3 bool "Show timing information on printks"
6 Selecting this option causes time stamps of the printk()
7 messages to be added to the output of the syslog() system
8 call and at the console.
10 The timestamp is always recorded internally, and exported
11 to /dev/kmsg. This flag just specifies if the timestamp should
12 be included, not that the timestamp is recorded.
14 The behavior is also controlled by the kernel command line
15 parameter printk.time=1. See Documentation/kernel-parameters.txt
17 config DEFAULT_MESSAGE_LOGLEVEL
18 int "Default message log level (1-7)"
22 Default log level for printk statements with no specified priority.
24 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
25 that are auditing their logs closely may want to set it to a lower
28 config ENABLE_WARN_DEPRECATED
29 bool "Enable __deprecated logic"
32 Enable the __deprecated logic in the kernel build.
33 Disable this to suppress the "warning: 'foo' is deprecated
34 (declared at kernel/power/somefile.c:1234)" messages.
36 config ENABLE_MUST_CHECK
37 bool "Enable __must_check logic"
40 Enable the __must_check logic in the kernel build. Disable this to
41 suppress the "warning: ignoring return value of 'foo', declared with
42 attribute warn_unused_result" messages.
45 int "Warn for stack frames larger than (needs gcc 4.4)"
47 default 1024 if !64BIT
50 Tell gcc to warn at build time for stack frames larger than this.
51 Setting this too low will cause a lot of warnings.
52 Setting it to 0 disables the warning.
56 bool "Magic SysRq key"
59 If you say Y here, you will have some control over the system even
60 if the system crashes for example during kernel debugging (e.g., you
61 will be able to flush the buffer cache to disk, reboot the system
62 immediately or dump some status information). This is accomplished
63 by pressing various keys while holding SysRq (Alt+PrintScreen). It
64 also works on a serial console (on PC hardware at least), if you
65 send a BREAK and then within 5 seconds a command keypress. The
66 keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
67 unless you really know what this hack does.
70 bool "Strip assembler-generated symbols during link"
73 Strip internal assembler-generated symbols during a link (symbols
74 that look like '.Lxxx') so they don't pollute the output of
75 get_wchan() and suchlike.
78 bool "Generate readable assembler code"
79 depends on DEBUG_KERNEL
81 Disable some compiler optimizations that tend to generate human unreadable
82 assembler output. This may make the kernel slightly slower, but it helps
83 to keep kernel developers who have to stare a lot at assembler listings
87 bool "Enable unused/obsolete exported symbols"
90 Unused but exported symbols make the kernel needlessly bigger. For
91 that reason most of these unused exports will soon be removed. This
92 option is provided temporarily to provide a transition period in case
93 some external kernel module needs one of these symbols anyway. If you
94 encounter such a case in your module, consider if you are actually
95 using the right API. (rationale: since nobody in the kernel is using
96 this in a module, there is a pretty good chance it's actually the
97 wrong interface to use). If you really need the symbol, please send a
98 mail to the linux kernel mailing list mentioning the symbol and why
99 you really need it, and what the merge plan to the mainline kernel for
103 bool "Debug Filesystem"
105 debugfs is a virtual file system that kernel developers use to put
106 debugging files into. Enable this option to be able to read and
107 write to these files.
109 For detailed documentation on the debugfs API, see
110 Documentation/DocBook/filesystems.
115 bool "Run 'make headers_check' when building vmlinux"
118 This option will extract the user-visible kernel headers whenever
119 building the kernel, and will run basic sanity checks on them to
120 ensure that exported files do not attempt to include files which
121 were not exported, etc.
123 If you're making modifications to header files which are
124 relevant for userspace, say 'Y', and check the headers
125 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
126 your build tree), to make sure they're suitable.
128 config DEBUG_SECTION_MISMATCH
129 bool "Enable full Section mismatch analysis"
131 The section mismatch analysis checks if there are illegal
132 references from one section to another section.
133 During linktime or runtime, some sections are dropped;
134 any use of code/data previously in these sections would
135 most likely result in an oops.
136 In the code, functions and variables are annotated with
137 __init, __cpuinit, etc. (see the full list in include/linux/init.h),
138 which results in the code/data being placed in specific sections.
139 The section mismatch analysis is always performed after a full
140 kernel build, and enabling this option causes the following
141 additional steps to occur:
142 - Add the option -fno-inline-functions-called-once to gcc commands.
143 When inlining a function annotated with __init in a non-init
144 function, we would lose the section information and thus
145 the analysis would not catch the illegal reference.
146 This option tells gcc to inline less (but it does result in
148 - Run the section mismatch analysis for each module/built-in.o file.
149 When we run the section mismatch analysis on vmlinux.o, we
150 lose valueble information about where the mismatch was
152 Running the analysis for each module/built-in.o file
153 tells where the mismatch happens much closer to the
154 source. The drawback is that the same mismatch is
155 reported at least twice.
156 - Enable verbose reporting from modpost in order to help resolve
157 the section mismatches that are reported.
160 bool "Kernel debugging"
162 Say Y here if you are developing drivers or trying to debug and
163 identify kernel problems.
166 bool "Debug shared IRQ handlers"
167 depends on DEBUG_KERNEL && GENERIC_HARDIRQS
169 Enable this to generate a spurious interrupt as soon as a shared
170 interrupt handler is registered, and just before one is deregistered.
171 Drivers ought to be able to handle interrupts coming in at those
172 points; some don't and need to be caught.
174 config LOCKUP_DETECTOR
175 bool "Detect Hard and Soft Lockups"
176 depends on DEBUG_KERNEL && !S390
178 Say Y here to enable the kernel to act as a watchdog to detect
179 hard and soft lockups.
