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 HOTPLUG_SMT 28 bool 29 30config GENERIC_ENTRY 31 bool 32 33config KPROBES 34 bool "Kprobes" 35 depends on MODULES 36 depends on HAVE_KPROBES 37 select KALLSYMS 38 help 39 Kprobes allows you to trap at almost any kernel address and 40 execute a callback function. register_kprobe() establishes 41 a probepoint and specifies the callback. Kprobes is useful 42 for kernel debugging, non-intrusive instrumentation and testing. 43 If in doubt, say "N". 44 45config JUMP_LABEL 46 bool "Optimize very unlikely/likely branches" 47 depends on HAVE_ARCH_JUMP_LABEL 48 depends on CC_HAS_ASM_GOTO 49 help 50 This option enables a transparent branch optimization that 51 makes certain almost-always-true or almost-always-false branch 52 conditions even cheaper to execute within the kernel. 53 54 Certain performance-sensitive kernel code, such as trace points, 55 scheduler functionality, networking code and KVM have such 56 branches and include support for this optimization technique. 57 58 If it is detected that the compiler has support for "asm goto", 59 the kernel will compile such branches with just a nop 60 instruction. When the condition flag is toggled to true, the 61 nop will be converted to a jump instruction to execute the 62 conditional block of instructions. 63 64 This technique lowers overhead and stress on the branch prediction 65 of the processor and generally makes the kernel faster. The update 66 of the condition is slower, but those are always very rare. 67 68 ( On 32-bit x86, the necessary options added to the compiler 69 flags may increase the size of the kernel slightly. ) 70 71config STATIC_KEYS_SELFTEST 72 bool "Static key selftest" 73 depends on JUMP_LABEL 74 help 75 Boot time self-test of the branch patching code. 76 77config STATIC_CALL_SELFTEST 78 bool "Static call selftest" 79 depends on HAVE_STATIC_CALL 80 help 81 Boot time self-test of the call patching code. 82 83config OPTPROBES 84 def_bool y 85 depends on KPROBES && HAVE_OPTPROBES 86 select TASKS_RCU if PREEMPTION 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 depends on ARCH_SUPPORTS_UPROBES 100 help 101 Uprobes is the user-space counterpart to kprobes: they 102 enable instrumentation applications (such as 'perf probe') 103 to establish unintrusive probes in user-space binaries and 104 libraries, by executing handler functions when the probes 105 are hit by user-space applications. 106 107 ( These probes come in the form of single-byte breakpoints, 108 managed by the kernel and kept transparent to the probed 109 application. ) 110 111config HAVE_64BIT_ALIGNED_ACCESS 112 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 113 help 114 Some architectures require 64 bit accesses to be 64 bit 115 aligned, which also requires structs containing 64 bit values 116 to be 64 bit aligned too. This includes some 32 bit 117 architectures which can do 64 bit accesses, as well as 64 bit 118 architectures without unaligned access. 119 120 This symbol should be selected by an architecture if 64 bit 121 accesses are required to be 64 bit aligned in this way even 122 though it is not a 64 bit architecture. 123 124 See Documentation/core-api/unaligned-memory-access.rst for 125 more information on the topic of unaligned memory accesses. 126 127config HAVE_EFFICIENT_UNALIGNED_ACCESS 128 bool 129 help 130 Some architectures are unable to perform unaligned accesses 131 without the use of get_unaligned/put_unaligned. Others are 132 unable to perform such accesses efficiently (e.g. trap on 133 unaligned access and require fixing it up in the exception 134 handler.) 135 136 This symbol should be selected by an architecture if it can 137 perform unaligned accesses efficiently to allow different 138 code paths to be selected for these cases. Some network 139 drivers, for example, could opt to not fix up alignment 140 problems with received packets if doing so would not help 141 much. 142 143 See Documentation/core-api/unaligned-memory-access.rst for more 144 information on the topic of unaligned memory accesses. 145 146config ARCH_USE_BUILTIN_BSWAP 147 bool 148 help 149 Modern versions of GCC (since 4.4) have builtin functions 150 for handling byte-swapping. Using these, instead of the old 151 inline assembler that the architecture code provides in the 152 __arch_bswapXX() macros, allows the compiler to see what's 153 happening and offers more opportunity for optimisation. In 154 particular, the compiler will be able to combine the byteswap 155 with a nearby load or store and use load-and-swap or 156 store-and-swap instructions if the architecture has them. It 157 should almost *never* result in code which is worse than the 158 hand-coded assembler in <asm/swab.h>. But just in case it 159 does, the use of the builtins is optional. 160 161 Any architecture with load-and-swap or store-and-swap 162 instructions should set this. And it shouldn't hurt to set it 163 on architectures that don't have such instructions. 164 165config KRETPROBES 166 def_bool y 167 depends on KPROBES && HAVE_KRETPROBES 168 169config USER_RETURN_NOTIFIER 170 bool 171 depends on HAVE_USER_RETURN_NOTIFIER 172 help 173 Provide a kernel-internal notification when a cpu is about to 174 switch to user mode. 175 176config HAVE_IOREMAP_PROT 177 bool 178 179config HAVE_KPROBES 180 bool 181 182config HAVE_KRETPROBES 183 bool 184 185config HAVE_OPTPROBES 186 bool 187 188config HAVE_KPROBES_ON_FTRACE 189 bool 190 191config ARCH_CORRECT_STACKTRACE_ON_KRETPROBE 192 bool 193 help 194 Since kretprobes modifies return address on the stack, the 195 stacktrace may see the kretprobe trampoline address instead 196 of correct one. If the architecture stacktrace code and 197 unwinder can adjust such entries, select this configuration. 