1 # SPDX-License-Identifier: GPL-2.0-only
4 menu "printk and dmesg options"
7 bool "Show timing information on printks"
10 Selecting this option causes time stamps of the printk()
11 messages to be added to the output of the syslog() system
12 call and at the console.
14 The timestamp is always recorded internally, and exported
15 to /dev/kmsg. This flag just specifies if the timestamp should
16 be included, not that the timestamp is recorded.
18 The behavior is also controlled by the kernel command line
19 parameter printk.time=1. See Documentation/admin-guide/kernel-parameters.rst
22 bool "Show caller information on printks"
25 Selecting this option causes printk() to add a caller "thread id" (if
26 in task context) or a caller "processor id" (if not in task context)
29 This option is intended for environments where multiple threads
30 concurrently call printk() for many times, for it is difficult to
31 interpret without knowing where these lines (or sometimes individual
32 line which was divided into multiple lines due to race) came from.
34 Since toggling after boot makes the code racy, currently there is
35 no option to enable/disable at the kernel command line parameter or
38 config STACKTRACE_BUILD_ID
39 bool "Show build ID information in stacktraces"
42 Selecting this option adds build ID information for symbols in
43 stacktraces printed with the printk format '%p[SR]b'.
45 This option is intended for distros where debuginfo is not easily
46 accessible but can be downloaded given the build ID of the vmlinux or
47 kernel module where the function is located.
49 config CONSOLE_LOGLEVEL_DEFAULT
50 int "Default console loglevel (1-15)"
54 Default loglevel to determine what will be printed on the console.
56 Setting a default here is equivalent to passing in loglevel=<x> in
57 the kernel bootargs. loglevel=<x> continues to override whatever
58 value is specified here as well.
60 Note: This does not affect the log level of un-prefixed printk()
61 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT
64 config CONSOLE_LOGLEVEL_QUIET
65 int "quiet console loglevel (1-15)"
69 loglevel to use when "quiet" is passed on the kernel commandline.
71 When "quiet" is passed on the kernel commandline this loglevel
72 will be used as the loglevel. IOW passing "quiet" will be the
73 equivalent of passing "loglevel=<CONSOLE_LOGLEVEL_QUIET>"
75 config MESSAGE_LOGLEVEL_DEFAULT
76 int "Default message log level (1-7)"
80 Default log level for printk statements with no specified priority.
82 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
83 that are auditing their logs closely may want to set it to a lower
86 Note: This does not affect what message level gets printed on the console
87 by default. To change that, use loglevel=<x> in the kernel bootargs,
88 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value.
90 config BOOT_PRINTK_DELAY
91 bool "Delay each boot printk message by N milliseconds"
92 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
94 This build option allows you to read kernel boot messages
95 by inserting a short delay after each one. The delay is
96 specified in milliseconds on the kernel command line,
99 It is likely that you would also need to use "lpj=M" to preset
100 the "loops per jiffie" value.
101 See a previous boot log for the "lpj" value to use for your
102 system, and then set "lpj=M" before setting "boot_delay=N".
103 NOTE: Using this option may adversely affect SMP systems.
104 I.e., processors other than the first one may not boot up.
105 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
106 what it believes to be lockup conditions.
109 bool "Enable dynamic printk() support"
112 depends on (DEBUG_FS || PROC_FS)
113 select DYNAMIC_DEBUG_CORE
116 Compiles debug level messages into the kernel, which would not
117 otherwise be available at runtime. These messages can then be
118 enabled/disabled based on various levels of scope - per source file,
119 function, module, format string, and line number. This mechanism
120 implicitly compiles in all pr_debug() and dev_dbg() calls, which
121 enlarges the kernel text size by about 2%.
123 If a source file is compiled with DEBUG flag set, any
124 pr_debug() calls in it are enabled by default, but can be
125 disabled at runtime as below. Note that DEBUG flag is
126 turned on by many CONFIG_*DEBUG* options.
130 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
131 which is contained in the 'debugfs' filesystem or procfs.
132 Thus, the debugfs or procfs filesystem must first be mounted before
133 making use of this feature.
134 We refer the control file as: <debugfs>/dynamic_debug/control. This
135 file contains a list of the debug statements that can be enabled. The
136 format for each line of the file is:
138 filename:lineno [module]function flags format
140 filename : source file of the debug statement
141 lineno : line number of the debug statement
142 module : module that contains the debug statement
143 function : function that contains the debug statement
144 flags : '=p' means the line is turned 'on' for printing
145 format : the format used for the debug statement
149 nullarbor:~ # cat <debugfs>/dynamic_debug/control
150 # filename:lineno [module]function flags format
151 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012"
152 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012"
153 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012"
157 // enable the message at line 1603 of file svcsock.c
158 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
159 <debugfs>/dynamic_debug/control
161 // enable all the messages in file svcsock.c
162 nullarbor:~ # echo -n 'file svcsock.c +p' >
163 <debugfs>/dynamic_debug/control
165 // enable all the messages in the NFS server module
166 nullarbor:~ # echo -n 'module nfsd +p' >
167 <debugfs>/dynamic_debug/control
169 // enable all 12 messages in the function svc_process()
170 nullarbor:~ # echo -n 'func svc_process +p' >
171 <debugfs>/dynamic_debug/control
173 // disable all 12 messages in the function svc_process()
174 nullarbor:~ # echo -n 'func svc_process -p' >
175 <debugfs>/dynamic_debug/control
177 See Documentation/admin-guide/dynamic-debug-howto.rst for additional
180 config DYNAMIC_DEBUG_CORE
181 bool "Enable core function of dynamic debug support"
183 depends on (DEBUG_FS || PROC_FS)
185 Enable core functional support of dynamic debug. It is useful
186 when you want to tie dynamic debug to your kernel modules with
187 DYNAMIC_DEBUG_MODULE defined for each of them, especially for
188 the case of embedded system where the kernel image size is
189 sensitive for people.
191 config SYMBOLIC_ERRNAME
192 bool "Support symbolic error names in printf"
195 If you say Y here, the kernel's printf implementation will
196 be able to print symbolic error names such as ENOSPC instead
197 of the number 28. It makes the kernel image slightly larger
198 (about 3KB), but can make the kernel logs easier to read.
200 config DEBUG_BUGVERBOSE
201 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
202 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
205 Say Y here to make BUG() panics output the file name and line number
206 of the BUG call as well as the EIP and oops trace. This aids
207 debugging but costs about 70-100K of memory.
209 endmenu # "printk and dmesg options"
212 bool "Kernel debugging"
214 Say Y here if you are developing drivers or trying to debug and
215 identify kernel problems.
218 bool "Miscellaneous debug code"
220 depends on DEBUG_KERNEL
222 Say Y here if you need to enable miscellaneous debug code that should
223 be under a more specific debug option but isn't.
225 menu "Compile-time checks and compiler options"
230 A kernel debug info option other than "None" has been selected
231 in the "Debug information" choice below, indicating that debug
232 information will be generated for build targets.
234 # Clang generates .uleb128 with label differences for DWARF v5, a feature that
235 # older binutils ports do not support when utilizing RISC-V style linker
236 # relaxation: https://sourceware.org/bugzilla/show_bug.cgi?id=27215
237 config AS_HAS_NON_CONST_ULEB128
238 def_bool $(as-instr,.uleb128 .Lexpr_end4 - .Lexpr_start3\n.Lexpr_start3:\n.Lexpr_end4:)
241 prompt "Debug information"
242 depends on DEBUG_KERNEL
244 Selecting something other than "None" results in a kernel image
245 that will include debugging info resulting in a larger kernel image.
246 This adds debug symbols to the kernel and modules (gcc -g), and
247 is needed if you intend to use kernel crashdump or binary object
248 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
250 Choose which version of DWARF debug info to emit. If unsure,
251 select "Toolchain default".
253 config DEBUG_INFO_NONE
254 bool "Disable debug information"
256 Do not build the kernel with debugging information, which will
257 result in a faster and smaller build.
259 config DEBUG_INFO_DWARF_TOOLCHAIN_DEFAULT
260 bool "Rely on the toolchain's implicit default DWARF version"
262 depends on !CC_IS_CLANG || AS_IS_LLVM || CLANG_VERSION < 140000 || (AS_IS_GNU && AS_VERSION >= 23502 && AS_HAS_NON_CONST_ULEB128)
264 The implicit default version of DWARF debug info produced by a
265 toolchain changes over time.
267 This can break consumers of the debug info that haven't upgraded to
268 support newer revisions, and prevent testing newer versions, but
269 those should be less common scenarios.
271 config DEBUG_INFO_DWARF4
272 bool "Generate DWARF Version 4 debuginfo"
274 depends on !CC_IS_CLANG || AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502)
276 Generate DWARF v4 debug info. This requires gcc 4.5+, binutils 2.35.2
277 if using clang without clang's integrated assembler, and gdb 7.0+.
279 If you have consumers of DWARF debug info that are not ready for
280 newer revisions of DWARF, you may wish to choose this or have your
283 config DEBUG_INFO_DWARF5
284 bool "Generate DWARF Version 5 debuginfo"
286 depends on !ARCH_HAS_BROKEN_DWARF5
287 depends on !CC_IS_CLANG || AS_IS_LLVM || (AS_IS_GNU && AS_VERSION >= 23502 && AS_HAS_NON_CONST_ULEB128)
289 Generate DWARF v5 debug info. Requires binutils 2.35.2, gcc 5.0+ (gcc
290 5.0+ accepts the -gdwarf-5 flag but only had partial support for some
291 draft features until 7.0), and gdb 8.0+.
293 Changes to the structure of debug info in Version 5 allow for around
294 15-18% savings in resulting image and debug info section sizes as
295 compared to DWARF Version 4. DWARF Version 5 standardizes previous
296 extensions such as accelerators for symbol indexing and the format
297 for fission (.dwo/.dwp) files. Users may not want to select this
298 config if they rely on tooling that has not yet been updated to
299 support DWARF Version 5.
301 endchoice # "Debug information"
305 config DEBUG_INFO_REDUCED
306 bool "Reduce debugging information"
308 If you say Y here gcc is instructed to generate less debugging
309 information for structure types. This means that tools that
310 need full debugging information (like kgdb or systemtap) won't
311 be happy. But if you merely need debugging information to
312 resolve line numbers there is no loss. Advantage is that
313 build directory object sizes shrink dramatically over a full
314 DEBUG_INFO build and compile times are reduced too.
315 Only works with newer gcc versions.
318 prompt "Compressed Debug information"
320 Compress the resulting debug info. Results in smaller debug info sections,
321 but requires that consumers are able to decompress the results.
323 If unsure, choose DEBUG_INFO_COMPRESSED_NONE.
325 config DEBUG_INFO_COMPRESSED_NONE
326 bool "Don't compress debug information"
328 Don't compress debug info sections.
330 config DEBUG_INFO_COMPRESSED_ZLIB
331 bool "Compress debugging information with zlib"
332 depends on $(cc-option,-gz=zlib)
333 depends on $(ld-option,--compress-debug-sections=zlib)
335 Compress the debug information using zlib. Requires GCC 5.0+ or Clang
336 5.0+, binutils 2.26+, and zlib.
338 Users of dpkg-deb via scripts/package/builddeb may find an increase in
339 size of their debug .deb packages with this config set, due to the
340 debug info being compressed with zlib, then the object files being
341 recompressed with a different compression scheme. But this is still
342 preferable to setting $KDEB_COMPRESS to "none" which would be even
345 config DEBUG_INFO_COMPRESSED_ZSTD
346 bool "Compress debugging information with zstd"
347 depends on $(cc-option,-gz=zstd)
348 depends on $(ld-option,--compress-debug-sections=zstd)
350 Compress the debug information using zstd. This may provide better
351 compression than zlib, for about the same time costs, but requires newer
352 toolchain support. Requires GCC 13.0+ or Clang 16.0+, binutils 2.40+, and
355 endchoice # "Compressed Debug information"
357 config DEBUG_INFO_SPLIT
358 bool "Produce split debuginfo in .dwo files"
359 depends on $(cc-option,-gsplit-dwarf)
360 # RISC-V linker relaxation + -gsplit-dwarf has issues with LLVM and GCC
362 # https://github.com/llvm/llvm-project/issues/56642
363 # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99090
364 depends on !RISCV || GCC_VERSION >= 120000
366 Generate debug info into separate .dwo files. This significantly
367 reduces the build directory size for builds with DEBUG_INFO,
368 because it stores the information only once on disk in .dwo
369 files instead of multiple times in object files and executables.