181 Softlockups are bugs that cause the kernel to loop in kernel
182 mode for more than 20 seconds, without giving other tasks a
183 chance to run. The current stack trace is displayed upon
184 detection and the system will stay locked up.
186 Hardlockups are bugs that cause the CPU to loop in kernel mode
187 for more than 10 seconds, without letting other interrupts have a
188 chance to run. The current stack trace is displayed upon detection
189 and the system will stay locked up.
191 The overhead should be minimal. A periodic hrtimer runs to
192 generate interrupts and kick the watchdog task every 4 seconds.
193 An NMI is generated every 10 seconds or so to check for hardlockups.
195 The frequency of hrtimer and NMI events and the soft and hard lockup
196 thresholds can be controlled through the sysctl watchdog_thresh.
198 config HARDLOCKUP_DETECTOR
200 depends on LOCKUP_DETECTOR && !HAVE_NMI_WATCHDOG
201 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI
203 config BOOTPARAM_HARDLOCKUP_PANIC
204 bool "Panic (Reboot) On Hard Lockups"
205 depends on HARDLOCKUP_DETECTOR
207 Say Y here to enable the kernel to panic on "hard lockups",
208 which are bugs that cause the kernel to loop in kernel
209 mode with interrupts disabled for more than 10 seconds (configurable
210 using the watchdog_thresh sysctl).
214 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
216 depends on HARDLOCKUP_DETECTOR
218 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
219 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
221 config BOOTPARAM_SOFTLOCKUP_PANIC
222 bool "Panic (Reboot) On Soft Lockups"
223 depends on LOCKUP_DETECTOR
225 Say Y here to enable the kernel to panic on "soft lockups",
226 which are bugs that cause the kernel to loop in kernel
227 mode for more than 20 seconds (configurable using the watchdog_thresh
228 sysctl), without giving other tasks a chance to run.
230 The panic can be used in combination with panic_timeout,
231 to cause the system to reboot automatically after a
232 lockup has been detected. This feature is useful for
233 high-availability systems that have uptime guarantees and
234 where a lockup must be resolved ASAP.
238 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
240 depends on LOCKUP_DETECTOR
242 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
243 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
246 bool "Panic on Oops" if EXPERT
249 Say Y here to enable the kernel to panic when it oopses. This
250 has the same effect as setting oops=panic on the kernel command
253 This feature is useful to ensure that the kernel does not do
254 anything erroneous after an oops which could result in data
255 corruption or other issues.
259 config PANIC_ON_OOPS_VALUE
262 default 0 if !PANIC_ON_OOPS
263 default 1 if PANIC_ON_OOPS
265 config DETECT_HUNG_TASK
266 bool "Detect Hung Tasks"
267 depends on DEBUG_KERNEL
268 default LOCKUP_DETECTOR
270 Say Y here to enable the kernel to detect "hung tasks",
271 which are bugs that cause the task to be stuck in
272 uninterruptible "D" state indefinitiley.
274 When a hung task is detected, the kernel will print the
275 current stack trace (which you should report), but the
276 task will stay in uninterruptible state. If lockdep is
277 enabled then all held locks will also be reported. This
278 feature has negligible overhead.
280 config DEFAULT_HUNG_TASK_TIMEOUT
281 int "Default timeout for hung task detection (in seconds)"
282 depends on DETECT_HUNG_TASK
285 This option controls the default timeout (in seconds) used
286 to determine when a task has become non-responsive and should
289 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
290 sysctl or by writing a value to
291 /proc/sys/kernel/hung_task_timeout_secs.
293 A timeout of 0 disables the check. The default is two minutes.
294 Keeping the default should be fine in most cases.
296 config BOOTPARAM_HUNG_TASK_PANIC
297 bool "Panic (Reboot) On Hung Tasks"
298 depends on DETECT_HUNG_TASK
300 Say Y here to enable the kernel to panic on "hung tasks",
301 which are bugs that cause the kernel to leave a task stuck
302 in uninterruptible "D" state.
304 The panic can be used in combination with panic_timeout,
305 to cause the system to reboot automatically after a
306 hung task has been detected. This feature is useful for
307 high-availability systems that have uptime guarantees and
308 where a hung tasks must be resolved ASAP.
312 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
314 depends on DETECT_HUNG_TASK
316 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
317 default 1 if BOOTPARAM_HUNG_TASK_PANIC
320 bool "Collect scheduler debugging info"
321 depends on DEBUG_KERNEL && PROC_FS
324 If you say Y here, the /proc/sched_debug file will be provided
325 that can help debug the scheduler. The runtime overhead of this
329 bool "Collect scheduler statistics"
330 depends on DEBUG_KERNEL && PROC_FS
332 If you say Y here, additional code will be inserted into the
333 scheduler and related routines to collect statistics about
334 scheduler behavior and provide them in /proc/schedstat. These
335 stats may be useful for both tuning and debugging the scheduler
336 If you aren't debugging the scheduler or trying to tune a specific
337 application, you can say N to avoid the very slight overhead
341 bool "Collect kernel timers statistics"
342 depends on DEBUG_KERNEL && PROC_FS
344 If you say Y here, additional code will be inserted into the
345 timer routines to collect statistics about kernel timers being
346 reprogrammed. The statistics can be read from /proc/timer_stats.
347 The statistics collection is started by writing 1 to /proc/timer_stats,
348 writing 0 stops it. This feature is useful to collect information
349 about timer usage patterns in kernel and userspace. This feature
350 is lightweight if enabled in the kernel config but not activated
351 (it defaults to deactivated on bootup and will only be activated
352 if some application like powertop activates it explicitly).
355 bool "Debug object operations"
356 depends on DEBUG_KERNEL
358 If you say Y here, additional code will be inserted into the
359 kernel to track the life time of various objects and validate
360 the operations on those objects.
