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 config ARCH_CONFIGURES_CPU_MITIGATIONS
15 if !ARCH_CONFIGURES_CPU_MITIGATIONS
16 config CPU_MITIGATIONS
21 # Selected by architectures that need custom DMA operations for e.g. legacy
22 # IOMMUs not handled by dma-iommu. Drivers must never select this symbol.
24 config ARCH_HAS_DMA_OPS
26 select DMA_OPS_HELPERS
29 menu "General architecture-dependent options"
31 config ARCH_HAS_SUBPAGE_FAULTS
34 Select if the architecture can check permissions at sub-page
35 granularity (e.g. arm64 MTE). The probe_user_*() functions
41 config SMT_NUM_THREADS_DYNAMIC
44 # Selected by HOTPLUG_CORE_SYNC_DEAD or HOTPLUG_CORE_SYNC_FULL
45 config HOTPLUG_CORE_SYNC
48 # Basic CPU dead synchronization selected by architecture
49 config HOTPLUG_CORE_SYNC_DEAD
51 select HOTPLUG_CORE_SYNC
53 # Full CPU synchronization with alive state selected by architecture
54 config HOTPLUG_CORE_SYNC_FULL
56 select HOTPLUG_CORE_SYNC_DEAD if HOTPLUG_CPU
57 select HOTPLUG_CORE_SYNC
59 config HOTPLUG_SPLIT_STARTUP
61 select HOTPLUG_CORE_SYNC_FULL
63 config HOTPLUG_PARALLEL
65 select HOTPLUG_SPLIT_STARTUP
72 depends on HAVE_KPROBES
77 Kprobes allows you to trap at almost any kernel address and
78 execute a callback function. register_kprobe() establishes
79 a probepoint and specifies the callback. Kprobes is useful
80 for kernel debugging, non-intrusive instrumentation and testing.
84 bool "Optimize very unlikely/likely branches"
85 depends on HAVE_ARCH_JUMP_LABEL
86 select OBJTOOL if HAVE_JUMP_LABEL_HACK
88 This option enables a transparent branch optimization that
89 makes certain almost-always-true or almost-always-false branch
90 conditions even cheaper to execute within the kernel.
92 Certain performance-sensitive kernel code, such as trace points,
93 scheduler functionality, networking code and KVM have such
94 branches and include support for this optimization technique.
96 If it is detected that the compiler has support for "asm goto",
97 the kernel will compile such branches with just a nop
98 instruction. When the condition flag is toggled to true, the
99 nop will be converted to a jump instruction to execute the
100 conditional block of instructions.
102 This technique lowers overhead and stress on the branch prediction
103 of the processor and generally makes the kernel faster. The update
104 of the condition is slower, but those are always very rare.
106 ( On 32-bit x86, the necessary options added to the compiler
107 flags may increase the size of the kernel slightly. )
109 config STATIC_KEYS_SELFTEST
110 bool "Static key selftest"
111 depends on JUMP_LABEL
113 Boot time self-test of the branch patching code.
115 config STATIC_CALL_SELFTEST
116 bool "Static call selftest"
117 depends on HAVE_STATIC_CALL
119 Boot time self-test of the call patching code.
123 depends on KPROBES && HAVE_OPTPROBES
124 select NEED_TASKS_RCU
126 config KPROBES_ON_FTRACE
128 depends on KPROBES && HAVE_KPROBES_ON_FTRACE
129 depends on DYNAMIC_FTRACE_WITH_REGS
131 If function tracer is enabled and the arch supports full
132 passing of pt_regs to function tracing, then kprobes can
133 optimize on top of function tracing.
137 depends on ARCH_SUPPORTS_UPROBES
139 Uprobes is the user-space counterpart to kprobes: they
140 enable instrumentation applications (such as 'perf probe')
141 to establish unintrusive probes in user-space binaries and
142 libraries, by executing handler functions when the probes
143 are hit by user-space applications.
145 ( These probes come in the form of single-byte breakpoints,
146 managed by the kernel and kept transparent to the probed
149 config HAVE_64BIT_ALIGNED_ACCESS
150 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS
152 Some architectures require 64 bit accesses to be 64 bit
153 aligned, which also requires structs containing 64 bit values
154 to be 64 bit aligned too. This includes some 32 bit
155 architectures which can do 64 bit accesses, as well as 64 bit
156 architectures without unaligned access.
158 This symbol should be selected by an architecture if 64 bit
159 accesses are required to be 64 bit aligned in this way even
160 though it is not a 64 bit architecture.
162 See Documentation/core-api/unaligned-memory-access.rst for
163 more information on the topic of unaligned memory accesses.
165 config HAVE_EFFICIENT_UNALIGNED_ACCESS
168 Some architectures are unable to perform unaligned accesses
169 without the use of get_unaligned/put_unaligned. Others are
170 unable to perform such accesses efficiently (e.g. trap on
171 unaligned access and require fixing it up in the exception
174 This symbol should be selected by an architecture if it can
175 perform unaligned accesses efficiently to allow different
176 code paths to be selected for these cases. Some network
177 drivers, for example, could opt to not fix up alignment
178 problems with received packets if doing so would not help
181 See Documentation/core-api/unaligned-memory-access.rst for more
182 information on the topic of unaligned memory accesses.
184 config ARCH_USE_BUILTIN_BSWAP
187 Modern versions of GCC (since 4.4) have builtin functions
188 for handling byte-swapping. Using these, instead of the old
189 inline assembler that the architecture code provides in the
190 __arch_bswapXX() macros, allows the compiler to see what's
191 happening and offers more opportunity for optimisation. In
192 particular, the compiler will be able to combine the byteswap
193 with a nearby load or store and use load-and-swap or
194 store-and-swap instructions if the architecture has them. It
195 should almost *never* result in code which is worse than the
196 hand-coded assembler in <asm/swab.h>. But just in case it
197 does, the use of the builtins is optional.
199 Any architecture with load-and-swap or store-and-swap
200 instructions should set this. And it shouldn't hurt to set it
201 on architectures that don't have such instructions.