370 In addition the debug information is also compressed.
372 Requires recent gcc (4.7+) and recent gdb/binutils.
373 Any tool that packages or reads debug information would need
374 to know about the .dwo files and include them.
375 Incompatible with older versions of ccache.
377 config DEBUG_INFO_BTF
378 bool "Generate BTF typeinfo"
379 depends on !DEBUG_INFO_SPLIT && !DEBUG_INFO_REDUCED
380 depends on !GCC_PLUGIN_RANDSTRUCT || COMPILE_TEST
381 depends on BPF_SYSCALL
382 depends on !DEBUG_INFO_DWARF5 || PAHOLE_VERSION >= 121
383 # pahole uses elfutils, which does not have support for Hexagon relocations
386 Generate deduplicated BTF type information from DWARF debug info.
387 Turning this on expects presence of pahole tool, which will convert
388 DWARF type info into equivalent deduplicated BTF type info.
390 config PAHOLE_HAS_SPLIT_BTF
391 def_bool PAHOLE_VERSION >= 119
393 config PAHOLE_HAS_BTF_TAG
394 def_bool PAHOLE_VERSION >= 123
395 depends on CC_IS_CLANG
397 Decide whether pahole emits btf_tag attributes (btf_type_tag and
398 btf_decl_tag) or not. Currently only clang compiler implements
399 these attributes, so make the config depend on CC_IS_CLANG.
401 config PAHOLE_HAS_LANG_EXCLUDE
402 def_bool PAHOLE_VERSION >= 124
404 Support for the --lang_exclude flag which makes pahole exclude
405 compilation units from the supplied language. Used in Kbuild to
406 omit Rust CUs which are not supported in version 1.24 of pahole,
407 otherwise it would emit malformed kernel and module binaries when
408 using DEBUG_INFO_BTF_MODULES.
410 config DEBUG_INFO_BTF_MODULES
412 depends on DEBUG_INFO_BTF && MODULES && PAHOLE_HAS_SPLIT_BTF
414 Generate compact split BTF type information for kernel modules.
416 config MODULE_ALLOW_BTF_MISMATCH
417 bool "Allow loading modules with non-matching BTF type info"
418 depends on DEBUG_INFO_BTF_MODULES
420 For modules whose split BTF does not match vmlinux, load without
421 BTF rather than refusing to load. The default behavior with
422 module BTF enabled is to reject modules with such mismatches;
423 this option will still load module BTF where possible but ignore
424 it when a mismatch is found.
427 bool "Provide GDB scripts for kernel debugging"
429 This creates the required links to GDB helper scripts in the
430 build directory. If you load vmlinux into gdb, the helper
431 scripts will be automatically imported by gdb as well, and
432 additional functions are available to analyze a Linux kernel
433 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst
439 int "Warn for stack frames larger than"
442 default 2048 if GCC_PLUGIN_LATENT_ENTROPY
443 default 2048 if PARISC
444 default 1536 if (!64BIT && XTENSA)
445 default 1280 if KASAN && !64BIT
446 default 1024 if !64BIT
447 default 2048 if 64BIT
449 Tell the compiler to warn at build time for stack frames larger than this.
450 Setting this too low will cause a lot of warnings.
451 Setting it to 0 disables the warning.
453 config STRIP_ASM_SYMS
454 bool "Strip assembler-generated symbols during link"
457 Strip internal assembler-generated symbols during a link (symbols
458 that look like '.Lxxx') so they don't pollute the output of
459 get_wchan() and suchlike.
462 bool "Generate readable assembler code"
463 depends on DEBUG_KERNEL
466 Disable some compiler optimizations that tend to generate human unreadable
467 assembler output. This may make the kernel slightly slower, but it helps
468 to keep kernel developers who have to stare a lot at assembler listings
471 config HEADERS_INSTALL
472 bool "Install uapi headers to usr/include"
475 This option will install uapi headers (headers exported to user-space)
476 into the usr/include directory for use during the kernel build.
477 This is unneeded for building the kernel itself, but needed for some
478 user-space program samples. It is also needed by some features such
479 as uapi header sanity checks.
481 config DEBUG_SECTION_MISMATCH
482 bool "Enable full Section mismatch analysis"
485 The section mismatch analysis checks if there are illegal
486 references from one section to another section.
487 During linktime or runtime, some sections are dropped;
488 any use of code/data previously in these sections would
489 most likely result in an oops.
490 In the code, functions and variables are annotated with
491 __init,, etc. (see the full list in include/linux/init.h),
492 which results in the code/data being placed in specific sections.
493 The section mismatch analysis is always performed after a full
494 kernel build, and enabling this option causes the following
495 additional step to occur:
496 - Add the option -fno-inline-functions-called-once to gcc commands.
497 When inlining a function annotated with __init in a non-init
498 function, we would lose the section information and thus
499 the analysis would not catch the illegal reference.
500 This option tells gcc to inline less (but it does result in
503 config SECTION_MISMATCH_WARN_ONLY
504 bool "Make section mismatch errors non-fatal"
507 If you say N here, the build process will fail if there are any
508 section mismatch, instead of just throwing warnings.
512 config DEBUG_FORCE_FUNCTION_ALIGN_64B
513 bool "Force all function address 64B aligned"
514 depends on EXPERT && (X86_64 || ARM64 || PPC32 || PPC64 || ARC || RISCV || S390)
515 select FUNCTION_ALIGNMENT_64B
517 There are cases that a commit from one domain changes the function
518 address alignment of other domains, and cause magic performance
519 bump (regression or improvement). Enable this option will help to
520 verify if the bump is caused by function alignment changes, while
521 it will slightly increase the kernel size and affect icache usage.
523 It is mainly for debug and performance tuning use.
526 # Select this config option from the architecture Kconfig, if it
527 # is preferred to always offer frame pointers as a config
528 # option on the architecture (regardless of KERNEL_DEBUG):
530 config ARCH_WANT_FRAME_POINTERS
534 bool "Compile the kernel with frame pointers"
535 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS
536 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
538 If you say Y here the resulting kernel image will be slightly
539 larger and slower, but it gives very useful debugging information
540 in case of kernel bugs. (precise oopses/stacktraces/warnings)
545 config STACK_VALIDATION
546 bool "Compile-time stack metadata validation"
547 depends on HAVE_STACK_VALIDATION && UNWINDER_FRAME_POINTER
551 Validate frame pointer rules at compile-time. This helps ensure that
552 runtime stack traces are more reliable.
554 For more information, see
555 tools/objtool/Documentation/objtool.txt.
557 config NOINSTR_VALIDATION
559 depends on HAVE_NOINSTR_VALIDATION && DEBUG_ENTRY
564 bool "Generate vmlinux.map file when linking"
567 Selecting this option will pass "-Map=vmlinux.map" to ld
568 when linking vmlinux. That file can be useful for verifying
569 and debugging magic section games, and for seeing which
570 pieces of code get eliminated with
571 CONFIG_LD_DEAD_CODE_DATA_ELIMINATION.
573 config DEBUG_FORCE_WEAK_PER_CPU
574 bool "Force weak per-cpu definitions"
575 depends on DEBUG_KERNEL
577 s390 and alpha require percpu variables in modules to be
578 defined weak to work around addressing range issue which
579 puts the following two restrictions on percpu variable
582 1. percpu symbols must be unique whether static or not
583 2. percpu variables can't be defined inside a function
585 To ensure that generic code follows the above rules, this
586 option forces all percpu variables to be defined as weak.
588 endmenu # "Compiler options"
590 menu "Generic Kernel Debugging Instruments"
593 bool "Magic SysRq key"
596 If you say Y here, you will have some control over the system even
597 if the system crashes for example during kernel debugging (e.g., you
598 will be able to flush the buffer cache to disk, reboot the system
599 immediately or dump some status information). This is accomplished
600 by pressing various keys while holding SysRq (Alt+PrintScreen). It
601 also works on a serial console (on PC hardware at least), if you
602 send a BREAK and then within 5 seconds a command keypress. The
603 keys are documented in <file:Documentation/admin-guide/sysrq.rst>.
604 Don't say Y unless you really know what this hack does.
606 config MAGIC_SYSRQ_DEFAULT_ENABLE
607 hex "Enable magic SysRq key functions by default"
608 depends on MAGIC_SYSRQ
611 Specifies which SysRq key functions are enabled by default.
612 This may be set to 1 or 0 to enable or disable them all, or
613 to a bitmask as described in Documentation/admin-guide/sysrq.rst.
615 config MAGIC_SYSRQ_SERIAL
616 bool "Enable magic SysRq key over serial"
617 depends on MAGIC_SYSRQ
620 Many embedded boards have a disconnected TTL level serial which can
621 generate some garbage that can lead to spurious false sysrq detects.
622 This option allows you to decide whether you want to enable the
625 config MAGIC_SYSRQ_SERIAL_SEQUENCE
626 string "Char sequence that enables magic SysRq over serial"
627 depends on MAGIC_SYSRQ_SERIAL
630 Specifies a sequence of characters that can follow BREAK to enable
631 SysRq on a serial console.
633 If unsure, leave an empty string and the option will not be enabled.
636 bool "Debug Filesystem"
638 debugfs is a virtual file system that kernel developers use to put
639 debugging files into. Enable this option to be able to read and
640 write to these files.
642 For detailed documentation on the debugfs API, see
643 Documentation/filesystems/.
648 prompt "Debugfs default access"
650 default DEBUG_FS_ALLOW_ALL
652 This selects the default access restrictions for debugfs.
653 It can be overridden with kernel command line option
654 debugfs=[on,no-mount,off]. The restrictions apply for API access
655 and filesystem registration.
657 config DEBUG_FS_ALLOW_ALL
660 No restrictions apply. Both API and filesystem registration
661 is on. This is the normal default operation.
663 config DEBUG_FS_DISALLOW_MOUNT
664 bool "Do not register debugfs as filesystem"
666 The API is open but filesystem is not loaded. Clients can still do
667 their work and read with debug tools that do not need
670 config DEBUG_FS_ALLOW_NONE
673 Access is off. Clients get -PERM when trying to create nodes in
674 debugfs tree and debugfs is not registered as a filesystem.
675 Client can then back-off or continue without debugfs access.
679 source "lib/Kconfig.kgdb"
680 source "lib/Kconfig.ubsan"
681 source "lib/Kconfig.kcsan"
685 menu "Networking Debugging"
687 source "net/Kconfig.debug"
689 endmenu # "Networking Debugging"
691 menu "Memory Debugging"
693 source "mm/Kconfig.debug"
696 bool "Debug object operations"
697 depends on DEBUG_KERNEL
699 If you say Y here, additional code will be inserted into the
700 kernel to track the life time of various objects and validate
701 the operations on those objects.
703 config DEBUG_OBJECTS_SELFTEST
704 bool "Debug objects selftest"
705 depends on DEBUG_OBJECTS
707 This enables the selftest of the object debug code.
709 config DEBUG_OBJECTS_FREE
710 bool "Debug objects in freed memory"
711 depends on DEBUG_OBJECTS
713 This enables checks whether a k/v free operation frees an area
714 which contains an object which has not been deactivated
715 properly. This can make kmalloc/kfree-intensive workloads
718 config DEBUG_OBJECTS_TIMERS
719 bool "Debug timer objects"
720 depends on DEBUG_OBJECTS
722 If you say Y here, additional code will be inserted into the
723 timer routines to track the life time of timer objects and
724 validate the timer operations.
726 config DEBUG_OBJECTS_WORK
727 bool "Debug work objects"
728 depends on DEBUG_OBJECTS
730 If you say Y here, additional code will be inserted into the
731 work queue routines to track the life time of work objects and
732 validate the work operations.
