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. Or add printk.time=1 at boot-time.
11      See Documentation/kernel-parameters.txt
12
13config DEFAULT_MESSAGE_LOGLEVEL
14    int "Default message log level (1-7)"
15    range 1 7
16    default "4"
17    help
18      Default log level for printk statements with no specified priority.
19
20      This was hard-coded to KERN_WARNING since at least 2.6.10 but folks
21      that are auditing their logs closely may want to set it to a lower
22      priority.
23
24config ENABLE_WARN_DEPRECATED
25    bool "Enable __deprecated logic"
26    default y
27    help
28      Enable the __deprecated logic in the kernel build.
29      Disable this to suppress the "warning: 'foo' is deprecated
30      (declared at kernel/power/somefile.c:1234)" messages.
31
32config ENABLE_MUST_CHECK
33    bool "Enable __must_check logic"
34    default y
35    help
36      Enable the __must_check logic in the kernel build. Disable this to
37      suppress the "warning: ignoring return value of 'foo', declared with
38      attribute warn_unused_result" messages.
39
40config FRAME_WARN
41    int "Warn for stack frames larger than (needs gcc 4.4)"
42    range 0 8192
43    default 1024 if !64BIT
44    default 2048 if 64BIT
45    help
46      Tell gcc to warn at build time for stack frames larger than this.
47      Setting this too low will cause a lot of warnings.
48      Setting it to 0 disables the warning.
49      Requires gcc 4.4
50
51config MAGIC_SYSRQ
52    bool "Magic SysRq key"
53    depends on !UML
54    help
55      If you say Y here, you will have some control over the system even
56      if the system crashes for example during kernel debugging (e.g., you
57      will be able to flush the buffer cache to disk, reboot the system
58      immediately or dump some status information). This is accomplished
59      by pressing various keys while holding SysRq (Alt+PrintScreen). It
60      also works on a serial console (on PC hardware at least), if you
61      send a BREAK and then within 5 seconds a command keypress. The
62      keys are documented in <file:Documentation/sysrq.txt>. Don't say Y
63      unless you really know what this hack does.
64
65config STRIP_ASM_SYMS
66    bool "Strip assembler-generated symbols during link"
67    default n
68    help
69      Strip internal assembler-generated symbols during a link (symbols
70      that look like '.Lxxx') so they don't pollute the output of
71      get_wchan() and suchlike.
72
73config UNUSED_SYMBOLS
74    bool "Enable unused/obsolete exported symbols"
75    default y if X86
76    help
77      Unused but exported symbols make the kernel needlessly bigger. For
78      that reason most of these unused exports will soon be removed. This
79      option is provided temporarily to provide a transition period in case
80      some external kernel module needs one of these symbols anyway. If you
81      encounter such a case in your module, consider if you are actually
82      using the right API. (rationale: since nobody in the kernel is using
83      this in a module, there is a pretty good chance it's actually the
84      wrong interface to use). If you really need the symbol, please send a
85      mail to the linux kernel mailing list mentioning the symbol and why
86      you really need it, and what the merge plan to the mainline kernel for
87      your module is.
88
89config DEBUG_FS
90    bool "Debug Filesystem"
91    help
92      debugfs is a virtual file system that kernel developers use to put
93      debugging files into. Enable this option to be able to read and
94      write to these files.
95
96      For detailed documentation on the debugfs API, see
97      Documentation/DocBook/filesystems.
98
99      If unsure, say N.
100
101config HEADERS_CHECK
102    bool "Run 'make headers_check' when building vmlinux"
103    depends on !UML
104    help
105      This option will extract the user-visible kernel headers whenever
106      building the kernel, and will run basic sanity checks on them to
107      ensure that exported files do not attempt to include files which
108      were not exported, etc.
109
110      If you're making modifications to header files which are
111      relevant for userspace, say 'Y', and check the headers
112      exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in
113      your build tree), to make sure they're suitable.
114
115config DEBUG_SECTION_MISMATCH
116    bool "Enable full Section mismatch analysis"
117    help
118      The section mismatch analysis checks if there are illegal
119      references from one section to another section.
120      Linux will during link or during runtime drop some sections
121      and any use of code/data previously in these sections will
122      most likely result in an oops.
123      In the code functions and variables are annotated with
124      __init, __devinit etc. (see full list in include/linux/init.h)
125      which results in the code/data being placed in specific sections.
126      The section mismatch analysis is always done after a full
127      kernel build but enabling this option will in addition
128      do the following:
129      - Add the option -fno-inline-functions-called-once to gcc
130        When inlining a function annotated __init in a non-init
131        function we would lose the section information and thus
132        the analysis would not catch the illegal reference.
133        This option tells gcc to inline less but will also
134        result in a larger kernel.
135      - Run the section mismatch analysis for each module/built-in.o
136        When we run the section mismatch analysis on vmlinux.o we
137        lose valueble information about where the mismatch was
138        introduced.
139        Running the analysis for each module/built-in.o file
140        will tell where the mismatch happens much closer to the
141        source. The drawback is that we will report the same
142        mismatch at least twice.
143      - Enable verbose reporting from modpost to help solving
144        the section mismatches reported.
145
146config DEBUG_KERNEL
147    bool "Kernel debugging"
148    help
149      Say Y here if you are developing drivers or trying to debug and
150      identify kernel problems.
151
152config DEBUG_SHIRQ
153    bool "Debug shared IRQ handlers"
154    depends on DEBUG_KERNEL && GENERIC_HARDIRQS
155    help
156      Enable this to generate a spurious interrupt as soon as a shared
157      interrupt handler is registered, and just before one is deregistered.
158      Drivers ought to be able to handle interrupts coming in at those
159      points; some don't and need to be caught.
160
161config LOCKUP_DETECTOR
162    bool "Detect Hard and Soft Lockups"
163    depends on DEBUG_KERNEL && !S390
164    help
165      Say Y here to enable the kernel to act as a watchdog to detect
166      hard and soft lockups.