205 depends on KPROBES && (HAVE_KRETPROBES || HAVE_RETHOOK)
207 config KRETPROBE_ON_RETHOOK
209 depends on HAVE_RETHOOK
210 depends on KRETPROBES
213 config USER_RETURN_NOTIFIER
215 depends on HAVE_USER_RETURN_NOTIFIER
217 Provide a kernel-internal notification when a cpu is about to
220 config HAVE_IOREMAP_PROT
226 config HAVE_KRETPROBES
229 config HAVE_OPTPROBES
232 config HAVE_KPROBES_ON_FTRACE
235 config ARCH_CORRECT_STACKTRACE_ON_KRETPROBE
238 Since kretprobes modifies return address on the stack, the
239 stacktrace may see the kretprobe trampoline address instead
240 of correct one. If the architecture stacktrace code and
241 unwinder can adjust such entries, select this configuration.
243 config HAVE_FUNCTION_ERROR_INJECTION
249 config HAVE_FUNCTION_DESCRIPTORS
252 config TRACE_IRQFLAGS_SUPPORT
255 config TRACE_IRQFLAGS_NMI_SUPPORT
259 # An arch should select this if it provides all these things:
261 # task_pt_regs() in asm/processor.h or asm/ptrace.h
262 # arch_has_single_step() if there is hardware single-step support
263 # arch_has_block_step() if there is hardware block-step support
264 # asm/syscall.h supplying asm-generic/syscall.h interface
265 # linux/regset.h user_regset interfaces
266 # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
267 # TIF_SYSCALL_TRACE calls ptrace_report_syscall_{entry,exit}
268 # TIF_NOTIFY_RESUME calls resume_user_mode_work()
270 config HAVE_ARCH_TRACEHOOK
273 config HAVE_DMA_CONTIGUOUS
276 config GENERIC_SMP_IDLE_THREAD
279 config GENERIC_IDLE_POLL_SETUP
282 config ARCH_HAS_FORTIFY_SOURCE
285 An architecture should select this when it can successfully
286 build and run with CONFIG_FORTIFY_SOURCE.
289 # Select if the arch provides a historic keepinit alias for the retain_initrd
290 # command line option
292 config ARCH_HAS_KEEPINITRD
295 # Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h
296 config ARCH_HAS_SET_MEMORY
299 # Select if arch has all set_direct_map_invalid/default() functions
300 config ARCH_HAS_SET_DIRECT_MAP
304 # Select if the architecture provides the arch_dma_set_uncached symbol to
305 # either provide an uncached segment alias for a DMA allocation, or
306 # to remap the page tables in place.
308 config ARCH_HAS_DMA_SET_UNCACHED
312 # Select if the architectures provides the arch_dma_clear_uncached symbol
313 # to undo an in-place page table remap for uncached access.
315 config ARCH_HAS_DMA_CLEAR_UNCACHED
318 config ARCH_HAS_CPU_FINALIZE_INIT
321 # The architecture has a per-task state that includes the mm's PASID
322 config ARCH_HAS_CPU_PASID
326 config HAVE_ARCH_THREAD_STRUCT_WHITELIST
329 An architecture should select this to provide hardened usercopy
330 knowledge about what region of the thread_struct should be
331 whitelisted for copying to userspace. Normally this is only the
332 FPU registers. Specifically, arch_thread_struct_whitelist()
333 should be implemented. Without this, the entire thread_struct
334 field in task_struct will be left whitelisted.
336 # Select if arch wants to size task_struct dynamically via arch_task_struct_size:
337 config ARCH_WANTS_DYNAMIC_TASK_STRUCT
340 config ARCH_WANTS_NO_INSTR
343 An architecture should select this if the noinstr macro is being used on
344 functions to denote that the toolchain should avoid instrumenting such
345 functions and is required for correctness.
347 config ARCH_32BIT_OFF_T
351 All new 32-bit architectures should have 64-bit off_t type on
352 userspace side which corresponds to the loff_t kernel type. This
353 is the requirement for modern ABIs. Some existing architectures
354 still support 32-bit off_t. This option is enabled for all such
355 architectures explicitly.
357 # Selected by 64 bit architectures which have a 32 bit f_tinode in struct ustat
358 config ARCH_32BIT_USTAT_F_TINODE
361 config HAVE_ASM_MODVERSIONS
364 This symbol should be selected by an architecture if it provides
365 <asm/asm-prototypes.h> to support the module versioning for symbols
366 exported from assembly code.
368 config HAVE_REGS_AND_STACK_ACCESS_API
371 This symbol should be selected by an architecture if it supports
372 the API needed to access registers and stack entries from pt_regs,
373 declared in asm/ptrace.h
374 For example the kprobes-based event tracer needs this API.
378 depends on HAVE_REGS_AND_STACK_ACCESS_API
380 This symbol should be selected by an architecture if it
381 supports an implementation of restartable sequences.
386 This symbol should be selected by an architecture if it
389 config HAVE_FUNCTION_ARG_ACCESS_API
392 This symbol should be selected by an architecture if it supports
393 the API needed to access function arguments from pt_regs,
394 declared in asm/ptrace.h
396 config HAVE_HW_BREAKPOINT
398 depends on PERF_EVENTS
400 config HAVE_MIXED_BREAKPOINTS_REGS
402 depends on HAVE_HW_BREAKPOINT
404 Depending on the arch implementation of hardware breakpoints,
405 some of them have separate registers for data and instruction
406 breakpoints addresses, others have mixed registers to store
407 them but define the access type in a control register.
408 Select this option if your arch implements breakpoints under the
411 config HAVE_USER_RETURN_NOTIFIER
414 config HAVE_PERF_EVENTS_NMI
417 System hardware can generate an NMI using the perf event
418 subsystem. Also has support for calculating CPU cycle events
419 to determine how many clock cycles in a given period.
421 config HAVE_HARDLOCKUP_DETECTOR_PERF
423 depends on HAVE_PERF_EVENTS_NMI
425 The arch chooses to use the generic perf-NMI-based hardlockup
426 detector. Must define HAVE_PERF_EVENTS_NMI.
428 config HAVE_HARDLOCKUP_DETECTOR_ARCH
431 The arch provides its own hardlockup detector implementation instead
434 It uses the same command line parameters, and sysctl interface,
435 as the generic hardlockup detectors.
437 config HAVE_PERF_REGS
440 Support selective register dumps for perf events. This includes
441 bit-mapping of each registers and a unique architecture id.