198 199config HAVE_FUNCTION_ERROR_INJECTION 200 bool 201 202config HAVE_NMI 203 bool 204 205config TRACE_IRQFLAGS_SUPPORT 206 bool 207 208# 209# An arch should select this if it provides all these things: 210# 211# task_pt_regs() in asm/processor.h or asm/ptrace.h 212# arch_has_single_step() if there is hardware single-step support 213# arch_has_block_step() if there is hardware block-step support 214# asm/syscall.h supplying asm-generic/syscall.h interface 215# linux/regset.h user_regset interfaces 216# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 217# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 218# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 219# signal delivery calls tracehook_signal_handler() 220# 221config HAVE_ARCH_TRACEHOOK 222 bool 223 224config HAVE_DMA_CONTIGUOUS 225 bool 226 227config GENERIC_SMP_IDLE_THREAD 228 bool 229 230config GENERIC_IDLE_POLL_SETUP 231 bool 232 233config ARCH_HAS_FORTIFY_SOURCE 234 bool 235 help 236 An architecture should select this when it can successfully 237 build and run with CONFIG_FORTIFY_SOURCE. 238 239# 240# Select if the arch provides a historic keepinit alias for the retain_initrd 241# command line option 242# 243config ARCH_HAS_KEEPINITRD 244 bool 245 246# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 247config ARCH_HAS_SET_MEMORY 248 bool 249 250# Select if arch has all set_direct_map_invalid/default() functions 251config ARCH_HAS_SET_DIRECT_MAP 252 bool 253 254# 255# Select if the architecture provides the arch_dma_set_uncached symbol to 256# either provide an uncached segment alias for a DMA allocation, or 257# to remap the page tables in place. 258# 259config ARCH_HAS_DMA_SET_UNCACHED 260 bool 261 262# 263# Select if the architectures provides the arch_dma_clear_uncached symbol 264# to undo an in-place page table remap for uncached access. 265# 266config ARCH_HAS_DMA_CLEAR_UNCACHED 267 bool 268 269# Select if arch init_task must go in the __init_task_data section 270config ARCH_TASK_STRUCT_ON_STACK 271 bool 272 273# Select if arch has its private alloc_task_struct() function 274config ARCH_TASK_STRUCT_ALLOCATOR 275 bool 276 277config HAVE_ARCH_THREAD_STRUCT_WHITELIST 278 bool 279 depends on !ARCH_TASK_STRUCT_ALLOCATOR 280 help 281 An architecture should select this to provide hardened usercopy 282 knowledge about what region of the thread_struct should be 283 whitelisted for copying to userspace. Normally this is only the 284 FPU registers. Specifically, arch_thread_struct_whitelist() 285 should be implemented. Without this, the entire thread_struct 286 field in task_struct will be left whitelisted. 287 288# Select if arch has its private alloc_thread_stack() function 289config ARCH_THREAD_STACK_ALLOCATOR 290 bool 291 292# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 293config ARCH_WANTS_DYNAMIC_TASK_STRUCT 294 bool 295 296config ARCH_WANTS_NO_INSTR 297 bool 298 help 299 An architecture should select this if the noinstr macro is being used on 300 functions to denote that the toolchain should avoid instrumenting such 301 functions and is required for correctness. 302 303config ARCH_32BIT_OFF_T 304 bool 305 depends on !64BIT 306 help 307 All new 32-bit architectures should have 64-bit off_t type on 308 userspace side which corresponds to the loff_t kernel type. This 309 is the requirement for modern ABIs. Some existing architectures 310 still support 32-bit off_t. This option is enabled for all such 311 architectures explicitly. 312 313# Selected by 64 bit architectures which have a 32 bit f_tinode in struct ustat 314config ARCH_32BIT_USTAT_F_TINODE 315 bool 316 317config HAVE_ASM_MODVERSIONS 318 bool 319 help 320 This symbol should be selected by an architecture if it provides 321 <asm/asm-prototypes.h> to support the module versioning for symbols 322 exported from assembly code. 323 324config HAVE_REGS_AND_STACK_ACCESS_API 325 bool 326 help 327 This symbol should be selected by an architecture if it supports 328 the API needed to access registers and stack entries from pt_regs, 329 declared in asm/ptrace.h 330 For example the kprobes-based event tracer needs this API. 331 332config HAVE_RSEQ 333 bool 334 depends on HAVE_REGS_AND_STACK_ACCESS_API 335 help 336 This symbol should be selected by an architecture if it 337 supports an implementation of restartable sequences. 338 339config HAVE_FUNCTION_ARG_ACCESS_API 340 bool 341 help 342 This symbol should be selected by an architecture if it supports 343 the API needed to access function arguments from pt_regs, 344 declared in asm/ptrace.h 345 346config HAVE_HW_BREAKPOINT 347 bool 348 depends on PERF_EVENTS 349 350config HAVE_MIXED_BREAKPOINTS_REGS 351 bool 352 depends on HAVE_HW_BREAKPOINT 353 help 354 Depending on the arch implementation of hardware breakpoints, 355 some of them have separate registers for data and instruction 356 breakpoints addresses, others have mixed registers to store 357 them but define the access type in a control register. 358 Select this option if your arch implements breakpoints under the 359 latter fashion. 360 361config HAVE_USER_RETURN_NOTIFIER 362 bool 363 364config HAVE_PERF_EVENTS_NMI 365 bool 366 help 367 System hardware can generate an NMI using the perf event 368 subsystem. Also has support for calculating CPU cycle events 369 to determine how many clock cycles in a given period. 