1b2441318SGreg Kroah-Hartman# SPDX-License-Identifier: GPL-2.0 2fb32e03fSMathieu Desnoyers# 3fb32e03fSMathieu Desnoyers# General architecture dependent options 4fb32e03fSMathieu Desnoyers# 5125e5645SMathieu Desnoyers 6692f66f2SHari Bathiniconfig CRASH_CORE 7692f66f2SHari Bathini bool 8692f66f2SHari Bathini 92965faa5SDave Youngconfig KEXEC_CORE 10692f66f2SHari Bathini select CRASH_CORE 112965faa5SDave Young bool 122965faa5SDave Young 13467d2782SThiago Jung Bauermannconfig HAVE_IMA_KEXEC 14467d2782SThiago Jung Bauermann bool 15467d2782SThiago Jung Bauermann 16125e5645SMathieu Desnoyersconfig OPROFILE 17b309a294SRobert Richter tristate "OProfile system profiling" 18125e5645SMathieu Desnoyers depends on PROFILING 19125e5645SMathieu Desnoyers depends on HAVE_OPROFILE 20d69d59f4SIngo Molnar select RING_BUFFER 219a5963ebSChristian Borntraeger select RING_BUFFER_ALLOW_SWAP 22125e5645SMathieu Desnoyers help 23125e5645SMathieu Desnoyers OProfile is a profiling system capable of profiling the 24125e5645SMathieu Desnoyers whole system, include the kernel, kernel modules, libraries, 25125e5645SMathieu Desnoyers and applications. 26125e5645SMathieu Desnoyers 27125e5645SMathieu Desnoyers If unsure, say N. 28125e5645SMathieu Desnoyers 294d4036e0SJason Yehconfig OPROFILE_EVENT_MULTIPLEX 304d4036e0SJason Yeh bool "OProfile multiplexing support (EXPERIMENTAL)" 314d4036e0SJason Yeh default n 324d4036e0SJason Yeh depends on OPROFILE && X86 334d4036e0SJason Yeh help 344d4036e0SJason Yeh The number of hardware counters is limited. The multiplexing 354d4036e0SJason Yeh feature enables OProfile to gather more events than counters 364d4036e0SJason Yeh are provided by the hardware. This is realized by switching 379332ef9dSMasahiro Yamada between events at a user specified time interval. 384d4036e0SJason Yeh 394d4036e0SJason Yeh If unsure, say N. 404d4036e0SJason Yeh 41125e5645SMathieu Desnoyersconfig HAVE_OPROFILE 429ba16087SJan Beulich bool 43125e5645SMathieu Desnoyers 44dcfce4a0SRobert Richterconfig OPROFILE_NMI_TIMER 45dcfce4a0SRobert Richter def_bool y 46af9feebeSAnton Blanchard depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64 47dcfce4a0SRobert Richter 48125e5645SMathieu Desnoyersconfig KPROBES 49125e5645SMathieu Desnoyers bool "Kprobes" 5005ed160eSMasami Hiramatsu depends on MODULES 51125e5645SMathieu Desnoyers depends on HAVE_KPROBES 5205ed160eSMasami Hiramatsu select KALLSYMS 53125e5645SMathieu Desnoyers help 54125e5645SMathieu Desnoyers Kprobes allows you to trap at almost any kernel address and 55125e5645SMathieu Desnoyers execute a callback function. register_kprobe() establishes 56125e5645SMathieu Desnoyers a probepoint and specifies the callback. Kprobes is useful 57125e5645SMathieu Desnoyers for kernel debugging, non-intrusive instrumentation and testing. 58125e5645SMathieu Desnoyers If in doubt, say "N". 59125e5645SMathieu Desnoyers 6045f81b1cSSteven Rostedtconfig JUMP_LABEL 61c5905afbSIngo Molnar bool "Optimize very unlikely/likely branches" 6245f81b1cSSteven Rostedt depends on HAVE_ARCH_JUMP_LABEL 6345f81b1cSSteven Rostedt help 64c5905afbSIngo Molnar This option enables a transparent branch optimization that 65c5905afbSIngo Molnar makes certain almost-always-true or almost-always-false branch 66c5905afbSIngo Molnar conditions even cheaper to execute within the kernel. 