linux-6.8/arch/Kconfig

1 # SPDX-License-Identifier: GPL-2.0
12 menu "General architecture-dependent options"
17 	  Select if the architecture can check permissions at sub-page
63 	  for kernel debugging, non-intrusive instrumentation and testing.
72 	  makes certain almost-always-true or almost-always-false branch
75 	  Certain performance-sensitive kernel code, such as trace points,
89 	  ( On 32-bit x86, the necessary options added to the compiler
96 	  Boot time self-test of the branch patching code.
102 	  Boot time self-test of the call patching code.
122 	  Uprobes is the user-space counterpart to kprobes: they
124 	  to establish unintrusive probes in user-space binaries and
126 	  are hit by user-space applications.
128 	  ( These probes come in the form of single-byte breakpoints,
139 	  architectures without unaligned access.
145 	  See Documentation/core-api/unaligned-memory-access.rst for
146 	  more information on the topic of unaligned memory accesses.
151 	  Some architectures are unable to perform unaligned accesses
154 	  unaligned access and require fixing it up in the exception
158 	  perform unaligned accesses efficiently to allow different
164 	  See Documentation/core-api/unaligned-memory-access.rst for more
165 	  information on the topic of unaligned memory accesses.
171 	  for handling byte-swapping. Using these, instead of the old
176 	  with a nearby load or store and use load-and-swap or
177 	  store-and-swap instructions if the architecture has them. It
179 	  hand-coded assembler in <asm/swab.h>.  But just in case it
182 	  Any architecture with load-and-swap or store-and-swap
200 	  Provide a kernel-internal notification when a cpu is about to
245 #	arch_has_single_step()	if there is hardware single-step support
246 #	arch_has_block_step()	if there is hardware block-step support
247 #	asm/syscall.h		supplying asm-generic/syscall.h interface
296 # to undo an in-place page table remap for uncached access.
304 # The architecture has a per-task state that includes the mm's PASID
334 	  All new 32-bit architectures should have 64-bit off_t type on
337 	  still support 32-bit off_t. This option is enabled for all such
348 	  <asm/asm-prototypes.h> to support the module versioning for symbols
355 	  the API needed to access registers and stack entries from pt_regs,
357 	  For example the kprobes-based event tracer needs this API.
376 	  the API needed to access function arguments from pt_regs,
390 	  them but define the access type in a control register.
408 	  The arch chooses to use the generic perf-NMI-based hardlockup
424 	  bit-mapping of each registers and a unique architecture id.
430 	  access to the user stack pointer which is not unified across
473 # multi-threaded application), by reducing contention on the mm refcount.
475 # This can be disabled if the architecture ensures no CPUs are using an mm as a
482 # the lazy tlb reference of a kthread's ->active_mm (non-arch code has been
488 # This option allows MMU_LAZY_TLB_REFCOUNT=n. It ensures no CPUs are using an
496 # - At the time of the final mmdrop of the mm, ensure mm_cpumask(mm) contains
498 # - It must meet the requirements for MMU_LAZY_TLB_REFCOUNT=n (see above).
540 	  and compat syscalls if the asm-generic/seccomp.h defaults need adjustment:
541 	  - __NR_seccomp_read_32
542 	  - __NR_seccomp_write_32
543 	  - __NR_seccomp_exit_32
544 	  - __NR_seccomp_sigreturn_32
551 	  - all the requirements for HAVE_ARCH_SECCOMP
552 	  - syscall_get_arch()
553 	  - syscall_get_arguments()
554 	  - syscall_rollback()
555 	  - syscall_set_return_value()
556 	  - SIGSYS siginfo_t support
557 	  - secure_computing is called from a ptrace_event()-safe context
558 	  - secure_computing return value is checked and a return value of -1
560 	  - seccomp syscall wired up
561 	  - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE,
588 	  task-defined system call filtering polices.
590 	  See Documentation/userspace-api/seccomp_filter.rst for details.
617 	  - it has implemented a stack canary (e.g. __stack_chk_guard)
622 	depends on $(cc-option,-fstack-protector)
625 	  This option turns on the "stack-protector" GCC feature. This
633 	  Functions will have the stack-protector canary logic added if they
634 	  have an 8-byte or larger character array on the stack.
637 	  gcc with the feature backported ("-fstack-protector").