734 config DEBUG_OBJECTS_RCU_HEAD
735 bool "Debug RCU callbacks objects"
736 depends on DEBUG_OBJECTS
738 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
740 config DEBUG_OBJECTS_PERCPU_COUNTER
741 bool "Debug percpu counter objects"
742 depends on DEBUG_OBJECTS
744 If you say Y here, additional code will be inserted into the
745 percpu counter routines to track the life time of percpu counter
746 objects and validate the percpu counter operations.
748 config DEBUG_OBJECTS_ENABLE_DEFAULT
749 int "debug_objects bootup default value (0-1)"
752 depends on DEBUG_OBJECTS
754 Debug objects boot parameter default value
756 config SHRINKER_DEBUG
757 bool "Enable shrinker debugging support"
760 Say Y to enable the shrinker debugfs interface which provides
761 visibility into the kernel memory shrinkers subsystem.
762 Disable it to avoid an extra memory footprint.
764 config DEBUG_STACK_USAGE
765 bool "Stack utilization instrumentation"
766 depends on DEBUG_KERNEL
768 Enables the display of the minimum amount of free stack which each
769 task has ever had available in the sysrq-T and sysrq-P debug output.
770 Also emits a message to dmesg when a process exits if that process
771 used more stack space than previously exiting processes.
773 This option will slow down process creation somewhat.
775 config SCHED_STACK_END_CHECK
776 bool "Detect stack corruption on calls to schedule()"
777 depends on DEBUG_KERNEL
780 This option checks for a stack overrun on calls to schedule().
781 If the stack end location is found to be over written always panic as
782 the content of the corrupted region can no longer be trusted.
783 This is to ensure no erroneous behaviour occurs which could result in
784 data corruption or a sporadic crash at a later stage once the region
785 is examined. The runtime overhead introduced is minimal.
787 config ARCH_HAS_DEBUG_VM_PGTABLE
790 An architecture should select this when it can successfully
791 build and run DEBUG_VM_PGTABLE.
793 config DEBUG_VM_IRQSOFF
794 def_bool DEBUG_VM && !PREEMPT_RT
798 depends on DEBUG_KERNEL
800 Enable this to turn on extended checks in the virtual-memory system
801 that may impact performance.
805 config DEBUG_VM_SHOOT_LAZIES
806 bool "Debug MMU_LAZY_TLB_SHOOTDOWN implementation"
808 depends on MMU_LAZY_TLB_SHOOTDOWN
810 Enable additional IPIs that ensure lazy tlb mm references are removed
811 before the mm is freed.
815 config DEBUG_VM_MAPLE_TREE
816 bool "Debug VM maple trees"
818 select DEBUG_MAPLE_TREE
820 Enable VM maple tree debugging information and extra validations.
825 bool "Debug VM red-black trees"
828 Enable VM red-black tree debugging information and extra validations.
832 config DEBUG_VM_PGFLAGS
833 bool "Debug page-flags operations"
836 Enables extra validation on page flags operations.
840 config DEBUG_VM_PGTABLE
841 bool "Debug arch page table for semantics compliance"
843 depends on ARCH_HAS_DEBUG_VM_PGTABLE
844 default y if DEBUG_VM
846 This option provides a debug method which can be used to test
847 architecture page table helper functions on various platforms in
848 verifying if they comply with expected generic MM semantics. This
849 will help architecture code in making sure that any changes or
850 new additions of these helpers still conform to expected
851 semantics of the generic MM. Platforms will have to opt in for
852 this through ARCH_HAS_DEBUG_VM_PGTABLE.
856 config ARCH_HAS_DEBUG_VIRTUAL
860 bool "Debug VM translations"
861 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL
863 Enable some costly sanity checks in virtual to page code. This can
864 catch mistakes with virt_to_page() and friends.
868 config DEBUG_NOMMU_REGIONS
869 bool "Debug the global anon/private NOMMU mapping region tree"
870 depends on DEBUG_KERNEL && !MMU
872 This option causes the global tree of anonymous and private mapping
873 regions to be regularly checked for invalid topology.
875 config DEBUG_MEMORY_INIT
876 bool "Debug memory initialisation" if EXPERT
879 Enable this for additional checks during memory initialisation.
880 The sanity checks verify aspects of the VM such as the memory model
881 and other information provided by the architecture. Verbose
882 information will be printed at KERN_DEBUG loglevel depending
883 on the mminit_loglevel= command-line option.
887 config MEMORY_NOTIFIER_ERROR_INJECT
888 tristate "Memory hotplug notifier error injection module"
889 depends on MEMORY_HOTPLUG && NOTIFIER_ERROR_INJECTION
891 This option provides the ability to inject artificial errors to
892 memory hotplug notifier chain callbacks. It is controlled through
893 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
895 If the notifier call chain should be failed with some events
896 notified, write the error code to "actions/<notifier event>/error".
898 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
900 # cd /sys/kernel/debug/notifier-error-inject/memory
901 # echo -12 > actions/MEM_GOING_OFFLINE/error
902 # echo offline > /sys/devices/system/memory/memoryXXX/state
903 bash: echo: write error: Cannot allocate memory
905 To compile this code as a module, choose M here: the module will
906 be called memory-notifier-error-inject.
910 config DEBUG_PER_CPU_MAPS
911 bool "Debug access to per_cpu maps"
912 depends on DEBUG_KERNEL
915 Say Y to verify that the per_cpu map being accessed has
916 been set up. This adds a fair amount of code to kernel memory
917 and decreases performance.
921 config DEBUG_KMAP_LOCAL
922 bool "Debug kmap_local temporary mappings"
923 depends on DEBUG_KERNEL && KMAP_LOCAL
925 This option enables additional error checking for the kmap_local
926 infrastructure. Disable for production use.
928 config ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP
931 config DEBUG_KMAP_LOCAL_FORCE_MAP
932 bool "Enforce kmap_local temporary mappings"
933 depends on DEBUG_KERNEL && ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP
935 select DEBUG_KMAP_LOCAL
937 This option enforces temporary mappings through the kmap_local
938 mechanism for non-highmem pages and on non-highmem systems.
939 Disable this for production systems!
942 bool "Highmem debugging"
943 depends on DEBUG_KERNEL && HIGHMEM
944 select DEBUG_KMAP_LOCAL_FORCE_MAP if ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP
945 select DEBUG_KMAP_LOCAL
947 This option enables additional error checking for high memory
948 systems. Disable for production systems.
950 config HAVE_DEBUG_STACKOVERFLOW
953 config DEBUG_STACKOVERFLOW
954 bool "Check for stack overflows"
955 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW
957 Say Y here if you want to check for overflows of kernel, IRQ
958 and exception stacks (if your architecture uses them). This
959 option will show detailed messages if free stack space drops
960 below a certain limit.
962 These kinds of bugs usually occur when call-chains in the
963 kernel get too deep, especially when interrupts are
966 Use this in cases where you see apparently random memory
967 corruption, especially if it appears in 'struct thread_info'
969 If in doubt, say "N".
971 source "lib/Kconfig.kasan"
972 source "lib/Kconfig.kfence"
973 source "lib/Kconfig.kmsan"
975 endmenu # "Memory Debugging"
978 bool "Debug shared IRQ handlers"
979 depends on DEBUG_KERNEL
981 Enable this to generate a spurious interrupt just before a shared
982 interrupt handler is deregistered (generating one when registering
983 is currently disabled). Drivers need to handle this correctly. Some
984 don't and need to be caught.
986 menu "Debug Oops, Lockups and Hangs"
991 Say Y here to enable the kernel to panic when it oopses. This
992 has the same effect as setting oops=panic on the kernel command
995 This feature is useful to ensure that the kernel does not do
996 anything erroneous after an oops which could result in data
997 corruption or other issues.
1001 config PANIC_ON_OOPS_VALUE
1004 default 0 if !PANIC_ON_OOPS
1005 default 1 if PANIC_ON_OOPS
1007 config PANIC_TIMEOUT
1011 Set the timeout value (in seconds) until a reboot occurs when
1012 the kernel panics. If n = 0, then we wait forever. A timeout
1013 value n > 0 will wait n seconds before rebooting, while a timeout
1014 value n < 0 will reboot immediately.
1016 config LOCKUP_DETECTOR
1019 config SOFTLOCKUP_DETECTOR
1020 bool "Detect Soft Lockups"
1021 depends on DEBUG_KERNEL && !S390
1022 select LOCKUP_DETECTOR
1024 Say Y here to enable the kernel to act as a watchdog to detect
1027 Softlockups are bugs that cause the kernel to loop in kernel
1028 mode for more than 20 seconds, without giving other tasks a
1029 chance to run. The current stack trace is displayed upon
1030 detection and the system will stay locked up.
1032 config BOOTPARAM_SOFTLOCKUP_PANIC
1033 bool "Panic (Reboot) On Soft Lockups"
1034 depends on SOFTLOCKUP_DETECTOR
1036 Say Y here to enable the kernel to panic on "soft lockups",
1037 which are bugs that cause the kernel to loop in kernel
1038 mode for more than 20 seconds (configurable using the watchdog_thresh
1039 sysctl), without giving other tasks a chance to run.
1041 The panic can be used in combination with panic_timeout,
1042 to cause the system to reboot automatically after a
1043 lockup has been detected. This feature is useful for
1044 high-availability systems that have uptime guarantees and
1045 where a lockup must be resolved ASAP.
1049 config HAVE_HARDLOCKUP_DETECTOR_BUDDY
1055 # Global switch whether to build a hardlockup detector at all. It is available
1056 # only when the architecture supports at least one implementation. There are
1057 # two exceptions. The hardlockup detector is never enabled on:
1059 # s390: it reported many false positives there
1061 # sparc64: has a custom implementation which is not using the common
1062 # hardlockup command line options and sysctl interface.
1064 config HARDLOCKUP_DETECTOR
1065 bool "Detect Hard Lockups"
1066 depends on DEBUG_KERNEL && !S390 && !HARDLOCKUP_DETECTOR_SPARC64
1067 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_BUDDY || HAVE_HARDLOCKUP_DETECTOR_ARCH
1068 imply HARDLOCKUP_DETECTOR_PERF
1069 imply HARDLOCKUP_DETECTOR_BUDDY
1070 imply HARDLOCKUP_DETECTOR_ARCH
1071 select LOCKUP_DETECTOR
1074 Say Y here to enable the kernel to act as a watchdog to detect
1077 Hardlockups are bugs that cause the CPU to loop in kernel mode
1078 for more than 10 seconds, without letting other interrupts have a
1079 chance to run. The current stack trace is displayed upon detection
1080 and the system will stay locked up.
1083 # Note that arch-specific variants are always preferred.
1085 config HARDLOCKUP_DETECTOR_PREFER_BUDDY
1086 bool "Prefer the buddy CPU hardlockup detector"
1087 depends on HARDLOCKUP_DETECTOR
1088 depends on HAVE_HARDLOCKUP_DETECTOR_PERF && HAVE_HARDLOCKUP_DETECTOR_BUDDY
1089 depends on !HAVE_HARDLOCKUP_DETECTOR_ARCH
1091 Say Y here to prefer the buddy hardlockup detector over the perf one.
1093 With the buddy detector, each CPU uses its softlockup hrtimer
1094 to check that the next CPU is processing hrtimer interrupts by
1095 verifying that a counter is increasing.
1097 This hardlockup detector is useful on systems that don't have
1098 an arch-specific hardlockup detector or if resources needed
1099 for the hardlockup detector are better used for other things.