370 371config HAVE_HARDLOCKUP_DETECTOR_PERF 372 bool 373 depends on HAVE_PERF_EVENTS_NMI 374 help 375 The arch chooses to use the generic perf-NMI-based hardlockup 376 detector. Must define HAVE_PERF_EVENTS_NMI. 377 378config HAVE_NMI_WATCHDOG 379 depends on HAVE_NMI 380 bool 381 help 382 The arch provides a low level NMI watchdog. It provides 383 asm/nmi.h, and defines its own arch_touch_nmi_watchdog(). 384 385config HAVE_HARDLOCKUP_DETECTOR_ARCH 386 bool 387 select HAVE_NMI_WATCHDOG 388 help 389 The arch chooses to provide its own hardlockup detector, which is 390 a superset of the HAVE_NMI_WATCHDOG. It also conforms to config 391 interfaces and parameters provided by hardlockup detector subsystem. 392 393config HAVE_PERF_REGS 394 bool 395 help 396 Support selective register dumps for perf events. This includes 397 bit-mapping of each registers and a unique architecture id. 398 399config HAVE_PERF_USER_STACK_DUMP 400 bool 401 help 402 Support user stack dumps for perf event samples. This needs 403 access to the user stack pointer which is not unified across 404 architectures. 405 406config HAVE_ARCH_JUMP_LABEL 407 bool 408 409config HAVE_ARCH_JUMP_LABEL_RELATIVE 410 bool 411 412config MMU_GATHER_TABLE_FREE 413 bool 414 415config MMU_GATHER_RCU_TABLE_FREE 416 bool 417 select MMU_GATHER_TABLE_FREE 418 419config MMU_GATHER_PAGE_SIZE 420 bool 421 422config MMU_GATHER_NO_RANGE 423 bool 424 425config MMU_GATHER_NO_GATHER 426 bool 427 depends on MMU_GATHER_TABLE_FREE 428 429config ARCH_WANT_IRQS_OFF_ACTIVATE_MM 430 bool 431 help 432 Temporary select until all architectures can be converted to have 433 irqs disabled over activate_mm. Architectures that do IPI based TLB 434 shootdowns should enable this. 435 436config ARCH_HAVE_NMI_SAFE_CMPXCHG 437 bool 438 439config HAVE_ALIGNED_STRUCT_PAGE 440 bool 441 help 442 This makes sure that struct pages are double word aligned and that 443 e.g. the SLUB allocator can perform double word atomic operations 444 on a struct page for better performance. However selecting this 445 might increase the size of a struct page by a word. 446 447config HAVE_CMPXCHG_LOCAL 448 bool 449 450config HAVE_CMPXCHG_DOUBLE 451 bool 452 453config ARCH_WEAK_RELEASE_ACQUIRE 454 bool 455 456config ARCH_WANT_IPC_PARSE_VERSION 457 bool 458 459config ARCH_WANT_COMPAT_IPC_PARSE_VERSION 460 bool 461 462config ARCH_WANT_OLD_COMPAT_IPC 463 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 464 bool 465 466config HAVE_ARCH_SECCOMP 467 bool 468 help 469 An arch should select this symbol to support seccomp mode 1 (the fixed 470 syscall policy), and must provide an overrides for __NR_seccomp_sigreturn, 471 and compat syscalls if the asm-generic/seccomp.h defaults need adjustment: 472 - __NR_seccomp_read_32 473 - __NR_seccomp_write_32 474 - __NR_seccomp_exit_32 475 - __NR_seccomp_sigreturn_32 476 477config HAVE_ARCH_SECCOMP_FILTER 478 bool 479 select HAVE_ARCH_SECCOMP 480 help 481 An arch should select this symbol if it provides all of these things: 482 - all the requirements for HAVE_ARCH_SECCOMP 483 - syscall_get_arch() 484 - syscall_get_arguments() 485 - syscall_rollback() 486 - syscall_set_return_value() 487 - SIGSYS siginfo_t support 488 - secure_computing is called from a ptrace_event()-safe context 489 - secure_computing return value is checked and a return value of -1 490 results in the system call being skipped immediately. 491 - seccomp syscall wired up 492 - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE, 493 SECCOMP_ARCH_NATIVE_NR, SECCOMP_ARCH_NATIVE_NAME defined. If 494 COMPAT is supported, have the SECCOMP_ARCH_COMPAT* defines too. 495 496config SECCOMP 497 prompt "Enable seccomp to safely execute untrusted bytecode" 498 def_bool y 499 depends on HAVE_ARCH_SECCOMP 500 help 501 This kernel feature is useful for number crunching applications 502 that may need to handle untrusted bytecode during their 503 execution. By using pipes or other transports made available 504 to the process as file descriptors supporting the read/write 505 syscalls, it's possible to isolate those applications in their 506 own address space using seccomp. Once seccomp is enabled via 507 prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be 508 disabled and the task is only allowed to execute a few safe 509 syscalls defined by each seccomp mode. 510 511 If unsure, say Y. 512 513config SECCOMP_FILTER 514 def_bool y 515 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 516 help 517 Enable tasks to build secure computing environments defined 518 in terms of Berkeley Packet Filter programs which implement 519 task-defined system call filtering polices. 520 521 See Documentation/userspace-api/seccomp_filter.rst for details. 522 523config SECCOMP_CACHE_DEBUG 524 bool "Show seccomp filter cache status in /proc/pid/seccomp_cache" 525 depends on SECCOMP_FILTER && !HAVE_SPARSE_SYSCALL_NR 526 depends on PROC_FS 527 help 528 This enables the /proc/pid/seccomp_cache interface to monitor 529 seccomp cache data. The file format is subject to change. Reading 530 the file requires CAP_SYS_ADMIN. 531 532 This option is for debugging only. Enabling presents the risk that 533 an adversary may be able to infer the seccomp filter logic. 534 535 If unsure, say N. 536 537config HAVE_ARCH_STACKLEAK 538 bool 539 help 540 An architecture should select this if it has the code which 541 fills the used part of the kernel stack with the STACKLEAK_POISON 542 value before returning from system calls. 