167
168      Softlockups are bugs that cause the kernel to loop in kernel
169      mode for more than 60 seconds, without giving other tasks a
170      chance to run. The current stack trace is displayed upon
171      detection and the system will stay locked up.
172
173      Hardlockups are bugs that cause the CPU to loop in kernel mode
174      for more than 60 seconds, without letting other interrupts have a
175      chance to run. The current stack trace is displayed upon detection
176      and the system will stay locked up.
177
178      The overhead should be minimal. A periodic hrtimer runs to
179      generate interrupts and kick the watchdog task every 10-12 seconds.
180      An NMI is generated every 60 seconds or so to check for hardlockups.
181
182config HARDLOCKUP_DETECTOR
183    def_bool LOCKUP_DETECTOR && PERF_EVENTS && HAVE_PERF_EVENTS_NMI && \
184         !ARCH_HAS_NMI_WATCHDOG
185
186config BOOTPARAM_HARDLOCKUP_PANIC
187    bool "Panic (Reboot) On Hard Lockups"
188    depends on LOCKUP_DETECTOR
189    help
190      Say Y here to enable the kernel to panic on "hard lockups",
191      which are bugs that cause the kernel to loop in kernel
192      mode with interrupts disabled for more than 60 seconds.
193
194      Say N if unsure.
195
196config BOOTPARAM_HARDLOCKUP_PANIC_VALUE
197    int
198    depends on LOCKUP_DETECTOR
199    range 0 1
200    default 0 if !BOOTPARAM_HARDLOCKUP_PANIC
201    default 1 if BOOTPARAM_HARDLOCKUP_PANIC
202
203config BOOTPARAM_SOFTLOCKUP_PANIC
204    bool "Panic (Reboot) On Soft Lockups"
205    depends on LOCKUP_DETECTOR
206    help
207      Say Y here to enable the kernel to panic on "soft lockups",
208      which are bugs that cause the kernel to loop in kernel
209      mode for more than 60 seconds, without giving other tasks a
210      chance to run.
211
212      The panic can be used in combination with panic_timeout,
213      to cause the system to reboot automatically after a
214      lockup has been detected. This feature is useful for
215      high-availability systems that have uptime guarantees and
216      where a lockup must be resolved ASAP.
217
218      Say N if unsure.
219
220config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE
221    int
222    depends on LOCKUP_DETECTOR
223    range 0 1
224    default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC
225    default 1 if BOOTPARAM_SOFTLOCKUP_PANIC
226
227config DETECT_HUNG_TASK
228    bool "Detect Hung Tasks"
229    depends on DEBUG_KERNEL
230    default DETECT_SOFTLOCKUP
231    help
232      Say Y here to enable the kernel to detect "hung tasks",
233      which are bugs that cause the task to be stuck in
234      uninterruptible "D" state indefinitiley.
235
236      When a hung task is detected, the kernel will print the
237      current stack trace (which you should report), but the
238      task will stay in uninterruptible state. If lockdep is
239      enabled then all held locks will also be reported. This
240      feature has negligible overhead.
241
242config DEFAULT_HUNG_TASK_TIMEOUT
243    int "Default timeout for hung task detection (in seconds)"
244    depends on DETECT_HUNG_TASK
245    default 120
246    help
247      This option controls the default timeout (in seconds) used
248      to determine when a task has become non-responsive and should
249      be considered hung.
250
251      It can be adjusted at runtime via the kernel.hung_task_timeout
252      sysctl or by writing a value to /proc/sys/kernel/hung_task_timeout.
253
254      A timeout of 0 disables the check. The default is two minutes.
255      Keeping the default should be fine in most cases.
256
257config BOOTPARAM_HUNG_TASK_PANIC
258    bool "Panic (Reboot) On Hung Tasks"
259    depends on DETECT_HUNG_TASK
260    help
261      Say Y here to enable the kernel to panic on "hung tasks",
262      which are bugs that cause the kernel to leave a task stuck
263      in uninterruptible "D" state.
264
265      The panic can be used in combination with panic_timeout,
266      to cause the system to reboot automatically after a
267      hung task has been detected. This feature is useful for
268      high-availability systems that have uptime guarantees and
269      where a hung tasks must be resolved ASAP.
270
271      Say N if unsure.
272
273config BOOTPARAM_HUNG_TASK_PANIC_VALUE
274    int
275    depends on DETECT_HUNG_TASK
276    range 0 1
277    default 0 if !BOOTPARAM_HUNG_TASK_PANIC
278    default 1 if BOOTPARAM_HUNG_TASK_PANIC
279
280config SCHED_DEBUG
281    bool "Collect scheduler debugging info"
282    depends on DEBUG_KERNEL && PROC_FS
283    default y
284    help
285      If you say Y here, the /proc/sched_debug file will be provided
286      that can help debug the scheduler. The runtime overhead of this
287      option is minimal.
288
289config SCHEDSTATS
290    bool "Collect scheduler statistics"
291    depends on DEBUG_KERNEL && PROC_FS
292    help
293      If you say Y here, additional code will be inserted into the
294      scheduler and related routines to collect statistics about
295      scheduler behavior and provide them in /proc/schedstat. These
296      stats may be useful for both tuning and debugging the scheduler
297      If you aren't debugging the scheduler or trying to tune a specific
298      application, you can say N to avoid the very slight overhead
299      this adds.
300
301config TIMER_STATS
302    bool "Collect kernel timers statistics"
303    depends on DEBUG_KERNEL && PROC_FS
304    help
305      If you say Y here, additional code will be inserted into the
306      timer routines to collect statistics about kernel timers being
307      reprogrammed. The statistics can be read from /proc/timer_stats.
308      The statistics collection is started by writing 1 to /proc/timer_stats,
309      writing 0 stops it. This feature is useful to collect information
310      about timer usage patterns in kernel and userspace. This feature
311      is lightweight if enabled in the kernel config but not activated
312      (it defaults to deactivated on bootup and will only be activated
313      if some application like powertop activates it explicitly).
