1# SPDX-License-Identifier: GPL-2.0-only 2menu "Kernel hacking" 3 4menu "printk and dmesg options" 5 6config PRINTK_TIME 7 bool "Show timing information on printks" 8 depends on PRINTK 9 help 10 Selecting this option causes time stamps of the printk() 11 messages to be added to the output of the syslog() system 12 call and at the console. 13 14 The timestamp is always recorded internally, and exported 15 to /dev/kmsg. This flag just specifies if the timestamp should 16 be included, not that the timestamp is recorded. 17 18 The behavior is also controlled by the kernel command line 19 parameter printk.time=1. See Documentation/admin-guide/kernel-parameters.rst 20 21config PRINTK_CALLER 22 bool "Show caller information on printks" 23 depends on PRINTK 24 help 25 Selecting this option causes printk() to add a caller "thread id" (if 26 in task context) or a caller "processor id" (if not in task context) 27 to every message. 28 29 This option is intended for environments where multiple threads 30 concurrently call printk() for many times, for it is difficult to 31 interpret without knowing where these lines (or sometimes individual 32 line which was divided into multiple lines due to race) came from. 33 34 Since toggling after boot makes the code racy, currently there is 35 no option to enable/disable at the kernel command line parameter or 36 sysfs interface. 37 38config CONSOLE_LOGLEVEL_DEFAULT 39 int "Default console loglevel (1-15)" 40 range 1 15 41 default "7" 42 help 43 Default loglevel to determine what will be printed on the console. 44 45 Setting a default here is equivalent to passing in loglevel=<x> in 46 the kernel bootargs. loglevel=<x> continues to override whatever 47 value is specified here as well. 48 49 Note: This does not affect the log level of un-prefixed printk() 50 usage in the kernel. That is controlled by the MESSAGE_LOGLEVEL_DEFAULT 51 option. 52 53config CONSOLE_LOGLEVEL_QUIET 54 int "quiet console loglevel (1-15)" 55 range 1 15 56 default "4" 57 help 58 loglevel to use when "quiet" is passed on the kernel commandline. 59 60 When "quiet" is passed on the kernel commandline this loglevel 61 will be used as the loglevel. IOW passing "quiet" will be the 62 equivalent of passing "loglevel=<CONSOLE_LOGLEVEL_QUIET>" 63 64config MESSAGE_LOGLEVEL_DEFAULT 65 int "Default message log level (1-7)" 66 range 1 7 67 default "4" 68 help 69 Default log level for printk statements with no specified priority. 70 71 This was hard-coded to KERN_WARNING since at least 2.6.10 but folks 72 that are auditing their logs closely may want to set it to a lower 73 priority. 74 75 Note: This does not affect what message level gets printed on the console 76 by default. To change that, use loglevel=<x> in the kernel bootargs, 77 or pick a different CONSOLE_LOGLEVEL_DEFAULT configuration value. 78 79config BOOT_PRINTK_DELAY 80 bool "Delay each boot printk message by N milliseconds" 81 depends on DEBUG_KERNEL && PRINTK && GENERIC_CALIBRATE_DELAY 82 help 83 This build option allows you to read kernel boot messages 84 by inserting a short delay after each one. The delay is 85 specified in milliseconds on the kernel command line, 86 using "boot_delay=N". 87 88 It is likely that you would also need to use "lpj=M" to preset 89 the "loops per jiffie" value. 90 See a previous boot log for the "lpj" value to use for your 91 system, and then set "lpj=M" before setting "boot_delay=N". 92 NOTE: Using this option may adversely affect SMP systems. 93 I.e., processors other than the first one may not boot up. 94 BOOT_PRINTK_DELAY also may cause LOCKUP_DETECTOR to detect 95 what it believes to be lockup conditions. 96 97config DYNAMIC_DEBUG 98 bool "Enable dynamic printk() support" 99 default n 100 depends on PRINTK 101 depends on (DEBUG_FS || PROC_FS) 102 select DYNAMIC_DEBUG_CORE 103 help 104 105 Compiles debug level messages into the kernel, which would not 106 otherwise be available at runtime. These messages can then be 107 enabled/disabled based on various levels of scope - per source file, 108 function, module, format string, and line number. This mechanism 109 implicitly compiles in all pr_debug() and dev_dbg() calls, which 110 enlarges the kernel text size by about 2%. 111 112 If a source file is compiled with DEBUG flag set, any 113 pr_debug() calls in it are enabled by default, but can be 114 disabled at runtime as below. Note that DEBUG flag is 115 turned on by many CONFIG_*DEBUG* options. 116 117 Usage: 118 119 Dynamic debugging is controlled via the 'dynamic_debug/control' file, 120 which is contained in the 'debugfs' filesystem or procfs. 121 Thus, the debugfs or procfs filesystem must first be mounted before 122 making use of this feature. 123 We refer the control file as: <debugfs>/dynamic_debug/control. This 124 file contains a list of the debug statements that can be enabled. The 125 format for each line of the file is: 126 127 filename:lineno [module]function flags format 128 129 filename : source file of the debug statement 130 lineno : line number of the debug statement 131 module : module that contains the debug statement 132 function : function that contains the debug statement 133 flags : '=p' means the line is turned 'on' for printing 134 format : the format used for the debug statement 135 136 From a live system: 137 138 nullarbor:~ # cat <debugfs>/dynamic_debug/control 139 # filename:lineno [module]function flags format 140 fs/aio.c:222 [aio]__put_ioctx =_ "__put_ioctx:\040freeing\040%p\012" 141 fs/aio.c:248 [aio]ioctx_alloc =_ "ENOMEM:\040nr_events\040too\040high\012" 142 fs/aio.c:1770 [aio]sys_io_cancel =_ "calling\040cancel\012" 143 144 Example usage: 145 146 // enable the message at line 1603 of file svcsock.c 147 nullarbor:~ # echo -n 'file svcsock.c line 1603 +p' > 148 <debugfs>/dynamic_debug/control 149 150 // enable all the messages in file svcsock.c 151 nullarbor:~ # echo -n 'file svcsock.c +p' > 152 <debugfs>/dynamic_debug/control 153 154 // enable all the messages in the NFS server module 155 nullarbor:~ # echo -n 'module nfsd +p' > 156 <debugfs>/dynamic_debug/control 157 158 // enable all 12 messages in the function svc_process() 159 nullarbor:~ # echo -n 'func svc_process +p' > 160 <debugfs>/dynamic_debug/control 161 162 // disable all 12 messages in the function svc_process() 163 nullarbor:~ # echo -n 'func svc_process -p' > 164 <debugfs>/dynamic_debug/control 165 166 See Documentation/admin-guide/dynamic-debug-howto.rst for additional 167 information. 168 169config DYNAMIC_DEBUG_CORE 170 bool "Enable core function of dynamic debug support" 171 depends on PRINTK 172 depends on (DEBUG_FS || PROC_FS) 173 help 174 Enable core functional support of dynamic debug. It is useful 175 when you want to tie dynamic debug to your kernel modules with 176 DYNAMIC_DEBUG_MODULE defined for each of them, especially for 177 the case of embedded system where the kernel image size is 178 sensitive for people. 179 180config SYMBOLIC_ERRNAME 181 bool "Support symbolic error names in printf" 182 default y if PRINTK 183 help 184 If you say Y here, the kernel's printf implementation will 185 be able to print symbolic error names such as ENOSPC instead 186 of the number 28. It makes the kernel image slightly larger 187 (about 3KB), but can make the kernel logs easier to read. 188 189config DEBUG_BUGVERBOSE 190 bool "Verbose BUG() reporting (adds 70K)" if DEBUG_KERNEL && EXPERT 191 depends on BUG && (GENERIC_BUG || HAVE_DEBUG_BUGVERBOSE) 192 default y 193 help 194 Say Y here to make BUG() panics output the file name and line number 195 of the BUG call as well as the EIP and oops trace. This aids 196 debugging but costs about 70-100K of memory. 197 198endmenu # "printk and dmesg options" 199 200menu "Compile-time checks and compiler options" 201 202config DEBUG_INFO 203 bool "Compile the kernel with debug info" 204 depends on DEBUG_KERNEL && !COMPILE_TEST 205 help 206 If you say Y here the resulting kernel image will include 207 debugging info resulting in a larger kernel image. 208 This adds debug symbols to the kernel and modules (gcc -g), and 209 is needed if you intend to use kernel crashdump or binary object 210 tools like crash, kgdb, LKCD, gdb, etc on the kernel. 211 Say Y here only if you plan to debug the kernel. 212 213 If unsure, say N. 214 215if DEBUG_INFO 216 217config DEBUG_INFO_REDUCED 218 bool "Reduce debugging information" 219 help 220 If you say Y here gcc is instructed to generate less debugging 221 information for structure types. This means that tools that 222 need full debugging information (like kgdb or systemtap) won't 223 be happy. But if you merely need debugging information to 224 resolve line numbers there is no loss. Advantage is that 225 build directory object sizes shrink dramatically over a full 226 DEBUG_INFO build and compile times are reduced too. 227 Only works with newer gcc versions. 228 229config DEBUG_INFO_COMPRESSED 230 bool "Compressed debugging information" 231 depends on $(cc-option,-gz=zlib) 232 depends on $(ld-option,--compress-debug-sections=zlib) 233 help 234 Compress the debug information using zlib. Requires GCC 5.0+ or Clang 235 5.0+, binutils 2.26+, and zlib. 236 237 Users of dpkg-deb via scripts/package/builddeb may find an increase in 238 size of their debug .deb packages with this config set, due to the 239 debug info being compressed with zlib, then the object files being 240 recompressed with a different compression scheme. But this is still 241 preferable to setting $KDEB_COMPRESS to "none" which would be even 242 larger. 243 244config DEBUG_INFO_SPLIT 245 bool "Produce split debuginfo in .dwo files" 246 depends on $(cc-option,-gsplit-dwarf) 247 help 248 Generate debug info into separate .dwo files. This significantly 249 reduces the build directory size for builds with DEBUG_INFO, 250 because it stores the information only once on disk in .dwo 251 files instead of multiple times in object files and executables. 252 In addition the debug information is also compressed. 253 254 Requires recent gcc (4.7+) and recent gdb/binutils. 255 Any tool that packages or reads debug information would need 256 to know about the .dwo files and include them. 257 Incompatible with older versions of ccache. 258 259choice 260 prompt "DWARF version" 261 help 262 Which version of DWARF debug info to emit. 263 264config DEBUG_INFO_DWARF_TOOLCHAIN_DEFAULT 265 bool "Rely on the toolchain's implicit default DWARF version" 266 help 267 The implicit default version of DWARF debug info produced by a 268 toolchain changes over time. 269 270 This can break consumers of the debug info that haven't upgraded to 271 support newer revisions, and prevent testing newer versions, but 272 those should be less common scenarios. 273 274 If unsure, say Y. 275 276config DEBUG_INFO_DWARF4 277 bool "Generate DWARF Version 4 debuginfo" 278 help 279 Generate DWARF v4 debug info. This requires gcc 4.5+ and gdb 7.0+. 280 281 If you have consumers of DWARF debug info that are not ready for 282 newer revisions of DWARF, you may wish to choose this or have your 283 config select this. 284 285config DEBUG_INFO_DWARF5 286 bool "Generate DWARF Version 5 debuginfo" 287 depends on GCC_VERSION >= 50000 || CC_IS_CLANG 288 depends on CC_IS_GCC || $(success,$(srctree)/scripts/test_dwarf5_support.sh $(CC) $(CLANG_FLAGS)) 289 depends on !DEBUG_INFO_BTF 290 help 291 Generate DWARF v5 debug info. Requires binutils 2.35.2, gcc 5.0+ (gcc 292 5.0+ accepts the -gdwarf-5 flag but only had partial support for some 293 draft features until 7.0), and gdb 8.0+. 