1 # SPDX-License-Identifier: GPL-2.0
3 # General architecture dependent options
20 tristate "OProfile system profiling"
22 depends on HAVE_OPROFILE
24 select RING_BUFFER_ALLOW_SWAP
26 OProfile is a profiling system capable of profiling the
27 whole system, include the kernel, kernel modules, libraries,
32 config OPROFILE_EVENT_MULTIPLEX
33 bool "OProfile multiplexing support (EXPERIMENTAL)"
35 depends on OPROFILE && X86
37 The number of hardware counters is limited. The multiplexing
38 feature enables OProfile to gather more events than counters
39 are provided by the hardware. This is realized by switching
40 between events at a user specified time interval.
47 config OPROFILE_NMI_TIMER
49 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64
54 depends on HAVE_KPROBES
57 Kprobes allows you to trap at almost any kernel address and
58 execute a callback function. register_kprobe() establishes
59 a probepoint and specifies the callback. Kprobes is useful
60 for kernel debugging, non-intrusive instrumentation and testing.
64 bool "Optimize very unlikely/likely branches"
65 depends on HAVE_ARCH_JUMP_LABEL
67 This option enables a transparent branch optimization that
68 makes certain almost-always-true or almost-always-false branch
69 conditions even cheaper to execute within the kernel.
71 Certain performance-sensitive kernel code, such as trace points,
72 scheduler functionality, networking code and KVM have such
73 branches and include support for this optimization technique.
75 If it is detected that the compiler has support for "asm goto",
76 the kernel will compile such branches with just a nop
77 instruction. When the condition flag is toggled to true, the
78 nop will be converted to a jump instruction to execute the
79 conditional block of instructions.
81 This technique lowers overhead and stress on the branch prediction
82 of the processor and generally makes the kernel faster. The update
83 of the condition is slower, but those are always very rare.
85 ( On 32-bit x86, the necessary options added to the compiler
86 flags may increase the size of the kernel slightly. )
88 config STATIC_KEYS_SELFTEST
89 bool "Static key selftest"
92 Boot time self-test of the branch patching code.
96 depends on KPROBES && HAVE_OPTPROBES
97 select TASKS_RCU if PREEMPT
99 config KPROBES_ON_FTRACE
101 depends on KPROBES && HAVE_KPROBES_ON_FTRACE
102 depends on DYNAMIC_FTRACE_WITH_REGS
104 If function tracer is enabled and the arch supports full
105 passing of pt_regs to function tracing, then kprobes can
106 optimize on top of function tracing.
110 depends on ARCH_SUPPORTS_UPROBES
112 Uprobes is the user-space counterpart to kprobes: they
113 enable instrumentation applications (such as 'perf probe')
114 to establish unintrusive probes in user-space binaries and
115 libraries, by executing handler functions when the probes
116 are hit by user-space applications.
118 ( These probes come in the form of single-byte breakpoints,
119 managed by the kernel and kept transparent to the probed
122 config HAVE_64BIT_ALIGNED_ACCESS
123 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS
125 Some architectures require 64 bit accesses to be 64 bit
126 aligned, which also requires structs containing 64 bit values
127 to be 64 bit aligned too. This includes some 32 bit
128 architectures which can do 64 bit accesses, as well as 64 bit
129 architectures without unaligned access.
131 This symbol should be selected by an architecture if 64 bit
132 accesses are required to be 64 bit aligned in this way even
133 though it is not a 64 bit architecture.
135 See Documentation/unaligned-memory-access.txt for more
136 information on the topic of unaligned memory accesses.
138 config HAVE_EFFICIENT_UNALIGNED_ACCESS
141 Some architectures are unable to perform unaligned accesses
142 without the use of get_unaligned/put_unaligned. Others are
143 unable to perform such accesses efficiently (e.g. trap on
144 unaligned access and require fixing it up in the exception
147 This symbol should be selected by an architecture if it can
148 perform unaligned accesses efficiently to allow different
149 code paths to be selected for these cases. Some network
150 drivers, for example, could opt to not fix up alignment
151 problems with received packets if doing so would not help
154 See Documentation/unaligned-memory-access.txt for more
155 information on the topic of unaligned memory accesses.