314
315config DEBUG_OBJECTS
316    bool "Debug object operations"
317    depends on DEBUG_KERNEL
318    help
319      If you say Y here, additional code will be inserted into the
320      kernel to track the life time of various objects and validate
321      the operations on those objects.
322
323config DEBUG_OBJECTS_SELFTEST
324    bool "Debug objects selftest"
325    depends on DEBUG_OBJECTS
326    help
327      This enables the selftest of the object debug code.
328
329config DEBUG_OBJECTS_FREE
330    bool "Debug objects in freed memory"
331    depends on DEBUG_OBJECTS
332    help
333      This enables checks whether a k/v free operation frees an area
334      which contains an object which has not been deactivated
335      properly. This can make kmalloc/kfree-intensive workloads
336      much slower.
337
338config DEBUG_OBJECTS_TIMERS
339    bool "Debug timer objects"
340    depends on DEBUG_OBJECTS
341    help
342      If you say Y here, additional code will be inserted into the
343      timer routines to track the life time of timer objects and
344      validate the timer operations.
345
346config DEBUG_OBJECTS_WORK
347    bool "Debug work objects"
348    depends on DEBUG_OBJECTS
349    help
350      If you say Y here, additional code will be inserted into the
351      work queue routines to track the life time of work objects and
352      validate the work operations.
353
354config DEBUG_OBJECTS_RCU_HEAD
355    bool "Debug RCU callbacks objects"
356    depends on DEBUG_OBJECTS
357    help
358      Enable this to turn on debugging of RCU list heads (call_rcu() usage).
359
360config DEBUG_OBJECTS_PERCPU_COUNTER
361    bool "Debug percpu counter objects"
362    depends on DEBUG_OBJECTS
363    help
364      If you say Y here, additional code will be inserted into the
365      percpu counter routines to track the life time of percpu counter
366      objects and validate the percpu counter operations.
367
368config DEBUG_OBJECTS_ENABLE_DEFAULT
369    int "debug_objects bootup default value (0-1)"
370        range 0 1
371        default "1"
372        depends on DEBUG_OBJECTS
373        help
374          Debug objects boot parameter default value
375
376config DEBUG_SLAB
377    bool "Debug slab memory allocations"
378    depends on DEBUG_KERNEL && SLAB && !KMEMCHECK
379    help
380      Say Y here to have the kernel do limited verification on memory
381      allocation as well as poisoning memory on free to catch use of freed
382      memory. This can make kmalloc/kfree-intensive workloads much slower.
383
384config DEBUG_SLAB_LEAK
385    bool "Memory leak debugging"
386    depends on DEBUG_SLAB
387
388config SLUB_DEBUG_ON
389    bool "SLUB debugging on by default"
390    depends on SLUB && SLUB_DEBUG && !KMEMCHECK
391    default n
392    help
393      Boot with debugging on by default. SLUB boots by default with
394      the runtime debug capabilities switched off. Enabling this is
395      equivalent to specifying the "slub_debug" parameter on boot.
396      There is no support for more fine grained debug control like
397      possible with slub_debug=xxx. SLUB debugging may be switched
398      off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying
399      "slub_debug=-".
400
401config SLUB_STATS
402    default n
403    bool "Enable SLUB performance statistics"
404    depends on SLUB && SYSFS
405    help
406      SLUB statistics are useful to debug SLUBs allocation behavior in
407      order find ways to optimize the allocator. This should never be
408      enabled for production use since keeping statistics slows down
409      the allocator by a few percentage points. The slabinfo command
410      supports the determination of the most active slabs to figure
411      out which slabs are relevant to a particular load.
412      Try running: slabinfo -DA
413
414config DEBUG_KMEMLEAK
415    bool "Kernel memory leak detector"
416    depends on DEBUG_KERNEL && EXPERIMENTAL && !MEMORY_HOTPLUG && \
417        (X86 || ARM || PPC || MIPS || S390 || SPARC64 || SUPERH || MICROBLAZE || TILE)
418
419    select DEBUG_FS
420    select STACKTRACE if STACKTRACE_SUPPORT
421    select KALLSYMS
422    select CRC32
423    help
424      Say Y here if you want to enable the memory leak
425      detector. The memory allocation/freeing is traced in a way
426      similar to the Boehm's conservative garbage collector, the
427      difference being that the orphan objects are not freed but
428      only shown in /sys/kernel/debug/kmemleak. Enabling this
429      feature will introduce an overhead to memory
430      allocations. See Documentation/kmemleak.txt for more
431      details.
432
433      Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances
434      of finding leaks due to the slab objects poisoning.
435
436      In order to access the kmemleak file, debugfs needs to be
437      mounted (usually at /sys/kernel/debug).
438
439config DEBUG_KMEMLEAK_EARLY_LOG_SIZE
440    int "Maximum kmemleak early log entries"
441    depends on DEBUG_KMEMLEAK
442    range 200 40000
443    default 400
444    help
445      Kmemleak must track all the memory allocations to avoid
446      reporting false positives. Since memory may be allocated or
447      freed before kmemleak is initialised, an early log buffer is
448      used to store these actions. If kmemleak reports "early log
449      buffer exceeded", please increase this value.
450
451config DEBUG_KMEMLEAK_TEST
452    tristate "Simple test for the kernel memory leak detector"
453    depends on DEBUG_KMEMLEAK && m
454    help
455      This option enables a module that explicitly leaks memory.
456
457      If unsure, say N.
458
459config DEBUG_KMEMLEAK_DEFAULT_OFF
460    bool "Default kmemleak to off"
461    depends on DEBUG_KMEMLEAK
462    help
463      Say Y here to disable kmemleak by default. It can then be enabled
464      on the command line via kmemleak=on.
