2 # General architecture dependent options
12 tristate "OProfile system profiling"
14 depends on HAVE_OPROFILE
16 select RING_BUFFER_ALLOW_SWAP
18 OProfile is a profiling system capable of profiling the
19 whole system, include the kernel, kernel modules, libraries,
24 config OPROFILE_EVENT_MULTIPLEX
25 bool "OProfile multiplexing support (EXPERIMENTAL)"
27 depends on OPROFILE && X86
29 The number of hardware counters is limited. The multiplexing
30 feature enables OProfile to gather more events than counters
31 are provided by the hardware. This is realized by switching
32 between events at an user specified time interval.
39 config OPROFILE_NMI_TIMER
41 depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64
46 depends on HAVE_KPROBES
49 Kprobes allows you to trap at almost any kernel address and
50 execute a callback function. register_kprobe() establishes
51 a probepoint and specifies the callback. Kprobes is useful
52 for kernel debugging, non-intrusive instrumentation and testing.
56 bool "Optimize very unlikely/likely branches"
57 depends on HAVE_ARCH_JUMP_LABEL
59 This option enables a transparent branch optimization that
60 makes certain almost-always-true or almost-always-false branch
61 conditions even cheaper to execute within the kernel.
63 Certain performance-sensitive kernel code, such as trace points,
64 scheduler functionality, networking code and KVM have such
65 branches and include support for this optimization technique.
67 If it is detected that the compiler has support for "asm goto",
68 the kernel will compile such branches with just a nop
69 instruction. When the condition flag is toggled to true, the
70 nop will be converted to a jump instruction to execute the
71 conditional block of instructions.
73 This technique lowers overhead and stress on the branch prediction
74 of the processor and generally makes the kernel faster. The update
75 of the condition is slower, but those are always very rare.
77 ( On 32-bit x86, the necessary options added to the compiler
78 flags may increase the size of the kernel slightly. )
80 config STATIC_KEYS_SELFTEST
81 bool "Static key selftest"
84 Boot time self-test of the branch patching code.
88 depends on KPROBES && HAVE_OPTPROBES
91 config KPROBES_ON_FTRACE
93 depends on KPROBES && HAVE_KPROBES_ON_FTRACE
94 depends on DYNAMIC_FTRACE_WITH_REGS
96 If function tracer is enabled and the arch supports full
97 passing of pt_regs to function tracing, then kprobes can
98 optimize on top of function tracing.
103 Uprobes is the user-space counterpart to kprobes: they
104 enable instrumentation applications (such as 'perf probe')
105 to establish unintrusive probes in user-space binaries and
106 libraries, by executing handler functions when the probes
107 are hit by user-space applications.
109 ( These probes come in the form of single-byte breakpoints,
110 managed by the kernel and kept transparent to the probed
113 config HAVE_64BIT_ALIGNED_ACCESS
114 def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS
116 Some architectures require 64 bit accesses to be 64 bit
117 aligned, which also requires structs containing 64 bit values
118 to be 64 bit aligned too. This includes some 32 bit
119 architectures which can do 64 bit accesses, as well as 64 bit
120 architectures without unaligned access.
122 This symbol should be selected by an architecture if 64 bit
123 accesses are required to be 64 bit aligned in this way even
124 though it is not a 64 bit architecture.
126 See Documentation/unaligned-memory-access.txt for more
127 information on the topic of unaligned memory accesses.
129 config HAVE_EFFICIENT_UNALIGNED_ACCESS
132 Some architectures are unable to perform unaligned accesses
133 without the use of get_unaligned/put_unaligned. Others are
134 unable to perform such accesses efficiently (e.g. trap on
135 unaligned access and require fixing it up in the exception
138 This symbol should be selected by an architecture if it can
139 perform unaligned accesses efficiently to allow different
140 code paths to be selected for these cases. Some network
141 drivers, for example, could opt to not fix up alignment
142 problems with received packets if doing so would not help
145 See Documentation/unaligned-memory-access.txt for more
146 information on the topic of unaligned memory accesses.
148 config ARCH_USE_BUILTIN_BSWAP
151 Modern versions of GCC (since 4.4) have builtin functions
152 for handling byte-swapping. Using these, instead of the old
153 inline assembler that the architecture code provides in the
154 __arch_bswapXX() macros, allows the compiler to see what's
155 happening and offers more opportunity for optimisation. In
156 particular, the compiler will be able to combine the byteswap
157 with a nearby load or store and use load-and-swap or
158 store-and-swap instructions if the architecture has them. It
159 should almost *never* result in code which is worse than the
160 hand-coded assembler in <asm/swab.h>. But just in case it
161 does, the use of the builtins is optional.
