Root/lib/Kconfig.debug

1
2config PRINTK_TIME
3    bool "Show timing information on printks"
4    depends on PRINTK
5    help
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.
11
12config DEFAULT_MESSAGE_LOGLEVEL
13    int "Default message log level (1-7)"
14    range 1 7
15    default "4"
16    help
17      Default log level for printk statements with no specified priority.
18
19      This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
20      that are auditing their logs closely may want to set it to a lower
21      priority.
22
23config ENABLE_WARN_DEPRECATED
24    bool "Enable __deprecated logic"
25    default y
26    help
27      Enable the __deprecated logic in the kernel build.
28      Disable this to suppress the "warning: 'foo' is deprecated
29      (declared at kernel/power/somefile.c:1234)" messages.
30
31config ENABLE_MUST_CHECK
32    bool "Enable __must_check logic"
33    default y
34    help
35      Enable the __must_check logic in the kernel build. Disable this to
36      suppress the "warning: ignoring return value of 'foo', declared with
37      attribute warn_unused_result" messages.
38
39config FRAME_WARN
40    int "Warn for stack frames larger than (needs gcc 4.4)"
41    range 0 8192
42    default 1024 if !64BIT
43    default 2048 if 64BIT
44    help
45      Tell gcc to warn at build time for stack frames larger than this.
46      Setting this too low will cause a lot of warnings.
47      Setting it to 0 disables the warning.
48      Requires gcc 4.4
49
50config MAGIC_SYSRQ
51    bool "Magic SysRq key"
52    depends on !UML
53    help
54      If you say Y here, you will have some control over the system even
55      if the system crashes for example during kernel debugging (e.g., you
56      will be able to flush the buffer cache to disk, reboot the system
57      immediately or dump some status information). This is accomplished
58      by pressing various keys while holding SysRq (Alt+PrintScreen). It
59      also works on a serial console (on PC hardware at least), if you
60      send a BREAK and then within 5 seconds a command keypress. The
61      keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
62      unless you really know what this hack does.
63
64config STRIP_ASM_SYMS
65    bool "Strip assembler-generated symbols during link"
66    default n
67    help
68      Strip internal assembler-generated symbols during a link (symbols
69      that look like '.Lxxx') so they don't pollute the output of
70      get_wchan() and suchlike.
71
72config UNUSED_SYMBOLS
73    bool "Enable unused/obsolete exported symbols"
74    default y if X86
75    help
76      Unused but exported symbols make the kernel needlessly bigger. For
77      that reason most of these unused exports will soon be removed. This
78      option is provided temporarily to provide a transition period in case
79      some external kernel module needs one of these symbols anyway. If you
80      encounter such a case in your module, consider if you are actually
81      using the right API. (rationale: since nobody in the kernel is using
82      this in a module, there is a pretty good chance it's actually the
83      wrong interface to use). If you really need the symbol, please send a
84      mail to the linux kernel mailing list mentioning the symbol and why
85      you really need it, and what the merge plan to the mainline kernel for
86      your module is.
87
88config DEBUG_FS
89    bool "Debug Filesystem"
90    help
91      debugfs is a virtual file system that kernel developers use to put
92      debugging files into. Enable this option to be able to read and
93      write to these files.
94
95      For detailed documentation on the debugfs API, see
96      Documentation/DocBook/filesystems.
97
98      If unsure, say N.
99
100config HEADERS_CHECK
101    bool "Run 'make headers_check' when building vmlinux"
102    depends on !UML
103    help
104      This option will extract the user-visible kernel headers whenever
105      building the kernel, and will run basic sanity checks on them to
106      ensure that exported files do not attempt to include files which
107      were not exported, etc.
108
109      If you're making modifications to header files which are
110      relevant for userspace, say 'Y', and check the headers
111      exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
112      your build tree), to make sure they're suitable.
113
114config DEBUG_SECTION_MISMATCH
115    bool "Enable full Section mismatch analysis"
116    help
117      The section mismatch analysis checks if there are illegal
118      references from one section to another section.
119      Linux will during link or during runtime drop some sections
120      and any use of code/data previously in these sections will
121      most likely result in an oops.
122      In the code functions and variables are annotated with
123      __init, __devinit etc. (see full list in include/linux/init.h)
124      which results in the code/data being placed in specific sections.
125      The section mismatch analysis is always done after a full
126      kernel build but enabling this option will in addition
127      do the following:
128      - Add the option -fno-inline-functions-called-once to gcc
129        When inlining a function annotated __init in a non-init
130        function we would lose the section information and thus
131        the analysis would not catch the illegal reference.
132        This option tells gcc to inline less but will also
133        result in a larger kernel.
134      - Run the section mismatch analysis for each module/built-in.o
135        When we run the section mismatch analysis on vmlinux.o we
136        lose valueble information about where the mismatch was
137        introduced.
138        Running the analysis for each module/built-in.o file
139        will tell where the mismatch happens much closer to the
140        source. The drawback is that we will report the same
141        mismatch at least twice.
142      - Enable verbose reporting from modpost to help solving
143        the section mismatches reported.
144
145config DEBUG_KERNEL
146    bool "Kernel debugging"
147    help
148      Say Y here if you are developing drivers or trying to debug and
149      identify kernel problems.
150
151config DEBUG_SHIRQ
152    bool "Debug shared IRQ handlers"
153    depends on DEBUG_KERNEL && GENERIC_HARDIRQS
154    help
155      Enable this to generate a spurious interrupt as soon as a shared
156      interrupt handler is registered, and just before one is deregistered.
157      Drivers ought to be able to handle interrupts coming in at those
158      points; some don't and need to be caught.
159
160config LOCKUP_DETECTOR
161    bool "Detect Hard and Soft Lockups"
162    depends on DEBUG_KERNEL && !S390
163    help
164      Say Y here to enable the kernel to act as a watchdog to detect
165      hard and soft lockups.