6745f81b1cSSteven Rostedt 68c5905afbSIngo Molnar Certain performance-sensitive kernel code, such as trace points, 69c5905afbSIngo Molnar scheduler functionality, networking code and KVM have such 70c5905afbSIngo Molnar branches and include support for this optimization technique. 71c5905afbSIngo Molnar 72c5905afbSIngo Molnar If it is detected that the compiler has support for "asm goto", 73c5905afbSIngo Molnar the kernel will compile such branches with just a nop 74c5905afbSIngo Molnar instruction. When the condition flag is toggled to true, the 75c5905afbSIngo Molnar nop will be converted to a jump instruction to execute the 76c5905afbSIngo Molnar conditional block of instructions. 77c5905afbSIngo Molnar 78c5905afbSIngo Molnar This technique lowers overhead and stress on the branch prediction 79c5905afbSIngo Molnar of the processor and generally makes the kernel faster. The update 80c5905afbSIngo Molnar of the condition is slower, but those are always very rare. 81c5905afbSIngo Molnar 82c5905afbSIngo Molnar ( On 32-bit x86, the necessary options added to the compiler 83c5905afbSIngo Molnar flags may increase the size of the kernel slightly. ) 8445f81b1cSSteven Rostedt 851987c947SPeter Zijlstraconfig STATIC_KEYS_SELFTEST 861987c947SPeter Zijlstra bool "Static key selftest" 871987c947SPeter Zijlstra depends on JUMP_LABEL 881987c947SPeter Zijlstra help 891987c947SPeter Zijlstra Boot time self-test of the branch patching code. 901987c947SPeter Zijlstra 91afd66255SMasami Hiramatsuconfig OPTPROBES 925cc718b9SMasami Hiramatsu def_bool y 935cc718b9SMasami Hiramatsu depends on KPROBES && HAVE_OPTPROBES 94a30b85dfSMasami Hiramatsu select TASKS_RCU if PREEMPT 95afd66255SMasami Hiramatsu 96e7dbfe34SMasami Hiramatsuconfig KPROBES_ON_FTRACE 97e7dbfe34SMasami Hiramatsu def_bool y 98e7dbfe34SMasami Hiramatsu depends on KPROBES && HAVE_KPROBES_ON_FTRACE 99e7dbfe34SMasami Hiramatsu depends on DYNAMIC_FTRACE_WITH_REGS 100e7dbfe34SMasami Hiramatsu help 101e7dbfe34SMasami Hiramatsu If function tracer is enabled and the arch supports full 102e7dbfe34SMasami Hiramatsu passing of pt_regs to function tracing, then kprobes can 103e7dbfe34SMasami Hiramatsu optimize on top of function tracing. 104e7dbfe34SMasami Hiramatsu 1052b144498SSrikar Dronamrajuconfig UPROBES 10609294e31SDavid A. Long def_bool n 107e8f4aa60SAllen Pais depends on ARCH_SUPPORTS_UPROBES 1082b144498SSrikar Dronamraju help 1097b2d81d4SIngo Molnar Uprobes is the user-space counterpart to kprobes: they 1107b2d81d4SIngo Molnar enable instrumentation applications (such as 'perf probe') 1117b2d81d4SIngo Molnar to establish unintrusive probes in user-space binaries and 1127b2d81d4SIngo Molnar libraries, by executing handler functions when the probes 1137b2d81d4SIngo Molnar are hit by user-space applications. 1147b2d81d4SIngo Molnar 1157b2d81d4SIngo Molnar ( These probes come in the form of single-byte breakpoints, 1167b2d81d4SIngo Molnar managed by the kernel and kept transparent to the probed 1177b2d81d4SIngo Molnar application. ) 1182b144498SSrikar Dronamraju 119c19fa94aSJames Hoganconfig HAVE_64BIT_ALIGNED_ACCESS 120c19fa94aSJames Hogan def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 121c19fa94aSJames Hogan help 122c19fa94aSJames Hogan Some architectures require 64 bit accesses to be 64 bit 123c19fa94aSJames Hogan aligned, which also requires structs containing 64 bit values 124c19fa94aSJames Hogan to be 64 bit aligned too. This includes some 32 bit 125c19fa94aSJames Hogan architectures which can do 64 bit accesses, as well as 64 bit 126c19fa94aSJames Hogan architectures without unaligned access. 