443 config HAVE_PERF_USER_STACK_DUMP
446 Support user stack dumps for perf event samples. This needs
447 access to the user stack pointer which is not unified across
450 config HAVE_ARCH_JUMP_LABEL
453 config HAVE_ARCH_JUMP_LABEL_RELATIVE
456 config MMU_GATHER_TABLE_FREE
459 config MMU_GATHER_RCU_TABLE_FREE
461 select MMU_GATHER_TABLE_FREE
463 config MMU_GATHER_PAGE_SIZE
466 config MMU_GATHER_NO_RANGE
468 select MMU_GATHER_MERGE_VMAS
470 config MMU_GATHER_NO_FLUSH_CACHE
473 config MMU_GATHER_MERGE_VMAS
476 config MMU_GATHER_NO_GATHER
478 depends on MMU_GATHER_TABLE_FREE
480 config ARCH_WANT_IRQS_OFF_ACTIVATE_MM
483 Temporary select until all architectures can be converted to have
484 irqs disabled over activate_mm. Architectures that do IPI based TLB
485 shootdowns should enable this.
487 # Use normal mm refcounting for MMU_LAZY_TLB kernel thread references.
488 # MMU_LAZY_TLB_REFCOUNT=n can improve the scalability of context switching
489 # to/from kernel threads when the same mm is running on a lot of CPUs (a large
490 # multi-threaded application), by reducing contention on the mm refcount.
492 # This can be disabled if the architecture ensures no CPUs are using an mm as a
493 # "lazy tlb" beyond its final refcount (i.e., by the time __mmdrop frees the mm
494 # or its kernel page tables). This could be arranged by arch_exit_mmap(), or
495 # final exit(2) TLB flush, for example.
497 # To implement this, an arch *must*:
498 # Ensure the _lazy_tlb variants of mmgrab/mmdrop are used when manipulating
499 # the lazy tlb reference of a kthread's ->active_mm (non-arch code has been
500 # converted already).
501 config MMU_LAZY_TLB_REFCOUNT
503 depends on !MMU_LAZY_TLB_SHOOTDOWN
505 # This option allows MMU_LAZY_TLB_REFCOUNT=n. It ensures no CPUs are using an
506 # mm as a lazy tlb beyond its last reference count, by shooting down these
507 # users before the mm is deallocated. __mmdrop() first IPIs all CPUs that may
508 # be using the mm as a lazy tlb, so that they may switch themselves to using
509 # init_mm for their active mm. mm_cpumask(mm) is used to determine which CPUs
510 # may be using mm as a lazy tlb mm.
512 # To implement this, an arch *must*:
513 # - At the time of the final mmdrop of the mm, ensure mm_cpumask(mm) contains
514 # at least all possible CPUs in which the mm is lazy.
515 # - It must meet the requirements for MMU_LAZY_TLB_REFCOUNT=n (see above).
516 config MMU_LAZY_TLB_SHOOTDOWN
519 config ARCH_HAVE_NMI_SAFE_CMPXCHG
522 config ARCH_HAVE_EXTRA_ELF_NOTES
525 An architecture should select this in order to enable adding an
526 arch-specific ELF note section to core files. It must provide two
527 functions: elf_coredump_extra_notes_size() and
528 elf_coredump_extra_notes_write() which are invoked by the ELF core
531 config ARCH_HAS_NMI_SAFE_THIS_CPU_OPS
534 config HAVE_ALIGNED_STRUCT_PAGE
537 This makes sure that struct pages are double word aligned and that
538 e.g. the SLUB allocator can perform double word atomic operations
539 on a struct page for better performance. However selecting this
540 might increase the size of a struct page by a word.
542 config HAVE_CMPXCHG_LOCAL
545 config HAVE_CMPXCHG_DOUBLE
548 config ARCH_WEAK_RELEASE_ACQUIRE
551 config ARCH_WANT_IPC_PARSE_VERSION
554 config ARCH_WANT_COMPAT_IPC_PARSE_VERSION
557 config ARCH_WANT_OLD_COMPAT_IPC
558 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
561 config HAVE_ARCH_SECCOMP
564 An arch should select this symbol to support seccomp mode 1 (the fixed
565 syscall policy), and must provide an overrides for __NR_seccomp_sigreturn,
566 and compat syscalls if the asm-generic/seccomp.h defaults need adjustment:
567 - __NR_seccomp_read_32
568 - __NR_seccomp_write_32
569 - __NR_seccomp_exit_32
570 - __NR_seccomp_sigreturn_32
572 config HAVE_ARCH_SECCOMP_FILTER
574 select HAVE_ARCH_SECCOMP
576 An arch should select this symbol if it provides all of these things:
577 - all the requirements for HAVE_ARCH_SECCOMP
579 - syscall_get_arguments()
581 - syscall_set_return_value()
582 - SIGSYS siginfo_t support
583 - secure_computing is called from a ptrace_event()-safe context
584 - secure_computing return value is checked and a return value of -1
585 results in the system call being skipped immediately.
586 - seccomp syscall wired up
587 - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE,
588 SECCOMP_ARCH_NATIVE_NR, SECCOMP_ARCH_NATIVE_NAME defined. If
589 COMPAT is supported, have the SECCOMP_ARCH_COMPAT* defines too.
592 prompt "Enable seccomp to safely execute untrusted bytecode"
594 depends on HAVE_ARCH_SECCOMP
596 This kernel feature is useful for number crunching applications
597 that may need to handle untrusted bytecode during their
598 execution. By using pipes or other transports made available
599 to the process as file descriptors supporting the read/write
600 syscalls, it's possible to isolate those applications in their
601 own address space using seccomp. Once seccomp is enabled via
602 prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be
603 disabled and the task is only allowed to execute a few safe
604 syscalls defined by each seccomp mode.
608 config SECCOMP_FILTER
610 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
612 Enable tasks to build secure computing environments defined
613 in terms of Berkeley Packet Filter programs which implement
614 task-defined system call filtering polices.
616 See Documentation/userspace-api/seccomp_filter.rst for details.
618 config SECCOMP_CACHE_DEBUG
619 bool "Show seccomp filter cache status in /proc/pid/seccomp_cache"
620 depends on SECCOMP_FILTER && !HAVE_SPARSE_SYSCALL_NR
623 This enables the /proc/pid/seccomp_cache interface to monitor
624 seccomp cache data. The file format is subject to change. Reading
625 the file requires CAP_SYS_ADMIN.
627 This option is for debugging only. Enabling presents the risk that
628 an adversary may be able to infer the seccomp filter logic.
