1# SPDX-License-Identifier: GPL-2.0 2# 3# General architecture dependent options 4# 5 6# 7# Note: arch/$(SRCARCH)/Kconfig needs to be included first so that it can 8# override the default values in this file. 9# 10source "arch/$(SRCARCH)/Kconfig" 11 12menu "General architecture-dependent options" 13 14config CRASH_CORE 15 bool 16 17config KEXEC_CORE 18 select CRASH_CORE 19 bool 20 21config KEXEC_ELF 22 bool 23 24config HAVE_IMA_KEXEC 25 bool 26 27config SET_FS 28 bool 29 30config HOTPLUG_SMT 31 bool 32 33config GENERIC_ENTRY 34 bool 35 36config OPROFILE 37 tristate "OProfile system profiling" 38 depends on PROFILING 39 depends on HAVE_OPROFILE 40 select RING_BUFFER 41 select RING_BUFFER_ALLOW_SWAP 42 help 43 OProfile is a profiling system capable of profiling the 44 whole system, include the kernel, kernel modules, libraries, 45 and applications. 46 47 If unsure, say N. 48 49config OPROFILE_EVENT_MULTIPLEX 50 bool "OProfile multiplexing support (EXPERIMENTAL)" 51 default n 52 depends on OPROFILE && X86 53 help 54 The number of hardware counters is limited. The multiplexing 55 feature enables OProfile to gather more events than counters 56 are provided by the hardware. This is realized by switching 57 between events at a user specified time interval. 58 59 If unsure, say N. 60 61config HAVE_OPROFILE 62 bool 63 64config OPROFILE_NMI_TIMER 65 def_bool y 66 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64 67 68config KPROBES 69 bool "Kprobes" 70 depends on MODULES 71 depends on HAVE_KPROBES 72 select KALLSYMS 73 help 74 Kprobes allows you to trap at almost any kernel address and 75 execute a callback function. register_kprobe() establishes 76 a probepoint and specifies the callback. Kprobes is useful 77 for kernel debugging, non-intrusive instrumentation and testing. 78 If in doubt, say "N". 79 80config JUMP_LABEL 81 bool "Optimize very unlikely/likely branches" 82 depends on HAVE_ARCH_JUMP_LABEL 83 depends on CC_HAS_ASM_GOTO 84 help 85 This option enables a transparent branch optimization that 86 makes certain almost-always-true or almost-always-false branch 87 conditions even cheaper to execute within the kernel. 88 89 Certain performance-sensitive kernel code, such as trace points, 90 scheduler functionality, networking code and KVM have such 91 branches and include support for this optimization technique. 92 93 If it is detected that the compiler has support for "asm goto", 94 the kernel will compile such branches with just a nop 95 instruction. When the condition flag is toggled to true, the 96 nop will be converted to a jump instruction to execute the 97 conditional block of instructions. 98 99 This technique lowers overhead and stress on the branch prediction 100 of the processor and generally makes the kernel faster. The update 101 of the condition is slower, but those are always very rare. 102 103 ( On 32-bit x86, the necessary options added to the compiler 104 flags may increase the size of the kernel slightly. ) 105 106config STATIC_KEYS_SELFTEST 107 bool "Static key selftest" 108 depends on JUMP_LABEL 109 help 110 Boot time self-test of the branch patching code. 111 112config STATIC_CALL_SELFTEST 113 bool "Static call selftest" 114 depends on HAVE_STATIC_CALL 115 help 116 Boot time self-test of the call patching code. 117 118config OPTPROBES 119 def_bool y 120 depends on KPROBES && HAVE_OPTPROBES 121 select TASKS_RCU if PREEMPTION 122 123config KPROBES_ON_FTRACE 124 def_bool y 125 depends on KPROBES && HAVE_KPROBES_ON_FTRACE 126 depends on DYNAMIC_FTRACE_WITH_REGS 127 help 128 If function tracer is enabled and the arch supports full 129 passing of pt_regs to function tracing, then kprobes can 130 optimize on top of function tracing. 131 132config UPROBES 133 def_bool n 134 depends on ARCH_SUPPORTS_UPROBES 135 help 136 Uprobes is the user-space counterpart to kprobes: they 137 enable instrumentation applications (such as 'perf probe') 138 to establish unintrusive probes in user-space binaries and 139 libraries, by executing handler functions when the probes 140 are hit by user-space applications. 141 142 ( These probes come in the form of single-byte breakpoints, 143 managed by the kernel and kept transparent to the probed 144 application. ) 145 146config HAVE_64BIT_ALIGNED_ACCESS 147 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 148 help 149 Some architectures require 64 bit accesses to be 64 bit 150 aligned, which also requires structs containing 64 bit values 151 to be 64 bit aligned too. This includes some 32 bit 152 architectures which can do 64 bit accesses, as well as 64 bit 153 architectures without unaligned access. 154 155 This symbol should be selected by an architecture if 64 bit 156 accesses are required to be 64 bit aligned in this way even 157 though it is not a 64 bit architecture. 