294 295 Changes to the structure of debug info in Version 5 allow for around 296 15-18% savings in resulting image and debug info section sizes as 297 compared to DWARF Version 4. DWARF Version 5 standardizes previous 298 extensions such as accelerators for symbol indexing and the format 299 for fission (.dwo/.dwp) files. Users may not want to select this 300 config if they rely on tooling that has not yet been updated to 301 support DWARF Version 5. 302 303endchoice # "DWARF version" 304 305config DEBUG_INFO_BTF 306 bool "Generate BTF typeinfo" 307 depends on !DEBUG_INFO_SPLIT && !DEBUG_INFO_REDUCED 308 depends on !GCC_PLUGIN_RANDSTRUCT || COMPILE_TEST 309 help 310 Generate deduplicated BTF type information from DWARF debug info. 311 Turning this on expects presence of pahole tool, which will convert 312 DWARF type info into equivalent deduplicated BTF type info. 313 314config PAHOLE_HAS_SPLIT_BTF 315 def_bool $(success, test `$(PAHOLE) --version | sed -E 's/v([0-9]+)\.([0-9]+)/\1\2/'` -ge "119") 316 317config DEBUG_INFO_BTF_MODULES 318 def_bool y 319 depends on DEBUG_INFO_BTF && MODULES && PAHOLE_HAS_SPLIT_BTF 320 help 321 Generate compact split BTF type information for kernel modules. 322 323config GDB_SCRIPTS 324 bool "Provide GDB scripts for kernel debugging" 325 help 326 This creates the required links to GDB helper scripts in the 327 build directory. If you load vmlinux into gdb, the helper 328 scripts will be automatically imported by gdb as well, and 329 additional functions are available to analyze a Linux kernel 330 instance. See Documentation/dev-tools/gdb-kernel-debugging.rst 331 for further details. 332 333endif # DEBUG_INFO 334 335config FRAME_WARN 336 int "Warn for stack frames larger than" 337 range 0 8192 338 default 2048 if GCC_PLUGIN_LATENT_ENTROPY 339 default 1280 if (!64BIT && PARISC) 340 default 1024 if (!64BIT && !PARISC) 341 default 2048 if 64BIT 342 help 343 Tell gcc to warn at build time for stack frames larger than this. 344 Setting this too low will cause a lot of warnings. 345 Setting it to 0 disables the warning. 346 347config STRIP_ASM_SYMS 348 bool "Strip assembler-generated symbols during link" 349 default n 350 help 351 Strip internal assembler-generated symbols during a link (symbols 352 that look like '.Lxxx') so they don't pollute the output of 353 get_wchan() and suchlike. 354 355config READABLE_ASM 356 bool "Generate readable assembler code" 357 depends on DEBUG_KERNEL 358 help 359 Disable some compiler optimizations that tend to generate human unreadable 360 assembler output. This may make the kernel slightly slower, but it helps 361 to keep kernel developers who have to stare a lot at assembler listings 362 sane. 363 364config HEADERS_INSTALL 365 bool "Install uapi headers to usr/include" 366 depends on !UML 367 help 368 This option will install uapi headers (headers exported to user-space) 369 into the usr/include directory for use during the kernel build. 370 This is unneeded for building the kernel itself, but needed for some 371 user-space program samples. It is also needed by some features such 372 as uapi header sanity checks. 373 374config DEBUG_SECTION_MISMATCH 375 bool "Enable full Section mismatch analysis" 376 help 377 The section mismatch analysis checks if there are illegal 378 references from one section to another section. 379 During linktime or runtime, some sections are dropped; 380 any use of code/data previously in these sections would 381 most likely result in an oops. 382 In the code, functions and variables are annotated with 383 __init,, etc. (see the full list in include/linux/init.h), 384 which results in the code/data being placed in specific sections. 385 The section mismatch analysis is always performed after a full 386 kernel build, and enabling this option causes the following 387 additional step to occur: 388 - Add the option -fno-inline-functions-called-once to gcc commands. 389 When inlining a function annotated with __init in a non-init 390 function, we would lose the section information and thus 391 the analysis would not catch the illegal reference. 392 This option tells gcc to inline less (but it does result in 393 a larger kernel). 394 395config SECTION_MISMATCH_WARN_ONLY 396 bool "Make section mismatch errors non-fatal" 397 default y 398 help 399 If you say N here, the build process will fail if there are any 400 section mismatch, instead of just throwing warnings. 401 402 If unsure, say Y. 403 404config DEBUG_FORCE_FUNCTION_ALIGN_32B 405 bool "Force all function address 32B aligned" if EXPERT 406 help 407 There are cases that a commit from one domain changes the function 408 address alignment of other domains, and cause magic performance 409 bump (regression or improvement). Enable this option will help to 410 verify if the bump is caused by function alignment changes, while 411 it will slightly increase the kernel size and affect icache usage. 412 413 It is mainly for debug and performance tuning use. 414 415# 416# Select this config option from the architecture Kconfig, if it 417# is preferred to always offer frame pointers as a config 418# option on the architecture (regardless of KERNEL_DEBUG): 419# 420config ARCH_WANT_FRAME_POINTERS 421 bool 422 423config FRAME_POINTER 424 bool "Compile the kernel with frame pointers" 425 depends on DEBUG_KERNEL && (M68K || UML || SUPERH) || ARCH_WANT_FRAME_POINTERS 426 default y if (DEBUG_INFO && UML) || ARCH_WANT_FRAME_POINTERS 427 help 428 If you say Y here the resulting kernel image will be slightly 429 larger and slower, but it gives very useful debugging information 430 in case of kernel bugs. (precise oopses/stacktraces/warnings) 431 432config STACK_VALIDATION 433 bool "Compile-time stack metadata validation" 434 depends on HAVE_STACK_VALIDATION 435 default n 436 help 437 Add compile-time checks to validate stack metadata, including frame 438 pointers (if CONFIG_FRAME_POINTER is enabled). This helps ensure 439 that runtime stack traces are more reliable. 440 441 This is also a prerequisite for generation of ORC unwind data, which 442 is needed for CONFIG_UNWINDER_ORC. 443 444 For more information, see 445 tools/objtool/Documentation/stack-validation.txt. 446 447config VMLINUX_VALIDATION 448 bool 449 depends on STACK_VALIDATION && DEBUG_ENTRY && !PARAVIRT 450 default y 451 452config DEBUG_FORCE_WEAK_PER_CPU 453 bool "Force weak per-cpu definitions" 454 depends on DEBUG_KERNEL 455 help 456 s390 and alpha require percpu variables in modules to be 457 defined weak to work around addressing range issue which 458 puts the following two restrictions on percpu variable 459 definitions. 460 461 1. percpu symbols must be unique whether static or not 462 2. percpu variables can't be defined inside a function 463 464 To ensure that generic code follows the above rules, this 465 option forces all percpu variables to be defined as weak. 466 467endmenu # "Compiler options" 468 469menu "Generic Kernel Debugging Instruments" 470 471config MAGIC_SYSRQ 472 bool "Magic SysRq key" 473 depends on !UML 474 help 475 If you say Y here, you will have some control over the system even 476 if the system crashes for example during kernel debugging (e.g., you 477 will be able to flush the buffer cache to disk, reboot the system 478 immediately or dump some status information). This is accomplished 479 by pressing various keys while holding SysRq (Alt+PrintScreen). It 480 also works on a serial console (on PC hardware at least), if you 481 send a BREAK and then within 5 seconds a command keypress. The 482 keys are documented in <file:Documentation/admin-guide/sysrq.rst>. 483 Don't say Y unless you really know what this hack does. 484 485config MAGIC_SYSRQ_DEFAULT_ENABLE 486 hex "Enable magic SysRq key functions by default" 487 depends on MAGIC_SYSRQ 488 default 0x1 489 help 490 Specifies which SysRq key functions are enabled by default. 491 This may be set to 1 or 0 to enable or disable them all, or 492 to a bitmask as described in Documentation/admin-guide/sysrq.rst. 493 494config MAGIC_SYSRQ_SERIAL 495 bool "Enable magic SysRq key over serial" 496 depends on MAGIC_SYSRQ 497 default y 498 help 499 Many embedded boards have a disconnected TTL level serial which can 500 generate some garbage that can lead to spurious false sysrq detects. 501 This option allows you to decide whether you want to enable the 502 magic SysRq key. 503 504config MAGIC_SYSRQ_SERIAL_SEQUENCE 505 string "Char sequence that enables magic SysRq over serial" 506 depends on MAGIC_SYSRQ_SERIAL 507 default "" 508 help 509 Specifies a sequence of characters that can follow BREAK to enable 510 SysRq on a serial console. 511 512 If unsure, leave an empty string and the option will not be enabled. 513 514config DEBUG_FS 515 bool "Debug Filesystem" 516 help 517 debugfs is a virtual file system that kernel developers use to put 518 debugging files into. Enable this option to be able to read and 519 write to these files. 520 521 For detailed documentation on the debugfs API, see 522 Documentation/filesystems/. 523 524 If unsure, say N. 525 526choice 527 prompt "Debugfs default access" 528 depends on DEBUG_FS 529 default DEBUG_FS_ALLOW_ALL 530 help 531 This selects the default access restrictions for debugfs. 532 It can be overridden with kernel command line option 533 debugfs=[on,no-mount,off]. The restrictions apply for API access 534 and filesystem registration. 535 536config DEBUG_FS_ALLOW_ALL 537 bool "Access normal" 538 help 539 No restrictions apply. Both API and filesystem registration 540 is on. This is the normal default operation. 541 542config DEBUG_FS_DISALLOW_MOUNT 543 bool "Do not register debugfs as filesystem" 544 help 545 The API is open but filesystem is not loaded. Clients can still do 546 their work and read with debug tools that do not need 547 debugfs filesystem. 548 549config DEBUG_FS_ALLOW_NONE 550 bool "No access" 551 help 552 Access is off. Clients get -PERM when trying to create nodes in 553 debugfs tree and debugfs is not registered as a filesystem. 554 Client can then back-off or continue without debugfs access. 555 556endchoice 557 558source "lib/Kconfig.kgdb" 559source "lib/Kconfig.ubsan" 560source "lib/Kconfig.kcsan" 561 562endmenu 563 564config DEBUG_KERNEL 565 bool "Kernel debugging" 566 help 567 Say Y here if you are developing drivers or trying to debug and 568 identify kernel problems. 569 570config DEBUG_MISC 571 bool "Miscellaneous debug code" 572 default DEBUG_KERNEL 573 depends on DEBUG_KERNEL 574 help 575 Say Y here if you need to enable miscellaneous debug code that should 576 be under a more specific debug option but isn't. 577 578 579menu "Memory Debugging" 580 581source "mm/Kconfig.debug" 582 583config DEBUG_OBJECTS 584 bool "Debug object operations" 585 depends on DEBUG_KERNEL 586 help 587 If you say Y here, additional code will be inserted into the 588 kernel to track the life time of various objects and validate 589 the operations on those objects. 590 591config DEBUG_OBJECTS_SELFTEST 592 bool "Debug objects selftest" 593 depends on DEBUG_OBJECTS 594 help 595 This enables the selftest of the object debug code. 596 597config DEBUG_OBJECTS_FREE 598 bool "Debug objects in freed memory" 599 depends on DEBUG_OBJECTS 600 help 601 This enables checks whether a k/v free operation frees an area 602 which contains an object which has not been deactivated 603 properly. This can make kmalloc/kfree-intensive workloads 604 much slower. 