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 ENABLE_WARN_DEPRECATED
13    bool "Enable __deprecated logic"
14    default y
15    help
16      Enable the __deprecated logic in the kernel build.
17      Disable this to suppress the "warning: 'foo' is deprecated
18      (declared at kernel/power/somefile.c:1234)" messages.
19
20config ENABLE_MUST_CHECK
21    bool "Enable __must_check logic"
22    default y
23    help
24      Enable the __must_check logic in the kernel build. Disable this to
25      suppress the "warning: ignoring return value of 'foo', declared with
26      attribute warn_unused_result" messages.
27
28config FRAME_WARN
29    int "Warn for stack frames larger than (needs gcc 4.4)"
30    range 0 8192
31    default 1024 if !64BIT
32    default 2048 if 64BIT
33    help
34      Tell gcc to warn at build time for stack frames larger than this.
35      Setting this too low will cause a lot of warnings.
36      Setting it to 0 disables the warning.
37      Requires gcc 4.4
38
39config MAGIC_SYSRQ
40    bool "Magic SysRq key"
41    depends on !UML
42    help
43      If you say Y here, you will have some control over the system even
44      if the system crashes for example during kernel debugging (e.g., you
45      will be able to flush the buffer cache to disk, reboot the system
46      immediately or dump some status information). This is accomplished
47      by pressing various keys while holding SysRq (Alt+PrintScreen). It
48      also works on a serial console (on PC hardware at least), if you
49      send a BREAK and then within 5 seconds a command keypress. The
50      keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
51      unless you really know what this hack does.
52
53config STRIP_ASM_SYMS
54    bool "Strip assembler-generated symbols during link"
55    default n
56    help
57      Strip internal assembler-generated symbols during a link (symbols
58      that look like '.Lxxx') so they don't pollute the output of
59      get_wchan() and suchlike.
60
61config UNUSED_SYMBOLS
62    bool "Enable unused/obsolete exported symbols"
63    default y if X86
64    help
65      Unused but exported symbols make the kernel needlessly bigger. For
66      that reason most of these unused exports will soon be removed. This
67      option is provided temporarily to provide a transition period in case
68      some external kernel module needs one of these symbols anyway. If you
69      encounter such a case in your module, consider if you are actually
70      using the right API. (rationale: since nobody in the kernel is using
71      this in a module, there is a pretty good chance it's actually the
72      wrong interface to use). If you really need the symbol, please send a
73      mail to the linux kernel mailing list mentioning the symbol and why
74      you really need it, and what the merge plan to the mainline kernel for
75      your module is.
76
77config DEBUG_FS
78    bool "Debug Filesystem"
79    help
80      debugfs is a virtual file system that kernel developers use to put
81      debugging files into. Enable this option to be able to read and
82      write to these files.
83
84      For detailed documentation on the debugfs API, see
85      Documentation/DocBook/filesystems.
86
87      If unsure, say N.
88
89config HEADERS_CHECK
90    bool "Run 'make headers_check' when building vmlinux"
91    depends on !UML
92    help
93      This option will extract the user-visible kernel headers whenever
94      building the kernel, and will run basic sanity checks on them to
95      ensure that exported files do not attempt to include files which
96      were not exported, etc.
97
98      If you're making modifications to header files which are
99      relevant for userspace, say 'Y', and check the headers
100      exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
101      your build tree), to make sure they're suitable.
102
103config DEBUG_SECTION_MISMATCH
104    bool "Enable full Section mismatch analysis"
105    depends on UNDEFINED || (BLACKFIN)
106    default y
107    # This option is on purpose disabled for now.
108    # It will be enabled when we are down to a reasonable number
109    # of section mismatch warnings (< 10 for an allyesconfig build)
110    help
111      The section mismatch analysis checks if there are illegal
112      references from one section to another section.
113      Linux will during link or during runtime drop some sections
114      and any use of code/data previously in these sections will
115      most likely result in an oops.
116      In the code functions and variables are annotated with
117      __init, __devinit etc. (see full list in include/linux/init.h)
118      which results in the code/data being placed in specific sections.
119      The section mismatch analysis is always done after a full
120      kernel build but enabling this option will in addition
121      do the following:
122      - Add the option -fno-inline-functions-called-once to gcc
123        When inlining a function annotated __init in a non-init
124        function we would lose the section information and thus
125        the analysis would not catch the illegal reference.
126        This option tells gcc to inline less but will also
127        result in a larger kernel.
128      - Run the section mismatch analysis for each module/built-in.o
129        When we run the section mismatch analysis on vmlinux.o we
130        lose valueble information about where the mismatch was
131        introduced.
132        Running the analysis for each module/built-in.o file
133        will tell where the mismatch happens much closer to the
134        source. The drawback is that we will report the same
135        mismatch at least twice.
136      - Enable verbose reporting from modpost to help solving
137        the section mismatches reported.
138
139config DEBUG_KERNEL
140    bool "Kernel debugging"
141    help
142      Say Y here if you are developing drivers or trying to debug and
143      identify kernel problems.
144
145config DEBUG_SHIRQ
146    bool "Debug shared IRQ handlers"
147    depends on DEBUG_KERNEL && GENERIC_HARDIRQS
148    help
149      Enable this to generate a spurious interrupt as soon as a shared
150      interrupt handler is registered, and just before one is deregistered.