362 config DEBUG_OBJECTS_SELFTEST
363 bool "Debug objects selftest"
364 depends on DEBUG_OBJECTS
366 This enables the selftest of the object debug code.
368 config DEBUG_OBJECTS_FREE
369 bool "Debug objects in freed memory"
370 depends on DEBUG_OBJECTS
372 This enables checks whether a k/v free operation frees an area
373 which contains an object which has not been deactivated
374 properly. This can make kmalloc/kfree-intensive workloads
377 config DEBUG_OBJECTS_TIMERS
378 bool "Debug timer objects"
379 depends on DEBUG_OBJECTS
381 If you say Y here, additional code will be inserted into the
382 timer routines to track the life time of timer objects and
383 validate the timer operations.
385 config DEBUG_OBJECTS_WORK
386 bool "Debug work objects"
387 depends on DEBUG_OBJECTS
389 If you say Y here, additional code will be inserted into the
390 work queue routines to track the life time of work objects and
391 validate the work operations.
393 config DEBUG_OBJECTS_RCU_HEAD
394 bool "Debug RCU callbacks objects"
395 depends on DEBUG_OBJECTS
397 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
399 config DEBUG_OBJECTS_PERCPU_COUNTER
400 bool "Debug percpu counter objects"
401 depends on DEBUG_OBJECTS
403 If you say Y here, additional code will be inserted into the
404 percpu counter routines to track the life time of percpu counter
405 objects and validate the percpu counter operations.
407 config DEBUG_OBJECTS_ENABLE_DEFAULT
408 int "debug_objects bootup default value (0-1)"
411 depends on DEBUG_OBJECTS
413 Debug objects boot parameter default value
416 bool "Debug slab memory allocations"
417 depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
419 Say Y here to have the kernel do limited verification on memory
420 allocation as well as poisoning memory on free to catch use of freed
421 memory. This can make kmalloc/kfree-intensive workloads much slower.
423 config DEBUG_SLAB_LEAK
424 bool "Memory leak debugging"
425 depends on DEBUG_SLAB
428 bool "SLUB debugging on by default"
429 depends on SLUB && SLUB_DEBUG && !KMEMCHECK
432 Boot with debugging on by default. SLUB boots by default with
433 the runtime debug capabilities switched off. Enabling this is
434 equivalent to specifying the "slub_debug" parameter on boot.
435 There is no support for more fine grained debug control like
436 possible with slub_debug=xxx. SLUB debugging may be switched
437 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
442 bool "Enable SLUB performance statistics"
443 depends on SLUB && SYSFS
445 SLUB statistics are useful to debug SLUBs allocation behavior in
446 order find ways to optimize the allocator. This should never be
447 enabled for production use since keeping statistics slows down
448 the allocator by a few percentage points. The slabinfo command
449 supports the determination of the most active slabs to figure
450 out which slabs are relevant to a particular load.
451 Try running: slabinfo -DA
453 config HAVE_DEBUG_KMEMLEAK
456 config DEBUG_KMEMLEAK
457 bool "Kernel memory leak detector"
458 depends on DEBUG_KERNEL && EXPERIMENTAL && HAVE_DEBUG_KMEMLEAK
460 select STACKTRACE if STACKTRACE_SUPPORT
464 Say Y here if you want to enable the memory leak
465 detector. The memory allocation/freeing is traced in a way
466 similar to the Boehm's conservative garbage collector, the
467 difference being that the orphan objects are not freed but
468 only shown in /sys/kernel/debug/kmemleak. Enabling this
469 feature will introduce an overhead to memory
470 allocations. See Documentation/kmemleak.txt for more
473 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
474 of finding leaks due to the slab objects poisoning.
476 In order to access the kmemleak file, debugfs needs to be
477 mounted (usually at /sys/kernel/debug).
479 config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
480 int "Maximum kmemleak early log entries"
481 depends on DEBUG_KMEMLEAK
485 Kmemleak must track all the memory allocations to avoid
486 reporting false positives. Since memory may be allocated or
487 freed before kmemleak is initialised, an early log buffer is
488 used to store these actions. If kmemleak reports "early log
489 buffer exceeded", please increase this value.
491 config DEBUG_KMEMLEAK_TEST
492 tristate "Simple test for the kernel memory leak detector"
493 depends on DEBUG_KMEMLEAK && m
495 This option enables a module that explicitly leaks memory.
499 config DEBUG_KMEMLEAK_DEFAULT_OFF
500 bool "Default kmemleak to off"
501 depends on DEBUG_KMEMLEAK
503 Say Y here to disable kmemleak by default. It can then be enabled
504 on the command line via kmemleak=on.
507 bool "Debug preemptible kernel"
508 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
511 If you say Y here then the kernel will use a debug variant of the
512 commonly used smp_processor_id() function and will print warnings
513 if kernel code uses it in a preemption-unsafe way. Also, the kernel
514 will detect preemption count underflows.
516 config DEBUG_RT_MUTEXES
517 bool "RT Mutex debugging, deadlock detection"
518 depends on DEBUG_KERNEL && RT_MUTEXES
520 This allows rt mutex semantics violations and rt mutex related
521 deadlocks (lockups) to be detected and reported automatically.
526 depends on DEBUG_RT_MUTEXES
528 config RT_MUTEX_TESTER
529 bool "Built-in scriptable tester for rt-mutexes"
530 depends on DEBUG_KERNEL && RT_MUTEXES
532 This option enables a rt-mutex tester.
534 config DEBUG_SPINLOCK
535 bool "Spinlock and rw-lock debugging: basic checks"
536 depends on DEBUG_KERNEL
537 select UNINLINE_SPIN_UNLOCK
539 Say Y here and build SMP to catch missing spinlock initialization
540 and certain other kinds of spinlock errors commonly made. This is
541 best used in conjunction with the NMI watchdog so that spinlock
542 deadlocks are also debuggable.
