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 ARCH_HAS_SUBPAGE_FAULTS 15 bool 16 help 17 Select if the architecture can check permissions at sub-page 18 granularity (e.g. arm64 MTE). The probe_user_*() functions 19 must be implemented. 20 21config HOTPLUG_SMT 22 bool 23 24# Selected by HOTPLUG_CORE_SYNC_DEAD or HOTPLUG_CORE_SYNC_FULL 25config HOTPLUG_CORE_SYNC 26 bool 27 28# Basic CPU dead synchronization selected by architecture 29config HOTPLUG_CORE_SYNC_DEAD 30 bool 31 select HOTPLUG_CORE_SYNC 32 33# Full CPU synchronization with alive state selected by architecture 34config HOTPLUG_CORE_SYNC_FULL 35 bool 36 select HOTPLUG_CORE_SYNC_DEAD if HOTPLUG_CPU 37 select HOTPLUG_CORE_SYNC 38 39config HOTPLUG_SPLIT_STARTUP 40 bool 41 select HOTPLUG_CORE_SYNC_FULL 42 43config HOTPLUG_PARALLEL 44 bool 45 select HOTPLUG_SPLIT_STARTUP 46 47config GENERIC_ENTRY 48 bool 49 50config KPROBES 51 bool "Kprobes" 52 depends on MODULES 53 depends on HAVE_KPROBES 54 select KALLSYMS 55 select TASKS_RCU if PREEMPTION 56 help 57 Kprobes allows you to trap at almost any kernel address and 58 execute a callback function. register_kprobe() establishes 59 a probepoint and specifies the callback. Kprobes is useful 60 for kernel debugging, non-intrusive instrumentation and testing. 61 If in doubt, say "N". 62 63config JUMP_LABEL 64 bool "Optimize very unlikely/likely branches" 65 depends on HAVE_ARCH_JUMP_LABEL 66 select OBJTOOL if HAVE_JUMP_LABEL_HACK 67 help 68 This option enables a transparent branch optimization that 69 makes certain almost-always-true or almost-always-false branch 70 conditions even cheaper to execute within the kernel. 71 72 Certain performance-sensitive kernel code, such as trace points, 73 scheduler functionality, networking code and KVM have such 74 branches and include support for this optimization technique. 75 76 If it is detected that the compiler has support for "asm goto", 77 the kernel will compile such branches with just a nop 78 instruction. When the condition flag is toggled to true, the 79 nop will be converted to a jump instruction to execute the 80 conditional block of instructions. 81 82 This technique lowers overhead and stress on the branch prediction 83 of the processor and generally makes the kernel faster. The update 84 of the condition is slower, but those are always very rare. 85 86 ( On 32-bit x86, the necessary options added to the compiler 87 flags may increase the size of the kernel slightly. ) 88 89config STATIC_KEYS_SELFTEST 90 bool "Static key selftest" 91 depends on JUMP_LABEL 92 help 93 Boot time self-test of the branch patching code. 94 95config STATIC_CALL_SELFTEST 96 bool "Static call selftest" 97 depends on HAVE_STATIC_CALL 98 help 99 Boot time self-test of the call patching code. 100 101config OPTPROBES 102 def_bool y 103 depends on KPROBES && HAVE_OPTPROBES 104 select TASKS_RCU if PREEMPTION 105 106config KPROBES_ON_FTRACE 107 def_bool y 108 depends on KPROBES && HAVE_KPROBES_ON_FTRACE 109 depends on DYNAMIC_FTRACE_WITH_REGS 110 help 111 If function tracer is enabled and the arch supports full 112 passing of pt_regs to function tracing, then kprobes can 113 optimize on top of function tracing. 114 115config UPROBES 116 def_bool n 117 depends on ARCH_SUPPORTS_UPROBES 118 help 119 Uprobes is the user-space counterpart to kprobes: they 120 enable instrumentation applications (such as 'perf probe') 121 to establish unintrusive probes in user-space binaries and 122 libraries, by executing handler functions when the probes 123 are hit by user-space applications. 124 125 ( These probes come in the form of single-byte breakpoints, 126 managed by the kernel and kept transparent to the probed 127 application. ) 128 129config HAVE_64BIT_ALIGNED_ACCESS 130 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 131 help 132 Some architectures require 64 bit accesses to be 64 bit 133 aligned, which also requires structs containing 64 bit values 134 to be 64 bit aligned too. This includes some 32 bit 135 architectures which can do 64 bit accesses, as well as 64 bit 136 architectures without unaligned access. 137 138 This symbol should be selected by an architecture if 64 bit 139 accesses are required to be 64 bit aligned in this way even 140 though it is not a 64 bit architecture. 141 142 See Documentation/core-api/unaligned-memory-access.rst for 143 more information on the topic of unaligned memory accesses. 144 145config HAVE_EFFICIENT_UNALIGNED_ACCESS 146 bool 147 help 148 Some architectures are unable to perform unaligned accesses 149 without the use of get_unaligned/put_unaligned. Others are 150 unable to perform such accesses efficiently (e.g. trap on 151 unaligned access and require fixing it up in the exception 152 handler.) 153 154 This symbol should be selected by an architecture if it can 155 perform unaligned accesses efficiently to allow different 156 code paths to be selected for these cases. Some network 157 drivers, for example, could opt to not fix up alignment 158 problems with received packets if doing so would not help 159 much. 160 161 See Documentation/core-api/unaligned-memory-access.rst for more 162 information on the topic of unaligned memory accesses. 163 164config ARCH_USE_BUILTIN_BSWAP 165 bool 166 help 167 Modern versions of GCC (since 4.4) have builtin functions 168 for handling byte-swapping. Using these, instead of the old 169 inline assembler that the architecture code provides in the 170 __arch_bswapXX() macros, allows the compiler to see what's 171 happening and offers more opportunity for optimisation. In 172 particular, the compiler will be able to combine the byteswap 173 with a nearby load or store and use load-and-swap or 174 store-and-swap instructions if the architecture has them. It 175 should almost *never* result in code which is worse than the 176 hand-coded assembler in <asm/swab.h>. But just in case it 177 does, the use of the builtins is optional. 178 179 Any architecture with load-and-swap or store-and-swap 180 instructions should set this. And it shouldn't hurt to set it 181 on architectures that don't have such instructions. 