157 config ARCH_USE_BUILTIN_BSWAP
160 Modern versions of GCC (since 4.4) have builtin functions
161 for handling byte-swapping. Using these, instead of the old
162 inline assembler that the architecture code provides in the
163 __arch_bswapXX() macros, allows the compiler to see what's
164 happening and offers more opportunity for optimisation. In
165 particular, the compiler will be able to combine the byteswap
166 with a nearby load or store and use load-and-swap or
167 store-and-swap instructions if the architecture has them. It
168 should almost *never* result in code which is worse than the
169 hand-coded assembler in <asm/swab.h>. But just in case it
170 does, the use of the builtins is optional.
172 Any architecture with load-and-swap or store-and-swap
173 instructions should set this. And it shouldn't hurt to set it
174 on architectures that don't have such instructions.
178 depends on KPROBES && HAVE_KRETPROBES
180 config USER_RETURN_NOTIFIER
182 depends on HAVE_USER_RETURN_NOTIFIER
184 Provide a kernel-internal notification when a cpu is about to
187 config HAVE_IOREMAP_PROT
193 config HAVE_KRETPROBES
196 config HAVE_OPTPROBES
199 config HAVE_KPROBES_ON_FTRACE
206 # An arch should select this if it provides all these things:
208 # task_pt_regs() in asm/processor.h or asm/ptrace.h
209 # arch_has_single_step() if there is hardware single-step support
210 # arch_has_block_step() if there is hardware block-step support
211 # asm/syscall.h supplying asm-generic/syscall.h interface
212 # linux/regset.h user_regset interfaces
213 # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
214 # TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit}
215 # TIF_NOTIFY_RESUME calls tracehook_notify_resume()
216 # signal delivery calls tracehook_signal_handler()
218 config HAVE_ARCH_TRACEHOOK
221 config HAVE_DMA_CONTIGUOUS
224 config GENERIC_SMP_IDLE_THREAD
227 config GENERIC_IDLE_POLL_SETUP
230 config ARCH_HAS_FORTIFY_SOURCE
233 An architecture should select this when it can successfully
234 build and run with CONFIG_FORTIFY_SOURCE.
236 # Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h
237 config ARCH_HAS_SET_MEMORY
240 # Select if arch init_task initializer is different to init/init_task.c
241 config ARCH_INIT_TASK
244 # Select if arch has its private alloc_task_struct() function
245 config ARCH_TASK_STRUCT_ALLOCATOR
248 # Select if arch has its private alloc_thread_stack() function
249 config ARCH_THREAD_STACK_ALLOCATOR
252 # Select if arch wants to size task_struct dynamically via arch_task_struct_size:
253 config ARCH_WANTS_DYNAMIC_TASK_STRUCT
256 config HAVE_REGS_AND_STACK_ACCESS_API
259 This symbol should be selected by an architecure if it supports
260 the API needed to access registers and stack entries from pt_regs,
261 declared in asm/ptrace.h
262 For example the kprobes-based event tracer needs this API.
267 The <linux/clk.h> calls support software clock gating and
268 thus are a key power management tool on many systems.
270 config HAVE_DMA_API_DEBUG
273 config HAVE_HW_BREAKPOINT
275 depends on PERF_EVENTS
277 config HAVE_MIXED_BREAKPOINTS_REGS
279 depends on HAVE_HW_BREAKPOINT
281 Depending on the arch implementation of hardware breakpoints,
282 some of them have separate registers for data and instruction
283 breakpoints addresses, others have mixed registers to store
284 them but define the access type in a control register.
285 Select this option if your arch implements breakpoints under the
288 config HAVE_USER_RETURN_NOTIFIER
291 config HAVE_PERF_EVENTS_NMI
294 System hardware can generate an NMI using the perf event
295 subsystem. Also has support for calculating CPU cycle events
296 to determine how many clock cycles in a given period.
298 config HAVE_HARDLOCKUP_DETECTOR_PERF
300 depends on HAVE_PERF_EVENTS_NMI
302 The arch chooses to use the generic perf-NMI-based hardlockup
303 detector. Must define HAVE_PERF_EVENTS_NMI.
305 config HAVE_NMI_WATCHDOG
309 The arch provides a low level NMI watchdog. It provides
310 asm/nmi.h, and defines its own arch_touch_nmi_watchdog().