158 159 See Documentation/unaligned-memory-access.txt for more 160 information on the topic of unaligned memory accesses. 161 162config HAVE_EFFICIENT_UNALIGNED_ACCESS 163 bool 164 help 165 Some architectures are unable to perform unaligned accesses 166 without the use of get_unaligned/put_unaligned. Others are 167 unable to perform such accesses efficiently (e.g. trap on 168 unaligned access and require fixing it up in the exception 169 handler.) 170 171 This symbol should be selected by an architecture if it can 172 perform unaligned accesses efficiently to allow different 173 code paths to be selected for these cases. Some network 174 drivers, for example, could opt to not fix up alignment 175 problems with received packets if doing so would not help 176 much. 177 178 See Documentation/core-api/unaligned-memory-access.rst for more 179 information on the topic of unaligned memory accesses. 180 181config ARCH_USE_BUILTIN_BSWAP 182 bool 183 help 184 Modern versions of GCC (since 4.4) have builtin functions 185 for handling byte-swapping. Using these, instead of the old 186 inline assembler that the architecture code provides in the 187 __arch_bswapXX() macros, allows the compiler to see what's 188 happening and offers more opportunity for optimisation. In 189 particular, the compiler will be able to combine the byteswap 190 with a nearby load or store and use load-and-swap or 191 store-and-swap instructions if the architecture has them. It 192 should almost *never* result in code which is worse than the 193 hand-coded assembler in <asm/swab.h>. But just in case it 194 does, the use of the builtins is optional. 195 196 Any architecture with load-and-swap or store-and-swap 197 instructions should set this. And it shouldn't hurt to set it 198 on architectures that don't have such instructions. 199 200config KRETPROBES 201 def_bool y 202 depends on KPROBES && HAVE_KRETPROBES 203 204config USER_RETURN_NOTIFIER 205 bool 206 depends on HAVE_USER_RETURN_NOTIFIER 207 help 208 Provide a kernel-internal notification when a cpu is about to 209 switch to user mode. 210 211config HAVE_IOREMAP_PROT 212 bool 213 214config HAVE_KPROBES 215 bool 216 217config HAVE_KRETPROBES 218 bool 219 220config HAVE_OPTPROBES 221 bool 222 223config HAVE_KPROBES_ON_FTRACE 224 bool 225 226config HAVE_FUNCTION_ERROR_INJECTION 227 bool 228 229config HAVE_NMI 230 bool 231 232# 233# An arch should select this if it provides all these things: 234# 235# task_pt_regs() in asm/processor.h or asm/ptrace.h 236# arch_has_single_step() if there is hardware single-step support 237# arch_has_block_step() if there is hardware block-step support 238# asm/syscall.h supplying asm-generic/syscall.h interface 239# linux/regset.h user_regset interfaces 240# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 241# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 242# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 243# signal delivery calls tracehook_signal_handler() 244# 245config HAVE_ARCH_TRACEHOOK 246 bool 247 248config HAVE_DMA_CONTIGUOUS 249 bool 250 251config GENERIC_SMP_IDLE_THREAD 252 bool 253 254config GENERIC_IDLE_POLL_SETUP 255 bool 256 257config ARCH_HAS_FORTIFY_SOURCE 258 bool 259 help 260 An architecture should select this when it can successfully 261 build and run with CONFIG_FORTIFY_SOURCE. 262 263# 264# Select if the arch provides a historic keepinit alias for the retain_initrd 265# command line option 266# 267config ARCH_HAS_KEEPINITRD 268 bool 269 270# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 271config ARCH_HAS_SET_MEMORY 272 bool 273 274# Select if arch has all set_direct_map_invalid/default() functions 275config ARCH_HAS_SET_DIRECT_MAP 276 bool 277 278# 279# Select if the architecture provides the arch_dma_set_uncached symbol to 280# either provide an uncached segement alias for a DMA allocation, or 281# to remap the page tables in place. 282# 283config ARCH_HAS_DMA_SET_UNCACHED 284 bool 285 286# 287# Select if the architectures provides the arch_dma_clear_uncached symbol 288# to undo an in-place page table remap for uncached access. 289# 290config ARCH_HAS_DMA_CLEAR_UNCACHED 291 bool 292 293# Select if arch init_task must go in the __init_task_data section 294config ARCH_TASK_STRUCT_ON_STACK 295 bool 296 297# Select if arch has its private alloc_task_struct() function 298config ARCH_TASK_STRUCT_ALLOCATOR 299 bool 300 301config HAVE_ARCH_THREAD_STRUCT_WHITELIST 302 bool 303 depends on !ARCH_TASK_STRUCT_ALLOCATOR 304 help 305 An architecture should select this to provide hardened usercopy 306 knowledge about what region of the thread_struct should be 307 whitelisted for copying to userspace. Normally this is only the 308 FPU registers. Specifically, arch_thread_struct_whitelist() 309 should be implemented. Without this, the entire thread_struct 310 field in task_struct will be left whitelisted. 