182 183config KRETPROBES 184 def_bool y 185 depends on KPROBES && (HAVE_KRETPROBES || HAVE_RETHOOK) 186 187config KRETPROBE_ON_RETHOOK 188 def_bool y 189 depends on HAVE_RETHOOK 190 depends on KRETPROBES 191 select RETHOOK 192 193config USER_RETURN_NOTIFIER 194 bool 195 depends on HAVE_USER_RETURN_NOTIFIER 196 help 197 Provide a kernel-internal notification when a cpu is about to 198 switch to user mode. 199 200config HAVE_IOREMAP_PROT 201 bool 202 203config HAVE_KPROBES 204 bool 205 206config HAVE_KRETPROBES 207 bool 208 209config HAVE_OPTPROBES 210 bool 211 212config HAVE_KPROBES_ON_FTRACE 213 bool 214 215config ARCH_CORRECT_STACKTRACE_ON_KRETPROBE 216 bool 217 help 218 Since kretprobes modifies return address on the stack, the 219 stacktrace may see the kretprobe trampoline address instead 220 of correct one. If the architecture stacktrace code and 221 unwinder can adjust such entries, select this configuration. 222 223config HAVE_FUNCTION_ERROR_INJECTION 224 bool 225 226config HAVE_NMI 227 bool 228 229config HAVE_FUNCTION_DESCRIPTORS 230 bool 231 232config TRACE_IRQFLAGS_SUPPORT 233 bool 234 235config TRACE_IRQFLAGS_NMI_SUPPORT 236 bool 237 238# 239# An arch should select this if it provides all these things: 240# 241# task_pt_regs() in asm/processor.h or asm/ptrace.h 242# arch_has_single_step() if there is hardware single-step support 243# arch_has_block_step() if there is hardware block-step support 244# asm/syscall.h supplying asm-generic/syscall.h interface 245# linux/regset.h user_regset interfaces 246# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 247# TIF_SYSCALL_TRACE calls ptrace_report_syscall_{entry,exit} 248# TIF_NOTIFY_RESUME calls resume_user_mode_work() 249# 250config HAVE_ARCH_TRACEHOOK 251 bool 252 253config HAVE_DMA_CONTIGUOUS 254 bool 255 256config GENERIC_SMP_IDLE_THREAD 257 bool 258 259config GENERIC_IDLE_POLL_SETUP 260 bool 261 262config ARCH_HAS_FORTIFY_SOURCE 263 bool 264 help 265 An architecture should select this when it can successfully 266 build and run with CONFIG_FORTIFY_SOURCE. 267 268# 269# Select if the arch provides a historic keepinit alias for the retain_initrd 270# command line option 271# 272config ARCH_HAS_KEEPINITRD 273 bool 274 275# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 276config ARCH_HAS_SET_MEMORY 277 bool 278 279# Select if arch has all set_direct_map_invalid/default() functions 280config ARCH_HAS_SET_DIRECT_MAP 281 bool 282 283# 284# Select if the architecture provides the arch_dma_set_uncached symbol to 285# either provide an uncached segment alias for a DMA allocation, or 286# to remap the page tables in place. 287# 288config ARCH_HAS_DMA_SET_UNCACHED 289 bool 290 291# 292# Select if the architectures provides the arch_dma_clear_uncached symbol 293# to undo an in-place page table remap for uncached access. 294# 295config ARCH_HAS_DMA_CLEAR_UNCACHED 296 bool 297 298config ARCH_HAS_CPU_FINALIZE_INIT 299 bool 300 301# Select if arch init_task must go in the __init_task_data section 302config ARCH_TASK_STRUCT_ON_STACK 303 bool 304 305# Select if arch has its private alloc_task_struct() function 306config ARCH_TASK_STRUCT_ALLOCATOR 307 bool 308 309config HAVE_ARCH_THREAD_STRUCT_WHITELIST 310 bool 311 depends on !ARCH_TASK_STRUCT_ALLOCATOR 312 help 313 An architecture should select this to provide hardened usercopy 314 knowledge about what region of the thread_struct should be 315 whitelisted for copying to userspace. Normally this is only the 316 FPU registers. Specifically, arch_thread_struct_whitelist() 317 should be implemented. Without this, the entire thread_struct 318 field in task_struct will be left whitelisted. 319 320# Select if arch has its private alloc_thread_stack() function 321config ARCH_THREAD_STACK_ALLOCATOR 322 bool 323 324# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 325config ARCH_WANTS_DYNAMIC_TASK_STRUCT 326 bool 327 328config ARCH_WANTS_NO_INSTR 329 bool 330 help 331 An architecture should select this if the noinstr macro is being used on 332 functions to denote that the toolchain should avoid instrumenting such 333 functions and is required for correctness. 334 335config ARCH_32BIT_OFF_T 336 bool 337 depends on !64BIT 338 help 339 All new 32-bit architectures should have 64-bit off_t type on 340 userspace side which corresponds to the loff_t kernel type. This 341 is the requirement for modern ABIs. Some existing architectures 342 still support 32-bit off_t. This option is enabled for all such 343 architectures explicitly. 344 345# Selected by 64 bit architectures which have a 32 bit f_tinode in struct ustat 346config ARCH_32BIT_USTAT_F_TINODE 347 bool 348 349config HAVE_ASM_MODVERSIONS 350 bool 351 help 352 This symbol should be selected by an architecture if it provides 353 <asm/asm-prototypes.h> to support the module versioning for symbols 354 exported from assembly code. 355 356config HAVE_REGS_AND_STACK_ACCESS_API 357 bool 358 help 359 This symbol should be selected by an architecture if it supports 360 the API needed to access registers and stack entries from pt_regs, 361 declared in asm/ptrace.h 362 For example the kprobes-based event tracer needs this API. 363 364config HAVE_RSEQ 365 bool 366 depends on HAVE_REGS_AND_STACK_ACCESS_API 367 help 368 This symbol should be selected by an architecture if it 369 supports an implementation of restartable sequences. 370 371config HAVE_RUST 372 bool 373 help 374 This symbol should be selected by an architecture if it 375 supports Rust. 376 377config HAVE_FUNCTION_ARG_ACCESS_API 378 bool 379 help 380 This symbol should be selected by an architecture if it supports 381 the API needed to access function arguments from pt_regs, 382 declared in asm/ptrace.h 383 384config HAVE_HW_BREAKPOINT 385 bool 386 depends on PERF_EVENTS 387 388config HAVE_MIXED_BREAKPOINTS_REGS 389 bool 390 depends on HAVE_HW_BREAKPOINT 391 help 392 Depending on the arch implementation of hardware breakpoints, 393 some of them have separate registers for data and instruction 394 breakpoints addresses, others have mixed registers to store 395 them but define the access type in a control register. 396 Select this option if your arch implements breakpoints under the 397 latter fashion. 