1 # SPDX-License-Identifier: GPL-2.0
3 # General architecture dependent options
7 # Note: arch/$(SRCARCH)/Kconfig needs to be included first so that it can
8 # override the default values in this file.
10 source "arch/$(SRCARCH)/Kconfig"
12 menu "General architecture-dependent options"
27 config ARCH_HAS_SUBPAGE_FAULTS
30 Select if the architecture can check permissions at sub-page
31 granularity (e.g. arm64 MTE). The probe_user_*() functions
43 depends on HAVE_KPROBES
45 select TASKS_RCU if PREEMPTION
47 Kprobes allows you to trap at almost any kernel address and
48 execute a callback function. register_kprobe() establishes
49 a probepoint and specifies the callback. Kprobes is useful
50 for kernel debugging, non-intrusive instrumentation and testing.
54 bool "Optimize very unlikely/likely branches"
55 depends on HAVE_ARCH_JUMP_LABEL
56 select OBJTOOL if HAVE_JUMP_LABEL_HACK
58 This option enables a transparent branch optimization that
59 makes certain almost-always-true or almost-always-false branch
60 conditions even cheaper to execute within the kernel.
62 Certain performance-sensitive kernel code, such as trace points,
63 scheduler functionality, networking code and KVM have such
64 branches and include support for this optimization technique.
66 If it is detected that the compiler has support for "asm goto",
67 the kernel will compile such branches with just a nop
68 instruction. When the condition flag is toggled to true, the
69 nop will be converted to a jump instruction to execute the
70 conditional block of instructions.
72 This technique lowers overhead and stress on the branch prediction
73 of the processor and generally makes the kernel faster. The update
74 of the condition is slower, but those are always very rare.
76 ( On 32-bit x86, the necessary options added to the compiler
77 flags may increase the size of the kernel slightly. )
79 config STATIC_KEYS_SELFTEST
80 bool "Static key selftest"
83 Boot time self-test of the branch patching code.
85 config STATIC_CALL_SELFTEST
86 bool "Static call selftest"
87 depends on HAVE_STATIC_CALL
89 Boot time self-test of the call patching code.
93 depends on KPROBES && HAVE_OPTPROBES
94 select TASKS_RCU if PREEMPTION
96 config KPROBES_ON_FTRACE
98 depends on KPROBES && HAVE_KPROBES_ON_FTRACE
99 depends on DYNAMIC_FTRACE_WITH_REGS
101 If function tracer is enabled and the arch supports full
102 passing of pt_regs to function tracing, then kprobes can
103 optimize on top of function tracing.
107 depends on ARCH_SUPPORTS_UPROBES
109 Uprobes is the user-space counterpart to kprobes: they
110 enable instrumentation applications (such as 'perf probe')
111 to establish unintrusive probes in user-space binaries and
112 libraries, by executing handler functions when the probes
113 are hit by user-space applications.
115 ( These probes come in the form of single-byte breakpoints,
116 managed by the kernel and kept transparent to the probed
119 config HAVE_64BIT_ALIGNED_ACCESS
120 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS
122 Some architectures require 64 bit accesses to be 64 bit
123 aligned, which also requires structs containing 64 bit values
124 to be 64 bit aligned too. This includes some 32 bit
125 architectures which can do 64 bit accesses, as well as 64 bit
126 architectures without unaligned access.
128 This symbol should be selected by an architecture if 64 bit
129 accesses are required to be 64 bit aligned in this way even
130 though it is not a 64 bit architecture.
132 See Documentation/core-api/unaligned-memory-access.rst for
133 more information on the topic of unaligned memory accesses.
135 config HAVE_EFFICIENT_UNALIGNED_ACCESS
138 Some architectures are unable to perform unaligned accesses
139 without the use of get_unaligned/put_unaligned. Others are
140 unable to perform such accesses efficiently (e.g. trap on
141 unaligned access and require fixing it up in the exception
144 This symbol should be selected by an architecture if it can
145 perform unaligned accesses efficiently to allow different
146 code paths to be selected for these cases. Some network
147 drivers, for example, could opt to not fix up alignment
148 problems with received packets if doing so would not help
151 See Documentation/core-api/unaligned-memory-access.rst for more
152 information on the topic of unaligned memory accesses.
154 config ARCH_USE_BUILTIN_BSWAP
157 Modern versions of GCC (since 4.4) have builtin functions
158 for handling byte-swapping. Using these, instead of the old
159 inline assembler that the architecture code provides in the
160 __arch_bswapXX() macros, allows the compiler to see what's
161 happening and offers more opportunity for optimisation. In
162 particular, the compiler will be able to combine the byteswap
163 with a nearby load or store and use load-and-swap or
164 store-and-swap instructions if the architecture has them. It
165 should almost *never* result in code which is worse than the
166 hand-coded assembler in <asm/swab.h>. But just in case it
167 does, the use of the builtins is optional.
169 Any architecture with load-and-swap or store-and-swap
170 instructions should set this. And it shouldn't hurt to set it
171 on architectures that don't have such instructions.