605 606config DEBUG_OBJECTS_TIMERS 607 bool "Debug timer objects" 608 depends on DEBUG_OBJECTS 609 help 610 If you say Y here, additional code will be inserted into the 611 timer routines to track the life time of timer objects and 612 validate the timer operations. 613 614config DEBUG_OBJECTS_WORK 615 bool "Debug work objects" 616 depends on DEBUG_OBJECTS 617 help 618 If you say Y here, additional code will be inserted into the 619 work queue routines to track the life time of work objects and 620 validate the work operations. 621 622config DEBUG_OBJECTS_RCU_HEAD 623 bool "Debug RCU callbacks objects" 624 depends on DEBUG_OBJECTS 625 help 626 Enable this to turn on debugging of RCU list heads (call_rcu() usage). 627 628config DEBUG_OBJECTS_PERCPU_COUNTER 629 bool "Debug percpu counter objects" 630 depends on DEBUG_OBJECTS 631 help 632 If you say Y here, additional code will be inserted into the 633 percpu counter routines to track the life time of percpu counter 634 objects and validate the percpu counter operations. 635 636config DEBUG_OBJECTS_ENABLE_DEFAULT 637 int "debug_objects bootup default value (0-1)" 638 range 0 1 639 default "1" 640 depends on DEBUG_OBJECTS 641 help 642 Debug objects boot parameter default value 643 644config DEBUG_SLAB 645 bool "Debug slab memory allocations" 646 depends on DEBUG_KERNEL && SLAB 647 help 648 Say Y here to have the kernel do limited verification on memory 649 allocation as well as poisoning memory on free to catch use of freed 650 memory. This can make kmalloc/kfree-intensive workloads much slower. 651 652config SLUB_DEBUG_ON 653 bool "SLUB debugging on by default" 654 depends on SLUB && SLUB_DEBUG 655 default n 656 help 657 Boot with debugging on by default. SLUB boots by default with 658 the runtime debug capabilities switched off. Enabling this is 659 equivalent to specifying the "slub_debug" parameter on boot. 660 There is no support for more fine grained debug control like 661 possible with slub_debug=xxx. SLUB debugging may be switched 662 off in a kernel built with CONFIG_SLUB_DEBUG_ON by specifying 663 "slub_debug=-". 664 665config SLUB_STATS 666 default n 667 bool "Enable SLUB performance statistics" 668 depends on SLUB && SYSFS 669 help 670 SLUB statistics are useful to debug SLUBs allocation behavior in 671 order find ways to optimize the allocator. This should never be 672 enabled for production use since keeping statistics slows down 673 the allocator by a few percentage points. The slabinfo command 674 supports the determination of the most active slabs to figure 675 out which slabs are relevant to a particular load. 676 Try running: slabinfo -DA 677 678config HAVE_DEBUG_KMEMLEAK 679 bool 680 681config DEBUG_KMEMLEAK 682 bool "Kernel memory leak detector" 683 depends on DEBUG_KERNEL && HAVE_DEBUG_KMEMLEAK 684 select DEBUG_FS 685 select STACKTRACE if STACKTRACE_SUPPORT 686 select KALLSYMS 687 select CRC32 688 help 689 Say Y here if you want to enable the memory leak 690 detector. The memory allocation/freeing is traced in a way 691 similar to the Boehm's conservative garbage collector, the 692 difference being that the orphan objects are not freed but 693 only shown in /sys/kernel/debug/kmemleak. Enabling this 694 feature will introduce an overhead to memory 695 allocations. See Documentation/dev-tools/kmemleak.rst for more 696 details. 697 698 Enabling DEBUG_SLAB or SLUB_DEBUG may increase the chances 699 of finding leaks due to the slab objects poisoning. 700 701 In order to access the kmemleak file, debugfs needs to be 702 mounted (usually at /sys/kernel/debug). 703 704config DEBUG_KMEMLEAK_MEM_POOL_SIZE 705 int "Kmemleak memory pool size" 706 depends on DEBUG_KMEMLEAK 707 range 200 1000000 708 default 16000 709 help 710 Kmemleak must track all the memory allocations to avoid 711 reporting false positives. Since memory may be allocated or 712 freed before kmemleak is fully initialised, use a static pool 713 of metadata objects to track such callbacks. After kmemleak is 714 fully initialised, this memory pool acts as an emergency one 715 if slab allocations fail. 716 717config DEBUG_KMEMLEAK_TEST 718 tristate "Simple test for the kernel memory leak detector" 719 depends on DEBUG_KMEMLEAK && m 720 help 721 This option enables a module that explicitly leaks memory. 722 723 If unsure, say N. 724 725config DEBUG_KMEMLEAK_DEFAULT_OFF 726 bool "Default kmemleak to off" 727 depends on DEBUG_KMEMLEAK 728 help 729 Say Y here to disable kmemleak by default. It can then be enabled 730 on the command line via kmemleak=on. 731 732config DEBUG_KMEMLEAK_AUTO_SCAN 733 bool "Enable kmemleak auto scan thread on boot up" 734 default y 735 depends on DEBUG_KMEMLEAK 736 help 737 Depending on the cpu, kmemleak scan may be cpu intensive and can 738 stall user tasks at times. This option enables/disables automatic 739 kmemleak scan at boot up. 740 741 Say N here to disable kmemleak auto scan thread to stop automatic 742 scanning. Disabling this option disables automatic reporting of 743 memory leaks. 744 745 If unsure, say Y. 746 747config DEBUG_STACK_USAGE 748 bool "Stack utilization instrumentation" 749 depends on DEBUG_KERNEL && !IA64 750 help 751 Enables the display of the minimum amount of free stack which each 752 task has ever had available in the sysrq-T and sysrq-P debug output. 753 754 This option will slow down process creation somewhat. 755 756config SCHED_STACK_END_CHECK 757 bool "Detect stack corruption on calls to schedule()" 758 depends on DEBUG_KERNEL 759 default n 760 help 761 This option checks for a stack overrun on calls to schedule(). 762 If the stack end location is found to be over written always panic as 763 the content of the corrupted region can no longer be trusted. 764 This is to ensure no erroneous behaviour occurs which could result in 765 data corruption or a sporadic crash at a later stage once the region 766 is examined. The runtime overhead introduced is minimal. 767 768config ARCH_HAS_DEBUG_VM_PGTABLE 769 bool 770 help 771 An architecture should select this when it can successfully 772 build and run DEBUG_VM_PGTABLE. 773 774config DEBUG_VM 775 bool "Debug VM" 776 depends on DEBUG_KERNEL 777 help 778 Enable this to turn on extended checks in the virtual-memory system 779 that may impact performance. 780 781 If unsure, say N. 782 783config DEBUG_VM_VMACACHE 784 bool "Debug VMA caching" 785 depends on DEBUG_VM 786 help 787 Enable this to turn on VMA caching debug information. Doing so 788 can cause significant overhead, so only enable it in non-production 789 environments. 790 791 If unsure, say N. 792 793config DEBUG_VM_RB 794 bool "Debug VM red-black trees" 795 depends on DEBUG_VM 796 help 797 Enable VM red-black tree debugging information and extra validations. 798 799 If unsure, say N. 800 801config DEBUG_VM_PGFLAGS 802 bool "Debug page-flags operations" 803 depends on DEBUG_VM 804 help 805 Enables extra validation on page flags operations. 806 807 If unsure, say N. 808 809config DEBUG_VM_PGTABLE 810 bool "Debug arch page table for semantics compliance" 811 depends on MMU 812 depends on ARCH_HAS_DEBUG_VM_PGTABLE 813 default y if DEBUG_VM 814 help 815 This option provides a debug method which can be used to test 816 architecture page table helper functions on various platforms in 817 verifying if they comply with expected generic MM semantics. This 818 will help architecture code in making sure that any changes or 819 new additions of these helpers still conform to expected 820 semantics of the generic MM. Platforms will have to opt in for 821 this through ARCH_HAS_DEBUG_VM_PGTABLE. 822 823 If unsure, say N. 824 825config ARCH_HAS_DEBUG_VIRTUAL 826 bool 827 828config DEBUG_VIRTUAL 829 bool "Debug VM translations" 830 depends on DEBUG_KERNEL && ARCH_HAS_DEBUG_VIRTUAL 831 help 832 Enable some costly sanity checks in virtual to page code. This can 833 catch mistakes with virt_to_page() and friends. 834 835 If unsure, say N. 836 837config DEBUG_NOMMU_REGIONS 838 bool "Debug the global anon/private NOMMU mapping region tree" 839 depends on DEBUG_KERNEL && !MMU 840 help 841 This option causes the global tree of anonymous and private mapping 842 regions to be regularly checked for invalid topology. 843 844config DEBUG_MEMORY_INIT 845 bool "Debug memory initialisation" if EXPERT 846 default !EXPERT 847 help 848 Enable this for additional checks during memory initialisation. 849 The sanity checks verify aspects of the VM such as the memory model 850 and other information provided by the architecture. Verbose 851 information will be printed at KERN_DEBUG loglevel depending 852 on the mminit_loglevel= command-line option. 853 854 If unsure, say Y 855 856config MEMORY_NOTIFIER_ERROR_INJECT 857 tristate "Memory hotplug notifier error injection module" 858 depends on MEMORY_HOTPLUG_SPARSE && NOTIFIER_ERROR_INJECTION 859 help 860 This option provides the ability to inject artificial errors to 861 memory hotplug notifier chain callbacks. It is controlled through 862 debugfs interface under /sys/kernel/debug/notifier-error-inject/memory 863 864 If the notifier call chain should be failed with some events 865 notified, write the error code to "actions/<notifier event>/error". 866 867 Example: Inject memory hotplug offline error (-12 == -ENOMEM) 868 869 # cd /sys/kernel/debug/notifier-error-inject/memory 870 # echo -12 > actions/MEM_GOING_OFFLINE/error 871 # echo offline > /sys/devices/system/memory/memoryXXX/state 872 bash: echo: write error: Cannot allocate memory 873 874 To compile this code as a module, choose M here: the module will 875 be called memory-notifier-error-inject. 876 877 If unsure, say N. 878 879config DEBUG_PER_CPU_MAPS 880 bool "Debug access to per_cpu maps" 881 depends on DEBUG_KERNEL 882 depends on SMP 883 help 884 Say Y to verify that the per_cpu map being accessed has 885 been set up. This adds a fair amount of code to kernel memory 886 and decreases performance. 887 888 Say N if unsure. 889 890config DEBUG_KMAP_LOCAL 891 bool "Debug kmap_local temporary mappings" 892 depends on DEBUG_KERNEL && KMAP_LOCAL 893 help 894 This option enables additional error checking for the kmap_local 895 infrastructure. Disable for production use. 896 897config ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP 898 bool 899 900config DEBUG_KMAP_LOCAL_FORCE_MAP 901 bool "Enforce kmap_local temporary mappings" 902 depends on DEBUG_KERNEL && ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP 903 select KMAP_LOCAL 904 select DEBUG_KMAP_LOCAL 905 help 906 This option enforces temporary mappings through the kmap_local 907 mechanism for non-highmem pages and on non-highmem systems. 908 Disable this for production systems! 909 910config DEBUG_HIGHMEM 911 bool "Highmem debugging" 912 depends on DEBUG_KERNEL && HIGHMEM 913 select DEBUG_KMAP_LOCAL_FORCE_MAP if ARCH_SUPPORTS_KMAP_LOCAL_FORCE_MAP 914 select DEBUG_KMAP_LOCAL 915 help 916 This option enables additional error checking for high memory 917 systems. Disable for production systems. 918 919config HAVE_DEBUG_STACKOVERFLOW 920 bool 921 922config DEBUG_STACKOVERFLOW 923 bool "Check for stack overflows" 924 depends on DEBUG_KERNEL && HAVE_DEBUG_STACKOVERFLOW 925 help 926 Say Y here if you want to check for overflows of kernel, IRQ 927 and exception stacks (if your architecture uses them). This 928 option will show detailed messages if free stack space drops 929 below a certain limit. 