163 Any architecture with load-and-swap or store-and-swap
164 instructions should set this. And it shouldn't hurt to set it
165 on architectures that don't have such instructions.
169 depends on KPROBES && HAVE_KRETPROBES
171 config USER_RETURN_NOTIFIER
173 depends on HAVE_USER_RETURN_NOTIFIER
175 Provide a kernel-internal notification when a cpu is about to
178 config HAVE_IOREMAP_PROT
184 config HAVE_KRETPROBES
187 config HAVE_OPTPROBES
190 config HAVE_KPROBES_ON_FTRACE
196 config HAVE_NMI_WATCHDOG
200 # An arch should select this if it provides all these things:
202 # task_pt_regs() in asm/processor.h or asm/ptrace.h
203 # arch_has_single_step() if there is hardware single-step support
204 # arch_has_block_step() if there is hardware block-step support
205 # asm/syscall.h supplying asm-generic/syscall.h interface
206 # linux/regset.h user_regset interfaces
207 # CORE_DUMP_USE_REGSET #define'd in linux/elf.h
208 # TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit}
209 # TIF_NOTIFY_RESUME calls tracehook_notify_resume()
210 # signal delivery calls tracehook_signal_handler()
212 config HAVE_ARCH_TRACEHOOK
215 config HAVE_DMA_CONTIGUOUS
218 config GENERIC_SMP_IDLE_THREAD
221 config GENERIC_IDLE_POLL_SETUP
224 # Select if arch init_task initializer is different to init/init_task.c
225 config ARCH_INIT_TASK
228 # Select if arch has its private alloc_task_struct() function
229 config ARCH_TASK_STRUCT_ALLOCATOR
232 # Select if arch has its private alloc_thread_stack() function
233 config ARCH_THREAD_STACK_ALLOCATOR
236 # Select if arch wants to size task_struct dynamically via arch_task_struct_size:
237 config ARCH_WANTS_DYNAMIC_TASK_STRUCT
240 config HAVE_REGS_AND_STACK_ACCESS_API
243 This symbol should be selected by an architecure if it supports
244 the API needed to access registers and stack entries from pt_regs,
245 declared in asm/ptrace.h
246 For example the kprobes-based event tracer needs this API.
251 The <linux/clk.h> calls support software clock gating and
252 thus are a key power management tool on many systems.
254 config HAVE_DMA_API_DEBUG
257 config HAVE_HW_BREAKPOINT
259 depends on PERF_EVENTS
261 config HAVE_MIXED_BREAKPOINTS_REGS
263 depends on HAVE_HW_BREAKPOINT
265 Depending on the arch implementation of hardware breakpoints,
266 some of them have separate registers for data and instruction
267 breakpoints addresses, others have mixed registers to store
268 them but define the access type in a control register.
269 Select this option if your arch implements breakpoints under the
272 config HAVE_USER_RETURN_NOTIFIER
275 config HAVE_PERF_EVENTS_NMI
278 System hardware can generate an NMI using the perf event
279 subsystem. Also has support for calculating CPU cycle events
280 to determine how many clock cycles in a given period.
282 config HAVE_PERF_REGS
285 Support selective register dumps for perf events. This includes
286 bit-mapping of each registers and a unique architecture id.
288 config HAVE_PERF_USER_STACK_DUMP
291 Support user stack dumps for perf event samples. This needs
292 access to the user stack pointer which is not unified across
295 config HAVE_ARCH_JUMP_LABEL
298 config HAVE_RCU_TABLE_FREE
301 config ARCH_HAVE_NMI_SAFE_CMPXCHG
304 config HAVE_ALIGNED_STRUCT_PAGE
307 This makes sure that struct pages are double word aligned and that
308 e.g. the SLUB allocator can perform double word atomic operations
309 on a struct page for better performance. However selecting this
310 might increase the size of a struct page by a word.
