3 bool "Show timing information on printks"
6 Selecting this option causes timing information to be
7 included in printk output. This allows you to measure
8 the interval between kernel operations, including bootup
9 operations. This is useful for identifying long delays
10 in kernel startup. Or add printk.time=1 at boot-time.
11 See Documentation/kernel-parameters.txt
13 config DEFAULT_MESSAGE_LOGLEVEL
14 int "Default message log level (1-7)"
18 Default log level for printk statements with no specified priority.
20 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
21 that are auditing their logs closely may want to set it to a lower
24 config ENABLE_WARN_DEPRECATED
25 bool "Enable __deprecated logic"
28 Enable the __deprecated logic in the kernel build.
29 Disable this to suppress the "warning: 'foo' is deprecated
30 (declared at kernel/power/somefile.c:1234)" messages.
32 config ENABLE_MUST_CHECK
33 bool "Enable __must_check logic"
36 Enable the __must_check logic in the kernel build. Disable this to
37 suppress the "warning: ignoring return value of 'foo', declared with
38 attribute warn_unused_result" messages.
41 int "Warn for stack frames larger than (needs gcc 4.4)"
43 default 1024 if !64BIT
46 Tell gcc to warn at build time for stack frames larger than this.
47 Setting this too low will cause a lot of warnings.
48 Setting it to 0 disables the warning.
52 bool "Magic SysRq key"
55 If you say Y here, you will have some control over the system even
56 if the system crashes for example during kernel debugging (e.g., you
57 will be able to flush the buffer cache to disk, reboot the system
58 immediately or dump some status information). This is accomplished
59 by pressing various keys while holding SysRq (Alt+PrintScreen). It
60 also works on a serial console (on PC hardware at least), if you
61 send a BREAK and then within 5 seconds a command keypress. The
62 keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
63 unless you really know what this hack does.
66 bool "Strip assembler-generated symbols during link"
69 Strip internal assembler-generated symbols during a link (symbols
70 that look like '.Lxxx') so they don't pollute the output of
71 get_wchan() and suchlike.
74 bool "Enable unused/obsolete exported symbols"
77 Unused but exported symbols make the kernel needlessly bigger. For
78 that reason most of these unused exports will soon be removed. This
79 option is provided temporarily to provide a transition period in case
80 some external kernel module needs one of these symbols anyway. If you
81 encounter such a case in your module, consider if you are actually
82 using the right API. (rationale: since nobody in the kernel is using
83 this in a module, there is a pretty good chance it's actually the
84 wrong interface to use). If you really need the symbol, please send a
85 mail to the linux kernel mailing list mentioning the symbol and why
86 you really need it, and what the merge plan to the mainline kernel for
90 bool "Debug Filesystem"
92 debugfs is a virtual file system that kernel developers use to put
93 debugging files into. Enable this option to be able to read and
96 For detailed documentation on the debugfs API, see
97 Documentation/DocBook/filesystems.
102 bool "Run 'make headers_check' when building vmlinux"
105 This option will extract the user-visible kernel headers whenever
106 building the kernel, and will run basic sanity checks on them to
107 ensure that exported files do not attempt to include files which
108 were not exported, etc.
110 If you're making modifications to header files which are
111 relevant for userspace, say 'Y', and check the headers
112 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
113 your build tree), to make sure they're suitable.
115 config DEBUG_SECTION_MISMATCH
116 bool "Enable full Section mismatch analysis"
118 The section mismatch analysis checks if there are illegal
119 references from one section to another section.
120 During linktime or runtime, some sections are dropped;
121 any use of code/data previously in these sections would
122 most likely result in an oops.
123 In the code, functions and variables are annotated with
124 __init, __devinit, etc. (see the full list in include/linux/init.h),
125 which results in the code/data being placed in specific sections.
126 The section mismatch analysis is always performed after a full
127 kernel build, and enabling this option causes the following
128 additional steps to occur:
129 - Add the option -fno-inline-functions-called-once to gcc commands.
130 When inlining a function annotated with __init in a non-init
131 function, we would lose the section information and thus
132 the analysis would not catch the illegal reference.
133 This option tells gcc to inline less (but it does result in
135 - Run the section mismatch analysis for each module/built-in.o file.
136 When we run the section mismatch analysis on vmlinux.o, we
137 lose valueble information about where the mismatch was
139 Running the analysis for each module/built-in.o file
140 tells where the mismatch happens much closer to the
141 source. The drawback is that the same mismatch is
142 reported at least twice.
143 - Enable verbose reporting from modpost in order to help resolve
144 the section mismatches that are reported.
147 bool "Kernel debugging"
149 Say Y here if you are developing drivers or trying to debug and
150 identify kernel problems.
153 bool "Debug shared IRQ handlers"
154 depends on DEBUG_KERNEL && GENERIC_HARDIRQS
156 Enable this to generate a spurious interrupt as soon as a shared
157 interrupt handler is registered, and just before one is deregistered.
158 Drivers ought to be able to handle interrupts coming in at those
159 points; some don't and need to be caught.
161 config LOCKUP_DETECTOR
162 bool "Detect Hard and Soft Lockups"
163 depends on DEBUG_KERNEL && !S390
165 Say Y here to enable the kernel to act as a watchdog to detect
166 hard and soft lockups.