398 399config HAVE_USER_RETURN_NOTIFIER 400 bool 401 402config HAVE_PERF_EVENTS_NMI 403 bool 404 help 405 System hardware can generate an NMI using the perf event 406 subsystem. Also has support for calculating CPU cycle events 407 to determine how many clock cycles in a given period. 408 409config HAVE_HARDLOCKUP_DETECTOR_PERF 410 bool 411 depends on HAVE_PERF_EVENTS_NMI 412 help 413 The arch chooses to use the generic perf-NMI-based hardlockup 414 detector. Must define HAVE_PERF_EVENTS_NMI. 415 416config HAVE_HARDLOCKUP_DETECTOR_ARCH 417 bool 418 help 419 The arch provides its own hardlockup detector implementation instead 420 of the generic ones. 421 422 It uses the same command line parameters, and sysctl interface, 423 as the generic hardlockup detectors. 424 425config HAVE_PERF_REGS 426 bool 427 help 428 Support selective register dumps for perf events. This includes 429 bit-mapping of each registers and a unique architecture id. 430 431config HAVE_PERF_USER_STACK_DUMP 432 bool 433 help 434 Support user stack dumps for perf event samples. This needs 435 access to the user stack pointer which is not unified across 436 architectures. 437 438config HAVE_ARCH_JUMP_LABEL 439 bool 440 441config HAVE_ARCH_JUMP_LABEL_RELATIVE 442 bool 443 444config MMU_GATHER_TABLE_FREE 445 bool 446 447config MMU_GATHER_RCU_TABLE_FREE 448 bool 449 select MMU_GATHER_TABLE_FREE 450 451config MMU_GATHER_PAGE_SIZE 452 bool 453 454config MMU_GATHER_NO_RANGE 455 bool 456 select MMU_GATHER_MERGE_VMAS 457 458config MMU_GATHER_NO_FLUSH_CACHE 459 bool 460 461config MMU_GATHER_MERGE_VMAS 462 bool 463 464config MMU_GATHER_NO_GATHER 465 bool 466 depends on MMU_GATHER_TABLE_FREE 467 468config ARCH_WANT_IRQS_OFF_ACTIVATE_MM 469 bool 470 help 471 Temporary select until all architectures can be converted to have 472 irqs disabled over activate_mm. Architectures that do IPI based TLB 473 shootdowns should enable this. 474 475# Use normal mm refcounting for MMU_LAZY_TLB kernel thread references. 476# MMU_LAZY_TLB_REFCOUNT=n can improve the scalability of context switching 477# to/from kernel threads when the same mm is running on a lot of CPUs (a large 478# multi-threaded application), by reducing contention on the mm refcount. 479# 480# This can be disabled if the architecture ensures no CPUs are using an mm as a 481# "lazy tlb" beyond its final refcount (i.e., by the time __mmdrop frees the mm 482# or its kernel page tables). This could be arranged by arch_exit_mmap(), or 483# final exit(2) TLB flush, for example. 484# 485# To implement this, an arch *must*: 486# Ensure the _lazy_tlb variants of mmgrab/mmdrop are used when manipulating 487# the lazy tlb reference of a kthread's ->active_mm (non-arch code has been 488# converted already). 489config MMU_LAZY_TLB_REFCOUNT 490 def_bool y 491 depends on !MMU_LAZY_TLB_SHOOTDOWN 492 493# This option allows MMU_LAZY_TLB_REFCOUNT=n. It ensures no CPUs are using an 494# mm as a lazy tlb beyond its last reference count, by shooting down these 495# users before the mm is deallocated. __mmdrop() first IPIs all CPUs that may 496# be using the mm as a lazy tlb, so that they may switch themselves to using 497# init_mm for their active mm. mm_cpumask(mm) is used to determine which CPUs 498# may be using mm as a lazy tlb mm. 499# 500# To implement this, an arch *must*: 501# - At the time of the final mmdrop of the mm, ensure mm_cpumask(mm) contains 502# at least all possible CPUs in which the mm is lazy. 503# - It must meet the requirements for MMU_LAZY_TLB_REFCOUNT=n (see above). 504config MMU_LAZY_TLB_SHOOTDOWN 505 bool 506 507config ARCH_HAVE_NMI_SAFE_CMPXCHG 508 bool 509 510config ARCH_HAS_NMI_SAFE_THIS_CPU_OPS 511 bool 512 513config HAVE_ALIGNED_STRUCT_PAGE 514 bool 515 help 516 This makes sure that struct pages are double word aligned and that 517 e.g. the SLUB allocator can perform double word atomic operations 518 on a struct page for better performance. However selecting this 519 might increase the size of a struct page by a word. 520 521config HAVE_CMPXCHG_LOCAL 522 bool 523 524config HAVE_CMPXCHG_DOUBLE 525 bool 526 527config ARCH_WEAK_RELEASE_ACQUIRE 528 bool 529 530config ARCH_WANT_IPC_PARSE_VERSION 531 bool 532 533config ARCH_WANT_COMPAT_IPC_PARSE_VERSION 534 bool 535 536config ARCH_WANT_OLD_COMPAT_IPC 537 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 538 bool 539 540config HAVE_ARCH_SECCOMP 541 bool 542 help 543 An arch should select this symbol to support seccomp mode 1 (the fixed 544 syscall policy), and must provide an overrides for __NR_seccomp_sigreturn, 545 and compat syscalls if the asm-generic/seccomp.h defaults need adjustment: 546 - __NR_seccomp_read_32 547 - __NR_seccomp_write_32 548 - __NR_seccomp_exit_32 549 - __NR_seccomp_sigreturn_32 550 551config HAVE_ARCH_SECCOMP_FILTER 552 bool 553 select HAVE_ARCH_SECCOMP 554 help 555 An arch should select this symbol if it provides all of these things: 556 - all the requirements for HAVE_ARCH_SECCOMP 557 - syscall_get_arch() 558 - syscall_get_arguments() 559 - syscall_rollback() 560 - syscall_set_return_value() 561 - SIGSYS siginfo_t support 562 - secure_computing is called from a ptrace_event()-safe context 563 - secure_computing return value is checked and a return value of -1 564 results in the system call being skipped immediately. 565 - seccomp syscall wired up 566 - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE, 567 SECCOMP_ARCH_NATIVE_NR, SECCOMP_ARCH_NATIVE_NAME defined. If 568 COMPAT is supported, have the SECCOMP_ARCH_COMPAT* defines too. 569 570config SECCOMP 571 prompt "Enable seccomp to safely execute untrusted bytecode" 572 def_bool y 573 depends on HAVE_ARCH_SECCOMP 574 help 575 This kernel feature is useful for number crunching applications 576 that may need to handle untrusted bytecode during their 577 execution. By using pipes or other transports made available 578 to the process as file descriptors supporting the read/write 579 syscalls, it's possible to isolate those applications in their 580 own address space using seccomp. Once seccomp is enabled via 581 prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be 582 disabled and the task is only allowed to execute a few safe 583 syscalls defined by each seccomp mode. 584 585 If unsure, say Y. 586 587config SECCOMP_FILTER 588 def_bool y 589 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 590 help 591 Enable tasks to build secure computing environments defined 592 in terms of Berkeley Packet Filter programs which implement 593 task-defined system call filtering polices. 