311 312# Select if arch has its private alloc_thread_stack() function 313config ARCH_THREAD_STACK_ALLOCATOR 314 bool 315 316# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 317config ARCH_WANTS_DYNAMIC_TASK_STRUCT 318 bool 319 320config ARCH_32BIT_OFF_T 321 bool 322 depends on !64BIT 323 help 324 All new 32-bit architectures should have 64-bit off_t type on 325 userspace side which corresponds to the loff_t kernel type. This 326 is the requirement for modern ABIs. Some existing architectures 327 still support 32-bit off_t. This option is enabled for all such 328 architectures explicitly. 329 330config HAVE_ASM_MODVERSIONS 331 bool 332 help 333 This symbol should be selected by an architecure if it provides 334 <asm/asm-prototypes.h> to support the module versioning for symbols 335 exported from assembly code. 336 337config HAVE_REGS_AND_STACK_ACCESS_API 338 bool 339 help 340 This symbol should be selected by an architecure if it supports 341 the API needed to access registers and stack entries from pt_regs, 342 declared in asm/ptrace.h 343 For example the kprobes-based event tracer needs this API. 344 345config HAVE_RSEQ 346 bool 347 depends on HAVE_REGS_AND_STACK_ACCESS_API 348 help 349 This symbol should be selected by an architecture if it 350 supports an implementation of restartable sequences. 351 352config HAVE_FUNCTION_ARG_ACCESS_API 353 bool 354 help 355 This symbol should be selected by an architecure if it supports 356 the API needed to access function arguments from pt_regs, 357 declared in asm/ptrace.h 358 359config HAVE_HW_BREAKPOINT 360 bool 361 depends on PERF_EVENTS 362 363config HAVE_MIXED_BREAKPOINTS_REGS 364 bool 365 depends on HAVE_HW_BREAKPOINT 366 help 367 Depending on the arch implementation of hardware breakpoints, 368 some of them have separate registers for data and instruction 369 breakpoints addresses, others have mixed registers to store 370 them but define the access type in a control register. 371 Select this option if your arch implements breakpoints under the 372 latter fashion. 373 374config HAVE_USER_RETURN_NOTIFIER 375 bool 376 377config HAVE_PERF_EVENTS_NMI 378 bool 379 help 380 System hardware can generate an NMI using the perf event 381 subsystem. Also has support for calculating CPU cycle events 382 to determine how many clock cycles in a given period. 383 384config HAVE_HARDLOCKUP_DETECTOR_PERF 385 bool 386 depends on HAVE_PERF_EVENTS_NMI 387 help 388 The arch chooses to use the generic perf-NMI-based hardlockup 389 detector. Must define HAVE_PERF_EVENTS_NMI. 390 391config HAVE_NMI_WATCHDOG 392 depends on HAVE_NMI 393 bool 394 help 395 The arch provides a low level NMI watchdog. It provides 396 asm/nmi.h, and defines its own arch_touch_nmi_watchdog(). 397 398config HAVE_HARDLOCKUP_DETECTOR_ARCH 399 bool 400 select HAVE_NMI_WATCHDOG 401 help 402 The arch chooses to provide its own hardlockup detector, which is 403 a superset of the HAVE_NMI_WATCHDOG. It also conforms to config 404 interfaces and parameters provided by hardlockup detector subsystem. 405 406config HAVE_PERF_REGS 407 bool 408 help 409 Support selective register dumps for perf events. This includes 410 bit-mapping of each registers and a unique architecture id. 411 412config HAVE_PERF_USER_STACK_DUMP 413 bool 414 help 415 Support user stack dumps for perf event samples. This needs 416 access to the user stack pointer which is not unified across 417 architectures. 418 419config HAVE_ARCH_JUMP_LABEL 420 bool 421 422config HAVE_ARCH_JUMP_LABEL_RELATIVE 423 bool 424 425config MMU_GATHER_TABLE_FREE 426 bool 427 428config MMU_GATHER_RCU_TABLE_FREE 429 bool 430 select MMU_GATHER_TABLE_FREE 431 432config MMU_GATHER_PAGE_SIZE 433 bool 434 435config MMU_GATHER_NO_RANGE 436 bool 437 438config MMU_GATHER_NO_GATHER 439 bool 440 depends on MMU_GATHER_TABLE_FREE 441 442config ARCH_WANT_IRQS_OFF_ACTIVATE_MM 443 bool 444 help 445 Temporary select until all architectures can be converted to have 446 irqs disabled over activate_mm. Architectures that do IPI based TLB 447 shootdowns should enable this. 448 449config ARCH_HAVE_NMI_SAFE_CMPXCHG 450 bool 451 452config HAVE_ALIGNED_STRUCT_PAGE 453 bool 454 help 455 This makes sure that struct pages are double word aligned and that 456 e.g. the SLUB allocator can perform double word atomic operations 457 on a struct page for better performance. However selecting this 458 might increase the size of a struct page by a word. 459 460config HAVE_CMPXCHG_LOCAL 461 bool 462 463config HAVE_CMPXCHG_DOUBLE 464 bool 465 466config ARCH_WEAK_RELEASE_ACQUIRE 467 bool 468 469config ARCH_WANT_IPC_PARSE_VERSION 470 bool 471 472config ARCH_WANT_COMPAT_IPC_PARSE_VERSION 473 bool 474 475config ARCH_WANT_OLD_COMPAT_IPC 476 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 