312 config HAVE_HARDLOCKUP_DETECTOR_ARCH
314 select HAVE_NMI_WATCHDOG
316 The arch chooses to provide its own hardlockup detector, which is
317 a superset of the HAVE_NMI_WATCHDOG. It also conforms to config
318 interfaces and parameters provided by hardlockup detector subsystem.
320 config HAVE_PERF_REGS
323 Support selective register dumps for perf events. This includes
324 bit-mapping of each registers and a unique architecture id.
326 config HAVE_PERF_USER_STACK_DUMP
329 Support user stack dumps for perf event samples. This needs
330 access to the user stack pointer which is not unified across
333 config HAVE_ARCH_JUMP_LABEL
336 config HAVE_RCU_TABLE_FREE
339 config ARCH_HAVE_NMI_SAFE_CMPXCHG
342 config HAVE_ALIGNED_STRUCT_PAGE
345 This makes sure that struct pages are double word aligned and that
346 e.g. the SLUB allocator can perform double word atomic operations
347 on a struct page for better performance. However selecting this
348 might increase the size of a struct page by a word.
350 config HAVE_CMPXCHG_LOCAL
353 config HAVE_CMPXCHG_DOUBLE
356 config ARCH_WEAK_RELEASE_ACQUIRE
359 config ARCH_WANT_IPC_PARSE_VERSION
362 config ARCH_WANT_COMPAT_IPC_PARSE_VERSION
365 config ARCH_WANT_OLD_COMPAT_IPC
366 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
369 config HAVE_ARCH_SECCOMP_FILTER
372 An arch should select this symbol if it provides all of these things:
374 - syscall_get_arguments()
376 - syscall_set_return_value()
377 - SIGSYS siginfo_t support
378 - secure_computing is called from a ptrace_event()-safe context
379 - secure_computing return value is checked and a return value of -1
380 results in the system call being skipped immediately.
381 - seccomp syscall wired up
383 config SECCOMP_FILTER
385 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
387 Enable tasks to build secure computing environments defined
388 in terms of Berkeley Packet Filter programs which implement
389 task-defined system call filtering polices.
391 See Documentation/prctl/seccomp_filter.txt for details.
393 config HAVE_GCC_PLUGINS
396 An arch should select this symbol if it supports building with
399 menuconfig GCC_PLUGINS
401 depends on HAVE_GCC_PLUGINS
402 depends on !COMPILE_TEST
404 GCC plugins are loadable modules that provide extra features to the
405 compiler. They are useful for runtime instrumentation and static analysis.
407 See Documentation/gcc-plugins.txt for details.
409 config GCC_PLUGIN_CYC_COMPLEXITY
410 bool "Compute the cyclomatic complexity of a function" if EXPERT
411 depends on GCC_PLUGINS
412 depends on !COMPILE_TEST
414 The complexity M of a function's control flow graph is defined as:
418 E = the number of edges
419 N = the number of nodes
420 P = the number of connected components (exit nodes).
422 Enabling this plugin reports the complexity to stderr during the
423 build. It mainly serves as a simple example of how to create a
424 gcc plugin for the kernel.
426 config GCC_PLUGIN_SANCOV
428 depends on GCC_PLUGINS
430 This plugin inserts a __sanitizer_cov_trace_pc() call at the start of
431 basic blocks. It supports all gcc versions with plugin support (from
432 gcc-4.5 on). It is based on the commit "Add fuzzing coverage support"
433 by Dmitry Vyukov <dvyukov@google.com>.
435 config GCC_PLUGIN_LATENT_ENTROPY
436 bool "Generate some entropy during boot and runtime"
437 depends on GCC_PLUGINS
439 By saying Y here the kernel will instrument some kernel code to
440 extract some entropy from both original and artificially created
441 program state. This will help especially embedded systems where
442 there is little 'natural' source of entropy normally. The cost
443 is some slowdown of the boot process (about 0.5%) and fork and
446 Note that entropy extracted this way is not cryptographically
449 This plugin was ported from grsecurity/PaX. More information at:
450 * https://grsecurity.net/
451 * https://pax.grsecurity.net/
453 config GCC_PLUGIN_STRUCTLEAK
454 bool "Force initialization of variables containing userspace addresses"
455 depends on GCC_PLUGINS
457 This plugin zero-initializes any structures containing a
458 __user attribute. This can prevent some classes of information
461 This plugin was ported from grsecurity/PaX. More information at:
462 * https://grsecurity.net/
463 * https://pax.grsecurity.net/
465 config GCC_PLUGIN_STRUCTLEAK_BYREF_ALL
466 bool "Force initialize all struct type variables passed by reference"
467 depends on GCC_PLUGIN_STRUCTLEAK
469 Zero initialize any struct type local variable that may be passed by
470 reference without having been initialized.