175 depends on KPROBES && (HAVE_KRETPROBES || HAVE_RETHOOK)
177 config KRETPROBE_ON_RETHOOK
179 depends on HAVE_RETHOOK
180 depends on KRETPROBES
183 config USER_RETURN_NOTIFIER
185 depends on HAVE_USER_RETURN_NOTIFIER
187 Provide a kernel-internal notification when a cpu is about to
190 config HAVE_IOREMAP_PROT
196 config HAVE_KRETPROBES
199 config HAVE_OPTPROBES
202 config HAVE_KPROBES_ON_FTRACE
205 config ARCH_CORRECT_STACKTRACE_ON_KRETPROBE
208 Since kretprobes modifies return address on the stack, the
209 stacktrace may see the kretprobe trampoline address instead
210 of correct one. If the architecture stacktrace code and
211 unwinder can adjust such entries, select this configuration.
213 config HAVE_FUNCTION_ERROR_INJECTION
219 config HAVE_FUNCTION_DESCRIPTORS
222 config TRACE_IRQFLAGS_SUPPORT
225 config TRACE_IRQFLAGS_NMI_SUPPORT
229 # An arch should select this if it provides all these things:
231 # task_pt_regs() in asm/processor.h or asm/ptrace.h
232 # arch_has_single_step() if there is hardware single-step support
233 # arch_has_block_step() if there is hardware block-step support
234 # asm/syscall.h supplying asm-generic/syscall.h interface
235 # linux/regset.h user_regset interfaces
236 # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
237 # TIF_SYSCALL_TRACE calls ptrace_report_syscall_{entry,exit}
238 # TIF_NOTIFY_RESUME calls resume_user_mode_work()
240 config HAVE_ARCH_TRACEHOOK
243 config HAVE_DMA_CONTIGUOUS
246 config GENERIC_SMP_IDLE_THREAD
249 config GENERIC_IDLE_POLL_SETUP
252 config ARCH_HAS_FORTIFY_SOURCE
255 An architecture should select this when it can successfully
256 build and run with CONFIG_FORTIFY_SOURCE.
259 # Select if the arch provides a historic keepinit alias for the retain_initrd
260 # command line option
262 config ARCH_HAS_KEEPINITRD
265 # Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h
266 config ARCH_HAS_SET_MEMORY
269 # Select if arch has all set_direct_map_invalid/default() functions
270 config ARCH_HAS_SET_DIRECT_MAP
274 # Select if the architecture provides the arch_dma_set_uncached symbol to
275 # either provide an uncached segment alias for a DMA allocation, or
276 # to remap the page tables in place.
278 config ARCH_HAS_DMA_SET_UNCACHED
282 # Select if the architectures provides the arch_dma_clear_uncached symbol
283 # to undo an in-place page table remap for uncached access.
285 config ARCH_HAS_DMA_CLEAR_UNCACHED
288 # Select if arch init_task must go in the __init_task_data section
289 config ARCH_TASK_STRUCT_ON_STACK
292 # Select if arch has its private alloc_task_struct() function
293 config ARCH_TASK_STRUCT_ALLOCATOR
296 config HAVE_ARCH_THREAD_STRUCT_WHITELIST
298 depends on !ARCH_TASK_STRUCT_ALLOCATOR
300 An architecture should select this to provide hardened usercopy
301 knowledge about what region of the thread_struct should be
302 whitelisted for copying to userspace. Normally this is only the
303 FPU registers. Specifically, arch_thread_struct_whitelist()
304 should be implemented. Without this, the entire thread_struct
305 field in task_struct will be left whitelisted.
307 # Select if arch has its private alloc_thread_stack() function
308 config ARCH_THREAD_STACK_ALLOCATOR
311 # Select if arch wants to size task_struct dynamically via arch_task_struct_size:
312 config ARCH_WANTS_DYNAMIC_TASK_STRUCT
315 config ARCH_WANTS_NO_INSTR
318 An architecture should select this if the noinstr macro is being used on
319 functions to denote that the toolchain should avoid instrumenting such
320 functions and is required for correctness.
322 config ARCH_32BIT_OFF_T
326 All new 32-bit architectures should have 64-bit off_t type on
327 userspace side which corresponds to the loff_t kernel type. This
328 is the requirement for modern ABIs. Some existing architectures
329 still support 32-bit off_t. This option is enabled for all such
330 architectures explicitly.
332 # Selected by 64 bit architectures which have a 32 bit f_tinode in struct ustat
333 config ARCH_32BIT_USTAT_F_TINODE
336 config HAVE_ASM_MODVERSIONS
339 This symbol should be selected by an architecture if it provides
340 <asm/asm-prototypes.h> to support the module versioning for symbols
341 exported from assembly code.
343 config HAVE_REGS_AND_STACK_ACCESS_API
346 This symbol should be selected by an architecture if it supports
347 the API needed to access registers and stack entries from pt_regs,
348 declared in asm/ptrace.h
349 For example the kprobes-based event tracer needs this API.
353 depends on HAVE_REGS_AND_STACK_ACCESS_API
355 This symbol should be selected by an architecture if it
356 supports an implementation of restartable sequences.
361 This symbol should be selected by an architecture if it
364 config HAVE_FUNCTION_ARG_ACCESS_API
367 This symbol should be selected by an architecture if it supports
368 the API needed to access function arguments from pt_regs,
369 declared in asm/ptrace.h
371 config HAVE_HW_BREAKPOINT
373 depends on PERF_EVENTS
375 config HAVE_MIXED_BREAKPOINTS_REGS
377 depends on HAVE_HW_BREAKPOINT
379 Depending on the arch implementation of hardware breakpoints,
380 some of them have separate registers for data and instruction
381 breakpoints addresses, others have mixed registers to store
382 them but define the access type in a control register.