151      Drivers ought to be able to handle interrupts coming in at those
152      points; some don't and need to be caught.
153
154config LOCKUP_DETECTOR
155    bool "Detect Hard and Soft Lockups"
156    depends on DEBUG_KERNEL && !S390
157    help
158      Say Y here to enable the kernel to act as a watchdog to detect
159      hard and soft lockups.
160
161      Softlockups are bugs that cause the kernel to loop in kernel
162      mode for more than 60 seconds, without giving other tasks a
163      chance to run. The current stack trace is displayed upon
164      detection and the system will stay locked up.
165
166      Hardlockups are bugs that cause the CPU to loop in kernel mode
167      for more than 60 seconds, without letting other interrupts have a
168      chance to run. The current stack trace is displayed upon detection
169      and the system will stay locked up.
170
171      The overhead should be minimal. A periodic hrtimer runs to
172      generate interrupts and kick the watchdog task every 10-12 seconds.
173      An NMI is generated every 60 seconds or so to check for hardlockups.
174
175config HARDLOCKUP_DETECTOR
176    def_bool LOCKUP_DETECTOR && PERF_EVENTS && HAVE_PERF_EVENTS_NMI
177
178config BOOTPARAM_SOFTLOCKUP_PANIC
179    bool "Panic (Reboot) On Soft Lockups"
180    depends on LOCKUP_DETECTOR
181    help
182      Say Y here to enable the kernel to panic on "soft lockups",
183      which are bugs that cause the kernel to loop in kernel
184      mode for more than 60 seconds, without giving other tasks a
185      chance to run.
186
187      The panic can be used in combination with panic_timeout,
188      to cause the system to reboot automatically after a
189      lockup has been detected. This feature is useful for
190      high-availability systems that have uptime guarantees and
191      where a lockup must be resolved ASAP.
192
193      Say N if unsure.
194
195config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
196    int
197    depends on LOCKUP_DETECTOR
198    range 0 1
199    default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
200    default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
201
202config DETECT_HUNG_TASK
203    bool "Detect Hung Tasks"
204    depends on DEBUG_KERNEL
205    default DETECT_SOFTLOCKUP
206    help
207      Say Y here to enable the kernel to detect "hung tasks",
208      which are bugs that cause the task to be stuck in
209      uninterruptible "D" state indefinitiley.
210
211      When a hung task is detected, the kernel will print the
212      current stack trace (which you should report), but the
213      task will stay in uninterruptible state. If lockdep is
214      enabled then all held locks will also be reported. This
215      feature has negligible overhead.
216
217config BOOTPARAM_HUNG_TASK_PANIC
218    bool "Panic (Reboot) On Hung Tasks"
219    depends on DETECT_HUNG_TASK
220    help
221      Say Y here to enable the kernel to panic on "hung tasks",
222      which are bugs that cause the kernel to leave a task stuck
223      in uninterruptible "D" state.
224
225      The panic can be used in combination with panic_timeout,
226      to cause the system to reboot automatically after a
227      hung task has been detected. This feature is useful for
228      high-availability systems that have uptime guarantees and
229      where a hung tasks must be resolved ASAP.
230
231      Say N if unsure.
232
233config BOOTPARAM_HUNG_TASK_PANIC_VALUE
234    int
235    depends on DETECT_HUNG_TASK
236    range 0 1
237    default 0 if !BOOTPARAM_HUNG_TASK_PANIC
238    default 1 if BOOTPARAM_HUNG_TASK_PANIC
239
240config SCHED_DEBUG
241    bool "Collect scheduler debugging info"
242    depends on DEBUG_KERNEL && PROC_FS
243    default y
244    help
245      If you say Y here, the /proc/sched_debug file will be provided
246      that can help debug the scheduler. The runtime overhead of this
247      option is minimal.
248
249config SCHEDSTATS
250    bool "Collect scheduler statistics"
251    depends on DEBUG_KERNEL && PROC_FS
252    help
253      If you say Y here, additional code will be inserted into the
254      scheduler and related routines to collect statistics about
255      scheduler behavior and provide them in /proc/schedstat. These
256      stats may be useful for both tuning and debugging the scheduler
257      If you aren't debugging the scheduler or trying to tune a specific
258      application, you can say N to avoid the very slight overhead
259      this adds.
260
261config TIMER_STATS
262    bool "Collect kernel timers statistics"
263    depends on DEBUG_KERNEL && PROC_FS
264    help
265      If you say Y here, additional code will be inserted into the
266      timer routines to collect statistics about kernel timers being
267      reprogrammed. The statistics can be read from /proc/timer_stats.
268      The statistics collection is started by writing 1 to /proc/timer_stats,
269      writing 0 stops it. This feature is useful to collect information
270      about timer usage patterns in kernel and userspace. This feature
271      is lightweight if enabled in the kernel config but not activated
272      (it defaults to deactivated on bootup and will only be activated
273      if some application like powertop activates it explicitly).
274
275config DEBUG_OBJECTS
276    bool "Debug object operations"
277    depends on DEBUG_KERNEL
278    help
279      If you say Y here, additional code will be inserted into the
280      kernel to track the life time of various objects and validate
281      the operations on those objects.
