1menu "Kernel hacking" 2 3menu "printk and dmesg options" 4 5config PRINTK_TIME 6 bool "Show timing information on printks" 7 depends on PRINTK 8 help 9 Selecting this option causes time stamps of the printk() 10 messages to be added to the output of the syslog() system 11 call and at the console. 12 13 The timestamp is always recorded internally, and exported 14 to /dev/kmsg. This flag just specifies if the timestamp should 15 be included, not that the timestamp is recorded. 16 17 The behavior is also controlled by the kernel command line 18 parameter printk.time=1. See Documentation/admin-guide/kernel-parameters.rst 19 20config CONSOLE_LOGLEVEL_DEFAULT 21 int "Default console loglevel (1-15)" 22 range 1 15 23 default "7" 24 help 25 Default loglevel to determine what will be printed on the console. 26 27 Setting a default here is equivalent to passing in loglevel=<x> in 28 the kernel bootargs. loglevel=<x> continues to override whatever 29 value is specified here as well. 30 31 Note: This does not affect the log level of un-prefixed printk() 32 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT 33 option. 34 35config MESSAGE_LOGLEVEL_DEFAULT 36 int "Default message log level (1-7)" 37 range 1 7 38 default "4" 39 help 40 Default log level for printk statements with no specified priority. 41 42 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks 43 that are auditing their logs closely may want to set it to a lower 44 priority. 45 46 Note: This does not affect what message level gets printed on the console 47 by default. To change that, use loglevel=<x> in the kernel bootargs, 48 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value. 49 50config BOOT_PRINTK_DELAY 51 bool "Delay each boot printk message by N milliseconds" 52 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY 53 help 54 This build option allows you to read kernel boot messages 55 by inserting a short delay after each one. The delay is 56 specified in milliseconds on the kernel command line, 57 using "boot_delay=N". 58 59 It is likely that you would also need to use "lpj=M" to preset 60 the "loops per jiffie" value. 61 See a previous boot log for the "lpj" value to use for your 62 system, and then set "lpj=M" before setting "boot_delay=N". 63 NOTE: Using this option may adversely affect SMP systems. 64 I.e., processors other than the first one may not boot up. 65 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect 66 what it believes to be lockup conditions. 67 68config DYNAMIC_DEBUG 69 bool "Enable dynamic printk() support" 70 default n 71 depends on PRINTK 72 depends on DEBUG_FS 73 help 74 75 Compiles debug level messages into the kernel, which would not 76 otherwise be available at runtime. These messages can then be 77 enabled/disabled based on various levels of scope - per source file, 78 function, module, format string, and line number. This mechanism 79 implicitly compiles in all pr_debug() and dev_dbg() calls, which 80 enlarges the kernel text size by about 2%. 81 82 If a source file is compiled with DEBUG flag set, any 83 pr_debug() calls in it are enabled by default, but can be 84 disabled at runtime as below. Note that DEBUG flag is 85 turned on by many CONFIG_*DEBUG* options. 86 87 Usage: 88 89 Dynamic debugging is controlled via the 'dynamic_debug/control' file, 90 which is contained in the 'debugfs' filesystem. Thus, the debugfs 91 filesystem must first be mounted before making use of this feature. 92 We refer the control file as: <debugfs>/dynamic_debug/control. This 93 file contains a list of the debug statements that can be enabled. The 94 format for each line of the file is: 95 96 filename:lineno [module]function flags format 97 98 filename : source file of the debug statement 99 lineno : line number of the debug statement 100 module : module that contains the debug statement 101 function : function that contains the debug statement 102 flags : '=p' means the line is turned 'on' for printing 103 format : the format used for the debug statement 104 105 From a live system: 106 107 nullarbor:~ # cat <debugfs>/dynamic_debug/control 108 # filename:lineno [module]function flags format 109 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012" 110 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012" 111 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012" 112 113 Example usage: 114 115 // enable the message at line 1603 of file svcsock.c 116 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' > 117 <debugfs>/dynamic_debug/control 118 119 // enable all the messages in file svcsock.c 120 nullarbor:~ # echo -n 'file svcsock.c +p' > 121 <debugfs>/dynamic_debug/control 122 123 // enable all the messages in the NFS server module 124 nullarbor:~ # echo -n 'module nfsd +p' > 125 <debugfs>/dynamic_debug/control 126 127 // enable all 12 messages in the function svc_process() 128 nullarbor:~ # echo -n 'func svc_process +p' > 129 <debugfs>/dynamic_debug/control 130 131 // disable all 12 messages in the function svc_process() 132 nullarbor:~ # echo -n 'func svc_process -p' > 133 <debugfs>/dynamic_debug/control 134 135 See Documentation/admin-guide/dynamic-debug-howto.rst for additional 136 information. 137 138endmenu # "printk and dmesg options" 139 140menu "Compile-time checks and compiler options" 141 142config DEBUG_INFO 143 bool "Compile the kernel with debug info" 144 depends on DEBUG_KERNEL && !COMPILE_TEST 145 help 146 If you say Y here the resulting kernel image will include 147 debugging info resulting in a larger kernel image. 148 This adds debug symbols to the kernel and modules (gcc -g), and 149 is needed if you intend to use kernel crashdump or binary object 150 tools like crash, kgdb, LKCD, gdb, etc on the kernel. 151 Say Y here only if you plan to debug the kernel. 152 153 If unsure, say N. 154 155config DEBUG_INFO_REDUCED 156 bool "Reduce debugging information" 157 depends on DEBUG_INFO 158 help 159 If you say Y here gcc is instructed to generate less debugging 160 information for structure types. This means that tools that 161 need full debugging information (like kgdb or systemtap) won't 162 be happy. But if you merely need debugging information to 163 resolve line numbers there is no loss. Advantage is that 164 build directory object sizes shrink dramatically over a full 165 DEBUG_INFO build and compile times are reduced too. 166 Only works with newer gcc versions. 167 168config DEBUG_INFO_SPLIT 169 bool "Produce split debuginfo in .dwo files" 170 depends on DEBUG_INFO 171 help 172 Generate debug info into separate .dwo files. This significantly 173 reduces the build directory size for builds with DEBUG_INFO, 174 because it stores the information only once on disk in .dwo 175 files instead of multiple times in object files and executables. 176 In addition the debug information is also compressed. 177 178 Requires recent gcc (4.7+) and recent gdb/binutils. 179 Any tool that packages or reads debug information would need 180 to know about the .dwo files and include them. 181 Incompatible with older versions of ccache. 182 183config DEBUG_INFO_DWARF4 184 bool "Generate dwarf4 debuginfo" 185 depends on DEBUG_INFO 186 help 187 Generate dwarf4 debug info. This requires recent versions 188 of gcc and gdb. It makes the debug information larger. 189 But it significantly improves the success of resolving 190 variables in gdb on optimized code. 191 192config GDB_SCRIPTS 193 bool "Provide GDB scripts for kernel debugging" 194 depends on DEBUG_INFO 195 help 196 This creates the required links to GDB helper scripts in the 197 build directory. If you load vmlinux into gdb, the helper 198 scripts will be automatically imported by gdb as well, and 199 additional functions are available to analyze a Linux kernel 200 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst 201 for further details. 202 203config ENABLE_WARN_DEPRECATED 204 bool "Enable __deprecated logic" 205 default y 206 help 207 Enable the __deprecated logic in the kernel build. 208 Disable this to suppress the "warning: 'foo' is deprecated 209 (declared at kernel/power/somefile.c:1234)" messages. 210 211config ENABLE_MUST_CHECK 212 bool "Enable __must_check logic" 213 default y 214 help 215 Enable the __must_check logic in the kernel build. Disable this to 216 suppress the "warning: ignoring return value of 'foo', declared with 217 attribute warn_unused_result" messages. 218 219config FRAME_WARN 220 int "Warn for stack frames larger than (needs gcc 4.4)" 221 range 0 8192 222 default 3072 if KASAN_EXTRA 223 default 2048 if GCC_PLUGIN_LATENT_ENTROPY 224 default 1280 if (!64BIT && PARISC) 225 default 1024 if (!64BIT && !PARISC) 226 default 2048 if 64BIT 227 help 228 Tell gcc to warn at build time for stack frames larger than this. 229 Setting this too low will cause a lot of warnings. 230 Setting it to 0 disables the warning. 231 Requires gcc 4.4 232 233config STRIP_ASM_SYMS 234 bool "Strip assembler-generated symbols during link" 235 default n 236 help 237 Strip internal assembler-generated symbols during a link (symbols 238 that look like '.Lxxx') so they don't pollute the output of 239 get_wchan() and suchlike. 