465
466config DEBUG_PREEMPT
467    bool "Debug preemptible kernel"
468    depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT
469    default y
470    help
471      If you say Y here then the kernel will use a debug variant of the
472      commonly used smp_processor_id() function and will print warnings
473      if kernel code uses it in a preemption-unsafe way. Also, the kernel
474      will detect preemption count underflows.
475
476config DEBUG_RT_MUTEXES
477    bool "RT Mutex debugging, deadlock detection"
478    depends on DEBUG_KERNEL && RT_MUTEXES
479    help
480     This allows rt mutex semantics violations and rt mutex related
481     deadlocks (lockups) to be detected and reported automatically.
482
483config DEBUG_PI_LIST
484    bool
485    default y
486    depends on DEBUG_RT_MUTEXES
487
488config RT_MUTEX_TESTER
489    bool "Built-in scriptable tester for rt-mutexes"
490    depends on DEBUG_KERNEL && RT_MUTEXES
491    help
492      This option enables a rt-mutex tester.
493
494config DEBUG_SPINLOCK
495    bool "Spinlock and rw-lock debugging: basic checks"
496    depends on DEBUG_KERNEL
497    help
498      Say Y here and build SMP to catch missing spinlock initialization
499      and certain other kinds of spinlock errors commonly made. This is
500      best used in conjunction with the NMI watchdog so that spinlock
501      deadlocks are also debuggable.
502
503config DEBUG_MUTEXES
504    bool "Mutex debugging: basic checks"
505    depends on DEBUG_KERNEL
506    help
507     This feature allows mutex semantics violations to be detected and
508     reported.
509
510config DEBUG_LOCK_ALLOC
511    bool "Lock debugging: detect incorrect freeing of live locks"
512    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
513    select DEBUG_SPINLOCK
514    select DEBUG_MUTEXES
515    select LOCKDEP
516    help
517     This feature will check whether any held lock (spinlock, rwlock,
518     mutex or rwsem) is incorrectly freed by the kernel, via any of the
519     memory-freeing routines (kfree(), kmem_cache_free(), free_pages(),
520     vfree(), etc.), whether a live lock is incorrectly reinitialized via
521     spin_lock_init()/mutex_init()/etc., or whether there is any lock
522     held during task exit.
523
524config PROVE_LOCKING
525    bool "Lock debugging: prove locking correctness"
526    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
527    select LOCKDEP
528    select DEBUG_SPINLOCK
529    select DEBUG_MUTEXES
530    select DEBUG_LOCK_ALLOC
531    select TRACE_IRQFLAGS
532    default n
533    help
534     This feature enables the kernel to prove that all locking
535     that occurs in the kernel runtime is mathematically
536     correct: that under no circumstance could an arbitrary (and
537     not yet triggered) combination of observed locking
538     sequences (on an arbitrary number of CPUs, running an
539     arbitrary number of tasks and interrupt contexts) cause a
540     deadlock.
541
542     In short, this feature enables the kernel to report locking
543     related deadlocks before they actually occur.
544
545     The proof does not depend on how hard and complex a
546     deadlock scenario would be to trigger: how many
547     participant CPUs, tasks and irq-contexts would be needed
548     for it to trigger. The proof also does not depend on
549     timing: if a race and a resulting deadlock is possible
550     theoretically (no matter how unlikely the race scenario
551     is), it will be proven so and will immediately be
552     reported by the kernel (once the event is observed that
553     makes the deadlock theoretically possible).
554
555     If a deadlock is impossible (i.e. the locking rules, as
556     observed by the kernel, are mathematically correct), the
557     kernel reports nothing.
558
559     NOTE: this feature can also be enabled for rwlocks, mutexes
560     and rwsems - in which case all dependencies between these
561     different locking variants are observed and mapped too, and
562     the proof of observed correctness is also maintained for an
563     arbitrary combination of these separate locking variants.
564
565     For more details, see Documentation/lockdep-design.txt.
566
567config PROVE_RCU
568    bool "RCU debugging: prove RCU correctness"
569    depends on PROVE_LOCKING
570    default n
571    help
572     This feature enables lockdep extensions that check for correct
573     use of RCU APIs. This is currently under development. Say Y
574     if you want to debug RCU usage or help work on the PROVE_RCU
575     feature.
576
577     Say N if you are unsure.
578
579config PROVE_RCU_REPEATEDLY
580    bool "RCU debugging: don't disable PROVE_RCU on first splat"
581    depends on PROVE_RCU
582    default n
583    help
584     By itself, PROVE_RCU will disable checking upon issuing the
585     first warning (or "splat"). This feature prevents such
586     disabling, allowing multiple RCU-lockdep warnings to be printed
587     on a single reboot.
588
589     Say Y to allow multiple RCU-lockdep warnings per boot.
590
591     Say N if you are unsure.
592
593config SPARSE_RCU_POINTER
594    bool "RCU debugging: sparse-based checks for pointer usage"
595    default n
596    help
597     This feature enables the __rcu sparse annotation for
598     RCU-protected pointers. This annotation will cause sparse
599     to flag any non-RCU used of annotated pointers. This can be
600     helpful when debugging RCU usage. Please note that this feature
601     is not intended to enforce code cleanliness; it is instead merely
602     a debugging aid.
603
604     Say Y to make sparse flag questionable use of RCU-protected pointers
605
606     Say N if you are unsure.
607
608config LOCKDEP
609    bool
610    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
611    select STACKTRACE
612    select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE
613    select KALLSYMS
614    select KALLSYMS_ALL
615
616config LOCK_STAT
617    bool "Lock usage statistics"
618    depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
619    select LOCKDEP
620    select DEBUG_SPINLOCK
621    select DEBUG_MUTEXES
622    select DEBUG_LOCK_ALLOC
623    default n
624    help
625     This feature enables tracking lock contention points
626
627     For more details, see Documentation/lockstat.txt
628
629     This also enables lock events required by "perf lock",
630     subcommand of perf.
