2 # General architecture dependent options
16 tristate "OProfile system profiling"
18 depends on HAVE_OPROFILE
20 select RING_BUFFER_ALLOW_SWAP
22 OProfile is a profiling system capable of profiling the
23 whole system, include the kernel, kernel modules, libraries,
28 config OPROFILE_EVENT_MULTIPLEX
29 bool "OProfile multiplexing support (EXPERIMENTAL)"
31 depends on OPROFILE && X86
33 The number of hardware counters is limited. The multiplexing
34 feature enables OProfile to gather more events than counters
35 are provided by the hardware. This is realized by switching
36 between events at a user specified time interval.
43 config OPROFILE_NMI_TIMER
45 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64
50 depends on HAVE_KPROBES
53 Kprobes allows you to trap at almost any kernel address and
54 execute a callback function. register_kprobe() establishes
55 a probepoint and specifies the callback. Kprobes is useful
56 for kernel debugging, non-intrusive instrumentation and testing.
60 bool "Optimize very unlikely/likely branches"
61 depends on HAVE_ARCH_JUMP_LABEL
63 This option enables a transparent branch optimization that
64 makes certain almost-always-true or almost-always-false branch
65 conditions even cheaper to execute within the kernel.
67 Certain performance-sensitive kernel code, such as trace points,
68 scheduler functionality, networking code and KVM have such
69 branches and include support for this optimization technique.
71 If it is detected that the compiler has support for "asm goto",
72 the kernel will compile such branches with just a nop
73 instruction. When the condition flag is toggled to true, the
74 nop will be converted to a jump instruction to execute the
75 conditional block of instructions.
77 This technique lowers overhead and stress on the branch prediction
78 of the processor and generally makes the kernel faster. The update
79 of the condition is slower, but those are always very rare.
81 ( On 32-bit x86, the necessary options added to the compiler
82 flags may increase the size of the kernel slightly. )
84 config STATIC_KEYS_SELFTEST
85 bool "Static key selftest"
88 Boot time self-test of the branch patching code.
92 depends on KPROBES && HAVE_OPTPROBES
95 config KPROBES_ON_FTRACE
97 depends on KPROBES && HAVE_KPROBES_ON_FTRACE
98 depends on DYNAMIC_FTRACE_WITH_REGS
100 If function tracer is enabled and the arch supports full
101 passing of pt_regs to function tracing, then kprobes can
102 optimize on top of function tracing.
106 depends on ARCH_SUPPORTS_UPROBES
108 Uprobes is the user-space counterpart to kprobes: they
109 enable instrumentation applications (such as 'perf probe')
110 to establish unintrusive probes in user-space binaries and
111 libraries, by executing handler functions when the probes
112 are hit by user-space applications.
114 ( These probes come in the form of single-byte breakpoints,
115 managed by the kernel and kept transparent to the probed
118 config HAVE_64BIT_ALIGNED_ACCESS
119 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS
121 Some architectures require 64 bit accesses to be 64 bit
122 aligned, which also requires structs containing 64 bit values
123 to be 64 bit aligned too. This includes some 32 bit
124 architectures which can do 64 bit accesses, as well as 64 bit
125 architectures without unaligned access.
127 This symbol should be selected by an architecture if 64 bit
128 accesses are required to be 64 bit aligned in this way even
129 though it is not a 64 bit architecture.
131 See Documentation/unaligned-memory-access.txt for more
132 information on the topic of unaligned memory accesses.
134 config HAVE_EFFICIENT_UNALIGNED_ACCESS
137 Some architectures are unable to perform unaligned accesses
138 without the use of get_unaligned/put_unaligned. Others are
139 unable to perform such accesses efficiently (e.g. trap on
140 unaligned access and require fixing it up in the exception
143 This symbol should be selected by an architecture if it can
144 perform unaligned accesses efficiently to allow different
145 code paths to be selected for these cases. Some network
146 drivers, for example, could opt to not fix up alignment
147 problems with received packets if doing so would not help
150 See Documentation/unaligned-memory-access.txt for more
151 information on the topic of unaligned memory accesses.
