3 menu "printk and dmesg options"
6 bool "Show timing information on printks"
9 Selecting this option causes time stamps of the printk()
10 messages to be added to the output of the syslog() system
11 call and at the console.
13 The timestamp is always recorded internally, and exported
14 to /dev/kmsg. This flag just specifies if the timestamp should
15 be included, not that the timestamp is recorded.
17 The behavior is also controlled by the kernel command line
18 parameter printk.time=1. See Documentation/admin-guide/kernel-parameters.rst
20 config CONSOLE_LOGLEVEL_DEFAULT
21 int "Default console loglevel (1-15)"
25 Default loglevel to determine what will be printed on the console.
27 Setting a default here is equivalent to passing in loglevel=<x> in
28 the kernel bootargs. loglevel=<x> continues to override whatever
29 value is specified here as well.
31 Note: This does not affect the log level of un-prefixed printk()
32 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT
35 config CONSOLE_LOGLEVEL_QUIET
36 int "quiet console loglevel (1-15)"
40 loglevel to use when "quiet" is passed on the kernel commandline.
42 When "quiet" is passed on the kernel commandline this loglevel
43 will be used as the loglevel. IOW passing "quiet" will be the
44 equivalent of passing "loglevel=<CONSOLE_LOGLEVEL_QUIET>"
46 config MESSAGE_LOGLEVEL_DEFAULT
47 int "Default message log level (1-7)"
51 Default log level for printk statements with no specified priority.
53 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
54 that are auditing their logs closely may want to set it to a lower
57 Note: This does not affect what message level gets printed on the console
58 by default. To change that, use loglevel=<x> in the kernel bootargs,
59 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value.
61 config BOOT_PRINTK_DELAY
62 bool "Delay each boot printk message by N milliseconds"
63 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
65 This build option allows you to read kernel boot messages
66 by inserting a short delay after each one. The delay is
67 specified in milliseconds on the kernel command line,
70 It is likely that you would also need to use "lpj=M" to preset
71 the "loops per jiffie" value.
72 See a previous boot log for the "lpj" value to use for your
73 system, and then set "lpj=M" before setting "boot_delay=N".
74 NOTE: Using this option may adversely affect SMP systems.
75 I.e., processors other than the first one may not boot up.
76 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
77 what it believes to be lockup conditions.
80 bool "Enable dynamic printk() support"
86 Compiles debug level messages into the kernel, which would not
87 otherwise be available at runtime. These messages can then be
88 enabled/disabled based on various levels of scope - per source file,
89 function, module, format string, and line number. This mechanism
90 implicitly compiles in all pr_debug() and dev_dbg() calls, which
91 enlarges the kernel text size by about 2%.
93 If a source file is compiled with DEBUG flag set, any
94 pr_debug() calls in it are enabled by default, but can be
95 disabled at runtime as below. Note that DEBUG flag is
96 turned on by many CONFIG_*DEBUG* options.
100 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
101 which is contained in the 'debugfs' filesystem. Thus, the debugfs
102 filesystem must first be mounted before making use of this feature.
103 We refer the control file as: <debugfs>/dynamic_debug/control. This
104 file contains a list of the debug statements that can be enabled. The
105 format for each line of the file is:
107 filename:lineno [module]function flags format
109 filename : source file of the debug statement
110 lineno : line number of the debug statement
111 module : module that contains the debug statement
112 function : function that contains the debug statement
113 flags : '=p' means the line is turned 'on' for printing
114 format : the format used for the debug statement
118 nullarbor:~ # cat <debugfs>/dynamic_debug/control
119 # filename:lineno [module]function flags format
120 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012"
121 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012"
122 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012"
126 // enable the message at line 1603 of file svcsock.c
127 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
128 <debugfs>/dynamic_debug/control
130 // enable all the messages in file svcsock.c
131 nullarbor:~ # echo -n 'file svcsock.c +p' >
132 <debugfs>/dynamic_debug/control
134 // enable all the messages in the NFS server module
135 nullarbor:~ # echo -n 'module nfsd +p' >
136 <debugfs>/dynamic_debug/control
138 // enable all 12 messages in the function svc_process()
139 nullarbor:~ # echo -n 'func svc_process +p' >
140 <debugfs>/dynamic_debug/control
142 // disable all 12 messages in the function svc_process()
143 nullarbor:~ # echo -n 'func svc_process -p' >
144 <debugfs>/dynamic_debug/control
146 See Documentation/admin-guide/dynamic-debug-howto.rst for additional
149 endmenu # "printk and dmesg options"
151 menu "Compile-time checks and compiler options"
154 bool "Compile the kernel with debug info"
155 depends on DEBUG_KERNEL && !COMPILE_TEST
157 If you say Y here the resulting kernel image will include
158 debugging info resulting in a larger kernel image.
159 This adds debug symbols to the kernel and modules (gcc -g), and
160 is needed if you intend to use kernel crashdump or binary object
161 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
162 Say Y here only if you plan to debug the kernel.
166 config DEBUG_INFO_REDUCED
167 bool "Reduce debugging information"
168 depends on DEBUG_INFO
170 If you say Y here gcc is instructed to generate less debugging
171 information for structure types. This means that tools that
172 need full debugging information (like kgdb or systemtap) won't
173 be happy. But if you merely need debugging information to
174 resolve line numbers there is no loss. Advantage is that
175 build directory object sizes shrink dramatically over a full
176 DEBUG_INFO build and compile times are reduced too.
177 Only works with newer gcc versions.
179 config DEBUG_INFO_SPLIT
180 bool "Produce split debuginfo in .dwo files"
181 depends on DEBUG_INFO
183 Generate debug info into separate .dwo files. This significantly
184 reduces the build directory size for builds with DEBUG_INFO,
185 because it stores the information only once on disk in .dwo
186 files instead of multiple times in object files and executables.
187 In addition the debug information is also compressed.
189 Requires recent gcc (4.7+) and recent gdb/binutils.
190 Any tool that packages or reads debug information would need
191 to know about the .dwo files and include them.
192 Incompatible with older versions of ccache.
194 config DEBUG_INFO_DWARF4
195 bool "Generate dwarf4 debuginfo"
196 depends on DEBUG_INFO
198 Generate dwarf4 debug info. This requires recent versions
199 of gcc and gdb. It makes the debug information larger.
200 But it significantly improves the success of resolving
201 variables in gdb on optimized code.
204 bool "Provide GDB scripts for kernel debugging"
205 depends on DEBUG_INFO
207 This creates the required links to GDB helper scripts in the
208 build directory. If you load vmlinux into gdb, the helper
209 scripts will be automatically imported by gdb as well, and
210 additional functions are available to analyze a Linux kernel
211 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst
214 config ENABLE_MUST_CHECK
215 bool "Enable __must_check logic"
218 Enable the __must_check logic in the kernel build. Disable this to
219 suppress the "warning: ignoring return value of 'foo', declared with
220 attribute warn_unused_result" messages.
223 int "Warn for stack frames larger than (needs gcc 4.4)"
225 default 3072 if KASAN_EXTRA
226 default 2048 if GCC_PLUGIN_LATENT_ENTROPY
227 default 1280 if (!64BIT && PARISC)
228 default 1024 if (!64BIT && !PARISC)
229 default 2048 if 64BIT
231 Tell gcc to warn at build time for stack frames larger than this.
232 Setting this too low will cause a lot of warnings.
233 Setting it to 0 disables the warning.
236 config STRIP_ASM_SYMS
237 bool "Strip assembler-generated symbols during link"
240 Strip internal assembler-generated symbols during a link (symbols
241 that look like '.Lxxx') so they don't pollute the output of
242 get_wchan() and suchlike.