594 595 See Documentation/userspace-api/seccomp_filter.rst for details. 596 597config SECCOMP_CACHE_DEBUG 598 bool "Show seccomp filter cache status in /proc/pid/seccomp_cache" 599 depends on SECCOMP_FILTER && !HAVE_SPARSE_SYSCALL_NR 600 depends on PROC_FS 601 help 602 This enables the /proc/pid/seccomp_cache interface to monitor 603 seccomp cache data. The file format is subject to change. Reading 604 the file requires CAP_SYS_ADMIN. 605 606 This option is for debugging only. Enabling presents the risk that 607 an adversary may be able to infer the seccomp filter logic. 608 609 If unsure, say N. 610 611config HAVE_ARCH_STACKLEAK 612 bool 613 help 614 An architecture should select this if it has the code which 615 fills the used part of the kernel stack with the STACKLEAK_POISON 616 value before returning from system calls. 617 618config HAVE_STACKPROTECTOR 619 bool 620 help 621 An arch should select this symbol if: 622 - it has implemented a stack canary (e.g. __stack_chk_guard) 623 624config STACKPROTECTOR 625 bool "Stack Protector buffer overflow detection" 626 depends on HAVE_STACKPROTECTOR 627 depends on $(cc-option,-fstack-protector) 628 default y 629 help 630 This option turns on the "stack-protector" GCC feature. This 631 feature puts, at the beginning of functions, a canary value on 632 the stack just before the return address, and validates 633 the value just before actually returning. Stack based buffer 634 overflows (that need to overwrite this return address) now also 635 overwrite the canary, which gets detected and the attack is then 636 neutralized via a kernel panic. 637 638 Functions will have the stack-protector canary logic added if they 639 have an 8-byte or larger character array on the stack. 640 641 This feature requires gcc version 4.2 or above, or a distribution 642 gcc with the feature backported ("-fstack-protector"). 643 644 On an x86 "defconfig" build, this feature adds canary checks to 645 about 3% of all kernel functions, which increases kernel code size 646 by about 0.3%. 647 648config STACKPROTECTOR_STRONG 649 bool "Strong Stack Protector" 650 depends on STACKPROTECTOR 651 depends on $(cc-option,-fstack-protector-strong) 652 default y 653 help 654 Functions will have the stack-protector canary logic added in any 655 of the following conditions: 656 657 - local variable's address used as part of the right hand side of an 658 assignment or function argument 659 - local variable is an array (or union containing an array), 660 regardless of array type or length 661 - uses register local variables 662 663 This feature requires gcc version 4.9 or above, or a distribution 664 gcc with the feature backported ("-fstack-protector-strong"). 665 666 On an x86 "defconfig" build, this feature adds canary checks to 667 about 20% of all kernel functions, which increases the kernel code 668 size by about 2%. 669 670config ARCH_SUPPORTS_SHADOW_CALL_STACK 671 bool 672 help 673 An architecture should select this if it supports the compiler's 674 Shadow Call Stack and implements runtime support for shadow stack 675 switching. 676 677config SHADOW_CALL_STACK 678 bool "Shadow Call Stack" 679 depends on ARCH_SUPPORTS_SHADOW_CALL_STACK 680 depends on DYNAMIC_FTRACE_WITH_ARGS || DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER 681 help 682 This option enables the compiler's Shadow Call Stack, which 683 uses a shadow stack to protect function return addresses from 684 being overwritten by an attacker. More information can be found 685 in the compiler's documentation: 686 687 - Clang: https://clang.llvm.org/docs/ShadowCallStack.html 688 - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options 689 690 Note that security guarantees in the kernel differ from the 691 ones documented for user space. The kernel must store addresses 692 of shadow stacks in memory, which means an attacker capable of 693 reading and writing arbitrary memory may be able to locate them 694 and hijack control flow by modifying the stacks. 695 696config DYNAMIC_SCS 697 bool 698 help 699 Set by the arch code if it relies on code patching to insert the 700 shadow call stack push and pop instructions rather than on the 701 compiler. 702 703config LTO 704 bool 705 help 706 Selected if the kernel will be built using the compiler's LTO feature. 707 708config LTO_CLANG 709 bool 710 select LTO 711 help 712 Selected if the kernel will be built using Clang's LTO feature. 713 714config ARCH_SUPPORTS_LTO_CLANG 715 bool 716 help 717 An architecture should select this option if it supports: 718 - compiling with Clang, 719 - compiling inline assembly with Clang's integrated assembler, 720 - and linking with LLD. 721 722config ARCH_SUPPORTS_LTO_CLANG_THIN 723 bool 724 help 725 An architecture should select this option if it can support Clang's 726 ThinLTO mode. 727 728config HAS_LTO_CLANG 729 def_bool y 730 depends on CC_IS_CLANG && LD_IS_LLD && AS_IS_LLVM 731 depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm) 732 depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm) 733 depends on ARCH_SUPPORTS_LTO_CLANG 734 depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT 735 # https://github.com/ClangBuiltLinux/linux/issues/1721 736 depends on (!KASAN || KASAN_HW_TAGS || CLANG_VERSION >= 170000) || !DEBUG_INFO 737 depends on (!KCOV || CLANG_VERSION >= 170000) || !DEBUG_INFO 738 depends on !GCOV_KERNEL 739 help 740 The compiler and Kconfig options support building with Clang's 741 LTO. 742 743choice 744 prompt "Link Time Optimization (LTO)" 745 default LTO_NONE 746 help 747 This option enables Link Time Optimization (LTO), which allows the 748 compiler to optimize binaries globally. 749 750 If unsure, select LTO_NONE. Note that LTO is very resource-intensive 751 so it's disabled by default. 