1# 2# General architecture dependent options 3# 4 5config KEXEC_CORE 6 bool 7 8config OPROFILE 9 tristate "OProfile system profiling" 10 depends on PROFILING 11 depends on HAVE_OPROFILE 12 select RING_BUFFER 13 select RING_BUFFER_ALLOW_SWAP 14 help 15 OProfile is a profiling system capable of profiling the 16 whole system, include the kernel, kernel modules, libraries, 17 and applications. 18 19 If unsure, say N. 20 21config OPROFILE_EVENT_MULTIPLEX 22 bool "OProfile multiplexing support (EXPERIMENTAL)" 23 default n 24 depends on OPROFILE && X86 25 help 26 The number of hardware counters is limited. The multiplexing 27 feature enables OProfile to gather more events than counters 28 are provided by the hardware. This is realized by switching 29 between events at an user specified time interval. 30 31 If unsure, say N. 32 33config HAVE_OPROFILE 34 bool 35 36config OPROFILE_NMI_TIMER 37 def_bool y 38 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64 39 40config KPROBES 41 bool "Kprobes" 42 depends on MODULES 43 depends on HAVE_KPROBES 44 select KALLSYMS 45 help 46 Kprobes allows you to trap at almost any kernel address and 47 execute a callback function. register_kprobe() establishes 48 a probepoint and specifies the callback. Kprobes is useful 49 for kernel debugging, non-intrusive instrumentation and testing. 50 If in doubt, say "N". 51 52config JUMP_LABEL 53 bool "Optimize very unlikely/likely branches" 54 depends on HAVE_ARCH_JUMP_LABEL 55 help 56 This option enables a transparent branch optimization that 57 makes certain almost-always-true or almost-always-false branch 58 conditions even cheaper to execute within the kernel. 59 60 Certain performance-sensitive kernel code, such as trace points, 61 scheduler functionality, networking code and KVM have such 62 branches and include support for this optimization technique. 63 64 If it is detected that the compiler has support for "asm goto", 65 the kernel will compile such branches with just a nop 66 instruction. When the condition flag is toggled to true, the 67 nop will be converted to a jump instruction to execute the 68 conditional block of instructions. 69 70 This technique lowers overhead and stress on the branch prediction 71 of the processor and generally makes the kernel faster. The update 72 of the condition is slower, but those are always very rare. 73 74 ( On 32-bit x86, the necessary options added to the compiler 75 flags may increase the size of the kernel slightly. ) 76 77config STATIC_KEYS_SELFTEST 78 bool "Static key selftest" 79 depends on JUMP_LABEL 80 help 81 Boot time self-test of the branch patching code. 82 83config OPTPROBES 84 def_bool y 85 depends on KPROBES && HAVE_OPTPROBES 86 depends on !PREEMPT 87 88config KPROBES_ON_FTRACE 89 def_bool y 90 depends on KPROBES && HAVE_KPROBES_ON_FTRACE 91 depends on DYNAMIC_FTRACE_WITH_REGS 92 help 93 If function tracer is enabled and the arch supports full 94 passing of pt_regs to function tracing, then kprobes can 95 optimize on top of function tracing. 96 97config UPROBES 98 def_bool n 99 help 100 Uprobes is the user-space counterpart to kprobes: they 101 enable instrumentation applications (such as 'perf probe') 102 to establish unintrusive probes in user-space binaries and 103 libraries, by executing handler functions when the probes 104 are hit by user-space applications. 105 106 ( These probes come in the form of single-byte breakpoints, 107 managed by the kernel and kept transparent to the probed 108 application. ) 109 110config HAVE_64BIT_ALIGNED_ACCESS 111 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 112 help 113 Some architectures require 64 bit accesses to be 64 bit 114 aligned, which also requires structs containing 64 bit values 115 to be 64 bit aligned too. This includes some 32 bit 116 architectures which can do 64 bit accesses, as well as 64 bit 117 architectures without unaligned access. 118 119 This symbol should be selected by an architecture if 64 bit 120 accesses are required to be 64 bit aligned in this way even 121 though it is not a 64 bit architecture. 122 123 See Documentation/unaligned-memory-access.txt for more 124 information on the topic of unaligned memory accesses. 