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 CONSOLE_LOGLEVEL_DEFAULT
39 int "Default console loglevel (1-15)"
43 Default loglevel to determine what will be printed on the console.
45 Setting a default here is equivalent to passing in loglevel=<x> in
46 the kernel bootargs. loglevel=<x> continues to override whatever
47 value is specified here as well.
49 Note: This does not affect the log level of un-prefixed printk()
50 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT
53 config CONSOLE_LOGLEVEL_QUIET
54 int "quiet console loglevel (1-15)"
58 loglevel to use when "quiet" is passed on the kernel commandline.
60 When "quiet" is passed on the kernel commandline this loglevel
61 will be used as the loglevel. IOW passing "quiet" will be the
62 equivalent of passing "loglevel=<CONSOLE_LOGLEVEL_QUIET>"
64 config MESSAGE_LOGLEVEL_DEFAULT
65 int "Default message log level (1-7)"
69 Default log level for printk statements with no specified priority.
71 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
72 that are auditing their logs closely may want to set it to a lower
75 Note: This does not affect what message level gets printed on the console
76 by default. To change that, use loglevel=<x> in the kernel bootargs,
77 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value.
79 config BOOT_PRINTK_DELAY
80 bool "Delay each boot printk message by N milliseconds"
81 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
83 This build option allows you to read kernel boot messages
84 by inserting a short delay after each one. The delay is
85 specified in milliseconds on the kernel command line,
88 It is likely that you would also need to use "lpj=M" to preset
89 the "loops per jiffie" value.
90 See a previous boot log for the "lpj" value to use for your
91 system, and then set "lpj=M" before setting "boot_delay=N".
92 NOTE: Using this option may adversely affect SMP systems.
93 I.e., processors other than the first one may not boot up.
94 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
95 what it believes to be lockup conditions.
98 bool "Enable dynamic printk() support"
104 Compiles debug level messages into the kernel, which would not
105 otherwise be available at runtime. These messages can then be
106 enabled/disabled based on various levels of scope - per source file,
107 function, module, format string, and line number. This mechanism
108 implicitly compiles in all pr_debug() and dev_dbg() calls, which
109 enlarges the kernel text size by about 2%.
111 If a source file is compiled with DEBUG flag set, any
112 pr_debug() calls in it are enabled by default, but can be
113 disabled at runtime as below. Note that DEBUG flag is
114 turned on by many CONFIG_*DEBUG* options.
118 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
119 which is contained in the 'debugfs' filesystem. Thus, the debugfs
120 filesystem must first be mounted before making use of this feature.
121 We refer the control file as: <debugfs>/dynamic_debug/control. This
122 file contains a list of the debug statements that can be enabled. The
123 format for each line of the file is:
125 filename:lineno [module]function flags format
127 filename : source file of the debug statement
128 lineno : line number of the debug statement
129 module : module that contains the debug statement
130 function : function that contains the debug statement
131 flags : '=p' means the line is turned 'on' for printing
132 format : the format used for the debug statement
136 nullarbor:~ # cat <debugfs>/dynamic_debug/control
137 # filename:lineno [module]function flags format
138 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012"
139 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012"
140 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012"
144 // enable the message at line 1603 of file svcsock.c
145 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
146 <debugfs>/dynamic_debug/control
148 // enable all the messages in file svcsock.c
149 nullarbor:~ # echo -n 'file svcsock.c +p' >
150 <debugfs>/dynamic_debug/control
152 // enable all the messages in the NFS server module
153 nullarbor:~ # echo -n 'module nfsd +p' >
154 <debugfs>/dynamic_debug/control
156 // enable all 12 messages in the function svc_process()
157 nullarbor:~ # echo -n 'func svc_process +p' >
158 <debugfs>/dynamic_debug/control
160 // disable all 12 messages in the function svc_process()
161 nullarbor:~ # echo -n 'func svc_process -p' >
162 <debugfs>/dynamic_debug/control
164 See Documentation/admin-guide/dynamic-debug-howto.rst for additional
167 config SYMBOLIC_ERRNAME
168 bool "Support symbolic error names in printf"
171 If you say Y here, the kernel's printf implementation will
172 be able to print symbolic error names such as ENOSPC instead
173 of the number 28. It makes the kernel image slightly larger
174 (about 3KB), but can make the kernel logs easier to read.
176 config DEBUG_BUGVERBOSE
177 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
178 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
181 Say Y here to make BUG() panics output the file name and line number
182 of the BUG call as well as the EIP and oops trace. This aids
183 debugging but costs about 70-100K of memory.
185 endmenu # "printk and dmesg options"
187 menu "Compile-time checks and compiler options"
190 bool "Compile the kernel with debug info"
191 depends on DEBUG_KERNEL && !COMPILE_TEST
193 If you say Y here the resulting kernel image will include
194 debugging info resulting in a larger kernel image.
195 This adds debug symbols to the kernel and modules (gcc -g), and
196 is needed if you intend to use kernel crashdump or binary object
197 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
198 Say Y here only if you plan to debug the kernel.
202 config DEBUG_INFO_REDUCED
203 bool "Reduce debugging information"
204 depends on DEBUG_INFO
206 If you say Y here gcc is instructed to generate less debugging
207 information for structure types. This means that tools that
208 need full debugging information (like kgdb or systemtap) won't
209 be happy. But if you merely need debugging information to
210 resolve line numbers there is no loss. Advantage is that
211 build directory object sizes shrink dramatically over a full
212 DEBUG_INFO build and compile times are reduced too.
213 Only works with newer gcc versions.
215 config DEBUG_INFO_SPLIT
216 bool "Produce split debuginfo in .dwo files"
217 depends on DEBUG_INFO
218 depends on $(cc-option,-gsplit-dwarf)
220 Generate debug info into separate .dwo files. This significantly
221 reduces the build directory size for builds with DEBUG_INFO,
222 because it stores the information only once on disk in .dwo
223 files instead of multiple times in object files and executables.
224 In addition the debug information is also compressed.
226 Requires recent gcc (4.7+) and recent gdb/binutils.
227 Any tool that packages or reads debug information would need
228 to know about the .dwo files and include them.
229 Incompatible with older versions of ccache.
231 config DEBUG_INFO_DWARF4
232 bool "Generate dwarf4 debuginfo"
233 depends on DEBUG_INFO
234 depends on $(cc-option,-gdwarf-4)
236 Generate dwarf4 debug info. This requires recent versions
237 of gcc and gdb. It makes the debug information larger.
238 But it significantly improves the success of resolving
239 variables in gdb on optimized code.
241 config DEBUG_INFO_BTF
242 bool "Generate BTF typeinfo"
243 depends on DEBUG_INFO
245 Generate deduplicated BTF type information from DWARF debug info.
246 Turning this on expects presence of pahole tool, which will convert
247 DWARF type info into equivalent deduplicated BTF type info.
250 bool "Provide GDB scripts for kernel debugging"
251 depends on DEBUG_INFO
253 This creates the required links to GDB helper scripts in the
254 build directory. If you load vmlinux into gdb, the helper
255 scripts will be automatically imported by gdb as well, and
256 additional functions are available to analyze a Linux kernel
257 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst
260 config ENABLE_MUST_CHECK
261 bool "Enable __must_check logic"
264 Enable the __must_check logic in the kernel build. Disable this to
265 suppress the "warning: ignoring return value of 'foo', declared with
266 attribute warn_unused_result" messages.
269 int "Warn for stack frames larger than (needs gcc 4.4)"
271 default 2048 if GCC_PLUGIN_LATENT_ENTROPY
272 default 1280 if (!64BIT && PARISC)
273 default 1024 if (!64BIT && !PARISC)
274 default 2048 if 64BIT
276 Tell gcc to warn at build time for stack frames larger than this.
277 Setting this too low will cause a lot of warnings.
278 Setting it to 0 disables the warning.
281 config STRIP_ASM_SYMS
282 bool "Strip assembler-generated symbols during link"
285 Strip internal assembler-generated symbols during a link (symbols
286 that look like '.Lxxx') so they don't pollute the output of
287 get_wchan() and suchlike.