312 config HAVE_CMPXCHG_LOCAL
315 config HAVE_CMPXCHG_DOUBLE
318 config ARCH_WANT_IPC_PARSE_VERSION
321 config ARCH_WANT_COMPAT_IPC_PARSE_VERSION
324 config ARCH_WANT_OLD_COMPAT_IPC
325 select ARCH_WANT_COMPAT_IPC_PARSE_VERSION
328 config HAVE_ARCH_SECCOMP_FILTER
331 An arch should select this symbol if it provides all of these things:
333 - syscall_get_arguments()
335 - syscall_set_return_value()
336 - SIGSYS siginfo_t support
337 - secure_computing is called from a ptrace_event()-safe context
338 - secure_computing return value is checked and a return value of -1
339 results in the system call being skipped immediately.
340 - seccomp syscall wired up
342 config SECCOMP_FILTER
344 depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET
346 Enable tasks to build secure computing environments defined
347 in terms of Berkeley Packet Filter programs which implement
348 task-defined system call filtering polices.
350 See Documentation/prctl/seccomp_filter.txt for details.
352 config HAVE_GCC_PLUGINS
355 An arch should select this symbol if it supports building with
358 menuconfig GCC_PLUGINS
360 depends on HAVE_GCC_PLUGINS
361 depends on !COMPILE_TEST
363 GCC plugins are loadable modules that provide extra features to the
364 compiler. They are useful for runtime instrumentation and static analysis.
366 See Documentation/gcc-plugins.txt for details.
368 config GCC_PLUGIN_CYC_COMPLEXITY
369 bool "Compute the cyclomatic complexity of a function"
370 depends on GCC_PLUGINS
372 The complexity M of a function's control flow graph is defined as:
376 E = the number of edges
377 N = the number of nodes
378 P = the number of connected components (exit nodes).
380 config GCC_PLUGIN_SANCOV
382 depends on GCC_PLUGINS
384 This plugin inserts a __sanitizer_cov_trace_pc() call at the start of
385 basic blocks. It supports all gcc versions with plugin support (from
386 gcc-4.5 on). It is based on the commit "Add fuzzing coverage support"
387 by Dmitry Vyukov <dvyukov@google.com>.
389 config GCC_PLUGIN_LATENT_ENTROPY
390 bool "Generate some entropy during boot and runtime"
391 depends on GCC_PLUGINS
393 By saying Y here the kernel will instrument some kernel code to
394 extract some entropy from both original and artificially created
395 program state. This will help especially embedded systems where
396 there is little 'natural' source of entropy normally. The cost
397 is some slowdown of the boot process (about 0.5%) and fork and
400 Note that entropy extracted this way is not cryptographically
403 This plugin was ported from grsecurity/PaX. More information at:
404 * https://grsecurity.net/
405 * https://pax.grsecurity.net/
407 config HAVE_CC_STACKPROTECTOR
410 An arch should select this symbol if:
411 - its compiler supports the -fstack-protector option
412 - it has implemented a stack canary (e.g. __stack_chk_guard)
414 config CC_STACKPROTECTOR
417 Set when a stack-protector mode is enabled, so that the build
418 can enable kernel-side support for the GCC feature.
421 prompt "Stack Protector buffer overflow detection"
422 depends on HAVE_CC_STACKPROTECTOR
423 default CC_STACKPROTECTOR_NONE
425 This option turns on the "stack-protector" GCC feature. This
426 feature puts, at the beginning of functions, a canary value on
427 the stack just before the return address, and validates
428 the value just before actually returning. Stack based buffer
429 overflows (that need to overwrite this return address) now also
430 overwrite the canary, which gets detected and the attack is then
431 neutralized via a kernel panic.
433 config CC_STACKPROTECTOR_NONE
436 Disable "stack-protector" GCC feature.
438 config CC_STACKPROTECTOR_REGULAR
440 select CC_STACKPROTECTOR
442 Functions will have the stack-protector canary logic added if they
443 have an 8-byte or larger character array on the stack.
445 This feature requires gcc version 4.2 or above, or a distribution
446 gcc with the feature backported ("-fstack-protector").
448 On an x86 "defconfig" build, this feature adds canary checks to
449 about 3% of all kernel functions, which increases kernel code size
452 config CC_STACKPROTECTOR_STRONG
454 select CC_STACKPROTECTOR
456 Functions will have the stack-protector canary logic added in any
457 of the following conditions:
459 - local variable's address used as part of the right hand side of an
460 assignment or function argument
461 - local variable is an array (or union containing an array),
462 regardless of array type or length
463 - uses register local variables
465 This feature requires gcc version 4.9 or above, or a distribution
466 gcc with the feature backported ("-fstack-protector-strong").