166
167      Softlockups are bugs that cause the kernel to loop in kernel
168      mode for more than 60 seconds, without giving other tasks a
169      chance to run. The current stack trace is displayed upon
170      detection and the system will stay locked up.
171
172      Hardlockups are bugs that cause the CPU to loop in kernel mode
173      for more than 60 seconds, without letting other interrupts have a
174      chance to run. The current stack trace is displayed upon detection
175      and the system will stay locked up.
176
177      The overhead should be minimal. A periodic hrtimer runs to
178      generate interrupts and kick the watchdog task every 10-12 seconds.
179      An NMI is generated every 60 seconds or so to check for hardlockups.
180
181config HARDLOCKUP_DETECTOR
182    def_bool LOCKUP_DETECTOR && PERF_EVENTS && HAVE_PERF_EVENTS_NMI && \
183         !ARCH_HAS_NMI_WATCHDOG
184
185config BOOTPARAM_HARDLOCKUP_PANIC
186    bool "Panic (Reboot) On Hard Lockups"
187    depends on LOCKUP_DETECTOR
188    help
189      Say Y here to enable the kernel to panic on "hard lockups",
190      which are bugs that cause the kernel to loop in kernel
191      mode with interrupts disabled for more than 60 seconds.
192
193      Say N if unsure.
194
195config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
196    int
197    depends on LOCKUP_DETECTOR
198    range 0 1
199    default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
200    default 1 if BOOTPARAM_HARDLOCKUP_PANIC
201
202config BOOTPARAM_SOFTLOCKUP_PANIC
203    bool "Panic (Reboot) On Soft Lockups"
204    depends on LOCKUP_DETECTOR
205    help
206      Say Y here to enable the kernel to panic on "soft lockups",
207      which are bugs that cause the kernel to loop in kernel
208      mode for more than 60 seconds, without giving other tasks a
209      chance to run.
210
211      The panic can be used in combination with panic_timeout,
212      to cause the system to reboot automatically after a
213      lockup has been detected. This feature is useful for
214      high-availability systems that have uptime guarantees and
215      where a lockup must be resolved ASAP.
216
217      Say N if unsure.
218
219config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
220    int
221    depends on LOCKUP_DETECTOR
222    range 0 1
223    default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
224    default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
225
226config DETECT_HUNG_TASK
227    bool "Detect Hung Tasks"
228    depends on DEBUG_KERNEL
229    default DETECT_SOFTLOCKUP
230    help
231      Say Y here to enable the kernel to detect "hung tasks",
232      which are bugs that cause the task to be stuck in
233      uninterruptible "D" state indefinitiley.
234
235      When a hung task is detected, the kernel will print the
236      current stack trace (which you should report), but the
237      task will stay in uninterruptible state. If lockdep is
238      enabled then all held locks will also be reported. This
239      feature has negligible overhead.
240
241config BOOTPARAM_HUNG_TASK_PANIC
242    bool "Panic (Reboot) On Hung Tasks"
243    depends on DETECT_HUNG_TASK
244    help
245      Say Y here to enable the kernel to panic on "hung tasks",
246      which are bugs that cause the kernel to leave a task stuck
247      in uninterruptible "D" state.
248
249      The panic can be used in combination with panic_timeout,
250      to cause the system to reboot automatically after a
251      hung task has been detected. This feature is useful for
252      high-availability systems that have uptime guarantees and
253      where a hung tasks must be resolved ASAP.
254
255      Say N if unsure.
256
257config BOOTPARAM_HUNG_TASK_PANIC_VALUE
258    int
259    depends on DETECT_HUNG_TASK
260    range 0 1
261    default 0 if !BOOTPARAM_HUNG_TASK_PANIC
262    default 1 if BOOTPARAM_HUNG_TASK_PANIC
263
264config SCHED_DEBUG
265    bool "Collect scheduler debugging info"
266    depends on DEBUG_KERNEL && PROC_FS
267    default y
268    help
269      If you say Y here, the /proc/sched_debug file will be provided
270      that can help debug the scheduler. The runtime overhead of this
271      option is minimal.
272
273config SCHEDSTATS
274    bool "Collect scheduler statistics"
275    depends on DEBUG_KERNEL && PROC_FS
276    help
277      If you say Y here, additional code will be inserted into the
278      scheduler and related routines to collect statistics about
279      scheduler behavior and provide them in /proc/schedstat. These
280      stats may be useful for both tuning and debugging the scheduler
281      If you aren't debugging the scheduler or trying to tune a specific
282      application, you can say N to avoid the very slight overhead
283      this adds.
284
285config TIMER_STATS
286    bool "Collect kernel timers statistics"
287    depends on DEBUG_KERNEL && PROC_FS
288    help
289      If you say Y here, additional code will be inserted into the
290      timer routines to collect statistics about kernel timers being
291      reprogrammed. The statistics can be read from /proc/timer_stats.
292      The statistics collection is started by writing 1 to /proc/timer_stats,
293      writing 0 stops it. This feature is useful to collect information
294      about timer usage patterns in kernel and userspace. This feature
295      is lightweight if enabled in the kernel config but not activated
296      (it defaults to deactivated on bootup and will only be activated
297      if some application like powertop activates it explicitly).
298
299config DEBUG_OBJECTS
300    bool "Debug object operations"
301    depends on DEBUG_KERNEL
302    help
303      If you say Y here, additional code will be inserted into the
304      kernel to track the life time of various objects and validate
305      the operations on those objects.