1101 config HARDLOCKUP_DETECTOR_PERF
1103 depends on HARDLOCKUP_DETECTOR
1104 depends on HAVE_HARDLOCKUP_DETECTOR_PERF && !HARDLOCKUP_DETECTOR_PREFER_BUDDY
1105 depends on !HAVE_HARDLOCKUP_DETECTOR_ARCH
1106 select HARDLOCKUP_DETECTOR_COUNTS_HRTIMER
1108 config HARDLOCKUP_DETECTOR_BUDDY
1110 depends on HARDLOCKUP_DETECTOR
1111 depends on HAVE_HARDLOCKUP_DETECTOR_BUDDY
1112 depends on !HAVE_HARDLOCKUP_DETECTOR_PERF || HARDLOCKUP_DETECTOR_PREFER_BUDDY
1113 depends on !HAVE_HARDLOCKUP_DETECTOR_ARCH
1114 select HARDLOCKUP_DETECTOR_COUNTS_HRTIMER
1116 config HARDLOCKUP_DETECTOR_ARCH
1118 depends on HARDLOCKUP_DETECTOR
1119 depends on HAVE_HARDLOCKUP_DETECTOR_ARCH
1121 The arch-specific implementation of the hardlockup detector will
1125 # Both the "perf" and "buddy" hardlockup detectors count hrtimer
1126 # interrupts. This config enables functions managing this common code.
1128 config HARDLOCKUP_DETECTOR_COUNTS_HRTIMER
1130 select SOFTLOCKUP_DETECTOR
1133 # Enables a timestamp based low pass filter to compensate for perf based
1134 # hard lockup detection which runs too fast due to turbo modes.
1136 config HARDLOCKUP_CHECK_TIMESTAMP
1139 config BOOTPARAM_HARDLOCKUP_PANIC
1140 bool "Panic (Reboot) On Hard Lockups"
1141 depends on HARDLOCKUP_DETECTOR
1143 Say Y here to enable the kernel to panic on "hard lockups",
1144 which are bugs that cause the kernel to loop in kernel
1145 mode with interrupts disabled for more than 10 seconds (configurable
1146 using the watchdog_thresh sysctl).
1150 config DETECT_HUNG_TASK
1151 bool "Detect Hung Tasks"
1152 depends on DEBUG_KERNEL
1153 default SOFTLOCKUP_DETECTOR
1155 Say Y here to enable the kernel to detect "hung tasks",
1156 which are bugs that cause the task to be stuck in
1157 uninterruptible "D" state indefinitely.
1159 When a hung task is detected, the kernel will print the
1160 current stack trace (which you should report), but the
1161 task will stay in uninterruptible state. If lockdep is
1162 enabled then all held locks will also be reported. This
1163 feature has negligible overhead.
1165 config DEFAULT_HUNG_TASK_TIMEOUT
1166 int "Default timeout for hung task detection (in seconds)"
1167 depends on DETECT_HUNG_TASK
1170 This option controls the default timeout (in seconds) used
1171 to determine when a task has become non-responsive and should
1174 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
1175 sysctl or by writing a value to
1176 /proc/sys/kernel/hung_task_timeout_secs.
1178 A timeout of 0 disables the check. The default is two minutes.
1179 Keeping the default should be fine in most cases.
1181 config BOOTPARAM_HUNG_TASK_PANIC
1182 bool "Panic (Reboot) On Hung Tasks"
1183 depends on DETECT_HUNG_TASK
1185 Say Y here to enable the kernel to panic on "hung tasks",
1186 which are bugs that cause the kernel to leave a task stuck
1187 in uninterruptible "D" state.
1189 The panic can be used in combination with panic_timeout,
1190 to cause the system to reboot automatically after a
1191 hung task has been detected. This feature is useful for
1192 high-availability systems that have uptime guarantees and
1193 where a hung tasks must be resolved ASAP.
1198 bool "Detect Workqueue Stalls"
1199 depends on DEBUG_KERNEL
1201 Say Y here to enable stall detection on workqueues. If a
1202 worker pool doesn't make forward progress on a pending work
1203 item for over a given amount of time, 30s by default, a
1204 warning message is printed along with dump of workqueue
1205 state. This can be configured through kernel parameter
1206 "workqueue.watchdog_thresh" and its sysfs counterpart.
1208 config WQ_CPU_INTENSIVE_REPORT
1209 bool "Report per-cpu work items which hog CPU for too long"
1210 depends on DEBUG_KERNEL
1212 Say Y here to enable reporting of concurrency-managed per-cpu work
1213 items that hog CPUs for longer than
1214 workqueue.cpu_intensive_thresh_us. Workqueue automatically
1215 detects and excludes them from concurrency management to prevent
1216 them from stalling other per-cpu work items. Occassional
1217 triggering may not necessarily indicate a problem. Repeated
1218 triggering likely indicates that the work item should be switched
1219 to use an unbound workqueue.
1222 tristate "Test module to generate lockups"
1225 This builds the "test_lockup" module that helps to make sure
1226 that watchdogs and lockup detectors are working properly.
1228 Depending on module parameters it could emulate soft or hard
1229 lockup, "hung task", or locking arbitrary lock for a long time.
1230 Also it could generate series of lockups with cooling-down periods.
1234 endmenu # "Debug lockups and hangs"
1236 menu "Scheduler Debugging"
1239 bool "Collect scheduler debugging info"
1240 depends on DEBUG_KERNEL && DEBUG_FS
1243 If you say Y here, the /sys/kernel/debug/sched file will be provided
1244 that can help debug the scheduler. The runtime overhead of this
1252 bool "Collect scheduler statistics"
1253 depends on DEBUG_KERNEL && PROC_FS
1256 If you say Y here, additional code will be inserted into the
1257 scheduler and related routines to collect statistics about
1258 scheduler behavior and provide them in /proc/schedstat. These
1259 stats may be useful for both tuning and debugging the scheduler
1260 If you aren't debugging the scheduler or trying to tune a specific
1261 application, you can say N to avoid the very slight overhead
1266 config DEBUG_TIMEKEEPING
1267 bool "Enable extra timekeeping sanity checking"
1269 This option will enable additional timekeeping sanity checks
1270 which may be helpful when diagnosing issues where timekeeping
1271 problems are suspected.
1273 This may include checks in the timekeeping hotpaths, so this
1274 option may have a (very small) performance impact to some
1279 config DEBUG_PREEMPT
1280 bool "Debug preemptible kernel"
1281 depends on DEBUG_KERNEL && PREEMPTION && TRACE_IRQFLAGS_SUPPORT
1283 If you say Y here then the kernel will use a debug variant of the
1284 commonly used smp_processor_id() function and will print warnings
1285 if kernel code uses it in a preemption-unsafe way. Also, the kernel
1286 will detect preemption count underflows.
1288 This option has potential to introduce high runtime overhead,
1289 depending on workload as it triggers debugging routines for each
1290 this_cpu operation. It should only be used for debugging purposes.
1292 menu "Lock Debugging (spinlocks, mutexes, etc...)"
1294 config LOCK_DEBUGGING_SUPPORT
1296 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1299 config PROVE_LOCKING
1300 bool "Lock debugging: prove locking correctness"
1301 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1303 select DEBUG_SPINLOCK
1304 select DEBUG_MUTEXES if !PREEMPT_RT
1305 select DEBUG_RT_MUTEXES if RT_MUTEXES
1306 select DEBUG_RWSEMS if !PREEMPT_RT
1307 select DEBUG_WW_MUTEX_SLOWPATH
1308 select DEBUG_LOCK_ALLOC
1309 select PREEMPT_COUNT if !ARCH_NO_PREEMPT
1310 select TRACE_IRQFLAGS
1313 This feature enables the kernel to prove that all locking
1314 that occurs in the kernel runtime is mathematically
1315 correct: that under no circumstance could an arbitrary (and
1316 not yet triggered) combination of observed locking
1317 sequences (on an arbitrary number of CPUs, running an
1318 arbitrary number of tasks and interrupt contexts) cause a
1321 In short, this feature enables the kernel to report locking
1322 related deadlocks before they actually occur.
1324 The proof does not depend on how hard and complex a
1325 deadlock scenario would be to trigger: how many
1326 participant CPUs, tasks and irq-contexts would be needed
1327 for it to trigger. The proof also does not depend on
1328 timing: if a race and a resulting deadlock is possible
1329 theoretically (no matter how unlikely the race scenario
1330 is), it will be proven so and will immediately be
1331 reported by the kernel (once the event is observed that
1332 makes the deadlock theoretically possible).
1334 If a deadlock is impossible (i.e. the locking rules, as
1335 observed by the kernel, are mathematically correct), the
1336 kernel reports nothing.
1338 NOTE: this feature can also be enabled for rwlocks, mutexes
1339 and rwsems - in which case all dependencies between these
1340 different locking variants are observed and mapped too, and
1341 the proof of observed correctness is also maintained for an
1342 arbitrary combination of these separate locking variants.
1344 For more details, see Documentation/locking/lockdep-design.rst.
1346 config PROVE_RAW_LOCK_NESTING
1347 bool "Enable raw_spinlock - spinlock nesting checks"
1348 depends on PROVE_LOCKING
1351 Enable the raw_spinlock vs. spinlock nesting checks which ensure
1352 that the lock nesting rules for PREEMPT_RT enabled kernels are
1355 NOTE: There are known nesting problems. So if you enable this
1356 option expect lockdep splats until these problems have been fully
1357 addressed which is work in progress. This config switch allows to
1358 identify and analyze these problems. It will be removed and the
1359 check permanently enabled once the main issues have been fixed.
1361 If unsure, select N.
1364 bool "Lock usage statistics"
1365 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1367 select DEBUG_SPINLOCK
1368 select DEBUG_MUTEXES if !PREEMPT_RT
1369 select DEBUG_RT_MUTEXES if RT_MUTEXES
1370 select DEBUG_LOCK_ALLOC
1373 This feature enables tracking lock contention points
1375 For more details, see Documentation/locking/lockstat.rst
1377 This also enables lock events required by "perf lock",
1379 If you want to use "perf lock", you also need to turn on
1380 CONFIG_EVENT_TRACING.
1382 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
1383 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
1385 config DEBUG_RT_MUTEXES
1386 bool "RT Mutex debugging, deadlock detection"
1387 depends on DEBUG_KERNEL && RT_MUTEXES
1389 This allows rt mutex semantics violations and rt mutex related
1390 deadlocks (lockups) to be detected and reported automatically.
1392 config DEBUG_SPINLOCK
1393 bool "Spinlock and rw-lock debugging: basic checks"
1394 depends on DEBUG_KERNEL
1395 select UNINLINE_SPIN_UNLOCK
1397 Say Y here and build SMP to catch missing spinlock initialization
1398 and certain other kinds of spinlock errors commonly made. This is
1399 best used in conjunction with the NMI watchdog so that spinlock
1400 deadlocks are also debuggable.
1402 config DEBUG_MUTEXES
1403 bool "Mutex debugging: basic checks"
1404 depends on DEBUG_KERNEL && !PREEMPT_RT
1406 This feature allows mutex semantics violations to be detected and
1409 config DEBUG_WW_MUTEX_SLOWPATH
1410 bool "Wait/wound mutex debugging: Slowpath testing"
1411 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1412 select DEBUG_LOCK_ALLOC
1413 select DEBUG_SPINLOCK
1414 select DEBUG_MUTEXES if !PREEMPT_RT
1415 select DEBUG_RT_MUTEXES if PREEMPT_RT
1417 This feature enables slowpath testing for w/w mutex users by
1418 injecting additional -EDEADLK wound/backoff cases. Together with
1419 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this
1420 will test all possible w/w mutex interface abuse with the
1421 exception of simply not acquiring all the required locks.
1422 Note that this feature can introduce significant overhead, so
1423 it really should not be enabled in a production or distro kernel,
1424 even a debug kernel. If you are a driver writer, enable it. If
1425 you are a distro, do not.
1428 bool "RW Semaphore debugging: basic checks"
1429 depends on DEBUG_KERNEL && !PREEMPT_RT
1431 This debugging feature allows mismatched rw semaphore locks
1432 and unlocks to be detected and reported.
1434 config DEBUG_LOCK_ALLOC
1435 bool "Lock debugging: detect incorrect freeing of live locks"
1436 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1437 select DEBUG_SPINLOCK
1438 select DEBUG_MUTEXES if !PREEMPT_RT
1439 select DEBUG_RT_MUTEXES if RT_MUTEXES
1442 This feature will check whether any held lock (spinlock, rwlock,
1443 mutex or rwsem) is incorrectly freed by the kernel, via any of the
1444 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
1445 vfree(), etc.), whether a live lock is incorrectly reinitialized via
1446 spin_lock_init()/mutex_init()/etc., or whether there is any lock
1447 held during task exit.