383 Select this option if your arch implements breakpoints under the
386 config HAVE_USER_RETURN_NOTIFIER
389 config HAVE_PERF_EVENTS_NMI
392 System hardware can generate an NMI using the perf event
393 subsystem. Also has support for calculating CPU cycle events
394 to determine how many clock cycles in a given period.
396 config HAVE_HARDLOCKUP_DETECTOR_PERF
398 depends on HAVE_PERF_EVENTS_NMI
400 The arch chooses to use the generic perf-NMI-based hardlockup
401 detector. Must define HAVE_PERF_EVENTS_NMI.
403 config HAVE_NMI_WATCHDOG
407 The arch provides a low level NMI watchdog. It provides
408 asm/nmi.h, and defines its own arch_touch_nmi_watchdog().
410 config HAVE_HARDLOCKUP_DETECTOR_ARCH
412 select HAVE_NMI_WATCHDOG
414 The arch chooses to provide its own hardlockup detector, which is
415 a superset of the HAVE_NMI_WATCHDOG. It also conforms to config
416 interfaces and parameters provided by hardlockup detector subsystem.
418 config HAVE_PERF_REGS
421 Support selective register dumps for perf events. This includes
422 bit-mapping of each registers and a unique architecture id.
424 config HAVE_PERF_USER_STACK_DUMP
427 Support user stack dumps for perf event samples. This needs
428 access to the user stack pointer which is not unified across
431 config HAVE_ARCH_JUMP_LABEL
434 config HAVE_ARCH_JUMP_LABEL_RELATIVE
437 config MMU_GATHER_TABLE_FREE
440 config MMU_GATHER_RCU_TABLE_FREE
442 select MMU_GATHER_TABLE_FREE
444 config MMU_GATHER_PAGE_SIZE
447 config MMU_GATHER_NO_RANGE
449 select MMU_GATHER_MERGE_VMAS
451 config MMU_GATHER_NO_FLUSH_CACHE
454 config MMU_GATHER_MERGE_VMAS
457 config MMU_GATHER_NO_GATHER
459 depends on MMU_GATHER_TABLE_FREE
461 config ARCH_WANT_IRQS_OFF_ACTIVATE_MM
464 Temporary select until all architectures can be converted to have
465 irqs disabled over activate_mm. Architectures that do IPI based TLB
466 shootdowns should enable this.
468 config ARCH_HAVE_NMI_SAFE_CMPXCHG
471 config ARCH_HAS_NMI_SAFE_THIS_CPU_OPS
474 config HAVE_ALIGNED_STRUCT_PAGE
477 This makes sure that struct pages are double word aligned and that
478 e.g. the SLUB allocator can perform double word atomic operations
479 on a struct page for better performance. However selecting this
480 might increase the size of a struct page by a word.
482 config HAVE_CMPXCHG_LOCAL
485 config HAVE_CMPXCHG_DOUBLE
488 config ARCH_WEAK_RELEASE_ACQUIRE
491 config ARCH_WANT_IPC_PARSE_VERSION
494 config ARCH_WANT_COMPAT_IPC_PARSE_VERSION
497 config ARCH_WANT_OLD_COMPAT_IPC
498 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
501 config HAVE_ARCH_SECCOMP
504 An arch should select this symbol to support seccomp mode 1 (the fixed
505 syscall policy), and must provide an overrides for __NR_seccomp_sigreturn,
506 and compat syscalls if the asm-generic/seccomp.h defaults need adjustment:
507 - __NR_seccomp_read_32
508 - __NR_seccomp_write_32
509 - __NR_seccomp_exit_32
510 - __NR_seccomp_sigreturn_32
512 config HAVE_ARCH_SECCOMP_FILTER
514 select HAVE_ARCH_SECCOMP
516 An arch should select this symbol if it provides all of these things:
517 - all the requirements for HAVE_ARCH_SECCOMP
519 - syscall_get_arguments()
521 - syscall_set_return_value()
522 - SIGSYS siginfo_t support
523 - secure_computing is called from a ptrace_event()-safe context
524 - secure_computing return value is checked and a return value of -1
525 results in the system call being skipped immediately.
526 - seccomp syscall wired up
527 - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE,
528 SECCOMP_ARCH_NATIVE_NR, SECCOMP_ARCH_NATIVE_NAME defined. If
529 COMPAT is supported, have the SECCOMP_ARCH_COMPAT* defines too.
532 prompt "Enable seccomp to safely execute untrusted bytecode"
534 depends on HAVE_ARCH_SECCOMP
536 This kernel feature is useful for number crunching applications
537 that may need to handle untrusted bytecode during their
538 execution. By using pipes or other transports made available
539 to the process as file descriptors supporting the read/write
540 syscalls, it's possible to isolate those applications in their
541 own address space using seccomp. Once seccomp is enabled via
542 prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be
543 disabled and the task is only allowed to execute a few safe
544 syscalls defined by each seccomp mode.
548 config SECCOMP_FILTER
550 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
552 Enable tasks to build secure computing environments defined
553 in terms of Berkeley Packet Filter programs which implement
554 task-defined system call filtering polices.