127c19fa94aSJames Hogan 128c19fa94aSJames Hogan This symbol should be selected by an architecture if 64 bit 129c19fa94aSJames Hogan accesses are required to be 64 bit aligned in this way even 130c19fa94aSJames Hogan though it is not a 64 bit architecture. 131c19fa94aSJames Hogan 132c19fa94aSJames Hogan See Documentation/unaligned-memory-access.txt for more 133c19fa94aSJames Hogan information on the topic of unaligned memory accesses. 134c19fa94aSJames Hogan 13558340a07SJohannes Bergconfig HAVE_EFFICIENT_UNALIGNED_ACCESS 1369ba16087SJan Beulich bool 13758340a07SJohannes Berg help 13858340a07SJohannes Berg Some architectures are unable to perform unaligned accesses 13958340a07SJohannes Berg without the use of get_unaligned/put_unaligned. Others are 14058340a07SJohannes Berg unable to perform such accesses efficiently (e.g. trap on 14158340a07SJohannes Berg unaligned access and require fixing it up in the exception 14258340a07SJohannes Berg handler.) 14358340a07SJohannes Berg 14458340a07SJohannes Berg This symbol should be selected by an architecture if it can 14558340a07SJohannes Berg perform unaligned accesses efficiently to allow different 14658340a07SJohannes Berg code paths to be selected for these cases. Some network 14758340a07SJohannes Berg drivers, for example, could opt to not fix up alignment 14858340a07SJohannes Berg problems with received packets if doing so would not help 14958340a07SJohannes Berg much. 15058340a07SJohannes Berg 15158340a07SJohannes Berg See Documentation/unaligned-memory-access.txt for more 15258340a07SJohannes Berg information on the topic of unaligned memory accesses. 15358340a07SJohannes Berg 154cf66bb93SDavid Woodhouseconfig ARCH_USE_BUILTIN_BSWAP 155cf66bb93SDavid Woodhouse bool 156cf66bb93SDavid Woodhouse help 157cf66bb93SDavid Woodhouse Modern versions of GCC (since 4.4) have builtin functions 158cf66bb93SDavid Woodhouse for handling byte-swapping. Using these, instead of the old 159cf66bb93SDavid Woodhouse inline assembler that the architecture code provides in the 160cf66bb93SDavid Woodhouse __arch_bswapXX() macros, allows the compiler to see what's 161cf66bb93SDavid Woodhouse happening and offers more opportunity for optimisation. In 162cf66bb93SDavid Woodhouse particular, the compiler will be able to combine the byteswap 163cf66bb93SDavid Woodhouse with a nearby load or store and use load-and-swap or 164cf66bb93SDavid Woodhouse store-and-swap instructions if the architecture has them. It 165cf66bb93SDavid Woodhouse should almost *never* result in code which is worse than the 166cf66bb93SDavid Woodhouse hand-coded assembler in <asm/swab.h>. But just in case it 167cf66bb93SDavid Woodhouse does, the use of the builtins is optional. 168cf66bb93SDavid Woodhouse 169cf66bb93SDavid Woodhouse Any architecture with load-and-swap or store-and-swap 170cf66bb93SDavid Woodhouse instructions should set this. And it shouldn't hurt to set it 171cf66bb93SDavid Woodhouse on architectures that don't have such instructions. 