545 bool "Mutex debugging: basic checks"
546 depends on DEBUG_KERNEL
548 This feature allows mutex semantics violations to be detected and
551 config DEBUG_LOCK_ALLOC
552 bool "Lock debugging: detect incorrect freeing of live locks"
553 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
554 select DEBUG_SPINLOCK
558 This feature will check whether any held lock (spinlock, rwlock,
559 mutex or rwsem) is incorrectly freed by the kernel, via any of the
560 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
561 vfree(), etc.), whether a live lock is incorrectly reinitialized via
562 spin_lock_init()/mutex_init()/etc., or whether there is any lock
563 held during task exit.
566 bool "Lock debugging: prove locking correctness"
567 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
569 select DEBUG_SPINLOCK
571 select DEBUG_LOCK_ALLOC
572 select TRACE_IRQFLAGS
575 This feature enables the kernel to prove that all locking
576 that occurs in the kernel runtime is mathematically
577 correct: that under no circumstance could an arbitrary (and
578 not yet triggered) combination of observed locking
579 sequences (on an arbitrary number of CPUs, running an
580 arbitrary number of tasks and interrupt contexts) cause a
583 In short, this feature enables the kernel to report locking
584 related deadlocks before they actually occur.
586 The proof does not depend on how hard and complex a
587 deadlock scenario would be to trigger: how many
588 participant CPUs, tasks and irq-contexts would be needed
589 for it to trigger. The proof also does not depend on
590 timing: if a race and a resulting deadlock is possible
591 theoretically (no matter how unlikely the race scenario
592 is), it will be proven so and will immediately be
593 reported by the kernel (once the event is observed that
594 makes the deadlock theoretically possible).
596 If a deadlock is impossible (i.e. the locking rules, as
597 observed by the kernel, are mathematically correct), the
598 kernel reports nothing.
600 NOTE: this feature can also be enabled for rwlocks, mutexes
601 and rwsems - in which case all dependencies between these
602 different locking variants are observed and mapped too, and
603 the proof of observed correctness is also maintained for an
604 arbitrary combination of these separate locking variants.
606 For more details, see Documentation/lockdep-design.txt.
610 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
612 select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE
617 bool "Lock usage statistics"
618 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
620 select DEBUG_SPINLOCK
622 select DEBUG_LOCK_ALLOC
625 This feature enables tracking lock contention points
627 For more details, see Documentation/lockstat.txt
629 This also enables lock events required by "perf lock",
631 If you want to use "perf lock", you also need to turn on
632 CONFIG_EVENT_TRACING.
634 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
635 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
638 bool "Lock dependency engine debugging"
639 depends on DEBUG_KERNEL && LOCKDEP
641 If you say Y here, the lock dependency engine will do
642 additional runtime checks to debug itself, at the price
643 of more runtime overhead.
645 config TRACE_IRQFLAGS
648 Enables hooks to interrupt enabling and disabling for
649 either tracing or lock debugging.
651 config DEBUG_ATOMIC_SLEEP
652 bool "Sleep inside atomic section checking"
654 depends on DEBUG_KERNEL
656 If you say Y here, various routines which may sleep will become very
657 noisy if they are called inside atomic sections: when a spinlock is
658 held, inside an rcu read side critical section, inside preempt disabled
659 sections, inside an interrupt, etc...
661 config DEBUG_LOCKING_API_SELFTESTS
662 bool "Locking API boot-time self-tests"
663 depends on DEBUG_KERNEL
665 Say Y here if you want the kernel to run a short self-test during
666 bootup. The self-test checks whether common types of locking bugs
667 are detected by debugging mechanisms or not. (if you disable
668 lock debugging then those bugs wont be detected of course.)
669 The following locking APIs are covered: spinlocks, rwlocks,
674 depends on STACKTRACE_SUPPORT
676 config DEBUG_STACK_USAGE
677 bool "Stack utilization instrumentation"
678 depends on DEBUG_KERNEL && !IA64 && !PARISC
680 Enables the display of the minimum amount of free stack which each
681 task has ever had available in the sysrq-T and sysrq-P debug output.
683 This option will slow down process creation somewhat.
686 bool "kobject debugging"
687 depends on DEBUG_KERNEL
689 If you say Y here, some extra kobject debugging messages will be sent
693 bool "Highmem debugging"
694 depends on DEBUG_KERNEL && HIGHMEM
696 This options enables addition error checking for high memory systems.
697 Disable for production systems.
699 config HAVE_DEBUG_BUGVERBOSE
702 config DEBUG_BUGVERBOSE
703 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
704 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE)
707 Say Y here to make BUG() panics output the file name and line number
708 of the BUG call as well as the EIP and oops trace. This aids
709 debugging but costs about 70-100K of memory.
712 bool "Compile the kernel with debug info"
713 depends on DEBUG_KERNEL
715 If you say Y here the resulting kernel image will include
716 debugging info resulting in a larger kernel image.
717 This adds debug symbols to the kernel and modules (gcc -g), and
718 is needed if you intend to use kernel crashdump or binary object
719 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
720 Say Y here only if you plan to debug the kernel.
724 config DEBUG_INFO_REDUCED
725 bool "Reduce debugging information"
726 depends on DEBUG_INFO
728 If you say Y here gcc is instructed to generate less debugging
729 information for structure types. This means that tools that
730 need full debugging information (like kgdb or systemtap) won't
731 be happy. But if you merely need debugging information to
732 resolve line numbers there is no loss. Advantage is that
733 build directory object sizes shrink dramatically over a full
734 DEBUG_INFO build and compile times are reduced too.
735 Only works with newer gcc versions.
739 depends on DEBUG_KERNEL
741 Enable this to turn on extended checks in the virtual-memory system
742 that may impact performance.