153 config ARCH_USE_BUILTIN_BSWAP
156 Modern versions of GCC (since 4.4) have builtin functions
157 for handling byte-swapping. Using these, instead of the old
158 inline assembler that the architecture code provides in the
159 __arch_bswapXX() macros, allows the compiler to see what's
160 happening and offers more opportunity for optimisation. In
161 particular, the compiler will be able to combine the byteswap
162 with a nearby load or store and use load-and-swap or
163 store-and-swap instructions if the architecture has them. It
164 should almost *never* result in code which is worse than the
165 hand-coded assembler in <asm/swab.h>. But just in case it
166 does, the use of the builtins is optional.
168 Any architecture with load-and-swap or store-and-swap
169 instructions should set this. And it shouldn't hurt to set it
170 on architectures that don't have such instructions.
174 depends on KPROBES && HAVE_KRETPROBES
176 config USER_RETURN_NOTIFIER
178 depends on HAVE_USER_RETURN_NOTIFIER
180 Provide a kernel-internal notification when a cpu is about to
183 config HAVE_IOREMAP_PROT
189 config HAVE_KRETPROBES
192 config HAVE_OPTPROBES
195 config HAVE_KPROBES_ON_FTRACE
202 # An arch should select this if it provides all these things:
204 # task_pt_regs() in asm/processor.h or asm/ptrace.h
205 # arch_has_single_step() if there is hardware single-step support
206 # arch_has_block_step() if there is hardware block-step support
207 # asm/syscall.h supplying asm-generic/syscall.h interface
208 # linux/regset.h user_regset interfaces
209 # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
210 # TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit}
211 # TIF_NOTIFY_RESUME calls tracehook_notify_resume()
212 # signal delivery calls tracehook_signal_handler()
214 config HAVE_ARCH_TRACEHOOK
217 config HAVE_DMA_CONTIGUOUS
220 config GENERIC_SMP_IDLE_THREAD
223 config GENERIC_IDLE_POLL_SETUP
226 config ARCH_HAS_FORTIFY_SOURCE
229 An architecture should select this when it can successfully
230 build and run with CONFIG_FORTIFY_SOURCE.
232 # Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h
233 config ARCH_HAS_SET_MEMORY
236 # Select if arch init_task initializer is different to init/init_task.c
237 config ARCH_INIT_TASK
240 # Select if arch has its private alloc_task_struct() function
241 config ARCH_TASK_STRUCT_ALLOCATOR
244 # Select if arch has its private alloc_thread_stack() function
245 config ARCH_THREAD_STACK_ALLOCATOR
248 # Select if arch wants to size task_struct dynamically via arch_task_struct_size:
249 config ARCH_WANTS_DYNAMIC_TASK_STRUCT
252 config HAVE_REGS_AND_STACK_ACCESS_API
255 This symbol should be selected by an architecure if it supports
256 the API needed to access registers and stack entries from pt_regs,
257 declared in asm/ptrace.h
258 For example the kprobes-based event tracer needs this API.
263 The <linux/clk.h> calls support software clock gating and
264 thus are a key power management tool on many systems.
266 config HAVE_DMA_API_DEBUG
269 config HAVE_HW_BREAKPOINT
271 depends on PERF_EVENTS
273 config HAVE_MIXED_BREAKPOINTS_REGS
275 depends on HAVE_HW_BREAKPOINT
277 Depending on the arch implementation of hardware breakpoints,
278 some of them have separate registers for data and instruction
279 breakpoints addresses, others have mixed registers to store
280 them but define the access type in a control register.
281 Select this option if your arch implements breakpoints under the
284 config HAVE_USER_RETURN_NOTIFIER
287 config HAVE_PERF_EVENTS_NMI
290 System hardware can generate an NMI using the perf event
291 subsystem. Also has support for calculating CPU cycle events
292 to determine how many clock cycles in a given period.