282
283config DEBUG_OBJECTS_SELFTEST
284    bool "Debug objects selftest"
285    depends on DEBUG_OBJECTS
286    help
287      This enables the selftest of the object debug code.
288
289config DEBUG_OBJECTS_FREE
290    bool "Debug objects in freed memory"
291    depends on DEBUG_OBJECTS
292    help
293      This enables checks whether a k/v free operation frees an area
294      which contains an object which has not been deactivated
295      properly. This can make kmalloc/kfree-intensive workloads
296      much slower.
297
298config DEBUG_OBJECTS_TIMERS
299    bool "Debug timer objects"
300    depends on DEBUG_OBJECTS
301    help
302      If you say Y here, additional code will be inserted into the
303      timer routines to track the life time of timer objects and
304      validate the timer operations.
305
306config DEBUG_OBJECTS_WORK
307    bool "Debug work objects"
308    depends on DEBUG_OBJECTS
309    help
310      If you say Y here, additional code will be inserted into the
311      work queue routines to track the life time of work objects and
312      validate the work operations.
313
314config DEBUG_OBJECTS_RCU_HEAD
315    bool "Debug RCU callbacks objects"
316    depends on DEBUG_OBJECTS && PREEMPT
317    help
318      Enable this to turn on debugging of RCU list heads (call_rcu() usage).
319
320config DEBUG_OBJECTS_ENABLE_DEFAULT
321    int "debug_objects bootup default value (0-1)"
322        range 0 1
323        default "1"
324        depends on DEBUG_OBJECTS
325        help
326          Debug objects boot parameter default value
327
328config DEBUG_SLAB
329    bool "Debug slab memory allocations"
330    depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
331    help
332      Say Y here to have the kernel do limited verification on memory
333      allocation as well as poisoning memory on free to catch use of freed
334      memory. This can make kmalloc/kfree-intensive workloads much slower.
335
336config DEBUG_SLAB_LEAK
337    bool "Memory leak debugging"
338    depends on DEBUG_SLAB
339
340config SLUB_DEBUG_ON
341    bool "SLUB debugging on by default"
342    depends on SLUB && SLUB_DEBUG && !KMEMCHECK
343    default n
344    help
345      Boot with debugging on by default. SLUB boots by default with
346      the runtime debug capabilities switched off. Enabling this is
347      equivalent to specifying the "slub_debug" parameter on boot.
348      There is no support for more fine grained debug control like
349      possible with slub_debug=xxx. SLUB debugging may be switched
350      off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
351      "slub_debug=-".
352
353config SLUB_STATS
354    default n
355    bool "Enable SLUB performance statistics"
356    depends on SLUB && SLUB_DEBUG && SYSFS
357    help
358      SLUB statistics are useful to debug SLUBs allocation behavior in
359      order find ways to optimize the allocator. This should never be
360      enabled for production use since keeping statistics slows down
361      the allocator by a few percentage points. The slabinfo command
362      supports the determination of the most active slabs to figure
363      out which slabs are relevant to a particular load.
364      Try running: slabinfo -DA
365
366config DEBUG_KMEMLEAK
367    bool "Kernel memory leak detector"
368    depends on DEBUG_KERNEL && EXPERIMENTAL && !MEMORY_HOTPLUG && \
369        (X86 || ARM || PPC || S390 || SPARC64 || SUPERH || MICROBLAZE)
370
371    select DEBUG_FS if SYSFS
372    select STACKTRACE if STACKTRACE_SUPPORT
373    select KALLSYMS
374    select CRC32
375    help
376      Say Y here if you want to enable the memory leak
377      detector. The memory allocation/freeing is traced in a way
378      similar to the Boehm's conservative garbage collector, the
379      difference being that the orphan objects are not freed but
380      only shown in /sys/kernel/debug/kmemleak. Enabling this
381      feature will introduce an overhead to memory
382      allocations. See Documentation/kmemleak.txt for more
383      details.
384
385      Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
386      of finding leaks due to the slab objects poisoning.
387
388      In order to access the kmemleak file, debugfs needs to be
389      mounted (usually at /sys/kernel/debug).
390
391config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
392    int "Maximum kmemleak early log entries"
393    depends on DEBUG_KMEMLEAK
394    range 200 40000
395    default 400
396    help
397      Kmemleak must track all the memory allocations to avoid
398      reporting false positives. Since memory may be allocated or
399      freed before kmemleak is initialised, an early log buffer is
400      used to store these actions. If kmemleak reports "early log
401      buffer exceeded", please increase this value.
402
403config DEBUG_KMEMLEAK_TEST
404    tristate "Simple test for the kernel memory leak detector"
405    depends on DEBUG_KMEMLEAK
406    help
407      Say Y or M here to build a test for the kernel memory leak
408      detector. This option enables a module that explicitly leaks
409      memory.
410
411      If unsure, say N.
412
413config DEBUG_KMEMLEAK_DEFAULT_OFF
414    bool "Default kmemleak to off"
415    depends on DEBUG_KMEMLEAK
416    help
417      Say Y here to disable kmemleak by default. It can then be enabled
418      on the command line via kmemleak=on.
419
420config DEBUG_PREEMPT
421    bool "Debug preemptible kernel"
422    depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
423    default y
424    help
425      If you say Y here then the kernel will use a debug variant of the
426      commonly used smp_processor_id() function and will print warnings
427      if kernel code uses it in a preemption-unsafe way. Also, the kernel
428      will detect preemption count underflows.