125 126config HAVE_EFFICIENT_UNALIGNED_ACCESS 127 bool 128 help 129 Some architectures are unable to perform unaligned accesses 130 without the use of get_unaligned/put_unaligned. Others are 131 unable to perform such accesses efficiently (e.g. trap on 132 unaligned access and require fixing it up in the exception 133 handler.) 134 135 This symbol should be selected by an architecture if it can 136 perform unaligned accesses efficiently to allow different 137 code paths to be selected for these cases. Some network 138 drivers, for example, could opt to not fix up alignment 139 problems with received packets if doing so would not help 140 much. 141 142 See Documentation/unaligned-memory-access.txt for more 143 information on the topic of unaligned memory accesses. 144 145config ARCH_USE_BUILTIN_BSWAP 146 bool 147 help 148 Modern versions of GCC (since 4.4) have builtin functions 149 for handling byte-swapping. Using these, instead of the old 150 inline assembler that the architecture code provides in the 151 __arch_bswapXX() macros, allows the compiler to see what's 152 happening and offers more opportunity for optimisation. In 153 particular, the compiler will be able to combine the byteswap 154 with a nearby load or store and use load-and-swap or 155 store-and-swap instructions if the architecture has them. It 156 should almost *never* result in code which is worse than the 157 hand-coded assembler in <asm/swab.h>. But just in case it 158 does, the use of the builtins is optional. 159 160 Any architecture with load-and-swap or store-and-swap 161 instructions should set this. And it shouldn't hurt to set it 162 on architectures that don't have such instructions. 163 164config KRETPROBES 165 def_bool y 166 depends on KPROBES && HAVE_KRETPROBES 167 168config USER_RETURN_NOTIFIER 169 bool 170 depends on HAVE_USER_RETURN_NOTIFIER 171 help 172 Provide a kernel-internal notification when a cpu is about to 173 switch to user mode. 174 175config HAVE_IOREMAP_PROT 176 bool 177 178config HAVE_KPROBES 179 bool 180 181config HAVE_KRETPROBES 182 bool 183 184config HAVE_OPTPROBES 185 bool 186 187config HAVE_KPROBES_ON_FTRACE 188 bool 189 190config HAVE_NMI 191 bool 192 193config HAVE_NMI_WATCHDOG 194 depends on HAVE_NMI 195 bool 196# 197# An arch should select this if it provides all these things: 198# 199# task_pt_regs() in asm/processor.h or asm/ptrace.h 200# arch_has_single_step() if there is hardware single-step support 201# arch_has_block_step() if there is hardware block-step support 202# asm/syscall.h supplying asm-generic/syscall.h interface 203# linux/regset.h user_regset interfaces 204# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 205# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 206# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 207# signal delivery calls tracehook_signal_handler() 208# 209config HAVE_ARCH_TRACEHOOK 210 bool 211 212config HAVE_DMA_CONTIGUOUS 213 bool 214 215config GENERIC_SMP_IDLE_THREAD 216 bool 217 218config GENERIC_IDLE_POLL_SETUP 219 bool 220 221# Select if arch init_task initializer is different to init/init_task.c 222config ARCH_INIT_TASK 223 bool 224 225# Select if arch has its private alloc_task_struct() function 226config ARCH_TASK_STRUCT_ALLOCATOR 227 bool 228 229# Select if arch has its private alloc_thread_stack() function 230config ARCH_THREAD_STACK_ALLOCATOR 231 bool 232 233# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 234config ARCH_WANTS_DYNAMIC_TASK_STRUCT 235 bool 236 237config HAVE_REGS_AND_STACK_ACCESS_API 238 bool 239 help 240 This symbol should be selected by an architecure if it supports 241 the API needed to access registers and stack entries from pt_regs, 242 declared in asm/ptrace.h 243 For example the kprobes-based event tracer needs this API. 244 245config HAVE_CLK 246 bool 247 help 248 The <linux/clk.h> calls support software clock gating and 249 thus are a key power management tool on many systems. 