468 On an x86 "defconfig" build, this feature adds canary checks to
469 about 20% of all kernel functions, which increases the kernel code
477 Select this if the architecture wants to use thin archives
478 instead of ld -r to create the built-in.o files.
480 config LD_DEAD_CODE_DATA_ELIMINATION
483 Select this if the architecture wants to do dead code and
484 data elimination with the linker by compiling with
485 -ffunction-sections -fdata-sections and linking with
488 This requires that the arch annotates or otherwise protects
489 its external entry points from being discarded. Linker scripts
490 must also merge .text.*, .data.*, and .bss.* correctly into
491 output sections. Care must be taken not to pull in unrelated
492 sections (e.g., '.text.init'). Typically '.' in section names
493 is used to distinguish them from label names / C identifiers.
495 config HAVE_ARCH_WITHIN_STACK_FRAMES
498 An architecture should select this if it can walk the kernel stack
499 frames to determine if an object is part of either the arguments
500 or local variables (i.e. that it excludes saved return addresses,
501 and similar) by implementing an inline arch_within_stack_frames(),
502 which is used by CONFIG_HARDENED_USERCOPY.
504 config HAVE_CONTEXT_TRACKING
507 Provide kernel/user boundaries probes necessary for subsystems
508 that need it, such as userspace RCU extended quiescent state.
509 Syscalls need to be wrapped inside user_exit()-user_enter() through
510 the slow path using TIF_NOHZ flag. Exceptions handlers must be
511 wrapped as well. Irqs are already protected inside
512 rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on
513 irq exit still need to be protected.
515 config HAVE_VIRT_CPU_ACCOUNTING
518 config HAVE_VIRT_CPU_ACCOUNTING_GEN
522 With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit.
523 Before enabling this option, arch code must be audited
524 to ensure there are no races in concurrent read/write of
525 cputime_t. For example, reading/writing 64-bit cputime_t on
526 some 32-bit arches may require multiple accesses, so proper
527 locking is needed to protect against concurrent accesses.
530 config HAVE_IRQ_TIME_ACCOUNTING
533 Archs need to ensure they use a high enough resolution clock to
534 support irq time accounting and then call enable_sched_clock_irqtime().
536 config HAVE_ARCH_TRANSPARENT_HUGEPAGE
539 config HAVE_ARCH_HUGE_VMAP
542 config HAVE_ARCH_SOFT_DIRTY
545 config HAVE_MOD_ARCH_SPECIFIC
548 The arch uses struct mod_arch_specific to store data. Many arches
549 just need a simple module loader without arch specific data - those
550 should not enable this.
552 config MODULES_USE_ELF_RELA
555 Modules only use ELF RELA relocations. Modules with ELF REL
556 relocations will give an error.
558 config MODULES_USE_ELF_REL
561 Modules only use ELF REL relocations. Modules with ELF RELA
562 relocations will give an error.
564 config HAVE_UNDERSCORE_SYMBOL_PREFIX
567 Some architectures generate an _ in front of C symbols; things like
568 module loading and assembly files need to know about this.
570 config HAVE_IRQ_EXIT_ON_IRQ_STACK
573 Architecture doesn't only execute the irq handler on the irq stack
574 but also irq_exit(). This way we can process softirqs on this irq
575 stack instead of switching to a new one when we call __do_softirq()
576 in the end of an hardirq.
577 This spares a stack switch and improves cache usage on softirq
580 config PGTABLE_LEVELS
584 config ARCH_HAS_ELF_RANDOMIZE
587 An architecture supports choosing randomized locations for
588 stack, mmap, brk, and ET_DYN. Defined functions:
590 - arch_randomize_brk()
592 config HAVE_ARCH_MMAP_RND_BITS
595 An arch should select this symbol if it supports setting a variable
596 number of bits for use in establishing the base address for mmap
597 allocations, has MMU enabled and provides values for both:
598 - ARCH_MMAP_RND_BITS_MIN
599 - ARCH_MMAP_RND_BITS_MAX
601 config HAVE_EXIT_THREAD
604 An architecture implements exit_thread.
606 config ARCH_MMAP_RND_BITS_MIN
609 config ARCH_MMAP_RND_BITS_MAX
612 config ARCH_MMAP_RND_BITS_DEFAULT
615 config ARCH_MMAP_RND_BITS
616 int "Number of bits to use for ASLR of mmap base address" if EXPERT
617 range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX
618 default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT
619 default ARCH_MMAP_RND_BITS_MIN
620 depends on HAVE_ARCH_MMAP_RND_BITS
622 This value can be used to select the number of bits to use to
623 determine the random offset to the base address of vma regions
624 resulting from mmap allocations. This value will be bounded
625 by the architecture's minimum and maximum supported values.