556 See Documentation/userspace-api/seccomp_filter.rst for details.
558 config SECCOMP_CACHE_DEBUG
559 bool "Show seccomp filter cache status in /proc/pid/seccomp_cache"
560 depends on SECCOMP_FILTER && !HAVE_SPARSE_SYSCALL_NR
563 This enables the /proc/pid/seccomp_cache interface to monitor
564 seccomp cache data. The file format is subject to change. Reading
565 the file requires CAP_SYS_ADMIN.
567 This option is for debugging only. Enabling presents the risk that
568 an adversary may be able to infer the seccomp filter logic.
572 config HAVE_ARCH_STACKLEAK
575 An architecture should select this if it has the code which
576 fills the used part of the kernel stack with the STACKLEAK_POISON
577 value before returning from system calls.
579 config HAVE_STACKPROTECTOR
582 An arch should select this symbol if:
583 - it has implemented a stack canary (e.g. __stack_chk_guard)
585 config STACKPROTECTOR
586 bool "Stack Protector buffer overflow detection"
587 depends on HAVE_STACKPROTECTOR
588 depends on $(cc-option,-fstack-protector)
591 This option turns on the "stack-protector" GCC feature. This
592 feature puts, at the beginning of functions, a canary value on
593 the stack just before the return address, and validates
594 the value just before actually returning. Stack based buffer
595 overflows (that need to overwrite this return address) now also
596 overwrite the canary, which gets detected and the attack is then
597 neutralized via a kernel panic.
599 Functions will have the stack-protector canary logic added if they
600 have an 8-byte or larger character array on the stack.
602 This feature requires gcc version 4.2 or above, or a distribution
603 gcc with the feature backported ("-fstack-protector").
605 On an x86 "defconfig" build, this feature adds canary checks to
606 about 3% of all kernel functions, which increases kernel code size
609 config STACKPROTECTOR_STRONG
610 bool "Strong Stack Protector"
611 depends on STACKPROTECTOR
612 depends on $(cc-option,-fstack-protector-strong)
615 Functions will have the stack-protector canary logic added in any
616 of the following conditions:
618 - local variable's address used as part of the right hand side of an
619 assignment or function argument
620 - local variable is an array (or union containing an array),
621 regardless of array type or length
622 - uses register local variables
624 This feature requires gcc version 4.9 or above, or a distribution
625 gcc with the feature backported ("-fstack-protector-strong").
627 On an x86 "defconfig" build, this feature adds canary checks to
628 about 20% of all kernel functions, which increases the kernel code
631 config ARCH_SUPPORTS_SHADOW_CALL_STACK
634 An architecture should select this if it supports the compiler's
635 Shadow Call Stack and implements runtime support for shadow stack
638 config SHADOW_CALL_STACK
639 bool "Shadow Call Stack"
640 depends on ARCH_SUPPORTS_SHADOW_CALL_STACK
641 depends on DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER
643 This option enables the compiler's Shadow Call Stack, which
644 uses a shadow stack to protect function return addresses from
645 being overwritten by an attacker. More information can be found
646 in the compiler's documentation:
648 - Clang: https://clang.llvm.org/docs/ShadowCallStack.html
649 - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options
651 Note that security guarantees in the kernel differ from the
652 ones documented for user space. The kernel must store addresses
653 of shadow stacks in memory, which means an attacker capable of
654 reading and writing arbitrary memory may be able to locate them
655 and hijack control flow by modifying the stacks.
660 Set by the arch code if it relies on code patching to insert the
661 shadow call stack push and pop instructions rather than on the
667 Selected if the kernel will be built using the compiler's LTO feature.
673 Selected if the kernel will be built using Clang's LTO feature.
675 config ARCH_SUPPORTS_LTO_CLANG
678 An architecture should select this option if it supports:
679 - compiling with Clang,
680 - compiling inline assembly with Clang's integrated assembler,
681 - and linking with LLD.
683 config ARCH_SUPPORTS_LTO_CLANG_THIN
686 An architecture should select this option if it can support Clang's
691 depends on CC_IS_CLANG && LD_IS_LLD && AS_IS_LLVM
692 depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm)
693 depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm)
694 depends on ARCH_SUPPORTS_LTO_CLANG
695 depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT
696 depends on !KASAN || KASAN_HW_TAGS
697 depends on !GCOV_KERNEL
699 The compiler and Kconfig options support building with Clang's
703 prompt "Link Time Optimization (LTO)"
706 This option enables Link Time Optimization (LTO), which allows the
707 compiler to optimize binaries globally.
709 If unsure, select LTO_NONE. Note that LTO is very resource-intensive
710 so it's disabled by default.
715 Build the kernel normally, without Link Time Optimization (LTO).
717 config LTO_CLANG_FULL
718 bool "Clang Full LTO (EXPERIMENTAL)"
719 depends on HAS_LTO_CLANG
720 depends on !COMPILE_TEST
723 This option enables Clang's full Link Time Optimization (LTO), which
724 allows the compiler to optimize the kernel globally. If you enable
725 this option, the compiler generates LLVM bitcode instead of ELF
726 object files, and the actual compilation from bitcode happens at
727 the LTO link step, which may take several minutes depending on the
728 kernel configuration. More information can be found from LLVM's
731 https://llvm.org/docs/LinkTimeOptimization.html
733 During link time, this option can use a large amount of RAM, and
734 may take much longer than the ThinLTO option.