631     If you want to use "perf lock", you also need to turn on
632     CONFIG_EVENT_TRACING.
633
634     CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
635     (CONFIG_LOCKDEP defines "acquire" and "release" events.)
636
637config DEBUG_LOCKDEP
638    bool "Lock dependency engine debugging"
639    depends on DEBUG_KERNEL && LOCKDEP
640    help
641      If you say Y here, the lock dependency engine will do
642      additional runtime checks to debug itself, at the price
643      of more runtime overhead.
644
645config TRACE_IRQFLAGS
646    bool
647    help
648      Enables hooks to interrupt enabling and disabling for
649      either tracing or lock debugging.
650
651config DEBUG_SPINLOCK_SLEEP
652    bool "Spinlock debugging: sleep-inside-spinlock checking"
653    depends on DEBUG_KERNEL
654    help
655      If you say Y here, various routines which may sleep will become very
656      noisy if they are called with a spinlock held.
657
658config DEBUG_LOCKING_API_SELFTESTS
659    bool "Locking API boot-time self-tests"
660    depends on DEBUG_KERNEL
661    help
662      Say Y here if you want the kernel to run a short self-test during
663      bootup. The self-test checks whether common types of locking bugs
664      are detected by debugging mechanisms or not. (if you disable
665      lock debugging then those bugs wont be detected of course.)
666      The following locking APIs are covered: spinlocks, rwlocks,
667      mutexes and rwsems.
668
669config STACKTRACE
670    bool
671    depends on STACKTRACE_SUPPORT
672
673config DEBUG_STACK_USAGE
674    bool "Stack utilization instrumentation"
675    depends on DEBUG_KERNEL
676    help
677      Enables the display of the minimum amount of free stack which each
678      task has ever had available in the sysrq-T and sysrq-P debug output.
679
680      This option will slow down process creation somewhat.
681
682config DEBUG_KOBJECT
683    bool "kobject debugging"
684    depends on DEBUG_KERNEL
685    help
686      If you say Y here, some extra kobject debugging messages will be sent
687      to the syslog.
688
689config DEBUG_HIGHMEM
690    bool "Highmem debugging"
691    depends on DEBUG_KERNEL && HIGHMEM
692    help
693      This options enables addition error checking for high memory systems.
694      Disable for production systems.
695
696config DEBUG_BUGVERBOSE
697    bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT
698    depends on BUG
699    depends on ARM || AVR32 || M32R || M68K || SPARC32 || SPARC64 || \
700           FRV || SUPERH || GENERIC_BUG || BLACKFIN || MN10300 || TILE
701    default y
702    help
703      Say Y here to make BUG() panics output the file name and line number
704      of the BUG call as well as the EIP and oops trace. This aids
705      debugging but costs about 70-100K of memory.
706
707config DEBUG_INFO
708    bool "Compile the kernel with debug info"
709    depends on DEBUG_KERNEL
710    help
711          If you say Y here the resulting kernel image will include
712      debugging info resulting in a larger kernel image.
713      This adds debug symbols to the kernel and modules (gcc -g), and
714      is needed if you intend to use kernel crashdump or binary object
715      tools like crash, kgdb, LKCD, gdb, etc on the kernel.
716      Say Y here only if you plan to debug the kernel.
717
718      If unsure, say N.
719
720config DEBUG_INFO_REDUCED
721    bool "Reduce debugging information"
722    depends on DEBUG_INFO
723    help
724      If you say Y here gcc is instructed to generate less debugging
725      information for structure types. This means that tools that
726      need full debugging information (like kgdb or systemtap) won't
727      be happy. But if you merely need debugging information to
728      resolve line numbers there is no loss. Advantage is that
729      build directory object sizes shrink dramatically over a full
730      DEBUG_INFO build and compile times are reduced too.
731      Only works with newer gcc versions.
732
733config DEBUG_VM
734    bool "Debug VM"
735    depends on DEBUG_KERNEL
736    help
737      Enable this to turn on extended checks in the virtual-memory system
738          that may impact performance.
739
740      If unsure, say N.
741
742config DEBUG_VIRTUAL
743    bool "Debug VM translations"
744    depends on DEBUG_KERNEL && X86
745    help
746      Enable some costly sanity checks in virtual to page code. This can
747      catch mistakes with virt_to_page() and friends.
748
749      If unsure, say N.
750
751config DEBUG_NOMMU_REGIONS
752    bool "Debug the global anon/private NOMMU mapping region tree"
753    depends on DEBUG_KERNEL && !MMU
754    help
755      This option causes the global tree of anonymous and private mapping
756      regions to be regularly checked for invalid topology.
757
758config DEBUG_WRITECOUNT
759    bool "Debug filesystem writers count"
760    depends on DEBUG_KERNEL
761    help
762      Enable this to catch wrong use of the writers count in struct
763      vfsmount. This will increase the size of each file struct by
764      32 bits.
765
766      If unsure, say N.
767
768config DEBUG_MEMORY_INIT
769    bool "Debug memory initialisation" if EXPERT
770    default !EXPERT
771    help
772      Enable this for additional checks during memory initialisation.
773      The sanity checks verify aspects of the VM such as the memory model
774      and other information provided by the architecture. Verbose
775      information will be printed at KERN_DEBUG loglevel depending
776      on the mminit_loglevel= command-line option.
777
778      If unsure, say Y
779
780config DEBUG_LIST
781    bool "Debug linked list manipulation"
782    depends on DEBUG_KERNEL
783    help
784      Enable this to turn on extended checks in the linked-list
785      walking routines.
786
787      If unsure, say N.
788
789config TEST_LIST_SORT
790    bool "Linked list sorting test"
791    depends on DEBUG_KERNEL
792    help
793      Enable this to turn on 'list_sort()' function test. This test is
794      executed only once during system boot, so affects only boot time.