477 bool 478 479config HAVE_ARCH_SECCOMP 480 bool 481 help 482 An arch should select this symbol to support seccomp mode 1 (the fixed 483 syscall policy), and must provide an overrides for __NR_seccomp_sigreturn, 484 and compat syscalls if the asm-generic/seccomp.h defaults need adjustment: 485 - __NR_seccomp_read_32 486 - __NR_seccomp_write_32 487 - __NR_seccomp_exit_32 488 - __NR_seccomp_sigreturn_32 489 490config HAVE_ARCH_SECCOMP_FILTER 491 bool 492 select HAVE_ARCH_SECCOMP 493 help 494 An arch should select this symbol if it provides all of these things: 495 - all the requirements for HAVE_ARCH_SECCOMP 496 - syscall_get_arch() 497 - syscall_get_arguments() 498 - syscall_rollback() 499 - syscall_set_return_value() 500 - SIGSYS siginfo_t support 501 - secure_computing is called from a ptrace_event()-safe context 502 - secure_computing return value is checked and a return value of -1 503 results in the system call being skipped immediately. 504 - seccomp syscall wired up 505 506config SECCOMP 507 prompt "Enable seccomp to safely execute untrusted bytecode" 508 def_bool y 509 depends on HAVE_ARCH_SECCOMP 510 help 511 This kernel feature is useful for number crunching applications 512 that may need to handle untrusted bytecode during their 513 execution. By using pipes or other transports made available 514 to the process as file descriptors supporting the read/write 515 syscalls, it's possible to isolate those applications in their 516 own address space using seccomp. Once seccomp is enabled via 517 prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be 518 disabled and the task is only allowed to execute a few safe 519 syscalls defined by each seccomp mode. 520 521 If unsure, say Y. 522 523config SECCOMP_FILTER 524 def_bool y 525 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 526 help 527 Enable tasks to build secure computing environments defined 528 in terms of Berkeley Packet Filter programs which implement 529 task-defined system call filtering polices. 530 531 See Documentation/userspace-api/seccomp_filter.rst for details. 532 533config HAVE_ARCH_STACKLEAK 534 bool 535 help 536 An architecture should select this if it has the code which 537 fills the used part of the kernel stack with the STACKLEAK_POISON 538 value before returning from system calls. 539 540config HAVE_STACKPROTECTOR 541 bool 542 help 543 An arch should select this symbol if: 544 - it has implemented a stack canary (e.g. __stack_chk_guard) 545 546config STACKPROTECTOR 547 bool "Stack Protector buffer overflow detection" 548 depends on HAVE_STACKPROTECTOR 549 depends on $(cc-option,-fstack-protector) 550 default y 551 help 552 This option turns on the "stack-protector" GCC feature. This 553 feature puts, at the beginning of functions, a canary value on 554 the stack just before the return address, and validates 555 the value just before actually returning. Stack based buffer 556 overflows (that need to overwrite this return address) now also 557 overwrite the canary, which gets detected and the attack is then 558 neutralized via a kernel panic. 559 560 Functions will have the stack-protector canary logic added if they 561 have an 8-byte or larger character array on the stack. 562 563 This feature requires gcc version 4.2 or above, or a distribution 564 gcc with the feature backported ("-fstack-protector"). 565 566 On an x86 "defconfig" build, this feature adds canary checks to 567 about 3% of all kernel functions, which increases kernel code size 568 by about 0.3%. 569 570config STACKPROTECTOR_STRONG 571 bool "Strong Stack Protector" 572 depends on STACKPROTECTOR 573 depends on $(cc-option,-fstack-protector-strong) 574 default y 575 help 576 Functions will have the stack-protector canary logic added in any 577 of the following conditions: 578 579 - local variable's address used as part of the right hand side of an 580 assignment or function argument 581 - local variable is an array (or union containing an array), 582 regardless of array type or length 583 - uses register local variables 584 585 This feature requires gcc version 4.9 or above, or a distribution 586 gcc with the feature backported ("-fstack-protector-strong"). 587 588 On an x86 "defconfig" build, this feature adds canary checks to 589 about 20% of all kernel functions, which increases the kernel code 590 size by about 2%. 591 592config ARCH_SUPPORTS_SHADOW_CALL_STACK 593 bool 594 help 595 An architecture should select this if it supports Clang's Shadow 596 Call Stack and implements runtime support for shadow stack 597 switching. 