306
307config DEBUG_OBJECTS_SELFTEST
308    bool "Debug objects selftest"
309    depends on DEBUG_OBJECTS
310    help
311      This enables the selftest of the object debug code.
312
313config DEBUG_OBJECTS_FREE
314    bool "Debug objects in freed memory"
315    depends on DEBUG_OBJECTS
316    help
317      This enables checks whether a k/v free operation frees an area
318      which contains an object which has not been deactivated
319      properly. This can make kmalloc/kfree-intensive workloads
320      much slower.
321
322config DEBUG_OBJECTS_TIMERS
323    bool "Debug timer objects"
324    depends on DEBUG_OBJECTS
325    help
326      If you say Y here, additional code will be inserted into the
327      timer routines to track the life time of timer objects and
328      validate the timer operations.
329
330config DEBUG_OBJECTS_WORK
331    bool "Debug work objects"
332    depends on DEBUG_OBJECTS
333    help
334      If you say Y here, additional code will be inserted into the
335      work queue routines to track the life time of work objects and
336      validate the work operations.
337
338config DEBUG_OBJECTS_RCU_HEAD
339    bool "Debug RCU callbacks objects"
340    depends on DEBUG_OBJECTS && PREEMPT
341    help
342      Enable this to turn on debugging of RCU list heads (call_rcu() usage).
343
344config DEBUG_OBJECTS_PERCPU_COUNTER
345    bool "Debug percpu counter objects"
346    depends on DEBUG_OBJECTS
347    help
348      If you say Y here, additional code will be inserted into the
349      percpu counter routines to track the life time of percpu counter
350      objects and validate the percpu counter operations.
351
352config DEBUG_OBJECTS_ENABLE_DEFAULT
353    int "debug_objects bootup default value (0-1)"
354        range 0 1
355        default "1"
356        depends on DEBUG_OBJECTS
357        help
358          Debug objects boot parameter default value
359
360config DEBUG_SLAB
361    bool "Debug slab memory allocations"
362    depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
363    help
364      Say Y here to have the kernel do limited verification on memory
365      allocation as well as poisoning memory on free to catch use of freed
366      memory. This can make kmalloc/kfree-intensive workloads much slower.
367
368config DEBUG_SLAB_LEAK
369    bool "Memory leak debugging"
370    depends on DEBUG_SLAB
371
372config SLUB_DEBUG_ON
373    bool "SLUB debugging on by default"
374    depends on SLUB && SLUB_DEBUG && !KMEMCHECK
375    default n
376    help
377      Boot with debugging on by default. SLUB boots by default with
378      the runtime debug capabilities switched off. Enabling this is
379      equivalent to specifying the "slub_debug" parameter on boot.
380      There is no support for more fine grained debug control like
381      possible with slub_debug=xxx. SLUB debugging may be switched
382      off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
383      "slub_debug=-".
384
385config SLUB_STATS
386    default n
387    bool "Enable SLUB performance statistics"
388    depends on SLUB && SYSFS
389    help
390      SLUB statistics are useful to debug SLUBs allocation behavior in
391      order find ways to optimize the allocator. This should never be
392      enabled for production use since keeping statistics slows down
393      the allocator by a few percentage points. The slabinfo command
394      supports the determination of the most active slabs to figure
395      out which slabs are relevant to a particular load.
396      Try running: slabinfo -DA
397
398config DEBUG_KMEMLEAK
399    bool "Kernel memory leak detector"
400    depends on DEBUG_KERNEL && EXPERIMENTAL && !MEMORY_HOTPLUG && \
401        (X86 || ARM || PPC || S390 || SPARC64 || SUPERH || MICROBLAZE || TILE)
402
403    select DEBUG_FS if SYSFS
404    select STACKTRACE if STACKTRACE_SUPPORT
405    select KALLSYMS
406    select CRC32
407    help
408      Say Y here if you want to enable the memory leak
409      detector. The memory allocation/freeing is traced in a way
410      similar to the Boehm's conservative garbage collector, the
411      difference being that the orphan objects are not freed but
412      only shown in /sys/kernel/debug/kmemleak. Enabling this
413      feature will introduce an overhead to memory
414      allocations. See Documentation/kmemleak.txt for more
415      details.
416
417      Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
418      of finding leaks due to the slab objects poisoning.
419
420      In order to access the kmemleak file, debugfs needs to be
421      mounted (usually at /sys/kernel/debug).
422
423config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
424    int "Maximum kmemleak early log entries"
425    depends on DEBUG_KMEMLEAK
426    range 200 40000
427    default 400
428    help
429      Kmemleak must track all the memory allocations to avoid
430      reporting false positives. Since memory may be allocated or
431      freed before kmemleak is initialised, an early log buffer is
432      used to store these actions. If kmemleak reports "early log
433      buffer exceeded", please increase this value.
434
435config DEBUG_KMEMLEAK_TEST
436    tristate "Simple test for the kernel memory leak detector"
437    depends on DEBUG_KMEMLEAK && m
438    help
439      This option enables a module that explicitly leaks memory.
440
441      If unsure, say N.
442
443config DEBUG_KMEMLEAK_DEFAULT_OFF
444    bool "Default kmemleak to off"
445    depends on DEBUG_KMEMLEAK
446    help
447      Say Y here to disable kmemleak by default. It can then be enabled
448      on the command line via kmemleak=on.
449
450config DEBUG_PREEMPT
451    bool "Debug preemptible kernel"
452    depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
453    default y
454    help
455      If you say Y here then the kernel will use a debug variant of the
456      commonly used smp_processor_id() function and will print warnings
457      if kernel code uses it in a preemption-unsafe way. Also, the kernel
458      will detect preemption count underflows.
459
460config DEBUG_RT_MUTEXES
461    bool "RT Mutex debugging, deadlock detection"
462    depends on DEBUG_KERNEL && RT_MUTEXES
463    help
464     This allows rt mutex semantics violations and rt mutex related
465     deadlocks (lockups) to be detected and reported automatically.