290 bool "Generate readable assembler code"
291 depends on DEBUG_KERNEL
293 Disable some compiler optimizations that tend to generate human unreadable
294 assembler output. This may make the kernel slightly slower, but it helps
295 to keep kernel developers who have to stare a lot at assembler listings
298 config HEADERS_INSTALL
299 bool "Install uapi headers to usr/include"
302 This option will install uapi headers (headers exported to user-space)
303 into the usr/include directory for use during the kernel build.
304 This is unneeded for building the kernel itself, but needed for some
305 user-space program samples. It is also needed by some features such
306 as uapi header sanity checks.
308 config OPTIMIZE_INLINING
311 This option determines if the kernel forces gcc to inline the functions
312 developers have marked 'inline'. Doing so takes away freedom from gcc to
313 do what it thinks is best, which is desirable for the gcc 3.x series of
314 compilers. The gcc 4.x series have a rewritten inlining algorithm and
315 enabling this option will generate a smaller kernel there. Hopefully
316 this algorithm is so good that allowing gcc 4.x and above to make the
317 decision will become the default in the future. Until then this option
318 is there to test gcc for this.
320 config DEBUG_SECTION_MISMATCH
321 bool "Enable full Section mismatch analysis"
323 The section mismatch analysis checks if there are illegal
324 references from one section to another section.
325 During linktime or runtime, some sections are dropped;
326 any use of code/data previously in these sections would
327 most likely result in an oops.
328 In the code, functions and variables are annotated with
329 __init,, etc. (see the full list in include/linux/init.h),
330 which results in the code/data being placed in specific sections.
331 The section mismatch analysis is always performed after a full
332 kernel build, and enabling this option causes the following
333 additional step to occur:
334 - Add the option -fno-inline-functions-called-once to gcc commands.
335 When inlining a function annotated with __init in a non-init
336 function, we would lose the section information and thus
337 the analysis would not catch the illegal reference.
338 This option tells gcc to inline less (but it does result in
341 config SECTION_MISMATCH_WARN_ONLY
342 bool "Make section mismatch errors non-fatal"
345 If you say N here, the build process will fail if there are any
346 section mismatch, instead of just throwing warnings.
351 # Select this config option from the architecture Kconfig, if it
352 # is preferred to always offer frame pointers as a config
353 # option on the architecture (regardless of KERNEL_DEBUG):
355 config ARCH_WANT_FRAME_POINTERS
359 bool "Compile the kernel with frame pointers"
360 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS
361 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
363 If you say Y here the resulting kernel image will be slightly
364 larger and slower, but it gives very useful debugging information
365 in case of kernel bugs. (precise oopses/stacktraces/warnings)
367 config STACK_VALIDATION
368 bool "Compile-time stack metadata validation"
369 depends on HAVE_STACK_VALIDATION
372 Add compile-time checks to validate stack metadata, including frame
373 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure
374 that runtime stack traces are more reliable.
376 This is also a prerequisite for generation of ORC unwind data, which
377 is needed for CONFIG_UNWINDER_ORC.
379 For more information, see
380 tools/objtool/Documentation/stack-validation.txt.
382 config DEBUG_FORCE_WEAK_PER_CPU
383 bool "Force weak per-cpu definitions"
384 depends on DEBUG_KERNEL
386 s390 and alpha require percpu variables in modules to be
387 defined weak to work around addressing range issue which
388 puts the following two restrictions on percpu variable
391 1. percpu symbols must be unique whether static or not
392 2. percpu variables can't be defined inside a function
394 To ensure that generic code follows the above rules, this
395 option forces all percpu variables to be defined as weak.
397 endmenu # "Compiler options"
399 menu "Generic Kernel Debugging Instruments"
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 "Debug Filesystem"
437 debugfs is a virtual file system that kernel developers use to put
438 debugging files into. Enable this option to be able to read and
439 write to these files.
441 For detailed documentation on the debugfs API, see
442 Documentation/filesystems/.
446 source "lib/Kconfig.kgdb"
448 source "lib/Kconfig.ubsan"
453 bool "Kernel debugging"
455 Say Y here if you are developing drivers or trying to debug and
456 identify kernel problems.
459 bool "Miscellaneous debug code"
461 depends on DEBUG_KERNEL
463 Say Y here if you need to enable miscellaneous debug code that should
464 be under a more specific debug option but isn't.
467 menu "Memory Debugging"
469 source "mm/Kconfig.debug"
472 bool "Debug object operations"
473 depends on DEBUG_KERNEL
475 If you say Y here, additional code will be inserted into the
476 kernel to track the life time of various objects and validate
477 the operations on those objects.
479 config DEBUG_OBJECTS_SELFTEST
480 bool "Debug objects selftest"
481 depends on DEBUG_OBJECTS
483 This enables the selftest of the object debug code.
485 config DEBUG_OBJECTS_FREE
486 bool "Debug objects in freed memory"
487 depends on DEBUG_OBJECTS
489 This enables checks whether a k/v free operation frees an area
490 which contains an object which has not been deactivated
491 properly. This can make kmalloc/kfree-intensive workloads
494 config DEBUG_OBJECTS_TIMERS
495 bool "Debug timer objects"
496 depends on DEBUG_OBJECTS
498 If you say Y here, additional code will be inserted into the
499 timer routines to track the life time of timer objects and
500 validate the timer operations.
502 config DEBUG_OBJECTS_WORK
503 bool "Debug work objects"
504 depends on DEBUG_OBJECTS
506 If you say Y here, additional code will be inserted into the
507 work queue routines to track the life time of work objects and
508 validate the work operations.
510 config DEBUG_OBJECTS_RCU_HEAD
511 bool "Debug RCU callbacks objects"
512 depends on DEBUG_OBJECTS
514 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
516 config DEBUG_OBJECTS_PERCPU_COUNTER
517 bool "Debug percpu counter objects"
518 depends on DEBUG_OBJECTS
520 If you say Y here, additional code will be inserted into the
521 percpu counter routines to track the life time of percpu counter
522 objects and validate the percpu counter operations.
524 config DEBUG_OBJECTS_ENABLE_DEFAULT
525 int "debug_objects bootup default value (0-1)"
528 depends on DEBUG_OBJECTS
530 Debug objects boot parameter default value
533 bool "Debug slab memory allocations"
534 depends on DEBUG_KERNEL && SLAB
536 Say Y here to have the kernel do limited verification on memory
537 allocation as well as poisoning memory on free to catch use of freed
538 memory. This can make kmalloc/kfree-intensive workloads much slower.
541 bool "SLUB debugging on by default"
542 depends on SLUB && SLUB_DEBUG
545 Boot with debugging on by default. SLUB boots by default with
546 the runtime debug capabilities switched off. Enabling this is
547 equivalent to specifying the "slub_debug" parameter on boot.
548 There is no support for more fine grained debug control like
549 possible with slub_debug=xxx. SLUB debugging may be switched
550 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
555 bool "Enable SLUB performance statistics"
556 depends on SLUB && SYSFS
558 SLUB statistics are useful to debug SLUBs allocation behavior in
559 order find ways to optimize the allocator. This should never be
560 enabled for production use since keeping statistics slows down
561 the allocator by a few percentage points. The slabinfo command
562 supports the determination of the most active slabs to figure
563 out which slabs are relevant to a particular load.
564 Try running: slabinfo -DA
566 config HAVE_DEBUG_KMEMLEAK
569 config DEBUG_KMEMLEAK
570 bool "Kernel memory leak detector"
571 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK
573 select STACKTRACE if STACKTRACE_SUPPORT
577 Say Y here if you want to enable the memory leak
578 detector. The memory allocation/freeing is traced in a way
579 similar to the Boehm's conservative garbage collector, the
580 difference being that the orphan objects are not freed but
581 only shown in /sys/kernel/debug/kmemleak. Enabling this
582 feature will introduce an overhead to memory
583 allocations. See Documentation/dev-tools/kmemleak.rst for more
586 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
587 of finding leaks due to the slab objects poisoning.