736 config LTO_CLANG_THIN
737 bool "Clang ThinLTO (EXPERIMENTAL)"
738 depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN
741 This option enables Clang's ThinLTO, which allows for parallel
742 optimization and faster incremental compiles compared to the
743 CONFIG_LTO_CLANG_FULL option. More information can be found
744 from Clang's documentation:
746 https://clang.llvm.org/docs/ThinLTO.html
751 config ARCH_SUPPORTS_CFI_CLANG
754 An architecture should select this option if it can support Clang's
755 Control-Flow Integrity (CFI) checking.
757 config ARCH_USES_CFI_TRAPS
761 bool "Use Clang's Control Flow Integrity (CFI)"
762 depends on ARCH_SUPPORTS_CFI_CLANG
763 depends on $(cc-option,-fsanitize=kcfi)
765 This option enables Clang’s forward-edge Control Flow Integrity
766 (CFI) checking, where the compiler injects a runtime check to each
767 indirect function call to ensure the target is a valid function with
768 the correct static type. This restricts possible call targets and
769 makes it more difficult for an attacker to exploit bugs that allow
770 the modification of stored function pointers. More information can be
771 found from Clang's documentation:
773 https://clang.llvm.org/docs/ControlFlowIntegrity.html
775 config CFI_PERMISSIVE
776 bool "Use CFI in permissive mode"
779 When selected, Control Flow Integrity (CFI) violations result in a
780 warning instead of a kernel panic. This option should only be used
781 for finding indirect call type mismatches during development.
785 config HAVE_ARCH_WITHIN_STACK_FRAMES
788 An architecture should select this if it can walk the kernel stack
789 frames to determine if an object is part of either the arguments
790 or local variables (i.e. that it excludes saved return addresses,
791 and similar) by implementing an inline arch_within_stack_frames(),
792 which is used by CONFIG_HARDENED_USERCOPY.
794 config HAVE_CONTEXT_TRACKING_USER
797 Provide kernel/user boundaries probes necessary for subsystems
798 that need it, such as userspace RCU extended quiescent state.
799 Syscalls need to be wrapped inside user_exit()-user_enter(), either
800 optimized behind static key or through the slow path using TIF_NOHZ
801 flag. Exceptions handlers must be wrapped as well. Irqs are already
802 protected inside ct_irq_enter/ct_irq_exit() but preemption or signal
803 handling on irq exit still need to be protected.
805 config HAVE_CONTEXT_TRACKING_USER_OFFSTACK
808 Architecture neither relies on exception_enter()/exception_exit()
809 nor on schedule_user(). Also preempt_schedule_notrace() and
810 preempt_schedule_irq() can't be called in a preemptible section
811 while context tracking is CONTEXT_USER. This feature reflects a sane
812 entry implementation where the following requirements are met on
813 critical entry code, ie: before user_exit() or after user_enter():
815 - Critical entry code isn't preemptible (or better yet:
817 - No use of RCU read side critical sections, unless ct_nmi_enter()
819 - No use of instrumentation, unless instrumentation_begin() got
825 Arch relies on TIF_NOHZ and syscall slow path to implement context
826 tracking calls to user_enter()/user_exit().
828 config HAVE_VIRT_CPU_ACCOUNTING
831 config HAVE_VIRT_CPU_ACCOUNTING_IDLE
834 Architecture has its own way to account idle CPU time and therefore
835 doesn't implement vtime_account_idle().
837 config ARCH_HAS_SCALED_CPUTIME
840 config HAVE_VIRT_CPU_ACCOUNTING_GEN
844 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
845 Before enabling this option, arch code must be audited
846 to ensure there are no races in concurrent read/write of
847 cputime_t. For example, reading/writing 64-bit cputime_t on
848 some 32-bit arches may require multiple accesses, so proper
849 locking is needed to protect against concurrent accesses.
851 config HAVE_IRQ_TIME_ACCOUNTING
854 Archs need to ensure they use a high enough resolution clock to
855 support irq time accounting and then call enable_sched_clock_irqtime().
860 Architectures that select this are able to move page tables at the
861 PUD level. If there are only 3 page table levels, the move effectively
862 happens at the PGD level.
867 Archs that select this are able to move page tables at the PMD level.
869 config HAVE_ARCH_TRANSPARENT_HUGEPAGE
872 config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
875 config HAVE_ARCH_HUGE_VMAP
879 # Archs that select this would be capable of PMD-sized vmaps (i.e.,
880 # arch_vmap_pmd_supported() returns true). The VM_ALLOW_HUGE_VMAP flag
881 # must be used to enable allocations to use hugepages.
883 config HAVE_ARCH_HUGE_VMALLOC
884 depends on HAVE_ARCH_HUGE_VMAP
887 config ARCH_WANT_HUGE_PMD_SHARE
890 config HAVE_ARCH_SOFT_DIRTY
893 config HAVE_MOD_ARCH_SPECIFIC
896 The arch uses struct mod_arch_specific to store data. Many arches
897 just need a simple module loader without arch specific data - those
898 should not enable this.
900 config MODULES_USE_ELF_RELA
903 Modules only use ELF RELA relocations. Modules with ELF REL
904 relocations will give an error.