466
467config DEBUG_PI_LIST
468    bool
469    default y
470    depends on DEBUG_RT_MUTEXES
471
472config RT_MUTEX_TESTER
473    bool "Built-in scriptable tester for rt-mutexes"
474    depends on DEBUG_KERNEL && RT_MUTEXES
475    help
476      This option enables a rt-mutex tester.
477
478config DEBUG_SPINLOCK
479    bool "Spinlock and rw-lock debugging: basic checks"
480    depends on DEBUG_KERNEL
481    help
482      Say Y here and build SMP to catch missing spinlock initialization
483      and certain other kinds of spinlock errors commonly made. This is
484      best used in conjunction with the NMI watchdog so that spinlock
485      deadlocks are also debuggable.
486
487config DEBUG_MUTEXES
488    bool "Mutex debugging: basic checks"
489    depends on DEBUG_KERNEL
490    help
491     This feature allows mutex semantics violations to be detected and
492     reported.
493
494config DEBUG_LOCK_ALLOC
495    bool "Lock debugging: detect incorrect freeing of live locks"
496    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
497    select DEBUG_SPINLOCK
498    select DEBUG_MUTEXES
499    select LOCKDEP
500    help
501     This feature will check whether any held lock (spinlock, rwlock,
502     mutex or rwsem) is incorrectly freed by the kernel, via any of the
503     memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
504     vfree(), etc.), whether a live lock is incorrectly reinitialized via
505     spin_lock_init()/mutex_init()/etc., or whether there is any lock
506     held during task exit.
507
508config PROVE_LOCKING
509    bool "Lock debugging: prove locking correctness"
510    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
511    select LOCKDEP
512    select DEBUG_SPINLOCK
513    select DEBUG_MUTEXES
514    select DEBUG_LOCK_ALLOC
515    select TRACE_IRQFLAGS
516    default n
517    help
518     This feature enables the kernel to prove that all locking
519     that occurs in the kernel runtime is mathematically
520     correct: that under no circumstance could an arbitrary (and
521     not yet triggered) combination of observed locking
522     sequences (on an arbitrary number of CPUs, running an
523     arbitrary number of tasks and interrupt contexts) cause a
524     deadlock.
525
526     In short, this feature enables the kernel to report locking
527     related deadlocks before they actually occur.
528
529     The proof does not depend on how hard and complex a
530     deadlock scenario would be to trigger: how many
531     participant CPUs, tasks and irq-contexts would be needed
532     for it to trigger. The proof also does not depend on
533     timing: if a race and a resulting deadlock is possible
534     theoretically (no matter how unlikely the race scenario
535     is), it will be proven so and will immediately be
536     reported by the kernel (once the event is observed that
537     makes the deadlock theoretically possible).
538
539     If a deadlock is impossible (i.e. the locking rules, as
540     observed by the kernel, are mathematically correct), the
541     kernel reports nothing.
542
543     NOTE: this feature can also be enabled for rwlocks, mutexes
544     and rwsems - in which case all dependencies between these
545     different locking variants are observed and mapped too, and
546     the proof of observed correctness is also maintained for an
547     arbitrary combination of these separate locking variants.
548
549     For more details, see Documentation/lockdep-design.txt.
550
551config PROVE_RCU
552    bool "RCU debugging: prove RCU correctness"
553    depends on PROVE_LOCKING
554    default n
555    help
556     This feature enables lockdep extensions that check for correct
557     use of RCU APIs. This is currently under development. Say Y
558     if you want to debug RCU usage or help work on the PROVE_RCU
559     feature.
560
561     Say N if you are unsure.
562
563config PROVE_RCU_REPEATEDLY
564    bool "RCU debugging: don't disable PROVE_RCU on first splat"
565    depends on PROVE_RCU
566    default n
567    help
568     By itself, PROVE_RCU will disable checking upon issuing the
569     first warning (or "splat"). This feature prevents such
570     disabling, allowing multiple RCU-lockdep warnings to be printed
571     on a single reboot.
572
573     Say Y to allow multiple RCU-lockdep warnings per boot.
574
575     Say N if you are unsure.
576
577config SPARSE_RCU_POINTER
578    bool "RCU debugging: sparse-based checks for pointer usage"
579    default n
580    help
581     This feature enables the __rcu sparse annotation for
582     RCU-protected pointers. This annotation will cause sparse
583     to flag any non-RCU used of annotated pointers. This can be
584     helpful when debugging RCU usage. Please note that this feature
585     is not intended to enforce code cleanliness; it is instead merely
586     a debugging aid.
587
588     Say Y to make sparse flag questionable use of RCU-protected pointers
589
590     Say N if you are unsure.
591
592config LOCKDEP
593    bool
594    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
595    select STACKTRACE
596    select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE
597    select KALLSYMS
598    select KALLSYMS_ALL
599
600config LOCK_STAT
601    bool "Lock usage statistics"
602    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
603    select LOCKDEP
604    select DEBUG_SPINLOCK
605    select DEBUG_MUTEXES
606    select DEBUG_LOCK_ALLOC
607    default n
608    help
609     This feature enables tracking lock contention points
610
611     For more details, see Documentation/lockstat.txt
612
613     This also enables lock events required by "perf lock",
614     subcommand of perf.
615     If you want to use "perf lock", you also need to turn on
616     CONFIG_EVENT_TRACING.
617
618     CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
619     (CONFIG_LOCKDEP defines "acquire" and "release" events.)
620
621config DEBUG_LOCKDEP
622    bool "Lock dependency engine debugging"
623    depends on DEBUG_KERNEL && LOCKDEP
624    help
625      If you say Y here, the lock dependency engine will do
626      additional runtime checks to debug itself, at the price
627      of more runtime overhead.
628
629config TRACE_IRQFLAGS
630    bool
631    help
632      Enables hooks to interrupt enabling and disabling for
633      either tracing or lock debugging.