168 Softlockups are bugs that cause the kernel to loop in kernel
169 mode for more than 20 seconds, without giving other tasks a
170 chance to run. The current stack trace is displayed upon
171 detection and the system will stay locked up.
173 Hardlockups are bugs that cause the CPU to loop in kernel mode
174 for more than 10 seconds, without letting other interrupts have a
175 chance to run. The current stack trace is displayed upon detection
176 and the system will stay locked up.
178 The overhead should be minimal. A periodic hrtimer runs to
179 generate interrupts and kick the watchdog task every 4 seconds.
180 An NMI is generated every 10 seconds or so to check for hardlockups.
182 The frequency of hrtimer and NMI events and the soft and hard lockup
183 thresholds can be controlled through the sysctl watchdog_thresh.
185 config HARDLOCKUP_DETECTOR
186 def_bool LOCKUP_DETECTOR && PERF_EVENTS && HAVE_PERF_EVENTS_NMI && \
189 config BOOTPARAM_HARDLOCKUP_PANIC
190 bool "Panic (Reboot) On Hard Lockups"
191 depends on LOCKUP_DETECTOR
193 Say Y here to enable the kernel to panic on "hard lockups",
194 which are bugs that cause the kernel to loop in kernel
195 mode with interrupts disabled for more than 10 seconds (configurable
196 using the watchdog_thresh sysctl).
200 config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
202 depends on LOCKUP_DETECTOR
204 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
205 default 1 if BOOTPARAM_HARDLOCKUP_PANIC
207 config BOOTPARAM_SOFTLOCKUP_PANIC
208 bool "Panic (Reboot) On Soft Lockups"
209 depends on LOCKUP_DETECTOR
211 Say Y here to enable the kernel to panic on "soft lockups",
212 which are bugs that cause the kernel to loop in kernel
213 mode for more than 20 seconds (configurable using the watchdog_thresh
214 sysctl), without giving other tasks a chance to run.
216 The panic can be used in combination with panic_timeout,
217 to cause the system to reboot automatically after a
218 lockup has been detected. This feature is useful for
219 high-availability systems that have uptime guarantees and
220 where a lockup must be resolved ASAP.
224 config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
226 depends on LOCKUP_DETECTOR
228 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
229 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
231 config DETECT_HUNG_TASK
232 bool "Detect Hung Tasks"
233 depends on DEBUG_KERNEL
234 default LOCKUP_DETECTOR
236 Say Y here to enable the kernel to detect "hung tasks",
237 which are bugs that cause the task to be stuck in
238 uninterruptible "D" state indefinitiley.
240 When a hung task is detected, the kernel will print the
241 current stack trace (which you should report), but the
242 task will stay in uninterruptible state. If lockdep is
243 enabled then all held locks will also be reported. This
244 feature has negligible overhead.
246 config DEFAULT_HUNG_TASK_TIMEOUT
247 int "Default timeout for hung task detection (in seconds)"
248 depends on DETECT_HUNG_TASK
251 This option controls the default timeout (in seconds) used
252 to determine when a task has become non-responsive and should
255 It can be adjusted at runtime via the kernel.hung_task_timeout_secs
256 sysctl or by writing a value to
257 /proc/sys/kernel/hung_task_timeout_secs.
259 A timeout of 0 disables the check. The default is two minutes.
260 Keeping the default should be fine in most cases.
262 config BOOTPARAM_HUNG_TASK_PANIC
263 bool "Panic (Reboot) On Hung Tasks"
264 depends on DETECT_HUNG_TASK
266 Say Y here to enable the kernel to panic on "hung tasks",
267 which are bugs that cause the kernel to leave a task stuck
268 in uninterruptible "D" state.
270 The panic can be used in combination with panic_timeout,
271 to cause the system to reboot automatically after a
272 hung task has been detected. This feature is useful for
273 high-availability systems that have uptime guarantees and
274 where a hung tasks must be resolved ASAP.
278 config BOOTPARAM_HUNG_TASK_PANIC_VALUE
280 depends on DETECT_HUNG_TASK
282 default 0 if !BOOTPARAM_HUNG_TASK_PANIC
283 default 1 if BOOTPARAM_HUNG_TASK_PANIC
286 bool "Collect scheduler debugging info"
287 depends on DEBUG_KERNEL && PROC_FS
290 If you say Y here, the /proc/sched_debug file will be provided
291 that can help debug the scheduler. The runtime overhead of this
295 bool "Collect scheduler statistics"
296 depends on DEBUG_KERNEL && PROC_FS
298 If you say Y here, additional code will be inserted into the
299 scheduler and related routines to collect statistics about
300 scheduler behavior and provide them in /proc/schedstat. These
301 stats may be useful for both tuning and debugging the scheduler
302 If you aren't debugging the scheduler or trying to tune a specific
303 application, you can say N to avoid the very slight overhead
307 bool "Collect kernel timers statistics"
308 depends on DEBUG_KERNEL && PROC_FS
310 If you say Y here, additional code will be inserted into the
311 timer routines to collect statistics about kernel timers being
312 reprogrammed. The statistics can be read from /proc/timer_stats.
313 The statistics collection is started by writing 1 to /proc/timer_stats,
314 writing 0 stops it. This feature is useful to collect information
315 about timer usage patterns in kernel and userspace. This feature
316 is lightweight if enabled in the kernel config but not activated
317 (it defaults to deactivated on bootup and will only be activated
318 if some application like powertop activates it explicitly).