930 931 These kinds of bugs usually occur when call-chains in the 932 kernel get too deep, especially when interrupts are 933 involved. 934 935 Use this in cases where you see apparently random memory 936 corruption, especially if it appears in 'struct thread_info' 937 938 If in doubt, say "N". 939 940source "lib/Kconfig.kasan" 941 942endmenu # "Memory Debugging" 943 944config DEBUG_SHIRQ 945 bool "Debug shared IRQ handlers" 946 depends on DEBUG_KERNEL 947 help 948 Enable this to generate a spurious interrupt just before a shared 949 interrupt handler is deregistered (generating one when registering 950 is currently disabled). Drivers need to handle this correctly. Some 951 don't and need to be caught. 952 953menu "Debug Oops, Lockups and Hangs" 954 955config PANIC_ON_OOPS 956 bool "Panic on Oops" 957 help 958 Say Y here to enable the kernel to panic when it oopses. This 959 has the same effect as setting oops=panic on the kernel command 960 line. 961 962 This feature is useful to ensure that the kernel does not do 963 anything erroneous after an oops which could result in data 964 corruption or other issues. 965 966 Say N if unsure. 967 968config PANIC_ON_OOPS_VALUE 969 int 970 range 0 1 971 default 0 if !PANIC_ON_OOPS 972 default 1 if PANIC_ON_OOPS 973 974config PANIC_TIMEOUT 975 int "panic timeout" 976 default 0 977 help 978 Set the timeout value (in seconds) until a reboot occurs when 979 the kernel panics. If n = 0, then we wait forever. A timeout 980 value n > 0 will wait n seconds before rebooting, while a timeout 981 value n < 0 will reboot immediately. 982 983config LOCKUP_DETECTOR 984 bool 985 986config SOFTLOCKUP_DETECTOR 987 bool "Detect Soft Lockups" 988 depends on DEBUG_KERNEL && !S390 989 select LOCKUP_DETECTOR 990 help 991 Say Y here to enable the kernel to act as a watchdog to detect 992 soft lockups. 993 994 Softlockups are bugs that cause the kernel to loop in kernel 995 mode for more than 20 seconds, without giving other tasks a 996 chance to run. The current stack trace is displayed upon 997 detection and the system will stay locked up. 998 999config BOOTPARAM_SOFTLOCKUP_PANIC 1000 bool "Panic (Reboot) On Soft Lockups" 1001 depends on SOFTLOCKUP_DETECTOR 1002 help 1003 Say Y here to enable the kernel to panic on "soft lockups", 1004 which are bugs that cause the kernel to loop in kernel 1005 mode for more than 20 seconds (configurable using the watchdog_thresh 1006 sysctl), without giving other tasks a chance to run. 1007 1008 The panic can be used in combination with panic_timeout, 1009 to cause the system to reboot automatically after a 1010 lockup has been detected. This feature is useful for 1011 high-availability systems that have uptime guarantees and 1012 where a lockup must be resolved ASAP. 1013 1014 Say N if unsure. 1015 1016config BOOTPARAM_SOFTLOCKUP_PANIC_VALUE 1017 int 1018 depends on SOFTLOCKUP_DETECTOR 1019 range 0 1 1020 default 0 if !BOOTPARAM_SOFTLOCKUP_PANIC 1021 default 1 if BOOTPARAM_SOFTLOCKUP_PANIC 1022 1023config HARDLOCKUP_DETECTOR_PERF 1024 bool 1025 select SOFTLOCKUP_DETECTOR 1026 1027# 1028# Enables a timestamp based low pass filter to compensate for perf based 1029# hard lockup detection which runs too fast due to turbo modes. 1030# 1031config HARDLOCKUP_CHECK_TIMESTAMP 1032 bool 1033 1034# 1035# arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard 1036# lockup detector rather than the perf based detector. 1037# 1038config HARDLOCKUP_DETECTOR 1039 bool "Detect Hard Lockups" 1040 depends on DEBUG_KERNEL && !S390 1041 depends on HAVE_HARDLOCKUP_DETECTOR_PERF || HAVE_HARDLOCKUP_DETECTOR_ARCH 1042 select LOCKUP_DETECTOR 1043 select HARDLOCKUP_DETECTOR_PERF if HAVE_HARDLOCKUP_DETECTOR_PERF 1044 select HARDLOCKUP_DETECTOR_ARCH if HAVE_HARDLOCKUP_DETECTOR_ARCH 1045 help 1046 Say Y here to enable the kernel to act as a watchdog to detect 1047 hard lockups. 1048 1049 Hardlockups are bugs that cause the CPU to loop in kernel mode 1050 for more than 10 seconds, without letting other interrupts have a 1051 chance to run. The current stack trace is displayed upon detection 1052 and the system will stay locked up. 1053 1054config BOOTPARAM_HARDLOCKUP_PANIC 1055 bool "Panic (Reboot) On Hard Lockups" 1056 depends on HARDLOCKUP_DETECTOR 1057 help 1058 Say Y here to enable the kernel to panic on "hard lockups", 1059 which are bugs that cause the kernel to loop in kernel 1060 mode with interrupts disabled for more than 10 seconds (configurable 1061 using the watchdog_thresh sysctl). 1062 1063 Say N if unsure. 1064 1065config BOOTPARAM_HARDLOCKUP_PANIC_VALUE 1066 int 1067 depends on HARDLOCKUP_DETECTOR 1068 range 0 1 1069 default 0 if !BOOTPARAM_HARDLOCKUP_PANIC 1070 default 1 if BOOTPARAM_HARDLOCKUP_PANIC 1071 1072config DETECT_HUNG_TASK 1073 bool "Detect Hung Tasks" 1074 depends on DEBUG_KERNEL 1075 default SOFTLOCKUP_DETECTOR 1076 help 1077 Say Y here to enable the kernel to detect "hung tasks", 1078 which are bugs that cause the task to be stuck in 1079 uninterruptible "D" state indefinitely. 1080 1081 When a hung task is detected, the kernel will print the 1082 current stack trace (which you should report), but the 1083 task will stay in uninterruptible state. If lockdep is 1084 enabled then all held locks will also be reported. This 1085 feature has negligible overhead. 1086 1087config DEFAULT_HUNG_TASK_TIMEOUT 1088 int "Default timeout for hung task detection (in seconds)" 1089 depends on DETECT_HUNG_TASK 1090 default 120 1091 help 1092 This option controls the default timeout (in seconds) used 1093 to determine when a task has become non-responsive and should 1094 be considered hung. 1095 1096 It can be adjusted at runtime via the kernel.hung_task_timeout_secs 1097 sysctl or by writing a value to 1098 /proc/sys/kernel/hung_task_timeout_secs. 1099 1100 A timeout of 0 disables the check. The default is two minutes. 1101 Keeping the default should be fine in most cases. 1102 1103config BOOTPARAM_HUNG_TASK_PANIC 1104 bool "Panic (Reboot) On Hung Tasks" 1105 depends on DETECT_HUNG_TASK 1106 help 1107 Say Y here to enable the kernel to panic on "hung tasks", 1108 which are bugs that cause the kernel to leave a task stuck 1109 in uninterruptible "D" state. 1110 1111 The panic can be used in combination with panic_timeout, 1112 to cause the system to reboot automatically after a 1113 hung task has been detected. This feature is useful for 1114 high-availability systems that have uptime guarantees and 1115 where a hung tasks must be resolved ASAP. 1116 1117 Say N if unsure. 1118 1119config BOOTPARAM_HUNG_TASK_PANIC_VALUE 1120 int 1121 depends on DETECT_HUNG_TASK 1122 range 0 1 1123 default 0 if !BOOTPARAM_HUNG_TASK_PANIC 1124 default 1 if BOOTPARAM_HUNG_TASK_PANIC 1125 1126config WQ_WATCHDOG 1127 bool "Detect Workqueue Stalls" 1128 depends on DEBUG_KERNEL 1129 help 1130 Say Y here to enable stall detection on workqueues. If a 1131 worker pool doesn't make forward progress on a pending work 1132 item for over a given amount of time, 30s by default, a 1133 warning message is printed along with dump of workqueue 1134 state. This can be configured through kernel parameter 1135 "workqueue.watchdog_thresh" and its sysfs counterpart. 1136 1137config TEST_LOCKUP 1138 tristate "Test module to generate lockups" 1139 depends on m 1140 help 1141 This builds the "test_lockup" module that helps to make sure 1142 that watchdogs and lockup detectors are working properly. 1143 1144 Depending on module parameters it could emulate soft or hard 1145 lockup, "hung task", or locking arbitrary lock for a long time. 1146 Also it could generate series of lockups with cooling-down periods. 1147 1148 If unsure, say N. 1149 1150endmenu # "Debug lockups and hangs" 1151 1152menu "Scheduler Debugging" 1153 1154config SCHED_DEBUG 1155 bool "Collect scheduler debugging info" 1156 depends on DEBUG_KERNEL && PROC_FS 1157 default y 1158 help 1159 If you say Y here, the /proc/sched_debug file will be provided 1160 that can help debug the scheduler. The runtime overhead of this 1161 option is minimal. 1162 1163config SCHED_INFO 1164 bool 1165 default n 1166 1167config SCHEDSTATS 1168 bool "Collect scheduler statistics" 1169 depends on DEBUG_KERNEL && PROC_FS 1170 select SCHED_INFO 1171 help 1172 If you say Y here, additional code will be inserted into the 1173 scheduler and related routines to collect statistics about 1174 scheduler behavior and provide them in /proc/schedstat. These 1175 stats may be useful for both tuning and debugging the scheduler 1176 If you aren't debugging the scheduler or trying to tune a specific 1177 application, you can say N to avoid the very slight overhead 1178 this adds. 1179 1180endmenu 1181 1182config DEBUG_TIMEKEEPING 1183 bool "Enable extra timekeeping sanity checking" 1184 help 1185 This option will enable additional timekeeping sanity checks 1186 which may be helpful when diagnosing issues where timekeeping 1187 problems are suspected. 1188 1189 This may include checks in the timekeeping hotpaths, so this 1190 option may have a (very small) performance impact to some 1191 workloads. 1192 1193 If unsure, say N. 1194 1195config DEBUG_PREEMPT 1196 bool "Debug preemptible kernel" 1197 depends on DEBUG_KERNEL && PREEMPTION && TRACE_IRQFLAGS_SUPPORT 1198 default y 1199 help 1200 If you say Y here then the kernel will use a debug variant of the 1201 commonly used smp_processor_id() function and will print warnings 1202 if kernel code uses it in a preemption-unsafe way. Also, the kernel 1203 will detect preemption count underflows. 1204 1205menu "Lock Debugging (spinlocks, mutexes, etc...)" 1206 1207config LOCK_DEBUGGING_SUPPORT 1208 bool 1209 depends on TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT 1210 default y 1211 1212config PROVE_LOCKING 1213 bool "Lock debugging: prove locking correctness" 1214 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1215 select LOCKDEP 1216 select DEBUG_SPINLOCK 1217 select DEBUG_MUTEXES 1218 select DEBUG_RT_MUTEXES if RT_MUTEXES 1219 select DEBUG_RWSEMS 1220 select DEBUG_WW_MUTEX_SLOWPATH 1221 select DEBUG_LOCK_ALLOC 1222 select PREEMPT_COUNT if !ARCH_NO_PREEMPT 1223 select TRACE_IRQFLAGS 1224 default n 1225 help 1226 This feature enables the kernel to prove that all locking 1227 that occurs in the kernel runtime is mathematically 1228 correct: that under no circumstance could an arbitrary (and 1229 not yet triggered) combination of observed locking 1230 sequences (on an arbitrary number of CPUs, running an 1231 arbitrary number of tasks and interrupt contexts) cause a 1232 deadlock. 1233 1234 In short, this feature enables the kernel to report locking 1235 related deadlocks before they actually occur. 1236 1237 The proof does not depend on how hard and complex a 1238 deadlock scenario would be to trigger: how many 1239 participant CPUs, tasks and irq-contexts would be needed 1240 for it to trigger. The proof also does not depend on 1241 timing: if a race and a resulting deadlock is possible 1242 theoretically (no matter how unlikely the race scenario 1243 is), it will be proven so and will immediately be 1244 reported by the kernel (once the event is observed that 1245 makes the deadlock theoretically possible). 1246 1247 If a deadlock is impossible (i.e. the locking rules, as 1248 observed by the kernel, are mathematically correct), the 1249 kernel reports nothing. 