250 251config HAVE_DMA_API_DEBUG 252 bool 253 254config HAVE_HW_BREAKPOINT 255 bool 256 depends on PERF_EVENTS 257 258config HAVE_MIXED_BREAKPOINTS_REGS 259 bool 260 depends on HAVE_HW_BREAKPOINT 261 help 262 Depending on the arch implementation of hardware breakpoints, 263 some of them have separate registers for data and instruction 264 breakpoints addresses, others have mixed registers to store 265 them but define the access type in a control register. 266 Select this option if your arch implements breakpoints under the 267 latter fashion. 268 269config HAVE_USER_RETURN_NOTIFIER 270 bool 271 272config HAVE_PERF_EVENTS_NMI 273 bool 274 help 275 System hardware can generate an NMI using the perf event 276 subsystem. Also has support for calculating CPU cycle events 277 to determine how many clock cycles in a given period. 278 279config HAVE_PERF_REGS 280 bool 281 help 282 Support selective register dumps for perf events. This includes 283 bit-mapping of each registers and a unique architecture id. 284 285config HAVE_PERF_USER_STACK_DUMP 286 bool 287 help 288 Support user stack dumps for perf event samples. This needs 289 access to the user stack pointer which is not unified across 290 architectures. 291 292config HAVE_ARCH_JUMP_LABEL 293 bool 294 295config HAVE_RCU_TABLE_FREE 296 bool 297 298config ARCH_HAVE_NMI_SAFE_CMPXCHG 299 bool 300 301config HAVE_ALIGNED_STRUCT_PAGE 302 bool 303 help 304 This makes sure that struct pages are double word aligned and that 305 e.g. the SLUB allocator can perform double word atomic operations 306 on a struct page for better performance. However selecting this 307 might increase the size of a struct page by a word. 308 309config HAVE_CMPXCHG_LOCAL 310 bool 311 312config HAVE_CMPXCHG_DOUBLE 313 bool 314 315config ARCH_WANT_IPC_PARSE_VERSION 316 bool 317 318config ARCH_WANT_COMPAT_IPC_PARSE_VERSION 319 bool 320 321config ARCH_WANT_OLD_COMPAT_IPC 322 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 323 bool 324 325config HAVE_ARCH_SECCOMP_FILTER 326 bool 327 help 328 An arch should select this symbol if it provides all of these things: 329 - syscall_get_arch() 330 - syscall_get_arguments() 331 - syscall_rollback() 332 - syscall_set_return_value() 333 - SIGSYS siginfo_t support 334 - secure_computing is called from a ptrace_event()-safe context 335 - secure_computing return value is checked and a return value of -1 336 results in the system call being skipped immediately. 337 - seccomp syscall wired up 338 339config SECCOMP_FILTER 340 def_bool y 341 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 342 help 343 Enable tasks to build secure computing environments defined 344 in terms of Berkeley Packet Filter programs which implement 345 task-defined system call filtering polices. 346 347 See Documentation/prctl/seccomp_filter.txt for details. 348 349config HAVE_GCC_PLUGINS 350 bool 351 help 352 An arch should select this symbol if it supports building with 353 GCC plugins. 354 355menuconfig GCC_PLUGINS 356 bool "GCC plugins" 357 depends on HAVE_GCC_PLUGINS 358 depends on !COMPILE_TEST 359 help 360 GCC plugins are loadable modules that provide extra features to the 361 compiler. They are useful for runtime instrumentation and static analysis. 362 363 See Documentation/gcc-plugins.txt for details. 364 365config GCC_PLUGIN_CYC_COMPLEXITY 366 bool "Compute the cyclomatic complexity of a function" 367 depends on GCC_PLUGINS 368 help 369 The complexity M of a function's control flow graph is defined as: 370 M = E - N + 2P 371 where 372 373 E = the number of edges 374 N = the number of nodes 375 P = the number of connected components (exit nodes). 376 377config GCC_PLUGIN_SANCOV 378 bool 379 depends on GCC_PLUGINS 380 help 381 This plugin inserts a __sanitizer_cov_trace_pc() call at the start of 382 basic blocks. It supports all gcc versions with plugin support (from 383 gcc-4.5 on). It is based on the commit "Add fuzzing coverage support" 384 by Dmitry Vyukov <dvyukov@google.com>. 385 386config GCC_PLUGIN_LATENT_ENTROPY 387 bool "Generate some entropy during boot and runtime" 388 depends on GCC_PLUGINS 389 help 390 By saying Y here the kernel will instrument some kernel code to 391 extract some entropy from both original and artificially created 392 program state. This will help especially embedded systems where 393 there is little 'natural' source of entropy normally. The cost 394 is some slowdown of the boot process (about 0.5%) and fork and 395 irq processing. 