240 241config READABLE_ASM 242 bool "Generate readable assembler code" 243 depends on DEBUG_KERNEL 244 help 245 Disable some compiler optimizations that tend to generate human unreadable 246 assembler output. This may make the kernel slightly slower, but it helps 247 to keep kernel developers who have to stare a lot at assembler listings 248 sane. 249 250config UNUSED_SYMBOLS 251 bool "Enable unused/obsolete exported symbols" 252 default y if X86 253 help 254 Unused but exported symbols make the kernel needlessly bigger. For 255 that reason most of these unused exports will soon be removed. This 256 option is provided temporarily to provide a transition period in case 257 some external kernel module needs one of these symbols anyway. If you 258 encounter such a case in your module, consider if you are actually 259 using the right API. (rationale: since nobody in the kernel is using 260 this in a module, there is a pretty good chance it's actually the 261 wrong interface to use). If you really need the symbol, please send a 262 mail to the linux kernel mailing list mentioning the symbol and why 263 you really need it, and what the merge plan to the mainline kernel for 264 your module is. 265 266config PAGE_OWNER 267 bool "Track page owner" 268 depends on DEBUG_KERNEL && STACKTRACE_SUPPORT 269 select DEBUG_FS 270 select STACKTRACE 271 select STACKDEPOT 272 select PAGE_EXTENSION 273 help 274 This keeps track of what call chain is the owner of a page, may 275 help to find bare alloc_page(s) leaks. Even if you include this 276 feature on your build, it is disabled in default. You should pass 277 "page_owner=on" to boot parameter in order to enable it. Eats 278 a fair amount of memory if enabled. See tools/vm/page_owner_sort.c 279 for user-space helper. 280 281 If unsure, say N. 282 283config DEBUG_FS 284 bool "Debug Filesystem" 285 help 286 debugfs is a virtual file system that kernel developers use to put 287 debugging files into. Enable this option to be able to read and 288 write to these files. 289 290 For detailed documentation on the debugfs API, see 291 Documentation/filesystems/. 292 293 If unsure, say N. 294 295config HEADERS_CHECK 296 bool "Run 'make headers_check' when building vmlinux" 297 depends on !UML 298 help 299 This option will extract the user-visible kernel headers whenever 300 building the kernel, and will run basic sanity checks on them to 301 ensure that exported files do not attempt to include files which 302 were not exported, etc. 303 304 If you're making modifications to header files which are 305 relevant for userspace, say 'Y', and check the headers 306 exported to $(INSTALL_HDR_PATH) (usually 'usr/include' in 307 your build tree), to make sure they're suitable. 308 309config DEBUG_SECTION_MISMATCH 310 bool "Enable full Section mismatch analysis" 311 help 312 The section mismatch analysis checks if there are illegal 313 references from one section to another section. 314 During linktime or runtime, some sections are dropped; 315 any use of code/data previously in these sections would 316 most likely result in an oops. 317 In the code, functions and variables are annotated with 318 __init,, etc. (see the full list in include/linux/init.h), 319 which results in the code/data being placed in specific sections. 320 The section mismatch analysis is always performed after a full 321 kernel build, and enabling this option causes the following 322 additional steps to occur: 323 - Add the option -fno-inline-functions-called-once to gcc commands. 324 When inlining a function annotated with __init in a non-init 325 function, we would lose the section information and thus 326 the analysis would not catch the illegal reference. 327 This option tells gcc to inline less (but it does result in 328 a larger kernel). 329 - Run the section mismatch analysis for each module/built-in.a file. 330 When we run the section mismatch analysis on vmlinux.o, we 331 lose valuable information about where the mismatch was 332 introduced. 333 Running the analysis for each module/built-in.a file 334 tells where the mismatch happens much closer to the 335 source. The drawback is that the same mismatch is 336 reported at least twice. 337 - Enable verbose reporting from modpost in order to help resolve 338 the section mismatches that are reported. 339 340config SECTION_MISMATCH_WARN_ONLY 341 bool "Make section mismatch errors non-fatal" 342 default y 343 help 344 If you say N here, the build process will fail if there are any 345 section mismatch, instead of just throwing warnings. 346 347 If unsure, say Y. 348 349# 350# Select this config option from the architecture Kconfig, if it 351# is preferred to always offer frame pointers as a config 352# option on the architecture (regardless of KERNEL_DEBUG): 353# 354config ARCH_WANT_FRAME_POINTERS 355 bool 356 357config FRAME_POINTER 358 bool "Compile the kernel with frame pointers" 359 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS 360 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS 361 help 362 If you say Y here the resulting kernel image will be slightly 363 larger and slower, but it gives very useful debugging information 364 in case of kernel bugs. (precise oopses/stacktraces/warnings) 365 366config STACK_VALIDATION 367 bool "Compile-time stack metadata validation" 368 depends on HAVE_STACK_VALIDATION 369 default n 370 help 371 Add compile-time checks to validate stack metadata, including frame 372 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure 373 that runtime stack traces are more reliable. 374 375 This is also a prerequisite for generation of ORC unwind data, which 376 is needed for CONFIG_UNWINDER_ORC. 377 378 For more information, see 379 tools/objtool/Documentation/stack-validation.txt. 380 381config DEBUG_FORCE_WEAK_PER_CPU 382 bool "Force weak per-cpu definitions" 383 depends on DEBUG_KERNEL 384 help 385 s390 and alpha require percpu variables in modules to be 386 defined weak to work around addressing range issue which 387 puts the following two restrictions on percpu variable 388 definitions. 389 390 1. percpu symbols must be unique whether static or not 391 2. percpu variables can't be defined inside a function 392 393 To ensure that generic code follows the above rules, this 394 option forces all percpu variables to be defined as weak. 395 396endmenu # "Compiler options" 397 398config MAGIC_SYSRQ 399 bool "Magic SysRq key" 400 depends on !UML 401 help 402 If you say Y here, you will have some control over the system even 403 if the system crashes for example during kernel debugging (e.g., you 404 will be able to flush the buffer cache to disk, reboot the system 405 immediately or dump some status information). This is accomplished 406 by pressing various keys while holding SysRq (Alt+PrintScreen). It 407 also works on a serial console (on PC hardware at least), if you 408 send a BREAK and then within 5 seconds a command keypress. The 409 keys are documented in <file:Documentation/admin-guide/sysrq.rst>. 410 Don't say Y unless you really know what this hack does. 411 412config MAGIC_SYSRQ_DEFAULT_ENABLE 413 hex "Enable magic SysRq key functions by default" 414 depends on MAGIC_SYSRQ 415 default 0x1 416 help 417 Specifies which SysRq key functions are enabled by default. 418 This may be set to 1 or 0 to enable or disable them all, or 419 to a bitmask as described in Documentation/admin-guide/sysrq.rst. 420 421config MAGIC_SYSRQ_SERIAL 422 bool "Enable magic SysRq key over serial" 423 depends on MAGIC_SYSRQ 424 default y 425 help 426 Many embedded boards have a disconnected TTL level serial which can 427 generate some garbage that can lead to spurious false sysrq detects. 428 This option allows you to decide whether you want to enable the 429 magic SysRq key. 430 431config DEBUG_KERNEL 432 bool "Kernel debugging" 433 help 434 Say Y here if you are developing drivers or trying to debug and 435 identify kernel problems. 436 437menu "Memory Debugging" 438 439source mm/Kconfig.debug 440 441config DEBUG_OBJECTS 442 bool "Debug object operations" 443 depends on DEBUG_KERNEL 444 help 445 If you say Y here, additional code will be inserted into the 446 kernel to track the life time of various objects and validate 447 the operations on those objects. 448 449config DEBUG_OBJECTS_SELFTEST 450 bool "Debug objects selftest" 451 depends on DEBUG_OBJECTS 452 help 453 This enables the selftest of the object debug code. 454 455config DEBUG_OBJECTS_FREE 456 bool "Debug objects in freed memory" 457 depends on DEBUG_OBJECTS 458 help 459 This enables checks whether a k/v free operation frees an area 460 which contains an object which has not been deactivated 461 properly. This can make kmalloc/kfree-intensive workloads 462 much slower. 463 464config DEBUG_OBJECTS_TIMERS 465 bool "Debug timer objects" 466 depends on DEBUG_OBJECTS 467 help 468 If you say Y here, additional code will be inserted into the 469 timer routines to track the life time of timer objects and 470 validate the timer operations. 471 472config DEBUG_OBJECTS_WORK 473 bool "Debug work objects" 474 depends on DEBUG_OBJECTS 475 help 476 If you say Y here, additional code will be inserted into the 477 work queue routines to track the life time of work objects and 478 validate the work operations. 479 480config DEBUG_OBJECTS_RCU_HEAD 481 bool "Debug RCU callbacks objects" 482 depends on DEBUG_OBJECTS 483 help 484 Enable this to turn on debugging of RCU list heads (call_rcu() usage). 