589 In order to access the kmemleak file, debugfs needs to be
590 mounted (usually at /sys/kernel/debug).
592 config DEBUG_KMEMLEAK_MEM_POOL_SIZE
593 int "Kmemleak memory pool size"
594 depends on DEBUG_KMEMLEAK
598 Kmemleak must track all the memory allocations to avoid
599 reporting false positives. Since memory may be allocated or
600 freed before kmemleak is fully initialised, use a static pool
601 of metadata objects to track such callbacks. After kmemleak is
602 fully initialised, this memory pool acts as an emergency one
603 if slab allocations fail.
605 config DEBUG_KMEMLEAK_TEST
606 tristate "Simple test for the kernel memory leak detector"
607 depends on DEBUG_KMEMLEAK && m
609 This option enables a module that explicitly leaks memory.
613 config DEBUG_KMEMLEAK_DEFAULT_OFF
614 bool "Default kmemleak to off"
615 depends on DEBUG_KMEMLEAK
617 Say Y here to disable kmemleak by default. It can then be enabled
618 on the command line via kmemleak=on.
620 config DEBUG_KMEMLEAK_AUTO_SCAN
621 bool "Enable kmemleak auto scan thread on boot up"
623 depends on DEBUG_KMEMLEAK
625 Depending on the cpu, kmemleak scan may be cpu intensive and can
626 stall user tasks at times. This option enables/disables automatic
627 kmemleak scan at boot up.
629 Say N here to disable kmemleak auto scan thread to stop automatic
630 scanning. Disabling this option disables automatic reporting of
635 config DEBUG_STACK_USAGE
636 bool "Stack utilization instrumentation"
637 depends on DEBUG_KERNEL && !IA64
639 Enables the display of the minimum amount of free stack which each
640 task has ever had available in the sysrq-T and sysrq-P debug output.
642 This option will slow down process creation somewhat.
644 config SCHED_STACK_END_CHECK
645 bool "Detect stack corruption on calls to schedule()"
646 depends on DEBUG_KERNEL
649 This option checks for a stack overrun on calls to schedule().
650 If the stack end location is found to be over written always panic as
651 the content of the corrupted region can no longer be trusted.
652 This is to ensure no erroneous behaviour occurs which could result in
653 data corruption or a sporadic crash at a later stage once the region
654 is examined. The runtime overhead introduced is minimal.
658 depends on DEBUG_KERNEL
660 Enable this to turn on extended checks in the virtual-memory system
661 that may impact performance.
665 config DEBUG_VM_VMACACHE
666 bool "Debug VMA caching"
669 Enable this to turn on VMA caching debug information. Doing so
670 can cause significant overhead, so only enable it in non-production
676 bool "Debug VM red-black trees"
679 Enable VM red-black tree debugging information and extra validations.
683 config DEBUG_VM_PGFLAGS
684 bool "Debug page-flags operations"
687 Enables extra validation on page flags operations.
691 config ARCH_HAS_DEBUG_VIRTUAL
695 bool "Debug VM translations"
696 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL
698 Enable some costly sanity checks in virtual to page code. This can
699 catch mistakes with virt_to_page() and friends.
703 config DEBUG_NOMMU_REGIONS
704 bool "Debug the global anon/private NOMMU mapping region tree"
705 depends on DEBUG_KERNEL && !MMU
707 This option causes the global tree of anonymous and private mapping
708 regions to be regularly checked for invalid topology.
710 config DEBUG_MEMORY_INIT
711 bool "Debug memory initialisation" if EXPERT
714 Enable this for additional checks during memory initialisation.
715 The sanity checks verify aspects of the VM such as the memory model
716 and other information provided by the architecture. Verbose
717 information will be printed at KERN_DEBUG loglevel depending
718 on the mminit_loglevel= command-line option.
722 config MEMORY_NOTIFIER_ERROR_INJECT
723 tristate "Memory hotplug notifier error injection module"
724 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION
726 This option provides the ability to inject artificial errors to
727 memory hotplug notifier chain callbacks. It is controlled through
728 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
730 If the notifier call chain should be failed with some events
731 notified, write the error code to "actions/<notifier event>/error".
733 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
735 # cd /sys/kernel/debug/notifier-error-inject/memory
736 # echo -12 > actions/MEM_GOING_OFFLINE/error
737 # echo offline > /sys/devices/system/memory/memoryXXX/state
738 bash: echo: write error: Cannot allocate memory
740 To compile this code as a module, choose M here: the module will
741 be called memory-notifier-error-inject.
745 config DEBUG_PER_CPU_MAPS
746 bool "Debug access to per_cpu maps"
747 depends on DEBUG_KERNEL
750 Say Y to verify that the per_cpu map being accessed has
751 been set up. This adds a fair amount of code to kernel memory
752 and decreases performance.
757 bool "Highmem debugging"
758 depends on DEBUG_KERNEL && HIGHMEM
760 This option enables additional error checking for high memory
761 systems. Disable for production systems.
763 config HAVE_DEBUG_STACKOVERFLOW
766 config DEBUG_STACKOVERFLOW
767 bool "Check for stack overflows"
768 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW
770 Say Y here if you want to check for overflows of kernel, IRQ
771 and exception stacks (if your architecture uses them). This
772 option will show detailed messages if free stack space drops
773 below a certain limit.
775 These kinds of bugs usually occur when call-chains in the
776 kernel get too deep, especially when interrupts are
779 Use this in cases where you see apparently random memory
780 corruption, especially if it appears in 'struct thread_info'
782 If in doubt, say "N".
784 source "lib/Kconfig.kasan"
786 endmenu # "Memory Debugging"
789 bool "Debug shared IRQ handlers"
790 depends on DEBUG_KERNEL
792 Enable this to generate a spurious interrupt as soon as a shared
793 interrupt handler is registered, and just before one is deregistered.
794 Drivers ought to be able to handle interrupts coming in at those
795 points; some don't and need to be caught.
797 menu "Debug Oops, Lockups and Hangs"
802 Say Y here to enable the kernel to panic when it oopses. This
803 has the same effect as setting oops=panic on the kernel command
806 This feature is useful to ensure that the kernel does not do
807 anything erroneous after an oops which could result in data
808 corruption or other issues.
812 config PANIC_ON_OOPS_VALUE
815 default 0 if !PANIC_ON_OOPS
816 default 1 if PANIC_ON_OOPS
822 Set the timeout value (in seconds) until a reboot occurs when the
823 the kernel panics. If n = 0, then we wait forever. A timeout
824 value n > 0 will wait n seconds before rebooting, while a timeout
825 value n < 0 will reboot immediately.
827 config LOCKUP_DETECTOR
830 config SOFTLOCKUP_DETECTOR
831 bool "Detect Soft Lockups"
832 depends on DEBUG_KERNEL && !S390
833 select LOCKUP_DETECTOR
835 Say Y here to enable the kernel to act as a watchdog to detect
838 Softlockups are bugs that cause the kernel to loop in kernel
839 mode for more than 20 seconds, without giving other tasks a
840 chance to run. The current stack trace is displayed upon
841 detection and the system will stay locked up.
843 config BOOTPARAM_SOFTLOCKUP_PANIC
844 bool "Panic (Reboot) On Soft Lockups"
845 depends on SOFTLOCKUP_DETECTOR
847 Say Y here to enable the kernel to panic on "soft lockups",
848 which are bugs that cause the kernel to loop in kernel
849 mode for more than 20 seconds (configurable using the watchdog_thresh
850 sysctl), without giving other tasks a chance to run.
852 The panic can be used in combination with panic_timeout,
853 to cause the system to reboot automatically after a
854 lockup has been detected. This feature is useful for
855 high-availability systems that have uptime guarantees and
856 where a lockup must be resolved ASAP.