598 599config SHADOW_CALL_STACK 600 bool "Clang Shadow Call Stack" 601 depends on CC_IS_CLANG && ARCH_SUPPORTS_SHADOW_CALL_STACK 602 depends on DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER 603 help 604 This option enables Clang's Shadow Call Stack, which uses a 605 shadow stack to protect function return addresses from being 606 overwritten by an attacker. More information can be found in 607 Clang's documentation: 608 609 https://clang.llvm.org/docs/ShadowCallStack.html 610 611 Note that security guarantees in the kernel differ from the 612 ones documented for user space. The kernel must store addresses 613 of shadow stacks in memory, which means an attacker capable of 614 reading and writing arbitrary memory may be able to locate them 615 and hijack control flow by modifying the stacks. 616 617config LTO 618 bool 619 help 620 Selected if the kernel will be built using the compiler's LTO feature. 621 622config LTO_CLANG 623 bool 624 select LTO 625 help 626 Selected if the kernel will be built using Clang's LTO feature. 627 628config ARCH_SUPPORTS_LTO_CLANG 629 bool 630 help 631 An architecture should select this option if it supports: 632 - compiling with Clang, 633 - compiling inline assembly with Clang's integrated assembler, 634 - and linking with LLD. 635 636config ARCH_SUPPORTS_LTO_CLANG_THIN 637 bool 638 help 639 An architecture should select this option if it can support Clang's 640 ThinLTO mode. 641 642config HAS_LTO_CLANG 643 def_bool y 644 # Clang >= 11: https://github.com/ClangBuiltLinux/linux/issues/510 645 depends on CC_IS_CLANG && CLANG_VERSION >= 110000 && LD_IS_LLD 646 depends on $(success,test $(LLVM) -eq 1) 647 depends on $(success,test $(LLVM_IAS) -eq 1) 648 depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm) 649 depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm) 650 depends on ARCH_SUPPORTS_LTO_CLANG 651 depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT 652 depends on !KASAN || KASAN_HW_TAGS 653 depends on !GCOV_KERNEL 654 help 655 The compiler and Kconfig options support building with Clang's 656 LTO. 657 658choice 659 prompt "Link Time Optimization (LTO)" 660 default LTO_NONE 661 help 662 This option enables Link Time Optimization (LTO), which allows the 663 compiler to optimize binaries globally. 664 665 If unsure, select LTO_NONE. Note that LTO is very resource-intensive 666 so it's disabled by default. 667 668config LTO_NONE 669 bool "None" 670 help 671 Build the kernel normally, without Link Time Optimization (LTO). 672 673config LTO_CLANG_FULL 674 bool "Clang Full LTO (EXPERIMENTAL)" 675 depends on HAS_LTO_CLANG 676 depends on !COMPILE_TEST 677 select LTO_CLANG 678 help 679 This option enables Clang's full Link Time Optimization (LTO), which 680 allows the compiler to optimize the kernel globally. If you enable 681 this option, the compiler generates LLVM bitcode instead of ELF 682 object files, and the actual compilation from bitcode happens at 683 the LTO link step, which may take several minutes depending on the 684 kernel configuration. More information can be found from LLVM's 685 documentation: 686 687 https://llvm.org/docs/LinkTimeOptimization.html 688 689 During link time, this option can use a large amount of RAM, and 690 may take much longer than the ThinLTO option. 691 692config LTO_CLANG_THIN 693 bool "Clang ThinLTO (EXPERIMENTAL)" 694 depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN 695 select LTO_CLANG 696 help 697 This option enables Clang's ThinLTO, which allows for parallel 698 optimization and faster incremental compiles compared to the 699 CONFIG_LTO_CLANG_FULL option. More information can be found 700 from Clang's documentation: 701 702 https://clang.llvm.org/docs/ThinLTO.html 703 704 If unsure, say Y. 705endchoice 706 707config CFI_CLANG 708 bool "Use Clang's Control Flow Integrity (CFI)" 709 depends on LTO_CLANG && KALLSYMS 710 help 711 This option enables Clang's Control Flow Integrity (CFI), which adds 712 runtime checking for indirect function calls. 713 714config CFI_CLANG_SHADOW 715 bool "Use CFI shadow to speed up cross-module checks" 716 default y 717 depends on CFI_CLANG && MODULES 718 help 719 If you select this option, the kernel builds a fast look-up table of 720 CFI check functions in loaded modules to reduce overhead. 721 722config CFI_PERMISSIVE 723 bool "Use CFI in permissive mode" 724 depends on CFI_CLANG 725 help 726 When selected, Control Flow Integrity (CFI) violations result in a 727 warning instead of a kernel panic. This option is useful for finding 728 CFI violations during development. 729 730config HAVE_ARCH_WITHIN_STACK_FRAMES 731 bool 732 help 733 An architecture should select this if it can walk the kernel stack 734 frames to determine if an object is part of either the arguments 735 or local variables (i.e. that it excludes saved return addresses, 736 and similar) by implementing an inline arch_within_stack_frames(), 737 which is used by CONFIG_HARDENED_USERCOPY. 