testing/selftests/kmod/kmod.sh --help
1983 config TEST_DEBUG_VIRTUAL
1984 tristate "Test CONFIG_DEBUG_VIRTUAL feature"
1985 depends on DEBUG_VIRTUAL
1987 Test the kernel's ability to detect incorrect calls to
1988 virt_to_phys() done against the non-linear part of the
1989 kernel's virtual address map.
1993 config TEST_MEMCAT_P
1994 tristate "Test memcat_p() helper function"
1996 Test the memcat_p() helper for correctly merging two
1997 pointer arrays together.
2001 config TEST_LIVEPATCH
2002 tristate "Test livepatching"
2004 depends on DYNAMIC_DEBUG
2005 depends on LIVEPATCH
2008 Test kernel livepatching features for correctness. The tests will
2009 load test modules that will be livepatched in various scenarios.
2011 To run all the livepatching tests:
2013 make -C tools/testing/selftests TARGETS=livepatch run_tests
2015 Alternatively, individual tests may be invoked:
2017 tools/testing/selftests/livepatch/test-callbacks.sh
2018 tools/testing/selftests/livepatch/test-livepatch.sh
2019 tools/testing/selftests/livepatch/test-shadow-vars.sh
2024 tristate "Perform selftest on object aggreration manager"
2028 Enable this option to test object aggregation manager on boot
2032 config TEST_STACKINIT
2033 tristate "Test level of stack variable initialization"
2035 Test if the kernel is zero-initializing stack variables and
2036 padding. Coverage is controlled by compiler flags,
2037 CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF,
2038 or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL.
2042 endif # RUNTIME_TESTING_MENU
2047 This option adds a kernel parameter 'memtest', which allows memtest
2049 memtest=0, mean disabled; -- default
2050 memtest=1, mean do 1 test pattern;
2052 memtest=17, mean do 17 test patterns.
2053 If you are unsure how to answer this question, answer N.
2055 config BUG_ON_DATA_CORRUPTION
2056 bool "Trigger a BUG when data corruption is detected"
2059 Select this option if the kernel should BUG when it encounters
2060 data corruption in kernel memory structures when they get checked
2065 source "samples/Kconfig"
2067 source "lib/Kconfig.kgdb"
2069 source "lib/Kconfig.ubsan"
2071 config ARCH_HAS_DEVMEM_IS_ALLOWED
2074 config STRICT_DEVMEM
2075 bool "Filter access to /dev/mem"
2076 depends on MMU && DEVMEM
2077 depends on ARCH_HAS_DEVMEM_IS_ALLOWED
2078 default y if PPC || X86 || ARM64
2080 If this option is disabled, you allow userspace (root) access to all
2081 of memory, including kernel and userspace memory. Accidental
2082 access to this is obviously disastrous, but specific access can
2083 be used by people debugging the kernel. Note that with PAT support
2084 enabled, even in this case there are restrictions on /dev/mem
2085 use due to the cache aliasing requirements.
2087 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem
2088 file only allows userspace access to PCI space and the BIOS code and
2089 data regions. This is sufficient for dosemu and X and all common
2094 config IO_STRICT_DEVMEM
2095 bool "Filter I/O access to /dev/mem"
2096 depends on STRICT_DEVMEM
2098 If this option is disabled, you allow userspace (root) access to all
2099 io-memory regardless of whether a driver is actively using that
2100 range. Accidental access to this is obviously disastrous, but
2101 specific access can be used by people debugging kernel drivers.
2103 If this option is switched on, the /dev/mem file only allows
2104 userspace access to *idle* io-memory ranges (see /proc/iomem) This
2105 may break traditional users of /dev/mem (dosemu, legacy X, etc...)
2106 if the driver using a given range cannot be disabled.
2110 source "arch/$(SRCARCH)/Kconfig.debug"
2112 endmenu # Kernel hacking