747 bool "Debug VM red-black trees"
750 Enable this to turn on more extended checks in the virtual-memory
751 system that may impact performance.
756 bool "Debug VM translations"
757 depends on DEBUG_KERNEL && X86
759 Enable some costly sanity checks in virtual to page code. This can
760 catch mistakes with virt_to_page() and friends.
764 config DEBUG_NOMMU_REGIONS
765 bool "Debug the global anon/private NOMMU mapping region tree"
766 depends on DEBUG_KERNEL && !MMU
768 This option causes the global tree of anonymous and private mapping
769 regions to be regularly checked for invalid topology.
771 config DEBUG_WRITECOUNT
772 bool "Debug filesystem writers count"
773 depends on DEBUG_KERNEL
775 Enable this to catch wrong use of the writers count in struct
776 vfsmount. This will increase the size of each file struct by
781 config DEBUG_MEMORY_INIT
782 bool "Debug memory initialisation" if EXPERT
785 Enable this for additional checks during memory initialisation.
786 The sanity checks verify aspects of the VM such as the memory model
787 and other information provided by the architecture. Verbose
788 information will be printed at KERN_DEBUG loglevel depending
789 on the mminit_loglevel= command-line option.
794 bool "Debug linked list manipulation"
795 depends on DEBUG_KERNEL
797 Enable this to turn on extended checks in the linked-list
802 config TEST_LIST_SORT
803 bool "Linked list sorting test"
804 depends on DEBUG_KERNEL
806 Enable this to turn on 'list_sort()' function test. This test is
807 executed only once during system boot, so affects only boot time.
812 bool "Debug SG table operations"
813 depends on DEBUG_KERNEL
815 Enable this to turn on checks on scatter-gather tables. This can
816 help find problems with drivers that do not properly initialize
821 config DEBUG_NOTIFIERS
822 bool "Debug notifier call chains"
823 depends on DEBUG_KERNEL
825 Enable this to turn on sanity checking for notifier call chains.
826 This is most useful for kernel developers to make sure that
827 modules properly unregister themselves from notifier chains.
828 This is a relatively cheap check but if you care about maximum
831 config DEBUG_CREDENTIALS
832 bool "Debug credential management"
833 depends on DEBUG_KERNEL
835 Enable this to turn on some debug checking for credential
836 management. The additional code keeps track of the number of
837 pointers from task_structs to any given cred struct, and checks to
838 see that this number never exceeds the usage count of the cred
841 Furthermore, if SELinux is enabled, this also checks that the
842 security pointer in the cred struct is never seen to be invalid.
847 # Select this config option from the architecture Kconfig, if it
848 # is preferred to always offer frame pointers as a config
849 # option on the architecture (regardless of KERNEL_DEBUG):
851 config ARCH_WANT_FRAME_POINTERS
856 bool "Compile the kernel with frame pointers"
857 depends on DEBUG_KERNEL && \
858 (CRIS || M68K || FRV || UML || \
859 AVR32 || SUPERH || BLACKFIN || MN10300) || \
860 ARCH_WANT_FRAME_POINTERS
861 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
863 If you say Y here the resulting kernel image will be slightly
864 larger and slower, but it gives very useful debugging information
865 in case of kernel bugs. (precise oopses/stacktraces/warnings)
867 config BOOT_PRINTK_DELAY
868 bool "Delay each boot printk message by N milliseconds"
869 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
871 This build option allows you to read kernel boot messages
872 by inserting a short delay after each one. The delay is
873 specified in milliseconds on the kernel command line,
874 using "boot_delay=N".
876 It is likely that you would also need to use "lpj=M" to preset
877 the "loops per jiffie" value.
878 See a previous boot log for the "lpj" value to use for your
879 system, and then set "lpj=M" before setting "boot_delay=N".
880 NOTE: Using this option may adversely affect SMP systems.
881 I.e., processors other than the first one may not boot up.
882 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
883 what it believes to be lockup conditions.
888 bool "RCU debugging: prove RCU correctness"
889 depends on PROVE_LOCKING
892 This feature enables lockdep extensions that check for correct
893 use of RCU APIs. This is currently under development. Say Y
894 if you want to debug RCU usage or help work on the PROVE_RCU
897 Say N if you are unsure.
899 config PROVE_RCU_REPEATEDLY
900 bool "RCU debugging: don't disable PROVE_RCU on first splat"
904 By itself, PROVE_RCU will disable checking upon issuing the
905 first warning (or "splat"). This feature prevents such
906 disabling, allowing multiple RCU-lockdep warnings to be printed
909 Say Y to allow multiple RCU-lockdep warnings per boot.
911 Say N if you are unsure.
913 config PROVE_RCU_DELAY
914 bool "RCU debugging: preemptible RCU race provocation"
915 depends on DEBUG_KERNEL && PREEMPT_RCU
918 There is a class of races that involve an unlikely preemption
919 of __rcu_read_unlock() just after ->rcu_read_lock_nesting has
920 been set to INT_MIN. This feature inserts a delay at that
921 point to increase the probability of these races.
923 Say Y to increase probability of preemption of __rcu_read_unlock().
925 Say N if you are unsure.
927 config SPARSE_RCU_POINTER
928 bool "RCU debugging: sparse-based checks for pointer usage"
931 This feature enables the __rcu sparse annotation for
932 RCU-protected pointers. This annotation will cause sparse
933 to flag any non-RCU used of annotated pointers. This can be
934 helpful when debugging RCU usage. Please note that this feature
935 is not intended to enforce code cleanliness; it is instead merely
938 Say Y to make sparse flag questionable use of RCU-protected pointers
940 Say N if you are unsure.
942 config RCU_TORTURE_TEST
943 tristate "torture tests for RCU"
944 depends on DEBUG_KERNEL
947 This option provides a kernel module that runs torture tests
948 on the RCU infrastructure. The kernel module may be built
949 after the fact on the running kernel to be tested, if desired.