738 739config HAVE_CONTEXT_TRACKING 740 bool 741 help 742 Provide kernel/user boundaries probes necessary for subsystems 743 that need it, such as userspace RCU extended quiescent state. 744 Syscalls need to be wrapped inside user_exit()-user_enter(), either 745 optimized behind static key or through the slow path using TIF_NOHZ 746 flag. Exceptions handlers must be wrapped as well. Irqs are already 747 protected inside rcu_irq_enter/rcu_irq_exit() but preemption or signal 748 handling on irq exit still need to be protected. 749 750config HAVE_TIF_NOHZ 751 bool 752 help 753 Arch relies on TIF_NOHZ and syscall slow path to implement context 754 tracking calls to user_enter()/user_exit(). 755 756config HAVE_VIRT_CPU_ACCOUNTING 757 bool 758 759config ARCH_HAS_SCALED_CPUTIME 760 bool 761 762config HAVE_VIRT_CPU_ACCOUNTING_GEN 763 bool 764 default y if 64BIT 765 help 766 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 767 Before enabling this option, arch code must be audited 768 to ensure there are no races in concurrent read/write of 769 cputime_t. For example, reading/writing 64-bit cputime_t on 770 some 32-bit arches may require multiple accesses, so proper 771 locking is needed to protect against concurrent accesses. 772 773 774config HAVE_IRQ_TIME_ACCOUNTING 775 bool 776 help 777 Archs need to ensure they use a high enough resolution clock to 778 support irq time accounting and then call enable_sched_clock_irqtime(). 779 780config HAVE_MOVE_PUD 781 bool 782 help 783 Architectures that select this are able to move page tables at the 784 PUD level. If there are only 3 page table levels, the move effectively 785 happens at the PGD level. 786 787config HAVE_MOVE_PMD 788 bool 789 help 790 Archs that select this are able to move page tables at the PMD level. 791 792config HAVE_ARCH_TRANSPARENT_HUGEPAGE 793 bool 794 795config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 796 bool 797 798config HAVE_ARCH_HUGE_VMAP 799 bool 800 801config ARCH_WANT_HUGE_PMD_SHARE 802 bool 803 804config HAVE_ARCH_SOFT_DIRTY 805 bool 806 807config HAVE_MOD_ARCH_SPECIFIC 808 bool 809 help 810 The arch uses struct mod_arch_specific to store data. Many arches 811 just need a simple module loader without arch specific data - those 812 should not enable this. 813 814config MODULES_USE_ELF_RELA 815 bool 816 help 817 Modules only use ELF RELA relocations. Modules with ELF REL 818 relocations will give an error. 819 820config MODULES_USE_ELF_REL 821 bool 822 help 823 Modules only use ELF REL relocations. Modules with ELF RELA 824 relocations will give an error. 825 826config HAVE_IRQ_EXIT_ON_IRQ_STACK 827 bool 828 help 829 Architecture doesn't only execute the irq handler on the irq stack 830 but also irq_exit(). This way we can process softirqs on this irq 831 stack instead of switching to a new one when we call __do_softirq() 832 in the end of an hardirq. 833 This spares a stack switch and improves cache usage on softirq 834 processing. 835 836config PGTABLE_LEVELS 837 int 838 default 2 839 840config ARCH_HAS_ELF_RANDOMIZE 841 bool 842 help 843 An architecture supports choosing randomized locations for 844 stack, mmap, brk, and ET_DYN. Defined functions: 845 - arch_mmap_rnd() 846 - arch_randomize_brk() 847 848config HAVE_ARCH_MMAP_RND_BITS 849 bool 850 help 851 An arch should select this symbol if it supports setting a variable 852 number of bits for use in establishing the base address for mmap 853 allocations, has MMU enabled and provides values for both: 854 - ARCH_MMAP_RND_BITS_MIN 855 - ARCH_MMAP_RND_BITS_MAX 856 857config HAVE_EXIT_THREAD 858 bool 859 help 860 An architecture implements exit_thread. 861 862config ARCH_MMAP_RND_BITS_MIN 863 int 864 865config ARCH_MMAP_RND_BITS_MAX 866 int 867 868config ARCH_MMAP_RND_BITS_DEFAULT 869 int 870 871config ARCH_MMAP_RND_BITS 872 int "Number of bits to use for ASLR of mmap base address" if EXPERT 873 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 874 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 875 default ARCH_MMAP_RND_BITS_MIN 876 depends on HAVE_ARCH_MMAP_RND_BITS 877 help 878 This value can be used to select the number of bits to use to 879 determine the random offset to the base address of vma regions 880 resulting from mmap allocations. This value will be bounded 881 by the architecture's minimum and maximum supported values. 