1451 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1456 config LOCKDEP_SMALL
1460 int "Bitsize for MAX_LOCKDEP_ENTRIES"
1461 depends on LOCKDEP && !LOCKDEP_SMALL
1465 Try increasing this value if you hit "BUG: MAX_LOCKDEP_ENTRIES too low!" message.
1467 config LOCKDEP_CHAINS_BITS
1468 int "Bitsize for MAX_LOCKDEP_CHAINS"
1469 depends on LOCKDEP && !LOCKDEP_SMALL
1473 Try increasing this value if you hit "BUG: MAX_LOCKDEP_CHAINS too low!" message.
1475 config LOCKDEP_STACK_TRACE_BITS
1476 int "Bitsize for MAX_STACK_TRACE_ENTRIES"
1477 depends on LOCKDEP && !LOCKDEP_SMALL
1481 Try increasing this value if you hit "BUG: MAX_STACK_TRACE_ENTRIES too low!" message.
1483 config LOCKDEP_STACK_TRACE_HASH_BITS
1484 int "Bitsize for STACK_TRACE_HASH_SIZE"
1485 depends on LOCKDEP && !LOCKDEP_SMALL
1489 Try increasing this value if you need large STACK_TRACE_HASH_SIZE.
1491 config LOCKDEP_CIRCULAR_QUEUE_BITS
1492 int "Bitsize for elements in circular_queue struct"
1497 Try increasing this value if you hit "lockdep bfs error:-1" warning due to __cq_enqueue() failure.
1499 config DEBUG_LOCKDEP
1500 bool "Lock dependency engine debugging"
1501 depends on DEBUG_KERNEL && LOCKDEP
1502 select DEBUG_IRQFLAGS
1504 If you say Y here, the lock dependency engine will do
1505 additional runtime checks to debug itself, at the price
1506 of more runtime overhead.
1508 config DEBUG_ATOMIC_SLEEP
1509 bool "Sleep inside atomic section checking"
1510 select PREEMPT_COUNT
1511 depends on DEBUG_KERNEL
1512 depends on !ARCH_NO_PREEMPT
1514 If you say Y here, various routines which may sleep will become very
1515 noisy if they are called inside atomic sections: when a spinlock is
1516 held, inside an rcu read side critical section, inside preempt disabled
1517 sections, inside an interrupt, etc...
1519 config DEBUG_LOCKING_API_SELFTESTS
1520 bool "Locking API boot-time self-tests"
1521 depends on DEBUG_KERNEL
1523 Say Y here if you want the kernel to run a short self-test during
1524 bootup. The self-test checks whether common types of locking bugs
1525 are detected by debugging mechanisms or not. (if you disable
1526 lock debugging then those bugs won't be detected of course.)
1527 The following locking APIs are covered: spinlocks, rwlocks,
1530 config LOCK_TORTURE_TEST
1531 tristate "torture tests for locking"
1532 depends on DEBUG_KERNEL
1535 This option provides a kernel module that runs torture tests
1536 on kernel locking primitives. The kernel module may be built
1537 after the fact on the running kernel to be tested, if desired.
1539 Say Y here if you want kernel locking-primitive torture tests
1540 to be built into the kernel.
1541 Say M if you want these torture tests to build as a module.
1542 Say N if you are unsure.
1544 config WW_MUTEX_SELFTEST
1545 tristate "Wait/wound mutex selftests"
1547 This option provides a kernel module that runs tests on the
1548 on the struct ww_mutex locking API.
1550 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction
1551 with this test harness.
1553 Say M if you want these self tests to build as a module.
1554 Say N if you are unsure.
1556 config SCF_TORTURE_TEST
1557 tristate "torture tests for smp_call_function*()"
1558 depends on DEBUG_KERNEL
1561 This option provides a kernel module that runs torture tests
1562 on the smp_call_function() family of primitives. The kernel
1563 module may be built after the fact on the running kernel to
1564 be tested, if desired.
1566 config CSD_LOCK_WAIT_DEBUG
1567 bool "Debugging for csd_lock_wait(), called from smp_call_function*()"
1568 depends on DEBUG_KERNEL
1572 This option enables debug prints when CPUs are slow to respond
1573 to the smp_call_function*() IPI wrappers. These debug prints
1574 include the IPI handler function currently executing (if any)
1575 and relevant stack traces.
1577 config CSD_LOCK_WAIT_DEBUG_DEFAULT
1578 bool "Default csd_lock_wait() debugging on at boot time"
1579 depends on CSD_LOCK_WAIT_DEBUG
1583 This option causes the csdlock_debug= kernel boot parameter to
1584 default to 1 (basic debugging) instead of 0 (no debugging).
1586 endmenu # lock debugging
1588 config TRACE_IRQFLAGS
1589 depends on TRACE_IRQFLAGS_SUPPORT
1592 Enables hooks to interrupt enabling and disabling for
1593 either tracing or lock debugging.
1595 config TRACE_IRQFLAGS_NMI
1597 depends on TRACE_IRQFLAGS
1598 depends on TRACE_IRQFLAGS_NMI_SUPPORT
1600 config NMI_CHECK_CPU
1601 bool "Debugging for CPUs failing to respond to backtrace requests"
1602 depends on DEBUG_KERNEL
1606 Enables debug prints when a CPU fails to respond to a given
1607 backtrace NMI. These prints provide some reasons why a CPU
1608 might legitimately be failing to respond, for example, if it
1609 is offline of if ignore_nmis is set.
1611 config DEBUG_IRQFLAGS
1612 bool "Debug IRQ flag manipulation"
1614 Enables checks for potentially unsafe enabling or disabling of
1615 interrupts, such as calling raw_local_irq_restore() when interrupts
1619 bool "Stack backtrace support"
1620 depends on STACKTRACE_SUPPORT
1622 This option causes the kernel to create a /proc/pid/stack for
1623 every process, showing its current stack trace.
1624 It is also used by various kernel debugging features that require
1625 stack trace generation.
1627 config WARN_ALL_UNSEEDED_RANDOM
1628 bool "Warn for all uses of unseeded randomness"
1631 Some parts of the kernel contain bugs relating to their use of
1632 cryptographically secure random numbers before it's actually possible
1633 to generate those numbers securely. This setting ensures that these
1634 flaws don't go unnoticed, by enabling a message, should this ever
1635 occur. This will allow people with obscure setups to know when things
1636 are going wrong, so that they might contact developers about fixing
1639 Unfortunately, on some models of some architectures getting
1640 a fully seeded CRNG is extremely difficult, and so this can
1641 result in dmesg getting spammed for a surprisingly long
1642 time. This is really bad from a security perspective, and
1643 so architecture maintainers really need to do what they can
1644 to get the CRNG seeded sooner after the system is booted.
1645 However, since users cannot do anything actionable to
1646 address this, by default this option is disabled.
1648 Say Y here if you want to receive warnings for all uses of
1649 unseeded randomness. This will be of use primarily for
1650 those developers interested in improving the security of
1651 Linux kernels running on their architecture (or
1654 config DEBUG_KOBJECT
1655 bool "kobject debugging"
1656 depends on DEBUG_KERNEL
1658 If you say Y here, some extra kobject debugging messages will be sent
1661 config DEBUG_KOBJECT_RELEASE
1662 bool "kobject release debugging"
1663 depends on DEBUG_OBJECTS_TIMERS
1665 kobjects are reference counted objects. This means that their
1666 last reference count put is not predictable, and the kobject can
1667 live on past the point at which a driver decides to drop its
1668 initial reference to the kobject gained on allocation. An
1669 example of this would be a struct device which has just been
1672 However, some buggy drivers assume that after such an operation,
1673 the memory backing the kobject can be immediately freed. This
1674 goes completely against the principles of a refcounted object.
1676 If you say Y here, the kernel will delay the release of kobjects
1677 on the last reference count to improve the visibility of this
1678 kind of kobject release bug.
1680 config HAVE_DEBUG_BUGVERBOSE
1683 menu "Debug kernel data structures"
1686 bool "Debug linked list manipulation"
1687 depends on DEBUG_KERNEL
1688 select LIST_HARDENED
1690 Enable this to turn on extended checks in the linked-list walking
1693 This option trades better quality error reports for performance, and
1694 is more suitable for kernel debugging. If you care about performance,
1695 you should only enable CONFIG_LIST_HARDENED instead.
1700 bool "Debug priority linked list manipulation"
1701 depends on DEBUG_KERNEL
1703 Enable this to turn on extended checks in the priority-ordered
1704 linked-list (plist) walking routines. This checks the entire
1705 list multiple times during each manipulation.
1710 bool "Debug SG table operations"
1711 depends on DEBUG_KERNEL
1713 Enable this to turn on checks on scatter-gather tables. This can
1714 help find problems with drivers that do not properly initialize
1719 config DEBUG_NOTIFIERS
1720 bool "Debug notifier call chains"
1721 depends on DEBUG_KERNEL
1723 Enable this to turn on sanity checking for notifier call chains.
1724 This is most useful for kernel developers to make sure that
1725 modules properly unregister themselves from notifier chains.
1726 This is a relatively cheap check but if you care about maximum
1729 config DEBUG_CLOSURES
1730 bool "Debug closures (bcache async widgits)"
1734 Keeps all active closures in a linked list and provides a debugfs
1735 interface to list them, which makes it possible to see asynchronous
1736 operations that get stuck.
1738 config DEBUG_MAPLE_TREE
1739 bool "Debug maple trees"
1740 depends on DEBUG_KERNEL
1742 Enable maple tree debugging information and extra validations.
1748 source "kernel/rcu/Kconfig.debug"
1750 config DEBUG_WQ_FORCE_RR_CPU
1751 bool "Force round-robin CPU selection for unbound work items"
1752 depends on DEBUG_KERNEL
1755 Workqueue used to implicitly guarantee that work items queued
1756 without explicit CPU specified are put on the local CPU. This
1757 guarantee is no longer true and while local CPU is still
1758 preferred work items may be put on foreign CPUs. Kernel
1759 parameter "workqueue.debug_force_rr_cpu" is added to force
1760 round-robin CPU selection to flush out usages which depend on the
1761 now broken guarantee. This config option enables the debug
1762 feature by default. When enabled, memory and cache locality will
1765 config CPU_HOTPLUG_STATE_CONTROL
1766 bool "Enable CPU hotplug state control"
1767 depends on DEBUG_KERNEL
1768 depends on HOTPLUG_CPU
1771 Allows to write steps between "offline" and "online" to the CPUs
1772 sysfs target file so states can be stepped granular. This is a debug
1773 option for now as the hotplug machinery cannot be stopped and
1774 restarted at arbitrary points yet.
1776 Say N if your are unsure.
1779 bool "Latency measuring infrastructure"
1780 depends on DEBUG_KERNEL
1781 depends on STACKTRACE_SUPPORT
1783 depends on FRAME_POINTER || MIPS || PPC || S390 || MICROBLAZE || ARM || ARC || X86
1789 Enable this option if you want to use the LatencyTOP tool
1790 to find out which userspace is blocking on what kernel operations.
1792 config DEBUG_CGROUP_REF
1793 bool "Disable inlining of cgroup css reference count functions"
1794 depends on DEBUG_KERNEL
1799 Force cgroup css reference count functions to not be inlined so
1800 that they can be kprobed for debugging.
1802 source "kernel/trace/Kconfig"
1804 config PROVIDE_OHCI1394_DMA_INIT
1805 bool "Remote debugging over FireWire early on boot"
1806 depends on PCI && X86
1808 If you want to debug problems which hang or crash the kernel early
1809 on boot and the crashing machine has a FireWire port, you can use
1810 this feature to remotely access the memory of the crashed machine
1811 over FireWire. This employs remote DMA as part of the OHCI1394
1812 specification which is now the standard for FireWire controllers.
1814 With remote DMA, you can monitor the printk buffer remotely using
1815 firescope and access all memory below 4GB using fireproxy from gdb.
1816 Even controlling a kernel debugger is possible using remote DMA.
1820 If ohci1394_dma=early is used as boot parameter, it will initialize
1821 all OHCI1394 controllers which are found in the PCI config space.