321 bool "Debug object operations"
322 depends on DEBUG_KERNEL
324 If you say Y here, additional code will be inserted into the
325 kernel to track the life time of various objects and validate
326 the operations on those objects.
328 config DEBUG_OBJECTS_SELFTEST
329 bool "Debug objects selftest"
330 depends on DEBUG_OBJECTS
332 This enables the selftest of the object debug code.
334 config DEBUG_OBJECTS_FREE
335 bool "Debug objects in freed memory"
336 depends on DEBUG_OBJECTS
338 This enables checks whether a k/v free operation frees an area
339 which contains an object which has not been deactivated
340 properly. This can make kmalloc/kfree-intensive workloads
343 config DEBUG_OBJECTS_TIMERS
344 bool "Debug timer objects"
345 depends on DEBUG_OBJECTS
347 If you say Y here, additional code will be inserted into the
348 timer routines to track the life time of timer objects and
349 validate the timer operations.
351 config DEBUG_OBJECTS_WORK
352 bool "Debug work objects"
353 depends on DEBUG_OBJECTS
355 If you say Y here, additional code will be inserted into the
356 work queue routines to track the life time of work objects and
357 validate the work operations.
359 config DEBUG_OBJECTS_RCU_HEAD
360 bool "Debug RCU callbacks objects"
361 depends on DEBUG_OBJECTS
363 Enable this to turn on debugging of RCU list heads (call_rcu() usage).
365 config DEBUG_OBJECTS_PERCPU_COUNTER
366 bool "Debug percpu counter objects"
367 depends on DEBUG_OBJECTS
369 If you say Y here, additional code will be inserted into the
370 percpu counter routines to track the life time of percpu counter
371 objects and validate the percpu counter operations.
373 config DEBUG_OBJECTS_ENABLE_DEFAULT
374 int "debug_objects bootup default value (0-1)"
377 depends on DEBUG_OBJECTS
379 Debug objects boot parameter default value
382 bool "Debug slab memory allocations"
383 depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
385 Say Y here to have the kernel do limited verification on memory
386 allocation as well as poisoning memory on free to catch use of freed
387 memory. This can make kmalloc/kfree-intensive workloads much slower.
389 config DEBUG_SLAB_LEAK
390 bool "Memory leak debugging"
391 depends on DEBUG_SLAB
394 bool "SLUB debugging on by default"
395 depends on SLUB && SLUB_DEBUG && !KMEMCHECK
398 Boot with debugging on by default. SLUB boots by default with
399 the runtime debug capabilities switched off. Enabling this is
400 equivalent to specifying the "slub_debug" parameter on boot.
401 There is no support for more fine grained debug control like
402 possible with slub_debug=xxx. SLUB debugging may be switched
403 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
408 bool "Enable SLUB performance statistics"
409 depends on SLUB && SYSFS
411 SLUB statistics are useful to debug SLUBs allocation behavior in
412 order find ways to optimize the allocator. This should never be
413 enabled for production use since keeping statistics slows down
414 the allocator by a few percentage points. The slabinfo command
415 supports the determination of the most active slabs to figure
416 out which slabs are relevant to a particular load.
417 Try running: slabinfo -DA
419 config DEBUG_KMEMLEAK
420 bool "Kernel memory leak detector"
421 depends on DEBUG_KERNEL && EXPERIMENTAL && \
422 (X86 || ARM || PPC || MIPS || S390 || SPARC64 || SUPERH || MICROBLAZE || TILE)
425 select STACKTRACE if STACKTRACE_SUPPORT
429 Say Y here if you want to enable the memory leak
430 detector. The memory allocation/freeing is traced in a way
431 similar to the Boehm's conservative garbage collector, the
432 difference being that the orphan objects are not freed but
433 only shown in /sys/kernel/debug/kmemleak. Enabling this
434 feature will introduce an overhead to memory
435 allocations. See Documentation/kmemleak.txt for more
438 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
439 of finding leaks due to the slab objects poisoning.
441 In order to access the kmemleak file, debugfs needs to be
442 mounted (usually at /sys/kernel/debug).
444 config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
445 int "Maximum kmemleak early log entries"
446 depends on DEBUG_KMEMLEAK
450 Kmemleak must track all the memory allocations to avoid
451 reporting false positives. Since memory may be allocated or
452 freed before kmemleak is initialised, an early log buffer is
453 used to store these actions. If kmemleak reports "early log
454 buffer exceeded", please increase this value.
456 config DEBUG_KMEMLEAK_TEST
457 tristate "Simple test for the kernel memory leak detector"
458 depends on DEBUG_KMEMLEAK && m
460 This option enables a module that explicitly leaks memory.
464 config DEBUG_KMEMLEAK_DEFAULT_OFF
465 bool "Default kmemleak to off"
466 depends on DEBUG_KMEMLEAK
468 Say Y here to disable kmemleak by default. It can then be enabled
469 on the command line via kmemleak=on.
472 bool "Debug preemptible kernel"
473 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
476 If you say Y here then the kernel will use a debug variant of the
477 commonly used smp_processor_id() function and will print warnings
478 if kernel code uses it in a preemption-unsafe way. Also, the kernel
479 will detect preemption count underflows.