951 Say Y here if you want RCU torture tests to be built into
953 Say M if you want the RCU torture tests to build as a module.
954 Say N if you are unsure.
956 config RCU_TORTURE_TEST_RUNNABLE
957 bool "torture tests for RCU runnable by default"
958 depends on RCU_TORTURE_TEST = y
961 This option provides a way to build the RCU torture tests
962 directly into the kernel without them starting up at boot
963 time. You can use /proc/sys/kernel/rcutorture_runnable
964 to manually override this setting. This /proc file is
965 available only when the RCU torture tests have been built
968 Say Y here if you want the RCU torture tests to start during
969 boot (you probably don't).
970 Say N here if you want the RCU torture tests to start only
971 after being manually enabled via /proc.
973 config RCU_CPU_STALL_TIMEOUT
974 int "RCU CPU stall timeout in seconds"
975 depends on RCU_STALL_COMMON
979 If a given RCU grace period extends more than the specified
980 number of seconds, a CPU stall warning is printed. If the
981 RCU grace period persists, additional CPU stall warnings are
982 printed at more widely spaced intervals.
984 config RCU_CPU_STALL_VERBOSE
985 bool "Print additional per-task information for RCU_CPU_STALL_DETECTOR"
986 depends on TREE_PREEMPT_RCU
989 This option causes RCU to printk detailed per-task information
990 for any tasks that are stalling the current RCU grace period.
992 Say N if you are unsure.
994 Say Y if you want to enable such checks.
996 config RCU_CPU_STALL_INFO
997 bool "Print additional diagnostics on RCU CPU stall"
998 depends on (TREE_RCU || TREE_PREEMPT_RCU) && DEBUG_KERNEL
1001 For each stalled CPU that is aware of the current RCU grace
1002 period, print out additional per-CPU diagnostic information
1003 regarding scheduling-clock ticks, idle state, and,
1004 for RCU_FAST_NO_HZ kernels, idle-entry state.
1006 Say N if you are unsure.
1008 Say Y if you want to enable such diagnostics.
1011 bool "Enable tracing for RCU"
1012 depends on DEBUG_KERNEL
1015 This option provides tracing in RCU which presents stats
1016 in debugfs for debugging RCU implementation.
1018 Say Y here if you want to enable RCU tracing
1019 Say N if you are unsure.
1021 endmenu # "RCU Debugging"
1023 config KPROBES_SANITY_TEST
1024 bool "Kprobes sanity tests"
1025 depends on DEBUG_KERNEL
1029 This option provides for testing basic kprobes functionality on
1030 boot. A sample kprobe, jprobe and kretprobe are inserted and
1031 verified for functionality.
1033 Say N if you are unsure.
1035 config BACKTRACE_SELF_TEST
1036 tristate "Self test for the backtrace code"
1037 depends on DEBUG_KERNEL
1040 This option provides a kernel module that can be used to test
1041 the kernel stack backtrace code. This option is not useful
1042 for distributions or general kernels, but only for kernel
1043 developers working on architecture code.
1045 Note that if you want to also test saved backtraces, you will
1046 have to enable STACKTRACE as well.
1048 Say N if you are unsure.
1050 config DEBUG_BLOCK_EXT_DEVT
1051 bool "Force extended block device numbers and spread them"
1052 depends on DEBUG_KERNEL
1056 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1057 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1058 YOU ARE DOING. Distros, please enable this and fix whatever
1061 Conventionally, block device numbers are allocated from
1062 predetermined contiguous area. However, extended block area
1063 may introduce non-contiguous block device numbers. This
1064 option forces most block device numbers to be allocated from
1065 the extended space and spreads them to discover kernel or
1066 userland code paths which assume predetermined contiguous
1067 device number allocation.
1069 Note that turning on this debug option shuffles all the
1070 device numbers for all IDE and SCSI devices including libata
1071 ones, so root partition specified using device number
1072 directly (via rdev or root=MAJ:MIN) won't work anymore.
1073 Textual device names (root=/dev/sdXn) will continue to work.
1075 Say N if you are unsure.
1077 config DEBUG_FORCE_WEAK_PER_CPU
1078 bool "Force weak per-cpu definitions"
1079 depends on DEBUG_KERNEL
1081 s390 and alpha require percpu variables in modules to be
1082 defined weak to work around addressing range issue which
1083 puts the following two restrictions on percpu variable
1086 1. percpu symbols must be unique whether static or not
1087 2. percpu variables can't be defined inside a function
1089 To ensure that generic code follows the above rules, this
1090 option forces all percpu variables to be defined as weak.
1092 config DEBUG_PER_CPU_MAPS
1093 bool "Debug access to per_cpu maps"
1094 depends on DEBUG_KERNEL
1097 Say Y to verify that the per_cpu map being accessed has
1098 been set up. This adds a fair amount of code to kernel memory
1099 and decreases performance.
1104 tristate "Linux Kernel Dump Test Tool Module"
1109 This module enables testing of the different dumping mechanisms by
1110 inducing system failures at predefined crash points.
1111 If you don't need it: say N
1112 Choose M here to compile this code as a module. The module will be
1115 Documentation on how to use the module can be found in
1116 Documentation/fault-injection/provoke-crashes.txt
1118 config NOTIFIER_ERROR_INJECTION
1119 tristate "Notifier error injection"
1120 depends on DEBUG_KERNEL
1123 This option provides the ability to inject artificial errors to
1124 specified notifier chain callbacks. It is useful to test the error
1125 handling of notifier call chain failures.