543 544config HAVE_STACKPROTECTOR 545 bool 546 help 547 An arch should select this symbol if: 548 - it has implemented a stack canary (e.g. __stack_chk_guard) 549 550config STACKPROTECTOR 551 bool "Stack Protector buffer overflow detection" 552 depends on HAVE_STACKPROTECTOR 553 depends on $(cc-option,-fstack-protector) 554 default y 555 help 556 This option turns on the "stack-protector" GCC feature. This 557 feature puts, at the beginning of functions, a canary value on 558 the stack just before the return address, and validates 559 the value just before actually returning. Stack based buffer 560 overflows (that need to overwrite this return address) now also 561 overwrite the canary, which gets detected and the attack is then 562 neutralized via a kernel panic. 563 564 Functions will have the stack-protector canary logic added if they 565 have an 8-byte or larger character array on the stack. 566 567 This feature requires gcc version 4.2 or above, or a distribution 568 gcc with the feature backported ("-fstack-protector"). 569 570 On an x86 "defconfig" build, this feature adds canary checks to 571 about 3% of all kernel functions, which increases kernel code size 572 by about 0.3%. 573 574config STACKPROTECTOR_STRONG 575 bool "Strong Stack Protector" 576 depends on STACKPROTECTOR 577 depends on $(cc-option,-fstack-protector-strong) 578 default y 579 help 580 Functions will have the stack-protector canary logic added in any 581 of the following conditions: 582 583 - local variable's address used as part of the right hand side of an 584 assignment or function argument 585 - local variable is an array (or union containing an array), 586 regardless of array type or length 587 - uses register local variables 588 589 This feature requires gcc version 4.9 or above, or a distribution 590 gcc with the feature backported ("-fstack-protector-strong"). 591 592 On an x86 "defconfig" build, this feature adds canary checks to 593 about 20% of all kernel functions, which increases the kernel code 594 size by about 2%. 595 596config ARCH_SUPPORTS_SHADOW_CALL_STACK 597 bool 598 help 599 An architecture should select this if it supports the compiler's 600 Shadow Call Stack and implements runtime support for shadow stack 601 switching. 602 603config SHADOW_CALL_STACK 604 bool "Shadow Call Stack" 605 depends on ARCH_SUPPORTS_SHADOW_CALL_STACK 606 depends on DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER 607 help 608 This option enables the compiler's Shadow Call Stack, which 609 uses a shadow stack to protect function return addresses from 610 being overwritten by an attacker. More information can be found 611 in the compiler's documentation: 612 613 - Clang: https://clang.llvm.org/docs/ShadowCallStack.html 614 - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options 615 616 Note that security guarantees in the kernel differ from the 617 ones documented for user space. The kernel must store addresses 618 of shadow stacks in memory, which means an attacker capable of 619 reading and writing arbitrary memory may be able to locate them 620 and hijack control flow by modifying the stacks. 621 622config LTO 623 bool 624 help 625 Selected if the kernel will be built using the compiler's LTO feature. 626 627config LTO_CLANG 628 bool 629 select LTO 630 help 631 Selected if the kernel will be built using Clang's LTO feature. 632 633config ARCH_SUPPORTS_LTO_CLANG 634 bool 635 help 636 An architecture should select this option if it supports: 637 - compiling with Clang, 638 - compiling inline assembly with Clang's integrated assembler, 639 - and linking with LLD. 640 641config ARCH_SUPPORTS_LTO_CLANG_THIN 642 bool 643 help 644 An architecture should select this option if it can support Clang's 645 ThinLTO mode. 646 647config HAS_LTO_CLANG 648 def_bool y 649 depends on CC_IS_CLANG && LD_IS_LLD && AS_IS_LLVM 650 depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm) 651 depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm) 652 depends on ARCH_SUPPORTS_LTO_CLANG 653 depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT 654 depends on !KASAN || KASAN_HW_TAGS 655 depends on !GCOV_KERNEL 656 help 657 The compiler and Kconfig options support building with Clang's 658 LTO. 659 660choice 661 prompt "Link Time Optimization (LTO)" 662 default LTO_NONE 663 help 664 This option enables Link Time Optimization (LTO), which allows the 665 compiler to optimize binaries globally. 666 667 If unsure, select LTO_NONE. Note that LTO is very resource-intensive 668 so it's disabled by default. 669 670config LTO_NONE 671 bool "None" 672 help 673 Build the kernel normally, without Link Time Optimization (LTO). 674 675config LTO_CLANG_FULL 676 bool "Clang Full LTO (EXPERIMENTAL)" 677 depends on HAS_LTO_CLANG 678 depends on !COMPILE_TEST 679 select LTO_CLANG 680 help 681 This option enables Clang's full Link Time Optimization (LTO), which 682 allows the compiler to optimize the kernel globally. If you enable 683 this option, the compiler generates LLVM bitcode instead of ELF 684 object files, and the actual compilation from bitcode happens at 685 the LTO link step, which may take several minutes depending on the 686 kernel configuration. More information can be found from LLVM's 687 documentation: 688 689 https://llvm.org/docs/LinkTimeOptimization.html 690 691 During link time, this option can use a large amount of RAM, and 692 may take much longer than the ThinLTO option. 693 694config LTO_CLANG_THIN 695 bool "Clang ThinLTO (EXPERIMENTAL)" 696 depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN 697 select LTO_CLANG 698 help 699 This option enables Clang's ThinLTO, which allows for parallel 700 optimization and faster incremental compiles compared to the 701 CONFIG_LTO_CLANG_FULL option. More information can be found 702 from Clang's documentation: 703 704 https://clang.llvm.org/docs/ThinLTO.html 705 706 If unsure, say Y. 707endchoice 708 709config ARCH_SUPPORTS_CFI_CLANG 710 bool 711 help 712 An architecture should select this option if it can support Clang's 713 Control-Flow Integrity (CFI) checking. 