481 config DEBUG_RT_MUTEXES
482 bool "RT Mutex debugging, deadlock detection"
483 depends on DEBUG_KERNEL && RT_MUTEXES
485 This allows rt mutex semantics violations and rt mutex related
486 deadlocks (lockups) to be detected and reported automatically.
491 depends on DEBUG_RT_MUTEXES
493 config RT_MUTEX_TESTER
494 bool "Built-in scriptable tester for rt-mutexes"
495 depends on DEBUG_KERNEL && RT_MUTEXES
497 This option enables a rt-mutex tester.
499 config DEBUG_SPINLOCK
500 bool "Spinlock and rw-lock debugging: basic checks"
501 depends on DEBUG_KERNEL
502 select UNINLINE_SPIN_UNLOCK
504 Say Y here and build SMP to catch missing spinlock initialization
505 and certain other kinds of spinlock errors commonly made. This is
506 best used in conjunction with the NMI watchdog so that spinlock
507 deadlocks are also debuggable.
510 bool "Mutex debugging: basic checks"
511 depends on DEBUG_KERNEL
513 This feature allows mutex semantics violations to be detected and
516 config DEBUG_LOCK_ALLOC
517 bool "Lock debugging: detect incorrect freeing of live locks"
518 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
519 select DEBUG_SPINLOCK
523 This feature will check whether any held lock (spinlock, rwlock,
524 mutex or rwsem) is incorrectly freed by the kernel, via any of the
525 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
526 vfree(), etc.), whether a live lock is incorrectly reinitialized via
527 spin_lock_init()/mutex_init()/etc., or whether there is any lock
528 held during task exit.
531 bool "Lock debugging: prove locking correctness"
532 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
534 select DEBUG_SPINLOCK
536 select DEBUG_LOCK_ALLOC
537 select TRACE_IRQFLAGS
540 This feature enables the kernel to prove that all locking
541 that occurs in the kernel runtime is mathematically
542 correct: that under no circumstance could an arbitrary (and
543 not yet triggered) combination of observed locking
544 sequences (on an arbitrary number of CPUs, running an
545 arbitrary number of tasks and interrupt contexts) cause a
548 In short, this feature enables the kernel to report locking
549 related deadlocks before they actually occur.
551 The proof does not depend on how hard and complex a
552 deadlock scenario would be to trigger: how many
553 participant CPUs, tasks and irq-contexts would be needed
554 for it to trigger. The proof also does not depend on
555 timing: if a race and a resulting deadlock is possible
556 theoretically (no matter how unlikely the race scenario
557 is), it will be proven so and will immediately be
558 reported by the kernel (once the event is observed that
559 makes the deadlock theoretically possible).
561 If a deadlock is impossible (i.e. the locking rules, as
562 observed by the kernel, are mathematically correct), the
563 kernel reports nothing.
565 NOTE: this feature can also be enabled for rwlocks, mutexes
566 and rwsems - in which case all dependencies between these
567 different locking variants are observed and mapped too, and
568 the proof of observed correctness is also maintained for an
569 arbitrary combination of these separate locking variants.
571 For more details, see Documentation/lockdep-design.txt.
574 bool "RCU debugging: prove RCU correctness"
575 depends on PROVE_LOCKING
578 This feature enables lockdep extensions that check for correct
579 use of RCU APIs. This is currently under development. Say Y
580 if you want to debug RCU usage or help work on the PROVE_RCU
583 Say N if you are unsure.
585 config PROVE_RCU_REPEATEDLY
586 bool "RCU debugging: don't disable PROVE_RCU on first splat"
590 By itself, PROVE_RCU will disable checking upon issuing the
591 first warning (or "splat"). This feature prevents such
592 disabling, allowing multiple RCU-lockdep warnings to be printed
595 Say Y to allow multiple RCU-lockdep warnings per boot.
597 Say N if you are unsure.
599 config SPARSE_RCU_POINTER
600 bool "RCU debugging: sparse-based checks for pointer usage"
603 This feature enables the __rcu sparse annotation for
604 RCU-protected pointers. This annotation will cause sparse
605 to flag any non-RCU used of annotated pointers. This can be
606 helpful when debugging RCU usage. Please note that this feature
607 is not intended to enforce code cleanliness; it is instead merely
610 Say Y to make sparse flag questionable use of RCU-protected pointers
612 Say N if you are unsure.
616 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
618 select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE
623 bool "Lock usage statistics"
624 depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
626 select DEBUG_SPINLOCK
628 select DEBUG_LOCK_ALLOC
631 This feature enables tracking lock contention points
633 For more details, see Documentation/lockstat.txt
635 This also enables lock events required by "perf lock",
637 If you want to use "perf lock", you also need to turn on
638 CONFIG_EVENT_TRACING.
640 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
641 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
644 bool "Lock dependency engine debugging"
645 depends on DEBUG_KERNEL && LOCKDEP
647 If you say Y here, the lock dependency engine will do
648 additional runtime checks to debug itself, at the price
649 of more runtime overhead.
651 config TRACE_IRQFLAGS
654 Enables hooks to interrupt enabling and disabling for
655 either tracing or lock debugging.
657 config DEBUG_ATOMIC_SLEEP
658 bool "Sleep inside atomic section checking"
660 depends on DEBUG_KERNEL
662 If you say Y here, various routines which may sleep will become very
663 noisy if they are called inside atomic sections: when a spinlock is
664 held, inside an rcu read side critical section, inside preempt disabled
665 sections, inside an interrupt, etc...