294 config HAVE_HARDLOCKUP_DETECTOR_PERF
296 depends on HAVE_PERF_EVENTS_NMI
298 The arch chooses to use the generic perf-NMI-based hardlockup
299 detector. Must define HAVE_PERF_EVENTS_NMI.
301 config HAVE_NMI_WATCHDOG
305 The arch provides a low level NMI watchdog. It provides
306 asm/nmi.h, and defines its own arch_touch_nmi_watchdog().
308 config HAVE_HARDLOCKUP_DETECTOR_ARCH
310 select HAVE_NMI_WATCHDOG
312 The arch chooses to provide its own hardlockup detector, which is
313 a superset of the HAVE_NMI_WATCHDOG. It also conforms to config
314 interfaces and parameters provided by hardlockup detector subsystem.
316 config HAVE_PERF_REGS
319 Support selective register dumps for perf events. This includes
320 bit-mapping of each registers and a unique architecture id.
322 config HAVE_PERF_USER_STACK_DUMP
325 Support user stack dumps for perf event samples. This needs
326 access to the user stack pointer which is not unified across
329 config HAVE_ARCH_JUMP_LABEL
332 config HAVE_RCU_TABLE_FREE
335 config ARCH_HAVE_NMI_SAFE_CMPXCHG
338 config HAVE_ALIGNED_STRUCT_PAGE
341 This makes sure that struct pages are double word aligned and that
342 e.g. the SLUB allocator can perform double word atomic operations
343 on a struct page for better performance. However selecting this
344 might increase the size of a struct page by a word.
346 config HAVE_CMPXCHG_LOCAL
349 config HAVE_CMPXCHG_DOUBLE
352 config ARCH_WEAK_RELEASE_ACQUIRE
355 config ARCH_WANT_IPC_PARSE_VERSION
358 config ARCH_WANT_COMPAT_IPC_PARSE_VERSION
361 config ARCH_WANT_OLD_COMPAT_IPC
362 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
365 config HAVE_ARCH_SECCOMP_FILTER
368 An arch should select this symbol if it provides all of these things:
370 - syscall_get_arguments()
372 - syscall_set_return_value()
373 - SIGSYS siginfo_t support
374 - secure_computing is called from a ptrace_event()-safe context
375 - secure_computing return value is checked and a return value of -1
376 results in the system call being skipped immediately.
377 - seccomp syscall wired up
379 config SECCOMP_FILTER
381 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
383 Enable tasks to build secure computing environments defined
384 in terms of Berkeley Packet Filter programs which implement
385 task-defined system call filtering polices.
387 See Documentation/prctl/seccomp_filter.txt for details.
389 config HAVE_GCC_PLUGINS
392 An arch should select this symbol if it supports building with
395 menuconfig GCC_PLUGINS
397 depends on HAVE_GCC_PLUGINS
398 depends on !COMPILE_TEST
400 GCC plugins are loadable modules that provide extra features to the
401 compiler. They are useful for runtime instrumentation and static analysis.
403 See Documentation/gcc-plugins.txt for details.
405 config GCC_PLUGIN_CYC_COMPLEXITY
406 bool "Compute the cyclomatic complexity of a function" if EXPERT
407 depends on GCC_PLUGINS
408 depends on !COMPILE_TEST
410 The complexity M of a function's control flow graph is defined as:
414 E = the number of edges
415 N = the number of nodes
416 P = the number of connected components (exit nodes).
418 Enabling this plugin reports the complexity to stderr during the
419 build. It mainly serves as a simple example of how to create a
420 gcc plugin for the kernel.