1129 config CPU_NOTIFIER_ERROR_INJECT
1130 tristate "CPU notifier error injection module"
1131 depends on HOTPLUG_CPU && NOTIFIER_ERROR_INJECTION
1133 This option provides a kernel module that can be used to test
1134 the error handling of the cpu notifiers by injecting artificial
1135 errors to CPU notifier chain callbacks. It is controlled through
1136 debugfs interface under /sys/kernel/debug/notifier-error-inject/cpu
1138 If the notifier call chain should be failed with some events
1139 notified, write the error code to "actions/<notifier event>/error".
1141 Example: Inject CPU offline error (-1 == -EPERM)
1143 # cd /sys/kernel/debug/notifier-error-inject/cpu
1144 # echo -1 > actions/CPU_DOWN_PREPARE/error
1145 # echo 0 > /sys/devices/system/cpu/cpu1/online
1146 bash: echo: write error: Operation not permitted
1148 To compile this code as a module, choose M here: the module will
1149 be called cpu-notifier-error-inject.
1153 config PM_NOTIFIER_ERROR_INJECT
1154 tristate "PM notifier error injection module"
1155 depends on PM && NOTIFIER_ERROR_INJECTION
1156 default m if PM_DEBUG
1158 This option provides the ability to inject artificial errors to
1159 PM notifier chain callbacks. It is controlled through debugfs
1160 interface /sys/kernel/debug/notifier-error-inject/pm
1162 If the notifier call chain should be failed with some events
1163 notified, write the error code to "actions/<notifier event>/error".
1165 Example: Inject PM suspend error (-12 = -ENOMEM)
1167 # cd /sys/kernel/debug/notifier-error-inject/pm/
1168 # echo -12 > actions/PM_SUSPEND_PREPARE/error
1169 # echo mem > /sys/power/state
1170 bash: echo: write error: Cannot allocate memory
1172 To compile this code as a module, choose M here: the module will
1173 be called pm-notifier-error-inject.
1177 config MEMORY_NOTIFIER_ERROR_INJECT
1178 tristate "Memory hotplug notifier error injection module"
1179 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION
1181 This option provides the ability to inject artificial errors to
1182 memory hotplug notifier chain callbacks. It is controlled through
1183 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory
1185 If the notifier call chain should be failed with some events
1186 notified, write the error code to "actions/<notifier event>/error".
1188 Example: Inject memory hotplug offline error (-12 == -ENOMEM)
1190 # cd /sys/kernel/debug/notifier-error-inject/memory
1191 # echo -12 > actions/MEM_GOING_OFFLINE/error
1192 # echo offline > /sys/devices/system/memory/memoryXXX/state
1193 bash: echo: write error: Cannot allocate memory
1195 To compile this code as a module, choose M here: the module will
1196 be called pSeries-reconfig-notifier-error-inject.
1200 config OF_RECONFIG_NOTIFIER_ERROR_INJECT
1201 tristate "OF reconfig notifier error injection module"
1202 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION
1204 This option provides the ability to inject artificial errors to
1205 OF reconfig notifier chain callbacks. It is controlled
1206 through debugfs interface under
1207 /sys/kernel/debug/notifier-error-inject/OF-reconfig/
1209 If the notifier call chain should be failed with some events
1210 notified, write the error code to "actions/<notifier event>/error".
1212 To compile this code as a module, choose M here: the module will
1213 be called memory-notifier-error-inject.
1217 config FAULT_INJECTION
1218 bool "Fault-injection framework"
1219 depends on DEBUG_KERNEL
1221 Provide fault-injection framework.
1222 For more details, see Documentation/fault-injection/.
1225 bool "Fault-injection capability for kmalloc"
1226 depends on FAULT_INJECTION
1227 depends on SLAB || SLUB
1229 Provide fault-injection capability for kmalloc.
1231 config FAIL_PAGE_ALLOC
1232 bool "Fault-injection capabilitiy for alloc_pages()"
1233 depends on FAULT_INJECTION
1235 Provide fault-injection capability for alloc_pages().
1237 config FAIL_MAKE_REQUEST
1238 bool "Fault-injection capability for disk IO"
1239 depends on FAULT_INJECTION && BLOCK
1241 Provide fault-injection capability for disk IO.
1243 config FAIL_IO_TIMEOUT
1244 bool "Fault-injection capability for faking disk interrupts"
1245 depends on FAULT_INJECTION && BLOCK
1247 Provide fault-injection capability on end IO handling. This
1248 will make the block layer "forget" an interrupt as configured,
1249 thus exercising the error handling.
1251 Only works with drivers that use the generic timeout handling,
1252 for others it wont do anything.
1254 config FAIL_MMC_REQUEST
1255 bool "Fault-injection capability for MMC IO"
1257 depends on FAULT_INJECTION && MMC
1259 Provide fault-injection capability for MMC IO.
1260 This will make the mmc core return data errors. This is
1261 useful to test the error handling in the mmc block device
1262 and to test how the mmc host driver handles retries from
1265 config FAULT_INJECTION_DEBUG_FS
1266 bool "Debugfs entries for fault-injection capabilities"
1267 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1269 Enable configuration of fault-injection capabilities via debugfs.
1271 config FAULT_INJECTION_STACKTRACE_FILTER
1272 bool "stacktrace filter for fault-injection capabilities"
1273 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1276 select FRAME_POINTER if !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND
1278 Provide stacktrace filter for fault-injection capabilities
1281 bool "Latency measuring infrastructure"
1282 depends on HAVE_LATENCYTOP_SUPPORT
1283 depends on DEBUG_KERNEL
1284 depends on STACKTRACE_SUPPORT
1286 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND
1293 Enable this option if you want to use the LatencyTOP tool
1294 to find out which userspace is blocking on what kernel operations.