714 715config CFI_CLANG 716 bool "Use Clang's Control Flow Integrity (CFI)" 717 depends on LTO_CLANG && ARCH_SUPPORTS_CFI_CLANG 718 # Clang >= 12: 719 # - https://bugs.llvm.org/show_bug.cgi?id=46258 720 # - https://bugs.llvm.org/show_bug.cgi?id=47479 721 depends on CLANG_VERSION >= 120000 722 select KALLSYMS 723 help 724 This option enables Clang’s forward-edge Control Flow Integrity 725 (CFI) checking, where the compiler injects a runtime check to each 726 indirect function call to ensure the target is a valid function with 727 the correct static type. This restricts possible call targets and 728 makes it more difficult for an attacker to exploit bugs that allow 729 the modification of stored function pointers. More information can be 730 found from Clang's documentation: 731 732 https://clang.llvm.org/docs/ControlFlowIntegrity.html 733 734config CFI_CLANG_SHADOW 735 bool "Use CFI shadow to speed up cross-module checks" 736 default y 737 depends on CFI_CLANG && MODULES 738 help 739 If you select this option, the kernel builds a fast look-up table of 740 CFI check functions in loaded modules to reduce performance overhead. 741 742 If unsure, say Y. 743 744config CFI_PERMISSIVE 745 bool "Use CFI in permissive mode" 746 depends on CFI_CLANG 747 help 748 When selected, Control Flow Integrity (CFI) violations result in a 749 warning instead of a kernel panic. This option should only be used 750 for finding indirect call type mismatches during development. 751 752 If unsure, say N. 753 754config HAVE_ARCH_WITHIN_STACK_FRAMES 755 bool 756 help 757 An architecture should select this if it can walk the kernel stack 758 frames to determine if an object is part of either the arguments 759 or local variables (i.e. that it excludes saved return addresses, 760 and similar) by implementing an inline arch_within_stack_frames(), 761 which is used by CONFIG_HARDENED_USERCOPY. 762 763config HAVE_CONTEXT_TRACKING 764 bool 765 help 766 Provide kernel/user boundaries probes necessary for subsystems 767 that need it, such as userspace RCU extended quiescent state. 768 Syscalls need to be wrapped inside user_exit()-user_enter(), either 769 optimized behind static key or through the slow path using TIF_NOHZ 770 flag. Exceptions handlers must be wrapped as well. Irqs are already 771 protected inside rcu_irq_enter/rcu_irq_exit() but preemption or signal 772 handling on irq exit still need to be protected. 773 774config HAVE_CONTEXT_TRACKING_OFFSTACK 775 bool 776 help 777 Architecture neither relies on exception_enter()/exception_exit() 778 nor on schedule_user(). Also preempt_schedule_notrace() and 779 preempt_schedule_irq() can't be called in a preemptible section 780 while context tracking is CONTEXT_USER. This feature reflects a sane 781 entry implementation where the following requirements are met on 782 critical entry code, ie: before user_exit() or after user_enter(): 783 784 - Critical entry code isn't preemptible (or better yet: 785 not interruptible). 786 - No use of RCU read side critical sections, unless rcu_nmi_enter() 787 got called. 788 - No use of instrumentation, unless instrumentation_begin() got 789 called. 790 791config HAVE_TIF_NOHZ 792 bool 793 help 794 Arch relies on TIF_NOHZ and syscall slow path to implement context 795 tracking calls to user_enter()/user_exit(). 796 797config HAVE_VIRT_CPU_ACCOUNTING 798 bool 799 800config HAVE_VIRT_CPU_ACCOUNTING_IDLE 801 bool 802 help 803 Architecture has its own way to account idle CPU time and therefore 804 doesn't implement vtime_account_idle(). 805 806config ARCH_HAS_SCALED_CPUTIME 807 bool 808 809config HAVE_VIRT_CPU_ACCOUNTING_GEN 810 bool 811 default y if 64BIT 812 help 813 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 814 Before enabling this option, arch code must be audited 815 to ensure there are no races in concurrent read/write of 816 cputime_t. For example, reading/writing 64-bit cputime_t on 817 some 32-bit arches may require multiple accesses, so proper 818 locking is needed to protect against concurrent accesses. 819 820config HAVE_IRQ_TIME_ACCOUNTING 821 bool 822 help 823 Archs need to ensure they use a high enough resolution clock to 824 support irq time accounting and then call enable_sched_clock_irqtime(). 825 826config HAVE_MOVE_PUD 827 bool 828 help 829 Architectures that select this are able to move page tables at the 830 PUD level. If there are only 3 page table levels, the move effectively 831 happens at the PGD level. 832 833config HAVE_MOVE_PMD 834 bool 835 help 836 Archs that select this are able to move page tables at the PMD level. 837 838config HAVE_ARCH_TRANSPARENT_HUGEPAGE 839 bool 840 841config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 842 bool 843 844config HAVE_ARCH_HUGE_VMAP 845 bool 846 847# 848# Archs that select this would be capable of PMD-sized vmaps (i.e., 849# arch_vmap_pmd_supported() returns true), and they must make no assumptions 850# that vmalloc memory is mapped with PAGE_SIZE ptes. The VM_NO_HUGE_VMAP flag 851# can be used to prohibit arch-specific allocations from using hugepages to 852# help with this (e.g., modules may require it). 853# 854config HAVE_ARCH_HUGE_VMALLOC 855 depends on HAVE_ARCH_HUGE_VMAP 856 bool 857 858config ARCH_WANT_HUGE_PMD_SHARE 859 bool 860 861config HAVE_ARCH_SOFT_DIRTY 862 bool 863 864config HAVE_MOD_ARCH_SPECIFIC 865 bool 866 help 867 The arch uses struct mod_arch_specific to store data. Many arches 868 just need a simple module loader without arch specific data - those 869 should not enable this. 870 871config MODULES_USE_ELF_RELA 872 bool 873 help 874 Modules only use ELF RELA relocations. Modules with ELF REL 875 relocations will give an error. 