1250 1251 NOTE: this feature can also be enabled for rwlocks, mutexes 1252 and rwsems - in which case all dependencies between these 1253 different locking variants are observed and mapped too, and 1254 the proof of observed correctness is also maintained for an 1255 arbitrary combination of these separate locking variants. 1256 1257 For more details, see Documentation/locking/lockdep-design.rst. 1258 1259config PROVE_RAW_LOCK_NESTING 1260 bool "Enable raw_spinlock - spinlock nesting checks" 1261 depends on PROVE_LOCKING 1262 default n 1263 help 1264 Enable the raw_spinlock vs. spinlock nesting checks which ensure 1265 that the lock nesting rules for PREEMPT_RT enabled kernels are 1266 not violated. 1267 1268 NOTE: There are known nesting problems. So if you enable this 1269 option expect lockdep splats until these problems have been fully 1270 addressed which is work in progress. This config switch allows to 1271 identify and analyze these problems. It will be removed and the 1272 check permanentely enabled once the main issues have been fixed. 1273 1274 If unsure, select N. 1275 1276config LOCK_STAT 1277 bool "Lock usage statistics" 1278 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1279 select LOCKDEP 1280 select DEBUG_SPINLOCK 1281 select DEBUG_MUTEXES 1282 select DEBUG_RT_MUTEXES if RT_MUTEXES 1283 select DEBUG_LOCK_ALLOC 1284 default n 1285 help 1286 This feature enables tracking lock contention points 1287 1288 For more details, see Documentation/locking/lockstat.rst 1289 1290 This also enables lock events required by "perf lock", 1291 subcommand of perf. 1292 If you want to use "perf lock", you also need to turn on 1293 CONFIG_EVENT_TRACING. 1294 1295 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events. 1296 (CONFIG_LOCKDEP defines "acquire" and "release" events.) 1297 1298config DEBUG_RT_MUTEXES 1299 bool "RT Mutex debugging, deadlock detection" 1300 depends on DEBUG_KERNEL && RT_MUTEXES 1301 help 1302 This allows rt mutex semantics violations and rt mutex related 1303 deadlocks (lockups) to be detected and reported automatically. 1304 1305config DEBUG_SPINLOCK 1306 bool "Spinlock and rw-lock debugging: basic checks" 1307 depends on DEBUG_KERNEL 1308 select UNINLINE_SPIN_UNLOCK 1309 help 1310 Say Y here and build SMP to catch missing spinlock initialization 1311 and certain other kinds of spinlock errors commonly made. This is 1312 best used in conjunction with the NMI watchdog so that spinlock 1313 deadlocks are also debuggable. 1314 1315config DEBUG_MUTEXES 1316 bool "Mutex debugging: basic checks" 1317 depends on DEBUG_KERNEL 1318 help 1319 This feature allows mutex semantics violations to be detected and 1320 reported. 1321 1322config DEBUG_WW_MUTEX_SLOWPATH 1323 bool "Wait/wound mutex debugging: Slowpath testing" 1324 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1325 select DEBUG_LOCK_ALLOC 1326 select DEBUG_SPINLOCK 1327 select DEBUG_MUTEXES 1328 help 1329 This feature enables slowpath testing for w/w mutex users by 1330 injecting additional -EDEADLK wound/backoff cases. Together with 1331 the full mutex checks enabled with (CONFIG_PROVE_LOCKING) this 1332 will test all possible w/w mutex interface abuse with the 1333 exception of simply not acquiring all the required locks. 1334 Note that this feature can introduce significant overhead, so 1335 it really should not be enabled in a production or distro kernel, 1336 even a debug kernel. If you are a driver writer, enable it. If 1337 you are a distro, do not. 1338 1339config DEBUG_RWSEMS 1340 bool "RW Semaphore debugging: basic checks" 1341 depends on DEBUG_KERNEL 1342 help 1343 This debugging feature allows mismatched rw semaphore locks 1344 and unlocks to be detected and reported. 1345 1346config DEBUG_LOCK_ALLOC 1347 bool "Lock debugging: detect incorrect freeing of live locks" 1348 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1349 select DEBUG_SPINLOCK 1350 select DEBUG_MUTEXES 1351 select DEBUG_RT_MUTEXES if RT_MUTEXES 1352 select LOCKDEP 1353 help 1354 This feature will check whether any held lock (spinlock, rwlock, 1355 mutex or rwsem) is incorrectly freed by the kernel, via any of the 1356 memory-freeing routines (kfree(), kmem_cache_free(), free_pages(), 1357 vfree(), etc.), whether a live lock is incorrectly reinitialized via 1358 spin_lock_init()/mutex_init()/etc., or whether there is any lock 1359 held during task exit. 1360 1361config LOCKDEP 1362 bool 1363 depends on DEBUG_KERNEL && LOCK_DEBUGGING_SUPPORT 1364 select STACKTRACE 1365 select FRAME_POINTER if !MIPS && !PPC && !ARM && !S390 && !MICROBLAZE && !ARC && !X86 1366 select KALLSYMS 1367 select KALLSYMS_ALL 1368 1369config LOCKDEP_SMALL 1370 bool 1371 1372config DEBUG_LOCKDEP 1373 bool "Lock dependency engine debugging" 1374 depends on DEBUG_KERNEL && LOCKDEP 1375 select DEBUG_IRQFLAGS 1376 help 1377 If you say Y here, the lock dependency engine will do 1378 additional runtime checks to debug itself, at the price 1379 of more runtime overhead. 1380 1381config DEBUG_ATOMIC_SLEEP 1382 bool "Sleep inside atomic section checking" 1383 select PREEMPT_COUNT 1384 depends on DEBUG_KERNEL 1385 depends on !ARCH_NO_PREEMPT 1386 help 1387 If you say Y here, various routines which may sleep will become very 1388 noisy if they are called inside atomic sections: when a spinlock is 1389 held, inside an rcu read side critical section, inside preempt disabled 1390 sections, inside an interrupt, etc... 1391 1392config DEBUG_LOCKING_API_SELFTESTS 1393 bool "Locking API boot-time self-tests" 1394 depends on DEBUG_KERNEL 1395 help 1396 Say Y here if you want the kernel to run a short self-test during 1397 bootup. The self-test checks whether common types of locking bugs 1398 are detected by debugging mechanisms or not. (if you disable 1399 lock debugging then those bugs wont be detected of course.) 1400 The following locking APIs are covered: spinlocks, rwlocks, 1401 mutexes and rwsems. 1402 1403config LOCK_TORTURE_TEST 1404 tristate "torture tests for locking" 1405 depends on DEBUG_KERNEL 1406 select TORTURE_TEST 1407 help 1408 This option provides a kernel module that runs torture tests 1409 on kernel locking primitives. The kernel module may be built 1410 after the fact on the running kernel to be tested, if desired. 1411 1412 Say Y here if you want kernel locking-primitive torture tests 1413 to be built into the kernel. 1414 Say M if you want these torture tests to build as a module. 1415 Say N if you are unsure. 1416 1417config WW_MUTEX_SELFTEST 1418 tristate "Wait/wound mutex selftests" 1419 help 1420 This option provides a kernel module that runs tests on the 1421 on the struct ww_mutex locking API. 1422 1423 It is recommended to enable DEBUG_WW_MUTEX_SLOWPATH in conjunction 1424 with this test harness. 1425 1426 Say M if you want these self tests to build as a module. 1427 Say N if you are unsure. 1428 1429config SCF_TORTURE_TEST 1430 tristate "torture tests for smp_call_function*()" 1431 depends on DEBUG_KERNEL 1432 select TORTURE_TEST 1433 help 1434 This option provides a kernel module that runs torture tests 1435 on the smp_call_function() family of primitives. The kernel 1436 module may be built after the fact on the running kernel to 1437 be tested, if desired. 1438 1439config CSD_LOCK_WAIT_DEBUG 1440 bool "Debugging for csd_lock_wait(), called from smp_call_function*()" 1441 depends on DEBUG_KERNEL 1442 depends on 64BIT 1443 default n 1444 help 1445 This option enables debug prints when CPUs are slow to respond 1446 to the smp_call_function*() IPI wrappers. These debug prints 1447 include the IPI handler function currently executing (if any) 1448 and relevant stack traces. 1449 1450endmenu # lock debugging 1451 1452config TRACE_IRQFLAGS 1453 depends on TRACE_IRQFLAGS_SUPPORT 1454 bool 1455 help 1456 Enables hooks to interrupt enabling and disabling for 1457 either tracing or lock debugging. 1458 1459config TRACE_IRQFLAGS_NMI 1460 def_bool y 1461 depends on TRACE_IRQFLAGS 1462 depends on TRACE_IRQFLAGS_NMI_SUPPORT 1463 1464config DEBUG_IRQFLAGS 1465 bool "Debug IRQ flag manipulation" 1466 help 1467 Enables checks for potentially unsafe enabling or disabling of 1468 interrupts, such as calling raw_local_irq_restore() when interrupts 1469 are enabled. 1470 1471config STACKTRACE 1472 bool "Stack backtrace support" 1473 depends on STACKTRACE_SUPPORT 1474 help 1475 This option causes the kernel to create a /proc/pid/stack for 1476 every process, showing its current stack trace. 1477 It is also used by various kernel debugging features that require 1478 stack trace generation. 1479 1480config WARN_ALL_UNSEEDED_RANDOM 1481 bool "Warn for all uses of unseeded randomness" 1482 default n 1483 help 1484 Some parts of the kernel contain bugs relating to their use of 1485 cryptographically secure random numbers before it's actually possible 1486 to generate those numbers securely. This setting ensures that these 1487 flaws don't go unnoticed, by enabling a message, should this ever 1488 occur. This will allow people with obscure setups to know when things 1489 are going wrong, so that they might contact developers about fixing 1490 it. 1491 1492 Unfortunately, on some models of some architectures getting 1493 a fully seeded CRNG is extremely difficult, and so this can 1494 result in dmesg getting spammed for a surprisingly long 1495 time. This is really bad from a security perspective, and 1496 so architecture maintainers really need to do what they can 1497 to get the CRNG seeded sooner after the system is booted. 1498 However, since users cannot do anything actionable to 1499 address this, by default the kernel will issue only a single 1500 warning for the first use of unseeded randomness. 1501 1502 Say Y here if you want to receive warnings for all uses of 1503 unseeded randomness. This will be of use primarily for 1504 those developers interested in improving the security of 1505 Linux kernels running on their architecture (or 1506 subarchitecture). 1507 1508config DEBUG_KOBJECT 1509 bool "kobject debugging" 1510 depends on DEBUG_KERNEL 1511 help 1512 If you say Y here, some extra kobject debugging messages will be sent 1513 to the syslog. 1514 1515config DEBUG_KOBJECT_RELEASE 1516 bool "kobject release debugging" 1517 depends on DEBUG_OBJECTS_TIMERS 1518 help 1519 kobjects are reference counted objects. This means that their 1520 last reference count put is not predictable, and the kobject can 1521 live on past the point at which a driver decides to drop it's 1522 initial reference to the kobject gained on allocation. An 1523 example of this would be a struct device which has just been 1524 unregistered. 1525 1526 However, some buggy drivers assume that after such an operation, 1527 the memory backing the kobject can be immediately freed. This 1528 goes completely against the principles of a refcounted object. 1529 1530 If you say Y here, the kernel will delay the release of kobjects 1531 on the last reference count to improve the visibility of this 1532 kind of kobject release bug. 1533 1534config HAVE_DEBUG_BUGVERBOSE 1535 bool 1536 1537menu "Debug kernel data structures" 1538 1539config DEBUG_LIST 1540 bool "Debug linked list manipulation" 1541 depends on DEBUG_KERNEL || BUG_ON_DATA_CORRUPTION 1542 help 1543 Enable this to turn on extended checks in the linked-list 1544 walking routines. 1545 1546 If unsure, say N. 1547 1548config DEBUG_PLIST 1549 bool "Debug priority linked list manipulation" 1550 depends on DEBUG_KERNEL 1551 help 1552 Enable this to turn on extended checks in the priority-ordered 1553 linked-list (plist) walking routines. This checks the entire 1554 list multiple times during each manipulation. 