634
635config DEBUG_SPINLOCK_SLEEP
636    bool "Spinlock debugging: sleep-inside-spinlock checking"
637    depends on DEBUG_KERNEL
638    help
639      If you say Y here, various routines which may sleep will become very
640      noisy if they are called with a spinlock held.
641
642config DEBUG_LOCKING_API_SELFTESTS
643    bool "Locking API boot-time self-tests"
644    depends on DEBUG_KERNEL
645    help
646      Say Y here if you want the kernel to run a short self-test during
647      bootup. The self-test checks whether common types of locking bugs
648      are detected by debugging mechanisms or not. (if you disable
649      lock debugging then those bugs wont be detected of course.)
650      The following locking APIs are covered: spinlocks, rwlocks,
651      mutexes and rwsems.
652
653config STACKTRACE
654    bool
655    depends on STACKTRACE_SUPPORT
656
657config DEBUG_KOBJECT
658    bool "kobject debugging"
659    depends on DEBUG_KERNEL
660    help
661      If you say Y here, some extra kobject debugging messages will be sent
662      to the syslog.
663
664config DEBUG_HIGHMEM
665    bool "Highmem debugging"
666    depends on DEBUG_KERNEL && HIGHMEM
667    help
668      This options enables addition error checking for high memory systems.
669      Disable for production systems.
670
671config DEBUG_BUGVERBOSE
672    bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
673    depends on BUG
674    depends on ARM || AVR32 || M32R || M68K || SPARC32 || SPARC64 || \
675           FRV || SUPERH || GENERIC_BUG || BLACKFIN || MN10300
676    default y
677    help
678      Say Y here to make BUG() panics output the file name and line number
679      of the BUG call as well as the EIP and oops trace. This aids
680      debugging but costs about 70-100K of memory.
681
682config DEBUG_INFO
683    bool "Compile the kernel with debug info"
684    depends on DEBUG_KERNEL
685    help
686          If you say Y here the resulting kernel image will include
687      debugging info resulting in a larger kernel image.
688      This adds debug symbols to the kernel and modules (gcc -g), and
689      is needed if you intend to use kernel crashdump or binary object
690      tools like crash, kgdb, LKCD, gdb, etc on the kernel.
691      Say Y here only if you plan to debug the kernel.
692
693      If unsure, say N.
694
695config DEBUG_INFO_REDUCED
696    bool "Reduce debugging information"
697    depends on DEBUG_INFO
698    help
699      If you say Y here gcc is instructed to generate less debugging
700      information for structure types. This means that tools that
701      need full debugging information (like kgdb or systemtap) won't
702      be happy. But if you merely need debugging information to
703      resolve line numbers there is no loss. Advantage is that
704      build directory object sizes shrink dramatically over a full
705      DEBUG_INFO build and compile times are reduced too.
706      Only works with newer gcc versions.
707
708config DEBUG_VM
709    bool "Debug VM"
710    depends on DEBUG_KERNEL
711    help
712      Enable this to turn on extended checks in the virtual-memory system
713          that may impact performance.
714
715      If unsure, say N.
716
717config DEBUG_VIRTUAL
718    bool "Debug VM translations"
719    depends on DEBUG_KERNEL && X86
720    help
721      Enable some costly sanity checks in virtual to page code. This can
722      catch mistakes with virt_to_page() and friends.
723
724      If unsure, say N.
725
726config DEBUG_NOMMU_REGIONS
727    bool "Debug the global anon/private NOMMU mapping region tree"
728    depends on DEBUG_KERNEL && !MMU
729    help
730      This option causes the global tree of anonymous and private mapping
731      regions to be regularly checked for invalid topology.
732
733config DEBUG_WRITECOUNT
734    bool "Debug filesystem writers count"
735    depends on DEBUG_KERNEL
736    help
737      Enable this to catch wrong use of the writers count in struct
738      vfsmount. This will increase the size of each file struct by
739      32 bits.
740
741      If unsure, say N.
742
743config DEBUG_MEMORY_INIT
744    bool "Debug memory initialisation" if EXPERT
745    default !EXPERT
746    help
747      Enable this for additional checks during memory initialisation.
748      The sanity checks verify aspects of the VM such as the memory model
749      and other information provided by the architecture. Verbose
750      information will be printed at KERN_DEBUG loglevel depending
751      on the mminit_loglevel= command-line option.
752
753      If unsure, say Y
754
755config DEBUG_LIST
756    bool "Debug linked list manipulation"
757    depends on DEBUG_KERNEL
758    help
759      Enable this to turn on extended checks in the linked-list
760      walking routines.
761
762      If unsure, say N.
763
764config TEST_LIST_SORT
765    bool "Linked list sorting test"
766    depends on DEBUG_KERNEL
767    help
768      Enable this to turn on 'list_sort()' function test. This test is
769      executed only once during system boot, so affects only boot time.
770
771      If unsure, say N.
772
773config DEBUG_SG
774    bool "Debug SG table operations"
775    depends on DEBUG_KERNEL
776    help
777      Enable this to turn on checks on scatter-gather tables. This can
778      help find problems with drivers that do not properly initialize
779      their sg tables.
780
781      If unsure, say N.
782
783config DEBUG_NOTIFIERS
784    bool "Debug notifier call chains"
785    depends on DEBUG_KERNEL
786    help
787      Enable this to turn on sanity checking for notifier call chains.
788      This is most useful for kernel developers to make sure that
789      modules properly unregister themselves from notifier chains.
790      This is a relatively cheap check but if you care about maximum
791      performance, say N.
792
793config DEBUG_CREDENTIALS
794    bool "Debug credential management"
795    depends on DEBUG_KERNEL
796    help
797      Enable this to turn on some debug checking for credential
798      management. The additional code keeps track of the number of
799      pointers from task_structs to any given cred struct, and checks to
800      see that this number never exceeds the usage count of the cred
801      struct.