906 config MODULES_USE_ELF_REL
909 Modules only use ELF REL relocations. Modules with ELF RELA
910 relocations will give an error.
912 config ARCH_WANTS_MODULES_DATA_IN_VMALLOC
915 For architectures like powerpc/32 which have constraints on module
916 allocation and need to allocate module data outside of module area.
918 config HAVE_IRQ_EXIT_ON_IRQ_STACK
921 Architecture doesn't only execute the irq handler on the irq stack
922 but also irq_exit(). This way we can process softirqs on this irq
923 stack instead of switching to a new one when we call __do_softirq()
924 in the end of an hardirq.
925 This spares a stack switch and improves cache usage on softirq
928 config HAVE_SOFTIRQ_ON_OWN_STACK
931 Architecture provides a function to run __do_softirq() on a
934 config SOFTIRQ_ON_OWN_STACK
935 def_bool HAVE_SOFTIRQ_ON_OWN_STACK && !PREEMPT_RT
937 config ALTERNATE_USER_ADDRESS_SPACE
940 Architectures set this when the CPU uses separate address
941 spaces for kernel and user space pointers. In this case, the
942 access_ok() check on a __user pointer is skipped.
944 config PGTABLE_LEVELS
948 config ARCH_HAS_ELF_RANDOMIZE
951 An architecture supports choosing randomized locations for
952 stack, mmap, brk, and ET_DYN. Defined functions:
954 - arch_randomize_brk()
956 config HAVE_ARCH_MMAP_RND_BITS
959 An arch should select this symbol if it supports setting a variable
960 number of bits for use in establishing the base address for mmap
961 allocations, has MMU enabled and provides values for both:
962 - ARCH_MMAP_RND_BITS_MIN
963 - ARCH_MMAP_RND_BITS_MAX
965 config HAVE_EXIT_THREAD
968 An architecture implements exit_thread.
970 config ARCH_MMAP_RND_BITS_MIN
973 config ARCH_MMAP_RND_BITS_MAX
976 config ARCH_MMAP_RND_BITS_DEFAULT
979 config ARCH_MMAP_RND_BITS
980 int "Number of bits to use for ASLR of mmap base address" if EXPERT
981 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX
982 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT
983 default ARCH_MMAP_RND_BITS_MIN
984 depends on HAVE_ARCH_MMAP_RND_BITS
986 This value can be used to select the number of bits to use to
987 determine the random offset to the base address of vma regions
988 resulting from mmap allocations. This value will be bounded
989 by the architecture's minimum and maximum supported values.
991 This value can be changed after boot using the
992 /proc/sys/vm/mmap_rnd_bits tunable
994 config HAVE_ARCH_MMAP_RND_COMPAT_BITS
997 An arch should select this symbol if it supports running applications
998 in compatibility mode, supports setting a variable number of bits for
999 use in establishing the base address for mmap allocations, has MMU
1000 enabled and provides values for both:
1001 - ARCH_MMAP_RND_COMPAT_BITS_MIN
1002 - ARCH_MMAP_RND_COMPAT_BITS_MAX
1004 config ARCH_MMAP_RND_COMPAT_BITS_MIN
1007 config ARCH_MMAP_RND_COMPAT_BITS_MAX
1010 config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
1013 config ARCH_MMAP_RND_COMPAT_BITS
1014 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT
1015 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX
1016 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
1017 default ARCH_MMAP_RND_COMPAT_BITS_MIN
1018 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS
1020 This value can be used to select the number of bits to use to
1021 determine the random offset to the base address of vma regions
1022 resulting from mmap allocations for compatible applications This
1023 value will be bounded by the architecture's minimum and maximum
1026 This value can be changed after boot using the
1027 /proc/sys/vm/mmap_rnd_compat_bits tunable
1029 config HAVE_ARCH_COMPAT_MMAP_BASES
1032 This allows 64bit applications to invoke 32-bit mmap() syscall
1033 and vice-versa 32-bit applications to call 64-bit mmap().
1034 Required for applications doing different bitness syscalls.
1036 config PAGE_SIZE_LESS_THAN_64KB
1038 depends on !ARM64_64K_PAGES
1039 depends on !IA64_PAGE_SIZE_64KB
1040 depends on !PAGE_SIZE_64KB
1041 depends on !PARISC_PAGE_SIZE_64KB
1042 depends on PAGE_SIZE_LESS_THAN_256KB
1044 config PAGE_SIZE_LESS_THAN_256KB
1046 depends on !PAGE_SIZE_256KB
1048 # This allows to use a set of generic functions to determine mmap base
1049 # address by giving priority to top-down scheme only if the process
1050 # is not in legacy mode (compat task, unlimited stack size or
1051 # sysctl_legacy_va_layout).
1052 # Architecture that selects this option can provide its own version of:
1054 config ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT
1057 select ARCH_HAS_ELF_RANDOMIZE
1062 config HAVE_JUMP_LABEL_HACK
1065 config HAVE_NOINSTR_HACK
1068 config HAVE_NOINSTR_VALIDATION
1071 config HAVE_UACCESS_VALIDATION
1075 config HAVE_STACK_VALIDATION
1078 Architecture supports objtool compile-time frame pointer rule
1081 config HAVE_RELIABLE_STACKTRACE
1084 Architecture has either save_stack_trace_tsk_reliable() or
1085 arch_stack_walk_reliable() function which only returns a stack trace
1086 if it can guarantee the trace is reliable.