632 config HAVE_ARCH_STACKLEAK
635 An architecture should select this if it has the code which
636 fills the used part of the kernel stack with the STACKLEAK_POISON
637 value before returning from system calls.
639 config HAVE_STACKPROTECTOR
642 An arch should select this symbol if:
643 - it has implemented a stack canary (e.g. __stack_chk_guard)
645 config STACKPROTECTOR
646 bool "Stack Protector buffer overflow detection"
647 depends on HAVE_STACKPROTECTOR
648 depends on $(cc-option,-fstack-protector)
651 This option turns on the "stack-protector" GCC feature. This
652 feature puts, at the beginning of functions, a canary value on
653 the stack just before the return address, and validates
654 the value just before actually returning. Stack based buffer
655 overflows (that need to overwrite this return address) now also
656 overwrite the canary, which gets detected and the attack is then
657 neutralized via a kernel panic.
659 Functions will have the stack-protector canary logic added if they
660 have an 8-byte or larger character array on the stack.
662 This feature requires gcc version 4.2 or above, or a distribution
663 gcc with the feature backported ("-fstack-protector").
665 On an x86 "defconfig" build, this feature adds canary checks to
666 about 3% of all kernel functions, which increases kernel code size
669 config STACKPROTECTOR_STRONG
670 bool "Strong Stack Protector"
671 depends on STACKPROTECTOR
672 depends on $(cc-option,-fstack-protector-strong)
675 Functions will have the stack-protector canary logic added in any
676 of the following conditions:
678 - local variable's address used as part of the right hand side of an
679 assignment or function argument
680 - local variable is an array (or union containing an array),
681 regardless of array type or length
682 - uses register local variables
684 This feature requires gcc version 4.9 or above, or a distribution
685 gcc with the feature backported ("-fstack-protector-strong").
687 On an x86 "defconfig" build, this feature adds canary checks to
688 about 20% of all kernel functions, which increases the kernel code
691 config ARCH_SUPPORTS_SHADOW_CALL_STACK
694 An architecture should select this if it supports the compiler's
695 Shadow Call Stack and implements runtime support for shadow stack
698 config SHADOW_CALL_STACK
699 bool "Shadow Call Stack"
700 depends on ARCH_SUPPORTS_SHADOW_CALL_STACK
701 depends on DYNAMIC_FTRACE_WITH_ARGS || DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER
704 This option enables the compiler's Shadow Call Stack, which
705 uses a shadow stack to protect function return addresses from
706 being overwritten by an attacker. More information can be found
707 in the compiler's documentation:
709 - Clang: https://clang.llvm.org/docs/ShadowCallStack.html
710 - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options
712 Note that security guarantees in the kernel differ from the
713 ones documented for user space. The kernel must store addresses
714 of shadow stacks in memory, which means an attacker capable of
715 reading and writing arbitrary memory may be able to locate them
716 and hijack control flow by modifying the stacks.
721 Set by the arch code if it relies on code patching to insert the
722 shadow call stack push and pop instructions rather than on the
728 Selected if the kernel will be built using the compiler's LTO feature.
734 Selected if the kernel will be built using Clang's LTO feature.
736 config ARCH_SUPPORTS_LTO_CLANG
739 An architecture should select this option if it supports:
740 - compiling with Clang,
741 - compiling inline assembly with Clang's integrated assembler,
742 - and linking with LLD.
744 config ARCH_SUPPORTS_LTO_CLANG_THIN
747 An architecture should select this option if it can support Clang's
752 depends on CC_IS_CLANG && LD_IS_LLD && AS_IS_LLVM
753 depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm)
754 depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm)
755 depends on ARCH_SUPPORTS_LTO_CLANG
756 depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT
757 # https://github.com/ClangBuiltLinux/linux/issues/1721
758 depends on (!KASAN || KASAN_HW_TAGS || CLANG_VERSION >= 170000) || !DEBUG_INFO
759 depends on (!KCOV || CLANG_VERSION >= 170000) || !DEBUG_INFO
760 depends on !GCOV_KERNEL
762 The compiler and Kconfig options support building with Clang's
766 prompt "Link Time Optimization (LTO)"
769 This option enables Link Time Optimization (LTO), which allows the
770 compiler to optimize binaries globally.
772 If unsure, select LTO_NONE. Note that LTO is very resource-intensive
773 so it's disabled by default.
778 Build the kernel normally, without Link Time Optimization (LTO).
780 config LTO_CLANG_FULL
781 bool "Clang Full LTO (EXPERIMENTAL)"
782 depends on HAS_LTO_CLANG
783 depends on !COMPILE_TEST
786 This option enables Clang's full Link Time Optimization (LTO), which
787 allows the compiler to optimize the kernel globally. If you enable
788 this option, the compiler generates LLVM bitcode instead of ELF
789 object files, and the actual compilation from bitcode happens at
790 the LTO link step, which may take several minutes depending on the
791 kernel configuration. More information can be found from LLVM's
794 https://llvm.org/docs/LinkTimeOptimization.html
796 During link time, this option can use a large amount of RAM, and
797 may take much longer than the ThinLTO option.
799 config LTO_CLANG_THIN
800 bool "Clang ThinLTO (EXPERIMENTAL)"
801 depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN
804 This option enables Clang's ThinLTO, which allows for parallel
805 optimization and faster incremental compiles compared to the
806 CONFIG_LTO_CLANG_FULL option. More information can be found
807 from Clang's documentation:
809 https://clang.llvm.org/docs/ThinLTO.html
814 config ARCH_SUPPORTS_CFI_CLANG
817 An architecture should select this option if it can support Clang's
818 Control-Flow Integrity (CFI) checking.
820 config ARCH_USES_CFI_TRAPS
824 bool "Use Clang's Control Flow Integrity (CFI)"
825 depends on ARCH_SUPPORTS_CFI_CLANG
826 depends on $(cc-option,-fsanitize=kcfi)
828 This option enables Clang's forward-edge Control Flow Integrity
829 (CFI) checking, where the compiler injects a runtime check to each
830 indirect function call to ensure the target is a valid function with
831 the correct static type. This restricts possible call targets and
832 makes it more difficult for an attacker to exploit bugs that allow
833 the modification of stored function pointers. More information can be
834 found from Clang's documentation:
836 https://clang.llvm.org/docs/ControlFlowIntegrity.html
838 config CFI_ICALL_NORMALIZE_INTEGERS
839 bool "Normalize CFI tags for integers"
841 depends on $(cc-option,-fsanitize=kcfi -fsanitize-cfi-icall-experimental-normalize-integers)
843 This option normalizes the CFI tags for integer types so that all
844 integer types of the same size and signedness receive the same CFI
847 The option is separate from CONFIG_RUST because it affects the ABI.
848 When working with build systems that care about the ABI, it is
849 convenient to be able to turn on this flag first, before Rust is
852 This option is necessary for using CFI with Rust. If unsure, say N.
854 config CFI_PERMISSIVE
855 bool "Use CFI in permissive mode"
858 When selected, Control Flow Integrity (CFI) violations result in a
859 warning instead of a kernel panic. This option should only be used
860 for finding indirect call type mismatches during development.