667 config DEBUG_LOCKING_API_SELFTESTS
668 bool "Locking API boot-time self-tests"
669 depends on DEBUG_KERNEL
671 Say Y here if you want the kernel to run a short self-test during
672 bootup. The self-test checks whether common types of locking bugs
673 are detected by debugging mechanisms or not. (if you disable
674 lock debugging then those bugs wont be detected of course.)
675 The following locking APIs are covered: spinlocks, rwlocks,
680 depends on STACKTRACE_SUPPORT
682 config DEBUG_STACK_USAGE
683 bool "Stack utilization instrumentation"
684 depends on DEBUG_KERNEL
686 Enables the display of the minimum amount of free stack which each
687 task has ever had available in the sysrq-T and sysrq-P debug output.
689 This option will slow down process creation somewhat.
692 bool "kobject debugging"
693 depends on DEBUG_KERNEL
695 If you say Y here, some extra kobject debugging messages will be sent
699 bool "Highmem debugging"
700 depends on DEBUG_KERNEL && HIGHMEM
702 This options enables addition error checking for high memory systems.
703 Disable for production systems.
705 config DEBUG_BUGVERBOSE
706 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
708 depends on ARM || AVR32 || M32R || M68K || SPARC32 || SPARC64 || \
709 FRV || SUPERH || GENERIC_BUG || BLACKFIN || MN10300 || TILE
712 Say Y here to make BUG() panics output the file name and line number
713 of the BUG call as well as the EIP and oops trace. This aids
714 debugging but costs about 70-100K of memory.
717 bool "Compile the kernel with debug info"
718 depends on DEBUG_KERNEL
720 If you say Y here the resulting kernel image will include
721 debugging info resulting in a larger kernel image.
722 This adds debug symbols to the kernel and modules (gcc -g), and
723 is needed if you intend to use kernel crashdump or binary object
724 tools like crash, kgdb, LKCD, gdb, etc on the kernel.
725 Say Y here only if you plan to debug the kernel.
729 config DEBUG_INFO_REDUCED
730 bool "Reduce debugging information"
731 depends on DEBUG_INFO
733 If you say Y here gcc is instructed to generate less debugging
734 information for structure types. This means that tools that
735 need full debugging information (like kgdb or systemtap) won't
736 be happy. But if you merely need debugging information to
737 resolve line numbers there is no loss. Advantage is that
738 build directory object sizes shrink dramatically over a full
739 DEBUG_INFO build and compile times are reduced too.
740 Only works with newer gcc versions.
744 depends on DEBUG_KERNEL
746 Enable this to turn on extended checks in the virtual-memory system
747 that may impact performance.
752 bool "Debug VM translations"
753 depends on DEBUG_KERNEL && X86
755 Enable some costly sanity checks in virtual to page code. This can
756 catch mistakes with virt_to_page() and friends.
760 config DEBUG_NOMMU_REGIONS
761 bool "Debug the global anon/private NOMMU mapping region tree"
762 depends on DEBUG_KERNEL && !MMU
764 This option causes the global tree of anonymous and private mapping
765 regions to be regularly checked for invalid topology.
767 config DEBUG_WRITECOUNT
768 bool "Debug filesystem writers count"
769 depends on DEBUG_KERNEL
771 Enable this to catch wrong use of the writers count in struct
772 vfsmount. This will increase the size of each file struct by
777 config DEBUG_MEMORY_INIT
778 bool "Debug memory initialisation" if EXPERT
781 Enable this for additional checks during memory initialisation.
782 The sanity checks verify aspects of the VM such as the memory model
783 and other information provided by the architecture. Verbose
784 information will be printed at KERN_DEBUG loglevel depending
785 on the mminit_loglevel= command-line option.
790 bool "Debug linked list manipulation"
791 depends on DEBUG_KERNEL
793 Enable this to turn on extended checks in the linked-list
798 config TEST_LIST_SORT
799 bool "Linked list sorting test"
800 depends on DEBUG_KERNEL
802 Enable this to turn on 'list_sort()' function test. This test is
803 executed only once during system boot, so affects only boot time.
808 bool "Debug SG table operations"
809 depends on DEBUG_KERNEL
811 Enable this to turn on checks on scatter-gather tables. This can
812 help find problems with drivers that do not properly initialize
817 config DEBUG_NOTIFIERS
818 bool "Debug notifier call chains"
819 depends on DEBUG_KERNEL
821 Enable this to turn on sanity checking for notifier call chains.
822 This is most useful for kernel developers to make sure that
823 modules properly unregister themselves from notifier chains.
824 This is a relatively cheap check but if you care about maximum
827 config DEBUG_CREDENTIALS
828 bool "Debug credential management"
829 depends on DEBUG_KERNEL
831 Enable this to turn on some debug checking for credential
832 management. The additional code keeps track of the number of
833 pointers from task_structs to any given cred struct, and checks to
834 see that this number never exceeds the usage count of the cred
837 Furthermore, if SELinux is enabled, this also checks that the
838 security pointer in the cred struct is never seen to be invalid.