627 This value can be changed after boot using the
628 /proc/sys/vm/mmap_rnd_bits tunable
630 config HAVE_ARCH_MMAP_RND_COMPAT_BITS
633 An arch should select this symbol if it supports running applications
634 in compatibility mode, supports setting a variable number of bits for
635 use in establishing the base address for mmap allocations, has MMU
636 enabled and provides values for both:
637 - ARCH_MMAP_RND_COMPAT_BITS_MIN
638 - ARCH_MMAP_RND_COMPAT_BITS_MAX
640 config ARCH_MMAP_RND_COMPAT_BITS_MIN
643 config ARCH_MMAP_RND_COMPAT_BITS_MAX
646 config ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
649 config ARCH_MMAP_RND_COMPAT_BITS
650 int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT
651 range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX
652 default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT
653 default ARCH_MMAP_RND_COMPAT_BITS_MIN
654 depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS
656 This value can be used to select the number of bits to use to
657 determine the random offset to the base address of vma regions
658 resulting from mmap allocations for compatible applications This
659 value will be bounded by the architecture's minimum and maximum
662 This value can be changed after boot using the
663 /proc/sys/vm/mmap_rnd_compat_bits tunable
665 config HAVE_COPY_THREAD_TLS
668 Architecture provides copy_thread_tls to accept tls argument via
669 normal C parameter passing, rather than extracting the syscall
670 argument from pt_regs.
672 config HAVE_STACK_VALIDATION
675 Architecture supports the 'objtool check' host tool command, which
676 performs compile-time stack metadata validation.
678 config HAVE_ARCH_HASH
682 If this is set, the architecture provides an <asm/hash.h>
683 file which provides platform-specific implementations of some
684 functions in <linux/hash.h> or fs/namei.c.
692 config CLONE_BACKWARDS
695 Architecture has tls passed as the 4th argument of clone(2),
698 config CLONE_BACKWARDS2
701 Architecture has the first two arguments of clone(2) swapped.
703 config CLONE_BACKWARDS3
706 Architecture has tls passed as the 3rd argument of clone(2),
709 config ODD_RT_SIGACTION
712 Architecture has unusual rt_sigaction(2) arguments
714 config OLD_SIGSUSPEND
717 Architecture has old sigsuspend(2) syscall, of one-argument variety
719 config OLD_SIGSUSPEND3
722 Even weirder antique ABI - three-argument sigsuspend(2)
727 Architecture has old sigaction(2) syscall. Nope, not the same
728 as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2),
729 but fairly different variant of sigaction(2), thanks to OSF/1
732 config COMPAT_OLD_SIGACTION
735 config ARCH_NO_COHERENT_DMA_MMAP
738 config CPU_NO_EFFICIENT_FFS
741 config HAVE_ARCH_VMAP_STACK
744 An arch should select this symbol if it can support kernel stacks
745 in vmalloc space. This means:
747 - vmalloc space must be large enough to hold many kernel stacks.
748 This may rule out many 32-bit architectures.
750 - Stacks in vmalloc space need to work reliably. For example, if
751 vmap page tables are created on demand, either this mechanism
752 needs to work while the stack points to a virtual address with
753 unpopulated page tables or arch code (switch_to() and switch_mm(),
754 most likely) needs to ensure that the stack's page table entries
755 are populated before running on a possibly unpopulated stack.
757 - If the stack overflows into a guard page, something reasonable
758 should happen. The definition of "reasonable" is flexible, but
759 instantly rebooting without logging anything would be unfriendly.
763 bool "Use a virtually-mapped stack"
764 depends on HAVE_ARCH_VMAP_STACK && !KASAN
766 Enable this if you want the use virtually-mapped kernel stacks
767 with guard pages. This causes kernel stack overflows to be
768 caught immediately rather than causing difficult-to-diagnose
771 This is presently incompatible with KASAN because KASAN expects
772 the stack to map directly to the KASAN shadow map using a formula
773 that is incorrect if the stack is in vmalloc space.
775 source "kernel/gcov/Kconfig"