752 753config LTO_NONE 754 bool "None" 755 help 756 Build the kernel normally, without Link Time Optimization (LTO). 757 758config LTO_CLANG_FULL 759 bool "Clang Full LTO (EXPERIMENTAL)" 760 depends on HAS_LTO_CLANG 761 depends on !COMPILE_TEST 762 select LTO_CLANG 763 help 764 This option enables Clang's full Link Time Optimization (LTO), which 765 allows the compiler to optimize the kernel globally. If you enable 766 this option, the compiler generates LLVM bitcode instead of ELF 767 object files, and the actual compilation from bitcode happens at 768 the LTO link step, which may take several minutes depending on the 769 kernel configuration. More information can be found from LLVM's 770 documentation: 771 772 https://llvm.org/docs/LinkTimeOptimization.html 773 774 During link time, this option can use a large amount of RAM, and 775 may take much longer than the ThinLTO option. 776 777config LTO_CLANG_THIN 778 bool "Clang ThinLTO (EXPERIMENTAL)" 779 depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN 780 select LTO_CLANG 781 help 782 This option enables Clang's ThinLTO, which allows for parallel 783 optimization and faster incremental compiles compared to the 784 CONFIG_LTO_CLANG_FULL option. More information can be found 785 from Clang's documentation: 786 787 https://clang.llvm.org/docs/ThinLTO.html 788 789 If unsure, say Y. 790endchoice 791 792config ARCH_SUPPORTS_CFI_CLANG 793 bool 794 help 795 An architecture should select this option if it can support Clang's 796 Control-Flow Integrity (CFI) checking. 797 798config ARCH_USES_CFI_TRAPS 799 bool 800 801config CFI_CLANG 802 bool "Use Clang's Control Flow Integrity (CFI)" 803 depends on ARCH_SUPPORTS_CFI_CLANG 804 depends on $(cc-option,-fsanitize=kcfi) 805 help 806 This option enables Clang’s forward-edge Control Flow Integrity 807 (CFI) checking, where the compiler injects a runtime check to each 808 indirect function call to ensure the target is a valid function with 809 the correct static type. This restricts possible call targets and 810 makes it more difficult for an attacker to exploit bugs that allow 811 the modification of stored function pointers. More information can be 812 found from Clang's documentation: 813 814 https://clang.llvm.org/docs/ControlFlowIntegrity.html 815 816config CFI_PERMISSIVE 817 bool "Use CFI in permissive mode" 818 depends on CFI_CLANG 819 help 820 When selected, Control Flow Integrity (CFI) violations result in a 821 warning instead of a kernel panic. This option should only be used 822 for finding indirect call type mismatches during development. 823 824 If unsure, say N. 825 826config HAVE_ARCH_WITHIN_STACK_FRAMES 827 bool 828 help 829 An architecture should select this if it can walk the kernel stack 830 frames to determine if an object is part of either the arguments 831 or local variables (i.e. that it excludes saved return addresses, 832 and similar) by implementing an inline arch_within_stack_frames(), 833 which is used by CONFIG_HARDENED_USERCOPY. 834 835config HAVE_CONTEXT_TRACKING_USER 836 bool 837 help 838 Provide kernel/user boundaries probes necessary for subsystems 839 that need it, such as userspace RCU extended quiescent state. 840 Syscalls need to be wrapped inside user_exit()-user_enter(), either 841 optimized behind static key or through the slow path using TIF_NOHZ 842 flag. Exceptions handlers must be wrapped as well. Irqs are already 843 protected inside ct_irq_enter/ct_irq_exit() but preemption or signal 844 handling on irq exit still need to be protected. 845 846config HAVE_CONTEXT_TRACKING_USER_OFFSTACK 847 bool 848 help 849 Architecture neither relies on exception_enter()/exception_exit() 850 nor on schedule_user(). Also preempt_schedule_notrace() and 851 preempt_schedule_irq() can't be called in a preemptible section 852 while context tracking is CONTEXT_USER. This feature reflects a sane 853 entry implementation where the following requirements are met on 854 critical entry code, ie: before user_exit() or after user_enter(): 855 856 - Critical entry code isn't preemptible (or better yet: 857 not interruptible). 858 - No use of RCU read side critical sections, unless ct_nmi_enter() 859 got called. 860 - No use of instrumentation, unless instrumentation_begin() got 861 called. 862 863config HAVE_TIF_NOHZ 864 bool 865 help 866 Arch relies on TIF_NOHZ and syscall slow path to implement context 867 tracking calls to user_enter()/user_exit(). 868 869config HAVE_VIRT_CPU_ACCOUNTING 870 bool 871 872config HAVE_VIRT_CPU_ACCOUNTING_IDLE 873 bool 874 help 875 Architecture has its own way to account idle CPU time and therefore 876 doesn't implement vtime_account_idle(). 877 878config ARCH_HAS_SCALED_CPUTIME 879 bool 880 881config HAVE_VIRT_CPU_ACCOUNTING_GEN 882 bool 883 default y if 64BIT 884 help 885 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 886 Before enabling this option, arch code must be audited 887 to ensure there are no races in concurrent read/write of 888 cputime_t. For example, reading/writing 64-bit cputime_t on 889 some 32-bit arches may require multiple accesses, so proper 890 locking is needed to protect against concurrent accesses. 891 892config HAVE_IRQ_TIME_ACCOUNTING 893 bool 894 help 895 Archs need to ensure they use a high enough resolution clock to 896 support irq time accounting and then call enable_sched_clock_irqtime(). 897 898config HAVE_MOVE_PUD 899 bool 900 help 901 Architectures that select this are able to move page tables at the 902 PUD level. If there are only 3 page table levels, the move effectively 903 happens at the PGD level. 904 905config HAVE_MOVE_PMD 906 bool 907 help 908 Archs that select this are able to move page tables at the PMD level. 909 910config HAVE_ARCH_TRANSPARENT_HUGEPAGE 911 bool 912 913config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 914 bool 915 916config HAVE_ARCH_HUGE_VMAP 917 bool 918 919# 920# Archs that select this would be capable of PMD-sized vmaps (i.e., 921# arch_vmap_pmd_supported() returns true). The VM_ALLOW_HUGE_VMAP flag 922# must be used to enable allocations to use hugepages. 