882 883 This value can be changed after boot using the 884 /proc/sys/vm/mmap_rnd_bits tunable 885 886config HAVE_ARCH_MMAP_RND_COMPAT_BITS 887 bool 888 help 889 An arch should select this symbol if it supports running applications 890 in compatibility mode, supports setting a variable number of bits for 891 use in establishing the base address for mmap allocations, has MMU 892 enabled and provides values for both: 893 - ARCH_MMAP_RND_COMPAT_BITS_MIN 894 - ARCH_MMAP_RND_COMPAT_BITS_MAX 895 896config ARCH_MMAP_RND_COMPAT_BITS_MIN 897 int 898 899config ARCH_MMAP_RND_COMPAT_BITS_MAX 900 int 901 902config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 903 int 904 905config ARCH_MMAP_RND_COMPAT_BITS 906 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 907 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 908 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 909 default ARCH_MMAP_RND_COMPAT_BITS_MIN 910 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 911 help 912 This value can be used to select the number of bits to use to 913 determine the random offset to the base address of vma regions 914 resulting from mmap allocations for compatible applications This 915 value will be bounded by the architecture's minimum and maximum 916 supported values. 917 918 This value can be changed after boot using the 919 /proc/sys/vm/mmap_rnd_compat_bits tunable 920 921config HAVE_ARCH_COMPAT_MMAP_BASES 922 bool 923 help 924 This allows 64bit applications to invoke 32-bit mmap() syscall 925 and vice-versa 32-bit applications to call 64-bit mmap(). 926 Required for applications doing different bitness syscalls. 927 928# This allows to use a set of generic functions to determine mmap base 929# address by giving priority to top-down scheme only if the process 930# is not in legacy mode (compat task, unlimited stack size or 931# sysctl_legacy_va_layout). 932# Architecture that selects this option can provide its own version of: 933# - STACK_RND_MASK 934config ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT 935 bool 936 depends on MMU 937 select ARCH_HAS_ELF_RANDOMIZE 938 939config HAVE_STACK_VALIDATION 940 bool 941 help 942 Architecture supports the 'objtool check' host tool command, which 943 performs compile-time stack metadata validation. 944 945config HAVE_RELIABLE_STACKTRACE 946 bool 947 help 948 Architecture has either save_stack_trace_tsk_reliable() or 949 arch_stack_walk_reliable() function which only returns a stack trace 950 if it can guarantee the trace is reliable. 951 952config HAVE_ARCH_HASH 953 bool 954 default n 955 help 956 If this is set, the architecture provides an <asm/hash.h> 957 file which provides platform-specific implementations of some 958 functions in <linux/hash.h> or fs/namei.c. 959 960config HAVE_ARCH_NVRAM_OPS 961 bool 962 963config ISA_BUS_API 964 def_bool ISA 965 966# 967# ABI hall of shame 968# 969config CLONE_BACKWARDS 970 bool 971 help 972 Architecture has tls passed as the 4th argument of clone(2), 973 not the 5th one. 974 975config CLONE_BACKWARDS2 976 bool 977 help 978 Architecture has the first two arguments of clone(2) swapped. 979 980config CLONE_BACKWARDS3 981 bool 982 help 983 Architecture has tls passed as the 3rd argument of clone(2), 984 not the 5th one. 985 986config ODD_RT_SIGACTION 987 bool 988 help 989 Architecture has unusual rt_sigaction(2) arguments 990 991config OLD_SIGSUSPEND 992 bool 993 help 994 Architecture has old sigsuspend(2) syscall, of one-argument variety 995 996config OLD_SIGSUSPEND3 997 bool 998 help 999 Even weirder antique ABI - three-argument sigsuspend(2) 1000 1001config OLD_SIGACTION 1002 bool 1003 help 1004 Architecture has old sigaction(2) syscall. Nope, not the same 1005 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 1006 but fairly different variant of sigaction(2), thanks to OSF/1 1007 compatibility... 1008 1009config COMPAT_OLD_SIGACTION 1010 bool 1011 1012config COMPAT_32BIT_TIME 1013 bool "Provide system calls for 32-bit time_t" 1014 default !64BIT || COMPAT 1015 help 1016 This enables 32 bit time_t support in addition to 64 bit time_t support. 1017 This is relevant on all 32-bit architectures, and 64-bit architectures 1018 as part of compat syscall handling. 1019 1020config ARCH_NO_PREEMPT 1021 bool 1022 1023config ARCH_SUPPORTS_RT 1024 bool 1025 1026config CPU_NO_EFFICIENT_FFS 1027 def_bool n 1028 1029config HAVE_ARCH_VMAP_STACK 1030 def_bool n 1031 help 1032 An arch should select this symbol if it can support kernel stacks 1033 in vmalloc space. This means: 1034 1035 - vmalloc space must be large enough to hold many kernel stacks. 1036 This may rule out many 32-bit architectures. 1037 1038 - Stacks in vmalloc space need to work reliably. For example, if 1039 vmap page tables are created on demand, either this mechanism 1040 needs to work while the stack points to a virtual address with 1041 unpopulated page tables or arch code (switch_to() and switch_mm(), 1042 most likely) needs to ensure that the stack's page table entries 1043 are populated before running on a possibly unpopulated stack. 1044 1045 - If the stack overflows into a guard page, something reasonable 1046 should happen. The definition of "reasonable" is flexible, but 1047 instantly rebooting without logging anything would be unfriendly. 