172cf66bb93SDavid Woodhouse 1739edddaa2SAnanth N Mavinakayanahalliconfig KRETPROBES 1749edddaa2SAnanth N Mavinakayanahalli def_bool y 1759edddaa2SAnanth N Mavinakayanahalli depends on KPROBES && HAVE_KRETPROBES 1769edddaa2SAnanth N Mavinakayanahalli 1777c68af6eSAvi Kivityconfig USER_RETURN_NOTIFIER 1787c68af6eSAvi Kivity bool 1797c68af6eSAvi Kivity depends on HAVE_USER_RETURN_NOTIFIER 1807c68af6eSAvi Kivity help 1817c68af6eSAvi Kivity Provide a kernel-internal notification when a cpu is about to 1827c68af6eSAvi Kivity switch to user mode. 1837c68af6eSAvi Kivity 18428b2ee20SRik van Rielconfig HAVE_IOREMAP_PROT 1859ba16087SJan Beulich bool 18628b2ee20SRik van Riel 187125e5645SMathieu Desnoyersconfig HAVE_KPROBES 1889ba16087SJan Beulich bool 1899edddaa2SAnanth N Mavinakayanahalli 1909edddaa2SAnanth N Mavinakayanahalliconfig HAVE_KRETPROBES 1919ba16087SJan Beulich bool 19274bc7ceeSArthur Kepner 193afd66255SMasami Hiramatsuconfig HAVE_OPTPROBES 194afd66255SMasami Hiramatsu bool 195d314d74cSCong Wang 196e7dbfe34SMasami Hiramatsuconfig HAVE_KPROBES_ON_FTRACE 197e7dbfe34SMasami Hiramatsu bool 198e7dbfe34SMasami Hiramatsu 199540adea3SMasami Hiramatsuconfig HAVE_FUNCTION_ERROR_INJECTION 2009802d865SJosef Bacik bool 2019802d865SJosef Bacik 20242a0bb3fSPetr Mladekconfig HAVE_NMI 20342a0bb3fSPetr Mladek bool 20442a0bb3fSPetr Mladek 2051f5a4ad9SRoland McGrath# 2061f5a4ad9SRoland McGrath# An arch should select this if it provides all these things: 2071f5a4ad9SRoland McGrath# 2081f5a4ad9SRoland McGrath# task_pt_regs() in asm/processor.h or asm/ptrace.h 2091f5a4ad9SRoland McGrath# arch_has_single_step() if there is hardware single-step support 2101f5a4ad9SRoland McGrath# arch_has_block_step() if there is hardware block-step support 2111f5a4ad9SRoland McGrath# asm/syscall.h supplying asm-generic/syscall.h interface 2121f5a4ad9SRoland McGrath# linux/regset.h user_regset interfaces 2131f5a4ad9SRoland McGrath# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 2141f5a4ad9SRoland McGrath# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 2151f5a4ad9SRoland McGrath# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 2161f5a4ad9SRoland McGrath# signal delivery calls tracehook_signal_handler() 2171f5a4ad9SRoland McGrath# 2181f5a4ad9SRoland McGrathconfig HAVE_ARCH_TRACEHOOK 2199ba16087SJan Beulich bool 2201f5a4ad9SRoland McGrath 221c64be2bbSMarek Szyprowskiconfig HAVE_DMA_CONTIGUOUS 222c64be2bbSMarek Szyprowski bool 223c64be2bbSMarek Szyprowski 22429d5e047SThomas Gleixnerconfig GENERIC_SMP_IDLE_THREAD 22529d5e047SThomas Gleixner bool 22629d5e047SThomas Gleixner 227485cf5daSKevin Hilmanconfig GENERIC_IDLE_POLL_SETUP 228485cf5daSKevin Hilman bool 229485cf5daSKevin Hilman 2306974f0c4SDaniel Micayconfig ARCH_HAS_FORTIFY_SOURCE 2316974f0c4SDaniel Micay bool 2326974f0c4SDaniel Micay help 2336974f0c4SDaniel Micay An architecture should select this when it can successfully 2346974f0c4SDaniel Micay build and run with CONFIG_FORTIFY_SOURCE. 