245 bool "Generate readable assembler code"
246 depends on DEBUG_KERNEL
248 Disable some compiler optimizations that tend to generate human unreadable
249 assembler output. This may make the kernel slightly slower, but it helps
250 to keep kernel developers who have to stare a lot at assembler listings
253 config UNUSED_SYMBOLS
254 bool "Enable unused/obsolete exported symbols"
257 Unused but exported symbols make the kernel needlessly bigger. For
258 that reason most of these unused exports will soon be removed. This
259 option is provided temporarily to provide a transition period in case
260 some external kernel module needs one of these symbols anyway. If you
261 encounter such a case in your module, consider if you are actually
262 using the right API. (rationale: since nobody in the kernel is using
263 this in a module, there is a pretty good chance it's actually the
264 wrong interface to use). If you really need the symbol, please send a
265 mail to the linux kernel mailing list mentioning the symbol and why
266 you really need it, and what the merge plan to the mainline kernel for
270 bool "Track page owner"
271 depends on DEBUG_KERNEL && STACKTRACE_SUPPORT
275 select PAGE_EXTENSION
277 This keeps track of what call chain is the owner of a page, may
278 help to find bare alloc_page(s) leaks. Even if you include this
279 feature on your build, it is disabled in default. You should pass
280 "page_owner=on" to boot parameter in order to enable it. Eats
281 a fair amount of memory if enabled. See tools/vm/page_owner_sort.c
282 for user-space helper.
287 bool "Debug Filesystem"
289 debugfs is a virtual file system that kernel developers use to put
290 debugging files into. Enable this option to be able to read and
291 write to these files.
293 For detailed documentation on the debugfs API, see
294 Documentation/filesystems/.
299 bool "Run 'make headers_check' when building vmlinux"
302 This option will extract the user-visible kernel headers whenever
303 building the kernel, and will run basic sanity checks on them to
304 ensure that exported files do not attempt to include files which
305 were not exported, etc.
307 If you're making modifications to header files which are
308 relevant for userspace, say 'Y', and check the headers
309 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
310 your build tree), to make sure they're suitable.
312 config DEBUG_SECTION_MISMATCH
313 bool "Enable full Section mismatch analysis"
315 The section mismatch analysis checks if there are illegal
316 references from one section to another section.
317 During linktime or runtime, some sections are dropped;
318 any use of code/data previously in these sections would
319 most likely result in an oops.
320 In the code, functions and variables are annotated with
321 __init,, etc. (see the full list in include/linux/init.h),
322 which results in the code/data being placed in specific sections.
323 The section mismatch analysis is always performed after a full
324 kernel build, and enabling this option causes the following
325 additional steps to occur:
326 - Add the option -fno-inline-functions-called-once to gcc commands.
327 When inlining a function annotated with __init in a non-init
328 function, we would lose the section information and thus
329 the analysis would not catch the illegal reference.
330 This option tells gcc to inline less (but it does result in
332 - Run the section mismatch analysis for each module/built-in.a file.
333 When we run the section mismatch analysis on vmlinux.o, we
334 lose valuable information about where the mismatch was
336 Running the analysis for each module/built-in.a file
337 tells where the mismatch happens much closer to the
338 source. The drawback is that the same mismatch is
339 reported at least twice.
340 - Enable verbose reporting from modpost in order to help resolve
341 the section mismatches that are reported.
343 config SECTION_MISMATCH_WARN_ONLY
344 bool "Make section mismatch errors non-fatal"
347 If you say N here, the build process will fail if there are any
348 section mismatch, instead of just throwing warnings.
353 # Select this config option from the architecture Kconfig, if it
354 # is preferred to always offer frame pointers as a config
355 # option on the architecture (regardless of KERNEL_DEBUG):
357 config ARCH_WANT_FRAME_POINTERS
361 bool "Compile the kernel with frame pointers"
362 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS
363 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
365 If you say Y here the resulting kernel image will be slightly
366 larger and slower, but it gives very useful debugging information
367 in case of kernel bugs. (precise oopses/stacktraces/warnings)
369 config STACK_VALIDATION
370 bool "Compile-time stack metadata validation"
371 depends on HAVE_STACK_VALIDATION
374 Add compile-time checks to validate stack metadata, including frame
375 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure
376 that runtime stack traces are more reliable.
378 This is also a prerequisite for generation of ORC unwind data, which
379 is needed for CONFIG_UNWINDER_ORC.
381 For more information, see
382 tools/objtool/Documentation/stack-validation.txt.
384 config DEBUG_FORCE_WEAK_PER_CPU
385 bool "Force weak per-cpu definitions"
386 depends on DEBUG_KERNEL
388 s390 and alpha require percpu variables in modules to be
389 defined weak to work around addressing range issue which
390 puts the following two restrictions on percpu variable
393 1. percpu symbols must be unique whether static or not
394 2. percpu variables can't be defined inside a function
396 To ensure that generic code follows the above rules, this
397 option forces all percpu variables to be defined as weak.
399 endmenu # "Compiler options"
402 bool "Magic SysRq key"
405 If you say Y here, you will have some control over the system even
406 if the system crashes for example during kernel debugging (e.g., you
407 will be able to flush the buffer cache to disk, reboot the system
408 immediately or dump some status information). This is accomplished
409 by pressing various keys while holding SysRq (Alt+PrintScreen). It
410 also works on a serial console (on PC hardware at least), if you
411 send a BREAK and then within 5 seconds a command keypress. The
412 keys are documented in <file:Documentation/admin-guide/sysrq.rst>.
413 Don't say Y unless you really know what this hack does.
415 config MAGIC_SYSRQ_DEFAULT_ENABLE
416 hex "Enable magic SysRq key functions by default"
417 depends on MAGIC_SYSRQ
420 Specifies which SysRq key functions are enabled by default.
421 This may be set to 1 or 0 to enable or disable them all, or
422 to a bitmask as described in Documentation/admin-guide/sysrq.rst.
424 config MAGIC_SYSRQ_SERIAL
425 bool "Enable magic SysRq key over serial"
426 depends on MAGIC_SYSRQ
429 Many embedded boards have a disconnected TTL level serial which can
430 generate some garbage that can lead to spurious false sysrq detects.
431 This option allows you to decide whether you want to enable the
435 bool "Kernel debugging"
437 Say Y here if you are developing drivers or trying to debug and
438 identify kernel problems.
440 menu "Memory Debugging"
442 source "mm/Kconfig.debug"
445 bool "Debug object operations"
446 depends on DEBUG_KERNEL
448 If you say Y here, additional code will be inserted into the
449 kernel to track the life time of various objects and validate
450 the operations on those objects.
452 config DEBUG_OBJECTS_SELFTEST
453 bool "Debug objects selftest"
454 depends on DEBUG_OBJECTS
456 This enables the selftest of the object debug code.
458 config DEBUG_OBJECTS_FREE
459 bool "Debug objects in freed memory"
460 depends on DEBUG_OBJECTS
462 This enables checks whether a k/v free operation frees an area
463 which contains an object which has not been deactivated
464 properly. This can make kmalloc/kfree-intensive workloads
467 config DEBUG_OBJECTS_TIMERS
468 bool "Debug timer objects"
469 depends on DEBUG_OBJECTS
471 If you say Y here, additional code will be inserted into the
472 timer routines to track the life time of timer objects and
473 validate the timer operations.
475 config DEBUG_OBJECTS_WORK
476 bool "Debug work objects"
477 depends on DEBUG_OBJECTS
479 If you say Y here, additional code will be inserted into the
480 work queue routines to track the life time of work objects and
481 validate the work operations.
483 config DEBUG_OBJECTS_RCU_HEAD
484 bool "Debug RCU callbacks objects"
485 depends on DEBUG_OBJECTS
487 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
489 config DEBUG_OBJECTS_PERCPU_COUNTER
490 bool "Debug percpu counter objects"
491 depends on DEBUG_OBJECTS
493 If you say Y here, additional code will be inserted into the
494 percpu counter routines to track the life time of percpu counter
495 objects and validate the percpu counter operations.
497 config DEBUG_OBJECTS_ENABLE_DEFAULT
498 int "debug_objects bootup default value (0-1)"
501 depends on DEBUG_OBJECTS
503 Debug objects boot parameter default value
506 bool "Debug slab memory allocations"
507 depends on DEBUG_KERNEL && SLAB
509 Say Y here to have the kernel do limited verification on memory
510 allocation as well as poisoning memory on free to catch use of freed
511 memory. This can make kmalloc/kfree-intensive workloads much slower.