1555 1556 If unsure, say N. 1557 1558config DEBUG_SG 1559 bool "Debug SG table operations" 1560 depends on DEBUG_KERNEL 1561 help 1562 Enable this to turn on checks on scatter-gather tables. This can 1563 help find problems with drivers that do not properly initialize 1564 their sg tables. 1565 1566 If unsure, say N. 1567 1568config DEBUG_NOTIFIERS 1569 bool "Debug notifier call chains" 1570 depends on DEBUG_KERNEL 1571 help 1572 Enable this to turn on sanity checking for notifier call chains. 1573 This is most useful for kernel developers to make sure that 1574 modules properly unregister themselves from notifier chains. 1575 This is a relatively cheap check but if you care about maximum 1576 performance, say N. 1577 1578config BUG_ON_DATA_CORRUPTION 1579 bool "Trigger a BUG when data corruption is detected" 1580 select DEBUG_LIST 1581 help 1582 Select this option if the kernel should BUG when it encounters 1583 data corruption in kernel memory structures when they get checked 1584 for validity. 1585 1586 If unsure, say N. 1587 1588endmenu 1589 1590config DEBUG_CREDENTIALS 1591 bool "Debug credential management" 1592 depends on DEBUG_KERNEL 1593 help 1594 Enable this to turn on some debug checking for credential 1595 management. The additional code keeps track of the number of 1596 pointers from task_structs to any given cred struct, and checks to 1597 see that this number never exceeds the usage count of the cred 1598 struct. 1599 1600 Furthermore, if SELinux is enabled, this also checks that the 1601 security pointer in the cred struct is never seen to be invalid. 1602 1603 If unsure, say N. 1604 1605source "kernel/rcu/Kconfig.debug" 1606 1607config DEBUG_WQ_FORCE_RR_CPU 1608 bool "Force round-robin CPU selection for unbound work items" 1609 depends on DEBUG_KERNEL 1610 default n 1611 help 1612 Workqueue used to implicitly guarantee that work items queued 1613 without explicit CPU specified are put on the local CPU. This 1614 guarantee is no longer true and while local CPU is still 1615 preferred work items may be put on foreign CPUs. Kernel 1616 parameter "workqueue.debug_force_rr_cpu" is added to force 1617 round-robin CPU selection to flush out usages which depend on the 1618 now broken guarantee. This config option enables the debug 1619 feature by default. When enabled, memory and cache locality will 1620 be impacted. 1621 1622config DEBUG_BLOCK_EXT_DEVT 1623 bool "Force extended block device numbers and spread them" 1624 depends on DEBUG_KERNEL 1625 depends on BLOCK 1626 default n 1627 help 1628 BIG FAT WARNING: ENABLING THIS OPTION MIGHT BREAK BOOTING ON 1629 SOME DISTRIBUTIONS. DO NOT ENABLE THIS UNLESS YOU KNOW WHAT 1630 YOU ARE DOING. Distros, please enable this and fix whatever 1631 is broken. 1632 1633 Conventionally, block device numbers are allocated from 1634 predetermined contiguous area. However, extended block area 1635 may introduce non-contiguous block device numbers. This 1636 option forces most block device numbers to be allocated from 1637 the extended space and spreads them to discover kernel or 1638 userland code paths which assume predetermined contiguous 1639 device number allocation. 1640 1641 Note that turning on this debug option shuffles all the 1642 device numbers for all IDE and SCSI devices including libata 1643 ones, so root partition specified using device number 1644 directly (via rdev or root=MAJ:MIN) won't work anymore. 1645 Textual device names (root=/dev/sdXn) will continue to work. 1646 1647 Say N if you are unsure. 1648 1649config CPU_HOTPLUG_STATE_CONTROL 1650 bool "Enable CPU hotplug state control" 1651 depends on DEBUG_KERNEL 1652 depends on HOTPLUG_CPU 1653 default n 1654 help 1655 Allows to write steps between "offline" and "online" to the CPUs 1656 sysfs target file so states can be stepped granular. This is a debug 1657 option for now as the hotplug machinery cannot be stopped and 1658 restarted at arbitrary points yet. 1659 1660 Say N if your are unsure. 1661 1662config LATENCYTOP 1663 bool "Latency measuring infrastructure" 1664 depends on DEBUG_KERNEL 1665 depends on STACKTRACE_SUPPORT 1666 depends on PROC_FS 1667 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86 1668 select KALLSYMS 1669 select KALLSYMS_ALL 1670 select STACKTRACE 1671 select SCHEDSTATS 1672 select SCHED_DEBUG 1673 help 1674 Enable this option if you want to use the LatencyTOP tool 1675 to find out which userspace is blocking on what kernel operations. 1676 1677source "kernel/trace/Kconfig" 1678 1679config PROVIDE_OHCI1394_DMA_INIT 1680 bool "Remote debugging over FireWire early on boot" 1681 depends on PCI && X86 1682 help 1683 If you want to debug problems which hang or crash the kernel early 1684 on boot and the crashing machine has a FireWire port, you can use 1685 this feature to remotely access the memory of the crashed machine 1686 over FireWire. This employs remote DMA as part of the OHCI1394 1687 specification which is now the standard for FireWire controllers. 1688 1689 With remote DMA, you can monitor the printk buffer remotely using 1690 firescope and access all memory below 4GB using fireproxy from gdb. 1691 Even controlling a kernel debugger is possible using remote DMA. 1692 1693 Usage: 1694 1695 If ohci1394_dma=early is used as boot parameter, it will initialize 1696 all OHCI1394 controllers which are found in the PCI config space. 1697 1698 As all changes to the FireWire bus such as enabling and disabling 1699 devices cause a bus reset and thereby disable remote DMA for all 1700 devices, be sure to have the cable plugged and FireWire enabled on 1701 the debugging host before booting the debug target for debugging. 1702 1703 This code (~1k) is freed after boot. By then, the firewire stack 1704 in charge of the OHCI-1394 controllers should be used instead. 1705 1706 See Documentation/core-api/debugging-via-ohci1394.rst for more information. 1707 1708source "samples/Kconfig" 1709 1710config ARCH_HAS_DEVMEM_IS_ALLOWED 1711 bool 1712 1713config STRICT_DEVMEM 1714 bool "Filter access to /dev/mem" 1715 depends on MMU && DEVMEM 1716 depends on ARCH_HAS_DEVMEM_IS_ALLOWED || GENERIC_LIB_DEVMEM_IS_ALLOWED 1717 default y if PPC || X86 || ARM64 1718 help 1719 If this option is disabled, you allow userspace (root) access to all 1720 of memory, including kernel and userspace memory. Accidental 1721 access to this is obviously disastrous, but specific access can 1722 be used by people debugging the kernel. Note that with PAT support 1723 enabled, even in this case there are restrictions on /dev/mem 1724 use due to the cache aliasing requirements. 1725 1726 If this option is switched on, and IO_STRICT_DEVMEM=n, the /dev/mem 1727 file only allows userspace access to PCI space and the BIOS code and 1728 data regions. This is sufficient for dosemu and X and all common 1729 users of /dev/mem. 1730 1731 If in doubt, say Y. 1732 1733config IO_STRICT_DEVMEM 1734 bool "Filter I/O access to /dev/mem" 1735 depends on STRICT_DEVMEM 1736 help 1737 If this option is disabled, you allow userspace (root) access to all 1738 io-memory regardless of whether a driver is actively using that 1739 range. Accidental access to this is obviously disastrous, but 1740 specific access can be used by people debugging kernel drivers. 1741 1742 If this option is switched on, the /dev/mem file only allows 1743 userspace access to *idle* io-memory ranges (see /proc/iomem) This 1744 may break traditional users of /dev/mem (dosemu, legacy X, etc...) 1745 if the driver using a given range cannot be disabled. 1746 1747 If in doubt, say Y. 1748 1749menu "$(SRCARCH) Debugging" 1750 1751source "arch/$(SRCARCH)/Kconfig.debug" 1752 1753endmenu 1754 1755menu "Kernel Testing and Coverage" 1756 1757source "lib/kunit/Kconfig" 1758 1759config NOTIFIER_ERROR_INJECTION 1760 tristate "Notifier error injection" 1761 depends on DEBUG_KERNEL 1762 select DEBUG_FS 1763 help 1764 This option provides the ability to inject artificial errors to 1765 specified notifier chain callbacks. It is useful to test the error 1766 handling of notifier call chain failures. 1767 1768 Say N if unsure. 1769 1770config PM_NOTIFIER_ERROR_INJECT 1771 tristate "PM notifier error injection module" 1772 depends on PM && NOTIFIER_ERROR_INJECTION 1773 default m if PM_DEBUG 1774 help 1775 This option provides the ability to inject artificial errors to 1776 PM notifier chain callbacks. It is controlled through debugfs 1777 interface /sys/kernel/debug/notifier-error-inject/pm 1778 1779 If the notifier call chain should be failed with some events 1780 notified, write the error code to "actions/<notifier event>/error". 1781 1782 Example: Inject PM suspend error (-12 = -ENOMEM) 1783 1784 # cd /sys/kernel/debug/notifier-error-inject/pm/ 1785 # echo -12 > actions/PM_SUSPEND_PREPARE/error 1786 # echo mem > /sys/power/state 1787 bash: echo: write error: Cannot allocate memory 1788 1789 To compile this code as a module, choose M here: the module will 1790 be called pm-notifier-error-inject. 1791 1792 If unsure, say N. 1793 1794config OF_RECONFIG_NOTIFIER_ERROR_INJECT 1795 tristate "OF reconfig notifier error injection module" 1796 depends on OF_DYNAMIC && NOTIFIER_ERROR_INJECTION 1797 help 1798 This option provides the ability to inject artificial errors to 1799 OF reconfig notifier chain callbacks. It is controlled 1800 through debugfs interface under 1801 /sys/kernel/debug/notifier-error-inject/OF-reconfig/ 1802 1803 If the notifier call chain should be failed with some events 1804 notified, write the error code to "actions/<notifier event>/error". 1805 1806 To compile this code as a module, choose M here: the module will 1807 be called of-reconfig-notifier-error-inject. 1808 1809 If unsure, say N. 1810 1811config NETDEV_NOTIFIER_ERROR_INJECT 1812 tristate "Netdev notifier error injection module" 1813 depends on NET && NOTIFIER_ERROR_INJECTION 1814 help 1815 This option provides the ability to inject artificial errors to 1816 netdevice notifier chain callbacks. It is controlled through debugfs 1817 interface /sys/kernel/debug/notifier-error-inject/netdev 1818 1819 If the notifier call chain should be failed with some events 1820 notified, write the error code to "actions/<notifier event>/error". 1821 1822 Example: Inject netdevice mtu change error (-22 = -EINVAL) 1823 1824 # cd /sys/kernel/debug/notifier-error-inject/netdev 1825 # echo -22 > actions/NETDEV_CHANGEMTU/error 1826 # ip link set eth0 mtu 1024 1827 RTNETLINK answers: Invalid argument 1828 1829 To compile this code as a module, choose M here: the module will 1830 be called netdev-notifier-error-inject. 1831 1832 If unsure, say N. 1833 1834config FUNCTION_ERROR_INJECTION 1835 def_bool y 1836 depends on HAVE_FUNCTION_ERROR_INJECTION && KPROBES 1837 1838config FAULT_INJECTION 1839 bool "Fault-injection framework" 1840 depends on DEBUG_KERNEL 1841 help 1842 Provide fault-injection framework. 1843 For more details, see Documentation/fault-injection/. 1844 1845config FAILSLAB 1846 bool "Fault-injection capability for kmalloc" 1847 depends on FAULT_INJECTION 1848 depends on SLAB || SLUB 1849 help 1850 Provide fault-injection capability for kmalloc. 1851 1852config FAIL_PAGE_ALLOC 1853 bool "Fault-injection capability for alloc_pages()" 1854 depends on FAULT_INJECTION 1855 help 1856 Provide fault-injection capability for alloc_pages(). 