802
803      Furthermore, if SELinux is enabled, this also checks that the
804      security pointer in the cred struct is never seen to be invalid.
805
806      If unsure, say N.
807
808#
809# Select this config option from the architecture Kconfig, if it
810# it is preferred to always offer frame pointers as a config
811# option on the architecture (regardless of KERNEL_DEBUG):
812#
813config ARCH_WANT_FRAME_POINTERS
814    bool
815    help
816
817config FRAME_POINTER
818    bool "Compile the kernel with frame pointers"
819    depends on DEBUG_KERNEL && \
820        (CRIS || M68K || FRV || UML || \
821         AVR32 || SUPERH || BLACKFIN || MN10300) || \
822        ARCH_WANT_FRAME_POINTERS
823    default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
824    help
825      If you say Y here the resulting kernel image will be slightly
826      larger and slower, but it gives very useful debugging information
827      in case of kernel bugs. (precise oopses/stacktraces/warnings)
828
829config BOOT_PRINTK_DELAY
830    bool "Delay each boot printk message by N milliseconds"
831    depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
832    help
833      This build option allows you to read kernel boot messages
834      by inserting a short delay after each one. The delay is
835      specified in milliseconds on the kernel command line,
836      using "boot_delay=N".
837
838      It is likely that you would also need to use "lpj=M" to preset
839      the "loops per jiffie" value.
840      See a previous boot log for the "lpj" value to use for your
841      system, and then set "lpj=M" before setting "boot_delay=N".
842      NOTE: Using this option may adversely affect SMP systems.
843      I.e., processors other than the first one may not boot up.
844      BOOT_PRINTK_DELAY also may cause DETECT_SOFTLOCKUP to detect
845      what it believes to be lockup conditions.
846
847config RCU_TORTURE_TEST
848    tristate "torture tests for RCU"
849    depends on DEBUG_KERNEL
850    default n
851    help
852      This option provides a kernel module that runs torture tests
853      on the RCU infrastructure. The kernel module may be built
854      after the fact on the running kernel to be tested, if desired.
855
856      Say Y here if you want RCU torture tests to be built into
857      the kernel.
858      Say M if you want the RCU torture tests to build as a module.
859      Say N if you are unsure.
860
861config RCU_TORTURE_TEST_RUNNABLE
862    bool "torture tests for RCU runnable by default"
863    depends on RCU_TORTURE_TEST = y
864    default n
865    help
866      This option provides a way to build the RCU torture tests
867      directly into the kernel without them starting up at boot
868      time. You can use /proc/sys/kernel/rcutorture_runnable
869      to manually override this setting. This /proc file is
870      available only when the RCU torture tests have been built
871      into the kernel.
872
873      Say Y here if you want the RCU torture tests to start during
874      boot (you probably don't).
875      Say N here if you want the RCU torture tests to start only
876      after being manually enabled via /proc.
877
878config RCU_CPU_STALL_DETECTOR
879    bool "Check for stalled CPUs delaying RCU grace periods"
880    depends on TREE_RCU || TREE_PREEMPT_RCU
881    default y
882    help
883      This option causes RCU to printk information on which
884      CPUs are delaying the current grace period, but only when
885      the grace period extends for excessive time periods.
886
887      Say N if you want to disable such checks.
888
889      Say Y if you are unsure.
890
891config RCU_CPU_STALL_TIMEOUT
892    int "RCU CPU stall timeout in seconds"
893    depends on RCU_CPU_STALL_DETECTOR
894    range 3 300
895    default 60
896    help
897      If a given RCU grace period extends more than the specified
898      number of seconds, a CPU stall warning is printed. If the
899      RCU grace period persists, additional CPU stall warnings are
900      printed at more widely spaced intervals.
901
902config RCU_CPU_STALL_DETECTOR_RUNNABLE
903    bool "RCU CPU stall checking starts automatically at boot"
904    depends on RCU_CPU_STALL_DETECTOR
905    default y
906    help
907      If set, start checking for RCU CPU stalls immediately on
908      boot. Otherwise, RCU CPU stall checking must be manually
909      enabled.
910
911      Say Y if you are unsure.
912
913      Say N if you wish to suppress RCU CPU stall checking during boot.
914
915config RCU_CPU_STALL_VERBOSE
916    bool "Print additional per-task information for RCU_CPU_STALL_DETECTOR"
917    depends on RCU_CPU_STALL_DETECTOR && TREE_PREEMPT_RCU
918    default y
919    help
920      This option causes RCU to printk detailed per-task information
921      for any tasks that are stalling the current RCU grace period.
922
923      Say N if you are unsure.
924
925      Say Y if you want to enable such checks.
926
927config KPROBES_SANITY_TEST
928    bool "Kprobes sanity tests"
929    depends on DEBUG_KERNEL
930    depends on KPROBES
931    default n
932    help
933      This option provides for testing basic kprobes functionality on
934      boot. A sample kprobe, jprobe and kretprobe are inserted and
935      verified for functionality.
936
937      Say N if you are unsure.
938
939config BACKTRACE_SELF_TEST
940    tristate "Self test for the backtrace code"
941    depends on DEBUG_KERNEL
942    default n
943    help
944      This option provides a kernel module that can be used to test
945      the kernel stack backtrace code. This option is not useful
946      for distributions or general kernels, but only for kernel
947      developers working on architecture code.
948
949      Note that if you want to also test saved backtraces, you will
950      have to enable STACKTRACE as well.
951
952      Say N if you are unsure.
953
954config DEBUG_BLOCK_EXT_DEVT
955        bool "Force extended block device numbers and spread them"
956    depends on DEBUG_KERNEL
957    depends on BLOCK
958    default n
959    help
960      BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
961      SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
962      YOU ARE DOING. Distros, please enable this and fix whatever
963      is broken.