864 config HAVE_ARCH_WITHIN_STACK_FRAMES
867 An architecture should select this if it can walk the kernel stack
868 frames to determine if an object is part of either the arguments
869 or local variables (i.e. that it excludes saved return addresses,
870 and similar) by implementing an inline arch_within_stack_frames(),
871 which is used by CONFIG_HARDENED_USERCOPY.
873 config HAVE_CONTEXT_TRACKING_USER
876 Provide kernel/user boundaries probes necessary for subsystems
877 that need it, such as userspace RCU extended quiescent state.
878 Syscalls need to be wrapped inside user_exit()-user_enter(), either
879 optimized behind static key or through the slow path using TIF_NOHZ
880 flag. Exceptions handlers must be wrapped as well. Irqs are already
881 protected inside ct_irq_enter/ct_irq_exit() but preemption or signal
882 handling on irq exit still need to be protected.
884 config HAVE_CONTEXT_TRACKING_USER_OFFSTACK
887 Architecture neither relies on exception_enter()/exception_exit()
888 nor on schedule_user(). Also preempt_schedule_notrace() and
889 preempt_schedule_irq() can't be called in a preemptible section
890 while context tracking is CT_STATE_USER. This feature reflects a sane
891 entry implementation where the following requirements are met on
892 critical entry code, ie: before user_exit() or after user_enter():
894 - Critical entry code isn't preemptible (or better yet:
896 - No use of RCU read side critical sections, unless ct_nmi_enter()
898 - No use of instrumentation, unless instrumentation_begin() got
904 Arch relies on TIF_NOHZ and syscall slow path to implement context
905 tracking calls to user_enter()/user_exit().
907 config HAVE_VIRT_CPU_ACCOUNTING
910 config HAVE_VIRT_CPU_ACCOUNTING_IDLE
913 Architecture has its own way to account idle CPU time and therefore
914 doesn't implement vtime_account_idle().
916 config ARCH_HAS_SCALED_CPUTIME
919 config HAVE_VIRT_CPU_ACCOUNTING_GEN
923 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
924 Before enabling this option, arch code must be audited
925 to ensure there are no races in concurrent read/write of
926 cputime_t. For example, reading/writing 64-bit cputime_t on
927 some 32-bit arches may require multiple accesses, so proper
928 locking is needed to protect against concurrent accesses.
930 config HAVE_IRQ_TIME_ACCOUNTING
933 Archs need to ensure they use a high enough resolution clock to
934 support irq time accounting and then call enable_sched_clock_irqtime().
939 Architectures that select this are able to move page tables at the
940 PUD level. If there are only 3 page table levels, the move effectively
941 happens at the PGD level.
946 Archs that select this are able to move page tables at the PMD level.
948 config HAVE_ARCH_TRANSPARENT_HUGEPAGE
951 config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
954 config HAVE_ARCH_HUGE_VMAP
958 # Archs that select this would be capable of PMD-sized vmaps (i.e.,
959 # arch_vmap_pmd_supported() returns true). The VM_ALLOW_HUGE_VMAP flag
960 # must be used to enable allocations to use hugepages.
962 config HAVE_ARCH_HUGE_VMALLOC
963 depends on HAVE_ARCH_HUGE_VMAP
966 config ARCH_WANT_HUGE_PMD_SHARE
969 # Archs that want to use pmd_mkwrite on kernel memory need it defined even
970 # if there are no userspace memory management features that use it
971 config ARCH_WANT_KERNEL_PMD_MKWRITE
974 config ARCH_WANT_PMD_MKWRITE
975 def_bool TRANSPARENT_HUGEPAGE || ARCH_WANT_KERNEL_PMD_MKWRITE
977 config HAVE_ARCH_SOFT_DIRTY
980 config HAVE_MOD_ARCH_SPECIFIC
983 The arch uses struct mod_arch_specific to store data. Many arches
984 just need a simple module loader without arch specific data - those
985 should not enable this.
987 config MODULES_USE_ELF_RELA
990 Modules only use ELF RELA relocations. Modules with ELF REL
991 relocations will give an error.
993 config MODULES_USE_ELF_REL
996 Modules only use ELF REL relocations. Modules with ELF RELA
997 relocations will give an error.
999 config ARCH_WANTS_MODULES_DATA_IN_VMALLOC
1002 For architectures like powerpc/32 which have constraints on module
1003 allocation and need to allocate module data outside of module area.
1005 config ARCH_WANTS_EXECMEM_LATE
1008 For architectures that do not allocate executable memory early on
1009 boot, but rather require its initialization late when there is
1010 enough entropy for module space randomization, for instance
1013 config HAVE_IRQ_EXIT_ON_IRQ_STACK
1016 Architecture doesn't only execute the irq handler on the irq stack
1017 but also irq_exit(). This way we can process softirqs on this irq
1018 stack instead of switching to a new one when we call __do_softirq()
1019 in the end of an hardirq.
1020 This spares a stack switch and improves cache usage on softirq
1023 config HAVE_SOFTIRQ_ON_OWN_STACK
1026 Architecture provides a function to run __do_softirq() on a
1029 config SOFTIRQ_ON_OWN_STACK
1030 def_bool HAVE_SOFTIRQ_ON_OWN_STACK && !PREEMPT_RT
1032 config ALTERNATE_USER_ADDRESS_SPACE
1035 Architectures set this when the CPU uses separate address
1036 spaces for kernel and user space pointers. In this case, the
1037 access_ok() check on a __user pointer is skipped.