1296 source mm/Kconfig.debug
1297 source kernel/trace/Kconfig
1300 tristate "Red-Black tree test"
1301 depends on m && DEBUG_KERNEL
1303 A benchmark measuring the performance of the rbtree library.
1304 Also includes rbtree invariant checks.
1306 config INTERVAL_TREE_TEST
1307 tristate "Interval tree test"
1308 depends on m && DEBUG_KERNEL
1310 A benchmark measuring the performance of the interval tree library
1312 config PROVIDE_OHCI1394_DMA_INIT
1313 bool "Remote debugging over FireWire early on boot"
1314 depends on PCI && X86
1316 If you want to debug problems which hang or crash the kernel early
1317 on boot and the crashing machine has a FireWire port, you can use
1318 this feature to remotely access the memory of the crashed machine
1319 over FireWire. This employs remote DMA as part of the OHCI1394
1320 specification which is now the standard for FireWire controllers.
1322 With remote DMA, you can monitor the printk buffer remotely using
1323 firescope and access all memory below 4GB using fireproxy from gdb.
1324 Even controlling a kernel debugger is possible using remote DMA.
1328 If ohci1394_dma=early is used as boot parameter, it will initialize
1329 all OHCI1394 controllers which are found in the PCI config space.
1331 As all changes to the FireWire bus such as enabling and disabling
1332 devices cause a bus reset and thereby disable remote DMA for all
1333 devices, be sure to have the cable plugged and FireWire enabled on
1334 the debugging host before booting the debug target for debugging.
1336 This code (~1k) is freed after boot. By then, the firewire stack
1337 in charge of the OHCI-1394 controllers should be used instead.
1339 See Documentation/debugging-via-ohci1394.txt for more information.
1341 config FIREWIRE_OHCI_REMOTE_DMA
1342 bool "Remote debugging over FireWire with firewire-ohci"
1343 depends on FIREWIRE_OHCI
1345 This option lets you use the FireWire bus for remote debugging
1346 with help of the firewire-ohci driver. It enables unfiltered
1347 remote DMA in firewire-ohci.
1348 See Documentation/debugging-via-ohci1394.txt for more information.
1353 bool "Build targets in Documentation/ tree"
1354 depends on HEADERS_CHECK
1356 This option attempts to build objects from the source files in the
1357 kernel Documentation/ tree.
1359 Say N if you are unsure.
1361 config DYNAMIC_DEBUG
1362 bool "Enable dynamic printk() support"
1368 Compiles debug level messages into the kernel, which would not
1369 otherwise be available at runtime. These messages can then be
1370 enabled/disabled based on various levels of scope - per source file,
1371 function, module, format string, and line number. This mechanism
1372 implicitly compiles in all pr_debug() and dev_dbg() calls, which
1373 enlarges the kernel text size by about 2%.
1375 If a source file is compiled with DEBUG flag set, any
1376 pr_debug() calls in it are enabled by default, but can be
1377 disabled at runtime as below. Note that DEBUG flag is
1378 turned on by many CONFIG_*DEBUG* options.
1382 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
1383 which is contained in the 'debugfs' filesystem. Thus, the debugfs
1384 filesystem must first be mounted before making use of this feature.
1385 We refer the control file as: <debugfs>/dynamic_debug/control. This
1386 file contains a list of the debug statements that can be enabled. The
1387 format for each line of the file is:
1389 filename:lineno [module]function flags format
1391 filename : source file of the debug statement
1392 lineno : line number of the debug statement
1393 module : module that contains the debug statement
1394 function : function that contains the debug statement
1395 flags : '=p' means the line is turned 'on' for printing
1396 format : the format used for the debug statement
1400 nullarbor:~ # cat <debugfs>/dynamic_debug/control
1401 # filename:lineno [module]function flags format
1402 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012"
1403 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012"
1404 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012"
1408 // enable the message at line 1603 of file svcsock.c
1409 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
1410 <debugfs>/dynamic_debug/control
1412 // enable all the messages in file svcsock.c
1413 nullarbor:~ # echo -n 'file svcsock.c +p' >
1414 <debugfs>/dynamic_debug/control
1416 // enable all the messages in the NFS server module
1417 nullarbor:~ # echo -n 'module nfsd +p' >
1418 <debugfs>/dynamic_debug/control
1420 // enable all 12 messages in the function svc_process()
1421 nullarbor:~ # echo -n 'func svc_process +p' >
1422 <debugfs>/dynamic_debug/control
1424 // disable all 12 messages in the function svc_process()
1425 nullarbor:~ # echo -n 'func svc_process -p' >
1426 <debugfs>/dynamic_debug/control
1428 See Documentation/dynamic-debug-howto.txt for additional information.
1430 config DMA_API_DEBUG
1431 bool "Enable debugging of DMA-API usage"
1432 depends on HAVE_DMA_API_DEBUG
1434 Enable this option to debug the use of the DMA API by device drivers.
1435 With this option you will be able to detect common bugs in device
1436 drivers like double-freeing of DMA mappings or freeing mappings that
1437 were never allocated.
1438 This option causes a performance degredation. Use only if you want
1439 to debug device drivers. If unsure, say N.
1441 config ATOMIC64_SELFTEST
1442 bool "Perform an atomic64_t self-test at boot"
1444 Enable this option to test the atomic64_t functions at boot.
1448 config ASYNC_RAID6_TEST
1449 tristate "Self test for hardware accelerated raid6 recovery"
1450 depends on ASYNC_RAID6_RECOV
1453 This is a one-shot self test that permutes through the
1454 recovery of all the possible two disk failure scenarios for a
1455 N-disk array. Recovery is performed with the asynchronous
1456 raid6 recovery routines, and will optionally use an offload
1457 engine if one is available.
1461 source "samples/Kconfig"
1463 source "lib/Kconfig.kgdb"
1465 source "lib/Kconfig.kmemcheck"
1468 tristate "Test kstrto*() family of functions at runtime"