795
796      If unsure, say N.
797
798config DEBUG_SG
799    bool "Debug SG table operations"
800    depends on DEBUG_KERNEL
801    help
802      Enable this to turn on checks on scatter-gather tables. This can
803      help find problems with drivers that do not properly initialize
804      their sg tables.
805
806      If unsure, say N.
807
808config DEBUG_NOTIFIERS
809    bool "Debug notifier call chains"
810    depends on DEBUG_KERNEL
811    help
812      Enable this to turn on sanity checking for notifier call chains.
813      This is most useful for kernel developers to make sure that
814      modules properly unregister themselves from notifier chains.
815      This is a relatively cheap check but if you care about maximum
816      performance, say N.
817
818config DEBUG_CREDENTIALS
819    bool "Debug credential management"
820    depends on DEBUG_KERNEL
821    help
822      Enable this to turn on some debug checking for credential
823      management. The additional code keeps track of the number of
824      pointers from task_structs to any given cred struct, and checks to
825      see that this number never exceeds the usage count of the cred
826      struct.
827
828      Furthermore, if SELinux is enabled, this also checks that the
829      security pointer in the cred struct is never seen to be invalid.
830
831      If unsure, say N.
832
833#
834# Select this config option from the architecture Kconfig, if it
835# it is preferred to always offer frame pointers as a config
836# option on the architecture (regardless of KERNEL_DEBUG):
837#
838config ARCH_WANT_FRAME_POINTERS
839    bool
840    help
841
842config FRAME_POINTER
843    bool "Compile the kernel with frame pointers"
844    depends on DEBUG_KERNEL && \
845        (CRIS || M68K || FRV || UML || \
846         AVR32 || SUPERH || BLACKFIN || MN10300) || \
847        ARCH_WANT_FRAME_POINTERS
848    default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS
849    help
850      If you say Y here the resulting kernel image will be slightly
851      larger and slower, but it gives very useful debugging information
852      in case of kernel bugs. (precise oopses/stacktraces/warnings)
853
854config BOOT_PRINTK_DELAY
855    bool "Delay each boot printk message by N milliseconds"
856    depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY
857    help
858      This build option allows you to read kernel boot messages
859      by inserting a short delay after each one. The delay is
860      specified in milliseconds on the kernel command line,
861      using "boot_delay=N".
862
863      It is likely that you would also need to use "lpj=M" to preset
864      the "loops per jiffie" value.
865      See a previous boot log for the "lpj" value to use for your
866      system, and then set "lpj=M" before setting "boot_delay=N".
867      NOTE: Using this option may adversely affect SMP systems.
868      I.e., processors other than the first one may not boot up.
869      BOOT_PRINTK_DELAY also may cause DETECT_SOFTLOCKUP to detect
870      what it believes to be lockup conditions.
871
872config RCU_TORTURE_TEST
873    tristate "torture tests for RCU"
874    depends on DEBUG_KERNEL
875    default n
876    help
877      This option provides a kernel module that runs torture tests
878      on the RCU infrastructure. The kernel module may be built
879      after the fact on the running kernel to be tested, if desired.
880
881      Say Y here if you want RCU torture tests to be built into
882      the kernel.
883      Say M if you want the RCU torture tests to build as a module.
884      Say N if you are unsure.
885
886config RCU_TORTURE_TEST_RUNNABLE
887    bool "torture tests for RCU runnable by default"
888    depends on RCU_TORTURE_TEST = y
889    default n
890    help
891      This option provides a way to build the RCU torture tests
892      directly into the kernel without them starting up at boot
893      time. You can use /proc/sys/kernel/rcutorture_runnable
894      to manually override this setting. This /proc file is
895      available only when the RCU torture tests have been built
896      into the kernel.
897
898      Say Y here if you want the RCU torture tests to start during
899      boot (you probably don't).
900      Say N here if you want the RCU torture tests to start only
901      after being manually enabled via /proc.
902
903config RCU_CPU_STALL_TIMEOUT
904    int "RCU CPU stall timeout in seconds"
905    depends on TREE_RCU || TREE_PREEMPT_RCU
906    range 3 300
907    default 60
908    help
909      If a given RCU grace period extends more than the specified
910      number of seconds, a CPU stall warning is printed. If the
911      RCU grace period persists, additional CPU stall warnings are
912      printed at more widely spaced intervals.
913
914config RCU_CPU_STALL_VERBOSE
915    bool "Print additional per-task information for RCU_CPU_STALL_DETECTOR"
916    depends on TREE_PREEMPT_RCU
917    default y
918    help
919      This option causes RCU to printk detailed per-task information
920      for any tasks that are stalling the current RCU grace period.
921
922      Say N if you are unsure.
923
924      Say Y if you want to enable such checks.
925
926config KPROBES_SANITY_TEST
927    bool "Kprobes sanity tests"
928    depends on DEBUG_KERNEL
929    depends on KPROBES
930    default n
931    help
932      This option provides for testing basic kprobes functionality on
933      boot. A sample kprobe, jprobe and kretprobe are inserted and
934      verified for functionality.
935
936      Say N if you are unsure.
937
938config BACKTRACE_SELF_TEST
939    tristate "Self test for the backtrace code"
940    depends on DEBUG_KERNEL
941    default n
942    help
943      This option provides a kernel module that can be used to test
944      the kernel stack backtrace code. This option is not useful
945      for distributions or general kernels, but only for kernel
946      developers working on architecture code.
947
948      Note that if you want to also test saved backtraces, you will
949      have to enable STACKTRACE as well.
950
951      Say N if you are unsure.
952
953config DEBUG_BLOCK_EXT_DEVT
954        bool "Force extended block device numbers and spread them"
955    depends on DEBUG_KERNEL
956    depends on BLOCK
957    default n
958    help
959      BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON
960      SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT
961      YOU ARE DOING. Distros, please enable this and fix whatever
962      is broken.