1088 config HAVE_ARCH_HASH
1092 If this is set, the architecture provides an <asm/hash.h>
1093 file which provides platform-specific implementations of some
1094 functions in <linux/hash.h> or fs/namei.c.
1096 config HAVE_ARCH_NVRAM_OPS
1105 config CLONE_BACKWARDS
1108 Architecture has tls passed as the 4th argument of clone(2),
1111 config CLONE_BACKWARDS2
1114 Architecture has the first two arguments of clone(2) swapped.
1116 config CLONE_BACKWARDS3
1119 Architecture has tls passed as the 3rd argument of clone(2),
1122 config ODD_RT_SIGACTION
1125 Architecture has unusual rt_sigaction(2) arguments
1127 config OLD_SIGSUSPEND
1130 Architecture has old sigsuspend(2) syscall, of one-argument variety
1132 config OLD_SIGSUSPEND3
1135 Even weirder antique ABI - three-argument sigsuspend(2)
1137 config OLD_SIGACTION
1140 Architecture has old sigaction(2) syscall. Nope, not the same
1141 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
1142 but fairly different variant of sigaction(2), thanks to OSF/1
1145 config COMPAT_OLD_SIGACTION
1148 config COMPAT_32BIT_TIME
1149 bool "Provide system calls for 32-bit time_t"
1150 default !64BIT || COMPAT
1152 This enables 32 bit time_t support in addition to 64 bit time_t support.
1153 This is relevant on all 32-bit architectures, and 64-bit architectures
1154 as part of compat syscall handling.
1156 config ARCH_NO_PREEMPT
1159 config ARCH_EPHEMERAL_INODES
1162 An arch should select this symbol if it doesn't keep track of inode
1163 instances on its own, but instead relies on something else (e.g. the
1164 host kernel for an UML kernel).
1166 config ARCH_SUPPORTS_RT
1169 config CPU_NO_EFFICIENT_FFS
1172 config HAVE_ARCH_VMAP_STACK
1175 An arch should select this symbol if it can support kernel stacks
1176 in vmalloc space. This means:
1178 - vmalloc space must be large enough to hold many kernel stacks.
1179 This may rule out many 32-bit architectures.
1181 - Stacks in vmalloc space need to work reliably. For example, if
1182 vmap page tables are created on demand, either this mechanism
1183 needs to work while the stack points to a virtual address with
1184 unpopulated page tables or arch code (switch_to() and switch_mm(),
1185 most likely) needs to ensure that the stack's page table entries
1186 are populated before running on a possibly unpopulated stack.
1188 - If the stack overflows into a guard page, something reasonable
1189 should happen. The definition of "reasonable" is flexible, but
1190 instantly rebooting without logging anything would be unfriendly.
1194 bool "Use a virtually-mapped stack"
1195 depends on HAVE_ARCH_VMAP_STACK
1196 depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC
1198 Enable this if you want the use virtually-mapped kernel stacks
1199 with guard pages. This causes kernel stack overflows to be
1200 caught immediately rather than causing difficult-to-diagnose
1203 To use this with software KASAN modes, the architecture must support
1204 backing virtual mappings with real shadow memory, and KASAN_VMALLOC
1207 config HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET
1210 An arch should select this symbol if it can support kernel stack
1211 offset randomization with calls to add_random_kstack_offset()
1212 during syscall entry and choose_random_kstack_offset() during
1213 syscall exit. Careful removal of -fstack-protector-strong and
1214 -fstack-protector should also be applied to the entry code and
1215 closely examined, as the artificial stack bump looks like an array
1216 to the compiler, so it will attempt to add canary checks regardless
1217 of the static branch state.
1219 config RANDOMIZE_KSTACK_OFFSET
1220 bool "Support for randomizing kernel stack offset on syscall entry" if EXPERT
1222 depends on HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET
1223 depends on INIT_STACK_NONE || !CC_IS_CLANG || CLANG_VERSION >= 140000
1225 The kernel stack offset can be randomized (after pt_regs) by
1226 roughly 5 bits of entropy, frustrating memory corruption
1227 attacks that depend on stack address determinism or
1228 cross-syscall address exposures.
1230 The feature is controlled via the "randomize_kstack_offset=on/off"
1231 kernel boot param, and if turned off has zero overhead due to its use
1232 of static branches (see JUMP_LABEL).