429
430config DEBUG_RT_MUTEXES
431    bool "RT Mutex debugging, deadlock detection"
432    depends on DEBUG_KERNEL && RT_MUTEXES
433    help
434     This allows rt mutex semantics violations and rt mutex related
435     deadlocks (lockups) to be detected and reported automatically.
436
437config DEBUG_PI_LIST
438    bool
439    default y
440    depends on DEBUG_RT_MUTEXES
441
442config RT_MUTEX_TESTER
443    bool "Built-in scriptable tester for rt-mutexes"
444    depends on DEBUG_KERNEL && RT_MUTEXES
445    help
446      This option enables a rt-mutex tester.
447
448config DEBUG_SPINLOCK
449    bool "Spinlock and rw-lock debugging: basic checks"
450    depends on DEBUG_KERNEL
451    help
452      Say Y here and build SMP to catch missing spinlock initialization
453      and certain other kinds of spinlock errors commonly made. This is
454      best used in conjunction with the NMI watchdog so that spinlock
455      deadlocks are also debuggable.
456
457config DEBUG_MUTEXES
458    bool "Mutex debugging: basic checks"
459    depends on DEBUG_KERNEL
460    help
461     This feature allows mutex semantics violations to be detected and
462     reported.
463
464config DEBUG_LOCK_ALLOC
465    bool "Lock debugging: detect incorrect freeing of live locks"
466    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
467    select DEBUG_SPINLOCK
468    select DEBUG_MUTEXES
469    select LOCKDEP
470    help
471     This feature will check whether any held lock (spinlock, rwlock,
472     mutex or rwsem) is incorrectly freed by the kernel, via any of the
473     memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
474     vfree(), etc.), whether a live lock is incorrectly reinitialized via
475     spin_lock_init()/mutex_init()/etc., or whether there is any lock
476     held during task exit.
477
478config PROVE_LOCKING
479    bool "Lock debugging: prove locking correctness"
480    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
481    select LOCKDEP
482    select DEBUG_SPINLOCK
483    select DEBUG_MUTEXES
484    select DEBUG_LOCK_ALLOC
485    default n
486    help
487     This feature enables the kernel to prove that all locking
488     that occurs in the kernel runtime is mathematically
489     correct: that under no circumstance could an arbitrary (and
490     not yet triggered) combination of observed locking
491     sequences (on an arbitrary number of CPUs, running an
492     arbitrary number of tasks and interrupt contexts) cause a
493     deadlock.
494
495     In short, this feature enables the kernel to report locking
496     related deadlocks before they actually occur.
497
498     The proof does not depend on how hard and complex a
499     deadlock scenario would be to trigger: how many
500     participant CPUs, tasks and irq-contexts would be needed
501     for it to trigger. The proof also does not depend on
502     timing: if a race and a resulting deadlock is possible
503     theoretically (no matter how unlikely the race scenario
504     is), it will be proven so and will immediately be
505     reported by the kernel (once the event is observed that
506     makes the deadlock theoretically possible).
507
508     If a deadlock is impossible (i.e. the locking rules, as
509     observed by the kernel, are mathematically correct), the
510     kernel reports nothing.
511
512     NOTE: this feature can also be enabled for rwlocks, mutexes
513     and rwsems - in which case all dependencies between these
514     different locking variants are observed and mapped too, and
515     the proof of observed correctness is also maintained for an
516     arbitrary combination of these separate locking variants.
517
518     For more details, see Documentation/lockdep-design.txt.
519
520config PROVE_RCU
521    bool "RCU debugging: prove RCU correctness"
522    depends on PROVE_LOCKING
523    default n
524    help
525     This feature enables lockdep extensions that check for correct
526     use of RCU APIs. This is currently under development. Say Y
527     if you want to debug RCU usage or help work on the PROVE_RCU
528     feature.
529
530     Say N if you are unsure.
531
532config PROVE_RCU_REPEATEDLY
533    bool "RCU debugging: don't disable PROVE_RCU on first splat"
534    depends on PROVE_RCU
535    default n
536    help
537     By itself, PROVE_RCU will disable checking upon issuing the
538     first warning (or "splat"). This feature prevents such
539     disabling, allowing multiple RCU-lockdep warnings to be printed
540     on a single reboot.
541
542     Say N if you are unsure.
543
544config LOCKDEP
545    bool
546    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
547    select STACKTRACE
548    select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE
549    select KALLSYMS
550    select KALLSYMS_ALL
551
552config LOCK_STAT
553    bool "Lock usage statistics"
554    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
555    select LOCKDEP
556    select DEBUG_SPINLOCK
557    select DEBUG_MUTEXES
558    select DEBUG_LOCK_ALLOC
559    default n
560    help
561     This feature enables tracking lock contention points
562
563     For more details, see Documentation/lockstat.txt
564
565     This also enables lock events required by "perf lock",
566     subcommand of perf.
567     If you want to use "perf lock", you also need to turn on
568     CONFIG_EVENT_TRACING.
569
570     CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
571     (CONFIG_LOCKDEP defines "acquire" and "release" events.)
572
573config DEBUG_LOCKDEP
574    bool "Lock dependency engine debugging"
575    depends on DEBUG_KERNEL && LOCKDEP
576    help
577      If you say Y here, the lock dependency engine will do
578      additional runtime checks to debug itself, at the price
579      of more runtime overhead.