1039 config PGTABLE_LEVELS
1043 config ARCH_HAS_ELF_RANDOMIZE
1046 An architecture supports choosing randomized locations for
1047 stack, mmap, brk, and ET_DYN. Defined functions:
1049 - arch_randomize_brk()
1051 config HAVE_ARCH_MMAP_RND_BITS
1054 An arch should select this symbol if it supports setting a variable
1055 number of bits for use in establishing the base address for mmap
1056 allocations, has MMU enabled and provides values for both:
1057 - ARCH_MMAP_RND_BITS_MIN
1058 - ARCH_MMAP_RND_BITS_MAX
1060 config HAVE_EXIT_THREAD
1063 An architecture implements exit_thread.
1065 config ARCH_MMAP_RND_BITS_MIN
1068 config ARCH_MMAP_RND_BITS_MAX
1071 config ARCH_MMAP_RND_BITS_DEFAULT
1074 config ARCH_MMAP_RND_BITS
1075 int "Number of bits to use for ASLR of mmap base address" if EXPERT
1076 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX
1077 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT
1078 default ARCH_MMAP_RND_BITS_MIN
1079 depends on HAVE_ARCH_MMAP_RND_BITS
1081 This value can be used to select the number of bits to use to
1082 determine the random offset to the base address of vma regions
1083 resulting from mmap allocations. This value will be bounded
1084 by the architecture's minimum and maximum supported values.
1086 This value can be changed after boot using the
1087 /proc/sys/vm/mmap_rnd_bits tunable
1089 config HAVE_ARCH_MMAP_RND_COMPAT_BITS
1092 An arch should select this symbol if it supports running applications
1093 in compatibility mode, supports setting a variable number of bits for
1094 use in establishing the base address for mmap allocations, has MMU
1095 enabled and provides values for both:
1096 - ARCH_MMAP_RND_COMPAT_BITS_MIN
1097 - ARCH_MMAP_RND_COMPAT_BITS_MAX
1099 config ARCH_MMAP_RND_COMPAT_BITS_MIN
1102 config ARCH_MMAP_RND_COMPAT_BITS_MAX
1105 config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
1108 config ARCH_MMAP_RND_COMPAT_BITS
1109 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT
1110 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX
1111 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
1112 default ARCH_MMAP_RND_COMPAT_BITS_MIN
1113 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS
1115 This value can be used to select the number of bits to use to
1116 determine the random offset to the base address of vma regions
1117 resulting from mmap allocations for compatible applications This
1118 value will be bounded by the architecture's minimum and maximum
1121 This value can be changed after boot using the
1122 /proc/sys/vm/mmap_rnd_compat_bits tunable
1124 config HAVE_ARCH_COMPAT_MMAP_BASES
1127 This allows 64bit applications to invoke 32-bit mmap() syscall
1128 and vice-versa 32-bit applications to call 64-bit mmap().
1129 Required for applications doing different bitness syscalls.
1131 config HAVE_PAGE_SIZE_4KB
1134 config HAVE_PAGE_SIZE_8KB
1137 config HAVE_PAGE_SIZE_16KB
1140 config HAVE_PAGE_SIZE_32KB
1143 config HAVE_PAGE_SIZE_64KB
1146 config HAVE_PAGE_SIZE_256KB
1150 prompt "MMU page size"
1152 config PAGE_SIZE_4KB
1154 depends on HAVE_PAGE_SIZE_4KB
1156 This option select the standard 4KiB Linux page size and the only
1157 available option on many architectures. Using 4KiB page size will
1158 minimize memory consumption and is therefore recommended for low
1160 Some software that is written for x86 systems makes incorrect
1161 assumptions about the page size and only runs on 4KiB pages.
1163 config PAGE_SIZE_8KB
1165 depends on HAVE_PAGE_SIZE_8KB
1167 This option is the only supported page size on a few older
1168 processors, and can be slightly faster than 4KiB pages.
1170 config PAGE_SIZE_16KB
1172 depends on HAVE_PAGE_SIZE_16KB
1174 This option is usually a good compromise between memory
1175 consumption and performance for typical desktop and server
1176 workloads, often saving a level of page table lookups compared
1177 to 4KB pages as well as reducing TLB pressure and overhead of
1178 per-page operations in the kernel at the expense of a larger
1181 config PAGE_SIZE_32KB
1183 depends on HAVE_PAGE_SIZE_32KB
1185 Using 32KiB page size will result in slightly higher performance
1186 kernel at the price of higher memory consumption compared to
1187 16KiB pages. This option is available only on cnMIPS cores.
1188 Note that you will need a suitable Linux distribution to
1191 config PAGE_SIZE_64KB
1193 depends on HAVE_PAGE_SIZE_64KB
1195 Using 64KiB page size will result in slightly higher performance
1196 kernel at the price of much higher memory consumption compared to
1197 4KiB or 16KiB pages.
1198 This is not suitable for general-purpose workloads but the
1199 better performance may be worth the cost for certain types of
1200 supercomputing or database applications that work mostly with
1201 large in-memory data rather than small files.
1203 config PAGE_SIZE_256KB
1205 depends on HAVE_PAGE_SIZE_256KB
1207 256KiB pages have little practical value due to their extreme
1208 memory usage. The kernel will only be able to run applications
1209 that have been compiled with '-zmax-page-size' set to 256KiB
1210 (the default is 64KiB or 4KiB on most architectures).
1214 config PAGE_SIZE_LESS_THAN_64KB
1216 depends on !PAGE_SIZE_64KB
1217 depends on PAGE_SIZE_LESS_THAN_256KB
1219 config PAGE_SIZE_LESS_THAN_256KB
1221 depends on !PAGE_SIZE_256KB
1225 default 12 if PAGE_SIZE_4KB
1226 default 13 if PAGE_SIZE_8KB
1227 default 14 if PAGE_SIZE_16KB
1228 default 15 if PAGE_SIZE_32KB
1229 default 16 if PAGE_SIZE_64KB
1230 default 18 if PAGE_SIZE_256KB
1232 # This allows to use a set of generic functions to determine mmap base
1233 # address by giving priority to top-down scheme only if the process
1234 # is not in legacy mode (compat task, unlimited stack size or
1235 # sysctl_legacy_va_layout).