396 397 Note that entropy extracted this way is not cryptographically 398 secure! 399 400 This plugin was ported from grsecurity/PaX. More information at: 401 * https://grsecurity.net/ 402 * https://pax.grsecurity.net/ 403 404config HAVE_CC_STACKPROTECTOR 405 bool 406 help 407 An arch should select this symbol if: 408 - its compiler supports the -fstack-protector option 409 - it has implemented a stack canary (e.g. __stack_chk_guard) 410 411config CC_STACKPROTECTOR 412 def_bool n 413 help 414 Set when a stack-protector mode is enabled, so that the build 415 can enable kernel-side support for the GCC feature. 416 417choice 418 prompt "Stack Protector buffer overflow detection" 419 depends on HAVE_CC_STACKPROTECTOR 420 default CC_STACKPROTECTOR_NONE 421 help 422 This option turns on the "stack-protector" GCC feature. This 423 feature puts, at the beginning of functions, a canary value on 424 the stack just before the return address, and validates 425 the value just before actually returning. Stack based buffer 426 overflows (that need to overwrite this return address) now also 427 overwrite the canary, which gets detected and the attack is then 428 neutralized via a kernel panic. 429 430config CC_STACKPROTECTOR_NONE 431 bool "None" 432 help 433 Disable "stack-protector" GCC feature. 434 435config CC_STACKPROTECTOR_REGULAR 436 bool "Regular" 437 select CC_STACKPROTECTOR 438 help 439 Functions will have the stack-protector canary logic added if they 440 have an 8-byte or larger character array on the stack. 441 442 This feature requires gcc version 4.2 or above, or a distribution 443 gcc with the feature backported ("-fstack-protector"). 444 445 On an x86 "defconfig" build, this feature adds canary checks to 446 about 3% of all kernel functions, which increases kernel code size 447 by about 0.3%. 448 449config CC_STACKPROTECTOR_STRONG 450 bool "Strong" 451 select CC_STACKPROTECTOR 452 help 453 Functions will have the stack-protector canary logic added in any 454 of the following conditions: 455 456 - local variable's address used as part of the right hand side of an 457 assignment or function argument 458 - local variable is an array (or union containing an array), 459 regardless of array type or length 460 - uses register local variables 461 462 This feature requires gcc version 4.9 or above, or a distribution 463 gcc with the feature backported ("-fstack-protector-strong"). 464 465 On an x86 "defconfig" build, this feature adds canary checks to 466 about 20% of all kernel functions, which increases the kernel code 467 size by about 2%. 468 469endchoice 470 471config THIN_ARCHIVES 472 bool 473 help 474 Select this if the architecture wants to use thin archives 475 instead of ld -r to create the built-in.o files. 476 477config LD_DEAD_CODE_DATA_ELIMINATION 478 bool 479 help 480 Select this if the architecture wants to do dead code and 481 data elimination with the linker by compiling with 482 -ffunction-sections -fdata-sections and linking with 483 --gc-sections. 484 485 This requires that the arch annotates or otherwise protects 486 its external entry points from being discarded. Linker scripts 487 must also merge .text.*, .data.*, and .bss.* correctly into 488 output sections. Care must be taken not to pull in unrelated 489 sections (e.g., '.text.init'). Typically '.' in section names 490 is used to distinguish them from label names / C identifiers. 491 492config HAVE_ARCH_WITHIN_STACK_FRAMES 493 bool 494 help 495 An architecture should select this if it can walk the kernel stack 496 frames to determine if an object is part of either the arguments 497 or local variables (i.e. that it excludes saved return addresses, 498 and similar) by implementing an inline arch_within_stack_frames(), 499 which is used by CONFIG_HARDENED_USERCOPY. 500 501config HAVE_CONTEXT_TRACKING 502 bool 503 help 504 Provide kernel/user boundaries probes necessary for subsystems 505 that need it, such as userspace RCU extended quiescent state. 