843 # Select this config option from the architecture Kconfig, if it
844 # is preferred to always offer frame pointers as a config
845 # option on the architecture (regardless of KERNEL_DEBUG):
847 config ARCH_WANT_FRAME_POINTERS
852 bool "Compile the kernel with frame pointers"
853 depends on DEBUG_KERNEL && \
854 (CRIS || M68K || FRV || UML || \
855 AVR32 || SUPERH || BLACKFIN || MN10300) || \
856 ARCH_WANT_FRAME_POINTERS
857 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
859 If you say Y here the resulting kernel image will be slightly
860 larger and slower, but it gives very useful debugging information
861 in case of kernel bugs. (precise oopses/stacktraces/warnings)
864 bool "Compile the kernel with frame unwind information"
865 depends on !IA64 && !PARISC && !ARM
866 depends on !MODULES || !(MIPS || PPC || SUPERH || V850)
868 If you say Y here the resulting kernel image will be slightly larger
869 but not slower, and it will give very useful debugging information.
870 If you don't debug the kernel, you can say N, but we may not be able
871 to solve problems without frame unwind information or frame pointers.
874 bool "Stack unwind support"
875 depends on UNWIND_INFO
878 This enables more precise stack traces, omitting all unrelated
879 occurrences of pointers into kernel code from the dump.
881 config BOOT_PRINTK_DELAY
882 bool "Delay each boot printk message by N milliseconds"
883 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
885 This build option allows you to read kernel boot messages
886 by inserting a short delay after each one. The delay is
887 specified in milliseconds on the kernel command line,
888 using "boot_delay=N".
890 It is likely that you would also need to use "lpj=M" to preset
891 the "loops per jiffie" value.
892 See a previous boot log for the "lpj" value to use for your
893 system, and then set "lpj=M" before setting "boot_delay=N".
894 NOTE: Using this option may adversely affect SMP systems.
895 I.e., processors other than the first one may not boot up.
896 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect
897 what it believes to be lockup conditions.
899 config RCU_TORTURE_TEST
900 tristate "torture tests for RCU"
901 depends on DEBUG_KERNEL
904 This option provides a kernel module that runs torture tests
905 on the RCU infrastructure. The kernel module may be built
906 after the fact on the running kernel to be tested, if desired.
908 Say Y here if you want RCU torture tests to be built into
910 Say M if you want the RCU torture tests to build as a module.
911 Say N if you are unsure.
913 config RCU_TORTURE_TEST_RUNNABLE
914 bool "torture tests for RCU runnable by default"
915 depends on RCU_TORTURE_TEST = y
918 This option provides a way to build the RCU torture tests
919 directly into the kernel without them starting up at boot
920 time. You can use /proc/sys/kernel/rcutorture_runnable
921 to manually override this setting. This /proc file is
922 available only when the RCU torture tests have been built
925 Say Y here if you want the RCU torture tests to start during
926 boot (you probably don't).
927 Say N here if you want the RCU torture tests to start only
928 after being manually enabled via /proc.
930 config RCU_CPU_STALL_TIMEOUT
931 int "RCU CPU stall timeout in seconds"
932 depends on TREE_RCU || TREE_PREEMPT_RCU
936 If a given RCU grace period extends more than the specified
937 number of seconds, a CPU stall warning is printed. If the
938 RCU grace period persists, additional CPU stall warnings are
939 printed at more widely spaced intervals.
941 config RCU_CPU_STALL_VERBOSE
942 bool "Print additional per-task information for RCU_CPU_STALL_DETECTOR"
943 depends on TREE_PREEMPT_RCU
946 This option causes RCU to printk detailed per-task information
947 for any tasks that are stalling the current RCU grace period.
949 Say N if you are unsure.
951 Say Y if you want to enable such checks.
953 config RCU_CPU_STALL_INFO
954 bool "Print additional diagnostics on RCU CPU stall"
955 depends on (TREE_RCU || TREE_PREEMPT_RCU) && DEBUG_KERNEL
958 For each stalled CPU that is aware of the current RCU grace
959 period, print out additional per-CPU diagnostic information
960 regarding scheduling-clock ticks, idle state, and,
961 for RCU_FAST_NO_HZ kernels, idle-entry state.
963 Say N if you are unsure.
965 Say Y if you want to enable such diagnostics.
968 bool "Enable tracing for RCU"
969 depends on DEBUG_KERNEL
971 This option provides tracing in RCU which presents stats
972 in debugfs for debugging RCU implementation.
974 Say Y here if you want to enable RCU tracing
975 Say N if you are unsure.
977 config KPROBES_SANITY_TEST
978 bool "Kprobes sanity tests"
979 depends on DEBUG_KERNEL
983 This option provides for testing basic kprobes functionality on
984 boot. A sample kprobe, jprobe and kretprobe are inserted and
985 verified for functionality.
987 Say N if you are unsure.
989 config BACKTRACE_SELF_TEST
990 tristate "Self test for the backtrace code"
991 depends on DEBUG_KERNEL
994 This option provides a kernel module that can be used to test
995 the kernel stack backtrace code. This option is not useful
996 for distributions or general kernels, but only for kernel
997 developers working on architecture code.
999 Note that if you want to also test saved backtraces, you will
1000 have to enable STACKTRACE as well.