485 486config DEBUG_OBJECTS_PERCPU_COUNTER 487 bool "Debug percpu counter objects" 488 depends on DEBUG_OBJECTS 489 help 490 If you say Y here, additional code will be inserted into the 491 percpu counter routines to track the life time of percpu counter 492 objects and validate the percpu counter operations. 493 494config DEBUG_OBJECTS_ENABLE_DEFAULT 495 int "debug_objects bootup default value (0-1)" 496 range 0 1 497 default "1" 498 depends on DEBUG_OBJECTS 499 help 500 Debug objects boot parameter default value 501 502config DEBUG_SLAB 503 bool "Debug slab memory allocations" 504 depends on DEBUG_KERNEL && SLAB 505 help 506 Say Y here to have the kernel do limited verification on memory 507 allocation as well as poisoning memory on free to catch use of freed 508 memory. This can make kmalloc/kfree-intensive workloads much slower. 509 510config DEBUG_SLAB_LEAK 511 bool "Memory leak debugging" 512 depends on DEBUG_SLAB 513 514config SLUB_DEBUG_ON 515 bool "SLUB debugging on by default" 516 depends on SLUB && SLUB_DEBUG 517 default n 518 help 519 Boot with debugging on by default. SLUB boots by default with 520 the runtime debug capabilities switched off. Enabling this is 521 equivalent to specifying the "slub_debug" parameter on boot. 522 There is no support for more fine grained debug control like 523 possible with slub_debug=xxx. SLUB debugging may be switched 524 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying 525 "slub_debug=-". 526 527config SLUB_STATS 528 default n 529 bool "Enable SLUB performance statistics" 530 depends on SLUB && SYSFS 531 help 532 SLUB statistics are useful to debug SLUBs allocation behavior in 533 order find ways to optimize the allocator. This should never be 534 enabled for production use since keeping statistics slows down 535 the allocator by a few percentage points. The slabinfo command 536 supports the determination of the most active slabs to figure 537 out which slabs are relevant to a particular load. 538 Try running: slabinfo -DA 539 540config HAVE_DEBUG_KMEMLEAK 541 bool 542 543config DEBUG_KMEMLEAK 544 bool "Kernel memory leak detector" 545 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK 546 select DEBUG_FS 547 select STACKTRACE if STACKTRACE_SUPPORT 548 select KALLSYMS 549 select CRC32 550 help 551 Say Y here if you want to enable the memory leak 552 detector. The memory allocation/freeing is traced in a way 553 similar to the Boehm's conservative garbage collector, the 554 difference being that the orphan objects are not freed but 555 only shown in /sys/kernel/debug/kmemleak. Enabling this 556 feature will introduce an overhead to memory 557 allocations. See Documentation/dev-tools/kmemleak.rst for more 558 details. 559 560 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances 561 of finding leaks due to the slab objects poisoning. 562 563 In order to access the kmemleak file, debugfs needs to be 564 mounted (usually at /sys/kernel/debug). 565 566config DEBUG_KMEMLEAK_EARLY_LOG_SIZE 567 int "Maximum kmemleak early log entries" 568 depends on DEBUG_KMEMLEAK 569 range 200 40000 570 default 400 571 help 572 Kmemleak must track all the memory allocations to avoid 573 reporting false positives. Since memory may be allocated or 574 freed before kmemleak is initialised, an early log buffer is 575 used to store these actions. If kmemleak reports "early log 576 buffer exceeded", please increase this value. 577 578config DEBUG_KMEMLEAK_TEST 579 tristate "Simple test for the kernel memory leak detector" 580 depends on DEBUG_KMEMLEAK && m 581 help 582 This option enables a module that explicitly leaks memory. 583 584 If unsure, say N. 585 586config DEBUG_KMEMLEAK_DEFAULT_OFF 587 bool "Default kmemleak to off" 588 depends on DEBUG_KMEMLEAK 589 help 590 Say Y here to disable kmemleak by default. It can then be enabled 591 on the command line via kmemleak=on. 592 593config DEBUG_STACK_USAGE 594 bool "Stack utilization instrumentation" 595 depends on DEBUG_KERNEL && !IA64 596 help 597 Enables the display of the minimum amount of free stack which each 598 task has ever had available in the sysrq-T and sysrq-P debug output. 599 600 This option will slow down process creation somewhat. 601 602config DEBUG_VM 603 bool "Debug VM" 604 depends on DEBUG_KERNEL 605 help 606 Enable this to turn on extended checks in the virtual-memory system 607 that may impact performance. 608 609 If unsure, say N. 610 611config DEBUG_VM_VMACACHE 612 bool "Debug VMA caching" 613 depends on DEBUG_VM 614 help 615 Enable this to turn on VMA caching debug information. Doing so 616 can cause significant overhead, so only enable it in non-production 617 environments. 618 619 If unsure, say N. 620 621config DEBUG_VM_RB 622 bool "Debug VM red-black trees" 623 depends on DEBUG_VM 624 help 625 Enable VM red-black tree debugging information and extra validations. 626 627 If unsure, say N. 628 629config DEBUG_VM_PGFLAGS 630 bool "Debug page-flags operations" 631 depends on DEBUG_VM 632 help 633 Enables extra validation on page flags operations. 634 635 If unsure, say N. 636 637config ARCH_HAS_DEBUG_VIRTUAL 638 bool 639 640config DEBUG_VIRTUAL 641 bool "Debug VM translations" 642 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL 643 help 644 Enable some costly sanity checks in virtual to page code. This can 645 catch mistakes with virt_to_page() and friends. 646 647 If unsure, say N. 648 649config DEBUG_NOMMU_REGIONS 650 bool "Debug the global anon/private NOMMU mapping region tree" 651 depends on DEBUG_KERNEL && !MMU 652 help 653 This option causes the global tree of anonymous and private mapping 654 regions to be regularly checked for invalid topology. 655 656config DEBUG_MEMORY_INIT 657 bool "Debug memory initialisation" if EXPERT 658 default !EXPERT 659 help 660 Enable this for additional checks during memory initialisation. 661 The sanity checks verify aspects of the VM such as the memory model 662 and other information provided by the architecture. Verbose 663 information will be printed at KERN_DEBUG loglevel depending 664 on the mminit_loglevel= command-line option. 665 666 If unsure, say Y 667 668config MEMORY_NOTIFIER_ERROR_INJECT 669 tristate "Memory hotplug notifier error injection module" 670 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION 671 help 672 This option provides the ability to inject artificial errors to 673 memory hotplug notifier chain callbacks. It is controlled through 674 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory 675 676 If the notifier call chain should be failed with some events 677 notified, write the error code to "actions/<notifier event>/error". 678 679 Example: Inject memory hotplug offline error (-12 == -ENOMEM) 680 681 # cd /sys/kernel/debug/notifier-error-inject/memory 682 # echo -12 > actions/MEM_GOING_OFFLINE/error 683 # echo offline > /sys/devices/system/memory/memoryXXX/state 684 bash: echo: write error: Cannot allocate memory 685 686 To compile this code as a module, choose M here: the module will 687 be called memory-notifier-error-inject. 688 689 If unsure, say N. 690 691config DEBUG_PER_CPU_MAPS 692 bool "Debug access to per_cpu maps" 693 depends on DEBUG_KERNEL 694 depends on SMP 695 help 696 Say Y to verify that the per_cpu map being accessed has 697 been set up. This adds a fair amount of code to kernel memory 698 and decreases performance. 699 700 Say N if unsure. 701 702config DEBUG_HIGHMEM 703 bool "Highmem debugging" 704 depends on DEBUG_KERNEL && HIGHMEM 705 help 706 This option enables additional error checking for high memory 707 systems. Disable for production systems. 708 709config HAVE_DEBUG_STACKOVERFLOW 710 bool 711 712config DEBUG_STACKOVERFLOW 713 bool "Check for stack overflows" 714 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW 715 ---help--- 716 Say Y here if you want to check for overflows of kernel, IRQ 717 and exception stacks (if your architecture uses them). This 718 option will show detailed messages if free stack space drops 719 below a certain limit. 720 721 These kinds of bugs usually occur when call-chains in the 722 kernel get too deep, especially when interrupts are 723 involved. 724 725 Use this in cases where you see apparently random memory 726 corruption, especially if it appears in 'struct thread_info' 727 728 If in doubt, say "N". 729 730source "lib/Kconfig.kasan" 731 732endmenu # "Memory Debugging" 733 734config ARCH_HAS_KCOV 735 bool 736 help 737 KCOV does not have any arch-specific code, but currently it is enabled 738 only for x86_64. KCOV requires testing on other archs, and most likely 739 disabling of instrumentation for some early boot code. 740 741config CC_HAS_SANCOV_TRACE_PC 742 def_bool $(cc-option,-fsanitize-coverage=trace-pc) 743 744config KCOV 745 bool "Code coverage for fuzzing" 746 depends on ARCH_HAS_KCOV 747 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS 748 select DEBUG_FS 749 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC 750 help 751 KCOV exposes kernel code coverage information in a form suitable 752 for coverage-guided fuzzing (randomized testing). 