860 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
862 depends on SOFTLOCKUP_DETECTOR
864 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
865 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
867 config HARDLOCKUP_DETECTOR_PERF
869 select SOFTLOCKUP_DETECTOR
872 # Enables a timestamp based low pass filter to compensate for perf based
873 # hard lockup detection which runs too fast due to turbo modes.
875 config HARDLOCKUP_CHECK_TIMESTAMP
879 # arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard
880 # lockup detector rather than the perf based detector.
882 config HARDLOCKUP_DETECTOR
883 bool "Detect Hard Lockups"
884 depends on DEBUG_KERNEL && !S390
885 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_ARCH
886 select LOCKUP_DETECTOR
887 select HARDLOCKUP_DETECTOR_PERF if HAVE_HARDLOCKUP_DETECTOR_PERF
888 select HARDLOCKUP_DETECTOR_ARCH if HAVE_HARDLOCKUP_DETECTOR_ARCH
890 Say Y here to enable the kernel to act as a watchdog to detect
893 Hardlockups are bugs that cause the CPU to loop in kernel mode
894 for more than 10 seconds, without letting other interrupts have a
895 chance to run. The current stack trace is displayed upon detection
896 and the system will stay locked up.
898 config BOOTPARAM_HARDLOCKUP_PANIC
899 bool "Panic (Reboot) On Hard Lockups"
900 depends on HARDLOCKUP_DETECTOR
902 Say Y here to enable the kernel to panic on "hard lockups",
903 which are bugs that cause the kernel to loop in kernel
904 mode with interrupts disabled for more than 10 seconds (configurable
905 using the watchdog_thresh sysctl).
909 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
911 depends on HARDLOCKUP_DETECTOR
913 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
914 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
916 config DETECT_HUNG_TASK
917 bool "Detect Hung Tasks"
918 depends on DEBUG_KERNEL
919 default SOFTLOCKUP_DETECTOR
921 Say Y here to enable the kernel to detect "hung tasks",
922 which are bugs that cause the task to be stuck in
923 uninterruptible "D" state indefinitely.
925 When a hung task is detected, the kernel will print the
926 current stack trace (which you should report), but the
927 task will stay in uninterruptible state. If lockdep is
928 enabled then all held locks will also be reported. This
929 feature has negligible overhead.
931 config DEFAULT_HUNG_TASK_TIMEOUT
932 int "Default timeout for hung task detection (in seconds)"
933 depends on DETECT_HUNG_TASK
936 This option controls the default timeout (in seconds) used
937 to determine when a task has become non-responsive and should
940 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
941 sysctl or by writing a value to
942 /proc/sys/kernel/hung_task_timeout_secs.
944 A timeout of 0 disables the check. The default is two minutes.
945 Keeping the default should be fine in most cases.
947 config BOOTPARAM_HUNG_TASK_PANIC
948 bool "Panic (Reboot) On Hung Tasks"
949 depends on DETECT_HUNG_TASK
951 Say Y here to enable the kernel to panic on "hung tasks",
952 which are bugs that cause the kernel to leave a task stuck
953 in uninterruptible "D" state.
955 The panic can be used in combination with panic_timeout,
956 to cause the system to reboot automatically after a
957 hung task has been detected. This feature is useful for
958 high-availability systems that have uptime guarantees and
959 where a hung tasks must be resolved ASAP.
963 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
965 depends on DETECT_HUNG_TASK
967 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
968 default 1 if BOOTPARAM_HUNG_TASK_PANIC
971 bool "Detect Workqueue Stalls"
972 depends on DEBUG_KERNEL
974 Say Y here to enable stall detection on workqueues. If a
975 worker pool doesn't make forward progress on a pending work
976 item for over a given amount of time, 30s by default, a
977 warning message is printed along with dump of workqueue
978 state. This can be configured through kernel parameter
979 "workqueue.watchdog_thresh" and its sysfs counterpart.
981 endmenu # "Debug lockups and hangs"
983 menu "Scheduler Debugging"
986 bool "Collect scheduler debugging info"
987 depends on DEBUG_KERNEL && PROC_FS
990 If you say Y here, the /proc/sched_debug file will be provided
991 that can help debug the scheduler. The runtime overhead of this
999 bool "Collect scheduler statistics"
1000 depends on DEBUG_KERNEL && PROC_FS
1003 If you say Y here, additional code will be inserted into the
1004 scheduler and related routines to collect statistics about
1005 scheduler behavior and provide them in /proc/schedstat. These
1006 stats may be useful for both tuning and debugging the scheduler
1007 If you aren't debugging the scheduler or trying to tune a specific
1008 application, you can say N to avoid the very slight overhead
1013 config DEBUG_TIMEKEEPING
1014 bool "Enable extra timekeeping sanity checking"
1016 This option will enable additional timekeeping sanity checks
1017 which may be helpful when diagnosing issues where timekeeping
1018 problems are suspected.
1020 This may include checks in the timekeeping hotpaths, so this
1021 option may have a (very small) performance impact to some
1026 config DEBUG_PREEMPT
1027 bool "Debug preemptible kernel"
1028 depends on DEBUG_KERNEL && PREEMPTION && TRACE_IRQFLAGS_SUPPORT
1031 If you say Y here then the kernel will use a debug variant of the
1032 commonly used smp_processor_id() function and will print warnings
1033 if kernel code uses it in a preemption-unsafe way. Also, the kernel
1034 will detect preemption count underflows.
1036 menu "Lock Debugging (spinlocks, mutexes, etc...)"
1038 config LOCK_DEBUGGING_SUPPORT
1040 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
1043 config PROVE_LOCKING
1044 bool "Lock debugging: prove locking correctness"
1045 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1047 select DEBUG_SPINLOCK
1048 select DEBUG_MUTEXES
1049 select DEBUG_RT_MUTEXES if RT_MUTEXES
1051 select DEBUG_WW_MUTEX_SLOWPATH
1052 select DEBUG_LOCK_ALLOC
1053 select TRACE_IRQFLAGS
1056 This feature enables the kernel to prove that all locking
1057 that occurs in the kernel runtime is mathematically
1058 correct: that under no circumstance could an arbitrary (and
1059 not yet triggered) combination of observed locking
1060 sequences (on an arbitrary number of CPUs, running an
1061 arbitrary number of tasks and interrupt contexts) cause a
1064 In short, this feature enables the kernel to report locking
1065 related deadlocks before they actually occur.
1067 The proof does not depend on how hard and complex a
1068 deadlock scenario would be to trigger: how many
1069 participant CPUs, tasks and irq-contexts would be needed
1070 for it to trigger. The proof also does not depend on
1071 timing: if a race and a resulting deadlock is possible
1072 theoretically (no matter how unlikely the race scenario
1073 is), it will be proven so and will immediately be
1074 reported by the kernel (once the event is observed that
1075 makes the deadlock theoretically possible).
1077 If a deadlock is impossible (i.e. the locking rules, as
1078 observed by the kernel, are mathematically correct), the
1079 kernel reports nothing.
1081 NOTE: this feature can also be enabled for rwlocks, mutexes
1082 and rwsems - in which case all dependencies between these
1083 different locking variants are observed and mapped too, and
1084 the proof of observed correctness is also maintained for an
1085 arbitrary combination of these separate locking variants.
1087 For more details, see Documentation/locking/lockdep-design.rst.
1090 bool "Lock usage statistics"
1091 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1093 select DEBUG_SPINLOCK
1094 select DEBUG_MUTEXES
1095 select DEBUG_RT_MUTEXES if RT_MUTEXES
1096 select DEBUG_LOCK_ALLOC
1099 This feature enables tracking lock contention points
1101 For more details, see Documentation/locking/lockstat.rst
1103 This also enables lock events required by "perf lock",
1105 If you want to use "perf lock", you also need to turn on
1106 CONFIG_EVENT_TRACING.
1108 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
1109 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
1111 config DEBUG_RT_MUTEXES
1112 bool "RT Mutex debugging, deadlock detection"
1113 depends on DEBUG_KERNEL && RT_MUTEXES
1115 This allows rt mutex semantics violations and rt mutex related
1116 deadlocks (lockups) to be detected and reported automatically.