1002 Say N if you are unsure.
1004 config DEBUG_BLOCK_EXT_DEVT
1005 bool "Force extended block device numbers and spread them"
1006 depends on DEBUG_KERNEL
1010 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
1011 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
1012 YOU ARE DOING. Distros, please enable this and fix whatever
1015 Conventionally, block device numbers are allocated from
1016 predetermined contiguous area. However, extended block area
1017 may introduce non-contiguous block device numbers. This
1018 option forces most block device numbers to be allocated from
1019 the extended space and spreads them to discover kernel or
1020 userland code paths which assume predetermined contiguous
1021 device number allocation.
1023 Note that turning on this debug option shuffles all the
1024 device numbers for all IDE and SCSI devices including libata
1025 ones, so root partition specified using device number
1026 directly (via rdev or root=MAJ:MIN) won't work anymore.
1027 Textual device names (root=/dev/sdXn) will continue to work.
1029 Say N if you are unsure.
1031 config DEBUG_FORCE_WEAK_PER_CPU
1032 bool "Force weak per-cpu definitions"
1033 depends on DEBUG_KERNEL
1035 s390 and alpha require percpu variables in modules to be
1036 defined weak to work around addressing range issue which
1037 puts the following two restrictions on percpu variable
1040 1. percpu symbols must be unique whether static or not
1041 2. percpu variables can't be defined inside a function
1043 To ensure that generic code follows the above rules, this
1044 option forces all percpu variables to be defined as weak.
1046 config DEBUG_PER_CPU_MAPS
1047 bool "Debug access to per_cpu maps"
1048 depends on DEBUG_KERNEL
1051 Say Y to verify that the per_cpu map being accessed has
1052 been set up. This adds a fair amount of code to kernel memory
1053 and decreases performance.
1058 tristate "Linux Kernel Dump Test Tool Module"
1063 This module enables testing of the different dumping mechanisms by
1064 inducing system failures at predefined crash points.
1065 If you don't need it: say N
1066 Choose M here to compile this code as a module. The module will be
1069 Documentation on how to use the module can be found in
1070 Documentation/fault-injection/provoke-crashes.txt
1072 config CPU_NOTIFIER_ERROR_INJECT
1073 tristate "CPU notifier error injection module"
1074 depends on HOTPLUG_CPU && DEBUG_KERNEL
1076 This option provides a kernel module that can be used to test
1077 the error handling of the cpu notifiers
1079 To compile this code as a module, choose M here: the module will
1080 be called cpu-notifier-error-inject.
1084 config FAULT_INJECTION
1085 bool "Fault-injection framework"
1086 depends on DEBUG_KERNEL
1088 Provide fault-injection framework.
1089 For more details, see Documentation/fault-injection/.
1092 bool "Fault-injection capability for kmalloc"
1093 depends on FAULT_INJECTION
1094 depends on SLAB || SLUB
1096 Provide fault-injection capability for kmalloc.
1098 config FAIL_PAGE_ALLOC
1099 bool "Fault-injection capabilitiy for alloc_pages()"
1100 depends on FAULT_INJECTION
1102 Provide fault-injection capability for alloc_pages().
1104 config FAIL_MAKE_REQUEST
1105 bool "Fault-injection capability for disk IO"
1106 depends on FAULT_INJECTION && BLOCK
1108 Provide fault-injection capability for disk IO.
1110 config FAIL_IO_TIMEOUT
1111 bool "Fault-injection capability for faking disk interrupts"
1112 depends on FAULT_INJECTION && BLOCK
1114 Provide fault-injection capability on end IO handling. This
1115 will make the block layer "forget" an interrupt as configured,
1116 thus exercising the error handling.
1118 Only works with drivers that use the generic timeout handling,
1119 for others it wont do anything.
1121 config FAIL_MMC_REQUEST
1122 bool "Fault-injection capability for MMC IO"
1124 depends on FAULT_INJECTION && MMC
1126 Provide fault-injection capability for MMC IO.
1127 This will make the mmc core return data errors. This is
1128 useful to test the error handling in the mmc block device
1129 and to test how the mmc host driver handles retries from
1132 config FAULT_INJECTION_DEBUG_FS
1133 bool "Debugfs entries for fault-injection capabilities"
1134 depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1136 Enable configuration of fault-injection capabilities via debugfs.
1138 config FAULT_INJECTION_STACKTRACE_FILTER
1139 bool "stacktrace filter for fault-injection capabilities"
1140 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1143 select FRAME_POINTER if !PPC && !S390 && !MICROBLAZE && !X86 && !ARM_UNWIND
1144 select UNWIND_INFO if X86 && !FRAME_POINTER
1146 Provide stacktrace filter for fault-injection capabilities
1149 bool "Latency measuring infrastructure"
1150 depends on HAVE_LATENCYTOP_SUPPORT
1151 depends on DEBUG_KERNEL
1152 depends on STACKTRACE_SUPPORT
1154 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !X86 && !ARM_UNWIND
1155 select UNWIND_INFO if X86 && !FRAME_POINTER
1162 Enable this option if you want to use the LatencyTOP tool
1163 to find out which userspace is blocking on what kernel operations.
1165 source mm/Kconfig.debug
1166 source kernel/trace/Kconfig
1168 config PROVIDE_OHCI1394_DMA_INIT
1169 bool "Remote debugging over FireWire early on boot"
1170 depends on PCI && X86
1172 If you want to debug problems which hang or crash the kernel early
1173 on boot and the crashing machine has a FireWire port, you can use
1174 this feature to remotely access the memory of the crashed machine
1175 over FireWire. This employs remote DMA as part of the OHCI1394
1176 specification which is now the standard for FireWire controllers.