513 config DEBUG_SLAB_LEAK
514 bool "Memory leak debugging"
515 depends on DEBUG_SLAB
518 bool "SLUB debugging on by default"
519 depends on SLUB && SLUB_DEBUG
522 Boot with debugging on by default. SLUB boots by default with
523 the runtime debug capabilities switched off. Enabling this is
524 equivalent to specifying the "slub_debug" parameter on boot.
525 There is no support for more fine grained debug control like
526 possible with slub_debug=xxx. SLUB debugging may be switched
527 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
532 bool "Enable SLUB performance statistics"
533 depends on SLUB && SYSFS
535 SLUB statistics are useful to debug SLUBs allocation behavior in
536 order find ways to optimize the allocator. This should never be
537 enabled for production use since keeping statistics slows down
538 the allocator by a few percentage points. The slabinfo command
539 supports the determination of the most active slabs to figure
540 out which slabs are relevant to a particular load.
541 Try running: slabinfo -DA
543 config HAVE_DEBUG_KMEMLEAK
546 config DEBUG_KMEMLEAK
547 bool "Kernel memory leak detector"
548 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK
550 select STACKTRACE if STACKTRACE_SUPPORT
554 Say Y here if you want to enable the memory leak
555 detector. The memory allocation/freeing is traced in a way
556 similar to the Boehm's conservative garbage collector, the
557 difference being that the orphan objects are not freed but
558 only shown in /sys/kernel/debug/kmemleak. Enabling this
559 feature will introduce an overhead to memory
560 allocations. See Documentation/dev-tools/kmemleak.rst for more
563 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
564 of finding leaks due to the slab objects poisoning.
566 In order to access the kmemleak file, debugfs needs to be
567 mounted (usually at /sys/kernel/debug).
569 config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
570 int "Maximum kmemleak early log entries"
571 depends on DEBUG_KMEMLEAK
575 Kmemleak must track all the memory allocations to avoid
576 reporting false positives. Since memory may be allocated or
577 freed before kmemleak is initialised, an early log buffer is
578 used to store these actions. If kmemleak reports "early log
579 buffer exceeded", please increase this value.
581 config DEBUG_KMEMLEAK_TEST
582 tristate "Simple test for the kernel memory leak detector"
583 depends on DEBUG_KMEMLEAK && m
585 This option enables a module that explicitly leaks memory.
589 config DEBUG_KMEMLEAK_DEFAULT_OFF
590 bool "Default kmemleak to off"
591 depends on DEBUG_KMEMLEAK
593 Say Y here to disable kmemleak by default. It can then be enabled
594 on the command line via kmemleak=on.
596 config DEBUG_KMEMLEAK_AUTO_SCAN
597 bool "Enable kmemleak auto scan thread on boot up"
599 depends on DEBUG_KMEMLEAK
601 Depending on the cpu, kmemleak scan may be cpu intensive and can
602 stall user tasks at times. This option enables/disables automatic
603 kmemleak scan at boot up.
605 Say N here to disable kmemleak auto scan thread to stop automatic
606 scanning. Disabling this option disables automatic reporting of
611 config DEBUG_STACK_USAGE
612 bool "Stack utilization instrumentation"
613 depends on DEBUG_KERNEL && !IA64
615 Enables the display of the minimum amount of free stack which each
616 task has ever had available in the sysrq-T and sysrq-P debug output.
618 This option will slow down process creation somewhat.
622 depends on DEBUG_KERNEL
624 Enable this to turn on extended checks in the virtual-memory system
625 that may impact performance.
629 config DEBUG_VM_VMACACHE
630 bool "Debug VMA caching"
633 Enable this to turn on VMA caching debug information. Doing so
634 can cause significant overhead, so only enable it in non-production
640 bool "Debug VM red-black trees"
643 Enable VM red-black tree debugging information and extra validations.
647 config DEBUG_VM_PGFLAGS
648 bool "Debug page-flags operations"
651 Enables extra validation on page flags operations.
655 config ARCH_HAS_DEBUG_VIRTUAL
659 bool "Debug VM translations"
660 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL
662 Enable some costly sanity checks in virtual to page code. This can
663 catch mistakes with virt_to_page() and friends.
667 config DEBUG_NOMMU_REGIONS
668 bool "Debug the global anon/private NOMMU mapping region tree"
669 depends on DEBUG_KERNEL && !MMU
671 This option causes the global tree of anonymous and private mapping
672 regions to be regularly checked for invalid topology.
674 config DEBUG_MEMORY_INIT
675 bool "Debug memory initialisation" if EXPERT
678 Enable this for additional checks during memory initialisation.
679 The sanity checks verify aspects of the VM such as the memory model
680 and other information provided by the architecture. Verbose
681 information will be printed at KERN_DEBUG loglevel depending
682 on the mminit_loglevel= command-line option.
686 config MEMORY_NOTIFIER_ERROR_INJECT
687 tristate "Memory hotplug notifier error injection module"
688 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION
690 This option provides the ability to inject artificial errors to
691 memory hotplug notifier chain callbacks. It is controlled through
692 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
694 If the notifier call chain should be failed with some events
695 notified, write the error code to "actions/<notifier event>/error".
697 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
699 # cd /sys/kernel/debug/notifier-error-inject/memory
700 # echo -12 > actions/MEM_GOING_OFFLINE/error
701 # echo offline > /sys/devices/system/memory/memoryXXX/state
702 bash: echo: write error: Cannot allocate memory
704 To compile this code as a module, choose M here: the module will
705 be called memory-notifier-error-inject.
709 config DEBUG_PER_CPU_MAPS
710 bool "Debug access to per_cpu maps"
711 depends on DEBUG_KERNEL
714 Say Y to verify that the per_cpu map being accessed has
715 been set up. This adds a fair amount of code to kernel memory
716 and decreases performance.
721 bool "Highmem debugging"
722 depends on DEBUG_KERNEL && HIGHMEM
724 This option enables additional error checking for high memory
725 systems. Disable for production systems.
727 config HAVE_DEBUG_STACKOVERFLOW
730 config DEBUG_STACKOVERFLOW
731 bool "Check for stack overflows"
732 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW
734 Say Y here if you want to check for overflows of kernel, IRQ
735 and exception stacks (if your architecture uses them). This
736 option will show detailed messages if free stack space drops
737 below a certain limit.
739 These kinds of bugs usually occur when call-chains in the
740 kernel get too deep, especially when interrupts are
743 Use this in cases where you see apparently random memory
744 corruption, especially if it appears in 'struct thread_info'
746 If in doubt, say "N".
748 source "lib/Kconfig.kasan"
750 endmenu # "Memory Debugging"
755 KCOV does not have any arch-specific code, but currently it is enabled
756 only for x86_64. KCOV requires testing on other archs, and most likely
757 disabling of instrumentation for some early boot code.
759 config CC_HAS_SANCOV_TRACE_PC
760 def_bool $(cc-option,-fsanitize-coverage=trace-pc)
763 bool "Code coverage for fuzzing"
764 depends on ARCH_HAS_KCOV
765 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS
767 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC
769 KCOV exposes kernel code coverage information in a form suitable
770 for coverage-guided fuzzing (randomized testing).
772 If RANDOMIZE_BASE is enabled, PC values will not be stable across
773 different machines and across reboots. If you need stable PC values,
774 disable RANDOMIZE_BASE.
776 For more details, see Documentation/dev-tools/kcov.rst.
778 config KCOV_ENABLE_COMPARISONS
779 bool "Enable comparison operands collection by KCOV"
781 depends on $(cc-option,-fsanitize-coverage=trace-cmp)
783 KCOV also exposes operands of every comparison in the instrumented
784 code along with operand sizes and PCs of the comparison instructions.