472 config GCC_PLUGIN_STRUCTLEAK_VERBOSE
473 bool "Report forcefully initialized variables"
474 depends on GCC_PLUGIN_STRUCTLEAK
475 depends on !COMPILE_TEST
477 This option will cause a warning to be printed each time the
478 structleak plugin finds a variable it thinks needs to be
479 initialized. Since not all existing initializers are detected
480 by the plugin, this can produce false positive warnings.
482 config GCC_PLUGIN_RANDSTRUCT
483 bool "Randomize layout of sensitive kernel structures"
484 depends on GCC_PLUGINS
485 select MODVERSIONS if MODULES
487 If you say Y here, the layouts of structures that are entirely
488 function pointers (and have not been manually annotated with
489 __no_randomize_layout), or structures that have been explicitly
490 marked with __randomize_layout, will be randomized at compile-time.
491 This can introduce the requirement of an additional information
492 exposure vulnerability for exploits targeting these structure
495 Enabling this feature will introduce some performance impact,
496 slightly increase memory usage, and prevent the use of forensic
497 tools like Volatility against the system (unless the kernel
498 source tree isn't cleaned after kernel installation).
500 The seed used for compilation is located at
501 scripts/gcc-plgins/randomize_layout_seed.h. It remains after
502 a make clean to allow for external modules to be compiled with
503 the existing seed and will be removed by a make mrproper or
506 Note that the implementation requires gcc 4.7 or newer.
508 This plugin was ported from grsecurity/PaX. More information at:
509 * https://grsecurity.net/
510 * https://pax.grsecurity.net/
512 config GCC_PLUGIN_RANDSTRUCT_PERFORMANCE
513 bool "Use cacheline-aware structure randomization"
514 depends on GCC_PLUGIN_RANDSTRUCT
515 depends on !COMPILE_TEST
517 If you say Y here, the RANDSTRUCT randomization will make a
518 best effort at restricting randomization to cacheline-sized
519 groups of elements. It will further not randomize bitfields
520 in structures. This reduces the performance hit of RANDSTRUCT
521 at the cost of weakened randomization.
523 config HAVE_CC_STACKPROTECTOR
526 An arch should select this symbol if:
527 - its compiler supports the -fstack-protector option
528 - it has implemented a stack canary (e.g. __stack_chk_guard)
530 config CC_STACKPROTECTOR
533 Set when a stack-protector mode is enabled, so that the build
534 can enable kernel-side support for the GCC feature.
537 prompt "Stack Protector buffer overflow detection"
538 depends on HAVE_CC_STACKPROTECTOR
539 default CC_STACKPROTECTOR_NONE
541 This option turns on the "stack-protector" GCC feature. This
542 feature puts, at the beginning of functions, a canary value on
543 the stack just before the return address, and validates
544 the value just before actually returning. Stack based buffer
545 overflows (that need to overwrite this return address) now also
546 overwrite the canary, which gets detected and the attack is then
547 neutralized via a kernel panic.
549 config CC_STACKPROTECTOR_NONE
552 Disable "stack-protector" GCC feature.
554 config CC_STACKPROTECTOR_REGULAR
556 select CC_STACKPROTECTOR
558 Functions will have the stack-protector canary logic added if they
559 have an 8-byte or larger character array on the stack.
561 This feature requires gcc version 4.2 or above, or a distribution
562 gcc with the feature backported ("-fstack-protector").
564 On an x86 "defconfig" build, this feature adds canary checks to
565 about 3% of all kernel functions, which increases kernel code size
568 config CC_STACKPROTECTOR_STRONG
570 select CC_STACKPROTECTOR
572 Functions will have the stack-protector canary logic added in any
573 of the following conditions:
575 - local variable's address used as part of the right hand side of an
576 assignment or function argument
577 - local variable is an array (or union containing an array),
578 regardless of array type or length
579 - uses register local variables
581 This feature requires gcc version 4.9 or above, or a distribution
582 gcc with the feature backported ("-fstack-protector-strong").