2356974f0c4SDaniel Micay 236d2852a22SDaniel Borkmann# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 237d2852a22SDaniel Borkmannconfig ARCH_HAS_SET_MEMORY 238d2852a22SDaniel Borkmann bool 239d2852a22SDaniel Borkmann 2400500871fSDavid Howells# Select if arch init_task must go in the __init_task_data section 2410500871fSDavid Howellsconfig ARCH_TASK_STRUCT_ON_STACK 242a4a2eb49SThomas Gleixner bool 243a4a2eb49SThomas Gleixner 244f5e10287SThomas Gleixner# Select if arch has its private alloc_task_struct() function 245f5e10287SThomas Gleixnerconfig ARCH_TASK_STRUCT_ALLOCATOR 246f5e10287SThomas Gleixner bool 247f5e10287SThomas Gleixner 2485905429aSKees Cookconfig HAVE_ARCH_THREAD_STRUCT_WHITELIST 2495905429aSKees Cook bool 2505905429aSKees Cook depends on !ARCH_TASK_STRUCT_ALLOCATOR 2515905429aSKees Cook help 2525905429aSKees Cook An architecture should select this to provide hardened usercopy 2535905429aSKees Cook knowledge about what region of the thread_struct should be 2545905429aSKees Cook whitelisted for copying to userspace. Normally this is only the 2555905429aSKees Cook FPU registers. Specifically, arch_thread_struct_whitelist() 2565905429aSKees Cook should be implemented. Without this, the entire thread_struct 2575905429aSKees Cook field in task_struct will be left whitelisted. 2585905429aSKees Cook 259b235beeaSLinus Torvalds# Select if arch has its private alloc_thread_stack() function 260b235beeaSLinus Torvaldsconfig ARCH_THREAD_STACK_ALLOCATOR 261f5e10287SThomas Gleixner bool 262f5e10287SThomas Gleixner 2635aaeb5c0SIngo Molnar# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 2645aaeb5c0SIngo Molnarconfig ARCH_WANTS_DYNAMIC_TASK_STRUCT 2655aaeb5c0SIngo Molnar bool 2665aaeb5c0SIngo Molnar 267f850c30cSHeiko Carstensconfig HAVE_REGS_AND_STACK_ACCESS_API 268f850c30cSHeiko Carstens bool 269e01292b1SHeiko Carstens help 270e01292b1SHeiko Carstens This symbol should be selected by an architecure if it supports 271e01292b1SHeiko Carstens the API needed to access registers and stack entries from pt_regs, 272e01292b1SHeiko Carstens declared in asm/ptrace.h 273e01292b1SHeiko Carstens For example the kprobes-based event tracer needs this API. 274f850c30cSHeiko Carstens 275d7822b1eSMathieu Desnoyersconfig HAVE_RSEQ 276d7822b1eSMathieu Desnoyers bool 277d7822b1eSMathieu Desnoyers depends on HAVE_REGS_AND_STACK_ACCESS_API 278d7822b1eSMathieu Desnoyers help 279d7822b1eSMathieu Desnoyers This symbol should be selected by an architecture if it 280d7822b1eSMathieu Desnoyers supports an implementation of restartable sequences. 281d7822b1eSMathieu Desnoyers 2829483a578SDavid Brownellconfig HAVE_CLK 2839ba16087SJan Beulich bool 2849483a578SDavid Brownell help 2859483a578SDavid Brownell The <linux/clk.h> calls support software clock gating and 2869483a578SDavid Brownell thus are a key power management tool on many systems. 2879483a578SDavid Brownell 28862a038d3SK.Prasadconfig HAVE_HW_BREAKPOINT 28962a038d3SK.Prasad bool 29099e8c5a3SFrederic Weisbecker depends on PERF_EVENTS 29162a038d3SK.Prasad 2920102752eSFrederic Weisbeckerconfig HAVE_MIXED_BREAKPOINTS_REGS 2930102752eSFrederic Weisbecker bool 2940102752eSFrederic Weisbecker depends on HAVE_HW_BREAKPOINT 2950102752eSFrederic Weisbecker help 2960102752eSFrederic Weisbecker Depending on the arch implementation of hardware breakpoints, 2970102752eSFrederic Weisbecker some of them have separate registers for data and instruction 2980102752eSFrederic Weisbecker breakpoints addresses, others have mixed registers to store 2990102752eSFrederic Weisbecker them but define the access type in a control register. 3000102752eSFrederic Weisbecker Select this option if your arch implements breakpoints under the 3010102752eSFrederic Weisbecker latter fashion. 