1236 # Architecture that selects this option can provide its own version of:
1238 config ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT
1241 select ARCH_HAS_ELF_RANDOMIZE
1246 config HAVE_JUMP_LABEL_HACK
1249 config HAVE_NOINSTR_HACK
1252 config HAVE_NOINSTR_VALIDATION
1255 config HAVE_UACCESS_VALIDATION
1259 config HAVE_STACK_VALIDATION
1262 Architecture supports objtool compile-time frame pointer rule
1265 config HAVE_RELIABLE_STACKTRACE
1268 Architecture has either save_stack_trace_tsk_reliable() or
1269 arch_stack_walk_reliable() function which only returns a stack trace
1270 if it can guarantee the trace is reliable.
1272 config HAVE_ARCH_HASH
1276 If this is set, the architecture provides an <asm/hash.h>
1277 file which provides platform-specific implementations of some
1278 functions in <linux/hash.h> or fs/namei.c.
1280 config HAVE_ARCH_NVRAM_OPS
1289 config CLONE_BACKWARDS
1292 Architecture has tls passed as the 4th argument of clone(2),
1295 config CLONE_BACKWARDS2
1298 Architecture has the first two arguments of clone(2) swapped.
1300 config CLONE_BACKWARDS3
1303 Architecture has tls passed as the 3rd argument of clone(2),
1306 config ODD_RT_SIGACTION
1309 Architecture has unusual rt_sigaction(2) arguments
1311 config OLD_SIGSUSPEND
1314 Architecture has old sigsuspend(2) syscall, of one-argument variety
1316 config OLD_SIGSUSPEND3
1319 Even weirder antique ABI - three-argument sigsuspend(2)
1321 config OLD_SIGACTION
1324 Architecture has old sigaction(2) syscall. Nope, not the same
1325 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
1326 but fairly different variant of sigaction(2), thanks to OSF/1
1329 config COMPAT_OLD_SIGACTION
1332 config COMPAT_32BIT_TIME
1333 bool "Provide system calls for 32-bit time_t"
1334 default !64BIT || COMPAT
1336 This enables 32 bit time_t support in addition to 64 bit time_t support.
1337 This is relevant on all 32-bit architectures, and 64-bit architectures
1338 as part of compat syscall handling.
1340 config ARCH_NO_PREEMPT
1343 config ARCH_SUPPORTS_RT
1346 config CPU_NO_EFFICIENT_FFS
1349 config HAVE_ARCH_VMAP_STACK
1352 An arch should select this symbol if it can support kernel stacks
1353 in vmalloc space. This means:
1355 - vmalloc space must be large enough to hold many kernel stacks.
1356 This may rule out many 32-bit architectures.
1358 - Stacks in vmalloc space need to work reliably. For example, if
1359 vmap page tables are created on demand, either this mechanism
1360 needs to work while the stack points to a virtual address with
1361 unpopulated page tables or arch code (switch_to() and switch_mm(),
1362 most likely) needs to ensure that the stack's page table entries
1363 are populated before running on a possibly unpopulated stack.
1365 - If the stack overflows into a guard page, something reasonable
1366 should happen. The definition of "reasonable" is flexible, but
1367 instantly rebooting without logging anything would be unfriendly.
1371 bool "Use a virtually-mapped stack"
1372 depends on HAVE_ARCH_VMAP_STACK
1373 depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC
1375 Enable this if you want the use virtually-mapped kernel stacks
1376 with guard pages. This causes kernel stack overflows to be
1377 caught immediately rather than causing difficult-to-diagnose
1380 To use this with software KASAN modes, the architecture must support
1381 backing virtual mappings with real shadow memory, and KASAN_VMALLOC
1384 config HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET
1387 An arch should select this symbol if it can support kernel stack
1388 offset randomization with calls to add_random_kstack_offset()
1389 during syscall entry and choose_random_kstack_offset() during
1390 syscall exit. Careful removal of -fstack-protector-strong and
1391 -fstack-protector should also be applied to the entry code and
1392 closely examined, as the artificial stack bump looks like an array
1393 to the compiler, so it will attempt to add canary checks regardless
1394 of the static branch state.
1396 config RANDOMIZE_KSTACK_OFFSET
1397 bool "Support for randomizing kernel stack offset on syscall entry" if EXPERT
1399 depends on HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET
1400 depends on INIT_STACK_NONE || !CC_IS_CLANG || CLANG_VERSION >= 140000
1402 The kernel stack offset can be randomized (after pt_regs) by
1403 roughly 5 bits of entropy, frustrating memory corruption
1404 attacks that depend on stack address determinism or
1405 cross-syscall address exposures.
1407 The feature is controlled via the "randomize_kstack_offset=on/off"
1408 kernel boot param, and if turned off has zero overhead due to its use
1409 of static branches (see JUMP_LABEL).
1413 config RANDOMIZE_KSTACK_OFFSET_DEFAULT
1414 bool "Default state of kernel stack offset randomization"
1415 depends on RANDOMIZE_KSTACK_OFFSET
1417 Kernel stack offset randomization is controlled by kernel boot param
1418 "randomize_kstack_offset=on/off", and this config chooses the default
1421 config ARCH_OPTIONAL_KERNEL_RWX
1424 config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1427 config ARCH_HAS_STRICT_KERNEL_RWX
1430 config STRICT_KERNEL_RWX
1431 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX
1432 depends on ARCH_HAS_STRICT_KERNEL_RWX
1433 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1435 If this is set, kernel text and rodata memory will be made read-only,
1436 and non-text memory will be made non-executable. This provides
1437 protection against certain security exploits (e.g. executing the heap
1440 These features are considered standard security practice these days.
1441 You should say Y here in almost all cases.
1443 config ARCH_HAS_STRICT_MODULE_RWX
1446 config STRICT_MODULE_RWX
1447 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX
1448 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES
1449 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1451 If this is set, module text and rodata memory will be made read-only,
1452 and non-text memory will be made non-executable. This provides
1453 protection against certain security exploits (e.g. writing to text)
1455 # select if the architecture provides an asm/dma-direct.h header
1456 config ARCH_HAS_PHYS_TO_DMA
1459 config HAVE_ARCH_COMPILER_H
1462 An architecture can select this if it provides an
1463 asm/compiler.h header that should be included after
1464 linux/compiler-*.h in order to override macro definitions that those
1465 headers generally provide.