1823 As all changes to the FireWire bus such as enabling and disabling
1824 devices cause a bus reset and thereby disable remote DMA for all
1825 devices, be sure to have the cable plugged and FireWire enabled on
1826 the debugging host before booting the debug target for debugging.
1828 This code (~1k) is freed after boot. By then, the firewire stack
1829 in charge of the OHCI-1394 controllers should be used instead.
1831 See Documentation/core-api/debugging-via-ohci1394.rst for more information.
1833 source "samples/Kconfig"
1835 config ARCH_HAS_DEVMEM_IS_ALLOWED
1838 config STRICT_DEVMEM
1839 bool "Filter access to /dev/mem"
1840 depends on MMU && DEVMEM
1841 depends on ARCH_HAS_DEVMEM_IS_ALLOWED || GENERIC_LIB_DEVMEM_IS_ALLOWED
1842 default y if PPC || X86 || ARM64
1844 If this option is disabled, you allow userspace (root) access to all
1845 of memory, including kernel and userspace memory. Accidental
1846 access to this is obviously disastrous, but specific access can
1847 be used by people debugging the kernel. Note that with PAT support
1848 enabled, even in this case there are restrictions on /dev/mem
1849 use due to the cache aliasing requirements.
1851 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem
1852 file only allows userspace access to PCI space and the BIOS code and
1853 data regions. This is sufficient for dosemu and X and all common
1858 config IO_STRICT_DEVMEM
1859 bool "Filter I/O access to /dev/mem"
1860 depends on STRICT_DEVMEM
1862 If this option is disabled, you allow userspace (root) access to all
1863 io-memory regardless of whether a driver is actively using that
1864 range. Accidental access to this is obviously disastrous, but
1865 specific access can be used by people debugging kernel drivers.
1867 If this option is switched on, the /dev/mem file only allows
1868 userspace access to *idle* io-memory ranges (see /proc/iomem) This
1869 may break traditional users of /dev/mem (dosemu, legacy X, etc...)
1870 if the driver using a given range cannot be disabled.
1874 menu "$(SRCARCH) Debugging"
1876 source "arch/$(SRCARCH)/Kconfig.debug"
1880 menu "Kernel Testing and Coverage"
1882 source "lib/kunit/Kconfig"
1884 config NOTIFIER_ERROR_INJECTION
1885 tristate "Notifier error injection"
1886 depends on DEBUG_KERNEL
1889 This option provides the ability to inject artificial errors to
1890 specified notifier chain callbacks. It is useful to test the error
1891 handling of notifier call chain failures.
1895 config PM_NOTIFIER_ERROR_INJECT
1896 tristate "PM notifier error injection module"
1897 depends on PM && NOTIFIER_ERROR_INJECTION
1898 default m if PM_DEBUG
1900 This option provides the ability to inject artificial errors to
1901 PM notifier chain callbacks. It is controlled through debugfs
1902 interface /sys/kernel/debug/notifier-error-inject/pm
1904 If the notifier call chain should be failed with some events
1905 notified, write the error code to "actions/<notifier event>/error".
1907 Example: Inject PM suspend error (-12 = -ENOMEM)
1909 # cd /sys/kernel/debug/notifier-error-inject/pm/
1910 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1911 # echo mem > /sys/power/state
1912 bash: echo: write error: Cannot allocate memory
1914 To compile this code as a module, choose M here: the module will
1915 be called pm-notifier-error-inject.
1919 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1920 tristate "OF reconfig notifier error injection module"
1921 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1923 This option provides the ability to inject artificial errors to
1924 OF reconfig notifier chain callbacks. It is controlled
1925 through debugfs interface under
1926 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1928 If the notifier call chain should be failed with some events
1929 notified, write the error code to "actions/<notifier event>/error".
1931 To compile this code as a module, choose M here: the module will
1932 be called of-reconfig-notifier-error-inject.
1936 config NETDEV_NOTIFIER_ERROR_INJECT
1937 tristate "Netdev notifier error injection module"
1938 depends on NET && NOTIFIER_ERROR_INJECTION
1940 This option provides the ability to inject artificial errors to
1941 netdevice notifier chain callbacks. It is controlled through debugfs
1942 interface /sys/kernel/debug/notifier-error-inject/netdev
1944 If the notifier call chain should be failed with some events
1945 notified, write the error code to "actions/<notifier event>/error".
1947 Example: Inject netdevice mtu change error (-22 = -EINVAL)
1949 # cd /sys/kernel/debug/notifier-error-inject/netdev
1950 # echo -22 > actions/NETDEV_CHANGEMTU/error
1951 # ip link set eth0 mtu 1024
1952 RTNETLINK answers: Invalid argument
1954 To compile this code as a module, choose M here: the module will
1955 be called netdev-notifier-error-inject.
1959 config FUNCTION_ERROR_INJECTION
1960 bool "Fault-injections of functions"
1961 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES
1963 Add fault injections into various functions that are annotated with
1964 ALLOW_ERROR_INJECTION() in the kernel. BPF may also modify the return
1965 value of these functions. This is useful to test error paths of code.
1969 config FAULT_INJECTION
1970 bool "Fault-injection framework"
1971 depends on DEBUG_KERNEL
1973 Provide fault-injection framework.
1974 For more details, see Documentation/fault-injection/.
1977 bool "Fault-injection capability for kmalloc"
1978 depends on FAULT_INJECTION
1980 Provide fault-injection capability for kmalloc.
1982 config FAIL_PAGE_ALLOC
1983 bool "Fault-injection capability for alloc_pages()"
1984 depends on FAULT_INJECTION
1986 Provide fault-injection capability for alloc_pages().
1988 config FAULT_INJECTION_USERCOPY
1989 bool "Fault injection capability for usercopy functions"
1990 depends on FAULT_INJECTION
1992 Provides fault-injection capability to inject failures
1993 in usercopy functions (copy_from_user(), get_user(), ...).
1995 config FAIL_MAKE_REQUEST
1996 bool "Fault-injection capability for disk IO"
1997 depends on FAULT_INJECTION && BLOCK
1999 Provide fault-injection capability for disk IO.
2001 config FAIL_IO_TIMEOUT
2002 bool "Fault-injection capability for faking disk interrupts"
2003 depends on FAULT_INJECTION && BLOCK
2005 Provide fault-injection capability on end IO handling. This
2006 will make the block layer "forget" an interrupt as configured,
2007 thus exercising the error handling.
2009 Only works with drivers that use the generic timeout handling,
2010 for others it won't do anything.
2013 bool "Fault-injection capability for futexes"
2015 depends on FAULT_INJECTION && FUTEX
2017 Provide fault-injection capability for futexes.
2019 config FAULT_INJECTION_DEBUG_FS
2020 bool "Debugfs entries for fault-injection capabilities"
2021 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
2023 Enable configuration of fault-injection capabilities via debugfs.
2025 config FAIL_FUNCTION
2026 bool "Fault-injection capability for functions"
2027 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION
2029 Provide function-based fault-injection capability.
2030 This will allow you to override a specific function with a return
2031 with given return value. As a result, function caller will see
2032 an error value and have to handle it. This is useful to test the
2033 error handling in various subsystems.
2035 config FAIL_MMC_REQUEST
2036 bool "Fault-injection capability for MMC IO"
2037 depends on FAULT_INJECTION_DEBUG_FS && MMC
2039 Provide fault-injection capability for MMC IO.
2040 This will make the mmc core return data errors. This is
2041 useful to test the error handling in the mmc block device
2042 and to test how the mmc host driver handles retries from
2046 bool "Fault-injection capability for SunRPC"
2047 depends on FAULT_INJECTION_DEBUG_FS && SUNRPC_DEBUG
2049 Provide fault-injection capability for SunRPC and
2052 config FAULT_INJECTION_CONFIGFS
2053 bool "Configfs interface for fault-injection capabilities"
2054 depends on FAULT_INJECTION
2057 This option allows configfs-based drivers to dynamically configure
2058 fault-injection via configfs. Each parameter for driver-specific
2059 fault-injection can be made visible as a configfs attribute in a
2063 config FAULT_INJECTION_STACKTRACE_FILTER
2064 bool "stacktrace filter for fault-injection capabilities"
2065 depends on FAULT_INJECTION
2066 depends on (FAULT_INJECTION_DEBUG_FS || FAULT_INJECTION_CONFIGFS) && STACKTRACE_SUPPORT
2068 depends on FRAME_POINTER || MIPS || PPC || S390 || MICROBLAZE || ARM || ARC || X86
2070 Provide stacktrace filter for fault-injection capabilities
2072 config ARCH_HAS_KCOV
2075 An architecture should select this when it can successfully
2076 build and run with CONFIG_KCOV. This typically requires
2077 disabling instrumentation for some early boot code.
2079 config CC_HAS_SANCOV_TRACE_PC
2080 def_bool $(cc-option,-fsanitize-coverage=trace-pc)
2084 bool "Code coverage for fuzzing"
2085 depends on ARCH_HAS_KCOV
2086 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS
2087 depends on !ARCH_WANTS_NO_INSTR || HAVE_NOINSTR_HACK || \
2088 GCC_VERSION >= 120000 || CC_IS_CLANG
2090 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC
2091 select OBJTOOL if HAVE_NOINSTR_HACK
2093 KCOV exposes kernel code coverage information in a form suitable
2094 for coverage-guided fuzzing (randomized testing).
2096 For more details, see Documentation/dev-tools/kcov.rst.
2098 config KCOV_ENABLE_COMPARISONS
2099 bool "Enable comparison operands collection by KCOV"
2101 depends on $(cc-option,-fsanitize-coverage=trace-cmp)
2103 KCOV also exposes operands of every comparison in the instrumented
2104 code along with operand sizes and PCs of the comparison instructions.
2105 These operands can be used by fuzzing engines to improve the quality
2106 of fuzzing coverage.
2108 config KCOV_INSTRUMENT_ALL
2109 bool "Instrument all code by default"
2113 If you are doing generic system call fuzzing (like e.g. syzkaller),
2114 then you will want to instrument the whole kernel and you should
2115 say y here. If you are doing more targeted fuzzing (like e.g.
2116 filesystem fuzzing with AFL) then you will want to enable coverage
2117 for more specific subsets of files, and should say n here.
2119 config KCOV_IRQ_AREA_SIZE
2120 hex "Size of interrupt coverage collection area in words"
2124 KCOV uses preallocated per-cpu areas to collect coverage from
2125 soft interrupts. This specifies the size of those areas in the
2126 number of unsigned long words.
2128 menuconfig RUNTIME_TESTING_MENU
2129 bool "Runtime Testing"
2132 if RUNTIME_TESTING_MENU
2135 tristate "Dhrystone benchmark test"
2137 Enable this to include the Dhrystone 2.1 benchmark. This test
2138 calculates the number of Dhrystones per second, and the number of
2139 DMIPS (Dhrystone MIPS) obtained when the Dhrystone score is divided
2140 by 1757 (the number of Dhrystones per second obtained on the VAX
2141 11/780, nominally a 1 MIPS machine).
2143 To run the benchmark, it needs to be enabled explicitly, either from
2144 the kernel command line (when built-in), or from userspace (when
2145 built-in or modular).
2147 Run once during kernel boot:
2151 Set number of iterations from kernel command line:
2153 test_dhry.iterations=<n>
2155 Set number of iterations from userspace:
2157 echo <n> > /sys/module/test_dhry/parameters/iterations
2159 Trigger manual run from userspace:
2161 echo y > /sys/module/test_dhry/parameters/run
2163 If the number of iterations is <= 0, the test will devise a suitable
2164 number of iterations (test runs for at least 2s) automatically.
2165 This process takes ca. 4s.
2170 tristate "Linux Kernel Dump Test Tool Module"
2173 This module enables testing of the different dumping mechanisms by
2174 inducing system failures at predefined crash points.