646 	depends on $(cc-option,-fstack-protector-strong)
649 	  Functions will have the stack-protector canary logic added in any
652 	  - local variable's address used as part of the right hand side of an
654 	  - local variable is an array (or union containing an array),
656 	  - uses register local variables
659 	  gcc with the feature backported ("-fstack-protector-strong").
683 	  - Clang: https://clang.llvm.org/docs/ShadowCallStack.html
684 	  - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options
714 	  - compiling with Clang,
715 	  - compiling inline assembly with Clang's integrated assembler,
716 	  - and linking with LLD.
727 	depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm)
728 	depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm)
746 	  If unsure, select LTO_NONE. Note that LTO is very resource-intensive
792 	  Control-Flow Integrity (CFI) checking.
800 	depends on $(cc-option,-fsanitize=kcfi)
802 	  This option enables Clang’s forward-edge Control Flow Integrity
836 	  Syscalls need to be wrapped inside user_exit()-user_enter(), either
852 	  - Critical entry code isn't preemptible (or better yet:
854 	  - No use of RCU read side critical sections, unless ct_nmi_enter()
856 	  - No use of instrumentation, unless instrumentation_begin() got
881 	  With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
883 	  to ensure there are no races in concurrent read/write of
884 	  cputime_t. For example, reading/writing 64-bit cputime_t on
885 	  some 32-bit arches may require multiple accesses, so proper
916 #  Archs that select this would be capable of PMD-sized vmaps (i.e.,
928 # if there are no userspace memory management features that use it
942 	  just need a simple module loader without arch specific data - those
998 	  - arch_mmap_rnd()
999 	  - arch_randomize_brk()
1007 	  - ARCH_MMAP_RND_BITS_MIN
1008 	  - ARCH_MMAP_RND_BITS_MAX
1046 	  - ARCH_MMAP_RND_COMPAT_BITS_MIN
1047 	  - ARCH_MMAP_RND_COMPAT_BITS_MAX
1077 	  This allows 64bit applications to invoke 32-bit mmap() syscall
1078 	  and vice-versa 32-bit applications to call 64-bit mmap().
1093 # address by giving priority to top-down scheme only if the process
1097 # - STACK_RND_MASK
1122 	  Architecture supports objtool compile-time frame pointer rule
1137 	  file which provides platform-specific implementations of some
1174 	  Architecture has old sigsuspend(2) syscall, of one-argument variety
1179 	  Even weirder antique ABI - three-argument sigsuspend(2)
1185 	  as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
1193 	bool "Provide system calls for 32-bit time_t"
1197 	  This is relevant on all 32-bit architectures, and 64-bit architectures
1215 	  - vmalloc space must be large enough to hold many kernel stacks.
1216 	    This may rule out many 32-bit architectures.
1218 	  - Stacks in vmalloc space need to work reliably.  For example, if
1225 	  - If the stack overflows into a guard page, something reasonable
1231 	bool "Use a virtually-mapped stack"
1235 	  Enable this if you want the use virtually-mapped kernel stacks
1237 	  caught immediately rather than causing difficult-to-diagnose
1250 	  syscall exit. Careful removal of -fstack-protector-strong and
1251 	  -fstack-protector should also be applied to the entry code and
1265 	  cross-syscall address exposures.
1291 	bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX
1295 	  If this is set, kernel text and rodata memory will be made read-only,
1296 	  and non-text memory will be made non-executable. This provides
1311 	  If this is set, module text and rodata memory will be made read-only,
1312 	  and non-text memory will be made non-executable. This provides
1315 # select if the architecture provides an asm/dma-direct.h header
1324 	  linux/compiler-*.h in order to override macro definitions that those
1330 	  May be selected by an architecture if it supports place-relative
1331 	  32-bit relocations, both in the toolchain and in the module loader,
1344 	  Enable light-weight counting of various locking related events
1360 	  well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy
1418 	  static key. This should have slightly lower overhead than non-inline
1430 	  included, size-asserted, or discarded in the linker scripts. This is
1447 	  If a 32-bit architecture requires 64-bit arguments to be split into
1448 	  pairs of 32-bit arguments, select this option.
1478 	  accessed bit in non-leaf PMD entries when using them as part of linear
1484 source "scripts/gcc-plugins/Kconfig"