584 On an x86 "defconfig" build, this feature adds canary checks to
585 about 20% of all kernel functions, which increases the kernel code
593 Select this if the architecture wants to use thin archives
594 instead of ld -r to create the built-in.o files.
596 config LD_DEAD_CODE_DATA_ELIMINATION
599 Select this if the architecture wants to do dead code and
600 data elimination with the linker by compiling with
601 -ffunction-sections -fdata-sections and linking with
604 This requires that the arch annotates or otherwise protects
605 its external entry points from being discarded. Linker scripts
606 must also merge .text.*, .data.*, and .bss.* correctly into
607 output sections. Care must be taken not to pull in unrelated
608 sections (e.g., '.text.init'). Typically '.' in section names
609 is used to distinguish them from label names / C identifiers.
614 config ARCH_SUPPORTS_LTO_CLANG
617 An architecture should select this option it supports:
618 - compiling with clang,
619 - compiling inline assembly with clang's integrated assembler,
620 - and linking with either lld or GNU gold w/ LLVMgold.
623 prompt "Link-Time Optimization (LTO) (EXPERIMENTAL)"
626 This option turns on Link-Time Optimization (LTO).
632 bool "Use clang Link Time Optimization (LTO) (EXPERIMENTAL)"
633 depends on ARCH_SUPPORTS_LTO_CLANG
634 depends on !FTRACE_MCOUNT_RECORD || HAVE_C_RECORDMCOUNT
637 select LD_DEAD_CODE_DATA_ELIMINATION
639 This option enables clang's Link Time Optimization (LTO), which allows
640 the compiler to optimize the kernel globally at link time. If you
641 enable this option, the compiler generates LLVM IR instead of object
642 files, and the actual compilation from IR occurs at the LTO link step,
643 which may take several minutes.
645 If you select this option, you must compile the kernel with clang >=
646 5.0 (make CC=clang) and GNU gold from binutils >= 2.27, and have the
647 LLVMgold plug-in in LD_LIBRARY_PATH.
654 config CFI_PERMISSIVE
655 bool "Use CFI in permissive mode"
658 When selected, Control Flow Integrity (CFI) violations result in a
659 warning instead of a kernel panic. This option is useful for finding
660 CFI violations in drivers during development.
663 bool "Use clang Control Flow Integrity (CFI) (EXPERIMENTAL)"
668 This option enables clang Control Flow Integrity (CFI), which adds
669 runtime checking for indirect function calls.
671 config CFI_CLANG_SHADOW
672 bool "Use CFI shadow to speed up cross-module checks"
676 If you select this option, the kernel builds a fast look-up table of
677 CFI check functions in loaded modules to reduce overhead.
679 config HAVE_ARCH_WITHIN_STACK_FRAMES
682 An architecture should select this if it can walk the kernel stack
683 frames to determine if an object is part of either the arguments
684 or local variables (i.e. that it excludes saved return addresses,
685 and similar) by implementing an inline arch_within_stack_frames(),
686 which is used by CONFIG_HARDENED_USERCOPY.
688 config HAVE_CONTEXT_TRACKING
691 Provide kernel/user boundaries probes necessary for subsystems
692 that need it, such as userspace RCU extended quiescent state.
693 Syscalls need to be wrapped inside user_exit()-user_enter() through
694 the slow path using TIF_NOHZ flag. Exceptions handlers must be
695 wrapped as well. Irqs are already protected inside
696 rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on
697 irq exit still need to be protected.
699 config HAVE_VIRT_CPU_ACCOUNTING
702 config ARCH_HAS_SCALED_CPUTIME
705 config HAVE_VIRT_CPU_ACCOUNTING_GEN
709 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
710 Before enabling this option, arch code must be audited
711 to ensure there are no races in concurrent read/write of
712 cputime_t. For example, reading/writing 64-bit cputime_t on
713 some 32-bit arches may require multiple accesses, so proper
714 locking is needed to protect against concurrent accesses.
717 config HAVE_IRQ_TIME_ACCOUNTING
720 Archs need to ensure they use a high enough resolution clock to
721 support irq time accounting and then call enable_sched_clock_irqtime().