3020102752eSFrederic Weisbecker 3037c68af6eSAvi Kivityconfig HAVE_USER_RETURN_NOTIFIER 3047c68af6eSAvi Kivity bool 305a1922ed6SIngo Molnar 306c01d4323SFrederic Weisbeckerconfig HAVE_PERF_EVENTS_NMI 307c01d4323SFrederic Weisbecker bool 30823637d47SFrederic Weisbecker help 30923637d47SFrederic Weisbecker System hardware can generate an NMI using the perf event 31023637d47SFrederic Weisbecker subsystem. Also has support for calculating CPU cycle events 31123637d47SFrederic Weisbecker to determine how many clock cycles in a given period. 312c01d4323SFrederic Weisbecker 31305a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_PERF 31405a4a952SNicholas Piggin bool 31505a4a952SNicholas Piggin depends on HAVE_PERF_EVENTS_NMI 31605a4a952SNicholas Piggin help 31705a4a952SNicholas Piggin The arch chooses to use the generic perf-NMI-based hardlockup 31805a4a952SNicholas Piggin detector. Must define HAVE_PERF_EVENTS_NMI. 31905a4a952SNicholas Piggin 32005a4a952SNicholas Pigginconfig HAVE_NMI_WATCHDOG 32105a4a952SNicholas Piggin depends on HAVE_NMI 32205a4a952SNicholas Piggin bool 32305a4a952SNicholas Piggin help 32405a4a952SNicholas Piggin The arch provides a low level NMI watchdog. It provides 32505a4a952SNicholas Piggin asm/nmi.h, and defines its own arch_touch_nmi_watchdog(). 32605a4a952SNicholas Piggin 32705a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_ARCH 32805a4a952SNicholas Piggin bool 32905a4a952SNicholas Piggin select HAVE_NMI_WATCHDOG 33005a4a952SNicholas Piggin help 33105a4a952SNicholas Piggin The arch chooses to provide its own hardlockup detector, which is 33205a4a952SNicholas Piggin a superset of the HAVE_NMI_WATCHDOG. It also conforms to config 33305a4a952SNicholas Piggin interfaces and parameters provided by hardlockup detector subsystem. 33405a4a952SNicholas Piggin 335c5e63197SJiri Olsaconfig HAVE_PERF_REGS 336c5e63197SJiri Olsa bool 337c5e63197SJiri Olsa help 338c5e63197SJiri Olsa Support selective register dumps for perf events. This includes 339c5e63197SJiri Olsa bit-mapping of each registers and a unique architecture id. 340c5e63197SJiri Olsa 341c5ebcedbSJiri Olsaconfig HAVE_PERF_USER_STACK_DUMP 342c5ebcedbSJiri Olsa bool 343c5ebcedbSJiri Olsa help 344c5ebcedbSJiri Olsa Support user stack dumps for perf event samples. This needs 345c5ebcedbSJiri Olsa access to the user stack pointer which is not unified across 346c5ebcedbSJiri Olsa architectures. 347c5ebcedbSJiri Olsa 348bf5438fcSJason Baronconfig HAVE_ARCH_JUMP_LABEL 349bf5438fcSJason Baron bool 350bf5438fcSJason Baron 35126723911SPeter Zijlstraconfig HAVE_RCU_TABLE_FREE 35226723911SPeter Zijlstra bool 35326723911SPeter Zijlstra 354df013ffbSHuang Yingconfig ARCH_HAVE_NMI_SAFE_CMPXCHG 355df013ffbSHuang Ying bool 356df013ffbSHuang Ying 35743570fd2SHeiko Carstensconfig HAVE_ALIGNED_STRUCT_PAGE 35843570fd2SHeiko Carstens bool 35943570fd2SHeiko Carstens help 36043570fd2SHeiko Carstens This makes sure that struct pages are double word aligned and that 36143570fd2SHeiko Carstens e.g. the SLUB allocator can perform double word atomic operations 36243570fd2SHeiko Carstens on a struct page for better performance. However selecting this 36343570fd2SHeiko Carstens might increase the size of a struct page by a word. 36443570fd2SHeiko Carstens 3654156153cSHeiko Carstensconfig