506 Syscalls need to be wrapped inside user_exit()-user_enter() through 507 the slow path using TIF_NOHZ flag. Exceptions handlers must be 508 wrapped as well. Irqs are already protected inside 509 rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on 510 irq exit still need to be protected. 511 512config HAVE_VIRT_CPU_ACCOUNTING 513 bool 514 515config HAVE_VIRT_CPU_ACCOUNTING_GEN 516 bool 517 default y if 64BIT 518 help 519 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 520 Before enabling this option, arch code must be audited 521 to ensure there are no races in concurrent read/write of 522 cputime_t. For example, reading/writing 64-bit cputime_t on 523 some 32-bit arches may require multiple accesses, so proper 524 locking is needed to protect against concurrent accesses. 525 526 527config HAVE_IRQ_TIME_ACCOUNTING 528 bool 529 help 530 Archs need to ensure they use a high enough resolution clock to 531 support irq time accounting and then call enable_sched_clock_irqtime(). 532 533config HAVE_ARCH_TRANSPARENT_HUGEPAGE 534 bool 535 536config HAVE_ARCH_HUGE_VMAP 537 bool 538 539config HAVE_ARCH_SOFT_DIRTY 540 bool 541 542config HAVE_MOD_ARCH_SPECIFIC 543 bool 544 help 545 The arch uses struct mod_arch_specific to store data. Many arches 546 just need a simple module loader without arch specific data - those 547 should not enable this. 548 549config MODULES_USE_ELF_RELA 550 bool 551 help 552 Modules only use ELF RELA relocations. Modules with ELF REL 553 relocations will give an error. 554 555config MODULES_USE_ELF_REL 556 bool 557 help 558 Modules only use ELF REL relocations. Modules with ELF RELA 559 relocations will give an error. 560 561config HAVE_UNDERSCORE_SYMBOL_PREFIX 562 bool 563 help 564 Some architectures generate an _ in front of C symbols; things like 565 module loading and assembly files need to know about this. 566 567config HAVE_IRQ_EXIT_ON_IRQ_STACK 568 bool 569 help 570 Architecture doesn't only execute the irq handler on the irq stack 571 but also irq_exit(). This way we can process softirqs on this irq 572 stack instead of switching to a new one when we call __do_softirq() 573 in the end of an hardirq. 574 This spares a stack switch and improves cache usage on softirq 575 processing. 576 577config PGTABLE_LEVELS 578 int 579 default 2 580 581config ARCH_HAS_ELF_RANDOMIZE 582 bool 583 help 584 An architecture supports choosing randomized locations for 585 stack, mmap, brk, and ET_DYN. Defined functions: 586 - arch_mmap_rnd() 587 - arch_randomize_brk() 588 589config HAVE_ARCH_MMAP_RND_BITS 590 bool 591 help 592 An arch should select this symbol if it supports setting a variable 593 number of bits for use in establishing the base address for mmap 594 allocations, has MMU enabled and provides values for both: 595 - ARCH_MMAP_RND_BITS_MIN 596 - ARCH_MMAP_RND_BITS_MAX 597 598config HAVE_EXIT_THREAD 599 bool 600 help 601 An architecture implements exit_thread. 602 603config ARCH_MMAP_RND_BITS_MIN 604 int 605 606config ARCH_MMAP_RND_BITS_MAX 607 int 608 609config ARCH_MMAP_RND_BITS_DEFAULT 610 int 611 612config ARCH_MMAP_RND_BITS 613 int "Number of bits to use for ASLR of mmap base address" if EXPERT 614 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 615 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 616 default ARCH_MMAP_RND_BITS_MIN 617 depends on HAVE_ARCH_MMAP_RND_BITS 618 help 619 This value can be used to select the number of bits to use to 620 determine the random offset to the base address of vma regions 621 resulting from mmap allocations. This value will be bounded 622 by the architecture's minimum and maximum supported values. 