580
581config TRACE_IRQFLAGS
582    depends on DEBUG_KERNEL
583    bool
584    default y
585    depends on TRACE_IRQFLAGS_SUPPORT
586    depends on PROVE_LOCKING
587
588config DEBUG_SPINLOCK_SLEEP
589    bool "Spinlock debugging: sleep-inside-spinlock checking"
590    depends on DEBUG_KERNEL
591    help
592      If you say Y here, various routines which may sleep will become very
593      noisy if they are called with a spinlock held.
594
595config DEBUG_LOCKING_API_SELFTESTS
596    bool "Locking API boot-time self-tests"
597    depends on DEBUG_KERNEL
598    help
599      Say Y here if you want the kernel to run a short self-test during
600      bootup. The self-test checks whether common types of locking bugs
601      are detected by debugging mechanisms or not. (if you disable
602      lock debugging then those bugs wont be detected of course.)
603      The following locking APIs are covered: spinlocks, rwlocks,
604      mutexes and rwsems.
605
606config STACKTRACE
607    bool
608    depends on STACKTRACE_SUPPORT
609
610config DEBUG_KOBJECT
611    bool "kobject debugging"
612    depends on DEBUG_KERNEL
613    help
614      If you say Y here, some extra kobject debugging messages will be sent
615      to the syslog.
616
617config DEBUG_HIGHMEM
618    bool "Highmem debugging"
619    depends on DEBUG_KERNEL && HIGHMEM
620    help
621      This options enables addition error checking for high memory systems.
622      Disable for production systems.
623
624config DEBUG_BUGVERBOSE
625    bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EMBEDDED
626    depends on BUG
627    depends on ARM || AVR32 || M32R || M68K || SPARC32 || SPARC64 || \
628           FRV || SUPERH || GENERIC_BUG || BLACKFIN || MN10300
629    default y
630    help
631      Say Y here to make BUG() panics output the file name and line number
632      of the BUG call as well as the EIP and oops trace. This aids
633      debugging but costs about 70-100K of memory.
634
635config DEBUG_INFO
636    bool "Compile the kernel with debug info"
637    depends on DEBUG_KERNEL
638    help
639          If you say Y here the resulting kernel image will include
640      debugging info resulting in a larger kernel image.
641      This adds debug symbols to the kernel and modules (gcc -g), and
642      is needed if you intend to use kernel crashdump or binary object
643      tools like crash, kgdb, LKCD, gdb, etc on the kernel.
644      Say Y here only if you plan to debug the kernel.
645
646      If unsure, say N.
647
648config DEBUG_INFO_REDUCED
649    bool "Reduce debugging information"
650    depends on DEBUG_INFO
651    help
652      If you say Y here gcc is instructed to generate less debugging
653      information for structure types. This means that tools that
654      need full debugging information (like kgdb or systemtap) won't
655      be happy. But if you merely need debugging information to
656      resolve line numbers there is no loss. Advantage is that
657      build directory object sizes shrink dramatically over a full
658      DEBUG_INFO build and compile times are reduced too.
659      Only works with newer gcc versions.
660
661config DEBUG_VM
662    bool "Debug VM"
663    depends on DEBUG_KERNEL
664    help
665      Enable this to turn on extended checks in the virtual-memory system
666          that may impact performance.
667
668      If unsure, say N.
669
670config DEBUG_VIRTUAL
671    bool "Debug VM translations"
672    depends on DEBUG_KERNEL && X86
673    help
674      Enable some costly sanity checks in virtual to page code. This can
675      catch mistakes with virt_to_page() and friends.
676
677      If unsure, say N.
678
679config DEBUG_NOMMU_REGIONS
680    bool "Debug the global anon/private NOMMU mapping region tree"
681    depends on DEBUG_KERNEL && !MMU
682    help
683      This option causes the global tree of anonymous and private mapping
684      regions to be regularly checked for invalid topology.
685
686config DEBUG_WRITECOUNT
687    bool "Debug filesystem writers count"
688    depends on DEBUG_KERNEL
689    help
690      Enable this to catch wrong use of the writers count in struct
691      vfsmount. This will increase the size of each file struct by
692      32 bits.
693
694      If unsure, say N.
695
696config DEBUG_MEMORY_INIT
697    bool "Debug memory initialisation" if EMBEDDED
698    default !EMBEDDED
699    help
700      Enable this for additional checks during memory initialisation.
701      The sanity checks verify aspects of the VM such as the memory model
702      and other information provided by the architecture. Verbose
703      information will be printed at KERN_DEBUG loglevel depending
704      on the mminit_loglevel= command-line option.
705
706      If unsure, say Y
707
708config DEBUG_LIST
709    bool "Debug linked list manipulation"
710    depends on DEBUG_KERNEL
711    help
712      Enable this to turn on extended checks in the linked-list
713      walking routines.
714
715      If unsure, say N.
716
717config DEBUG_SG
718    bool "Debug SG table operations"
719    depends on DEBUG_KERNEL
720    help
721      Enable this to turn on checks on scatter-gather tables. This can
722      help find problems with drivers that do not properly initialize
723      their sg tables.
724
725      If unsure, say N.
726
727config DEBUG_NOTIFIERS
728    bool "Debug notifier call chains"
729    depends on DEBUG_KERNEL
730    help
731      Enable this to turn on sanity checking for notifier call chains.
732      This is most useful for kernel developers to make sure that
733      modules properly unregister themselves from notifier chains.