753 754 If RANDOMIZE_BASE is enabled, PC values will not be stable across 755 different machines and across reboots. If you need stable PC values, 756 disable RANDOMIZE_BASE. 757 758 For more details, see Documentation/dev-tools/kcov.rst. 759 760config KCOV_ENABLE_COMPARISONS 761 bool "Enable comparison operands collection by KCOV" 762 depends on KCOV 763 depends on $(cc-option,-fsanitize-coverage=trace-cmp) 764 help 765 KCOV also exposes operands of every comparison in the instrumented 766 code along with operand sizes and PCs of the comparison instructions. 767 These operands can be used by fuzzing engines to improve the quality 768 of fuzzing coverage. 769 770config KCOV_INSTRUMENT_ALL 771 bool "Instrument all code by default" 772 depends on KCOV 773 default y 774 help 775 If you are doing generic system call fuzzing (like e.g. syzkaller), 776 then you will want to instrument the whole kernel and you should 777 say y here. If you are doing more targeted fuzzing (like e.g. 778 filesystem fuzzing with AFL) then you will want to enable coverage 779 for more specific subsets of files, and should say n here. 780 781config DEBUG_SHIRQ 782 bool "Debug shared IRQ handlers" 783 depends on DEBUG_KERNEL 784 help 785 Enable this to generate a spurious interrupt as soon as a shared 786 interrupt handler is registered, and just before one is deregistered. 787 Drivers ought to be able to handle interrupts coming in at those 788 points; some don't and need to be caught. 789 790menu "Debug Lockups and Hangs" 791 792config LOCKUP_DETECTOR 793 bool 794 795config SOFTLOCKUP_DETECTOR 796 bool "Detect Soft Lockups" 797 depends on DEBUG_KERNEL && !S390 798 select LOCKUP_DETECTOR 799 help 800 Say Y here to enable the kernel to act as a watchdog to detect 801 soft lockups. 802 803 Softlockups are bugs that cause the kernel to loop in kernel 804 mode for more than 20 seconds, without giving other tasks a 805 chance to run. The current stack trace is displayed upon 806 detection and the system will stay locked up. 807 808config BOOTPARAM_SOFTLOCKUP_PANIC 809 bool "Panic (Reboot) On Soft Lockups" 810 depends on SOFTLOCKUP_DETECTOR 811 help 812 Say Y here to enable the kernel to panic on "soft lockups", 813 which are bugs that cause the kernel to loop in kernel 814 mode for more than 20 seconds (configurable using the watchdog_thresh 815 sysctl), without giving other tasks a chance to run. 816 817 The panic can be used in combination with panic_timeout, 818 to cause the system to reboot automatically after a 819 lockup has been detected. This feature is useful for 820 high-availability systems that have uptime guarantees and 821 where a lockup must be resolved ASAP. 822 823 Say N if unsure. 824 825config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE 826 int 827 depends on SOFTLOCKUP_DETECTOR 828 range 0 1 829 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC 830 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC 831 832config HARDLOCKUP_DETECTOR_PERF 833 bool 834 select SOFTLOCKUP_DETECTOR 835 836# 837# Enables a timestamp based low pass filter to compensate for perf based 838# hard lockup detection which runs too fast due to turbo modes. 839# 840config HARDLOCKUP_CHECK_TIMESTAMP 841 bool 842 843# 844# arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard 845# lockup detector rather than the perf based detector. 846# 847config HARDLOCKUP_DETECTOR 848 bool "Detect Hard Lockups" 849 depends on DEBUG_KERNEL && !S390 850 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_ARCH 851 select LOCKUP_DETECTOR 852 select HARDLOCKUP_DETECTOR_PERF if HAVE_HARDLOCKUP_DETECTOR_PERF 853 select HARDLOCKUP_DETECTOR_ARCH if HAVE_HARDLOCKUP_DETECTOR_ARCH 854 help 855 Say Y here to enable the kernel to act as a watchdog to detect 856 hard lockups. 857 858 Hardlockups are bugs that cause the CPU to loop in kernel mode 859 for more than 10 seconds, without letting other interrupts have a 860 chance to run. The current stack trace is displayed upon detection 861 and the system will stay locked up. 862 863config BOOTPARAM_HARDLOCKUP_PANIC 864 bool "Panic (Reboot) On Hard Lockups" 865 depends on HARDLOCKUP_DETECTOR 866 help 867 Say Y here to enable the kernel to panic on "hard lockups", 868 which are bugs that cause the kernel to loop in kernel 869 mode with interrupts disabled for more than 10 seconds (configurable 870 using the watchdog_thresh sysctl). 871 872 Say N if unsure. 873 874config BOOTPARAM_HARDLOCKUP_PANIC_VALUE 875 int 876 depends on HARDLOCKUP_DETECTOR 877 range 0 1 878 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC 879 default 1 if BOOTPARAM_HARDLOCKUP_PANIC 880 881config DETECT_HUNG_TASK 882 bool "Detect Hung Tasks" 883 depends on DEBUG_KERNEL 884 default SOFTLOCKUP_DETECTOR 885 help 886 Say Y here to enable the kernel to detect "hung tasks", 887 which are bugs that cause the task to be stuck in 888 uninterruptible "D" state indefinitely. 889 890 When a hung task is detected, the kernel will print the 891 current stack trace (which you should report), but the 892 task will stay in uninterruptible state. If lockdep is 893 enabled then all held locks will also be reported. This 894 feature has negligible overhead. 895 896config DEFAULT_HUNG_TASK_TIMEOUT 897 int "Default timeout for hung task detection (in seconds)" 898 depends on DETECT_HUNG_TASK 899 default 120 900 help 901 This option controls the default timeout (in seconds) used 902 to determine when a task has become non-responsive and should 903 be considered hung. 904 905 It can be adjusted at runtime via the kernel.hung_task_timeout_secs 906 sysctl or by writing a value to 907 /proc/sys/kernel/hung_task_timeout_secs. 908 909 A timeout of 0 disables the check. The default is two minutes. 910 Keeping the default should be fine in most cases. 911 912config BOOTPARAM_HUNG_TASK_PANIC 913 bool "Panic (Reboot) On Hung Tasks" 914 depends on DETECT_HUNG_TASK 915 help 916 Say Y here to enable the kernel to panic on "hung tasks", 917 which are bugs that cause the kernel to leave a task stuck 918 in uninterruptible "D" state. 919 920 The panic can be used in combination with panic_timeout, 921 to cause the system to reboot automatically after a 922 hung task has been detected. This feature is useful for 923 high-availability systems that have uptime guarantees and 924 where a hung tasks must be resolved ASAP. 925 926 Say N if unsure. 927 928config BOOTPARAM_HUNG_TASK_PANIC_VALUE 929 int 930 depends on DETECT_HUNG_TASK 931 range 0 1 932 default 0 if !BOOTPARAM_HUNG_TASK_PANIC 933 default 1 if BOOTPARAM_HUNG_TASK_PANIC 934 935config WQ_WATCHDOG 936 bool "Detect Workqueue Stalls" 937 depends on DEBUG_KERNEL 938 help 939 Say Y here to enable stall detection on workqueues. If a 940 worker pool doesn't make forward progress on a pending work 941 item for over a given amount of time, 30s by default, a 942 warning message is printed along with dump of workqueue 943 state. This can be configured through kernel parameter 944 "workqueue.watchdog_thresh" and its sysfs counterpart. 945 946endmenu # "Debug lockups and hangs" 947 948config PANIC_ON_OOPS 949 bool "Panic on Oops" 950 help 951 Say Y here to enable the kernel to panic when it oopses. This 952 has the same effect as setting oops=panic on the kernel command 953 line. 954 955 This feature is useful to ensure that the kernel does not do 956 anything erroneous after an oops which could result in data 957 corruption or other issues. 958 959 Say N if unsure. 960 961config PANIC_ON_OOPS_VALUE 962 int 963 range 0 1 964 default 0 if !PANIC_ON_OOPS 965 default 1 if PANIC_ON_OOPS 966 967config PANIC_TIMEOUT 968 int "panic timeout" 969 default 0 970 help 971 Set the timeout value (in seconds) until a reboot occurs when the 972 the kernel panics. If n = 0, then we wait forever. A timeout 973 value n > 0 will wait n seconds before rebooting, while a timeout 974 value n < 0 will reboot immediately. 975 976config SCHED_DEBUG 977 bool "Collect scheduler debugging info" 978 depends on DEBUG_KERNEL && PROC_FS 979 default y 980 help 981 If you say Y here, the /proc/sched_debug file will be provided 982 that can help debug the scheduler. The runtime overhead of this 983 option is minimal. 984 985config SCHED_INFO 986 bool 987 default n 988 989config SCHEDSTATS 990 bool "Collect scheduler statistics" 991 depends on DEBUG_KERNEL && PROC_FS 992 select SCHED_INFO 993 help 994 If you say Y here, additional code will be inserted into the 995 scheduler and related routines to collect statistics about 996 scheduler behavior and provide them in /proc/schedstat. These 997 stats may be useful for both tuning and debugging the scheduler 998 If you aren't debugging the scheduler or trying to tune a specific 999 application, you can say N to avoid the very slight overhead 1000 this adds. 1001 1002config SCHED_STACK_END_CHECK 1003 bool "Detect stack corruption on calls to schedule()" 1004 depends on DEBUG_KERNEL 1005 default n 1006 help 1007 This option checks for a stack overrun on calls to schedule(). 1008 If the stack end location is found to be over written always panic as 1009 the content of the corrupted region can no longer be trusted. 1010 This is to ensure no erroneous behaviour occurs which could result in 1011 data corruption or a sporadic crash at a later stage once the region 1012 is examined. The runtime overhead introduced is minimal. 1013 1014config DEBUG_TIMEKEEPING 1015 bool "Enable extra timekeeping sanity checking" 1016 help 1017 This option will enable additional timekeeping sanity checks 1018 which may be helpful when diagnosing issues where timekeeping 1019 problems are suspected. 