HAVE_CMPXCHG_LOCAL 3664156153cSHeiko Carstens bool 3674156153cSHeiko Carstens 3682565409fSHeiko Carstensconfig HAVE_CMPXCHG_DOUBLE 3692565409fSHeiko Carstens bool 3702565409fSHeiko Carstens 37177e58496SPaul E. McKenneyconfig ARCH_WEAK_RELEASE_ACQUIRE 37277e58496SPaul E. McKenney bool 37377e58496SPaul E. McKenney 374c1d7e01dSWill Deaconconfig ARCH_WANT_IPC_PARSE_VERSION 375c1d7e01dSWill Deacon bool 376c1d7e01dSWill Deacon 377c1d7e01dSWill Deaconconfig ARCH_WANT_COMPAT_IPC_PARSE_VERSION 378c1d7e01dSWill Deacon bool 379c1d7e01dSWill Deacon 38048b25c43SChris Metcalfconfig ARCH_WANT_OLD_COMPAT_IPC 381c1d7e01dSWill Deacon select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 38248b25c43SChris Metcalf bool 38348b25c43SChris Metcalf 384e2cfabdfSWill Drewryconfig HAVE_ARCH_SECCOMP_FILTER 385e2cfabdfSWill Drewry bool 386e2cfabdfSWill Drewry help 387fb0fadf9SWill Drewry An arch should select this symbol if it provides all of these things: 388bb6ea430SWill Drewry - syscall_get_arch() 389bb6ea430SWill Drewry - syscall_get_arguments() 390bb6ea430SWill Drewry - syscall_rollback() 391bb6ea430SWill Drewry - syscall_set_return_value() 392fb0fadf9SWill Drewry - SIGSYS siginfo_t support 393fb0fadf9SWill Drewry - secure_computing is called from a ptrace_event()-safe context 394fb0fadf9SWill Drewry - secure_computing return value is checked and a return value of -1 395fb0fadf9SWill Drewry results in the system call being skipped immediately. 39648dc92b9SKees Cook - seccomp syscall wired up 397e2cfabdfSWill Drewry 398e2cfabdfSWill Drewryconfig SECCOMP_FILTER 399e2cfabdfSWill Drewry def_bool y 400e2cfabdfSWill Drewry depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 401e2cfabdfSWill Drewry help 402e2cfabdfSWill Drewry Enable tasks to build secure computing environments defined 403e2cfabdfSWill Drewry in terms of Berkeley Packet Filter programs which implement 404e2cfabdfSWill Drewry task-defined system call filtering polices. 405e2cfabdfSWill Drewry 406e2cfabdfSWill Drewry See Documentation/prctl/seccomp_filter.txt for details. 407e2cfabdfSWill Drewry 4086b90bd4bSEmese Revfyconfig HAVE_GCC_PLUGINS 4096b90bd4bSEmese Revfy bool 4106b90bd4bSEmese Revfy help 4116b90bd4bSEmese Revfy An arch should select this symbol if it supports building with 4126b90bd4bSEmese Revfy GCC plugins. 4136b90bd4bSEmese Revfy 4146b90bd4bSEmese Revfymenuconfig GCC_PLUGINS 4156b90bd4bSEmese Revfy bool "GCC plugins" 4166b90bd4bSEmese Revfy depends on HAVE_GCC_PLUGINS 417a519167eSKees Cook depends on !COMPILE_TEST 4186b90bd4bSEmese Revfy help 4196b90bd4bSEmese Revfy GCC plugins are loadable modules that provide extra features to the 4206b90bd4bSEmese Revfy compiler. They are useful for runtime instrumentation and static analysis. 4216b90bd4bSEmese Revfy 4226b90bd4bSEmese Revfy See Documentation/gcc-plugins.txt for details. 4236b90bd4bSEmese Revfy 4240dae776cSEmese Revfyconfig GCC_PLUGIN_CYC_COMPLEXITY 425215e2aa6SKees Cook bool "Compute the cyclomatic complexity of a function" if EXPERT 4260dae776cSEmese Revfy depends on GCC_PLUGINS 427215e2aa6SKees Cook depends on !COMPILE_TEST 4280dae776cSEmese Revfy help 4290dae776cSEmese Revfy The complexity M of a function's control flow graph is defined as: 4300dae776cSEmese Revfy M = E - N + 2P 4310dae776cSEmese Revfy where 4320dae776cSEmese Revfy 4330dae776cSEmese Revfy E = the number of edges 4340dae776cSEmese Revfy N = the number of nodes 4350dae776cSEmese Revfy P = the number of connected components (exit nodes). 