734      This is a relatively cheap check but if you care about maximum
735      performance, say N.
736
737config DEBUG_CREDENTIALS
738    bool "Debug credential management"
739    depends on DEBUG_KERNEL
740    help
741      Enable this to turn on some debug checking for credential
742      management. The additional code keeps track of the number of
743      pointers from task_structs to any given cred struct, and checks to
744      see that this number never exceeds the usage count of the cred
745      struct.
746
747      Furthermore, if SELinux is enabled, this also checks that the
748      security pointer in the cred struct is never seen to be invalid.
749
750      If unsure, say N.
751
752#
753# Select this config option from the architecture Kconfig, if it
754# it is preferred to always offer frame pointers as a config
755# option on the architecture (regardless of KERNEL_DEBUG):
756#
757config ARCH_WANT_FRAME_POINTERS
758    bool
759    help
760
761config FRAME_POINTER
762    bool "Compile the kernel with frame pointers"
763    depends on DEBUG_KERNEL && \
764        (CRIS || M68K || M68KNOMMU || FRV || UML || \
765         AVR32 || SUPERH || BLACKFIN || MN10300) || \
766        ARCH_WANT_FRAME_POINTERS
767    default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
768    help
769      If you say Y here the resulting kernel image will be slightly
770      larger and slower, but it gives very useful debugging information
771      in case of kernel bugs. (precise oopses/stacktraces/warnings)
772
773config BOOT_PRINTK_DELAY
774    bool "Delay each boot printk message by N milliseconds"
775    depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
776    help
777      This build option allows you to read kernel boot messages
778      by inserting a short delay after each one. The delay is
779      specified in milliseconds on the kernel command line,
780      using "boot_delay=N".
781
782      It is likely that you would also need to use "lpj=M" to preset
783      the "loops per jiffie" value.
784      See a previous boot log for the "lpj" value to use for your
785      system, and then set "lpj=M" before setting "boot_delay=N".
786      NOTE: Using this option may adversely affect SMP systems.
787      I.e., processors other than the first one may not boot up.
788      BOOT_PRINTK_DELAY also may cause DETECT_SOFTLOCKUP to detect
789      what it believes to be lockup conditions.
790
791config RCU_TORTURE_TEST
792    tristate "torture tests for RCU"
793    depends on DEBUG_KERNEL
794    default n
795    help
796      This option provides a kernel module that runs torture tests
797      on the RCU infrastructure. The kernel module may be built
798      after the fact on the running kernel to be tested, if desired.
799
800      Say Y here if you want RCU torture tests to be built into
801      the kernel.
802      Say M if you want the RCU torture tests to build as a module.
803      Say N if you are unsure.
804
805config RCU_TORTURE_TEST_RUNNABLE
806    bool "torture tests for RCU runnable by default"
807    depends on RCU_TORTURE_TEST = y
808    default n
809    help
810      This option provides a way to build the RCU torture tests
811      directly into the kernel without them starting up at boot
812      time. You can use /proc/sys/kernel/rcutorture_runnable
813      to manually override this setting. This /proc file is
814      available only when the RCU torture tests have been built
815      into the kernel.
816
817      Say Y here if you want the RCU torture tests to start during
818      boot (you probably don't).
819      Say N here if you want the RCU torture tests to start only
820      after being manually enabled via /proc.
821
822config RCU_CPU_STALL_DETECTOR
823    bool "Check for stalled CPUs delaying RCU grace periods"
824    depends on TREE_RCU || TREE_PREEMPT_RCU
825    default y
826    help
827      This option causes RCU to printk information on which
828      CPUs are delaying the current grace period, but only when
829      the grace period extends for excessive time periods.
830
831      Say N if you want to disable such checks.
832
833      Say Y if you are unsure.
834
835config RCU_CPU_STALL_VERBOSE
836    bool "Print additional per-task information for RCU_CPU_STALL_DETECTOR"
837    depends on RCU_CPU_STALL_DETECTOR && TREE_PREEMPT_RCU
838    default y
839    help
840      This option causes RCU to printk detailed per-task information
841      for any tasks that are stalling the current RCU grace period.
842
843      Say N if you are unsure.
844
845      Say Y if you want to enable such checks.
846
847config KPROBES_SANITY_TEST
848    bool "Kprobes sanity tests"
849    depends on DEBUG_KERNEL
850    depends on KPROBES
851    default n
852    help
853      This option provides for testing basic kprobes functionality on
854      boot. A sample kprobe, jprobe and kretprobe are inserted and
855      verified for functionality.
856
857      Say N if you are unsure.
858
859config BACKTRACE_SELF_TEST
860    tristate "Self test for the backtrace code"
861    depends on DEBUG_KERNEL
862    default n
863    help
864      This option provides a kernel module that can be used to test
865      the kernel stack backtrace code. This option is not useful
866      for distributions or general kernels, but only for kernel
867      developers working on architecture code.
868
869      Note that if you want to also test saved backtraces, you will
870      have to enable STACKTRACE as well.
871
872      Say N if you are unsure.
873
874config DEBUG_BLOCK_EXT_DEVT
875        bool "Force extended block device numbers and spread them"
876    depends on DEBUG_KERNEL
877    depends on BLOCK
878    default n
879    help
880      BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
881      SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
882      YOU ARE DOING. Distros, please enable this and fix whatever
883      is broken.