1020 1021 This may include checks in the timekeeping hotpaths, so this 1022 option may have a (very small) performance impact to some 1023 workloads. 1024 1025 If unsure, say N. 1026 1027config DEBUG_PREEMPT 1028 bool "Debug preemptible kernel" 1029 depends on DEBUG_KERNEL && PREEMPT && TRACE_IRQFLAGS_SUPPORT 1030 default y 1031 help 1032 If you say Y here then the kernel will use a debug variant of the 1033 commonly used smp_processor_id() function and will print warnings 1034 if kernel code uses it in a preemption-unsafe way. Also, the kernel 1035 will detect preemption count underflows. 1036 1037menu "Lock Debugging (spinlocks, mutexes, etc...)" 1038 1039config LOCK_DEBUGGING_SUPPORT 1040 bool 1041 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 1042 default y 1043 1044config PROVE_LOCKING 1045 bool "Lock debugging: prove locking correctness" 1046 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1047 select LOCKDEP 1048 select DEBUG_SPINLOCK 1049 select DEBUG_MUTEXES 1050 select DEBUG_RT_MUTEXES if RT_MUTEXES 1051 select DEBUG_RWSEMS if RWSEM_SPIN_ON_OWNER 1052 select DEBUG_WW_MUTEX_SLOWPATH 1053 select DEBUG_LOCK_ALLOC 1054 select TRACE_IRQFLAGS 1055 default n 1056 help 1057 This feature enables the kernel to prove that all locking 1058 that occurs in the kernel runtime is mathematically 1059 correct: that under no circumstance could an arbitrary (and 1060 not yet triggered) combination of observed locking 1061 sequences (on an arbitrary number of CPUs, running an 1062 arbitrary number of tasks and interrupt contexts) cause a 1063 deadlock. 1064 1065 In short, this feature enables the kernel to report locking 1066 related deadlocks before they actually occur. 1067 1068 The proof does not depend on how hard and complex a 1069 deadlock scenario would be to trigger: how many 1070 participant CPUs, tasks and irq-contexts would be needed 1071 for it to trigger. The proof also does not depend on 1072 timing: if a race and a resulting deadlock is possible 1073 theoretically (no matter how unlikely the race scenario 1074 is), it will be proven so and will immediately be 1075 reported by the kernel (once the event is observed that 1076 makes the deadlock theoretically possible). 1077 1078 If a deadlock is impossible (i.e. the locking rules, as 1079 observed by the kernel, are mathematically correct), the 1080 kernel reports nothing. 1081 1082 NOTE: this feature can also be enabled for rwlocks, mutexes 1083 and rwsems - in which case all dependencies between these 1084 different locking variants are observed and mapped too, and 1085 the proof of observed correctness is also maintained for an 1086 arbitrary combination of these separate locking variants. 1087 1088 For more details, see Documentation/locking/lockdep-design.txt. 1089 1090config LOCK_STAT 1091 bool "Lock usage statistics" 1092 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1093 select LOCKDEP 1094 select DEBUG_SPINLOCK 1095 select DEBUG_MUTEXES 1096 select DEBUG_RT_MUTEXES if RT_MUTEXES 1097 select DEBUG_LOCK_ALLOC 1098 default n 1099 help 1100 This feature enables tracking lock contention points 1101 1102 For more details, see Documentation/locking/lockstat.txt 1103 1104 This also enables lock events required by "perf lock", 1105 subcommand of perf. 1106 If you want to use "perf lock", you also need to turn on 1107 CONFIG_EVENT_TRACING. 1108 1109 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events. 1110 (CONFIG_LOCKDEP defines "acquire" and "release" events.) 1111 1112config DEBUG_RT_MUTEXES 1113 bool "RT Mutex debugging, deadlock detection" 1114 depends on DEBUG_KERNEL && RT_MUTEXES 1115 help 1116 This allows rt mutex semantics violations and rt mutex related 1117 deadlocks (lockups) to be detected and reported automatically. 1118 1119config DEBUG_SPINLOCK 1120 bool "Spinlock and rw-lock debugging: basic checks" 1121 depends on DEBUG_KERNEL 1122 select UNINLINE_SPIN_UNLOCK 1123 help 1124 Say Y here and build SMP to catch missing spinlock initialization 1125 and certain other kinds of spinlock errors commonly made. This is 1126 best used in conjunction with the NMI watchdog so that spinlock 1127 deadlocks are also debuggable. 1128 1129config DEBUG_MUTEXES 1130 bool "Mutex debugging: basic checks" 1131 depends on DEBUG_KERNEL 1132 help 1133 This feature allows mutex semantics violations to be detected and 1134 reported. 1135 1136config DEBUG_WW_MUTEX_SLOWPATH 1137 bool "Wait/wound mutex debugging: Slowpath testing" 1138 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1139 select DEBUG_LOCK_ALLOC 1140 select DEBUG_SPINLOCK 1141 select DEBUG_MUTEXES 1142 help 1143 This feature enables slowpath testing for w/w mutex users by 1144 injecting additional -EDEADLK wound/backoff cases. Together with 1145 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this 1146 will test all possible w/w mutex interface abuse with the 1147 exception of simply not acquiring all the required locks. 1148 Note that this feature can introduce significant overhead, so 1149 it really should not be enabled in a production or distro kernel, 1150 even a debug kernel. If you are a driver writer, enable it. If 1151 you are a distro, do not. 1152 1153config DEBUG_RWSEMS 1154 bool "RW Semaphore debugging: basic checks" 1155 depends on DEBUG_KERNEL && RWSEM_SPIN_ON_OWNER 1156 help 1157 This debugging feature allows mismatched rw semaphore locks and unlocks 1158 to be detected and reported. 1159 1160config DEBUG_LOCK_ALLOC 1161 bool "Lock debugging: detect incorrect freeing of live locks" 1162 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1163 select DEBUG_SPINLOCK 1164 select DEBUG_MUTEXES 1165 select DEBUG_RT_MUTEXES if RT_MUTEXES 1166 select LOCKDEP 1167 help 1168 This feature will check whether any held lock (spinlock, rwlock, 1169 mutex or rwsem) is incorrectly freed by the kernel, via any of the 1170 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(), 1171 vfree(), etc.), whether a live lock is incorrectly reinitialized via 1172 spin_lock_init()/mutex_init()/etc., or whether there is any lock 1173 held during task exit. 1174 1175config LOCKDEP 1176 bool 1177 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1178 select STACKTRACE 1179 select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE && !ARC && !X86 1180 select KALLSYMS 1181 select KALLSYMS_ALL 1182 1183config LOCKDEP_SMALL 1184 bool 1185 1186config DEBUG_LOCKDEP 1187 bool "Lock dependency engine debugging" 1188 depends on DEBUG_KERNEL && LOCKDEP 1189 help 1190 If you say Y here, the lock dependency engine will do 1191 additional runtime checks to debug itself, at the price 1192 of more runtime overhead. 1193 1194config DEBUG_ATOMIC_SLEEP 1195 bool "Sleep inside atomic section checking" 1196 select PREEMPT_COUNT 1197 depends on DEBUG_KERNEL 1198 depends on !ARCH_NO_PREEMPT 1199 help 1200 If you say Y here, various routines which may sleep will become very 1201 noisy if they are called inside atomic sections: when a spinlock is 1202 held, inside an rcu read side critical section, inside preempt disabled 1203 sections, inside an interrupt, etc... 1204 1205config DEBUG_LOCKING_API_SELFTESTS 1206 bool "Locking API boot-time self-tests" 1207 depends on DEBUG_KERNEL 1208 help 1209 Say Y here if you want the kernel to run a short self-test during 1210 bootup. The self-test checks whether common types of locking bugs 1211 are detected by debugging mechanisms or not. (if you disable 1212 lock debugging then those bugs wont be detected of course.) 1213 The following locking APIs are covered: spinlocks, rwlocks, 1214 mutexes and rwsems. 1215 1216config LOCK_TORTURE_TEST 1217 tristate "torture tests for locking" 1218 depends on DEBUG_KERNEL 1219 select TORTURE_TEST 1220 default n 1221 help 1222 This option provides a kernel module that runs torture tests 1223 on kernel locking primitives. The kernel module may be built 1224 after the fact on the running kernel to be tested, if desired. 1225 1226 Say Y here if you want kernel locking-primitive torture tests 1227 to be built into the kernel. 1228 Say M if you want these torture tests to build as a module. 1229 Say N if you are unsure. 1230 1231config WW_MUTEX_SELFTEST 1232 tristate "Wait/wound mutex selftests" 1233 help 1234 This option provides a kernel module that runs tests on the 1235 on the struct ww_mutex locking API. 1236 1237 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction 1238 with this test harness. 1239 1240 Say M if you want these self tests to build as a module. 1241 Say N if you are unsure. 1242 1243endmenu # lock debugging 1244 1245config TRACE_IRQFLAGS 1246 bool 1247 help 1248 Enables hooks to interrupt enabling and disabling for 1249 either tracing or lock debugging. 1250 1251config STACKTRACE 1252 bool "Stack backtrace support" 1253 depends on STACKTRACE_SUPPORT 1254 help 1255 This option causes the kernel to create a /proc/pid/stack for 1256 every process, showing its current stack trace. 1257 It is also used by various kernel debugging features that require 1258 stack trace generation. 