876 877config MODULES_USE_ELF_REL 878 bool 879 help 880 Modules only use ELF REL relocations. Modules with ELF RELA 881 relocations will give an error. 882 883config HAVE_IRQ_EXIT_ON_IRQ_STACK 884 bool 885 help 886 Architecture doesn't only execute the irq handler on the irq stack 887 but also irq_exit(). This way we can process softirqs on this irq 888 stack instead of switching to a new one when we call __do_softirq() 889 in the end of an hardirq. 890 This spares a stack switch and improves cache usage on softirq 891 processing. 892 893config HAVE_SOFTIRQ_ON_OWN_STACK 894 bool 895 help 896 Architecture provides a function to run __do_softirq() on a 897 separate stack. 898 899config ALTERNATE_USER_ADDRESS_SPACE 900 bool 901 help 902 Architectures set this when the CPU uses separate address 903 spaces for kernel and user space pointers. In this case, the 904 access_ok() check on a __user pointer is skipped. 905 906config PGTABLE_LEVELS 907 int 908 default 2 909 910config ARCH_HAS_ELF_RANDOMIZE 911 bool 912 help 913 An architecture supports choosing randomized locations for 914 stack, mmap, brk, and ET_DYN. Defined functions: 915 - arch_mmap_rnd() 916 - arch_randomize_brk() 917 918config HAVE_ARCH_MMAP_RND_BITS 919 bool 920 help 921 An arch should select this symbol if it supports setting a variable 922 number of bits for use in establishing the base address for mmap 923 allocations, has MMU enabled and provides values for both: 924 - ARCH_MMAP_RND_BITS_MIN 925 - ARCH_MMAP_RND_BITS_MAX 926 927config HAVE_EXIT_THREAD 928 bool 929 help 930 An architecture implements exit_thread. 931 932config ARCH_MMAP_RND_BITS_MIN 933 int 934 935config ARCH_MMAP_RND_BITS_MAX 936 int 937 938config ARCH_MMAP_RND_BITS_DEFAULT 939 int 940 941config ARCH_MMAP_RND_BITS 942 int "Number of bits to use for ASLR of mmap base address" if EXPERT 943 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 944 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 945 default ARCH_MMAP_RND_BITS_MIN 946 depends on HAVE_ARCH_MMAP_RND_BITS 947 help 948 This value can be used to select the number of bits to use to 949 determine the random offset to the base address of vma regions 950 resulting from mmap allocations. This value will be bounded 951 by the architecture's minimum and maximum supported values. 952 953 This value can be changed after boot using the 954 /proc/sys/vm/mmap_rnd_bits tunable 955 956config HAVE_ARCH_MMAP_RND_COMPAT_BITS 957 bool 958 help 959 An arch should select this symbol if it supports running applications 960 in compatibility mode, supports setting a variable number of bits for 961 use in establishing the base address for mmap allocations, has MMU 962 enabled and provides values for both: 963 - ARCH_MMAP_RND_COMPAT_BITS_MIN 964 - ARCH_MMAP_RND_COMPAT_BITS_MAX 965 966config ARCH_MMAP_RND_COMPAT_BITS_MIN 967 int 968 969config ARCH_MMAP_RND_COMPAT_BITS_MAX 970 int 971 972config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 973 int 974 975config ARCH_MMAP_RND_COMPAT_BITS 976 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 977 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 978 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 979 default ARCH_MMAP_RND_COMPAT_BITS_MIN 980 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 981 help 982 This value can be used to select the number of bits to use to 983 determine the random offset to the base address of vma regions 984 resulting from mmap allocations for compatible applications This 985 value will be bounded by the architecture's minimum and maximum 986 supported values. 987 988 This value can be changed after boot using the 989 /proc/sys/vm/mmap_rnd_compat_bits tunable 990 991config HAVE_ARCH_COMPAT_MMAP_BASES 992 bool 993 help 994 This allows 64bit applications to invoke 32-bit mmap() syscall 995 and vice-versa 32-bit applications to call 64-bit mmap(). 996 Required for applications doing different bitness syscalls. 997 998config PAGE_SIZE_LESS_THAN_64KB 999 def_bool y 1000 depends on !ARM64_64K_PAGES 1001 depends on !IA64_PAGE_SIZE_64KB 1002 depends on !PAGE_SIZE_64KB 1003 depends on !PARISC_PAGE_SIZE_64KB 1004 depends on !PPC_64K_PAGES 1005 depends on PAGE_SIZE_LESS_THAN_256KB 1006 1007config PAGE_SIZE_LESS_THAN_256KB 1008 def_bool y 1009 depends on !PPC_256K_PAGES 1010 depends on !PAGE_SIZE_256KB 1011 1012# This allows to use a set of generic functions to determine mmap base 1013# address by giving priority to top-down scheme only if the process 1014# is not in legacy mode (compat task, unlimited stack size or 1015# sysctl_legacy_va_layout). 1016# Architecture that selects this option can provide its own version of: 1017# - STACK_RND_MASK 1018config ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT 1019 bool 1020 depends on MMU 1021 select ARCH_HAS_ELF_RANDOMIZE 1022 1023config HAVE_STACK_VALIDATION 1024 bool 1025 help 1026 Architecture supports the 'objtool check' host tool command, which 1027 performs compile-time stack metadata validation. 1028 1029config HAVE_RELIABLE_STACKTRACE 1030 bool 1031 help 1032 Architecture has either save_stack_trace_tsk_reliable() or 1033 arch_stack_walk_reliable() function which only returns a stack trace 1034 if it can guarantee the trace is reliable. 