1236 config RANDOMIZE_KSTACK_OFFSET_DEFAULT
1237 bool "Default state of kernel stack offset randomization"
1238 depends on RANDOMIZE_KSTACK_OFFSET
1240 Kernel stack offset randomization is controlled by kernel boot param
1241 "randomize_kstack_offset=on/off", and this config chooses the default
1244 config ARCH_OPTIONAL_KERNEL_RWX
1247 config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1250 config ARCH_HAS_STRICT_KERNEL_RWX
1253 config STRICT_KERNEL_RWX
1254 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX
1255 depends on ARCH_HAS_STRICT_KERNEL_RWX
1256 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1258 If this is set, kernel text and rodata memory will be made read-only,
1259 and non-text memory will be made non-executable. This provides
1260 protection against certain security exploits (e.g. executing the heap
1263 These features are considered standard security practice these days.
1264 You should say Y here in almost all cases.
1266 config ARCH_HAS_STRICT_MODULE_RWX
1269 config STRICT_MODULE_RWX
1270 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX
1271 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES
1272 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1274 If this is set, module text and rodata memory will be made read-only,
1275 and non-text memory will be made non-executable. This provides
1276 protection against certain security exploits (e.g. writing to text)
1278 # select if the architecture provides an asm/dma-direct.h header
1279 config ARCH_HAS_PHYS_TO_DMA
1282 config HAVE_ARCH_COMPILER_H
1285 An architecture can select this if it provides an
1286 asm/compiler.h header that should be included after
1287 linux/compiler-*.h in order to override macro definitions that those
1288 headers generally provide.
1290 config HAVE_ARCH_PREL32_RELOCATIONS
1293 May be selected by an architecture if it supports place-relative
1294 32-bit relocations, both in the toolchain and in the module loader,
1295 in which case relative references can be used in special sections
1296 for PCI fixup, initcalls etc which are only half the size on 64 bit
1297 architectures, and don't require runtime relocation on relocatable
1300 config ARCH_USE_MEMREMAP_PROT
1303 config LOCK_EVENT_COUNTS
1304 bool "Locking event counts collection"
1307 Enable light-weight counting of various locking related events
1308 in the system with minimal performance impact. This reduces
1309 the chance of application behavior change because of timing
1310 differences. The counts are reported via debugfs.
1312 # Select if the architecture has support for applying RELR relocations.
1313 config ARCH_HAS_RELR
1317 bool "Use RELR relocation packing"
1318 depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR
1321 Store the kernel's dynamic relocations in the RELR relocation packing
1322 format. Requires a compatible linker (LLD supports this feature), as
1323 well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy
1326 config ARCH_HAS_MEM_ENCRYPT
1329 config ARCH_HAS_CC_PLATFORM
1332 config HAVE_SPARSE_SYSCALL_NR
1335 An architecture should select this if its syscall numbering is sparse
1336 to save space. For example, MIPS architecture has a syscall array with
1337 entries at 4000, 5000 and 6000 locations. This option turns on syscall
1338 related optimizations for a given architecture.
1340 config ARCH_HAS_VDSO_DATA
1343 config HAVE_STATIC_CALL
1346 config HAVE_STATIC_CALL_INLINE
1348 depends on HAVE_STATIC_CALL
1351 config HAVE_PREEMPT_DYNAMIC
1354 config HAVE_PREEMPT_DYNAMIC_CALL
1356 depends on HAVE_STATIC_CALL
1357 select HAVE_PREEMPT_DYNAMIC
1359 An architecture should select this if it can handle the preemption
1360 model being selected at boot time using static calls.
1362 Where an architecture selects HAVE_STATIC_CALL_INLINE, any call to a
1363 preemption function will be patched directly.
1365 Where an architecture does not select HAVE_STATIC_CALL_INLINE, any
1366 call to a preemption function will go through a trampoline, and the
1367 trampoline will be patched.
1369 It is strongly advised to support inline static call to avoid any
1372 config HAVE_PREEMPT_DYNAMIC_KEY
1374 depends on HAVE_ARCH_JUMP_LABEL
1375 select HAVE_PREEMPT_DYNAMIC
1377 An architecture should select this if it can handle the preemption
1378 model being selected at boot time using static keys.
1380 Each preemption function will be given an early return based on a
1381 static key. This should have slightly lower overhead than non-inline
1382 static calls, as this effectively inlines each trampoline into the
1383 start of its callee. This may avoid redundant work, and may
1384 integrate better with CFI schemes.
1386 This will have greater overhead than using inline static calls as
1387 the call to the preemption function cannot be entirely elided.
1389 config ARCH_WANT_LD_ORPHAN_WARN
1392 An arch should select this symbol once all linker sections are explicitly
1393 included, size-asserted, or discarded in the linker scripts. This is
1394 important because we never want expected sections to be placed heuristically
1395 by the linker, since the locations of such sections can change between linker
1398 config HAVE_ARCH_PFN_VALID
1401 config ARCH_SUPPORTS_DEBUG_PAGEALLOC
1404 config ARCH_SUPPORTS_PAGE_TABLE_CHECK
1407 config ARCH_SPLIT_ARG64
1410 If a 32-bit architecture requires 64-bit arguments to be split into
1411 pairs of 32-bit arguments, select this option.
1413 config ARCH_HAS_ELFCORE_COMPAT
1416 config ARCH_HAS_PARANOID_L1D_FLUSH
1419 config ARCH_HAVE_TRACE_MMIO_ACCESS
1422 config DYNAMIC_SIGFRAME
1425 # Select, if arch has a named attribute group bound to NUMA device nodes.
1426 config HAVE_ARCH_NODE_DEV_GROUP
1429 config ARCH_HAS_NONLEAF_PMD_YOUNG
1432 Architectures that select this option are capable of setting the
1433 accessed bit in non-leaf PMD entries when using them as part of linear
1434 address translations. Page table walkers that clear the accessed bit
1435 may use this capability to reduce their search space.
1437 source "kernel/gcov/Kconfig"
1439 source "scripts/gcc-plugins/Kconfig"