1857 1858config FAULT_INJECTION_USERCOPY 1859 bool "Fault injection capability for usercopy functions" 1860 depends on FAULT_INJECTION 1861 help 1862 Provides fault-injection capability to inject failures 1863 in usercopy functions (copy_from_user(), get_user(), ...). 1864 1865config FAIL_MAKE_REQUEST 1866 bool "Fault-injection capability for disk IO" 1867 depends on FAULT_INJECTION && BLOCK 1868 help 1869 Provide fault-injection capability for disk IO. 1870 1871config FAIL_IO_TIMEOUT 1872 bool "Fault-injection capability for faking disk interrupts" 1873 depends on FAULT_INJECTION && BLOCK 1874 help 1875 Provide fault-injection capability on end IO handling. This 1876 will make the block layer "forget" an interrupt as configured, 1877 thus exercising the error handling. 1878 1879 Only works with drivers that use the generic timeout handling, 1880 for others it wont do anything. 1881 1882config FAIL_FUTEX 1883 bool "Fault-injection capability for futexes" 1884 select DEBUG_FS 1885 depends on FAULT_INJECTION && FUTEX 1886 help 1887 Provide fault-injection capability for futexes. 1888 1889config FAULT_INJECTION_DEBUG_FS 1890 bool "Debugfs entries for fault-injection capabilities" 1891 depends on FAULT_INJECTION && SYSFS && DEBUG_FS 1892 help 1893 Enable configuration of fault-injection capabilities via debugfs. 1894 1895config FAIL_FUNCTION 1896 bool "Fault-injection capability for functions" 1897 depends on FAULT_INJECTION_DEBUG_FS && FUNCTION_ERROR_INJECTION 1898 help 1899 Provide function-based fault-injection capability. 1900 This will allow you to override a specific function with a return 1901 with given return value. As a result, function caller will see 1902 an error value and have to handle it. This is useful to test the 1903 error handling in various subsystems. 1904 1905config FAIL_MMC_REQUEST 1906 bool "Fault-injection capability for MMC IO" 1907 depends on FAULT_INJECTION_DEBUG_FS && MMC 1908 help 1909 Provide fault-injection capability for MMC IO. 1910 This will make the mmc core return data errors. This is 1911 useful to test the error handling in the mmc block device 1912 and to test how the mmc host driver handles retries from 1913 the block device. 1914 1915config FAULT_INJECTION_STACKTRACE_FILTER 1916 bool "stacktrace filter for fault-injection capabilities" 1917 depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT 1918 depends on !X86_64 1919 select STACKTRACE 1920 select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM && !ARC && !X86 1921 help 1922 Provide stacktrace filter for fault-injection capabilities 1923 1924config ARCH_HAS_KCOV 1925 bool 1926 help 1927 An architecture should select this when it can successfully 1928 build and run with CONFIG_KCOV. This typically requires 1929 disabling instrumentation for some early boot code. 1930 1931config CC_HAS_SANCOV_TRACE_PC 1932 def_bool $(cc-option,-fsanitize-coverage=trace-pc) 1933 1934 1935config KCOV 1936 bool "Code coverage for fuzzing" 1937 depends on ARCH_HAS_KCOV 1938 depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS 1939 select DEBUG_FS 1940 select GCC_PLUGIN_SANCOV if !CC_HAS_SANCOV_TRACE_PC 1941 help 1942 KCOV exposes kernel code coverage information in a form suitable 1943 for coverage-guided fuzzing (randomized testing). 1944 1945 If RANDOMIZE_BASE is enabled, PC values will not be stable across 1946 different machines and across reboots. If you need stable PC values, 1947 disable RANDOMIZE_BASE. 1948 1949 For more details, see Documentation/dev-tools/kcov.rst. 1950 1951config KCOV_ENABLE_COMPARISONS 1952 bool "Enable comparison operands collection by KCOV" 1953 depends on KCOV 1954 depends on $(cc-option,-fsanitize-coverage=trace-cmp) 1955 help 1956 KCOV also exposes operands of every comparison in the instrumented 1957 code along with operand sizes and PCs of the comparison instructions. 1958 These operands can be used by fuzzing engines to improve the quality 1959 of fuzzing coverage. 1960 1961config KCOV_INSTRUMENT_ALL 1962 bool "Instrument all code by default" 1963 depends on KCOV 1964 default y 1965 help 1966 If you are doing generic system call fuzzing (like e.g. syzkaller), 1967 then you will want to instrument the whole kernel and you should 1968 say y here. If you are doing more targeted fuzzing (like e.g. 1969 filesystem fuzzing with AFL) then you will want to enable coverage 1970 for more specific subsets of files, and should say n here. 1971 1972config KCOV_IRQ_AREA_SIZE 1973 hex "Size of interrupt coverage collection area in words" 1974 depends on KCOV 1975 default 0x40000 1976 help 1977 KCOV uses preallocated per-cpu areas to collect coverage from 1978 soft interrupts. This specifies the size of those areas in the 1979 number of unsigned long words. 1980 1981menuconfig RUNTIME_TESTING_MENU 1982 bool "Runtime Testing" 1983 def_bool y 1984 1985if RUNTIME_TESTING_MENU 1986 1987config LKDTM 1988 tristate "Linux Kernel Dump Test Tool Module" 1989 depends on DEBUG_FS 1990 help 1991 This module enables testing of the different dumping mechanisms by 1992 inducing system failures at predefined crash points. 1993 If you don't need it: say N 1994 Choose M here to compile this code as a module. The module will be 1995 called lkdtm. 1996 1997 Documentation on how to use the module can be found in 1998 Documentation/fault-injection/provoke-crashes.rst 1999 2000config TEST_LIST_SORT 2001 tristate "Linked list sorting test" 2002 depends on DEBUG_KERNEL || m 2003 help 2004 Enable this to turn on 'list_sort()' function test. This test is 2005 executed only once during system boot (so affects only boot time), 2006 or at module load time. 2007 2008 If unsure, say N. 2009 2010config TEST_MIN_HEAP 2011 tristate "Min heap test" 2012 depends on DEBUG_KERNEL || m 2013 help 2014 Enable this to turn on min heap function tests. This test is 2015 executed only once during system boot (so affects only boot time), 2016 or at module load time. 2017 2018 If unsure, say N. 2019 2020config TEST_SORT 2021 tristate "Array-based sort test" 2022 depends on DEBUG_KERNEL || m 2023 help 2024 This option enables the self-test function of 'sort()' at boot, 2025 or at module load time. 2026 2027 If unsure, say N. 2028 2029config KPROBES_SANITY_TEST 2030 bool "Kprobes sanity tests" 2031 depends on DEBUG_KERNEL 2032 depends on KPROBES 2033 help 2034 This option provides for testing basic kprobes functionality on 2035 boot. Samples of kprobe and kretprobe are inserted and 2036 verified for functionality. 2037 2038 Say N if you are unsure. 2039 2040config BACKTRACE_SELF_TEST 2041 tristate "Self test for the backtrace code" 2042 depends on DEBUG_KERNEL 2043 help 2044 This option provides a kernel module that can be used to test 2045 the kernel stack backtrace code. This option is not useful 2046 for distributions or general kernels, but only for kernel 2047 developers working on architecture code. 2048 2049 Note that if you want to also test saved backtraces, you will 2050 have to enable STACKTRACE as well. 2051 2052 Say N if you are unsure. 2053 2054config RBTREE_TEST 2055 tristate "Red-Black tree test" 2056 depends on DEBUG_KERNEL 2057 help 2058 A benchmark measuring the performance of the rbtree library. 2059 Also includes rbtree invariant checks. 2060 2061config REED_SOLOMON_TEST 2062 tristate "Reed-Solomon library test" 2063 depends on DEBUG_KERNEL || m 2064 select REED_SOLOMON 2065 select REED_SOLOMON_ENC16 2066 select REED_SOLOMON_DEC16 2067 help 2068 This option enables the self-test function of rslib at boot, 2069 or at module load time. 2070 2071 If unsure, say N. 2072 2073config INTERVAL_TREE_TEST 2074 tristate "Interval tree test" 2075 depends on DEBUG_KERNEL 2076 select INTERVAL_TREE 2077 help 2078 A benchmark measuring the performance of the interval tree library 2079 2080config PERCPU_TEST 2081 tristate "Per cpu operations test" 2082 depends on m && DEBUG_KERNEL 2083 help 2084 Enable this option to build test module which validates per-cpu 2085 operations. 2086 2087 If unsure, say N. 2088 2089config ATOMIC64_SELFTEST 2090 tristate "Perform an atomic64_t self-test" 2091 help 2092 Enable this option to test the atomic64_t functions at boot or 2093 at module load time. 2094 2095 If unsure, say N. 2096 2097config ASYNC_RAID6_TEST 2098 tristate "Self test for hardware accelerated raid6 recovery" 2099 depends on ASYNC_RAID6_RECOV 2100 select ASYNC_MEMCPY 2101 help 2102 This is a one-shot self test that permutes through the 2103 recovery of all the possible two disk failure scenarios for a 2104 N-disk array. Recovery is performed with the asynchronous 2105 raid6 recovery routines, and will optionally use an offload 2106 engine if one is available. 2107 2108 If unsure, say N. 2109 2110config TEST_HEXDUMP 2111 tristate "Test functions located in the hexdump module at runtime" 2112 2113config TEST_STRING_HELPERS 2114 tristate "Test functions located in the string_helpers module at runtime" 2115 2116config TEST_STRSCPY 2117 tristate "Test strscpy*() family of functions at runtime" 2118 2119config TEST_KSTRTOX 2120 tristate "Test kstrto*() family of functions at runtime" 2121 2122config TEST_PRINTF 2123 tristate "Test printf() family of functions at runtime" 2124 2125config TEST_BITMAP 2126 tristate "Test bitmap_*() family of functions at runtime" 2127 help 2128 Enable this option to test the bitmap functions at boot. 2129 2130 If unsure, say N. 2131 2132config TEST_UUID 2133 tristate "Test functions located in the uuid module at runtime" 2134 2135config TEST_XARRAY 2136 tristate "Test the XArray code at runtime" 2137 2138config TEST_OVERFLOW 2139 tristate "Test check_*_overflow() functions at runtime" 2140 2141config TEST_RHASHTABLE 2142 tristate "Perform selftest on resizable hash table" 2143 help 2144 Enable this option to test the rhashtable functions at boot. 2145 2146 If unsure, say N. 2147 2148config TEST_HASH 2149 tristate "Perform selftest on hash functions" 2150 help 2151 Enable this option to test the kernel's integer (<linux/hash.h>), 2152 string (<linux/stringhash.h>), and siphash (<linux/siphash.h>) 2153 hash functions on boot (or module load). 2154 2155 This is intended to help people writing architecture-specific 2156 optimized versions. If unsure, say N. 2157 2158config TEST_IDA 2159 tristate "Perform selftest on IDA functions" 2160 2161config TEST_PARMAN 2162 tristate "Perform selftest on priority array manager" 2163 depends on PARMAN 2164 help 2165 Enable this option to test priority array manager on boot 2166 (or module load). 2167 2168 If unsure, say N. 2169 2170config TEST_IRQ_TIMINGS 2171 bool "IRQ timings selftest" 2172 depends on IRQ_TIMINGS 2173 help 2174 Enable this option to test the irq timings code on boot. 2175 2176 If unsure, say N. 2177 2178config TEST_LKM 2179 tristate "Test module loading with 'hello world' module" 2180 depends on m 2181 help 2182 This builds the "test_module" module that emits "Hello, world" 2183 on printk when loaded. It is designed to be used for basic 2184 evaluation of the module loading subsystem (for example when 2185 validating module verification). It lacks any extra dependencies, 2186 and will not normally be loaded by the system unless explicitly 2187 requested by name. 2188 2189 If unsure, say N. 2190 2191config TEST_BITOPS 2192 tristate "Test module for compilation of bitops operations" 2193 depends on m 2194 help 2195 This builds the "test_bitops" module that is much like the 2196 TEST_LKM module except that it does a basic exercise of the 2197 set/clear_bit macros and get_count_order/long to make sure there are 2198 no compiler warnings from C=1 sparse checker or -Wextra 2199 compilations. It has no dependencies and doesn't run or load unless 2200 explicitly requested by name. for example: modprobe test_bitops. 