964
965      Conventionally, block device numbers are allocated from
966      predetermined contiguous area. However, extended block area
967      may introduce non-contiguous block device numbers. This
968      option forces most block device numbers to be allocated from
969      the extended space and spreads them to discover kernel or
970      userland code paths which assume predetermined contiguous
971      device number allocation.
972
973      Note that turning on this debug option shuffles all the
974      device numbers for all IDE and SCSI devices including libata
975      ones, so root partition specified using device number
976      directly (via rdev or root=MAJ:MIN) won't work anymore.
977      Textual device names (root=/dev/sdXn) will continue to work.
978
979      Say N if you are unsure.
980
981config DEBUG_FORCE_WEAK_PER_CPU
982    bool "Force weak per-cpu definitions"
983    depends on DEBUG_KERNEL
984    help
985      s390 and alpha require percpu variables in modules to be
986      defined weak to work around addressing range issue which
987      puts the following two restrictions on percpu variable
988      definitions.
989
990      1. percpu symbols must be unique whether static or not
991      2. percpu variables can't be defined inside a function
992
993      To ensure that generic code follows the above rules, this
994      option forces all percpu variables to be defined as weak.
995
996config LKDTM
997    tristate "Linux Kernel Dump Test Tool Module"
998    depends on DEBUG_FS
999    depends on BLOCK
1000    default n
1001    help
1002    This module enables testing of the different dumping mechanisms by
1003    inducing system failures at predefined crash points.
1004    If you don't need it: say N
1005    Choose M here to compile this code as a module. The module will be
1006    called lkdtm.
1007
1008    Documentation on how to use the module can be found in
1009    Documentation/fault-injection/provoke-crashes.txt
1010
1011config CPU_NOTIFIER_ERROR_INJECT
1012    tristate "CPU notifier error injection module"
1013    depends on HOTPLUG_CPU && DEBUG_KERNEL
1014    help
1015      This option provides a kernel module that can be used to test
1016      the error handling of the cpu notifiers
1017
1018      To compile this code as a module, choose M here: the module will
1019      be called cpu-notifier-error-inject.
1020
1021      If unsure, say N.
1022
1023config FAULT_INJECTION
1024    bool "Fault-injection framework"
1025    depends on DEBUG_KERNEL
1026    help
1027      Provide fault-injection framework.
1028      For more details, see Documentation/fault-injection/.
1029
1030config FAILSLAB
1031    bool "Fault-injection capability for kmalloc"
1032    depends on FAULT_INJECTION
1033    depends on SLAB || SLUB
1034    help
1035      Provide fault-injection capability for kmalloc.
1036
1037config FAIL_PAGE_ALLOC
1038    bool "Fault-injection capabilitiy for alloc_pages()"
1039    depends on FAULT_INJECTION
1040    help
1041      Provide fault-injection capability for alloc_pages().
1042
1043config FAIL_MAKE_REQUEST
1044    bool "Fault-injection capability for disk IO"
1045    depends on FAULT_INJECTION && BLOCK
1046    help
1047      Provide fault-injection capability for disk IO.
1048
1049config FAIL_IO_TIMEOUT
1050    bool "Fault-injection capability for faking disk interrupts"
1051    depends on FAULT_INJECTION && BLOCK
1052    help
1053      Provide fault-injection capability on end IO handling. This
1054      will make the block layer "forget" an interrupt as configured,
1055      thus exercising the error handling.
1056
1057      Only works with drivers that use the generic timeout handling,
1058      for others it wont do anything.
1059
1060config FAULT_INJECTION_DEBUG_FS
1061    bool "Debugfs entries for fault-injection capabilities"
1062    depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1063    help
1064      Enable configuration of fault-injection capabilities via debugfs.
1065
1066config FAULT_INJECTION_STACKTRACE_FILTER
1067    bool "stacktrace filter for fault-injection capabilities"
1068    depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1069    depends on !X86_64
1070    select STACKTRACE
1071    select FRAME_POINTER if !PPC && !S390 && !MICROBLAZE
1072    help
1073      Provide stacktrace filter for fault-injection capabilities
1074
1075config LATENCYTOP
1076    bool "Latency measuring infrastructure"
1077    depends on HAVE_LATENCYTOP_SUPPORT
1078    depends on DEBUG_KERNEL
1079    depends on STACKTRACE_SUPPORT
1080    depends on PROC_FS
1081    select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE
1082    select KALLSYMS
1083    select KALLSYMS_ALL
1084    select STACKTRACE
1085    select SCHEDSTATS
1086    select SCHED_DEBUG
1087    help
1088      Enable this option if you want to use the LatencyTOP tool
1089      to find out which userspace is blocking on what kernel operations.
1090
1091config SYSCTL_SYSCALL_CHECK
1092    bool "Sysctl checks"
1093    depends on SYSCTL
1094    ---help---
1095      sys_sysctl uses binary paths that have been found challenging
1096      to properly maintain and use. This enables checks that help
1097      you to keep things correct.
1098
1099source mm/Kconfig.debug
1100source kernel/trace/Kconfig
1101
1102config PROVIDE_OHCI1394_DMA_INIT
1103    bool "Remote debugging over FireWire early on boot"
1104    depends on PCI && X86
1105    help
1106      If you want to debug problems which hang or crash the kernel early
1107      on boot and the crashing machine has a FireWire port, you can use
1108      this feature to remotely access the memory of the crashed machine
1109      over FireWire. This employs remote DMA as part of the OHCI1394
1110      specification which is now the standard for FireWire controllers.
1111
1112      With remote DMA, you can monitor the printk buffer remotely using
1113      firescope and access all memory below 4GB using fireproxy from gdb.