1259 1260config WARN_ALL_UNSEEDED_RANDOM 1261 bool "Warn for all uses of unseeded randomness" 1262 default n 1263 help 1264 Some parts of the kernel contain bugs relating to their use of 1265 cryptographically secure random numbers before it's actually possible 1266 to generate those numbers securely. This setting ensures that these 1267 flaws don't go unnoticed, by enabling a message, should this ever 1268 occur. This will allow people with obscure setups to know when things 1269 are going wrong, so that they might contact developers about fixing 1270 it. 1271 1272 Unfortunately, on some models of some architectures getting 1273 a fully seeded CRNG is extremely difficult, and so this can 1274 result in dmesg getting spammed for a surprisingly long 1275 time. This is really bad from a security perspective, and 1276 so architecture maintainers really need to do what they can 1277 to get the CRNG seeded sooner after the system is booted. 1278 However, since users can not do anything actionble to 1279 address this, by default the kernel will issue only a single 1280 warning for the first use of unseeded randomness. 1281 1282 Say Y here if you want to receive warnings for all uses of 1283 unseeded randomness. This will be of use primarily for 1284 those developers interersted in improving the security of 1285 Linux kernels running on their architecture (or 1286 subarchitecture). 1287 1288config DEBUG_KOBJECT 1289 bool "kobject debugging" 1290 depends on DEBUG_KERNEL 1291 help 1292 If you say Y here, some extra kobject debugging messages will be sent 1293 to the syslog. 1294 1295config DEBUG_KOBJECT_RELEASE 1296 bool "kobject release debugging" 1297 depends on DEBUG_OBJECTS_TIMERS 1298 help 1299 kobjects are reference counted objects. This means that their 1300 last reference count put is not predictable, and the kobject can 1301 live on past the point at which a driver decides to drop it's 1302 initial reference to the kobject gained on allocation. An 1303 example of this would be a struct device which has just been 1304 unregistered. 1305 1306 However, some buggy drivers assume that after such an operation, 1307 the memory backing the kobject can be immediately freed. This 1308 goes completely against the principles of a refcounted object. 1309 1310 If you say Y here, the kernel will delay the release of kobjects 1311 on the last reference count to improve the visibility of this 1312 kind of kobject release bug. 1313 1314config HAVE_DEBUG_BUGVERBOSE 1315 bool 1316 1317config DEBUG_BUGVERBOSE 1318 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT 1319 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE) 1320 default y 1321 help 1322 Say Y here to make BUG() panics output the file name and line number 1323 of the BUG call as well as the EIP and oops trace. This aids 1324 debugging but costs about 70-100K of memory. 1325 1326config DEBUG_LIST 1327 bool "Debug linked list manipulation" 1328 depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION 1329 help 1330 Enable this to turn on extended checks in the linked-list 1331 walking routines. 1332 1333 If unsure, say N. 1334 1335config DEBUG_PI_LIST 1336 bool "Debug priority linked list manipulation" 1337 depends on DEBUG_KERNEL 1338 help 1339 Enable this to turn on extended checks in the priority-ordered 1340 linked-list (plist) walking routines. This checks the entire 1341 list multiple times during each manipulation. 1342 1343 If unsure, say N. 1344 1345config DEBUG_SG 1346 bool "Debug SG table operations" 1347 depends on DEBUG_KERNEL 1348 help 1349 Enable this to turn on checks on scatter-gather tables. This can 1350 help find problems with drivers that do not properly initialize 1351 their sg tables. 1352 1353 If unsure, say N. 1354 1355config DEBUG_NOTIFIERS 1356 bool "Debug notifier call chains" 1357 depends on DEBUG_KERNEL 1358 help 1359 Enable this to turn on sanity checking for notifier call chains. 1360 This is most useful for kernel developers to make sure that 1361 modules properly unregister themselves from notifier chains. 1362 This is a relatively cheap check but if you care about maximum 1363 performance, say N. 1364 1365config DEBUG_CREDENTIALS 1366 bool "Debug credential management" 1367 depends on DEBUG_KERNEL 1368 help 1369 Enable this to turn on some debug checking for credential 1370 management. The additional code keeps track of the number of 1371 pointers from task_structs to any given cred struct, and checks to 1372 see that this number never exceeds the usage count of the cred 1373 struct. 1374 1375 Furthermore, if SELinux is enabled, this also checks that the 1376 security pointer in the cred struct is never seen to be invalid. 1377 1378 If unsure, say N. 1379 1380source "kernel/rcu/Kconfig.debug" 1381 1382config DEBUG_WQ_FORCE_RR_CPU 1383 bool "Force round-robin CPU selection for unbound work items" 1384 depends on DEBUG_KERNEL 1385 default n 1386 help 1387 Workqueue used to implicitly guarantee that work items queued 1388 without explicit CPU specified are put on the local CPU. This 1389 guarantee is no longer true and while local CPU is still 1390 preferred work items may be put on foreign CPUs. Kernel 1391 parameter "workqueue.debug_force_rr_cpu" is added to force 1392 round-robin CPU selection to flush out usages which depend on the 1393 now broken guarantee. This config option enables the debug 1394 feature by default. When enabled, memory and cache locality will 1395 be impacted. 1396 1397config DEBUG_BLOCK_EXT_DEVT 1398 bool "Force extended block device numbers and spread them" 1399 depends on DEBUG_KERNEL 1400 depends on BLOCK 1401 default n 1402 help 1403 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON 1404 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT 1405 YOU ARE DOING. Distros, please enable this and fix whatever 1406 is broken. 1407 1408 Conventionally, block device numbers are allocated from 1409 predetermined contiguous area. However, extended block area 1410 may introduce non-contiguous block device numbers. This 1411 option forces most block device numbers to be allocated from 1412 the extended space and spreads them to discover kernel or 1413 userland code paths which assume predetermined contiguous 1414 device number allocation. 1415 1416 Note that turning on this debug option shuffles all the 1417 device numbers for all IDE and SCSI devices including libata 1418 ones, so root partition specified using device number 1419 directly (via rdev or root=MAJ:MIN) won't work anymore. 1420 Textual device names (root=/dev/sdXn) will continue to work. 1421 1422 Say N if you are unsure. 1423 1424config CPU_HOTPLUG_STATE_CONTROL 1425 bool "Enable CPU hotplug state control" 1426 depends on DEBUG_KERNEL 1427 depends on HOTPLUG_CPU 1428 default n 1429 help 1430 Allows to write steps between "offline" and "online" to the CPUs 1431 sysfs target file so states can be stepped granular. This is a debug 1432 option for now as the hotplug machinery cannot be stopped and 1433 restarted at arbitrary points yet. 1434 1435 Say N if your are unsure. 1436 1437config NOTIFIER_ERROR_INJECTION 1438 tristate "Notifier error injection" 1439 depends on DEBUG_KERNEL 1440 select DEBUG_FS 1441 help 1442 This option provides the ability to inject artificial errors to 1443 specified notifier chain callbacks. It is useful to test the error 1444 handling of notifier call chain failures. 1445 1446 Say N if unsure. 1447 1448config PM_NOTIFIER_ERROR_INJECT 1449 tristate "PM notifier error injection module" 1450 depends on PM && NOTIFIER_ERROR_INJECTION 1451 default m if PM_DEBUG 1452 help 1453 This option provides the ability to inject artificial errors to 1454 PM notifier chain callbacks. It is controlled through debugfs 1455 interface /sys/kernel/debug/notifier-error-inject/pm 1456 1457 If the notifier call chain should be failed with some events 1458 notified, write the error code to "actions/<notifier event>/error". 1459 1460 Example: Inject PM suspend error (-12 = -ENOMEM) 1461 1462 # cd /sys/kernel/debug/notifier-error-inject/pm/ 1463 # echo -12 > actions/PM_SUSPEND_PREPARE/error 1464 # echo mem > /sys/power/state 1465 bash: echo: write error: Cannot allocate memory 1466 1467 To compile this code as a module, choose M here: the module will 1468 be called pm-notifier-error-inject. 1469 1470 If unsure, say N. 1471 1472config OF_RECONFIG_NOTIFIER_ERROR_INJECT 1473 tristate "OF reconfig notifier error injection module" 1474 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION 1475 help 1476 This option provides the ability to inject artificial errors to 1477 OF reconfig notifier chain callbacks. It is controlled 1478 through debugfs interface under 1479 /sys/kernel/debug/notifier-error-inject/OF-reconfig/ 1480 1481 If the notifier call chain should be failed with some events 1482 notified, write the error code to "actions/<notifier event>/error". 1483 1484 To compile this code as a module, choose M here: the module will 1485 be called of-reconfig-notifier-error-inject. 1486 1487 If unsure, say N. 1488 1489config NETDEV_NOTIFIER_ERROR_INJECT 1490 tristate "Netdev notifier error injection module" 1491 depends on NET && NOTIFIER_ERROR_INJECTION 1492 help 1493 This option provides the ability to inject artificial errors to 1494 netdevice notifier chain callbacks. It is controlled through debugfs 1495 interface /sys/kernel/debug/notifier-error-inject/netdev 1496 1497 If the notifier call chain should be failed with some events 1498 notified, write the error code to "actions/<notifier event>/error". 