422 config GCC_PLUGIN_SANCOV
424 depends on GCC_PLUGINS
426 This plugin inserts a __sanitizer_cov_trace_pc() call at the start of
427 basic blocks. It supports all gcc versions with plugin support (from
428 gcc-4.5 on). It is based on the commit "Add fuzzing coverage support"
429 by Dmitry Vyukov <dvyukov@google.com>.
431 config GCC_PLUGIN_LATENT_ENTROPY
432 bool "Generate some entropy during boot and runtime"
433 depends on GCC_PLUGINS
435 By saying Y here the kernel will instrument some kernel code to
436 extract some entropy from both original and artificially created
437 program state. This will help especially embedded systems where
438 there is little 'natural' source of entropy normally. The cost
439 is some slowdown of the boot process (about 0.5%) and fork and
442 Note that entropy extracted this way is not cryptographically
445 This plugin was ported from grsecurity/PaX. More information at:
446 * https://grsecurity.net/
447 * https://pax.grsecurity.net/
449 config GCC_PLUGIN_STRUCTLEAK
450 bool "Force initialization of variables containing userspace addresses"
451 depends on GCC_PLUGINS
453 This plugin zero-initializes any structures containing a
454 __user attribute. This can prevent some classes of information
457 This plugin was ported from grsecurity/PaX. More information at:
458 * https://grsecurity.net/
459 * https://pax.grsecurity.net/
461 config GCC_PLUGIN_STRUCTLEAK_VERBOSE
462 bool "Report forcefully initialized variables"
463 depends on GCC_PLUGIN_STRUCTLEAK
464 depends on !COMPILE_TEST
466 This option will cause a warning to be printed each time the
467 structleak plugin finds a variable it thinks needs to be
468 initialized. Since not all existing initializers are detected
469 by the plugin, this can produce false positive warnings.
471 config GCC_PLUGIN_RANDSTRUCT
472 bool "Randomize layout of sensitive kernel structures"
473 depends on GCC_PLUGINS
474 select MODVERSIONS if MODULES
476 If you say Y here, the layouts of structures explicitly
477 marked by __randomize_layout will be randomized at
478 compile-time. This can introduce the requirement of an
479 additional information exposure vulnerability for exploits
480 targeting these structure types.
482 Enabling this feature will introduce some performance impact,
483 slightly increase memory usage, and prevent the use of forensic
484 tools like Volatility against the system (unless the kernel
485 source tree isn't cleaned after kernel installation).
487 The seed used for compilation is located at
488 scripts/gcc-plgins/randomize_layout_seed.h. It remains after
489 a make clean to allow for external modules to be compiled with
490 the existing seed and will be removed by a make mrproper or
493 Note that the implementation requires gcc 4.7 or newer.
495 This plugin was ported from grsecurity/PaX. More information at:
496 * https://grsecurity.net/
497 * https://pax.grsecurity.net/
499 config GCC_PLUGIN_RANDSTRUCT_PERFORMANCE
500 bool "Use cacheline-aware structure randomization"
501 depends on GCC_PLUGIN_RANDSTRUCT
502 depends on !COMPILE_TEST
504 If you say Y here, the RANDSTRUCT randomization will make a
505 best effort at restricting randomization to cacheline-sized
506 groups of elements. It will further not randomize bitfields
507 in structures. This reduces the performance hit of RANDSTRUCT
508 at the cost of weakened randomization.
510 config HAVE_CC_STACKPROTECTOR
513 An arch should select this symbol if:
514 - its compiler supports the -fstack-protector option
515 - it has implemented a stack canary (e.g. __stack_chk_guard)
517 config CC_STACKPROTECTOR
520 Set when a stack-protector mode is enabled, so that the build
521 can enable kernel-side support for the GCC feature.
524 prompt "Stack Protector buffer overflow detection"
525 depends on HAVE_CC_STACKPROTECTOR
526 default CC_STACKPROTECTOR_NONE
528 This option turns on the "stack-protector" GCC feature. This
529 feature puts, at the beginning of functions, a canary value on
530 the stack just before the return address, and validates
531 the value just before actually returning. Stack based buffer
532 overflows (that need to overwrite this return address) now also
533 overwrite the canary, which gets detected and the attack is then
534 neutralized via a kernel panic.