2175 If you don't need it: say N
2176 Choose M here to compile this code as a module. The module will be
2179 Documentation on how to use the module can be found in
2180 Documentation/fault-injection/provoke-crashes.rst
2182 config CPUMASK_KUNIT_TEST
2183 tristate "KUnit test for cpumask" if !KUNIT_ALL_TESTS
2185 default KUNIT_ALL_TESTS
2187 Enable to turn on cpumask tests, running at boot or module load time.
2189 For more information on KUnit and unit tests in general, please refer
2190 to the KUnit documentation in Documentation/dev-tools/kunit/.
2194 config TEST_LIST_SORT
2195 tristate "Linked list sorting test" if !KUNIT_ALL_TESTS
2197 default KUNIT_ALL_TESTS
2199 Enable this to turn on 'list_sort()' function test. This test is
2200 executed only once during system boot (so affects only boot time),
2201 or at module load time.
2205 config TEST_MIN_HEAP
2206 tristate "Min heap test"
2207 depends on DEBUG_KERNEL || m
2209 Enable this to turn on min heap function tests. This test is
2210 executed only once during system boot (so affects only boot time),
2211 or at module load time.
2216 tristate "Array-based sort test" if !KUNIT_ALL_TESTS
2218 default KUNIT_ALL_TESTS
2220 This option enables the self-test function of 'sort()' at boot,
2221 or at module load time.
2226 tristate "64bit/32bit division and modulo test"
2227 depends on DEBUG_KERNEL || m
2229 Enable this to turn on 'do_div()' function test. This test is
2230 executed only once during system boot (so affects only boot time),
2231 or at module load time.
2235 config TEST_IOV_ITER
2236 tristate "Test iov_iter operation" if !KUNIT_ALL_TESTS
2239 default KUNIT_ALL_TESTS
2241 Enable this to turn on testing of the operation of the I/O iterator
2242 (iov_iter). This test is executed only once during system boot (so
2243 affects only boot time), or at module load time.
2247 config KPROBES_SANITY_TEST
2248 tristate "Kprobes sanity tests" if !KUNIT_ALL_TESTS
2249 depends on DEBUG_KERNEL
2252 select STACKTRACE if ARCH_CORRECT_STACKTRACE_ON_KRETPROBE
2253 default KUNIT_ALL_TESTS
2255 This option provides for testing basic kprobes functionality on
2256 boot. Samples of kprobe and kretprobe are inserted and
2257 verified for functionality.
2259 Say N if you are unsure.
2261 config FPROBE_SANITY_TEST
2262 bool "Self test for fprobe"
2263 depends on DEBUG_KERNEL
2267 This option will enable testing the fprobe when the system boot.
2268 A series of tests are made to verify that the fprobe is functioning
2271 Say N if you are unsure.
2273 config BACKTRACE_SELF_TEST
2274 tristate "Self test for the backtrace code"
2275 depends on DEBUG_KERNEL
2277 This option provides a kernel module that can be used to test
2278 the kernel stack backtrace code. This option is not useful
2279 for distributions or general kernels, but only for kernel
2280 developers working on architecture code.
2282 Note that if you want to also test saved backtraces, you will
2283 have to enable STACKTRACE as well.
2285 Say N if you are unsure.
2287 config TEST_REF_TRACKER
2288 tristate "Self test for reference tracker"
2289 depends on DEBUG_KERNEL && STACKTRACE_SUPPORT
2292 This option provides a kernel module performing tests
2293 using reference tracker infrastructure.
2295 Say N if you are unsure.
2298 tristate "Red-Black tree test"
2299 depends on DEBUG_KERNEL
2301 A benchmark measuring the performance of the rbtree library.
2302 Also includes rbtree invariant checks.
2304 config REED_SOLOMON_TEST
2305 tristate "Reed-Solomon library test"
2306 depends on DEBUG_KERNEL || m
2308 select REED_SOLOMON_ENC16
2309 select REED_SOLOMON_DEC16
2311 This option enables the self-test function of rslib at boot,
2312 or at module load time.
2316 config INTERVAL_TREE_TEST
2317 tristate "Interval tree test"
2318 depends on DEBUG_KERNEL
2319 select INTERVAL_TREE
2321 A benchmark measuring the performance of the interval tree library
2324 tristate "Per cpu operations test"
2325 depends on m && DEBUG_KERNEL
2327 Enable this option to build test module which validates per-cpu
2332 config ATOMIC64_SELFTEST
2333 tristate "Perform an atomic64_t self-test"
2335 Enable this option to test the atomic64_t functions at boot or
2336 at module load time.
2340 config ASYNC_RAID6_TEST
2341 tristate "Self test for hardware accelerated raid6 recovery"
2342 depends on ASYNC_RAID6_RECOV
2345 This is a one-shot self test that permutes through the
2346 recovery of all the possible two disk failure scenarios for a
2347 N-disk array. Recovery is performed with the asynchronous
2348 raid6 recovery routines, and will optionally use an offload
2349 engine if one is available.
2354 tristate "Test functions located in the hexdump module at runtime"
2356 config STRING_KUNIT_TEST
2357 tristate "KUnit test string functions at runtime" if !KUNIT_ALL_TESTS
2359 default KUNIT_ALL_TESTS
2361 config STRING_HELPERS_KUNIT_TEST
2362 tristate "KUnit test string helpers at runtime" if !KUNIT_ALL_TESTS
2364 default KUNIT_ALL_TESTS
2367 tristate "Test kstrto*() family of functions at runtime"
2370 tristate "Test printf() family of functions at runtime"
2373 tristate "Test scanf() family of functions at runtime"
2376 tristate "Test bitmap_*() family of functions at runtime"
2378 Enable this option to test the bitmap functions at boot.
2383 tristate "Test functions located in the uuid module at runtime"
2386 tristate "Test the XArray code at runtime"
2388 config TEST_MAPLE_TREE
2389 tristate "Test the Maple Tree code at runtime or module load"
2391 Enable this option to test the maple tree code functions at boot, or
2392 when the module is loaded. Enable "Debug Maple Trees" will enable
2393 more verbose output on failures.
2397 config TEST_RHASHTABLE
2398 tristate "Perform selftest on resizable hash table"
2400 Enable this option to test the rhashtable functions at boot.
2405 tristate "Perform selftest on IDA functions"
2408 tristate "Perform selftest on priority array manager"
2411 Enable this option to test priority array manager on boot
2416 config TEST_IRQ_TIMINGS
2417 bool "IRQ timings selftest"
2418 depends on IRQ_TIMINGS
2420 Enable this option to test the irq timings code on boot.
2425 tristate "Test module loading with 'hello world' module"
2428 This builds the "test_module" module that emits "Hello, world"
2429 on printk when loaded. It is designed to be used for basic
2430 evaluation of the module loading subsystem (for example when
2431 validating module verification). It lacks any extra dependencies,
2432 and will not normally be loaded by the system unless explicitly
2438 tristate "Test module for compilation of bitops operations"
2441 This builds the "test_bitops" module that is much like the
2442 TEST_LKM module except that it does a basic exercise of the
2443 set/clear_bit macros and get_count_order/long to make sure there are
2444 no compiler warnings from C=1 sparse checker or -Wextra
2445 compilations. It has no dependencies and doesn't run or load unless
2446 explicitly requested by name. for example: modprobe test_bitops.
2451 tristate "Test module for stress/performance analysis of vmalloc allocator"
2456 This builds the "test_vmalloc" module that should be used for
2457 stress and performance analysis. So, any new change for vmalloc
2458 subsystem can be evaluated from performance and stability point
2463 config TEST_USER_COPY
2464 tristate "Test user/kernel boundary protections"
2467 This builds the "test_user_copy" module that runs sanity checks
2468 on the copy_to/from_user infrastructure, making sure basic
2469 user/kernel boundary testing is working. If it fails to load,
2470 a regression has been detected in the user/kernel memory boundary
2476 tristate "Test BPF filter functionality"
2479 This builds the "test_bpf" module that runs various test vectors
2480 against the BPF interpreter or BPF JIT compiler depending on the
2481 current setting. This is in particular useful for BPF JIT compiler
2482 development, but also to run regression tests against changes in
2483 the interpreter code. It also enables test stubs for eBPF maps and
2484 verifier used by user space verifier testsuite.
2488 config TEST_BLACKHOLE_DEV
2489 tristate "Test blackhole netdev functionality"
2492 This builds the "test_blackhole_dev" module that validates the
2493 data path through this blackhole netdev.
2497 config FIND_BIT_BENCHMARK
2498 tristate "Test find_bit functions"
2500 This builds the "test_find_bit" module that measure find_*_bit()
2501 functions performance.
2505 config TEST_FIRMWARE
2506 tristate "Test firmware loading via userspace interface"
2507 depends on FW_LOADER
2509 This builds the "test_firmware" module that creates a userspace
2510 interface for testing firmware loading. This can be used to
2511 control the triggering of firmware loading without needing an
2512 actual firmware-using device. The contents can be rechecked by
2518 tristate "sysctl test driver"
2519 depends on PROC_SYSCTL
2521 This builds the "test_sysctl" module. This driver enables to test the
2522 proc sysctl interfaces available to drivers safely without affecting
2523 production knobs which might alter system functionality.
2527 config BITFIELD_KUNIT
2528 tristate "KUnit test bitfield functions at runtime" if !KUNIT_ALL_TESTS
2530 default KUNIT_ALL_TESTS
2532 Enable this option to test the bitfield functions at boot.
2534 KUnit tests run during boot and output the results to the debug log
2535 in TAP format (http://testanything.org/). Only useful for kernel devs
2536 running the KUnit test harness, and not intended for inclusion into a
2539 For more information on KUnit and unit tests in general please refer
2540 to the KUnit documentation in Documentation/dev-tools/kunit/.
2544 config CHECKSUM_KUNIT
2545 tristate "KUnit test checksum functions at runtime" if !KUNIT_ALL_TESTS
2547 default KUNIT_ALL_TESTS
2549 Enable this option to test the checksum functions at boot.
2551 KUnit tests run during boot and output the results to the debug log
2552 in TAP format (http://testanything.org/). Only useful for kernel devs
2553 running the KUnit test harness, and not intended for inclusion into a
2556 For more information on KUnit and unit tests in general please refer
2557 to the KUnit documentation in Documentation/dev-tools/kunit/.
2561 config HASH_KUNIT_TEST
2562 tristate "KUnit Test for integer hash functions" if !KUNIT_ALL_TESTS
2564 default KUNIT_ALL_TESTS
2566 Enable this option to test the kernel's string (<linux/stringhash.h>), and
2567 integer (<linux/hash.h>) hash functions on boot.
2569 KUnit tests run during boot and output the results to the debug log
2570 in TAP format (https://testanything.org/). Only useful for kernel devs
2571 running the KUnit test harness, and not intended for inclusion into a
2574 For more information on KUnit and unit tests in general please refer
2575 to the KUnit documentation in Documentation/dev-tools/kunit/.
2577 This is intended to help people writing architecture-specific
2578 optimized versions. If unsure, say N.
2580 config RESOURCE_KUNIT_TEST
2581 tristate "KUnit test for resource API" if !KUNIT_ALL_TESTS
2583 default KUNIT_ALL_TESTS
2585 This builds the resource API unit test.
2586 Tests the logic of API provided by resource.c and ioport.h.
2587 For more information on KUnit and unit tests in general please refer
2588 to the KUnit documentation in Documentation/dev-tools/kunit/.
2592 config SYSCTL_KUNIT_TEST
2593 tristate "KUnit test for sysctl" if !KUNIT_ALL_TESTS
2595 default KUNIT_ALL_TESTS
2597 This builds the proc sysctl unit test, which runs on boot.
2598 Tests the API contract and implementation correctness of sysctl.
2599 For more information on KUnit and unit tests in general please refer
2600 to the KUnit documentation in Documentation/dev-tools/kunit/.
2604 config LIST_KUNIT_TEST
2605 tristate "KUnit Test for Kernel Linked-list structures" if !KUNIT_ALL_TESTS
2607 default KUNIT_ALL_TESTS
2609 This builds the linked list KUnit test suite.
2610 It tests that the API and basic functionality of the list_head type
2611 and associated macros.