723 config HAVE_ARCH_TRANSPARENT_HUGEPAGE
726 config HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
729 config HAVE_ARCH_HUGE_VMAP
732 config HAVE_ARCH_SOFT_DIRTY
735 config HAVE_MOD_ARCH_SPECIFIC
738 The arch uses struct mod_arch_specific to store data. Many arches
739 just need a simple module loader without arch specific data - those
740 should not enable this.
742 config MODULES_USE_ELF_RELA
745 Modules only use ELF RELA relocations. Modules with ELF REL
746 relocations will give an error.
748 config MODULES_USE_ELF_REL
751 Modules only use ELF REL relocations. Modules with ELF RELA
752 relocations will give an error.
754 config HAVE_UNDERSCORE_SYMBOL_PREFIX
757 Some architectures generate an _ in front of C symbols; things like
758 module loading and assembly files need to know about this.
760 config HAVE_IRQ_EXIT_ON_IRQ_STACK
763 Architecture doesn't only execute the irq handler on the irq stack
764 but also irq_exit(). This way we can process softirqs on this irq
765 stack instead of switching to a new one when we call __do_softirq()
766 in the end of an hardirq.
767 This spares a stack switch and improves cache usage on softirq
770 config PGTABLE_LEVELS
774 config ARCH_HAS_ELF_RANDOMIZE
777 An architecture supports choosing randomized locations for
778 stack, mmap, brk, and ET_DYN. Defined functions:
780 - arch_randomize_brk()
782 config HAVE_ARCH_MMAP_RND_BITS
785 An arch should select this symbol if it supports setting a variable
786 number of bits for use in establishing the base address for mmap
787 allocations, has MMU enabled and provides values for both:
788 - ARCH_MMAP_RND_BITS_MIN
789 - ARCH_MMAP_RND_BITS_MAX
791 config HAVE_EXIT_THREAD
794 An architecture implements exit_thread.
796 config ARCH_MMAP_RND_BITS_MIN
799 config ARCH_MMAP_RND_BITS_MAX
802 config ARCH_MMAP_RND_BITS_DEFAULT
805 config ARCH_MMAP_RND_BITS
806 int "Number of bits to use for ASLR of mmap base address" if EXPERT
807 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX
808 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT
809 default ARCH_MMAP_RND_BITS_MIN
810 depends on HAVE_ARCH_MMAP_RND_BITS
812 This value can be used to select the number of bits to use to
813 determine the random offset to the base address of vma regions
814 resulting from mmap allocations. This value will be bounded
815 by the architecture's minimum and maximum supported values.
817 This value can be changed after boot using the
818 /proc/sys/vm/mmap_rnd_bits tunable
820 config HAVE_ARCH_MMAP_RND_COMPAT_BITS
823 An arch should select this symbol if it supports running applications
824 in compatibility mode, supports setting a variable number of bits for
825 use in establishing the base address for mmap allocations, has MMU
826 enabled and provides values for both:
827 - ARCH_MMAP_RND_COMPAT_BITS_MIN
828 - ARCH_MMAP_RND_COMPAT_BITS_MAX
830 config ARCH_MMAP_RND_COMPAT_BITS_MIN
833 config ARCH_MMAP_RND_COMPAT_BITS_MAX
836 config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
839 config ARCH_MMAP_RND_COMPAT_BITS
840 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT
841 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX
842 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
843 default ARCH_MMAP_RND_COMPAT_BITS_MIN
844 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS
846 This value can be used to select the number of bits to use to
847 determine the random offset to the base address of vma regions
848 resulting from mmap allocations for compatible applications This
849 value will be bounded by the architecture's minimum and maximum
852 This value can be changed after boot using the
853 /proc/sys/vm/mmap_rnd_compat_bits tunable
855 config HAVE_ARCH_COMPAT_MMAP_BASES
858 This allows 64bit applications to invoke 32-bit mmap() syscall
859 and vice-versa 32-bit applications to call 64-bit mmap().
860 Required for applications doing different bitness syscalls.
862 config HAVE_COPY_THREAD_TLS
865 Architecture provides copy_thread_tls to accept tls argument via
866 normal C parameter passing, rather than extracting the syscall
867 argument from pt_regs.
869 config HAVE_STACK_VALIDATION
872 Architecture supports the 'objtool check' host tool command, which
873 performs compile-time stack metadata validation.