1118 config DEBUG_SPINLOCK
1119 bool "Spinlock and rw-lock debugging: basic checks"
1120 depends on DEBUG_KERNEL
1121 select UNINLINE_SPIN_UNLOCK
1123 Say Y here and build SMP to catch missing spinlock initialization
1124 and certain other kinds of spinlock errors commonly made. This is
1125 best used in conjunction with the NMI watchdog so that spinlock
1126 deadlocks are also debuggable.
1128 config DEBUG_MUTEXES
1129 bool "Mutex debugging: basic checks"
1130 depends on DEBUG_KERNEL
1132 This feature allows mutex semantics violations to be detected and
1135 config DEBUG_WW_MUTEX_SLOWPATH
1136 bool "Wait/wound mutex debugging: Slowpath testing"
1137 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1138 select DEBUG_LOCK_ALLOC
1139 select DEBUG_SPINLOCK
1140 select DEBUG_MUTEXES
1142 This feature enables slowpath testing for w/w mutex users by
1143 injecting additional -EDEADLK wound/backoff cases. Together with
1144 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this
1145 will test all possible w/w mutex interface abuse with the
1146 exception of simply not acquiring all the required locks.
1147 Note that this feature can introduce significant overhead, so
1148 it really should not be enabled in a production or distro kernel,
1149 even a debug kernel. If you are a driver writer, enable it. If
1150 you are a distro, do not.
1153 bool "RW Semaphore debugging: basic checks"
1154 depends on DEBUG_KERNEL
1156 This debugging feature allows mismatched rw semaphore locks
1157 and unlocks to be detected and reported.
1159 config DEBUG_LOCK_ALLOC
1160 bool "Lock debugging: detect incorrect freeing of live locks"
1161 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1162 select DEBUG_SPINLOCK
1163 select DEBUG_MUTEXES
1164 select DEBUG_RT_MUTEXES if RT_MUTEXES
1167 This feature will check whether any held lock (spinlock, rwlock,
1168 mutex or rwsem) is incorrectly freed by the kernel, via any of the
1169 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
1170 vfree(), etc.), whether a live lock is incorrectly reinitialized via
1171 spin_lock_init()/mutex_init()/etc., or whether there is any lock
1172 held during task exit.
1176 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT
1178 select FRAME_POINTER if !MIPS && !PPC && !ARM && !S390 && !MICROBLAZE && !ARC && !X86
1182 config LOCKDEP_SMALL
1185 config DEBUG_LOCKDEP
1186 bool "Lock dependency engine debugging"
1187 depends on DEBUG_KERNEL && LOCKDEP
1189 If you say Y here, the lock dependency engine will do
1190 additional runtime checks to debug itself, at the price
1191 of more runtime overhead.
1193 config DEBUG_ATOMIC_SLEEP
1194 bool "Sleep inside atomic section checking"
1195 select PREEMPT_COUNT
1196 depends on DEBUG_KERNEL
1197 depends on !ARCH_NO_PREEMPT
1199 If you say Y here, various routines which may sleep will become very
1200 noisy if they are called inside atomic sections: when a spinlock is
1201 held, inside an rcu read side critical section, inside preempt disabled
1202 sections, inside an interrupt, etc...
1204 config DEBUG_LOCKING_API_SELFTESTS
1205 bool "Locking API boot-time self-tests"
1206 depends on DEBUG_KERNEL
1208 Say Y here if you want the kernel to run a short self-test during
1209 bootup. The self-test checks whether common types of locking bugs
1210 are detected by debugging mechanisms or not. (if you disable
1211 lock debugging then those bugs wont be detected of course.)
1212 The following locking APIs are covered: spinlocks, rwlocks,
1215 config LOCK_TORTURE_TEST
1216 tristate "torture tests for locking"
1217 depends on DEBUG_KERNEL
1220 This option provides a kernel module that runs torture tests
1221 on kernel locking primitives. The kernel module may be built
1222 after the fact on the running kernel to be tested, if desired.
1224 Say Y here if you want kernel locking-primitive torture tests
1225 to be built into the kernel.
1226 Say M if you want these torture tests to build as a module.
1227 Say N if you are unsure.
1229 config WW_MUTEX_SELFTEST
1230 tristate "Wait/wound mutex selftests"
1232 This option provides a kernel module that runs tests on the
1233 on the struct ww_mutex locking API.
1235 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction
1236 with this test harness.
1238 Say M if you want these self tests to build as a module.
1239 Say N if you are unsure.
1241 endmenu # lock debugging
1243 config TRACE_IRQFLAGS
1246 Enables hooks to interrupt enabling and disabling for
1247 either tracing or lock debugging.
1250 bool "Stack backtrace support"
1251 depends on STACKTRACE_SUPPORT
1253 This option causes the kernel to create a /proc/pid/stack for
1254 every process, showing its current stack trace.
1255 It is also used by various kernel debugging features that require
1256 stack trace generation.
1258 config WARN_ALL_UNSEEDED_RANDOM
1259 bool "Warn for all uses of unseeded randomness"
1262 Some parts of the kernel contain bugs relating to their use of
1263 cryptographically secure random numbers before it's actually possible
1264 to generate those numbers securely. This setting ensures that these
1265 flaws don't go unnoticed, by enabling a message, should this ever
1266 occur. This will allow people with obscure setups to know when things
1267 are going wrong, so that they might contact developers about fixing
1270 Unfortunately, on some models of some architectures getting
1271 a fully seeded CRNG is extremely difficult, and so this can
1272 result in dmesg getting spammed for a surprisingly long
1273 time. This is really bad from a security perspective, and
1274 so architecture maintainers really need to do what they can
1275 to get the CRNG seeded sooner after the system is booted.
1276 However, since users cannot do anything actionable to
1277 address this, by default the kernel will issue only a single
1278 warning for the first use of unseeded randomness.
1280 Say Y here if you want to receive warnings for all uses of
1281 unseeded randomness. This will be of use primarily for
1282 those developers interested in improving the security of
1283 Linux kernels running on their architecture (or
1286 config DEBUG_KOBJECT
1287 bool "kobject debugging"
1288 depends on DEBUG_KERNEL
1290 If you say Y here, some extra kobject debugging messages will be sent
1293 config DEBUG_KOBJECT_RELEASE
1294 bool "kobject release debugging"
1295 depends on DEBUG_OBJECTS_TIMERS
1297 kobjects are reference counted objects. This means that their
1298 last reference count put is not predictable, and the kobject can
1299 live on past the point at which a driver decides to drop it's
1300 initial reference to the kobject gained on allocation. An
1301 example of this would be a struct device which has just been
1304 However, some buggy drivers assume that after such an operation,
1305 the memory backing the kobject can be immediately freed. This
1306 goes completely against the principles of a refcounted object.
1308 If you say Y here, the kernel will delay the release of kobjects
1309 on the last reference count to improve the visibility of this
1310 kind of kobject release bug.
1312 config HAVE_DEBUG_BUGVERBOSE
1315 menu "Debug kernel data structures"
1318 bool "Debug linked list manipulation"
1319 depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION
1321 Enable this to turn on extended checks in the linked-list
1327 bool "Debug priority linked list manipulation"
1328 depends on DEBUG_KERNEL
1330 Enable this to turn on extended checks in the priority-ordered
1331 linked-list (plist) walking routines. This checks the entire
1332 list multiple times during each manipulation.
1337 bool "Debug SG table operations"
1338 depends on DEBUG_KERNEL
1340 Enable this to turn on checks on scatter-gather tables. This can
1341 help find problems with drivers that do not properly initialize
1346 config DEBUG_NOTIFIERS
1347 bool "Debug notifier call chains"
1348 depends on DEBUG_KERNEL
1350 Enable this to turn on sanity checking for notifier call chains.
1351 This is most useful for kernel developers to make sure that
1352 modules properly unregister themselves from notifier chains.