623 624 This value can be changed after boot using the 625 /proc/sys/vm/mmap_rnd_bits tunable 626 627config HAVE_ARCH_MMAP_RND_COMPAT_BITS 628 bool 629 help 630 An arch should select this symbol if it supports running applications 631 in compatibility mode, supports setting a variable number of bits for 632 use in establishing the base address for mmap allocations, has MMU 633 enabled and provides values for both: 634 - ARCH_MMAP_RND_COMPAT_BITS_MIN 635 - ARCH_MMAP_RND_COMPAT_BITS_MAX 636 637config ARCH_MMAP_RND_COMPAT_BITS_MIN 638 int 639 640config ARCH_MMAP_RND_COMPAT_BITS_MAX 641 int 642 643config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 644 int 645 646config ARCH_MMAP_RND_COMPAT_BITS 647 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 648 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 649 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 650 default ARCH_MMAP_RND_COMPAT_BITS_MIN 651 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 652 help 653 This value can be used to select the number of bits to use to 654 determine the random offset to the base address of vma regions 655 resulting from mmap allocations for compatible applications This 656 value will be bounded by the architecture's minimum and maximum 657 supported values. 658 659 This value can be changed after boot using the 660 /proc/sys/vm/mmap_rnd_compat_bits tunable 661 662config HAVE_COPY_THREAD_TLS 663 bool 664 help 665 Architecture provides copy_thread_tls to accept tls argument via 666 normal C parameter passing, rather than extracting the syscall 667 argument from pt_regs. 668 669config HAVE_STACK_VALIDATION 670 bool 671 help 672 Architecture supports the 'objtool check' host tool command, which 673 performs compile-time stack metadata validation. 674 675config HAVE_ARCH_HASH 676 bool 677 default n 678 help 679 If this is set, the architecture provides an <asm/hash.h> 680 file which provides platform-specific implementations of some 681 functions in <linux/hash.h> or fs/namei.c. 682 683config ISA_BUS_API 684 def_bool ISA 685 686# 687# ABI hall of shame 688# 689config CLONE_BACKWARDS 690 bool 691 help 692 Architecture has tls passed as the 4th argument of clone(2), 693 not the 5th one. 694 695config CLONE_BACKWARDS2 696 bool 697 help 698 Architecture has the first two arguments of clone(2) swapped. 699 700config CLONE_BACKWARDS3 701 bool 702 help 703 Architecture has tls passed as the 3rd argument of clone(2), 704 not the 5th one. 705 706config ODD_RT_SIGACTION 707 bool 708 help 709 Architecture has unusual rt_sigaction(2) arguments 710 711config OLD_SIGSUSPEND 712 bool 713 help 714 Architecture has old sigsuspend(2) syscall, of one-argument variety 715 716config OLD_SIGSUSPEND3 717 bool 718 help 719 Even weirder antique ABI - three-argument sigsuspend(2) 720 721config OLD_SIGACTION 722 bool 723 help 724 Architecture has old sigaction(2) syscall. Nope, not the same 725 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 726 but fairly different variant of sigaction(2), thanks to OSF/1 727 compatibility... 728 729config COMPAT_OLD_SIGACTION 730 bool 731 732config ARCH_NO_COHERENT_DMA_MMAP 733 bool 734 735config CPU_NO_EFFICIENT_FFS 736 def_bool n 737 738config HAVE_ARCH_VMAP_STACK 739 def_bool n 740 help 741 An arch should select this symbol if it can support kernel stacks 742 in vmalloc space. This means: 743 744 - vmalloc space must be large enough to hold many kernel stacks. 745 This may rule out many 32-bit architectures. 746 747 - Stacks in vmalloc space need to work reliably. For example, if 748 vmap page tables are created on demand, either this mechanism 749 needs to work while the stack points to a virtual address with 750 unpopulated page tables or arch code (switch_to() and switch_mm(), 751 most likely) needs to ensure that the stack's page table entries 752 are populated before running on a possibly unpopulated stack. 753 754 - If the stack overflows into a guard page, something reasonable 755 should happen. The definition of "reasonable" is flexible, but 756 instantly rebooting without logging anything would be unfriendly. 757 758config VMAP_STACK 759 default y 760 bool "Use a virtually-mapped stack" 761 depends on HAVE_ARCH_VMAP_STACK && !KASAN 762 ---help--- 763 Enable this if you want the use virtually-mapped kernel stacks 764 with guard pages. This causes kernel stack overflows to be 765 caught immediately rather than causing difficult-to-diagnose 766 corruption. 767 768 This is presently incompatible with KASAN because KASAN expects 769 the stack to map directly to the KASAN shadow map using a formula 770 that is incorrect if the stack is in vmalloc space. 771 772source "kernel/gcov/Kconfig" 773