785 These operands can be used by fuzzing engines to improve the quality
788 config KCOV_INSTRUMENT_ALL
789 bool "Instrument all code by default"
793 If you are doing generic system call fuzzing (like e.g. syzkaller),
794 then you will want to instrument the whole kernel and you should
795 say y here. If you are doing more targeted fuzzing (like e.g.
796 filesystem fuzzing with AFL) then you will want to enable coverage
797 for more specific subsets of files, and should say n here.
800 bool "Debug shared IRQ handlers"
801 depends on DEBUG_KERNEL
803 Enable this to generate a spurious interrupt as soon as a shared
804 interrupt handler is registered, and just before one is deregistered.
805 Drivers ought to be able to handle interrupts coming in at those
806 points; some don't and need to be caught.
808 menu "Debug Lockups and Hangs"
810 config LOCKUP_DETECTOR
813 config SOFTLOCKUP_DETECTOR
814 bool "Detect Soft Lockups"
815 depends on DEBUG_KERNEL && !S390
816 select LOCKUP_DETECTOR
818 Say Y here to enable the kernel to act as a watchdog to detect
821 Softlockups are bugs that cause the kernel to loop in kernel
822 mode for more than 20 seconds, without giving other tasks a
823 chance to run. The current stack trace is displayed upon
824 detection and the system will stay locked up.
826 config BOOTPARAM_SOFTLOCKUP_PANIC
827 bool "Panic (Reboot) On Soft Lockups"
828 depends on SOFTLOCKUP_DETECTOR
830 Say Y here to enable the kernel to panic on "soft lockups",
831 which are bugs that cause the kernel to loop in kernel
832 mode for more than 20 seconds (configurable using the watchdog_thresh
833 sysctl), without giving other tasks a chance to run.
835 The panic can be used in combination with panic_timeout,
836 to cause the system to reboot automatically after a
837 lockup has been detected. This feature is useful for
838 high-availability systems that have uptime guarantees and
839 where a lockup must be resolved ASAP.
843 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
845 depends on SOFTLOCKUP_DETECTOR
847 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
848 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
850 config HARDLOCKUP_DETECTOR_PERF
852 select SOFTLOCKUP_DETECTOR
855 # Enables a timestamp based low pass filter to compensate for perf based
856 # hard lockup detection which runs too fast due to turbo modes.
858 config HARDLOCKUP_CHECK_TIMESTAMP
862 # arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard
863 # lockup detector rather than the perf based detector.
865 config HARDLOCKUP_DETECTOR
866 bool "Detect Hard Lockups"
867 depends on DEBUG_KERNEL && !S390
868 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_ARCH
869 select LOCKUP_DETECTOR
870 select HARDLOCKUP_DETECTOR_PERF if HAVE_HARDLOCKUP_DETECTOR_PERF
871 select HARDLOCKUP_DETECTOR_ARCH if HAVE_HARDLOCKUP_DETECTOR_ARCH
873 Say Y here to enable the kernel to act as a watchdog to detect
876 Hardlockups are bugs that cause the CPU to loop in kernel mode
877 for more than 10 seconds, without letting other interrupts have a
878 chance to run. The current stack trace is displayed upon detection
879 and the system will stay locked up.
881 config BOOTPARAM_HARDLOCKUP_PANIC
882 bool "Panic (Reboot) On Hard Lockups"
883 depends on HARDLOCKUP_DETECTOR
885 Say Y here to enable the kernel to panic on "hard lockups",
886 which are bugs that cause the kernel to loop in kernel
887 mode with interrupts disabled for more than 10 seconds (configurable
888 using the watchdog_thresh sysctl).
892 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
894 depends on HARDLOCKUP_DETECTOR
896 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
897 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
899 config DETECT_HUNG_TASK
900 bool "Detect Hung Tasks"
901 depends on DEBUG_KERNEL
902 default SOFTLOCKUP_DETECTOR
904 Say Y here to enable the kernel to detect "hung tasks",
905 which are bugs that cause the task to be stuck in
906 uninterruptible "D" state indefinitely.
908 When a hung task is detected, the kernel will print the
909 current stack trace (which you should report), but the
910 task will stay in uninterruptible state. If lockdep is
911 enabled then all held locks will also be reported. This
912 feature has negligible overhead.
914 config DEFAULT_HUNG_TASK_TIMEOUT
915 int "Default timeout for hung task detection (in seconds)"
916 depends on DETECT_HUNG_TASK
919 This option controls the default timeout (in seconds) used
920 to determine when a task has become non-responsive and should
923 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
924 sysctl or by writing a value to
925 /proc/sys/kernel/hung_task_timeout_secs.
927 A timeout of 0 disables the check. The default is two minutes.
928 Keeping the default should be fine in most cases.
930 config BOOTPARAM_HUNG_TASK_PANIC
931 bool "Panic (Reboot) On Hung Tasks"
932 depends on DETECT_HUNG_TASK
934 Say Y here to enable the kernel to panic on "hung tasks",
935 which are bugs that cause the kernel to leave a task stuck
936 in uninterruptible "D" state.
938 The panic can be used in combination with panic_timeout,
939 to cause the system to reboot automatically after a
940 hung task has been detected. This feature is useful for
941 high-availability systems that have uptime guarantees and
942 where a hung tasks must be resolved ASAP.
946 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
948 depends on DETECT_HUNG_TASK
950 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
951 default 1 if BOOTPARAM_HUNG_TASK_PANIC
954 bool "Detect Workqueue Stalls"
955 depends on DEBUG_KERNEL
957 Say Y here to enable stall detection on workqueues. If a
958 worker pool doesn't make forward progress on a pending work
959 item for over a given amount of time, 30s by default, a
960 warning message is printed along with dump of workqueue
961 state. This can be configured through kernel parameter
962 "workqueue.watchdog_thresh" and its sysfs counterpart.
964 endmenu # "Debug lockups and hangs"
969 Say Y here to enable the kernel to panic when it oopses. This
970 has the same effect as setting oops=panic on the kernel command
973 This feature is useful to ensure that the kernel does not do
974 anything erroneous after an oops which could result in data
975 corruption or other issues.
979 config PANIC_ON_OOPS_VALUE
982 default 0 if !PANIC_ON_OOPS
983 default 1 if PANIC_ON_OOPS
989 Set the timeout value (in seconds) until a reboot occurs when the
990 the kernel panics. If n = 0, then we wait forever. A timeout
991 value n > 0 will wait n seconds before rebooting, while a timeout
992 value n < 0 will reboot immediately.
995 bool "Collect scheduler debugging info"
996 depends on DEBUG_KERNEL && PROC_FS
999 If you say Y here, the /proc/sched_debug file will be provided
1000 that can help debug the scheduler. The runtime overhead of this
1008 bool "Collect scheduler statistics"
1009 depends on DEBUG_KERNEL && PROC_FS
1012 If you say Y here, additional code will be inserted into the
1013 scheduler and related routines to collect statistics about
1014 scheduler behavior and provide them in /proc/schedstat. These
1015 stats may be useful for both tuning and debugging the scheduler
1016 If you aren't debugging the scheduler or trying to tune a specific
1017 application, you can say N to avoid the very slight overhead
1020 config SCHED_STACK_END_CHECK
1021 bool "Detect stack corruption on calls to schedule()"
1022 depends on DEBUG_KERNEL
1025 This option checks for a stack overrun on calls to schedule().
1026 If the stack end location is found to be over written always panic as
1027 the content of the corrupted region can no longer be trusted.
1028 This is to ensure no erroneous behaviour occurs which could result in
1029 data corruption or a sporadic crash at a later stage once the region
1030 is examined. The runtime overhead introduced is minimal.
1032 config DEBUG_TIMEKEEPING
1033 bool "Enable extra timekeeping sanity checking"
1035 This option will enable additional timekeeping sanity checks
1036 which may be helpful when diagnosing issues where timekeeping
1037 problems are suspected.