1353 This is a relatively cheap check but if you care about maximum
1356 config BUG_ON_DATA_CORRUPTION
1357 bool "Trigger a BUG when data corruption is detected"
1360 Select this option if the kernel should BUG when it encounters
1361 data corruption in kernel memory structures when they get checked
1368 config DEBUG_CREDENTIALS
1369 bool "Debug credential management"
1370 depends on DEBUG_KERNEL
1372 Enable this to turn on some debug checking for credential
1373 management. The additional code keeps track of the number of
1374 pointers from task_structs to any given cred struct, and checks to
1375 see that this number never exceeds the usage count of the cred
1378 Furthermore, if SELinux is enabled, this also checks that the
1379 security pointer in the cred struct is never seen to be invalid.
1383 source "kernel/rcu/Kconfig.debug"
1385 config DEBUG_WQ_FORCE_RR_CPU
1386 bool "Force round-robin CPU selection for unbound work items"
1387 depends on DEBUG_KERNEL
1390 Workqueue used to implicitly guarantee that work items queued
1391 without explicit CPU specified are put on the local CPU. This
1392 guarantee is no longer true and while local CPU is still
1393 preferred work items may be put on foreign CPUs. Kernel
1394 parameter "workqueue.debug_force_rr_cpu" is added to force
1395 round-robin CPU selection to flush out usages which depend on the
1396 now broken guarantee. This config option enables the debug
1397 feature by default. When enabled, memory and cache locality will
1400 config DEBUG_BLOCK_EXT_DEVT
1401 bool "Force extended block device numbers and spread them"
1402 depends on DEBUG_KERNEL
1406 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1407 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1408 YOU ARE DOING. Distros, please enable this and fix whatever
1411 Conventionally, block device numbers are allocated from
1412 predetermined contiguous area. However, extended block area
1413 may introduce non-contiguous block device numbers. This
1414 option forces most block device numbers to be allocated from
1415 the extended space and spreads them to discover kernel or
1416 userland code paths which assume predetermined contiguous
1417 device number allocation.
1419 Note that turning on this debug option shuffles all the
1420 device numbers for all IDE and SCSI devices including libata
1421 ones, so root partition specified using device number
1422 directly (via rdev or root=MAJ:MIN) won't work anymore.
1423 Textual device names (root=/dev/sdXn) will continue to work.
1425 Say N if you are unsure.
1427 config CPU_HOTPLUG_STATE_CONTROL
1428 bool "Enable CPU hotplug state control"
1429 depends on DEBUG_KERNEL
1430 depends on HOTPLUG_CPU
1433 Allows to write steps between "offline" and "online" to the CPUs
1434 sysfs target file so states can be stepped granular. This is a debug
1435 option for now as the hotplug machinery cannot be stopped and
1436 restarted at arbitrary points yet.
1438 Say N if your are unsure.
1441 bool "Latency measuring infrastructure"
1442 depends on DEBUG_KERNEL
1443 depends on STACKTRACE_SUPPORT
1445 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1452 Enable this option if you want to use the LatencyTOP tool
1453 to find out which userspace is blocking on what kernel operations.
1455 source "kernel/trace/Kconfig"
1457 config PROVIDE_OHCI1394_DMA_INIT
1458 bool "Remote debugging over FireWire early on boot"
1459 depends on PCI && X86
1461 If you want to debug problems which hang or crash the kernel early
1462 on boot and the crashing machine has a FireWire port, you can use
1463 this feature to remotely access the memory of the crashed machine
1464 over FireWire. This employs remote DMA as part of the OHCI1394
1465 specification which is now the standard for FireWire controllers.
1467 With remote DMA, you can monitor the printk buffer remotely using
1468 firescope and access all memory below 4GB using fireproxy from gdb.
1469 Even controlling a kernel debugger is possible using remote DMA.
1473 If ohci1394_dma=early is used as boot parameter, it will initialize
1474 all OHCI1394 controllers which are found in the PCI config space.
1476 As all changes to the FireWire bus such as enabling and disabling
1477 devices cause a bus reset and thereby disable remote DMA for all
1478 devices, be sure to have the cable plugged and FireWire enabled on
1479 the debugging host before booting the debug target for debugging.
1481 This code (~1k) is freed after boot. By then, the firewire stack
1482 in charge of the OHCI-1394 controllers should be used instead.
1484 See Documentation/debugging-via-ohci1394.txt for more information.
1486 source "samples/Kconfig"
1488 config ARCH_HAS_DEVMEM_IS_ALLOWED
1491 config STRICT_DEVMEM
1492 bool "Filter access to /dev/mem"
1493 depends on MMU && DEVMEM
1494 depends on ARCH_HAS_DEVMEM_IS_ALLOWED
1495 default y if PPC || X86 || ARM64
1497 If this option is disabled, you allow userspace (root) access to all
1498 of memory, including kernel and userspace memory. Accidental
1499 access to this is obviously disastrous, but specific access can
1500 be used by people debugging the kernel. Note that with PAT support
1501 enabled, even in this case there are restrictions on /dev/mem
1502 use due to the cache aliasing requirements.
1504 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem
1505 file only allows userspace access to PCI space and the BIOS code and
1506 data regions. This is sufficient for dosemu and X and all common
1511 config IO_STRICT_DEVMEM
1512 bool "Filter I/O access to /dev/mem"
1513 depends on STRICT_DEVMEM
1515 If this option is disabled, you allow userspace (root) access to all
1516 io-memory regardless of whether a driver is actively using that
1517 range. Accidental access to this is obviously disastrous, but
1518 specific access can be used by people debugging kernel drivers.
1520 If this option is switched on, the /dev/mem file only allows
1521 userspace access to *idle* io-memory ranges (see /proc/iomem) This
1522 may break traditional users of /dev/mem (dosemu, legacy X, etc...)
1523 if the driver using a given range cannot be disabled.
1527 menu "$(SRCARCH) Debugging"
1529 source "arch/$(SRCARCH)/Kconfig.debug"
1533 menu "Kernel Testing and Coverage"
1535 source "lib/kunit/Kconfig"
1537 config NOTIFIER_ERROR_INJECTION
1538 tristate "Notifier error injection"
1539 depends on DEBUG_KERNEL
1542 This option provides the ability to inject artificial errors to
1543 specified notifier chain callbacks. It is useful to test the error
1544 handling of notifier call chain failures.
1548 config PM_NOTIFIER_ERROR_INJECT
1549 tristate "PM notifier error injection module"
1550 depends on PM && NOTIFIER_ERROR_INJECTION
1551 default m if PM_DEBUG
1553 This option provides the ability to inject artificial errors to
1554 PM notifier chain callbacks. It is controlled through debugfs
1555 interface /sys/kernel/debug/notifier-error-inject/pm
1557 If the notifier call chain should be failed with some events
1558 notified, write the error code to "actions/<notifier event>/error".
1560 Example: Inject PM suspend error (-12 = -ENOMEM)
1562 # cd /sys/kernel/debug/notifier-error-inject/pm/
1563 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1564 # echo mem > /sys/power/state
1565 bash: echo: write error: Cannot allocate memory
1567 To compile this code as a module, choose M here: the module will
1568 be called pm-notifier-error-inject.
1572 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1573 tristate "OF reconfig notifier error injection module"
1574 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1576 This option provides the ability to inject artificial errors to
1577 OF reconfig notifier chain callbacks. It is controlled
1578 through debugfs interface under
1579 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1581 If the notifier call chain should be failed with some events
1582 notified, write the error code to "actions/<notifier event>/error".
1584 To compile this code as a module, choose M here: the module will
1585 be called of-reconfig-notifier-error-inject.
1589 config NETDEV_NOTIFIER_ERROR_INJECT
1590 tristate "Netdev notifier error injection module"
1591 depends on NET && NOTIFIER_ERROR_INJECTION
1593 This option provides the ability to inject artificial errors to
1594 netdevice notifier chain callbacks. It is controlled through debugfs
1595 interface /sys/kernel/debug/notifier-error-inject/netdev
1597 If the notifier call chain should be failed with some events
1598 notified, write the error code to "actions/<notifier event>/error".