536 config CC_STACKPROTECTOR_NONE
539 Disable "stack-protector" GCC feature.
541 config CC_STACKPROTECTOR_REGULAR
543 select CC_STACKPROTECTOR
545 Functions will have the stack-protector canary logic added if they
546 have an 8-byte or larger character array on the stack.
548 This feature requires gcc version 4.2 or above, or a distribution
549 gcc with the feature backported ("-fstack-protector").
551 On an x86 "defconfig" build, this feature adds canary checks to
552 about 3% of all kernel functions, which increases kernel code size
555 config CC_STACKPROTECTOR_STRONG
557 select CC_STACKPROTECTOR
559 Functions will have the stack-protector canary logic added in any
560 of the following conditions:
562 - local variable's address used as part of the right hand side of an
563 assignment or function argument
564 - local variable is an array (or union containing an array),
565 regardless of array type or length
566 - uses register local variables
568 This feature requires gcc version 4.9 or above, or a distribution
569 gcc with the feature backported ("-fstack-protector-strong").
571 On an x86 "defconfig" build, this feature adds canary checks to
572 about 20% of all kernel functions, which increases the kernel code
580 Select this if the architecture wants to use thin archives
581 instead of ld -r to create the built-in.o files.
583 config LD_DEAD_CODE_DATA_ELIMINATION
586 Select this if the architecture wants to do dead code and
587 data elimination with the linker by compiling with
588 -ffunction-sections -fdata-sections and linking with
591 This requires that the arch annotates or otherwise protects
592 its external entry points from being discarded. Linker scripts
593 must also merge .text.*, .data.*, and .bss.* correctly into
594 output sections. Care must be taken not to pull in unrelated
595 sections (e.g., '.text.init'). Typically '.' in section names
596 is used to distinguish them from label names / C identifiers.
598 config HAVE_ARCH_WITHIN_STACK_FRAMES
601 An architecture should select this if it can walk the kernel stack
602 frames to determine if an object is part of either the arguments
603 or local variables (i.e. that it excludes saved return addresses,
604 and similar) by implementing an inline arch_within_stack_frames(),
605 which is used by CONFIG_HARDENED_USERCOPY.
607 config HAVE_CONTEXT_TRACKING
610 Provide kernel/user boundaries probes necessary for subsystems
611 that need it, such as userspace RCU extended quiescent state.
612 Syscalls need to be wrapped inside user_exit()-user_enter() through
613 the slow path using TIF_NOHZ flag. Exceptions handlers must be
614 wrapped as well. Irqs are already protected inside
615 rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on
616 irq exit still need to be protected.
618 config HAVE_VIRT_CPU_ACCOUNTING
621 config ARCH_HAS_SCALED_CPUTIME
624 config HAVE_VIRT_CPU_ACCOUNTING_GEN
628 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
629 Before enabling this option, arch code must be audited
630 to ensure there are no races in concurrent read/write of
631 cputime_t. For example, reading/writing 64-bit cputime_t on
632 some 32-bit arches may require multiple accesses, so proper
633 locking is needed to protect against concurrent accesses.
636 config HAVE_IRQ_TIME_ACCOUNTING
639 Archs need to ensure they use a high enough resolution clock to
640 support irq time accounting and then call enable_sched_clock_irqtime().
642 config HAVE_ARCH_TRANSPARENT_HUGEPAGE
645 config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
648 config HAVE_ARCH_HUGE_VMAP
651 config HAVE_ARCH_SOFT_DIRTY
654 config HAVE_MOD_ARCH_SPECIFIC
657 The arch uses struct mod_arch_specific to store data. Many arches
658 just need a simple module loader without arch specific data - those
659 should not enable this.