1039 This may include checks in the timekeeping hotpaths, so this
1040 option may have a (very small) performance impact to some
1045 config DEBUG_PREEMPT
1046 bool "Debug preemptible kernel"
1047 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
1050 If you say Y here then the kernel will use a debug variant of the
1051 commonly used smp_processor_id() function and will print warnings
1052 if kernel code uses it in a preemption-unsafe way. Also, the kernel
1053 will detect preemption count underflows.
1055 menu "Lock Debugging (spinlocks, mutexes, etc...)"
1057 config LOCK_DEBUGGING_SUPPORT
1059 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1062 config PROVE_LOCKING
1063 bool "Lock debugging: prove locking correctness"
1064 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1066 select DEBUG_SPINLOCK
1067 select DEBUG_MUTEXES
1068 select DEBUG_RT_MUTEXES if RT_MUTEXES
1069 select DEBUG_RWSEMS if RWSEM_SPIN_ON_OWNER
1070 select DEBUG_WW_MUTEX_SLOWPATH
1071 select DEBUG_LOCK_ALLOC
1072 select TRACE_IRQFLAGS
1075 This feature enables the kernel to prove that all locking
1076 that occurs in the kernel runtime is mathematically
1077 correct: that under no circumstance could an arbitrary (and
1078 not yet triggered) combination of observed locking
1079 sequences (on an arbitrary number of CPUs, running an
1080 arbitrary number of tasks and interrupt contexts) cause a
1083 In short, this feature enables the kernel to report locking
1084 related deadlocks before they actually occur.
1086 The proof does not depend on how hard and complex a
1087 deadlock scenario would be to trigger: how many
1088 participant CPUs, tasks and irq-contexts would be needed
1089 for it to trigger. The proof also does not depend on
1090 timing: if a race and a resulting deadlock is possible
1091 theoretically (no matter how unlikely the race scenario
1092 is), it will be proven so and will immediately be
1093 reported by the kernel (once the event is observed that
1094 makes the deadlock theoretically possible).
1096 If a deadlock is impossible (i.e. the locking rules, as
1097 observed by the kernel, are mathematically correct), the
1098 kernel reports nothing.
1100 NOTE: this feature can also be enabled for rwlocks, mutexes
1101 and rwsems - in which case all dependencies between these
1102 different locking variants are observed and mapped too, and
1103 the proof of observed correctness is also maintained for an
1104 arbitrary combination of these separate locking variants.
1106 For more details, see Documentation/locking/lockdep-design.txt.
1109 bool "Lock usage statistics"
1110 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1112 select DEBUG_SPINLOCK
1113 select DEBUG_MUTEXES
1114 select DEBUG_RT_MUTEXES if RT_MUTEXES
1115 select DEBUG_LOCK_ALLOC
1118 This feature enables tracking lock contention points
1120 For more details, see Documentation/locking/lockstat.txt
1122 This also enables lock events required by "perf lock",
1124 If you want to use "perf lock", you also need to turn on
1125 CONFIG_EVENT_TRACING.
1127 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
1128 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
1130 config DEBUG_RT_MUTEXES
1131 bool "RT Mutex debugging, deadlock detection"
1132 depends on DEBUG_KERNEL && RT_MUTEXES
1134 This allows rt mutex semantics violations and rt mutex related
1135 deadlocks (lockups) to be detected and reported automatically.
1137 config DEBUG_SPINLOCK
1138 bool "Spinlock and rw-lock debugging: basic checks"
1139 depends on DEBUG_KERNEL
1140 select UNINLINE_SPIN_UNLOCK
1142 Say Y here and build SMP to catch missing spinlock initialization
1143 and certain other kinds of spinlock errors commonly made. This is
1144 best used in conjunction with the NMI watchdog so that spinlock
1145 deadlocks are also debuggable.
1147 config DEBUG_MUTEXES
1148 bool "Mutex debugging: basic checks"
1149 depends on DEBUG_KERNEL
1151 This feature allows mutex semantics violations to be detected and
1154 config DEBUG_WW_MUTEX_SLOWPATH
1155 bool "Wait/wound mutex debugging: Slowpath testing"
1156 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1157 select DEBUG_LOCK_ALLOC
1158 select DEBUG_SPINLOCK
1159 select DEBUG_MUTEXES
1161 This feature enables slowpath testing for w/w mutex users by
1162 injecting additional -EDEADLK wound/backoff cases. Together with
1163 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this
1164 will test all possible w/w mutex interface abuse with the
1165 exception of simply not acquiring all the required locks.
1166 Note that this feature can introduce significant overhead, so
1167 it really should not be enabled in a production or distro kernel,
1168 even a debug kernel. If you are a driver writer, enable it. If
1169 you are a distro, do not.
1172 bool "RW Semaphore debugging: basic checks"
1173 depends on DEBUG_KERNEL && RWSEM_SPIN_ON_OWNER
1175 This debugging feature allows mismatched rw semaphore locks and unlocks
1176 to be detected and reported.
1178 config DEBUG_LOCK_ALLOC
1179 bool "Lock debugging: detect incorrect freeing of live locks"
1180 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1181 select DEBUG_SPINLOCK
1182 select DEBUG_MUTEXES
1183 select DEBUG_RT_MUTEXES if RT_MUTEXES
1186 This feature will check whether any held lock (spinlock, rwlock,
1187 mutex or rwsem) is incorrectly freed by the kernel, via any of the
1188 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
1189 vfree(), etc.), whether a live lock is incorrectly reinitialized via
1190 spin_lock_init()/mutex_init()/etc., or whether there is any lock
1191 held during task exit.
1195 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1197 select FRAME_POINTER if !MIPS && !PPC && !ARM && !S390 && !MICROBLAZE && !ARC && !X86
1201 config LOCKDEP_SMALL
1204 config DEBUG_LOCKDEP
1205 bool "Lock dependency engine debugging"
1206 depends on DEBUG_KERNEL && LOCKDEP
1208 If you say Y here, the lock dependency engine will do
1209 additional runtime checks to debug itself, at the price
1210 of more runtime overhead.
1212 config DEBUG_ATOMIC_SLEEP
1213 bool "Sleep inside atomic section checking"
1214 select PREEMPT_COUNT
1215 depends on DEBUG_KERNEL
1216 depends on !ARCH_NO_PREEMPT
1218 If you say Y here, various routines which may sleep will become very
1219 noisy if they are called inside atomic sections: when a spinlock is
1220 held, inside an rcu read side critical section, inside preempt disabled
1221 sections, inside an interrupt, etc...
1223 config DEBUG_LOCKING_API_SELFTESTS
1224 bool "Locking API boot-time self-tests"
1225 depends on DEBUG_KERNEL
1227 Say Y here if you want the kernel to run a short self-test during
1228 bootup. The self-test checks whether common types of locking bugs
1229 are detected by debugging mechanisms or not. (if you disable
1230 lock debugging then those bugs wont be detected of course.)
1231 The following locking APIs are covered: spinlocks, rwlocks,
1234 config LOCK_TORTURE_TEST
1235 tristate "torture tests for locking"
1236 depends on DEBUG_KERNEL
1239 This option provides a kernel module that runs torture tests
1240 on kernel locking primitives. The kernel module may be built
1241 after the fact on the running kernel to be tested, if desired.
1243 Say Y here if you want kernel locking-primitive torture tests
1244 to be built into the kernel.
1245 Say M if you want these torture tests to build as a module.
1246 Say N if you are unsure.
1248 config WW_MUTEX_SELFTEST
1249 tristate "Wait/wound mutex selftests"
1251 This option provides a kernel module that runs tests on the
1252 on the struct ww_mutex locking API.
1254 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction
1255 with this test harness.
1257 Say M if you want these self tests to build as a module.
1258 Say N if you are unsure.
1260 endmenu # lock debugging
1262 config TRACE_IRQFLAGS
1265 Enables hooks to interrupt enabling and disabling for
1266 either tracing or lock debugging.
1269 bool "Stack backtrace support"
1270 depends on STACKTRACE_SUPPORT
1272 This option causes the kernel to create a /proc/pid/stack for
1273 every process, showing its current stack trace.