884
885      Conventionally, block device numbers are allocated from
886      predetermined contiguous area. However, extended block area
887      may introduce non-contiguous block device numbers. This
888      option forces most block device numbers to be allocated from
889      the extended space and spreads them to discover kernel or
890      userland code paths which assume predetermined contiguous
891      device number allocation.
892
893      Note that turning on this debug option shuffles all the
894      device numbers for all IDE and SCSI devices including libata
895      ones, so root partition specified using device number
896      directly (via rdev or root=MAJ:MIN) won't work anymore.
897      Textual device names (root=/dev/sdXn) will continue to work.
898
899      Say N if you are unsure.
900
901config DEBUG_FORCE_WEAK_PER_CPU
902    bool "Force weak per-cpu definitions"
903    depends on DEBUG_KERNEL
904    help
905      s390 and alpha require percpu variables in modules to be
906      defined weak to work around addressing range issue which
907      puts the following two restrictions on percpu variable
908      definitions.
909
910      1. percpu symbols must be unique whether static or not
911      2. percpu variables can't be defined inside a function
912
913      To ensure that generic code follows the above rules, this
914      option forces all percpu variables to be defined as weak.
915
916config LKDTM
917    tristate "Linux Kernel Dump Test Tool Module"
918    depends on DEBUG_FS
919    depends on BLOCK
920    default n
921    help
922    This module enables testing of the different dumping mechanisms by
923    inducing system failures at predefined crash points.
924    If you don't need it: say N
925    Choose M here to compile this code as a module. The module will be
926    called lkdtm.
927
928    Documentation on how to use the module can be found in
929    Documentation/fault-injection/provoke-crashes.txt
930
931config CPU_NOTIFIER_ERROR_INJECT
932    tristate "CPU notifier error injection module"
933    depends on HOTPLUG_CPU && DEBUG_KERNEL
934    help
935      This option provides a kernel module that can be used to test
936      the error handling of the cpu notifiers
937
938      To compile this code as a module, choose M here: the module will
939      be called cpu-notifier-error-inject.
940
941      If unsure, say N.
942
943config FAULT_INJECTION
944    bool "Fault-injection framework"
945    depends on DEBUG_KERNEL
946    help
947      Provide fault-injection framework.
948      For more details, see Documentation/fault-injection/.
949
950config FAILSLAB
951    bool "Fault-injection capability for kmalloc"
952    depends on FAULT_INJECTION
953    depends on SLAB || SLUB
954    help
955      Provide fault-injection capability for kmalloc.
956
957config FAIL_PAGE_ALLOC
958    bool "Fault-injection capabilitiy for alloc_pages()"
959    depends on FAULT_INJECTION
960    help
961      Provide fault-injection capability for alloc_pages().
962
963config FAIL_MAKE_REQUEST
964    bool "Fault-injection capability for disk IO"
965    depends on FAULT_INJECTION && BLOCK
966    help
967      Provide fault-injection capability for disk IO.
968
969config FAIL_IO_TIMEOUT
970    bool "Fault-injection capability for faking disk interrupts"
971    depends on FAULT_INJECTION && BLOCK
972    help
973      Provide fault-injection capability on end IO handling. This
974      will make the block layer "forget" an interrupt as configured,
975      thus exercising the error handling.
976
977      Only works with drivers that use the generic timeout handling,
978      for others it wont do anything.
979
980config FAULT_INJECTION_DEBUG_FS
981    bool "Debugfs entries for fault-injection capabilities"
982    depends on FAULT_INJECTION && SYSFS && DEBUG_FS
983    help
984      Enable configuration of fault-injection capabilities via debugfs.
985
986config FAULT_INJECTION_STACKTRACE_FILTER
987    bool "stacktrace filter for fault-injection capabilities"
988    depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
989    depends on !X86_64
990    select STACKTRACE
991    select FRAME_POINTER if !PPC && !S390 && !MICROBLAZE
992    help
993      Provide stacktrace filter for fault-injection capabilities
994
995config LATENCYTOP
996    bool "Latency measuring infrastructure"
997    depends on HAVE_LATENCYTOP_SUPPORT
998    depends on DEBUG_KERNEL
999    depends on STACKTRACE_SUPPORT
1000    depends on PROC_FS
1001    select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE
1002    select KALLSYMS
1003    select KALLSYMS_ALL
1004    select STACKTRACE
1005    select SCHEDSTATS
1006    select SCHED_DEBUG
1007    help
1008      Enable this option if you want to use the LatencyTOP tool
1009      to find out which userspace is blocking on what kernel operations.
1010
1011config SYSCTL_SYSCALL_CHECK
1012    bool "Sysctl checks"
1013    depends on SYSCTL
1014    ---help---
1015      sys_sysctl uses binary paths that have been found challenging
1016      to properly maintain and use. This enables checks that help
1017      you to keep things correct.
1018
1019source mm/Kconfig.debug
1020source kernel/trace/Kconfig
1021
1022config PROVIDE_OHCI1394_DMA_INIT
1023    bool "Remote debugging over FireWire early on boot"
1024    depends on PCI && X86
1025    help
1026      If you want to debug problems which hang or crash the kernel early
1027      on boot and the crashing machine has a FireWire port, you can use
1028      this feature to remotely access the memory of the crashed machine
1029      over FireWire. This employs remote DMA as part of the OHCI1394
1030      specification which is now the standard for FireWire controllers.