2613 KUnit tests run during boot and output the results to the debug log
2614 in TAP format (https://testanything.org/). Only useful for kernel devs
2615 running the KUnit test harness, and not intended for inclusion into a
2618 For more information on KUnit and unit tests in general please refer
2619 to the KUnit documentation in Documentation/dev-tools/kunit/.
2623 config HASHTABLE_KUNIT_TEST
2624 tristate "KUnit Test for Kernel Hashtable structures" if !KUNIT_ALL_TESTS
2626 default KUNIT_ALL_TESTS
2628 This builds the hashtable KUnit test suite.
2629 It tests the basic functionality of the API defined in
2630 include/linux/hashtable.h. For more information on KUnit and
2631 unit tests in general please refer to the KUnit documentation
2632 in Documentation/dev-tools/kunit/.
2636 config LINEAR_RANGES_TEST
2637 tristate "KUnit test for linear_ranges"
2639 select LINEAR_RANGES
2641 This builds the linear_ranges unit test, which runs on boot.
2642 Tests the linear_ranges logic correctness.
2643 For more information on KUnit and unit tests in general please refer
2644 to the KUnit documentation in Documentation/dev-tools/kunit/.
2648 config CMDLINE_KUNIT_TEST
2649 tristate "KUnit test for cmdline API" if !KUNIT_ALL_TESTS
2651 default KUNIT_ALL_TESTS
2653 This builds the cmdline API unit test.
2654 Tests the logic of API provided by cmdline.c.
2655 For more information on KUnit and unit tests in general please refer
2656 to the KUnit documentation in Documentation/dev-tools/kunit/.
2661 tristate "KUnit test for bits.h" if !KUNIT_ALL_TESTS
2663 default KUNIT_ALL_TESTS
2665 This builds the bits unit test.
2666 Tests the logic of macros defined in bits.h.
2667 For more information on KUnit and unit tests in general please refer
2668 to the KUnit documentation in Documentation/dev-tools/kunit/.
2672 config SLUB_KUNIT_TEST
2673 tristate "KUnit test for SLUB cache error detection" if !KUNIT_ALL_TESTS
2674 depends on SLUB_DEBUG && KUNIT
2675 default KUNIT_ALL_TESTS
2677 This builds SLUB allocator unit test.
2678 Tests SLUB cache debugging functionality.
2679 For more information on KUnit and unit tests in general please refer
2680 to the KUnit documentation in Documentation/dev-tools/kunit/.
2684 config RATIONAL_KUNIT_TEST
2685 tristate "KUnit test for rational.c" if !KUNIT_ALL_TESTS
2686 depends on KUNIT && RATIONAL
2687 default KUNIT_ALL_TESTS
2689 This builds the rational math unit test.
2690 For more information on KUnit and unit tests in general please refer
2691 to the KUnit documentation in Documentation/dev-tools/kunit/.
2695 config MEMCPY_KUNIT_TEST
2696 tristate "Test memcpy(), memmove(), and memset() functions at runtime" if !KUNIT_ALL_TESTS
2698 default KUNIT_ALL_TESTS
2700 Builds unit tests for memcpy(), memmove(), and memset() functions.
2701 For more information on KUnit and unit tests in general please refer
2702 to the KUnit documentation in Documentation/dev-tools/kunit/.
2706 config IS_SIGNED_TYPE_KUNIT_TEST
2707 tristate "Test is_signed_type() macro" if !KUNIT_ALL_TESTS
2709 default KUNIT_ALL_TESTS
2711 Builds unit tests for the is_signed_type() macro.
2713 For more information on KUnit and unit tests in general please refer
2714 to the KUnit documentation in Documentation/dev-tools/kunit/.
2718 config OVERFLOW_KUNIT_TEST
2719 tristate "Test check_*_overflow() functions at runtime" if !KUNIT_ALL_TESTS
2721 default KUNIT_ALL_TESTS
2723 Builds unit tests for the check_*_overflow(), size_*(), allocation, and
2726 For more information on KUnit and unit tests in general please refer
2727 to the KUnit documentation in Documentation/dev-tools/kunit/.
2731 config STACKINIT_KUNIT_TEST
2732 tristate "Test level of stack variable initialization" if !KUNIT_ALL_TESTS
2734 default KUNIT_ALL_TESTS
2736 Test if the kernel is zero-initializing stack variables and
2737 padding. Coverage is controlled by compiler flags,
2738 CONFIG_INIT_STACK_ALL_PATTERN, CONFIG_INIT_STACK_ALL_ZERO,
2739 CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF,
2740 or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL.
2742 config FORTIFY_KUNIT_TEST
2743 tristate "Test fortified str*() and mem*() function internals at runtime" if !KUNIT_ALL_TESTS
2745 default KUNIT_ALL_TESTS
2747 Builds unit tests for checking internals of FORTIFY_SOURCE as used
2748 by the str*() and mem*() family of functions. For testing runtime
2749 traps of FORTIFY_SOURCE, see LKDTM's "FORTIFY_*" tests.
2751 config HW_BREAKPOINT_KUNIT_TEST
2752 bool "Test hw_breakpoint constraints accounting" if !KUNIT_ALL_TESTS
2753 depends on HAVE_HW_BREAKPOINT
2755 default KUNIT_ALL_TESTS
2757 Tests for hw_breakpoint constraints accounting.
2761 config STRCAT_KUNIT_TEST
2762 tristate "Test strcat() family of functions at runtime" if !KUNIT_ALL_TESTS
2764 default KUNIT_ALL_TESTS
2766 config STRSCPY_KUNIT_TEST
2767 tristate "Test strscpy*() family of functions at runtime" if !KUNIT_ALL_TESTS
2769 default KUNIT_ALL_TESTS
2771 config SIPHASH_KUNIT_TEST
2772 tristate "Perform selftest on siphash functions" if !KUNIT_ALL_TESTS
2774 default KUNIT_ALL_TESTS
2776 Enable this option to test the kernel's siphash (<linux/siphash.h>) hash
2777 functions on boot (or module load).
2779 This is intended to help people writing architecture-specific
2780 optimized versions. If unsure, say N.
2783 tristate "udelay test driver"
2785 This builds the "udelay_test" module that helps to make sure
2786 that udelay() is working properly.
2790 config TEST_STATIC_KEYS
2791 tristate "Test static keys"
2794 Test the static key interfaces.
2798 config TEST_DYNAMIC_DEBUG
2799 tristate "Test DYNAMIC_DEBUG"
2800 depends on DYNAMIC_DEBUG
2802 This module registers a tracer callback to count enabled
2803 pr_debugs in a 'do_debugging' function, then alters their
2804 enablements, calls the function, and compares counts.
2809 tristate "kmod stress tester"
2811 depends on NETDEVICES && NET_CORE && INET # for TUN
2813 depends on PAGE_SIZE_LESS_THAN_256KB # for BTRFS
2819 Test the kernel's module loading mechanism: kmod. kmod implements
2820 support to load modules using the Linux kernel's usermode helper.
2821 This test provides a series of tests against kmod.
2823 Although technically you can either build test_kmod as a module or
2824 into the kernel we disallow building it into the kernel since
2825 it stress tests request_module() and this will very likely cause
2826 some issues by taking over precious threads available from other
2827 module load requests, ultimately this could be fatal.
2831 tools/testing/selftests/kmod/kmod.sh --help
2835 config TEST_DEBUG_VIRTUAL
2836 tristate "Test CONFIG_DEBUG_VIRTUAL feature"
2837 depends on DEBUG_VIRTUAL
2839 Test the kernel's ability to detect incorrect calls to
2840 virt_to_phys() done against the non-linear part of the
2841 kernel's virtual address map.
2845 config TEST_MEMCAT_P
2846 tristate "Test memcat_p() helper function"
2848 Test the memcat_p() helper for correctly merging two
2849 pointer arrays together.
2854 tristate "Perform selftest on object aggreration manager"
2858 Enable this option to test object aggregation manager on boot
2862 tristate "Test heap/page initialization"
2864 Test if the kernel is zero-initializing heap and page allocations.
2865 This can be useful to test init_on_alloc and init_on_free features.
2870 tristate "Test HMM (Heterogeneous Memory Management)"
2871 depends on TRANSPARENT_HUGEPAGE
2872 depends on DEVICE_PRIVATE
2876 This is a pseudo device driver solely for testing HMM.
2877 Say M here if you want to build the HMM test module.
2878 Doing so will allow you to run tools/testing/selftest/vm/hmm-tests.
2882 config TEST_FREE_PAGES
2883 tristate "Test freeing pages"
2885 Test that a memory leak does not occur due to a race between
2886 freeing a block of pages and a speculative page reference.
2887 Loading this module is safe if your kernel has the bug fixed.
2888 If the bug is not fixed, it will leak gigabytes of memory and
2889 probably OOM your system.
2892 tristate "Test floating point operations in kernel space"
2893 depends on X86 && !KCOV_INSTRUMENT_ALL
2895 Enable this option to add /sys/kernel/debug/selftest_helpers/test_fpu
2896 which will trigger a sequence of floating point operations. This is used
2897 for self-testing floating point control register setting in
2902 config TEST_CLOCKSOURCE_WATCHDOG
2903 tristate "Test clocksource watchdog in kernel space"
2904 depends on CLOCKSOURCE_WATCHDOG
2906 Enable this option to create a kernel module that will trigger
2907 a test of the clocksource watchdog. This module may be loaded
2908 via modprobe or insmod in which case it will run upon being
2909 loaded, or it may be built in, in which case it will run
2915 tristate "Test module for correctness and stress of objpool"
2917 depends on m && DEBUG_KERNEL
2919 This builds the "test_objpool" module that should be used for
2920 correctness verification and concurrent testings of objects
2921 allocation and reclamation.
2925 endif # RUNTIME_TESTING_MENU
2927 config ARCH_USE_MEMTEST
2930 An architecture should select this when it uses early_memtest()
2931 during boot process.
2935 depends on ARCH_USE_MEMTEST
2937 This option adds a kernel parameter 'memtest', which allows memtest
2938 to be set and executed.
2939 memtest=0, mean disabled; -- default
2940 memtest=1, mean do 1 test pattern;
2942 memtest=17, mean do 17 test patterns.
2943 If you are unsure how to answer this question, answer N.
2947 config HYPERV_TESTING
2948 bool "Microsoft Hyper-V driver testing"
2950 depends on HYPERV && DEBUG_FS
2952 Select this option to enable Hyper-V vmbus testing.
2954 endmenu # "Kernel Testing and Coverage"
2958 config RUST_DEBUG_ASSERTIONS
2959 bool "Debug assertions"
2962 Enables rustc's `-Cdebug-assertions` codegen option.
2964 This flag lets you turn `cfg(debug_assertions)` conditional
2965 compilation on or off. This can be used to enable extra debugging
2966 code in development but not in production. For example, it controls
2967 the behavior of the standard library's `debug_assert!` macro.
2969 Note that this will apply to all Rust code, including `core`.
2973 config RUST_OVERFLOW_CHECKS
2974 bool "Overflow checks"
2978 Enables rustc's `-Coverflow-checks` codegen option.
2980 This flag allows you to control the behavior of runtime integer
2981 overflow. When overflow-checks are enabled, a Rust panic will occur
2984 Note that this will apply to all Rust code, including `core`.
2988 config RUST_BUILD_ASSERT_ALLOW
2989 bool "Allow unoptimized build-time assertions"
2992 Controls how are `build_error!` and `build_assert!` handled during build.
2994 If calls to them exist in the binary, it may indicate a violated invariant
2995 or that the optimizer failed to verify the invariant during compilation.
2997 This should not happen, thus by default the build is aborted. However,
2998 as an escape hatch, you can choose Y here to ignore them during build
2999 and let the check be carried at runtime (with `panic!` being called if
3004 config RUST_KERNEL_DOCTESTS
3005 bool "Doctests for the `kernel` crate" if !KUNIT_ALL_TESTS
3006 depends on RUST && KUNIT=y
3007 default KUNIT_ALL_TESTS
3009 This builds the documentation tests of the `kernel` crate
3012 For more information on KUnit and unit tests in general,
3013 please refer to the KUnit documentation in Documentation/dev-tools/kunit/.
3019 endmenu # Kernel hacking