923# 924config HAVE_ARCH_HUGE_VMALLOC 925 depends on HAVE_ARCH_HUGE_VMAP 926 bool 927 928config ARCH_WANT_HUGE_PMD_SHARE 929 bool 930 931config HAVE_ARCH_SOFT_DIRTY 932 bool 933 934config HAVE_MOD_ARCH_SPECIFIC 935 bool 936 help 937 The arch uses struct mod_arch_specific to store data. Many arches 938 just need a simple module loader without arch specific data - those 939 should not enable this. 940 941config MODULES_USE_ELF_RELA 942 bool 943 help 944 Modules only use ELF RELA relocations. Modules with ELF REL 945 relocations will give an error. 946 947config MODULES_USE_ELF_REL 948 bool 949 help 950 Modules only use ELF REL relocations. Modules with ELF RELA 951 relocations will give an error. 952 953config ARCH_WANTS_MODULES_DATA_IN_VMALLOC 954 bool 955 help 956 For architectures like powerpc/32 which have constraints on module 957 allocation and need to allocate module data outside of module area. 958 959config HAVE_IRQ_EXIT_ON_IRQ_STACK 960 bool 961 help 962 Architecture doesn't only execute the irq handler on the irq stack 963 but also irq_exit(). This way we can process softirqs on this irq 964 stack instead of switching to a new one when we call __do_softirq() 965 in the end of an hardirq. 966 This spares a stack switch and improves cache usage on softirq 967 processing. 968 969config HAVE_SOFTIRQ_ON_OWN_STACK 970 bool 971 help 972 Architecture provides a function to run __do_softirq() on a 973 separate stack. 974 975config SOFTIRQ_ON_OWN_STACK 976 def_bool HAVE_SOFTIRQ_ON_OWN_STACK && !PREEMPT_RT 977 978config ALTERNATE_USER_ADDRESS_SPACE 979 bool 980 help 981 Architectures set this when the CPU uses separate address 982 spaces for kernel and user space pointers. In this case, the 983 access_ok() check on a __user pointer is skipped. 984 985config PGTABLE_LEVELS 986 int 987 default 2 988 989config ARCH_HAS_ELF_RANDOMIZE 990 bool 991 help 992 An architecture supports choosing randomized locations for 993 stack, mmap, brk, and ET_DYN. Defined functions: 994 - arch_mmap_rnd() 995 - arch_randomize_brk() 996 997config HAVE_ARCH_MMAP_RND_BITS 998 bool 999 help 1000 An arch should select this symbol if it supports setting a variable 1001 number of bits for use in establishing the base address for mmap 1002 allocations, has MMU enabled and provides values for both: 1003 - ARCH_MMAP_RND_BITS_MIN 1004 - ARCH_MMAP_RND_BITS_MAX 1005 1006config HAVE_EXIT_THREAD 1007 bool 1008 help 1009 An architecture implements exit_thread. 1010 1011config ARCH_MMAP_RND_BITS_MIN 1012 int 1013 1014config ARCH_MMAP_RND_BITS_MAX 1015 int 1016 1017config ARCH_MMAP_RND_BITS_DEFAULT 1018 int 1019 1020config ARCH_MMAP_RND_BITS 1021 int "Number of bits to use for ASLR of mmap base address" if EXPERT 1022 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 1023 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 1024 default ARCH_MMAP_RND_BITS_MIN 1025 depends on HAVE_ARCH_MMAP_RND_BITS 1026 help 1027 This value can be used to select the number of bits to use to 1028 determine the random offset to the base address of vma regions 1029 resulting from mmap allocations. This value will be bounded 1030 by the architecture's minimum and maximum supported values. 1031 1032 This value can be changed after boot using the 1033 /proc/sys/vm/mmap_rnd_bits tunable 1034 1035config HAVE_ARCH_MMAP_RND_COMPAT_BITS 1036 bool 1037 help 1038 An arch should select this symbol if it supports running applications 1039 in compatibility mode, supports setting a variable number of bits for 1040 use in establishing the base address for mmap allocations, has MMU 1041 enabled and provides values for both: 1042 - ARCH_MMAP_RND_COMPAT_BITS_MIN 1043 - ARCH_MMAP_RND_COMPAT_BITS_MAX 1044 1045config ARCH_MMAP_RND_COMPAT_BITS_MIN 1046 int 1047 1048config ARCH_MMAP_RND_COMPAT_BITS_MAX 1049 int 1050 1051config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 1052 int 1053 1054config ARCH_MMAP_RND_COMPAT_BITS 1055 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 1056 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 1057 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 1058 default ARCH_MMAP_RND_COMPAT_BITS_MIN 1059 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 1060 help 1061 This value can be used to select the number of bits to use to 1062 determine the random offset to the base address of vma regions 1063 resulting from mmap allocations for compatible applications This 1064 value will be bounded by the architecture's minimum and maximum 1065 supported values. 1066 1067 This value can be changed after boot using the 1068 /proc/sys/vm/mmap_rnd_compat_bits tunable 1069 1070config HAVE_ARCH_COMPAT_MMAP_BASES 1071 bool 1072 help 1073 This allows 64bit applications to invoke 32-bit mmap() syscall 1074 and vice-versa 32-bit applications to call 64-bit mmap(). 1075 Required for applications doing different bitness syscalls. 1076 1077config PAGE_SIZE_LESS_THAN_64KB 1078 def_bool y 1079 depends on !ARM64_64K_PAGES 1080 depends on !IA64_PAGE_SIZE_64KB 1081 depends on !PAGE_SIZE_64KB 1082 depends on !PARISC_PAGE_SIZE_64KB 1083 depends on PAGE_SIZE_LESS_THAN_256KB 1084 1085config PAGE_SIZE_LESS_THAN_256KB 1086 def_bool y 1087 depends on !PAGE_SIZE_256KB 1088 1089# This allows to use a set of generic functions to determine mmap base 1090# address by giving priority to top-down scheme only if the process 1091# is not in legacy mode (compat task, unlimited stack size or 1092# sysctl_legacy_va_layout). 1093# Architecture that selects this option can provide its own version of: 1094# - STACK_RND_MASK 1095config ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT 1096 bool 1097 depends on MMU 1098 select ARCH_HAS_ELF_RANDOMIZE 1099 1100config HAVE_OBJTOOL 1101 bool 1102 1103config HAVE_JUMP_LABEL_HACK 1104 bool 1105 1106config HAVE_NOINSTR_HACK 1107 bool 1108 1109config HAVE_NOINSTR_VALIDATION 1110 bool 1111 1112config HAVE_UACCESS_VALIDATION 1113 bool 1114 select OBJTOOL 1115 1116config HAVE_STACK_VALIDATION 1117 bool 1118 help 1119 Architecture supports objtool compile-time frame pointer rule 1120 validation. 