1031
1032      With remote DMA, you can monitor the printk buffer remotely using
1033      firescope and access all memory below 4GB using fireproxy from gdb.
1034      Even controlling a kernel debugger is possible using remote DMA.
1035
1036      Usage:
1037
1038      If ohci1394_dma=early is used as boot parameter, it will initialize
1039      all OHCI1394 controllers which are found in the PCI config space.
1040
1041      As all changes to the FireWire bus such as enabling and disabling
1042      devices cause a bus reset and thereby disable remote DMA for all
1043      devices, be sure to have the cable plugged and FireWire enabled on
1044      the debugging host before booting the debug target for debugging.
1045
1046      This code (~1k) is freed after boot. By then, the firewire stack
1047      in charge of the OHCI-1394 controllers should be used instead.
1048
1049      See Documentation/debugging-via-ohci1394.txt for more information.
1050
1051config FIREWIRE_OHCI_REMOTE_DMA
1052    bool "Remote debugging over FireWire with firewire-ohci"
1053    depends on FIREWIRE_OHCI
1054    help
1055      This option lets you use the FireWire bus for remote debugging
1056      with help of the firewire-ohci driver. It enables unfiltered
1057      remote DMA in firewire-ohci.
1058      See Documentation/debugging-via-ohci1394.txt for more information.
1059
1060      If unsure, say N.
1061
1062config BUILD_DOCSRC
1063    bool "Build targets in Documentation/ tree"
1064    depends on HEADERS_CHECK
1065    help
1066      This option attempts to build objects from the source files in the
1067      kernel Documentation/ tree.
1068
1069      Say N if you are unsure.
1070
1071config DYNAMIC_DEBUG
1072    bool "Enable dynamic printk() support"
1073    default n
1074    depends on PRINTK
1075    depends on DEBUG_FS
1076    help
1077
1078      Compiles debug level messages into the kernel, which would not
1079      otherwise be available at runtime. These messages can then be
1080      enabled/disabled based on various levels of scope - per source file,
1081      function, module, format string, and line number. This mechanism
1082      implicitly enables all pr_debug() and dev_dbg() calls. The impact of
1083      this compile option is a larger kernel text size of about 2%.
1084
1085      Usage:
1086
1087      Dynamic debugging is controlled via the 'dynamic_debug/control' file,
1088      which is contained in the 'debugfs' filesystem. Thus, the debugfs
1089      filesystem must first be mounted before making use of this feature.
1090      We refer the control file as: <debugfs>/dynamic_debug/control. This
1091      file contains a list of the debug statements that can be enabled. The
1092      format for each line of the file is:
1093
1094        filename:lineno [module]function flags format
1095
1096      filename : source file of the debug statement
1097      lineno : line number of the debug statement
1098      module : module that contains the debug statement
1099      function : function that contains the debug statement
1100          flags : 'p' means the line is turned 'on' for printing
1101          format : the format used for the debug statement
1102
1103      From a live system:
1104
1105        nullarbor:~ # cat <debugfs>/dynamic_debug/control
1106        # filename:lineno [module]function flags format
1107        fs/aio.c:222 [aio]__put_ioctx - "__put_ioctx:\040freeing\040%p\012"
1108        fs/aio.c:248 [aio]ioctx_alloc - "ENOMEM:\040nr_events\040too\040high\012"
1109        fs/aio.c:1770 [aio]sys_io_cancel - "calling\040cancel\012"
1110
1111      Example usage:
1112
1113        // enable the message at line 1603 of file svcsock.c
1114        nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
1115                        <debugfs>/dynamic_debug/control
1116
1117        // enable all the messages in file svcsock.c
1118        nullarbor:~ # echo -n 'file svcsock.c +p' >
1119                        <debugfs>/dynamic_debug/control
1120
1121        // enable all the messages in the NFS server module
1122        nullarbor:~ # echo -n 'module nfsd +p' >
1123                        <debugfs>/dynamic_debug/control
1124
1125        // enable all 12 messages in the function svc_process()
1126        nullarbor:~ # echo -n 'func svc_process +p' >
1127                        <debugfs>/dynamic_debug/control
1128
1129        // disable all 12 messages in the function svc_process()
1130        nullarbor:~ # echo -n 'func svc_process -p' >
1131                        <debugfs>/dynamic_debug/control
1132
1133      See Documentation/dynamic-debug-howto.txt for additional information.
1134
1135config DMA_API_DEBUG
1136    bool "Enable debugging of DMA-API usage"
1137    depends on HAVE_DMA_API_DEBUG
1138    help
1139      Enable this option to debug the use of the DMA API by device drivers.
1140      With this option you will be able to detect common bugs in device
1141      drivers like double-freeing of DMA mappings or freeing mappings that
1142      were never allocated.
1143      This option causes a performance degredation. Use only if you want
1144      to debug device drivers. If unsure, say N.
1145
1146config ATOMIC64_SELFTEST
1147    bool "Perform an atomic64_t self-test at boot"
1148    help
1149      Enable this option to test the atomic64_t functions at boot.
1150
1151      If unsure, say N.
1152
1153source "samples/Kconfig"
1154
1155source "lib/Kconfig.kgdb"
1156
1157source "lib/Kconfig.kmemcheck"
1158

Archive Download this file



interactive