1600 Example: Inject netdevice mtu change error (-22 = -EINVAL)
1602 # cd /sys/kernel/debug/notifier-error-inject/netdev
1603 # echo -22 > actions/NETDEV_CHANGEMTU/error
1604 # ip link set eth0 mtu 1024
1605 RTNETLINK answers: Invalid argument
1607 To compile this code as a module, choose M here: the module will
1608 be called netdev-notifier-error-inject.
1612 config FUNCTION_ERROR_INJECTION
1614 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES
1616 config FAULT_INJECTION
1617 bool "Fault-injection framework"
1618 depends on DEBUG_KERNEL
1620 Provide fault-injection framework.
1621 For more details, see Documentation/fault-injection/.
1624 bool "Fault-injection capability for kmalloc"
1625 depends on FAULT_INJECTION
1626 depends on SLAB || SLUB
1628 Provide fault-injection capability for kmalloc.
1630 config FAIL_PAGE_ALLOC
1631 bool "Fault-injection capabilitiy for alloc_pages()"
1632 depends on FAULT_INJECTION
1634 Provide fault-injection capability for alloc_pages().
1636 config FAIL_MAKE_REQUEST
1637 bool "Fault-injection capability for disk IO"
1638 depends on FAULT_INJECTION && BLOCK
1640 Provide fault-injection capability for disk IO.
1642 config FAIL_IO_TIMEOUT
1643 bool "Fault-injection capability for faking disk interrupts"
1644 depends on FAULT_INJECTION && BLOCK
1646 Provide fault-injection capability on end IO handling. This
1647 will make the block layer "forget" an interrupt as configured,
1648 thus exercising the error handling.
1650 Only works with drivers that use the generic timeout handling,
1651 for others it wont do anything.
1654 bool "Fault-injection capability for futexes"
1656 depends on FAULT_INJECTION && FUTEX
1658 Provide fault-injection capability for futexes.
1660 config FAULT_INJECTION_DEBUG_FS
1661 bool "Debugfs entries for fault-injection capabilities"
1662 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1664 Enable configuration of fault-injection capabilities via debugfs.
1666 config FAIL_FUNCTION
1667 bool "Fault-injection capability for functions"
1668 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION
1670 Provide function-based fault-injection capability.
1671 This will allow you to override a specific function with a return
1672 with given return value. As a result, function caller will see
1673 an error value and have to handle it. This is useful to test the
1674 error handling in various subsystems.
1676 config FAIL_MMC_REQUEST
1677 bool "Fault-injection capability for MMC IO"
1678 depends on FAULT_INJECTION_DEBUG_FS && MMC
1680 Provide fault-injection capability for MMC IO.
1681 This will make the mmc core return data errors. This is
1682 useful to test the error handling in the mmc block device
1683 and to test how the mmc host driver handles retries from
1686 config FAULT_INJECTION_STACKTRACE_FILTER
1687 bool "stacktrace filter for fault-injection capabilities"
1688 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1691 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86
1693 Provide stacktrace filter for fault-injection capabilities
1695 config ARCH_HAS_KCOV
1698 An architecture should select this when it can successfully
1699 build and run with CONFIG_KCOV. This typically requires
1700 disabling instrumentation for some early boot code.
1702 config CC_HAS_SANCOV_TRACE_PC
1703 def_bool $(cc-option,-fsanitize-coverage=trace-pc)
1707 bool "Code coverage for fuzzing"
1708 depends on ARCH_HAS_KCOV
1709 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS
1711 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC
1713 KCOV exposes kernel code coverage information in a form suitable
1714 for coverage-guided fuzzing (randomized testing).
1716 If RANDOMIZE_BASE is enabled, PC values will not be stable across
1717 different machines and across reboots. If you need stable PC values,
1718 disable RANDOMIZE_BASE.
1720 For more details, see Documentation/dev-tools/kcov.rst.
1722 config KCOV_ENABLE_COMPARISONS
1723 bool "Enable comparison operands collection by KCOV"
1725 depends on $(cc-option,-fsanitize-coverage=trace-cmp)
1727 KCOV also exposes operands of every comparison in the instrumented
1728 code along with operand sizes and PCs of the comparison instructions.
1729 These operands can be used by fuzzing engines to improve the quality
1730 of fuzzing coverage.
1732 config KCOV_INSTRUMENT_ALL
1733 bool "Instrument all code by default"
1737 If you are doing generic system call fuzzing (like e.g. syzkaller),
1738 then you will want to instrument the whole kernel and you should
1739 say y here. If you are doing more targeted fuzzing (like e.g.
1740 filesystem fuzzing with AFL) then you will want to enable coverage
1741 for more specific subsets of files, and should say n here.
1743 menuconfig RUNTIME_TESTING_MENU
1744 bool "Runtime Testing"
1747 if RUNTIME_TESTING_MENU
1750 tristate "Linux Kernel Dump Test Tool Module"
1753 This module enables testing of the different dumping mechanisms by
1754 inducing system failures at predefined crash points.
1755 If you don't need it: say N
1756 Choose M here to compile this code as a module. The module will be
1759 Documentation on how to use the module can be found in
1760 Documentation/fault-injection/provoke-crashes.rst
1762 config TEST_LIST_SORT
1763 tristate "Linked list sorting test"
1764 depends on DEBUG_KERNEL || m
1766 Enable this to turn on 'list_sort()' function test. This test is
1767 executed only once during system boot (so affects only boot time),
1768 or at module load time.
1773 tristate "Array-based sort test"
1774 depends on DEBUG_KERNEL || m
1776 This option enables the self-test function of 'sort()' at boot,
1777 or at module load time.
1781 config KPROBES_SANITY_TEST
1782 bool "Kprobes sanity tests"
1783 depends on DEBUG_KERNEL
1786 This option provides for testing basic kprobes functionality on
1787 boot. Samples of kprobe and kretprobe are inserted and
1788 verified for functionality.
1790 Say N if you are unsure.
1792 config BACKTRACE_SELF_TEST
1793 tristate "Self test for the backtrace code"
1794 depends on DEBUG_KERNEL
1796 This option provides a kernel module that can be used to test
1797 the kernel stack backtrace code. This option is not useful
1798 for distributions or general kernels, but only for kernel
1799 developers working on architecture code.
1801 Note that if you want to also test saved backtraces, you will
1802 have to enable STACKTRACE as well.
1804 Say N if you are unsure.
1807 tristate "Red-Black tree test"
1808 depends on DEBUG_KERNEL
1810 A benchmark measuring the performance of the rbtree library.
1811 Also includes rbtree invariant checks.
1813 config REED_SOLOMON_TEST
1814 tristate "Reed-Solomon library test"
1815 depends on DEBUG_KERNEL || m
1817 select REED_SOLOMON_ENC16
1818 select REED_SOLOMON_DEC16
1820 This option enables the self-test function of rslib at boot,
1821 or at module load time.
1825 config INTERVAL_TREE_TEST
1826 tristate "Interval tree test"
1827 depends on DEBUG_KERNEL
1828 select INTERVAL_TREE
1830 A benchmark measuring the performance of the interval tree library
1833 tristate "Per cpu operations test"
1834 depends on m && DEBUG_KERNEL
1836 Enable this option to build test module which validates per-cpu
1841 config ATOMIC64_SELFTEST
1842 tristate "Perform an atomic64_t self-test"
1844 Enable this option to test the atomic64_t functions at boot or
1845 at module load time.
1849 config ASYNC_RAID6_TEST
1850 tristate "Self test for hardware accelerated raid6 recovery"
1851 depends on ASYNC_RAID6_RECOV
1854 This is a one-shot self test that permutes through the
1855 recovery of all the possible two disk failure scenarios for a
1856 N-disk array. Recovery is performed with the asynchronous
1857 raid6 recovery routines, and will optionally use an offload
1858 engine if one is available.