963
964      Conventionally, block device numbers are allocated from
965      predetermined contiguous area. However, extended block area
966      may introduce non-contiguous block device numbers. This
967      option forces most block device numbers to be allocated from
968      the extended space and spreads them to discover kernel or
969      userland code paths which assume predetermined contiguous
970      device number allocation.
971
972      Note that turning on this debug option shuffles all the
973      device numbers for all IDE and SCSI devices including libata
974      ones, so root partition specified using device number
975      directly (via rdev or root=MAJ:MIN) won't work anymore.
976      Textual device names (root=/dev/sdXn) will continue to work.
977
978      Say N if you are unsure.
979
980config DEBUG_FORCE_WEAK_PER_CPU
981    bool "Force weak per-cpu definitions"
982    depends on DEBUG_KERNEL
983    help
984      s390 and alpha require percpu variables in modules to be
985      defined weak to work around addressing range issue which
986      puts the following two restrictions on percpu variable
987      definitions.
988
989      1. percpu symbols must be unique whether static or not
990      2. percpu variables can't be defined inside a function
991
992      To ensure that generic code follows the above rules, this
993      option forces all percpu variables to be defined as weak.
994
995config DEBUG_PER_CPU_MAPS
996    bool "Debug access to per_cpu maps"
997    depends on DEBUG_KERNEL
998    depends on SMP
999    help
1000      Say Y to verify that the per_cpu map being accessed has
1001      been set up. This adds a fair amount of code to kernel memory
1002      and decreases performance.
1003
1004      Say N if unsure.
1005
1006config LKDTM
1007    tristate "Linux Kernel Dump Test Tool Module"
1008    depends on DEBUG_FS
1009    depends on BLOCK
1010    default n
1011    help
1012    This module enables testing of the different dumping mechanisms by
1013    inducing system failures at predefined crash points.
1014    If you don't need it: say N
1015    Choose M here to compile this code as a module. The module will be
1016    called lkdtm.
1017
1018    Documentation on how to use the module can be found in
1019    Documentation/fault-injection/provoke-crashes.txt
1020
1021config CPU_NOTIFIER_ERROR_INJECT
1022    tristate "CPU notifier error injection module"
1023    depends on HOTPLUG_CPU && DEBUG_KERNEL
1024    help
1025      This option provides a kernel module that can be used to test
1026      the error handling of the cpu notifiers
1027
1028      To compile this code as a module, choose M here: the module will
1029      be called cpu-notifier-error-inject.
1030
1031      If unsure, say N.
1032
1033config FAULT_INJECTION
1034    bool "Fault-injection framework"
1035    depends on DEBUG_KERNEL
1036    help
1037      Provide fault-injection framework.
1038      For more details, see Documentation/fault-injection/.
1039
1040config FAILSLAB
1041    bool "Fault-injection capability for kmalloc"
1042    depends on FAULT_INJECTION
1043    depends on SLAB || SLUB
1044    help
1045      Provide fault-injection capability for kmalloc.
1046
1047config FAIL_PAGE_ALLOC
1048    bool "Fault-injection capabilitiy for alloc_pages()"
1049    depends on FAULT_INJECTION
1050    help
1051      Provide fault-injection capability for alloc_pages().
1052
1053config FAIL_MAKE_REQUEST
1054    bool "Fault-injection capability for disk IO"
1055    depends on FAULT_INJECTION && BLOCK
1056    help
1057      Provide fault-injection capability for disk IO.
1058
1059config FAIL_IO_TIMEOUT
1060    bool "Fault-injection capability for faking disk interrupts"
1061    depends on FAULT_INJECTION && BLOCK
1062    help
1063      Provide fault-injection capability on end IO handling. This
1064      will make the block layer "forget" an interrupt as configured,
1065      thus exercising the error handling.
1066
1067      Only works with drivers that use the generic timeout handling,
1068      for others it wont do anything.
1069
1070config FAULT_INJECTION_DEBUG_FS
1071    bool "Debugfs entries for fault-injection capabilities"
1072    depends on FAULT_INJECTION && SYSFS && DEBUG_FS
1073    help
1074      Enable configuration of fault-injection capabilities via debugfs.
1075
1076config FAULT_INJECTION_STACKTRACE_FILTER
1077    bool "stacktrace filter for fault-injection capabilities"
1078    depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
1079    depends on !X86_64
1080    select STACKTRACE
1081    select FRAME_POINTER if !PPC && !S390 && !MICROBLAZE
1082    help
1083      Provide stacktrace filter for fault-injection capabilities
1084
1085config LATENCYTOP
1086    bool "Latency measuring infrastructure"
1087    depends on HAVE_LATENCYTOP_SUPPORT
1088    depends on DEBUG_KERNEL
1089    depends on STACKTRACE_SUPPORT
1090    depends on PROC_FS
1091    select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE
1092    select KALLSYMS
1093    select KALLSYMS_ALL
1094    select STACKTRACE
1095    select SCHEDSTATS
1096    select SCHED_DEBUG
1097    help
1098      Enable this option if you want to use the LatencyTOP tool
1099      to find out which userspace is blocking on what kernel operations.
1100
1101config SYSCTL_SYSCALL_CHECK
1102    bool "Sysctl checks"
1103    depends on SYSCTL
1104    ---help---
1105      sys_sysctl uses binary paths that have been found challenging
1106      to properly maintain and use. This enables checks that help
1107      you to keep things correct.
1108
1109source mm/Kconfig.debug
1110source kernel/trace/Kconfig
1111
1112config PROVIDE_OHCI1394_DMA_INIT
1113    bool "Remote debugging over FireWire early on boot"
1114    depends on PCI && X86
1115    help
1116      If you want to debug problems which hang or crash the kernel early
1117      on boot and the crashing machine has a FireWire port, you can use
1118      this feature to remotely access the memory of the crashed machine
1119      over FireWire. This employs remote DMA as part of the OHCI1394
1120      specification which is now the standard for FireWire controllers.