1274 It is also used by various kernel debugging features that require
1275 stack trace generation.
1277 config WARN_ALL_UNSEEDED_RANDOM
1278 bool "Warn for all uses of unseeded randomness"
1281 Some parts of the kernel contain bugs relating to their use of
1282 cryptographically secure random numbers before it's actually possible
1283 to generate those numbers securely. This setting ensures that these
1284 flaws don't go unnoticed, by enabling a message, should this ever
1285 occur. This will allow people with obscure setups to know when things
1286 are going wrong, so that they might contact developers about fixing
1289 Unfortunately, on some models of some architectures getting
1290 a fully seeded CRNG is extremely difficult, and so this can
1291 result in dmesg getting spammed for a surprisingly long
1292 time. This is really bad from a security perspective, and
1293 so architecture maintainers really need to do what they can
1294 to get the CRNG seeded sooner after the system is booted.
1295 However, since users cannot do anything actionable to
1296 address this, by default the kernel will issue only a single
1297 warning for the first use of unseeded randomness.
1299 Say Y here if you want to receive warnings for all uses of
1300 unseeded randomness. This will be of use primarily for
1301 those developers interested in improving the security of
1302 Linux kernels running on their architecture (or
1305 config DEBUG_KOBJECT
1306 bool "kobject debugging"
1307 depends on DEBUG_KERNEL
1309 If you say Y here, some extra kobject debugging messages will be sent
1312 config DEBUG_KOBJECT_RELEASE
1313 bool "kobject release debugging"
1314 depends on DEBUG_OBJECTS_TIMERS
1316 kobjects are reference counted objects. This means that their
1317 last reference count put is not predictable, and the kobject can
1318 live on past the point at which a driver decides to drop it's
1319 initial reference to the kobject gained on allocation. An
1320 example of this would be a struct device which has just been
1323 However, some buggy drivers assume that after such an operation,
1324 the memory backing the kobject can be immediately freed. This
1325 goes completely against the principles of a refcounted object.
1327 If you say Y here, the kernel will delay the release of kobjects
1328 on the last reference count to improve the visibility of this
1329 kind of kobject release bug.
1331 config HAVE_DEBUG_BUGVERBOSE
1334 config DEBUG_BUGVERBOSE
1335 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
1336 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
1339 Say Y here to make BUG() panics output the file name and line number
1340 of the BUG call as well as the EIP and oops trace. This aids
1341 debugging but costs about 70-100K of memory.
1344 bool "Debug linked list manipulation"
1345 depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION
1347 Enable this to turn on extended checks in the linked-list
1352 config DEBUG_PI_LIST
1353 bool "Debug priority linked list manipulation"
1354 depends on DEBUG_KERNEL
1356 Enable this to turn on extended checks in the priority-ordered
1357 linked-list (plist) walking routines. This checks the entire
1358 list multiple times during each manipulation.
1363 bool "Debug SG table operations"
1364 depends on DEBUG_KERNEL
1366 Enable this to turn on checks on scatter-gather tables. This can
1367 help find problems with drivers that do not properly initialize
1372 config DEBUG_NOTIFIERS
1373 bool "Debug notifier call chains"
1374 depends on DEBUG_KERNEL
1376 Enable this to turn on sanity checking for notifier call chains.
1377 This is most useful for kernel developers to make sure that
1378 modules properly unregister themselves from notifier chains.
1379 This is a relatively cheap check but if you care about maximum
1382 config DEBUG_CREDENTIALS
1383 bool "Debug credential management"
1384 depends on DEBUG_KERNEL
1386 Enable this to turn on some debug checking for credential
1387 management. The additional code keeps track of the number of
1388 pointers from task_structs to any given cred struct, and checks to
1389 see that this number never exceeds the usage count of the cred
1392 Furthermore, if SELinux is enabled, this also checks that the
1393 security pointer in the cred struct is never seen to be invalid.
1397 source "kernel/rcu/Kconfig.debug"
1399 config DEBUG_WQ_FORCE_RR_CPU
1400 bool "Force round-robin CPU selection for unbound work items"
1401 depends on DEBUG_KERNEL
1404 Workqueue used to implicitly guarantee that work items queued
1405 without explicit CPU specified are put on the local CPU. This
1406 guarantee is no longer true and while local CPU is still
1407 preferred work items may be put on foreign CPUs. Kernel
1408 parameter "workqueue.debug_force_rr_cpu" is added to force
1409 round-robin CPU selection to flush out usages which depend on the
1410 now broken guarantee. This config option enables the debug
1411 feature by default. When enabled, memory and cache locality will
1414 config DEBUG_BLOCK_EXT_DEVT
1415 bool "Force extended block device numbers and spread them"
1416 depends on DEBUG_KERNEL
1420 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1421 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1422 YOU ARE DOING. Distros, please enable this and fix whatever
1425 Conventionally, block device numbers are allocated from
1426 predetermined contiguous area. However, extended block area
1427 may introduce non-contiguous block device numbers. This
1428 option forces most block device numbers to be allocated from
1429 the extended space and spreads them to discover kernel or
1430 userland code paths which assume predetermined contiguous
1431 device number allocation.
1433 Note that turning on this debug option shuffles all the
1434 device numbers for all IDE and SCSI devices including libata
1435 ones, so root partition specified using device number
1436 directly (via rdev or root=MAJ:MIN) won't work anymore.
1437 Textual device names (root=/dev/sdXn) will continue to work.
1439 Say N if you are unsure.
1441 config CPU_HOTPLUG_STATE_CONTROL
1442 bool "Enable CPU hotplug state control"
1443 depends on DEBUG_KERNEL
1444 depends on HOTPLUG_CPU
1447 Allows to write steps between "offline" and "online" to the CPUs
1448 sysfs target file so states can be stepped granular. This is a debug
1449 option for now as the hotplug machinery cannot be stopped and
1450 restarted at arbitrary points yet.
1452 Say N if your are unsure.
1454 config NOTIFIER_ERROR_INJECTION
1455 tristate "Notifier error injection"
1456 depends on DEBUG_KERNEL
1459 This option provides the ability to inject artificial errors to
1460 specified notifier chain callbacks. It is useful to test the error
1461 handling of notifier call chain failures.
1465 config PM_NOTIFIER_ERROR_INJECT
1466 tristate "PM notifier error injection module"
1467 depends on PM && NOTIFIER_ERROR_INJECTION
1468 default m if PM_DEBUG
1470 This option provides the ability to inject artificial errors to
1471 PM notifier chain callbacks. It is controlled through debugfs
1472 interface /sys/kernel/debug/notifier-error-inject/pm
1474 If the notifier call chain should be failed with some events
1475 notified, write the error code to "actions/<notifier event>/error".
1477 Example: Inject PM suspend error (-12 = -ENOMEM)
1479 # cd /sys/kernel/debug/notifier-error-inject/pm/
1480 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1481 # echo mem > /sys/power/state
1482 bash: echo: write error: Cannot allocate memory
1484 To compile this code as a module, choose M here: the module will
1485 be called pm-notifier-error-inject.
1489 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1490 tristate "OF reconfig notifier error injection module"
1491 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1493 This option provides the ability to inject artificial errors to
1494 OF reconfig notifier chain callbacks. It is controlled
1495 through debugfs interface under
1496 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1498 If the notifier call chain should be failed with some events
1499 notified, write the error code to "actions/<notifier event>/error".
1501 To compile this code as a module, choose M here: the module will
1502 be called of-reconfig-notifier-error-inject.
1506 config NETDEV_NOTIFIER_ERROR_INJECT
1507 tristate "Netdev notifier error injection module"
1508 depends on NET && NOTIFIER_ERROR_INJECTION
1510 This option provides the ability to inject artificial errors to
1511 netdevice notifier chain callbacks. It is controlled through debugfs
1512 interface /sys/kernel/debug/notifier-error-inject/netdev
1514 If the notifier call chain should be failed with some events
1515 notified, write the error code to "actions/<notifier event>/error".