875 config HAVE_RELIABLE_STACKTRACE
878 Architecture has a save_stack_trace_tsk_reliable() function which
879 only returns a stack trace if it can guarantee the trace is reliable.
881 config HAVE_ARCH_HASH
885 If this is set, the architecture provides an <asm/hash.h>
886 file which provides platform-specific implementations of some
887 functions in <linux/hash.h> or fs/namei.c.
895 config CLONE_BACKWARDS
898 Architecture has tls passed as the 4th argument of clone(2),
901 config CLONE_BACKWARDS2
904 Architecture has the first two arguments of clone(2) swapped.
906 config CLONE_BACKWARDS3
909 Architecture has tls passed as the 3rd argument of clone(2),
912 config ODD_RT_SIGACTION
915 Architecture has unusual rt_sigaction(2) arguments
917 config OLD_SIGSUSPEND
920 Architecture has old sigsuspend(2) syscall, of one-argument variety
922 config OLD_SIGSUSPEND3
925 Even weirder antique ABI - three-argument sigsuspend(2)
930 Architecture has old sigaction(2) syscall. Nope, not the same
931 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
932 but fairly different variant of sigaction(2), thanks to OSF/1
935 config COMPAT_OLD_SIGACTION
938 config ARCH_NO_COHERENT_DMA_MMAP
941 config CPU_NO_EFFICIENT_FFS
944 config HAVE_ARCH_VMAP_STACK
947 An arch should select this symbol if it can support kernel stacks
948 in vmalloc space. This means:
950 - vmalloc space must be large enough to hold many kernel stacks.
951 This may rule out many 32-bit architectures.
953 - Stacks in vmalloc space need to work reliably. For example, if
954 vmap page tables are created on demand, either this mechanism
955 needs to work while the stack points to a virtual address with
956 unpopulated page tables or arch code (switch_to() and switch_mm(),
957 most likely) needs to ensure that the stack's page table entries
958 are populated before running on a possibly unpopulated stack.
960 - If the stack overflows into a guard page, something reasonable
961 should happen. The definition of "reasonable" is flexible, but
962 instantly rebooting without logging anything would be unfriendly.
966 bool "Use a virtually-mapped stack"
967 depends on HAVE_ARCH_VMAP_STACK && !KASAN
969 Enable this if you want the use virtually-mapped kernel stacks
970 with guard pages. This causes kernel stack overflows to be
971 caught immediately rather than causing difficult-to-diagnose
974 This is presently incompatible with KASAN because KASAN expects
975 the stack to map directly to the KASAN shadow map using a formula
976 that is incorrect if the stack is in vmalloc space.
978 config ARCH_OPTIONAL_KERNEL_RWX
981 config ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
984 config ARCH_HAS_STRICT_KERNEL_RWX
987 config STRICT_KERNEL_RWX
988 bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX
989 depends on ARCH_HAS_STRICT_KERNEL_RWX
990 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
992 If this is set, kernel text and rodata memory will be made read-only,
993 and non-text memory will be made non-executable. This provides
994 protection against certain security exploits (e.g. executing the heap
997 These features are considered standard security practice these days.
998 You should say Y here in almost all cases.
1000 config ARCH_HAS_STRICT_MODULE_RWX
1003 config STRICT_MODULE_RWX
1004 bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX
1005 depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES
1006 default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT
1008 If this is set, module text and rodata memory will be made read-only,
1009 and non-text memory will be made non-executable. This provides
1010 protection against certain security exploits (e.g. writing to text)
1012 config ARCH_HAS_REFCOUNT
1015 An architecture selects this when it has implemented refcount_t
1016 using open coded assembly primitives that provide an optimized
1017 refcount_t implementation, possibly at the expense of some full
1018 refcount state checks of CONFIG_REFCOUNT_FULL=y.
1020 The refcount overflow check behavior, however, must be retained.
1021 Catching overflows is the primary security concern for protecting
1022 against bugs in reference counts.
1024 config REFCOUNT_FULL
1025 bool "Perform full reference count validation at the expense of speed"
1027 Enabling this switches the refcounting infrastructure from a fast
1028 unchecked atomic_t implementation to a fully state checked
1029 implementation, which can be (slightly) slower but provides protections
1030 against various use-after-free conditions that can be used in
1031 security flaw exploits.
1033 source "kernel/gcov/Kconfig"