1499 1500 Example: Inject netdevice mtu change error (-22 = -EINVAL) 1501 1502 # cd /sys/kernel/debug/notifier-error-inject/netdev 1503 # echo -22 > actions/NETDEV_CHANGEMTU/error 1504 # ip link set eth0 mtu 1024 1505 RTNETLINK answers: Invalid argument 1506 1507 To compile this code as a module, choose M here: the module will 1508 be called netdev-notifier-error-inject. 1509 1510 If unsure, say N. 1511 1512config FUNCTION_ERROR_INJECTION 1513 def_bool y 1514 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES 1515 1516config FAULT_INJECTION 1517 bool "Fault-injection framework" 1518 depends on DEBUG_KERNEL 1519 help 1520 Provide fault-injection framework. 1521 For more details, see Documentation/fault-injection/. 1522 1523config FAILSLAB 1524 bool "Fault-injection capability for kmalloc" 1525 depends on FAULT_INJECTION 1526 depends on SLAB || SLUB 1527 help 1528 Provide fault-injection capability for kmalloc. 1529 1530config FAIL_PAGE_ALLOC 1531 bool "Fault-injection capabilitiy for alloc_pages()" 1532 depends on FAULT_INJECTION 1533 help 1534 Provide fault-injection capability for alloc_pages(). 1535 1536config FAIL_MAKE_REQUEST 1537 bool "Fault-injection capability for disk IO" 1538 depends on FAULT_INJECTION && BLOCK 1539 help 1540 Provide fault-injection capability for disk IO. 1541 1542config FAIL_IO_TIMEOUT 1543 bool "Fault-injection capability for faking disk interrupts" 1544 depends on FAULT_INJECTION && BLOCK 1545 help 1546 Provide fault-injection capability on end IO handling. This 1547 will make the block layer "forget" an interrupt as configured, 1548 thus exercising the error handling. 1549 1550 Only works with drivers that use the generic timeout handling, 1551 for others it wont do anything. 1552 1553config FAIL_FUTEX 1554 bool "Fault-injection capability for futexes" 1555 select DEBUG_FS 1556 depends on FAULT_INJECTION && FUTEX 1557 help 1558 Provide fault-injection capability for futexes. 1559 1560config FAULT_INJECTION_DEBUG_FS 1561 bool "Debugfs entries for fault-injection capabilities" 1562 depends on FAULT_INJECTION && SYSFS && DEBUG_FS 1563 help 1564 Enable configuration of fault-injection capabilities via debugfs. 1565 1566config FAIL_FUNCTION 1567 bool "Fault-injection capability for functions" 1568 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION 1569 help 1570 Provide function-based fault-injection capability. 1571 This will allow you to override a specific function with a return 1572 with given return value. As a result, function caller will see 1573 an error value and have to handle it. This is useful to test the 1574 error handling in various subsystems. 1575 1576config FAIL_MMC_REQUEST 1577 bool "Fault-injection capability for MMC IO" 1578 depends on FAULT_INJECTION_DEBUG_FS && MMC 1579 help 1580 Provide fault-injection capability for MMC IO. 1581 This will make the mmc core return data errors. This is 1582 useful to test the error handling in the mmc block device 1583 and to test how the mmc host driver handles retries from 1584 the block device. 1585 1586config FAULT_INJECTION_STACKTRACE_FILTER 1587 bool "stacktrace filter for fault-injection capabilities" 1588 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT 1589 depends on !X86_64 1590 select STACKTRACE 1591 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !X86 1592 help 1593 Provide stacktrace filter for fault-injection capabilities 1594 1595config LATENCYTOP 1596 bool "Latency measuring infrastructure" 1597 depends on DEBUG_KERNEL 1598 depends on STACKTRACE_SUPPORT 1599 depends on PROC_FS 1600 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !X86 1601 select KALLSYMS 1602 select KALLSYMS_ALL 1603 select STACKTRACE 1604 select SCHEDSTATS 1605 select SCHED_DEBUG 1606 help 1607 Enable this option if you want to use the LatencyTOP tool 1608 to find out which userspace is blocking on what kernel operations. 1609 1610source kernel/trace/Kconfig 1611 1612config PROVIDE_OHCI1394_DMA_INIT 1613 bool "Remote debugging over FireWire early on boot" 1614 depends on PCI && X86 1615 help 1616 If you want to debug problems which hang or crash the kernel early 1617 on boot and the crashing machine has a FireWire port, you can use 1618 this feature to remotely access the memory of the crashed machine 1619 over FireWire. This employs remote DMA as part of the OHCI1394 1620 specification which is now the standard for FireWire controllers. 1621 1622 With remote DMA, you can monitor the printk buffer remotely using 1623 firescope and access all memory below 4GB using fireproxy from gdb. 1624 Even controlling a kernel debugger is possible using remote DMA. 1625 1626 Usage: 1627 1628 If ohci1394_dma=early is used as boot parameter, it will initialize 1629 all OHCI1394 controllers which are found in the PCI config space. 1630 1631 As all changes to the FireWire bus such as enabling and disabling 1632 devices cause a bus reset and thereby disable remote DMA for all 1633 devices, be sure to have the cable plugged and FireWire enabled on 1634 the debugging host before booting the debug target for debugging. 1635 1636 This code (~1k) is freed after boot. By then, the firewire stack 1637 in charge of the OHCI-1394 controllers should be used instead. 1638 1639 See Documentation/debugging-via-ohci1394.txt for more information. 1640 1641config DMA_API_DEBUG 1642 bool "Enable debugging of DMA-API usage" 1643 select NEED_DMA_MAP_STATE 1644 help 1645 Enable this option to debug the use of the DMA API by device drivers. 1646 With this option you will be able to detect common bugs in device 1647 drivers like double-freeing of DMA mappings or freeing mappings that 1648 were never allocated. 1649 1650 This also attempts to catch cases where a page owned by DMA is 1651 accessed by the cpu in a way that could cause data corruption. For 1652 example, this enables cow_user_page() to check that the source page is 1653 not undergoing DMA. 1654 1655 This option causes a performance degradation. Use only if you want to 1656 debug device drivers and dma interactions. 1657 1658 If unsure, say N. 1659 1660config DMA_API_DEBUG_SG 1661 bool "Debug DMA scatter-gather usage" 1662 default y 1663 depends on DMA_API_DEBUG 1664 help 1665 Perform extra checking that callers of dma_map_sg() have respected the 1666 appropriate segment length/boundary limits for the given device when 1667 preparing DMA scatterlists. 1668 1669 This is particularly likely to have been overlooked in cases where the 1670 dma_map_sg() API is used for general bulk mapping of pages rather than 1671 preparing literal scatter-gather descriptors, where there is a risk of 1672 unexpected behaviour from DMA API implementations if the scatterlist 1673 is technically out-of-spec. 1674 1675 If unsure, say N. 1676 1677menuconfig RUNTIME_TESTING_MENU 1678 bool "Runtime Testing" 1679 def_bool y 1680 1681if RUNTIME_TESTING_MENU 1682 1683config LKDTM 1684 tristate "Linux Kernel Dump Test Tool Module" 1685 depends on DEBUG_FS 1686 depends on BLOCK 1687 default n 1688 help 1689 This module enables testing of the different dumping mechanisms by 1690 inducing system failures at predefined crash points. 1691 If you don't need it: say N 1692 Choose M here to compile this code as a module. The module will be 1693 called lkdtm. 1694 1695 Documentation on how to use the module can be found in 1696 Documentation/fault-injection/provoke-crashes.txt 1697 1698config TEST_LIST_SORT 1699 tristate "Linked list sorting test" 1700 depends on DEBUG_KERNEL || m 1701 help 1702 Enable this to turn on 'list_sort()' function test. This test is 1703 executed only once during system boot (so affects only boot time), 1704 or at module load time. 1705 1706 If unsure, say N. 1707 1708config TEST_SORT 1709 tristate "Array-based sort test" 1710 depends on DEBUG_KERNEL || m 1711 help 1712 This option enables the self-test function of 'sort()' at boot, 1713 or at module load time. 1714 1715 If unsure, say N. 1716 1717config KPROBES_SANITY_TEST 1718 bool "Kprobes sanity tests" 1719 depends on DEBUG_KERNEL 1720 depends on KPROBES 1721 default n 1722 help 1723 This option provides for testing basic kprobes functionality on 1724 boot. A sample kprobe, jprobe and kretprobe are inserted and 1725 verified for functionality. 1726 1727 Say N if you are unsure. 1728 1729config BACKTRACE_SELF_TEST 1730 tristate "Self test for the backtrace code" 1731 depends on DEBUG_KERNEL 1732 default n 1733 help 1734 This option provides a kernel module that can be used to test 1735 the kernel stack backtrace code. This option is not useful 1736 for distributions or general kernels, but only for kernel 1737 developers working on architecture code. 1738 1739 Note that if you want to also test saved backtraces, you will 1740 have to enable STACKTRACE as well. 1741 1742 Say N if you are unsure. 1743 1744config RBTREE_TEST 1745 tristate "Red-Black tree test" 1746 depends on DEBUG_KERNEL 1747 help 1748 A benchmark measuring the performance of the rbtree library. 1749 Also includes rbtree invariant checks. 1750 1751config INTERVAL_TREE_TEST 1752 tristate "Interval tree test" 1753 depends on DEBUG_KERNEL 1754 select INTERVAL_TREE 1755 help 1756 A benchmark measuring the performance of the interval tree library 1757 1758config PERCPU_TEST 1759 tristate "Per cpu operations test" 1760 depends on m && DEBUG_KERNEL 1761 help 1762 Enable this option to build test module which validates per-cpu 1763 operations. 