1121 1122config HAVE_RELIABLE_STACKTRACE 1123 bool 1124 help 1125 Architecture has either save_stack_trace_tsk_reliable() or 1126 arch_stack_walk_reliable() function which only returns a stack trace 1127 if it can guarantee the trace is reliable. 1128 1129config HAVE_ARCH_HASH 1130 bool 1131 default n 1132 help 1133 If this is set, the architecture provides an <asm/hash.h> 1134 file which provides platform-specific implementations of some 1135 functions in <linux/hash.h> or fs/namei.c. 1136 1137config HAVE_ARCH_NVRAM_OPS 1138 bool 1139 1140config ISA_BUS_API 1141 def_bool ISA 1142 1143# 1144# ABI hall of shame 1145# 1146config CLONE_BACKWARDS 1147 bool 1148 help 1149 Architecture has tls passed as the 4th argument of clone(2), 1150 not the 5th one. 1151 1152config CLONE_BACKWARDS2 1153 bool 1154 help 1155 Architecture has the first two arguments of clone(2) swapped. 1156 1157config CLONE_BACKWARDS3 1158 bool 1159 help 1160 Architecture has tls passed as the 3rd argument of clone(2), 1161 not the 5th one. 1162 1163config ODD_RT_SIGACTION 1164 bool 1165 help 1166 Architecture has unusual rt_sigaction(2) arguments 1167 1168config OLD_SIGSUSPEND 1169 bool 1170 help 1171 Architecture has old sigsuspend(2) syscall, of one-argument variety 1172 1173config OLD_SIGSUSPEND3 1174 bool 1175 help 1176 Even weirder antique ABI - three-argument sigsuspend(2) 1177 1178config OLD_SIGACTION 1179 bool 1180 help 1181 Architecture has old sigaction(2) syscall. Nope, not the same 1182 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 1183 but fairly different variant of sigaction(2), thanks to OSF/1 1184 compatibility... 1185 1186config COMPAT_OLD_SIGACTION 1187 bool 1188 1189config COMPAT_32BIT_TIME 1190 bool "Provide system calls for 32-bit time_t" 1191 default !64BIT || COMPAT 1192 help 1193 This enables 32 bit time_t support in addition to 64 bit time_t support. 1194 This is relevant on all 32-bit architectures, and 64-bit architectures 1195 as part of compat syscall handling. 1196 1197config ARCH_NO_PREEMPT 1198 bool 1199 1200config ARCH_SUPPORTS_RT 1201 bool 1202 1203config CPU_NO_EFFICIENT_FFS 1204 def_bool n 1205 1206config HAVE_ARCH_VMAP_STACK 1207 def_bool n 1208 help 1209 An arch should select this symbol if it can support kernel stacks 1210 in vmalloc space. This means: 1211 1212 - vmalloc space must be large enough to hold many kernel stacks. 1213 This may rule out many 32-bit architectures. 1214 1215 - Stacks in vmalloc space need to work reliably. For example, if 1216 vmap page tables are created on demand, either this mechanism 1217 needs to work while the stack points to a virtual address with 1218 unpopulated page tables or arch code (switch_to() and switch_mm(), 1219 most likely) needs to ensure that the stack's page table entries 1220 are populated before running on a possibly unpopulated stack. 1221 1222 - If the stack overflows into a guard page, something reasonable 1223 should happen. The definition of "reasonable" is flexible, but 1224 instantly rebooting without logging anything would be unfriendly. 1225 1226config VMAP_STACK 1227 default y 1228 bool "Use a virtually-mapped stack" 1229 depends on HAVE_ARCH_VMAP_STACK 1230 depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC 1231 help 1232 Enable this if you want the use virtually-mapped kernel stacks 1233 with guard pages. This causes kernel stack overflows to be 1234 caught immediately rather than causing difficult-to-diagnose 1235 corruption. 1236 1237 To use this with software KASAN modes, the architecture must support 1238 backing virtual mappings with real shadow memory, and KASAN_VMALLOC 1239 must be enabled. 1240 1241config HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 1242 def_bool n 1243 help 1244 An arch should select this symbol if it can support kernel stack 1245 offset randomization with calls to add_random_kstack_offset() 1246 during syscall entry and choose_random_kstack_offset() during 1247 syscall exit. Careful removal of -fstack-protector-strong and 1248 -fstack-protector should also be applied to the entry code and 1249 closely examined, as the artificial stack bump looks like an array 1250 to the compiler, so it will attempt to add canary checks regardless 1251 of the static branch state. 1252 1253config RANDOMIZE_KSTACK_OFFSET 1254 bool "Support for randomizing kernel stack offset on syscall entry" if EXPERT 1255 default y 1256 depends on HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 1257 depends on INIT_STACK_NONE || !CC_IS_CLANG || CLANG_VERSION >= 140000 1258 help 1259 The kernel stack offset can be randomized (after pt_regs) by 1260 roughly 5 bits of entropy, frustrating memory corruption 1261 attacks that depend on stack address determinism or 1262 cross-syscall address exposures. 1263 1264 The feature is controlled via the "randomize_kstack_offset=on/off" 1265 kernel boot param, and if turned off has zero overhead due to its use 1266 of static branches (see JUMP_LABEL). 1267 1268 If unsure, say Y. 1269 1270config RANDOMIZE_KSTACK_OFFSET_DEFAULT 1271 bool "Default state of kernel stack offset randomization" 1272 depends on RANDOMIZE_KSTACK_OFFSET 1273 help 1274 Kernel stack offset randomization is controlled by kernel boot param 1275 "randomize_kstack_offset=on/off", and this config chooses the default 1276 boot state. 1277 1278config ARCH_OPTIONAL_KERNEL_RWX 1279 def_bool n 1280 1281config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1282 def_bool n 1283 1284config ARCH_HAS_STRICT_KERNEL_RWX 1285 def_bool n 1286 1287config STRICT_KERNEL_RWX 1288 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 1289 depends on ARCH_HAS_STRICT_KERNEL_RWX 1290 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1291 help 1292 If this is set, kernel text and rodata memory will be made read-only, 1293 and non-text memory will be made non-executable. This provides 1294 protection against certain security exploits (e.g. executing the heap 1295 or modifying text) 1296 1297 These features are considered standard security practice these days. 1298 You should say Y here in almost all cases. 