1121
1122      With remote DMA, you can monitor the printk buffer remotely using
1123      firescope and access all memory below 4GB using fireproxy from gdb.
1124      Even controlling a kernel debugger is possible using remote DMA.
1125
1126      Usage:
1127
1128      If ohci1394_dma=early is used as boot parameter, it will initialize
1129      all OHCI1394 controllers which are found in the PCI config space.
1130
1131      As all changes to the FireWire bus such as enabling and disabling
1132      devices cause a bus reset and thereby disable remote DMA for all
1133      devices, be sure to have the cable plugged and FireWire enabled on
1134      the debugging host before booting the debug target for debugging.
1135
1136      This code (~1k) is freed after boot. By then, the firewire stack
1137      in charge of the OHCI-1394 controllers should be used instead.
1138
1139      See Documentation/debugging-via-ohci1394.txt for more information.
1140
1141config FIREWIRE_OHCI_REMOTE_DMA
1142    bool "Remote debugging over FireWire with firewire-ohci"
1143    depends on FIREWIRE_OHCI
1144    help
1145      This option lets you use the FireWire bus for remote debugging
1146      with help of the firewire-ohci driver. It enables unfiltered
1147      remote DMA in firewire-ohci.
1148      See Documentation/debugging-via-ohci1394.txt for more information.
1149
1150      If unsure, say N.
1151
1152config BUILD_DOCSRC
1153    bool "Build targets in Documentation/ tree"
1154    depends on HEADERS_CHECK
1155    help
1156      This option attempts to build objects from the source files in the
1157      kernel Documentation/ tree.
1158
1159      Say N if you are unsure.
1160
1161config DYNAMIC_DEBUG
1162    bool "Enable dynamic printk() support"
1163    default n
1164    depends on PRINTK
1165    depends on DEBUG_FS
1166    help
1167
1168      Compiles debug level messages into the kernel, which would not
1169      otherwise be available at runtime. These messages can then be
1170      enabled/disabled based on various levels of scope - per source file,
1171      function, module, format string, and line number. This mechanism
1172      implicitly enables all pr_debug() and dev_dbg() calls. The impact of
1173      this compile option is a larger kernel text size of about 2%.
1174
1175      Usage:
1176
1177      Dynamic debugging is controlled via the 'dynamic_debug/control' file,
1178      which is contained in the 'debugfs' filesystem. Thus, the debugfs
1179      filesystem must first be mounted before making use of this feature.
1180      We refer the control file as: <debugfs>/dynamic_debug/control. This
1181      file contains a list of the debug statements that can be enabled. The
1182      format for each line of the file is:
1183
1184        filename:lineno [module]function flags format
1185
1186      filename : source file of the debug statement
1187      lineno : line number of the debug statement
1188      module : module that contains the debug statement
1189      function : function that contains the debug statement
1190          flags : 'p' means the line is turned 'on' for printing
1191          format : the format used for the debug statement
1192
1193      From a live system:
1194
1195        nullarbor:~ # cat <debugfs>/dynamic_debug/control
1196        # filename:lineno [module]function flags format
1197        fs/aio.c:222 [aio]__put_ioctx - "__put_ioctx:\040freeing\040%p\012"
1198        fs/aio.c:248 [aio]ioctx_alloc - "ENOMEM:\040nr_events\040too\040high\012"
1199        fs/aio.c:1770 [aio]sys_io_cancel - "calling\040cancel\012"
1200
1201      Example usage:
1202
1203        // enable the message at line 1603 of file svcsock.c
1204        nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' >
1205                        <debugfs>/dynamic_debug/control
1206
1207        // enable all the messages in file svcsock.c
1208        nullarbor:~ # echo -n 'file svcsock.c +p' >
1209                        <debugfs>/dynamic_debug/control
1210
1211        // enable all the messages in the NFS server module
1212        nullarbor:~ # echo -n 'module nfsd +p' >
1213                        <debugfs>/dynamic_debug/control
1214
1215        // enable all 12 messages in the function svc_process()
1216        nullarbor:~ # echo -n 'func svc_process +p' >
1217                        <debugfs>/dynamic_debug/control
1218
1219        // disable all 12 messages in the function svc_process()
1220        nullarbor:~ # echo -n 'func svc_process -p' >
1221                        <debugfs>/dynamic_debug/control
1222
1223      See Documentation/dynamic-debug-howto.txt for additional information.
1224
1225config DMA_API_DEBUG
1226    bool "Enable debugging of DMA-API usage"
1227    depends on HAVE_DMA_API_DEBUG
1228    help
1229      Enable this option to debug the use of the DMA API by device drivers.
1230      With this option you will be able to detect common bugs in device
1231      drivers like double-freeing of DMA mappings or freeing mappings that
1232      were never allocated.
1233      This option causes a performance degredation. Use only if you want
1234      to debug device drivers. If unsure, say N.
1235
1236config ATOMIC64_SELFTEST
1237    bool "Perform an atomic64_t self-test at boot"
1238    help
1239      Enable this option to test the atomic64_t functions at boot.
1240
1241      If unsure, say N.
1242
1243config ASYNC_RAID6_TEST
1244    tristate "Self test for hardware accelerated raid6 recovery"
1245    depends on ASYNC_RAID6_RECOV
1246    select ASYNC_MEMCPY
1247    ---help---
1248      This is a one-shot self test that permutes through the
1249      recovery of all the possible two disk failure scenarios for a
1250      N-disk array. Recovery is performed with the asynchronous
1251      raid6 recovery routines, and will optionally use an offload
1252      engine if one is available.
1253
1254      If unsure, say N.
1255
1256source "samples/Kconfig"
1257
1258source "lib/Kconfig.kgdb"
1259
1260source "lib/Kconfig.kmemcheck"
1261
1262config TEST_KSTRTOX
1263    tristate "Test kstrto*() family of functions at runtime"
1264

Archive Download this file



interactive