1517 Example: Inject netdevice mtu change error (-22 = -EINVAL)
1519 # cd /sys/kernel/debug/notifier-error-inject/netdev
1520 # echo -22 > actions/NETDEV_CHANGEMTU/error
1521 # ip link set eth0 mtu 1024
1522 RTNETLINK answers: Invalid argument
1524 To compile this code as a module, choose M here: the module will
1525 be called netdev-notifier-error-inject.
1529 config FUNCTION_ERROR_INJECTION
1531 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES
1533 config FAULT_INJECTION
1534 bool "Fault-injection framework"
1535 depends on DEBUG_KERNEL
1537 Provide fault-injection framework.
1538 For more details, see Documentation/fault-injection/.
1541 bool "Fault-injection capability for kmalloc"
1542 depends on FAULT_INJECTION
1543 depends on SLAB || SLUB
1545 Provide fault-injection capability for kmalloc.
1547 config FAIL_PAGE_ALLOC
1548 bool "Fault-injection capabilitiy for alloc_pages()"
1549 depends on FAULT_INJECTION
1551 Provide fault-injection capability for alloc_pages().
1553 config FAIL_MAKE_REQUEST
1554 bool "Fault-injection capability for disk IO"
1555 depends on FAULT_INJECTION && BLOCK
1557 Provide fault-injection capability for disk IO.
1559 config FAIL_IO_TIMEOUT
1560 bool "Fault-injection capability for faking disk interrupts"
1561 depends on FAULT_INJECTION && BLOCK
1563 Provide fault-injection capability on end IO handling. This
1564 will make the block layer "forget" an interrupt as configured,
1565 thus exercising the error handling.
1567 Only works with drivers that use the generic timeout handling,
1568 for others it wont do anything.
1571 bool "Fault-injection capability for futexes"
1573 depends on FAULT_INJECTION && FUTEX
1575 Provide fault-injection capability for futexes.
1577 config FAULT_INJECTION_DEBUG_FS
1578 bool "Debugfs entries for fault-injection capabilities"
1579 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1581 Enable configuration of fault-injection capabilities via debugfs.
1583 config FAIL_FUNCTION
1584 bool "Fault-injection capability for functions"
1585 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION
1587 Provide function-based fault-injection capability.
1588 This will allow you to override a specific function with a return
1589 with given return value. As a result, function caller will see
1590 an error value and have to handle it. This is useful to test the
1591 error handling in various subsystems.
1593 config FAIL_MMC_REQUEST
1594 bool "Fault-injection capability for MMC IO"
1595 depends on FAULT_INJECTION_DEBUG_FS && MMC
1597 Provide fault-injection capability for MMC IO.
1598 This will make the mmc core return data errors. This is
1599 useful to test the error handling in the mmc block device
1600 and to test how the mmc host driver handles retries from
1603 config FAULT_INJECTION_STACKTRACE_FILTER
1604 bool "stacktrace filter for fault-injection capabilities"
1605 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1608 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1610 Provide stacktrace filter for fault-injection capabilities
1613 bool "Latency measuring infrastructure"
1614 depends on DEBUG_KERNEL
1615 depends on STACKTRACE_SUPPORT
1617 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1624 Enable this option if you want to use the LatencyTOP tool
1625 to find out which userspace is blocking on what kernel operations.
1627 source "kernel/trace/Kconfig"
1629 config PROVIDE_OHCI1394_DMA_INIT
1630 bool "Remote debugging over FireWire early on boot"
1631 depends on PCI && X86
1633 If you want to debug problems which hang or crash the kernel early
1634 on boot and the crashing machine has a FireWire port, you can use
1635 this feature to remotely access the memory of the crashed machine
1636 over FireWire. This employs remote DMA as part of the OHCI1394
1637 specification which is now the standard for FireWire controllers.
1639 With remote DMA, you can monitor the printk buffer remotely using
1640 firescope and access all memory below 4GB using fireproxy from gdb.
1641 Even controlling a kernel debugger is possible using remote DMA.
1645 If ohci1394_dma=early is used as boot parameter, it will initialize
1646 all OHCI1394 controllers which are found in the PCI config space.
1648 As all changes to the FireWire bus such as enabling and disabling
1649 devices cause a bus reset and thereby disable remote DMA for all
1650 devices, be sure to have the cable plugged and FireWire enabled on
1651 the debugging host before booting the debug target for debugging.
1653 This code (~1k) is freed after boot. By then, the firewire stack
1654 in charge of the OHCI-1394 controllers should be used instead.
1656 See Documentation/debugging-via-ohci1394.txt for more information.
1658 config DMA_API_DEBUG
1659 bool "Enable debugging of DMA-API usage"
1660 select NEED_DMA_MAP_STATE
1662 Enable this option to debug the use of the DMA API by device drivers.
1663 With this option you will be able to detect common bugs in device
1664 drivers like double-freeing of DMA mappings or freeing mappings that
1665 were never allocated.
1667 This also attempts to catch cases where a page owned by DMA is
1668 accessed by the cpu in a way that could cause data corruption. For
1669 example, this enables cow_user_page() to check that the source page is
1672 This option causes a performance degradation. Use only if you want to
1673 debug device drivers and dma interactions.
1677 config DMA_API_DEBUG_SG
1678 bool "Debug DMA scatter-gather usage"
1680 depends on DMA_API_DEBUG
1682 Perform extra checking that callers of dma_map_sg() have respected the
1683 appropriate segment length/boundary limits for the given device when
1684 preparing DMA scatterlists.
1686 This is particularly likely to have been overlooked in cases where the
1687 dma_map_sg() API is used for general bulk mapping of pages rather than
1688 preparing literal scatter-gather descriptors, where there is a risk of
1689 unexpected behaviour from DMA API implementations if the scatterlist
1690 is technically out-of-spec.
1694 menuconfig RUNTIME_TESTING_MENU
1695 bool "Runtime Testing"
1698 if RUNTIME_TESTING_MENU
1701 tristate "Linux Kernel Dump Test Tool Module"
1705 This module enables testing of the different dumping mechanisms by
1706 inducing system failures at predefined crash points.
1707 If you don't need it: say N
1708 Choose M here to compile this code as a module. The module will be
1711 Documentation on how to use the module can be found in
1712 Documentation/fault-injection/provoke-crashes.txt
1714 config TEST_LIST_SORT
1715 tristate "Linked list sorting test"
1716 depends on DEBUG_KERNEL || m
1718 Enable this to turn on 'list_sort()' function test. This test is
1719 executed only once during system boot (so affects only boot time),
1720 or at module load time.
1725 tristate "Array-based sort test"
1726 depends on DEBUG_KERNEL || m
1728 This option enables the self-test function of 'sort()' at boot,
1729 or at module load time.
1733 config KPROBES_SANITY_TEST
1734 bool "Kprobes sanity tests"
1735 depends on DEBUG_KERNEL
1738 This option provides for testing basic kprobes functionality on
1739 boot. Samples of kprobe and kretprobe are inserted and
1740 verified for functionality.
1742 Say N if you are unsure.
1744 config BACKTRACE_SELF_TEST
1745 tristate "Self test for the backtrace code"
1746 depends on DEBUG_KERNEL
1748 This option provides a kernel module that can be used to test
1749 the kernel stack backtrace code. This option is not useful
1750 for distributions or general kernels, but only for kernel
1751 developers working on architecture code.
1753 Note that if you want to also test saved backtraces, you will
1754 have to enable STACKTRACE as well.
1756 Say N if you are unsure.
1759 tristate "Red-Black tree test"
1760 depends on DEBUG_KERNEL
1762 A benchmark measuring the performance of the rbtree library.
1763 Also includes rbtree invariant checks.
1765 config INTERVAL_TREE_TEST
1766 tristate "Interval tree test"
1767 depends on DEBUG_KERNEL
1768 select INTERVAL_TREE
1770 A benchmark measuring the performance of the interval tree library
1773 tristate "Per cpu operations test"
1774 depends on m && DEBUG_KERNEL
1776 Enable this option to build test module which validates per-cpu
1781 config ATOMIC64_SELFTEST
1782 tristate "Perform an atomic64_t self-test"
1784 Enable this option to test the atomic64_t functions at boot or
1785 at module load time.