1863 tristate "Test functions located in the hexdump module at runtime"
1865 config TEST_STRING_HELPERS
1866 tristate "Test functions located in the string_helpers module at runtime"
1869 tristate "Test strscpy*() family of functions at runtime"
1872 tristate "Test kstrto*() family of functions at runtime"
1875 tristate "Test printf() family of functions at runtime"
1878 tristate "Test bitmap_*() family of functions at runtime"
1880 Enable this option to test the bitmap functions at boot.
1884 config TEST_BITFIELD
1885 tristate "Test bitfield functions at runtime"
1887 Enable this option to test the bitfield functions at boot.
1892 tristate "Test functions located in the uuid module at runtime"
1895 tristate "Test the XArray code at runtime"
1897 config TEST_OVERFLOW
1898 tristate "Test check_*_overflow() functions at runtime"
1900 config TEST_RHASHTABLE
1901 tristate "Perform selftest on resizable hash table"
1903 Enable this option to test the rhashtable functions at boot.
1908 tristate "Perform selftest on hash functions"
1910 Enable this option to test the kernel's integer (<linux/hash.h>),
1911 string (<linux/stringhash.h>), and siphash (<linux/siphash.h>)
1912 hash functions on boot (or module load).
1914 This is intended to help people writing architecture-specific
1915 optimized versions. If unsure, say N.
1918 tristate "Perform selftest on IDA functions"
1921 tristate "Perform selftest on priority array manager"
1924 Enable this option to test priority array manager on boot
1929 config TEST_IRQ_TIMINGS
1930 bool "IRQ timings selftest"
1931 depends on IRQ_TIMINGS
1933 Enable this option to test the irq timings code on boot.
1938 tristate "Test module loading with 'hello world' module"
1941 This builds the "test_module" module that emits "Hello, world"
1942 on printk when loaded. It is designed to be used for basic
1943 evaluation of the module loading subsystem (for example when
1944 validating module verification). It lacks any extra dependencies,
1945 and will not normally be loaded by the system unless explicitly
1951 tristate "Test module for stress/performance analysis of vmalloc allocator"
1956 This builds the "test_vmalloc" module that should be used for
1957 stress and performance analysis. So, any new change for vmalloc
1958 subsystem can be evaluated from performance and stability point
1963 config TEST_USER_COPY
1964 tristate "Test user/kernel boundary protections"
1967 This builds the "test_user_copy" module that runs sanity checks
1968 on the copy_to/from_user infrastructure, making sure basic
1969 user/kernel boundary testing is working. If it fails to load,
1970 a regression has been detected in the user/kernel memory boundary
1976 tristate "Test BPF filter functionality"
1979 This builds the "test_bpf" module that runs various test vectors
1980 against the BPF interpreter or BPF JIT compiler depending on the
1981 current setting. This is in particular useful for BPF JIT compiler
1982 development, but also to run regression tests against changes in
1983 the interpreter code. It also enables test stubs for eBPF maps and
1984 verifier used by user space verifier testsuite.
1988 config TEST_BLACKHOLE_DEV
1989 tristate "Test blackhole netdev functionality"
1992 This builds the "test_blackhole_dev" module that validates the
1993 data path through this blackhole netdev.
1997 config FIND_BIT_BENCHMARK
1998 tristate "Test find_bit functions"
2000 This builds the "test_find_bit" module that measure find_*_bit()
2001 functions performance.
2005 config TEST_FIRMWARE
2006 tristate "Test firmware loading via userspace interface"
2007 depends on FW_LOADER
2009 This builds the "test_firmware" module that creates a userspace
2010 interface for testing firmware loading. This can be used to
2011 control the triggering of firmware loading without needing an
2012 actual firmware-using device. The contents can be rechecked by
2018 tristate "sysctl test driver"
2019 depends on PROC_SYSCTL
2021 This builds the "test_sysctl" module. This driver enables to test the
2022 proc sysctl interfaces available to drivers safely without affecting
2023 production knobs which might alter system functionality.
2027 config SYSCTL_KUNIT_TEST
2028 tristate "KUnit test for sysctl"
2031 This builds the proc sysctl unit test, which runs on boot.
2032 Tests the API contract and implementation correctness of sysctl.
2033 For more information on KUnit and unit tests in general please refer
2034 to the KUnit documentation in Documentation/dev-tools/kunit/.
2038 config LIST_KUNIT_TEST
2039 tristate "KUnit Test for Kernel Linked-list structures"
2042 This builds the linked list KUnit test suite.
2043 It tests that the API and basic functionality of the list_head type
2044 and associated macros.
2046 KUnit tests run during boot and output the results to the debug log
2047 in TAP format (http://testanything.org/). Only useful for kernel devs
2048 running the KUnit test harness, and not intended for inclusion into a
2051 For more information on KUnit and unit tests in general please refer
2052 to the KUnit documentation in Documentation/dev-tools/kunit/.
2057 tristate "udelay test driver"
2059 This builds the "udelay_test" module that helps to make sure
2060 that udelay() is working properly.
2064 config TEST_STATIC_KEYS
2065 tristate "Test static keys"
2068 Test the static key interfaces.
2073 tristate "kmod stress tester"
2075 depends on NETDEVICES && NET_CORE && INET # for TUN
2082 Test the kernel's module loading mechanism: kmod. kmod implements
2083 support to load modules using the Linux kernel's usermode helper.
2084 This test provides a series of tests against kmod.
2086 Although technically you can either build test_kmod as a module or
2087 into the kernel we disallow building it into the kernel since
2088 it stress tests request_module() and this will very likely cause
2089 some issues by taking over precious threads available from other
2090 module load requests, ultimately this could be fatal.
2094 tools/testing/selftests/kmod/kmod.sh --help
2098 config TEST_DEBUG_VIRTUAL
2099 tristate "Test CONFIG_DEBUG_VIRTUAL feature"
2100 depends on DEBUG_VIRTUAL
2102 Test the kernel's ability to detect incorrect calls to
2103 virt_to_phys() done against the non-linear part of the
2104 kernel's virtual address map.
2108 config TEST_MEMCAT_P
2109 tristate "Test memcat_p() helper function"
2111 Test the memcat_p() helper for correctly merging two
2112 pointer arrays together.
2116 config TEST_LIVEPATCH
2117 tristate "Test livepatching"
2119 depends on DYNAMIC_DEBUG
2120 depends on LIVEPATCH
2123 Test kernel livepatching features for correctness. The tests will
2124 load test modules that will be livepatched in various scenarios.
2126 To run all the livepatching tests:
2128 make -C tools/testing/selftests TARGETS=livepatch run_tests
2130 Alternatively, individual tests may be invoked:
2132 tools/testing/selftests/livepatch/test-callbacks.sh
2133 tools/testing/selftests/livepatch/test-livepatch.sh
2134 tools/testing/selftests/livepatch/test-shadow-vars.sh
2139 tristate "Perform selftest on object aggreration manager"
2143 Enable this option to test object aggregation manager on boot
2147 config TEST_STACKINIT
2148 tristate "Test level of stack variable initialization"
2150 Test if the kernel is zero-initializing stack variables and
2151 padding. Coverage is controlled by compiler flags,
2152 CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF,
2153 or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL.
2158 tristate "Test heap/page initialization"
2160 Test if the kernel is zero-initializing heap and page allocations.
2161 This can be useful to test init_on_alloc and init_on_free features.
2165 endif # RUNTIME_TESTING_MENU
2170 This option adds a kernel parameter 'memtest', which allows memtest
2172 memtest=0, mean disabled; -- default
2173 memtest=1, mean do 1 test pattern;
2175 memtest=17, mean do 17 test patterns.
2176 If you are unsure how to answer this question, answer N.
2180 config HYPERV_TESTING
2181 bool "Microsoft Hyper-V driver testing"
2183 depends on HYPERV && DEBUG_FS
2185 Select this option to enable Hyper-V vmbus testing.
2187 endmenu # "Kernel Testing and Coverage"
2189 endmenu # Kernel hacking