1114      Even controlling a kernel debugger is possible using remote DMA.
1115
1116      Usage:
1117
1118      If ohci1394_dma=early is used as boot parameter, it will initialize
1119      all OHCI1394 controllers which are found in the PCI config space.
1120
1121      As all changes to the FireWire bus such as enabling and disabling
1122      devices cause a bus reset and thereby disable remote DMA for all
1123      devices, be sure to have the cable plugged and FireWire enabled on
1124      the debugging host before booting the debug target for debugging.
1125
1126      This code (~1k) is freed after boot. By then, the firewire stack
1127      in charge of the OHCI-1394 controllers should be used instead.
1128
1129      See Documentation/debugging-via-ohci1394.txt for more information.
1130
1131config FIREWIRE_OHCI_REMOTE_DMA
1132    bool "Remote debugging over FireWire with firewire-ohci"
1133    depends on FIREWIRE_OHCI
1134    help
1135      This option lets you use the FireWire bus for remote debugging
1136      with help of the firewire-ohci driver. It enables unfiltered
1137      remote DMA in firewire-ohci.
1138      See Documentation/debugging-via-ohci1394.txt for more information.
1139
1140      If unsure, say N.
1141
1142config BUILD_DOCSRC
1143    bool "Build targets in Documentation/ tree"
1144    depends on HEADERS_CHECK
1145    help
1146      This option attempts to build objects from the source files in the
1147      kernel Documentation/ tree.
1148
1149      Say N if you are unsure.
1150
1151config DYNAMIC_DEBUG
1152    bool "Enable dynamic printk() support"
1153    default n
1154    depends on PRINTK
1155    depends on DEBUG_FS
1156    help
1157
1158      Compiles debug level messages into the kernel, which would not
1159      otherwise be available at runtime. These messages can then be
1160      enabled/disabled based on various levels of scope - per source file,
1161      function, module, format string, and line number. This mechanism
1162      implicitly enables all pr_debug() and dev_dbg() calls. The impact of
1163      this compile option is a larger kernel text size of about 2%.
1164
1165      Usage:
1166
1167      Dynamic debugging is controlled via the 'dynamic_debug/control' file,
1168      which is contained in the 'debugfs' filesystem. Thus, the debugfs
1169      filesystem must first be mounted before making use of this feature.
1170      We refer the control file as: <debugfs>/dynamic_debug/control. This
1171      file contains a list of the debug statements that can be enabled. The
1172      format for each line of the file is:
1173
1174        filename:lineno [module]function flags format
1175
1176      filename : source file of the debug statement
1177      lineno : line number of the debug statement
1178      module : module that contains the debug statement
1179      function : function that contains the debug statement
1180          flags : 'p' means the line is turned 'on' for printing
1181          format : the format used for the debug statement
1182
1183      From a live system:
1184
1185        nullarbor:~ # cat <debugfs>/dynamic_debug/control
1186        # filename:lineno [module]function flags format
1187        fs/aio.c:222 [aio]__put_ioctx - "__put_ioctx:\040freeing\040%p\012"
1188        fs/aio.c:248 [aio]ioctx_alloc - "ENOMEM:\040nr_events\040too\040high\012"
1189        fs/aio.c:1770 [aio]sys_io_cancel - "calling\040cancel\012"
1190
1191      Example usage:
1192
1193        // enable the message at line 1603 of file svcsock.c
1194        nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
1195                        <debugfs>/dynamic_debug/control
1196
1197        // enable all the messages in file svcsock.c
1198        nullarbor:~ # echo -n 'file svcsock.c +p' >
1199                        <debugfs>/dynamic_debug/control
1200
1201        // enable all the messages in the NFS server module
1202        nullarbor:~ # echo -n 'module nfsd +p' >
1203                        <debugfs>/dynamic_debug/control
1204
1205        // enable all 12 messages in the function svc_process()
1206        nullarbor:~ # echo -n 'func svc_process +p' >
1207                        <debugfs>/dynamic_debug/control
1208
1209        // disable all 12 messages in the function svc_process()
1210        nullarbor:~ # echo -n 'func svc_process -p' >
1211                        <debugfs>/dynamic_debug/control
1212
1213      See Documentation/dynamic-debug-howto.txt for additional information.
1214
1215config DMA_API_DEBUG
1216    bool "Enable debugging of DMA-API usage"
1217    depends on HAVE_DMA_API_DEBUG
1218    help
1219      Enable this option to debug the use of the DMA API by device drivers.
1220      With this option you will be able to detect common bugs in device
1221      drivers like double-freeing of DMA mappings or freeing mappings that
1222      were never allocated.
1223      This option causes a performance degredation. Use only if you want
1224      to debug device drivers. If unsure, say N.
1225
1226config ATOMIC64_SELFTEST
1227    bool "Perform an atomic64_t self-test at boot"
1228    help
1229      Enable this option to test the atomic64_t functions at boot.
1230
1231      If unsure, say N.
1232
1233config ASYNC_RAID6_TEST
1234    tristate "Self test for hardware accelerated raid6 recovery"
1235    depends on ASYNC_RAID6_RECOV
1236    select ASYNC_MEMCPY
1237    ---help---
1238      This is a one-shot self test that permutes through the
1239      recovery of all the possible two disk failure scenarios for a
1240      N-disk array. Recovery is performed with the asynchronous
1241      raid6 recovery routines, and will optionally use an offload
1242      engine if one is available.
1243
1244      If unsure, say N.
1245
1246source "samples/Kconfig"
1247
1248source "lib/Kconfig.kgdb"
1249
1250source "lib/Kconfig.kmemcheck"
1251
1252config TEST_KSTRTOX
1253    tristate "Test kstrto*() family of functions at runtime"
1254

Archive Download this file



interactive