661 config MODULES_USE_ELF_RELA
664 Modules only use ELF RELA relocations. Modules with ELF REL
665 relocations will give an error.
667 config MODULES_USE_ELF_REL
670 Modules only use ELF REL relocations. Modules with ELF RELA
671 relocations will give an error.
673 config HAVE_UNDERSCORE_SYMBOL_PREFIX
676 Some architectures generate an _ in front of C symbols; things like
677 module loading and assembly files need to know about this.
679 config HAVE_IRQ_EXIT_ON_IRQ_STACK
682 Architecture doesn't only execute the irq handler on the irq stack
683 but also irq_exit(). This way we can process softirqs on this irq
684 stack instead of switching to a new one when we call __do_softirq()
685 in the end of an hardirq.
686 This spares a stack switch and improves cache usage on softirq
689 config PGTABLE_LEVELS
693 config ARCH_HAS_ELF_RANDOMIZE
696 An architecture supports choosing randomized locations for
697 stack, mmap, brk, and ET_DYN. Defined functions:
699 - arch_randomize_brk()
701 config HAVE_ARCH_MMAP_RND_BITS
704 An arch should select this symbol if it supports setting a variable
705 number of bits for use in establishing the base address for mmap
706 allocations, has MMU enabled and provides values for both:
707 - ARCH_MMAP_RND_BITS_MIN
708 - ARCH_MMAP_RND_BITS_MAX
710 config HAVE_EXIT_THREAD
713 An architecture implements exit_thread.
715 config ARCH_MMAP_RND_BITS_MIN
718 config ARCH_MMAP_RND_BITS_MAX
721 config ARCH_MMAP_RND_BITS_DEFAULT
724 config ARCH_MMAP_RND_BITS
725 int "Number of bits to use for ASLR of mmap base address" if EXPERT
726 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX
727 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT
728 default ARCH_MMAP_RND_BITS_MIN
729 depends on HAVE_ARCH_MMAP_RND_BITS
731 This value can be used to select the number of bits to use to
732 determine the random offset to the base address of vma regions
733 resulting from mmap allocations. This value will be bounded
734 by the architecture's minimum and maximum supported values.
736 This value can be changed after boot using the
737 /proc/sys/vm/mmap_rnd_bits tunable
739 config HAVE_ARCH_MMAP_RND_COMPAT_BITS
742 An arch should select this symbol if it supports running applications
743 in compatibility mode, supports setting a variable number of bits for
744 use in establishing the base address for mmap allocations, has MMU
745 enabled and provides values for both:
746 - ARCH_MMAP_RND_COMPAT_BITS_MIN
747 - ARCH_MMAP_RND_COMPAT_BITS_MAX
749 config ARCH_MMAP_RND_COMPAT_BITS_MIN
752 config ARCH_MMAP_RND_COMPAT_BITS_MAX
755 config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
758 config ARCH_MMAP_RND_COMPAT_BITS
759 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT
760 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX
761 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
762 default ARCH_MMAP_RND_COMPAT_BITS_MIN
763 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS
765 This value can be used to select the number of bits to use to
766 determine the random offset to the base address of vma regions
767 resulting from mmap allocations for compatible applications This
768 value will be bounded by the architecture's minimum and maximum
771 This value can be changed after boot using the
772 /proc/sys/vm/mmap_rnd_compat_bits tunable
774 config HAVE_ARCH_COMPAT_MMAP_BASES
777 This allows 64bit applications to invoke 32-bit mmap() syscall
778 and vice-versa 32-bit applications to call 64-bit mmap().
779 Required for applications doing different bitness syscalls.
781 config HAVE_COPY_THREAD_TLS
784 Architecture provides copy_thread_tls to accept tls argument via
785 normal C parameter passing, rather than extracting the syscall
786 argument from pt_regs.
788 config HAVE_STACK_VALIDATION
791 Architecture supports the 'objtool check' host tool command, which
792 performs compile-time stack metadata validation.