1467 config HAVE_ARCH_PREL32_RELOCATIONS
1470 May be selected by an architecture if it supports place-relative
1471 32-bit relocations, both in the toolchain and in the module loader,
1472 in which case relative references can be used in special sections
1473 for PCI fixup, initcalls etc which are only half the size on 64 bit
1474 architectures, and don't require runtime relocation on relocatable
1477 config ARCH_USE_MEMREMAP_PROT
1480 config LOCK_EVENT_COUNTS
1481 bool "Locking event counts collection"
1484 Enable light-weight counting of various locking related events
1485 in the system with minimal performance impact. This reduces
1486 the chance of application behavior change because of timing
1487 differences. The counts are reported via debugfs.
1489 # Select if the architecture has support for applying RELR relocations.
1490 config ARCH_HAS_RELR
1494 bool "Use RELR relocation packing"
1495 depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR
1498 Store the kernel's dynamic relocations in the RELR relocation packing
1499 format. Requires a compatible linker (LLD supports this feature), as
1500 well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy
1503 config ARCH_HAS_MEM_ENCRYPT
1506 config ARCH_HAS_CC_PLATFORM
1509 config HAVE_SPARSE_SYSCALL_NR
1512 An architecture should select this if its syscall numbering is sparse
1513 to save space. For example, MIPS architecture has a syscall array with
1514 entries at 4000, 5000 and 6000 locations. This option turns on syscall
1515 related optimizations for a given architecture.
1517 config ARCH_HAS_VDSO_DATA
1520 config HAVE_STATIC_CALL
1523 config HAVE_STATIC_CALL_INLINE
1525 depends on HAVE_STATIC_CALL
1528 config HAVE_PREEMPT_DYNAMIC
1531 config HAVE_PREEMPT_DYNAMIC_CALL
1533 depends on HAVE_STATIC_CALL
1534 select HAVE_PREEMPT_DYNAMIC
1536 An architecture should select this if it can handle the preemption
1537 model being selected at boot time using static calls.
1539 Where an architecture selects HAVE_STATIC_CALL_INLINE, any call to a
1540 preemption function will be patched directly.
1542 Where an architecture does not select HAVE_STATIC_CALL_INLINE, any
1543 call to a preemption function will go through a trampoline, and the
1544 trampoline will be patched.
1546 It is strongly advised to support inline static call to avoid any
1549 config HAVE_PREEMPT_DYNAMIC_KEY
1551 depends on HAVE_ARCH_JUMP_LABEL
1552 select HAVE_PREEMPT_DYNAMIC
1554 An architecture should select this if it can handle the preemption
1555 model being selected at boot time using static keys.
1557 Each preemption function will be given an early return based on a
1558 static key. This should have slightly lower overhead than non-inline
1559 static calls, as this effectively inlines each trampoline into the
1560 start of its callee. This may avoid redundant work, and may
1561 integrate better with CFI schemes.
1563 This will have greater overhead than using inline static calls as
1564 the call to the preemption function cannot be entirely elided.
1566 config ARCH_WANT_LD_ORPHAN_WARN
1569 An arch should select this symbol once all linker sections are explicitly
1570 included, size-asserted, or discarded in the linker scripts. This is
1571 important because we never want expected sections to be placed heuristically
1572 by the linker, since the locations of such sections can change between linker
1575 config HAVE_ARCH_PFN_VALID
1578 config ARCH_SUPPORTS_DEBUG_PAGEALLOC
1581 config ARCH_SUPPORTS_PAGE_TABLE_CHECK
1584 config ARCH_SPLIT_ARG64
1587 If a 32-bit architecture requires 64-bit arguments to be split into
1588 pairs of 32-bit arguments, select this option.
1590 config ARCH_HAS_ELFCORE_COMPAT
1593 config ARCH_HAS_PARANOID_L1D_FLUSH
1596 config ARCH_HAVE_TRACE_MMIO_ACCESS
1599 config DYNAMIC_SIGFRAME
1602 # Select, if arch has a named attribute group bound to NUMA device nodes.
1603 config HAVE_ARCH_NODE_DEV_GROUP
1606 config ARCH_HAS_HW_PTE_YOUNG
1609 Architectures that select this option are capable of setting the
1610 accessed bit in PTE entries when using them as part of linear address
1611 translations. Architectures that require runtime check should select
1612 this option and override arch_has_hw_pte_young().
1614 config ARCH_HAS_NONLEAF_PMD_YOUNG
1617 Architectures that select this option are capable of setting the
1618 accessed bit in non-leaf PMD entries when using them as part of linear
1619 address translations. Page table walkers that clear the accessed bit
1620 may use this capability to reduce their search space.
1622 config ARCH_HAS_KERNEL_FPU_SUPPORT
1625 Architectures that select this option can run floating-point code in
1626 the kernel, as described in Documentation/core-api/floating-point.rst.
1628 source "kernel/gcov/Kconfig"
1630 source "scripts/gcc-plugins/Kconfig"
1632 config FUNCTION_ALIGNMENT_4B
1635 config FUNCTION_ALIGNMENT_8B
1638 config FUNCTION_ALIGNMENT_16B
1641 config FUNCTION_ALIGNMENT_32B
1644 config FUNCTION_ALIGNMENT_64B
1647 config FUNCTION_ALIGNMENT
1649 default 64 if FUNCTION_ALIGNMENT_64B
1650 default 32 if FUNCTION_ALIGNMENT_32B
1651 default 16 if FUNCTION_ALIGNMENT_16B
1652 default 8 if FUNCTION_ALIGNMENT_8B
1653 default 4 if FUNCTION_ALIGNMENT_4B
1656 config CC_HAS_MIN_FUNCTION_ALIGNMENT
1657 # Detect availability of the GCC option -fmin-function-alignment which
1658 # guarantees minimal alignment for all functions, unlike
1659 # -falign-functions which the compiler ignores for cold functions.
1660 def_bool $(cc-option, -fmin-function-alignment=8)
1662 config CC_HAS_SANE_FUNCTION_ALIGNMENT
1663 # Set if the guaranteed alignment with -fmin-function-alignment is
1664 # available or extra care is required in the kernel. Clang provides
1665 # strict alignment always, even with -falign-functions.
1666 def_bool CC_HAS_MIN_FUNCTION_ALIGNMENT || CC_IS_CLANG
1668 config ARCH_NEED_CMPXCHG_1_EMU