1299 1300config ARCH_HAS_STRICT_MODULE_RWX 1301 def_bool n 1302 1303config STRICT_MODULE_RWX 1304 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 1305 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 1306 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1307 help 1308 If this is set, module text and rodata memory will be made read-only, 1309 and non-text memory will be made non-executable. This provides 1310 protection against certain security exploits (e.g. writing to text) 1311 1312# select if the architecture provides an asm/dma-direct.h header 1313config ARCH_HAS_PHYS_TO_DMA 1314 bool 1315 1316config HAVE_ARCH_COMPILER_H 1317 bool 1318 help 1319 An architecture can select this if it provides an 1320 asm/compiler.h header that should be included after 1321 linux/compiler-*.h in order to override macro definitions that those 1322 headers generally provide. 1323 1324config HAVE_ARCH_PREL32_RELOCATIONS 1325 bool 1326 help 1327 May be selected by an architecture if it supports place-relative 1328 32-bit relocations, both in the toolchain and in the module loader, 1329 in which case relative references can be used in special sections 1330 for PCI fixup, initcalls etc which are only half the size on 64 bit 1331 architectures, and don't require runtime relocation on relocatable 1332 kernels. 1333 1334config ARCH_USE_MEMREMAP_PROT 1335 bool 1336 1337config LOCK_EVENT_COUNTS 1338 bool "Locking event counts collection" 1339 depends on DEBUG_FS 1340 help 1341 Enable light-weight counting of various locking related events 1342 in the system with minimal performance impact. This reduces 1343 the chance of application behavior change because of timing 1344 differences. The counts are reported via debugfs. 1345 1346# Select if the architecture has support for applying RELR relocations. 1347config ARCH_HAS_RELR 1348 bool 1349 1350config RELR 1351 bool "Use RELR relocation packing" 1352 depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR 1353 default y 1354 help 1355 Store the kernel's dynamic relocations in the RELR relocation packing 1356 format. Requires a compatible linker (LLD supports this feature), as 1357 well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy 1358 are compatible). 1359 1360config ARCH_HAS_MEM_ENCRYPT 1361 bool 1362 1363config ARCH_HAS_CC_PLATFORM 1364 bool 1365 1366config HAVE_SPARSE_SYSCALL_NR 1367 bool 1368 help 1369 An architecture should select this if its syscall numbering is sparse 1370 to save space. For example, MIPS architecture has a syscall array with 1371 entries at 4000, 5000 and 6000 locations. This option turns on syscall 1372 related optimizations for a given architecture. 1373 1374config ARCH_HAS_VDSO_DATA 1375 bool 1376 1377config HAVE_STATIC_CALL 1378 bool 1379 1380config HAVE_STATIC_CALL_INLINE 1381 bool 1382 depends on HAVE_STATIC_CALL 1383 select OBJTOOL 1384 1385config HAVE_PREEMPT_DYNAMIC 1386 bool 1387 1388config HAVE_PREEMPT_DYNAMIC_CALL 1389 bool 1390 depends on HAVE_STATIC_CALL 1391 select HAVE_PREEMPT_DYNAMIC 1392 help 1393 An architecture should select this if it can handle the preemption 1394 model being selected at boot time using static calls. 1395 1396 Where an architecture selects HAVE_STATIC_CALL_INLINE, any call to a 1397 preemption function will be patched directly. 1398 1399 Where an architecture does not select HAVE_STATIC_CALL_INLINE, any 1400 call to a preemption function will go through a trampoline, and the 1401 trampoline will be patched. 1402 1403 It is strongly advised to support inline static call to avoid any 1404 overhead. 1405 1406config HAVE_PREEMPT_DYNAMIC_KEY 1407 bool 1408 depends on HAVE_ARCH_JUMP_LABEL 1409 select HAVE_PREEMPT_DYNAMIC 1410 help 1411 An architecture should select this if it can handle the preemption 1412 model being selected at boot time using static keys. 1413 1414 Each preemption function will be given an early return based on a 1415 static key. This should have slightly lower overhead than non-inline 1416 static calls, as this effectively inlines each trampoline into the 1417 start of its callee. This may avoid redundant work, and may 1418 integrate better with CFI schemes. 1419 1420 This will have greater overhead than using inline static calls as 1421 the call to the preemption function cannot be entirely elided. 1422 1423config ARCH_WANT_LD_ORPHAN_WARN 1424 bool 1425 help 1426 An arch should select this symbol once all linker sections are explicitly 1427 included, size-asserted, or discarded in the linker scripts. This is 1428 important because we never want expected sections to be placed heuristically 1429 by the linker, since the locations of such sections can change between linker 1430 versions. 1431 1432config HAVE_ARCH_PFN_VALID 1433 bool 1434 1435config ARCH_SUPPORTS_DEBUG_PAGEALLOC 1436 bool 1437 1438config ARCH_SUPPORTS_PAGE_TABLE_CHECK 1439 bool 1440 1441config ARCH_SPLIT_ARG64 1442 bool 1443 help 1444 If a 32-bit architecture requires 64-bit arguments to be split into 1445 pairs of 32-bit arguments, select this option. 1446 1447config ARCH_HAS_ELFCORE_COMPAT 1448 bool 1449 1450config ARCH_HAS_PARANOID_L1D_FLUSH 1451 bool 1452 1453config ARCH_HAVE_TRACE_MMIO_ACCESS 1454 bool 1455 1456config DYNAMIC_SIGFRAME 1457 bool 1458 1459# Select, if arch has a named attribute group bound to NUMA device nodes. 1460config HAVE_ARCH_NODE_DEV_GROUP 1461 bool 1462 1463config ARCH_HAS_NONLEAF_PMD_YOUNG 1464 bool 1465 help 1466 Architectures that select this option are capable of setting the 1467 accessed bit in non-leaf PMD entries when using them as part of linear 1468 address translations. Page table walkers that clear the accessed bit 1469 may use this capability to reduce their search space. 1470 1471source "kernel/gcov/Kconfig" 1472 1473source "scripts/gcc-plugins/Kconfig" 1474 1475config FUNCTION_ALIGNMENT_4B 1476 bool 1477 1478config FUNCTION_ALIGNMENT_8B 1479 bool 1480 1481config FUNCTION_ALIGNMENT_16B 1482 bool 1483 1484config FUNCTION_ALIGNMENT_32B 1485 bool 1486 1487config FUNCTION_ALIGNMENT_64B 1488 bool 1489 1490config FUNCTION_ALIGNMENT 1491 int 1492 default 64 if FUNCTION_ALIGNMENT_64B 1493 default 32 if FUNCTION_ALIGNMENT_32B 1494 default 16 if FUNCTION_ALIGNMENT_16B 1495 default 8 if FUNCTION_ALIGNMENT_8B 1496 default 4 if FUNCTION_ALIGNMENT_4B 1497 default 0 1498 1499endmenu 1500