2201 2202 If unsure, say N. 2203 2204config TEST_VMALLOC 2205 tristate "Test module for stress/performance analysis of vmalloc allocator" 2206 default n 2207 depends on MMU 2208 depends on m 2209 help 2210 This builds the "test_vmalloc" module that should be used for 2211 stress and performance analysis. So, any new change for vmalloc 2212 subsystem can be evaluated from performance and stability point 2213 of view. 2214 2215 If unsure, say N. 2216 2217config TEST_USER_COPY 2218 tristate "Test user/kernel boundary protections" 2219 depends on m 2220 help 2221 This builds the "test_user_copy" module that runs sanity checks 2222 on the copy_to/from_user infrastructure, making sure basic 2223 user/kernel boundary testing is working. If it fails to load, 2224 a regression has been detected in the user/kernel memory boundary 2225 protections. 2226 2227 If unsure, say N. 2228 2229config TEST_BPF 2230 tristate "Test BPF filter functionality" 2231 depends on m && NET 2232 help 2233 This builds the "test_bpf" module that runs various test vectors 2234 against the BPF interpreter or BPF JIT compiler depending on the 2235 current setting. This is in particular useful for BPF JIT compiler 2236 development, but also to run regression tests against changes in 2237 the interpreter code. It also enables test stubs for eBPF maps and 2238 verifier used by user space verifier testsuite. 2239 2240 If unsure, say N. 2241 2242config TEST_BLACKHOLE_DEV 2243 tristate "Test blackhole netdev functionality" 2244 depends on m && NET 2245 help 2246 This builds the "test_blackhole_dev" module that validates the 2247 data path through this blackhole netdev. 2248 2249 If unsure, say N. 2250 2251config FIND_BIT_BENCHMARK 2252 tristate "Test find_bit functions" 2253 help 2254 This builds the "test_find_bit" module that measure find_*_bit() 2255 functions performance. 2256 2257 If unsure, say N. 2258 2259config TEST_FIRMWARE 2260 tristate "Test firmware loading via userspace interface" 2261 depends on FW_LOADER 2262 help 2263 This builds the "test_firmware" module that creates a userspace 2264 interface for testing firmware loading. This can be used to 2265 control the triggering of firmware loading without needing an 2266 actual firmware-using device. The contents can be rechecked by 2267 userspace. 2268 2269 If unsure, say N. 2270 2271config TEST_SYSCTL 2272 tristate "sysctl test driver" 2273 depends on PROC_SYSCTL 2274 help 2275 This builds the "test_sysctl" module. This driver enables to test the 2276 proc sysctl interfaces available to drivers safely without affecting 2277 production knobs which might alter system functionality. 2278 2279 If unsure, say N. 2280 2281config BITFIELD_KUNIT 2282 tristate "KUnit test bitfield functions at runtime" 2283 depends on KUNIT 2284 help 2285 Enable this option to test the bitfield functions at boot. 2286 2287 KUnit tests run during boot and output the results to the debug log 2288 in TAP format (http://testanything.org/). Only useful for kernel devs 2289 running the KUnit test harness, and not intended for inclusion into a 2290 production build. 2291 2292 For more information on KUnit and unit tests in general please refer 2293 to the KUnit documentation in Documentation/dev-tools/kunit/. 2294 2295 If unsure, say N. 2296 2297config RESOURCE_KUNIT_TEST 2298 tristate "KUnit test for resource API" 2299 depends on KUNIT 2300 help 2301 This builds the resource API unit test. 2302 Tests the logic of API provided by resource.c and ioport.h. 2303 For more information on KUnit and unit tests in general please refer 2304 to the KUnit documentation in Documentation/dev-tools/kunit/. 2305 2306 If unsure, say N. 2307 2308config SYSCTL_KUNIT_TEST 2309 tristate "KUnit test for sysctl" if !KUNIT_ALL_TESTS 2310 depends on KUNIT 2311 default KUNIT_ALL_TESTS 2312 help 2313 This builds the proc sysctl unit test, which runs on boot. 2314 Tests the API contract and implementation correctness of sysctl. 2315 For more information on KUnit and unit tests in general please refer 2316 to the KUnit documentation in Documentation/dev-tools/kunit/. 2317 2318 If unsure, say N. 2319 2320config LIST_KUNIT_TEST 2321 tristate "KUnit Test for Kernel Linked-list structures" if !KUNIT_ALL_TESTS 2322 depends on KUNIT 2323 default KUNIT_ALL_TESTS 2324 help 2325 This builds the linked list KUnit test suite. 2326 It tests that the API and basic functionality of the list_head type 2327 and associated macros. 2328 2329 KUnit tests run during boot and output the results to the debug log 2330 in TAP format (https://testanything.org/). Only useful for kernel devs 2331 running the KUnit test harness, and not intended for inclusion into a 2332 production build. 2333 2334 For more information on KUnit and unit tests in general please refer 2335 to the KUnit documentation in Documentation/dev-tools/kunit/. 2336 2337 If unsure, say N. 2338 2339config LINEAR_RANGES_TEST 2340 tristate "KUnit test for linear_ranges" 2341 depends on KUNIT 2342 select LINEAR_RANGES 2343 help 2344 This builds the linear_ranges unit test, which runs on boot. 2345 Tests the linear_ranges logic correctness. 2346 For more information on KUnit and unit tests in general please refer 2347 to the KUnit documentation in Documentation/dev-tools/kunit/. 2348 2349 If unsure, say N. 2350 2351config CMDLINE_KUNIT_TEST 2352 tristate "KUnit test for cmdline API" 2353 depends on KUNIT 2354 help 2355 This builds the cmdline API unit test. 2356 Tests the logic of API provided by cmdline.c. 2357 For more information on KUnit and unit tests in general please refer 2358 to the KUnit documentation in Documentation/dev-tools/kunit/. 2359 2360 If unsure, say N. 2361 2362config BITS_TEST 2363 tristate "KUnit test for bits.h" 2364 depends on KUNIT 2365 help 2366 This builds the bits unit test. 2367 Tests the logic of macros defined in bits.h. 2368 For more information on KUnit and unit tests in general please refer 2369 to the KUnit documentation in Documentation/dev-tools/kunit/. 2370 2371 If unsure, say N. 2372 2373config TEST_UDELAY 2374 tristate "udelay test driver" 2375 help 2376 This builds the "udelay_test" module that helps to make sure 2377 that udelay() is working properly. 2378 2379 If unsure, say N. 2380 2381config TEST_STATIC_KEYS 2382 tristate "Test static keys" 2383 depends on m 2384 help 2385 Test the static key interfaces. 2386 2387 If unsure, say N. 2388 2389config TEST_KMOD 2390 tristate "kmod stress tester" 2391 depends on m 2392 depends on NETDEVICES && NET_CORE && INET # for TUN 2393 depends on BLOCK 2394 select TEST_LKM 2395 select XFS_FS 2396 select TUN 2397 select BTRFS_FS 2398 help 2399 Test the kernel's module loading mechanism: kmod. kmod implements 2400 support to load modules using the Linux kernel's usermode helper. 2401 This test provides a series of tests against kmod. 2402 2403 Although technically you can either build test_kmod as a module or 2404 into the kernel we disallow building it into the kernel since 2405 it stress tests request_module() and this will very likely cause 2406 some issues by taking over precious threads available from other 2407 module load requests, ultimately this could be fatal. 2408 2409 To run tests run: 2410 2411 tools/testing/selftests/kmod/kmod.sh --help 2412 2413 If unsure, say N. 2414 2415config TEST_DEBUG_VIRTUAL 2416 tristate "Test CONFIG_DEBUG_VIRTUAL feature" 2417 depends on DEBUG_VIRTUAL 2418 help 2419 Test the kernel's ability to detect incorrect calls to 2420 virt_to_phys() done against the non-linear part of the 2421 kernel's virtual address map. 2422 2423 If unsure, say N. 2424 2425config TEST_MEMCAT_P 2426 tristate "Test memcat_p() helper function" 2427 help 2428 Test the memcat_p() helper for correctly merging two 2429 pointer arrays together. 2430 2431 If unsure, say N. 2432 2433config TEST_LIVEPATCH 2434 tristate "Test livepatching" 2435 default n 2436 depends on DYNAMIC_DEBUG 2437 depends on LIVEPATCH 2438 depends on m 2439 help 2440 Test kernel livepatching features for correctness. The tests will 2441 load test modules that will be livepatched in various scenarios. 2442 2443 To run all the livepatching tests: 2444 2445 make -C tools/testing/selftests TARGETS=livepatch run_tests 2446 2447 Alternatively, individual tests may be invoked: 2448 2449 tools/testing/selftests/livepatch/test-callbacks.sh 2450 tools/testing/selftests/livepatch/test-livepatch.sh 2451 tools/testing/selftests/livepatch/test-shadow-vars.sh 2452 2453 If unsure, say N. 2454 2455config TEST_OBJAGG 2456 tristate "Perform selftest on object aggreration manager" 2457 default n 2458 depends on OBJAGG 2459 help 2460 Enable this option to test object aggregation manager on boot 2461 (or module load). 2462 2463 2464config TEST_STACKINIT 2465 tristate "Test level of stack variable initialization" 2466 help 2467 Test if the kernel is zero-initializing stack variables and 2468 padding. Coverage is controlled by compiler flags, 2469 CONFIG_GCC_PLUGIN_STRUCTLEAK, CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF, 2470 or CONFIG_GCC_PLUGIN_STRUCTLEAK_BYREF_ALL. 2471 2472 If unsure, say N. 2473 2474config TEST_MEMINIT 2475 tristate "Test heap/page initialization" 2476 help 2477 Test if the kernel is zero-initializing heap and page allocations. 2478 This can be useful to test init_on_alloc and init_on_free features. 2479 2480 If unsure, say N. 2481 2482config TEST_HMM 2483 tristate "Test HMM (Heterogeneous Memory Management)" 2484 depends on TRANSPARENT_HUGEPAGE 2485 depends on DEVICE_PRIVATE 2486 select HMM_MIRROR 2487 select MMU_NOTIFIER 2488 help 2489 This is a pseudo device driver solely for testing HMM. 2490 Say M here if you want to build the HMM test module. 2491 Doing so will allow you to run tools/testing/selftest/vm/hmm-tests. 2492 2493 If unsure, say N. 2494 2495config TEST_FREE_PAGES 2496 tristate "Test freeing pages" 2497 help 2498 Test that a memory leak does not occur due to a race between 2499 freeing a block of pages and a speculative page reference. 2500 Loading this module is safe if your kernel has the bug fixed. 2501 If the bug is not fixed, it will leak gigabytes of memory and 2502 probably OOM your system. 2503 2504config TEST_FPU 2505 tristate "Test floating point operations in kernel space" 2506 depends on X86 && !KCOV_INSTRUMENT_ALL 2507 help 2508 Enable this option to add /sys/kernel/debug/selftest_helpers/test_fpu 2509 which will trigger a sequence of floating point operations. This is used 2510 for self-testing floating point control register setting in 2511 kernel_fpu_begin(). 2512 2513 If unsure, say N. 2514 2515endif # RUNTIME_TESTING_MENU 2516 2517config MEMTEST 2518 bool "Memtest" 2519 help 2520 This option adds a kernel parameter 'memtest', which allows memtest 2521 to be set. 2522 memtest=0, mean disabled; -- default 2523 memtest=1, mean do 1 test pattern; 2524 ... 2525 memtest=17, mean do 17 test patterns. 2526 If you are unsure how to answer this question, answer N. 2527 2528 2529 2530config HYPERV_TESTING 2531 bool "Microsoft Hyper-V driver testing" 2532 default n 2533 depends on HYPERV && DEBUG_FS 2534 help 2535 Select this option to enable Hyper-V vmbus testing. 2536 2537endmenu # "Kernel Testing and Coverage" 2538 2539source "Documentation/Kconfig" 2540 2541endmenu # Kernel hacking 2542