1178 With remote DMA, you can monitor the printk buffer remotely using
1179 firescope and access all memory below 4GB using fireproxy from gdb.
1180 Even controlling a kernel debugger is possible using remote DMA.
1184 If ohci1394_dma=early is used as boot parameter, it will initialize
1185 all OHCI1394 controllers which are found in the PCI config space.
1187 As all changes to the FireWire bus such as enabling and disabling
1188 devices cause a bus reset and thereby disable remote DMA for all
1189 devices, be sure to have the cable plugged and FireWire enabled on
1190 the debugging host before booting the debug target for debugging.
1192 This code (~1k) is freed after boot. By then, the firewire stack
1193 in charge of the OHCI-1394 controllers should be used instead.
1195 See Documentation/debugging-via-ohci1394.txt for more information.
1197 config FIREWIRE_OHCI_REMOTE_DMA
1198 bool "Remote debugging over FireWire with firewire-ohci"
1199 depends on FIREWIRE_OHCI
1201 This option lets you use the FireWire bus for remote debugging
1202 with help of the firewire-ohci driver. It enables unfiltered
1203 remote DMA in firewire-ohci.
1204 See Documentation/debugging-via-ohci1394.txt for more information.
1209 bool "Build targets in Documentation/ tree"
1210 depends on HEADERS_CHECK
1212 This option attempts to build objects from the source files in the
1213 kernel Documentation/ tree.
1215 Say N if you are unsure.
1217 config DYNAMIC_DEBUG
1218 bool "Enable dynamic printk() support"
1224 Compiles debug level messages into the kernel, which would not
1225 otherwise be available at runtime. These messages can then be
1226 enabled/disabled based on various levels of scope - per source file,
1227 function, module, format string, and line number. This mechanism
1228 implicitly enables all pr_debug() and dev_dbg() calls. The impact of
1229 this compile option is a larger kernel text size of about 2%.
1233 Dynamic debugging is controlled via the 'dynamic_debug/control' file,
1234 which is contained in the 'debugfs' filesystem. Thus, the debugfs
1235 filesystem must first be mounted before making use of this feature.
1236 We refer the control file as: <debugfs>/dynamic_debug/control. This
1237 file contains a list of the debug statements that can be enabled. The
1238 format for each line of the file is:
1240 filename:lineno [module]function flags format
1242 filename : source file of the debug statement
1243 lineno : line number of the debug statement
1244 module : module that contains the debug statement
1245 function : function that contains the debug statement
1246 flags : 'p' means the line is turned 'on' for printing
1247 format : the format used for the debug statement
1251 nullarbor:~ # cat <debugfs>/dynamic_debug/control
1252 # filename:lineno [module]function flags format
1253 fs/aio.c:222 [aio]__put_ioctx - "__put_ioctx:\040freeing\040%p\012"
1254 fs/aio.c:248 [aio]ioctx_alloc - "ENOMEM:\040nr_events\040too\040high\012"
1255 fs/aio.c:1770 [aio]sys_io_cancel - "calling\040cancel\012"
1259 // enable the message at line 1603 of file svcsock.c
1260 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
1261 <debugfs>/dynamic_debug/control
1263 // enable all the messages in file svcsock.c
1264 nullarbor:~ # echo -n 'file svcsock.c +p' >
1265 <debugfs>/dynamic_debug/control
1267 // enable all the messages in the NFS server module
1268 nullarbor:~ # echo -n 'module nfsd +p' >
1269 <debugfs>/dynamic_debug/control
1271 // enable all 12 messages in the function svc_process()
1272 nullarbor:~ # echo -n 'func svc_process +p' >
1273 <debugfs>/dynamic_debug/control
1275 // disable all 12 messages in the function svc_process()
1276 nullarbor:~ # echo -n 'func svc_process -p' >
1277 <debugfs>/dynamic_debug/control
1279 See Documentation/dynamic-debug-howto.txt for additional information.
1281 config DMA_API_DEBUG
1282 bool "Enable debugging of DMA-API usage"
1283 depends on HAVE_DMA_API_DEBUG
1285 Enable this option to debug the use of the DMA API by device drivers.
1286 With this option you will be able to detect common bugs in device
1287 drivers like double-freeing of DMA mappings or freeing mappings that
1288 were never allocated.
1289 This option causes a performance degredation. Use only if you want
1290 to debug device drivers. If unsure, say N.
1292 config ATOMIC64_SELFTEST
1293 bool "Perform an atomic64_t self-test at boot"
1295 Enable this option to test the atomic64_t functions at boot.
1299 config ASYNC_RAID6_TEST
1300 tristate "Self test for hardware accelerated raid6 recovery"
1301 depends on ASYNC_RAID6_RECOV
1304 This is a one-shot self test that permutes through the
1305 recovery of all the possible two disk failure scenarios for a
1306 N-disk array. Recovery is performed with the asynchronous
1307 raid6 recovery routines, and will optionally use an offload
1308 engine if one is available.
1312 source "samples/Kconfig"
1314 source "lib/Kconfig.kgdb"
1316 source "lib/Kconfig.kmemcheck"
1319 tristate "Test kstrto*() family of functions at runtime"