1764 1765 If unsure, say N. 1766 1767config ATOMIC64_SELFTEST 1768 tristate "Perform an atomic64_t self-test" 1769 help 1770 Enable this option to test the atomic64_t functions at boot or 1771 at module load time. 1772 1773 If unsure, say N. 1774 1775config ASYNC_RAID6_TEST 1776 tristate "Self test for hardware accelerated raid6 recovery" 1777 depends on ASYNC_RAID6_RECOV 1778 select ASYNC_MEMCPY 1779 ---help--- 1780 This is a one-shot self test that permutes through the 1781 recovery of all the possible two disk failure scenarios for a 1782 N-disk array. Recovery is performed with the asynchronous 1783 raid6 recovery routines, and will optionally use an offload 1784 engine if one is available. 1785 1786 If unsure, say N. 1787 1788config TEST_HEXDUMP 1789 tristate "Test functions located in the hexdump module at runtime" 1790 1791config TEST_STRING_HELPERS 1792 tristate "Test functions located in the string_helpers module at runtime" 1793 1794config TEST_KSTRTOX 1795 tristate "Test kstrto*() family of functions at runtime" 1796 1797config TEST_PRINTF 1798 tristate "Test printf() family of functions at runtime" 1799 1800config TEST_BITMAP 1801 tristate "Test bitmap_*() family of functions at runtime" 1802 default n 1803 help 1804 Enable this option to test the bitmap functions at boot. 1805 1806 If unsure, say N. 1807 1808config TEST_UUID 1809 tristate "Test functions located in the uuid module at runtime" 1810 1811config TEST_OVERFLOW 1812 tristate "Test check_*_overflow() functions at runtime" 1813 1814config TEST_RHASHTABLE 1815 tristate "Perform selftest on resizable hash table" 1816 default n 1817 help 1818 Enable this option to test the rhashtable functions at boot. 1819 1820 If unsure, say N. 1821 1822config TEST_HASH 1823 tristate "Perform selftest on hash functions" 1824 default n 1825 help 1826 Enable this option to test the kernel's integer (<linux/hash.h>), 1827 string (<linux/stringhash.h>), and siphash (<linux/siphash.h>) 1828 hash functions on boot (or module load). 1829 1830 This is intended to help people writing architecture-specific 1831 optimized versions. If unsure, say N. 1832 1833config TEST_PARMAN 1834 tristate "Perform selftest on priority array manager" 1835 default n 1836 depends on PARMAN 1837 help 1838 Enable this option to test priority array manager on boot 1839 (or module load). 1840 1841 If unsure, say N. 1842 1843config TEST_LKM 1844 tristate "Test module loading with 'hello world' module" 1845 default n 1846 depends on m 1847 help 1848 This builds the "test_module" module that emits "Hello, world" 1849 on printk when loaded. It is designed to be used for basic 1850 evaluation of the module loading subsystem (for example when 1851 validating module verification). It lacks any extra dependencies, 1852 and will not normally be loaded by the system unless explicitly 1853 requested by name. 1854 1855 If unsure, say N. 1856 1857config TEST_USER_COPY 1858 tristate "Test user/kernel boundary protections" 1859 default n 1860 depends on m 1861 help 1862 This builds the "test_user_copy" module that runs sanity checks 1863 on the copy_to/from_user infrastructure, making sure basic 1864 user/kernel boundary testing is working. If it fails to load, 1865 a regression has been detected in the user/kernel memory boundary 1866 protections. 1867 1868 If unsure, say N. 1869 1870config TEST_BPF 1871 tristate "Test BPF filter functionality" 1872 default n 1873 depends on m && NET 1874 help 1875 This builds the "test_bpf" module that runs various test vectors 1876 against the BPF interpreter or BPF JIT compiler depending on the 1877 current setting. This is in particular useful for BPF JIT compiler 1878 development, but also to run regression tests against changes in 1879 the interpreter code. It also enables test stubs for eBPF maps and 1880 verifier used by user space verifier testsuite. 1881 1882 If unsure, say N. 1883 1884config FIND_BIT_BENCHMARK 1885 tristate "Test find_bit functions" 1886 default n 1887 help 1888 This builds the "test_find_bit" module that measure find_*_bit() 1889 functions performance. 1890 1891 If unsure, say N. 1892 1893config TEST_FIRMWARE 1894 tristate "Test firmware loading via userspace interface" 1895 default n 1896 depends on FW_LOADER 1897 help 1898 This builds the "test_firmware" module that creates a userspace 1899 interface for testing firmware loading. This can be used to 1900 control the triggering of firmware loading without needing an 1901 actual firmware-using device. The contents can be rechecked by 1902 userspace. 1903 1904 If unsure, say N. 1905 1906config TEST_SYSCTL 1907 tristate "sysctl test driver" 1908 default n 1909 depends on PROC_SYSCTL 1910 help 1911 This builds the "test_sysctl" module. This driver enables to test the 1912 proc sysctl interfaces available to drivers safely without affecting 1913 production knobs which might alter system functionality. 1914 1915 If unsure, say N. 1916 1917config TEST_UDELAY 1918 tristate "udelay test driver" 1919 default n 1920 help 1921 This builds the "udelay_test" module that helps to make sure 1922 that udelay() is working properly. 1923 1924 If unsure, say N. 1925 1926config TEST_STATIC_KEYS 1927 tristate "Test static keys" 1928 default n 1929 depends on m 1930 help 1931 Test the static key interfaces. 1932 1933 If unsure, say N. 1934 1935config TEST_KMOD 1936 tristate "kmod stress tester" 1937 default n 1938 depends on m 1939 depends on BLOCK && (64BIT || LBDAF) # for XFS, BTRFS 1940 depends on NETDEVICES && NET_CORE && INET # for TUN 1941 select TEST_LKM 1942 select XFS_FS 1943 select TUN 1944 select BTRFS_FS 1945 help 1946 Test the kernel's module loading mechanism: kmod. kmod implements 1947 support to load modules using the Linux kernel's usermode helper. 1948 This test provides a series of tests against kmod. 1949 1950 Although technically you can either build test_kmod as a module or 1951 into the kernel we disallow building it into the kernel since 1952 it stress tests request_module() and this will very likely cause 1953 some issues by taking over precious threads available from other 1954 module load requests, ultimately this could be fatal. 1955 1956 To run tests run: 1957 1958 tools/testing/selftests/kmod/kmod.sh --help 1959 1960 If unsure, say N. 1961 1962config TEST_DEBUG_VIRTUAL 1963 tristate "Test CONFIG_DEBUG_VIRTUAL feature" 1964 depends on DEBUG_VIRTUAL 1965 help 1966 Test the kernel's ability to detect incorrect calls to 1967 virt_to_phys() done against the non-linear part of the 1968 kernel's virtual address map. 1969 1970 If unsure, say N. 1971 1972endif # RUNTIME_TESTING_MENU 1973 1974config MEMTEST 1975 bool "Memtest" 1976 depends on HAVE_MEMBLOCK 1977 ---help--- 1978 This option adds a kernel parameter 'memtest', which allows memtest 1979 to be set. 1980 memtest=0, mean disabled; -- default 1981 memtest=1, mean do 1 test pattern; 1982 ... 1983 memtest=17, mean do 17 test patterns. 1984 If you are unsure how to answer this question, answer N. 1985 1986config BUG_ON_DATA_CORRUPTION 1987 bool "Trigger a BUG when data corruption is detected" 1988 select DEBUG_LIST 1989 help 1990 Select this option if the kernel should BUG when it encounters 1991 data corruption in kernel memory structures when they get checked 1992 for validity. 1993 1994 If unsure, say N. 1995 1996source "samples/Kconfig" 1997 1998source "lib/Kconfig.kgdb" 1999 2000source "lib/Kconfig.ubsan" 2001 2002config ARCH_HAS_DEVMEM_IS_ALLOWED 2003 bool 2004 2005config STRICT_DEVMEM 2006 bool "Filter access to /dev/mem" 2007 depends on MMU && DEVMEM 2008 depends on ARCH_HAS_DEVMEM_IS_ALLOWED 2009 default y if PPC || X86 || ARM64 2010 ---help--- 2011 If this option is disabled, you allow userspace (root) access to all 2012 of memory, including kernel and userspace memory. Accidental 2013 access to this is obviously disastrous, but specific access can 2014 be used by people debugging the kernel. Note that with PAT support 2015 enabled, even in this case there are restrictions on /dev/mem 2016 use due to the cache aliasing requirements. 2017 2018 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem 2019 file only allows userspace access to PCI space and the BIOS code and 2020 data regions. This is sufficient for dosemu and X and all common 2021 users of /dev/mem. 2022 2023 If in doubt, say Y. 2024 2025config IO_STRICT_DEVMEM 2026 bool "Filter I/O access to /dev/mem" 2027 depends on STRICT_DEVMEM 2028 ---help--- 2029 If this option is disabled, you allow userspace (root) access to all 2030 io-memory regardless of whether a driver is actively using that 2031 range. Accidental access to this is obviously disastrous, but 2032 specific access can be used by people debugging kernel drivers. 2033 2034 If this option is switched on, the /dev/mem file only allows 2035 userspace access to *idle* io-memory ranges (see /proc/iomem) This 2036 may break traditional users of /dev/mem (dosemu, legacy X, etc...) 2037 if the driver using a given range cannot be disabled. 2038 2039 If in doubt, say Y. 2040 2041source "arch/$(SRCARCH)/Kconfig.debug" 2042 2043endmenu # Kernel hacking 2044