1035 1036config HAVE_ARCH_HASH 1037 bool 1038 default n 1039 help 1040 If this is set, the architecture provides an <asm/hash.h> 1041 file which provides platform-specific implementations of some 1042 functions in <linux/hash.h> or fs/namei.c. 1043 1044config HAVE_ARCH_NVRAM_OPS 1045 bool 1046 1047config ISA_BUS_API 1048 def_bool ISA 1049 1050# 1051# ABI hall of shame 1052# 1053config CLONE_BACKWARDS 1054 bool 1055 help 1056 Architecture has tls passed as the 4th argument of clone(2), 1057 not the 5th one. 1058 1059config CLONE_BACKWARDS2 1060 bool 1061 help 1062 Architecture has the first two arguments of clone(2) swapped. 1063 1064config CLONE_BACKWARDS3 1065 bool 1066 help 1067 Architecture has tls passed as the 3rd argument of clone(2), 1068 not the 5th one. 1069 1070config ODD_RT_SIGACTION 1071 bool 1072 help 1073 Architecture has unusual rt_sigaction(2) arguments 1074 1075config OLD_SIGSUSPEND 1076 bool 1077 help 1078 Architecture has old sigsuspend(2) syscall, of one-argument variety 1079 1080config OLD_SIGSUSPEND3 1081 bool 1082 help 1083 Even weirder antique ABI - three-argument sigsuspend(2) 1084 1085config OLD_SIGACTION 1086 bool 1087 help 1088 Architecture has old sigaction(2) syscall. Nope, not the same 1089 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 1090 but fairly different variant of sigaction(2), thanks to OSF/1 1091 compatibility... 1092 1093config COMPAT_OLD_SIGACTION 1094 bool 1095 1096config COMPAT_32BIT_TIME 1097 bool "Provide system calls for 32-bit time_t" 1098 default !64BIT || COMPAT 1099 help 1100 This enables 32 bit time_t support in addition to 64 bit time_t support. 1101 This is relevant on all 32-bit architectures, and 64-bit architectures 1102 as part of compat syscall handling. 1103 1104config ARCH_NO_PREEMPT 1105 bool 1106 1107config ARCH_EPHEMERAL_INODES 1108 def_bool n 1109 help 1110 An arch should select this symbol if it doesn't keep track of inode 1111 instances on its own, but instead relies on something else (e.g. the 1112 host kernel for an UML kernel). 1113 1114config ARCH_SUPPORTS_RT 1115 bool 1116 1117config CPU_NO_EFFICIENT_FFS 1118 def_bool n 1119 1120config HAVE_ARCH_VMAP_STACK 1121 def_bool n 1122 help 1123 An arch should select this symbol if it can support kernel stacks 1124 in vmalloc space. This means: 1125 1126 - vmalloc space must be large enough to hold many kernel stacks. 1127 This may rule out many 32-bit architectures. 1128 1129 - Stacks in vmalloc space need to work reliably. For example, if 1130 vmap page tables are created on demand, either this mechanism 1131 needs to work while the stack points to a virtual address with 1132 unpopulated page tables or arch code (switch_to() and switch_mm(), 1133 most likely) needs to ensure that the stack's page table entries 1134 are populated before running on a possibly unpopulated stack. 1135 1136 - If the stack overflows into a guard page, something reasonable 1137 should happen. The definition of "reasonable" is flexible, but 1138 instantly rebooting without logging anything would be unfriendly. 1139 1140config VMAP_STACK 1141 default y 1142 bool "Use a virtually-mapped stack" 1143 depends on HAVE_ARCH_VMAP_STACK 1144 depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC 1145 help 1146 Enable this if you want the use virtually-mapped kernel stacks 1147 with guard pages. This causes kernel stack overflows to be 1148 caught immediately rather than causing difficult-to-diagnose 1149 corruption. 1150 1151 To use this with software KASAN modes, the architecture must support 1152 backing virtual mappings with real shadow memory, and KASAN_VMALLOC 1153 must be enabled. 1154 1155config HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 1156 def_bool n 1157 help 1158 An arch should select this symbol if it can support kernel stack 1159 offset randomization with calls to add_random_kstack_offset() 1160 during syscall entry and choose_random_kstack_offset() during 1161 syscall exit. Careful removal of -fstack-protector-strong and 1162 -fstack-protector should also be applied to the entry code and 1163 closely examined, as the artificial stack bump looks like an array 1164 to the compiler, so it will attempt to add canary checks regardless 1165 of the static branch state. 1166 1167config RANDOMIZE_KSTACK_OFFSET 1168 bool "Support for randomizing kernel stack offset on syscall entry" if EXPERT 1169 default y 1170 depends on HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 1171 depends on INIT_STACK_NONE || !CC_IS_CLANG || CLANG_VERSION >= 140000 1172 help 1173 The kernel stack offset can be randomized (after pt_regs) by 1174 roughly 5 bits of entropy, frustrating memory corruption 1175 attacks that depend on stack address determinism or 1176 cross-syscall address exposures. 1177 1178 The feature is controlled via the "randomize_kstack_offset=on/off" 1179 kernel boot param, and if turned off has zero overhead due to its use 1180 of static branches (see JUMP_LABEL). 1181 1182 If unsure, say Y. 1183 1184config RANDOMIZE_KSTACK_OFFSET_DEFAULT 1185 bool "Default state of kernel stack offset randomization" 1186 depends on RANDOMIZE_KSTACK_OFFSET 1187 help 1188 Kernel stack offset randomization is controlled by kernel boot param 1189 "randomize_kstack_offset=on/off", and this config chooses the default 1190 boot state. 1191 1192config ARCH_OPTIONAL_KERNEL_RWX 1193 def_bool n 1194 1195config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1196 def_bool n 1197 1198config ARCH_HAS_STRICT_KERNEL_RWX 1199 def_bool n 1200 1201config STRICT_KERNEL_RWX 1202 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 1203 depends on ARCH_HAS_STRICT_KERNEL_RWX 1204 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1205 help 1206 If this is set, kernel text and rodata memory will be made read-only, 1207 and non-text memory will be made non-executable. This provides 1208 protection against certain security exploits (e.g. executing the heap 1209 or modifying text) 1210 1211 These features are considered standard security practice these days. 1212 You should say Y here in almost all cases. 