794 config HAVE_RELIABLE_STACKTRACE
797 Architecture has a save_stack_trace_tsk_reliable() function which
798 only returns a stack trace if it can guarantee the trace is reliable.
800 config HAVE_ARCH_HASH
804 If this is set, the architecture provides an <asm/hash.h>
805 file which provides platform-specific implementations of some
806 functions in <linux/hash.h> or fs/namei.c.
814 config CLONE_BACKWARDS
817 Architecture has tls passed as the 4th argument of clone(2),
820 config CLONE_BACKWARDS2
823 Architecture has the first two arguments of clone(2) swapped.
825 config CLONE_BACKWARDS3
828 Architecture has tls passed as the 3rd argument of clone(2),
831 config ODD_RT_SIGACTION
834 Architecture has unusual rt_sigaction(2) arguments
836 config OLD_SIGSUSPEND
839 Architecture has old sigsuspend(2) syscall, of one-argument variety
841 config OLD_SIGSUSPEND3
844 Even weirder antique ABI - three-argument sigsuspend(2)
849 Architecture has old sigaction(2) syscall. Nope, not the same
850 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
851 but fairly different variant of sigaction(2), thanks to OSF/1
854 config COMPAT_OLD_SIGACTION
857 config ARCH_NO_COHERENT_DMA_MMAP
860 config CPU_NO_EFFICIENT_FFS
863 config HAVE_ARCH_VMAP_STACK
866 An arch should select this symbol if it can support kernel stacks
867 in vmalloc space. This means:
869 - vmalloc space must be large enough to hold many kernel stacks.
870 This may rule out many 32-bit architectures.
872 - Stacks in vmalloc space need to work reliably. For example, if
873 vmap page tables are created on demand, either this mechanism
874 needs to work while the stack points to a virtual address with
875 unpopulated page tables or arch code (switch_to() and switch_mm(),
876 most likely) needs to ensure that the stack's page table entries
877 are populated before running on a possibly unpopulated stack.
879 - If the stack overflows into a guard page, something reasonable
880 should happen. The definition of "reasonable" is flexible, but
881 instantly rebooting without logging anything would be unfriendly.
885 bool "Use a virtually-mapped stack"
886 depends on HAVE_ARCH_VMAP_STACK && !KASAN
888 Enable this if you want the use virtually-mapped kernel stacks
889 with guard pages. This causes kernel stack overflows to be
890 caught immediately rather than causing difficult-to-diagnose
893 This is presently incompatible with KASAN because KASAN expects
894 the stack to map directly to the KASAN shadow map using a formula
895 that is incorrect if the stack is in vmalloc space.
897 config ARCH_OPTIONAL_KERNEL_RWX
900 config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
903 config ARCH_HAS_STRICT_KERNEL_RWX
906 config STRICT_KERNEL_RWX
907 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX
908 depends on ARCH_HAS_STRICT_KERNEL_RWX
909 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
911 If this is set, kernel text and rodata memory will be made read-only,
912 and non-text memory will be made non-executable. This provides
913 protection against certain security exploits (e.g. executing the heap
916 These features are considered standard security practice these days.
917 You should say Y here in almost all cases.
919 config ARCH_HAS_STRICT_MODULE_RWX
922 config STRICT_MODULE_RWX
923 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX
924 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES
925 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
927 If this is set, module text and rodata memory will be made read-only,
928 and non-text memory will be made non-executable. This provides
929 protection against certain security exploits (e.g. writing to text)
931 config ARCH_WANT_RELAX_ORDER
935 bool "Perform full reference count validation at the expense of speed"
937 Enabling this switches the refcounting infrastructure from a fast
938 unchecked atomic_t implementation to a fully state checked
939 implementation, which can be (slightly) slower but provides protections
940 against various use-after-free conditions that can be used in
941 security flaw exploits.
943 source "kernel/gcov/Kconfig"