1048 1049config VMAP_STACK 1050 default y 1051 bool "Use a virtually-mapped stack" 1052 depends on HAVE_ARCH_VMAP_STACK 1053 depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC 1054 help 1055 Enable this if you want the use virtually-mapped kernel stacks 1056 with guard pages. This causes kernel stack overflows to be 1057 caught immediately rather than causing difficult-to-diagnose 1058 corruption. 1059 1060 To use this with software KASAN modes, the architecture must support 1061 backing virtual mappings with real shadow memory, and KASAN_VMALLOC 1062 must be enabled. 1063 1064config ARCH_OPTIONAL_KERNEL_RWX 1065 def_bool n 1066 1067config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1068 def_bool n 1069 1070config ARCH_HAS_STRICT_KERNEL_RWX 1071 def_bool n 1072 1073config STRICT_KERNEL_RWX 1074 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 1075 depends on ARCH_HAS_STRICT_KERNEL_RWX 1076 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1077 help 1078 If this is set, kernel text and rodata memory will be made read-only, 1079 and non-text memory will be made non-executable. This provides 1080 protection against certain security exploits (e.g. executing the heap 1081 or modifying text) 1082 1083 These features are considered standard security practice these days. 1084 You should say Y here in almost all cases. 1085 1086config ARCH_HAS_STRICT_MODULE_RWX 1087 def_bool n 1088 1089config STRICT_MODULE_RWX 1090 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 1091 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 1092 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1093 help 1094 If this is set, module text and rodata memory will be made read-only, 1095 and non-text memory will be made non-executable. This provides 1096 protection against certain security exploits (e.g. writing to text) 1097 1098# select if the architecture provides an asm/dma-direct.h header 1099config ARCH_HAS_PHYS_TO_DMA 1100 bool 1101 1102config HAVE_ARCH_COMPILER_H 1103 bool 1104 help 1105 An architecture can select this if it provides an 1106 asm/compiler.h header that should be included after 1107 linux/compiler-*.h in order to override macro definitions that those 1108 headers generally provide. 1109 1110config HAVE_ARCH_PREL32_RELOCATIONS 1111 bool 1112 help 1113 May be selected by an architecture if it supports place-relative 1114 32-bit relocations, both in the toolchain and in the module loader, 1115 in which case relative references can be used in special sections 1116 for PCI fixup, initcalls etc which are only half the size on 64 bit 1117 architectures, and don't require runtime relocation on relocatable 1118 kernels. 1119 1120config ARCH_USE_MEMREMAP_PROT 1121 bool 1122 1123config LOCK_EVENT_COUNTS 1124 bool "Locking event counts collection" 1125 depends on DEBUG_FS 1126 help 1127 Enable light-weight counting of various locking related events 1128 in the system with minimal performance impact. This reduces 1129 the chance of application behavior change because of timing 1130 differences. The counts are reported via debugfs. 1131 1132# Select if the architecture has support for applying RELR relocations. 1133config ARCH_HAS_RELR 1134 bool 1135 1136config RELR 1137 bool "Use RELR relocation packing" 1138 depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR 1139 default y 1140 help 1141 Store the kernel's dynamic relocations in the RELR relocation packing 1142 format. Requires a compatible linker (LLD supports this feature), as 1143 well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy 1144 are compatible). 1145 1146config ARCH_HAS_MEM_ENCRYPT 1147 bool 1148 1149config ARCH_HAS_CC_PLATFORM 1150 bool 1151 1152config HAVE_SPARSE_SYSCALL_NR 1153 bool 1154 help 1155 An architecture should select this if its syscall numbering is sparse 1156 to save space. For example, MIPS architecture has a syscall array with 1157 entries at 4000, 5000 and 6000 locations. This option turns on syscall 1158 related optimizations for a given architecture. 1159 1160config ARCH_HAS_VDSO_DATA 1161 bool 1162 1163config HAVE_STATIC_CALL 1164 bool 1165 1166config HAVE_STATIC_CALL_INLINE 1167 bool 1168 depends on HAVE_STATIC_CALL 1169 1170config ARCH_WANT_LD_ORPHAN_WARN 1171 bool 1172 help 1173 An arch should select this symbol once all linker sections are explicitly 1174 included, size-asserted, or discarded in the linker scripts. This is 1175 important because we never want expected sections to be placed heuristically 1176 by the linker, since the locations of such sections can change between linker 1177 versions. 1178 1179config ARCH_SPLIT_ARG64 1180 bool 1181 help 1182 If a 32-bit architecture requires 64-bit arguments to be split into 1183 pairs of 32-bit arguments, select this option. 1184 1185source "kernel/gcov/Kconfig" 1186 1187source "scripts/gcc-plugins/Kconfig" 1188 1189endmenu 1190