1213 1214config ARCH_HAS_STRICT_MODULE_RWX 1215 def_bool n 1216 1217config STRICT_MODULE_RWX 1218 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 1219 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 1220 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1221 help 1222 If this is set, module text and rodata memory will be made read-only, 1223 and non-text memory will be made non-executable. This provides 1224 protection against certain security exploits (e.g. writing to text) 1225 1226# select if the architecture provides an asm/dma-direct.h header 1227config ARCH_HAS_PHYS_TO_DMA 1228 bool 1229 1230config HAVE_ARCH_COMPILER_H 1231 bool 1232 help 1233 An architecture can select this if it provides an 1234 asm/compiler.h header that should be included after 1235 linux/compiler-*.h in order to override macro definitions that those 1236 headers generally provide. 1237 1238config HAVE_ARCH_PREL32_RELOCATIONS 1239 bool 1240 help 1241 May be selected by an architecture if it supports place-relative 1242 32-bit relocations, both in the toolchain and in the module loader, 1243 in which case relative references can be used in special sections 1244 for PCI fixup, initcalls etc which are only half the size on 64 bit 1245 architectures, and don't require runtime relocation on relocatable 1246 kernels. 1247 1248config ARCH_USE_MEMREMAP_PROT 1249 bool 1250 1251config LOCK_EVENT_COUNTS 1252 bool "Locking event counts collection" 1253 depends on DEBUG_FS 1254 help 1255 Enable light-weight counting of various locking related events 1256 in the system with minimal performance impact. This reduces 1257 the chance of application behavior change because of timing 1258 differences. The counts are reported via debugfs. 1259 1260# Select if the architecture has support for applying RELR relocations. 1261config ARCH_HAS_RELR 1262 bool 1263 1264config RELR 1265 bool "Use RELR relocation packing" 1266 depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR 1267 default y 1268 help 1269 Store the kernel's dynamic relocations in the RELR relocation packing 1270 format. Requires a compatible linker (LLD supports this feature), as 1271 well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy 1272 are compatible). 1273 1274config ARCH_HAS_MEM_ENCRYPT 1275 bool 1276 1277config ARCH_HAS_CC_PLATFORM 1278 bool 1279 1280config HAVE_SPARSE_SYSCALL_NR 1281 bool 1282 help 1283 An architecture should select this if its syscall numbering is sparse 1284 to save space. For example, MIPS architecture has a syscall array with 1285 entries at 4000, 5000 and 6000 locations. This option turns on syscall 1286 related optimizations for a given architecture. 1287 1288config ARCH_HAS_VDSO_DATA 1289 bool 1290 1291config HAVE_STATIC_CALL 1292 bool 1293 1294config HAVE_STATIC_CALL_INLINE 1295 bool 1296 depends on HAVE_STATIC_CALL 1297 1298config HAVE_PREEMPT_DYNAMIC 1299 bool 1300 1301config HAVE_PREEMPT_DYNAMIC_CALL 1302 bool 1303 depends on HAVE_STATIC_CALL 1304 select HAVE_PREEMPT_DYNAMIC 1305 help 1306 An architecture should select this if it can handle the preemption 1307 model being selected at boot time using static calls. 1308 1309 Where an architecture selects HAVE_STATIC_CALL_INLINE, any call to a 1310 preemption function will be patched directly. 1311 1312 Where an architecture does not select HAVE_STATIC_CALL_INLINE, any 1313 call to a preemption function will go through a trampoline, and the 1314 trampoline will be patched. 1315 1316 It is strongly advised to support inline static call to avoid any 1317 overhead. 1318 1319config HAVE_PREEMPT_DYNAMIC_KEY 1320 bool 1321 depends on HAVE_ARCH_JUMP_LABEL && CC_HAS_ASM_GOTO 1322 select HAVE_PREEMPT_DYNAMIC 1323 help 1324 An architecture should select this if it can handle the preemption 1325 model being selected at boot time using static keys. 1326 1327 Each preemption function will be given an early return based on a 1328 static key. This should have slightly lower overhead than non-inline 1329 static calls, as this effectively inlines each trampoline into the 1330 start of its callee. This may avoid redundant work, and may 1331 integrate better with CFI schemes. 1332 1333 This will have greater overhead than using inline static calls as 1334 the call to the preemption function cannot be entirely elided. 1335 1336config ARCH_WANT_LD_ORPHAN_WARN 1337 bool 1338 help 1339 An arch should select this symbol once all linker sections are explicitly 1340 included, size-asserted, or discarded in the linker scripts. This is 1341 important because we never want expected sections to be placed heuristically 1342 by the linker, since the locations of such sections can change between linker 1343 versions. 1344 1345config HAVE_ARCH_PFN_VALID 1346 bool 1347 1348config ARCH_SUPPORTS_DEBUG_PAGEALLOC 1349 bool 1350 1351config ARCH_SUPPORTS_PAGE_TABLE_CHECK 1352 bool 1353 1354config ARCH_SPLIT_ARG64 1355 bool 1356 help 1357 If a 32-bit architecture requires 64-bit arguments to be split into 1358 pairs of 32-bit arguments, select this option. 1359 1360config ARCH_HAS_ELFCORE_COMPAT 1361 bool 1362 1363config ARCH_HAS_PARANOID_L1D_FLUSH 1364 bool 1365 1366config DYNAMIC_SIGFRAME 1367 bool 1368 1369# Select, if arch has a named attribute group bound to NUMA device nodes. 1370config HAVE_ARCH_NODE_DEV_GROUP 1371 bool 1372 1373source "kernel/gcov/Kconfig" 1374 1375source "scripts/gcc-plugins/Kconfig" 1376 1377endmenu 1378