1789 config ASYNC_RAID6_TEST
1790 tristate "Self test for hardware accelerated raid6 recovery"
1791 depends on ASYNC_RAID6_RECOV
1794 This is a one-shot self test that permutes through the
1795 recovery of all the possible two disk failure scenarios for a
1796 N-disk array. Recovery is performed with the asynchronous
1797 raid6 recovery routines, and will optionally use an offload
1798 engine if one is available.
1803 tristate "Test functions located in the hexdump module at runtime"
1805 config TEST_STRING_HELPERS
1806 tristate "Test functions located in the string_helpers module at runtime"
1809 tristate "Test kstrto*() family of functions at runtime"
1812 tristate "Test printf() family of functions at runtime"
1815 tristate "Test bitmap_*() family of functions at runtime"
1817 Enable this option to test the bitmap functions at boot.
1821 config TEST_BITFIELD
1822 tristate "Test bitfield functions at runtime"
1824 Enable this option to test the bitfield functions at boot.
1829 tristate "Test functions located in the uuid module at runtime"
1832 tristate "Test the XArray code at runtime"
1834 config TEST_OVERFLOW
1835 tristate "Test check_*_overflow() functions at runtime"
1837 config TEST_RHASHTABLE
1838 tristate "Perform selftest on resizable hash table"
1840 Enable this option to test the rhashtable functions at boot.
1845 tristate "Perform selftest on hash functions"
1847 Enable this option to test the kernel's integer (<linux/hash.h>),
1848 string (<linux/stringhash.h>), and siphash (<linux/siphash.h>)
1849 hash functions on boot (or module load).
1851 This is intended to help people writing architecture-specific
1852 optimized versions. If unsure, say N.
1855 tristate "Perform selftest on IDA functions"
1858 tristate "Perform selftest on priority array manager"
1861 Enable this option to test priority array manager on boot
1867 tristate "Test module loading with 'hello world' module"
1870 This builds the "test_module" module that emits "Hello, world"
1871 on printk when loaded. It is designed to be used for basic
1872 evaluation of the module loading subsystem (for example when
1873 validating module verification). It lacks any extra dependencies,
1874 and will not normally be loaded by the system unless explicitly
1879 config TEST_USER_COPY
1880 tristate "Test user/kernel boundary protections"
1883 This builds the "test_user_copy" module that runs sanity checks
1884 on the copy_to/from_user infrastructure, making sure basic
1885 user/kernel boundary testing is working. If it fails to load,
1886 a regression has been detected in the user/kernel memory boundary
1892 tristate "Test BPF filter functionality"
1895 This builds the "test_bpf" module that runs various test vectors
1896 against the BPF interpreter or BPF JIT compiler depending on the
1897 current setting. This is in particular useful for BPF JIT compiler
1898 development, but also to run regression tests against changes in
1899 the interpreter code. It also enables test stubs for eBPF maps and
1900 verifier used by user space verifier testsuite.
1904 config FIND_BIT_BENCHMARK
1905 tristate "Test find_bit functions"
1907 This builds the "test_find_bit" module that measure find_*_bit()
1908 functions performance.
1912 config TEST_FIRMWARE
1913 tristate "Test firmware loading via userspace interface"
1914 depends on FW_LOADER
1916 This builds the "test_firmware" module that creates a userspace
1917 interface for testing firmware loading. This can be used to
1918 control the triggering of firmware loading without needing an
1919 actual firmware-using device. The contents can be rechecked by
1925 tristate "sysctl test driver"
1926 depends on PROC_SYSCTL
1928 This builds the "test_sysctl" module. This driver enables to test the
1929 proc sysctl interfaces available to drivers safely without affecting
1930 production knobs which might alter system functionality.
1935 tristate "udelay test driver"
1937 This builds the "udelay_test" module that helps to make sure
1938 that udelay() is working properly.
1942 config TEST_STATIC_KEYS
1943 tristate "Test static keys"
1946 Test the static key interfaces.
1951 tristate "kmod stress tester"
1953 depends on BLOCK && (64BIT || LBDAF) # for XFS, BTRFS
1954 depends on NETDEVICES && NET_CORE && INET # for TUN
1960 Test the kernel's module loading mechanism: kmod. kmod implements
1961 support to load modules using the Linux kernel's usermode helper.
1962 This test provides a series of tests against kmod.
1964 Although technically you can either build test_kmod as a module or
1965 into the kernel we disallow building it into the kernel since
1966 it stress tests request_module() and this will very likely cause
1967 some issues by taking over precious threads available from other
1968 module load requests, ultimately this could be fatal.
1972 tools/testing/selftests/kmod/kmod.sh --help
1976 config TEST_DEBUG_VIRTUAL
1977 tristate "Test CONFIG_DEBUG_VIRTUAL feature"
1978 depends on DEBUG_VIRTUAL
1980 Test the kernel's ability to detect incorrect calls to
1981 virt_to_phys() done against the non-linear part of the
1982 kernel's virtual address map.
1986 config TEST_MEMCAT_P
1987 tristate "Test memcat_p() helper function"
1989 Test the memcat_p() helper for correctly merging two
1990 pointer arrays together.
1995 tristate "Perform selftest on object aggreration manager"
1999 Enable this option to test object aggregation manager on boot
2004 config TEST_STACKINIT
2005 tristate "Test level of stack variable initialization"
2007 Test if the kernel is zero-initializing stack variables and
2008 padding. Coverage is controlled by compiler flags,
2009 CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF,
2010 or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL.
2014 endif # RUNTIME_TESTING_MENU
2019 This option adds a kernel parameter 'memtest', which allows memtest
2021 memtest=0, mean disabled; -- default
2022 memtest=1, mean do 1 test pattern;
2024 memtest=17, mean do 17 test patterns.
2025 If you are unsure how to answer this question, answer N.
2027 config BUG_ON_DATA_CORRUPTION
2028 bool "Trigger a BUG when data corruption is detected"
2031 Select this option if the kernel should BUG when it encounters
2032 data corruption in kernel memory structures when they get checked
2037 source "samples/Kconfig"
2039 source "lib/Kconfig.kgdb"
2041 source "lib/Kconfig.ubsan"
2043 config ARCH_HAS_DEVMEM_IS_ALLOWED
2046 config STRICT_DEVMEM
2047 bool "Filter access to /dev/mem"
2048 depends on MMU && DEVMEM
2049 depends on ARCH_HAS_DEVMEM_IS_ALLOWED
2050 default y if PPC || X86 || ARM64
2052 If this option is disabled, you allow userspace (root) access to all
2053 of memory, including kernel and userspace memory. Accidental
2054 access to this is obviously disastrous, but specific access can
2055 be used by people debugging the kernel. Note that with PAT support
2056 enabled, even in this case there are restrictions on /dev/mem
2057 use due to the cache aliasing requirements.
2059 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem
2060 file only allows userspace access to PCI space and the BIOS code and
2061 data regions. This is sufficient for dosemu and X and all common
2066 config IO_STRICT_DEVMEM
2067 bool "Filter I/O access to /dev/mem"
2068 depends on STRICT_DEVMEM
2070 If this option is disabled, you allow userspace (root) access to all
2071 io-memory regardless of whether a driver is actively using that
2072 range. Accidental access to this is obviously disastrous, but
2073 specific access can be used by people debugging kernel drivers.
2075 If this option is switched on, the /dev/mem file only allows
2076 userspace access to *idle* io-memory ranges (see /proc/iomem) This
2077 may break traditional users of /dev/mem (dosemu, legacy X, etc...)
2078 if the driver using a given range cannot be disabled.
2082 source "arch/$(SRCARCH)/Kconfig.debug"
2084 endmenu # Kernel hacking