4360dae776cSEmese Revfy 437215e2aa6SKees Cook Enabling this plugin reports the complexity to stderr during the 438215e2aa6SKees Cook build. It mainly serves as a simple example of how to create a 439215e2aa6SKees Cook gcc plugin for the kernel. 440215e2aa6SKees Cook 441543c37cbSEmese Revfyconfig GCC_PLUGIN_SANCOV 442543c37cbSEmese Revfy bool 443543c37cbSEmese Revfy depends on GCC_PLUGINS 444543c37cbSEmese Revfy help 445543c37cbSEmese Revfy This plugin inserts a __sanitizer_cov_trace_pc() call at the start of 446543c37cbSEmese Revfy basic blocks. It supports all gcc versions with plugin support (from 447543c37cbSEmese Revfy gcc-4.5 on). It is based on the commit "Add fuzzing coverage support" 448543c37cbSEmese Revfy by Dmitry Vyukov <dvyukov@google.com>. 449543c37cbSEmese Revfy 45038addce8SEmese Revfyconfig GCC_PLUGIN_LATENT_ENTROPY 45138addce8SEmese Revfy bool "Generate some entropy during boot and runtime" 45238addce8SEmese Revfy depends on GCC_PLUGINS 45338addce8SEmese Revfy help 45438addce8SEmese Revfy By saying Y here the kernel will instrument some kernel code to 45538addce8SEmese Revfy extract some entropy from both original and artificially created 45638addce8SEmese Revfy program state. This will help especially embedded systems where 45738addce8SEmese Revfy there is little 'natural' source of entropy normally. The cost 45838addce8SEmese Revfy is some slowdown of the boot process (about 0.5%) and fork and 45938addce8SEmese Revfy irq processing. 46038addce8SEmese Revfy 46138addce8SEmese Revfy Note that entropy extracted this way is not cryptographically 46238addce8SEmese Revfy secure! 46338addce8SEmese Revfy 46438addce8SEmese Revfy This plugin was ported from grsecurity/PaX. More information at: 46538addce8SEmese Revfy * https://grsecurity.net/ 46638addce8SEmese Revfy * https://pax.grsecurity.net/ 46738addce8SEmese Revfy 468c61f13eaSKees Cookconfig GCC_PLUGIN_STRUCTLEAK 469c61f13eaSKees Cook bool "Force initialization of variables containing userspace addresses" 470c61f13eaSKees Cook depends on GCC_PLUGINS 471c9cf87eaSDmitry Vyukov # Currently STRUCTLEAK inserts initialization out of live scope of 472c9cf87eaSDmitry Vyukov # variables from KASAN point of view. This leads to KASAN false 473c9cf87eaSDmitry Vyukov # positive reports. Prohibit this combination for now. 474c9cf87eaSDmitry Vyukov depends on !KASAN_EXTRA 475c61f13eaSKees Cook help 476f136e090SJean Delvare This plugin zero-initializes any structures containing a 477c61f13eaSKees Cook __user attribute. This can prevent some classes of information 478c61f13eaSKees Cook exposures. 479c61f13eaSKees Cook 480c61f13eaSKees Cook This plugin was ported from grsecurity/PaX. More information at: 481c61f13eaSKees Cook * https://grsecurity.net/ 482c61f13eaSKees Cook * https://pax.grsecurity.net/ 483c61f13eaSKees Cook 484f7dd2507SArd Biesheuvelconfig GCC_PLUGIN_STRUCTLEAK_BYREF_ALL 485f7dd2507SArd Biesheuvel bool "Force initialize all struct type variables passed by reference" 486f7dd2507SArd Biesheuvel depends on GCC_PLUGIN_STRUCTLEAK 487f7dd2507SArd Biesheuvel help 488f7dd2507SArd Biesheuvel Zero initialize any struct type local variable that may be passed by 489f7dd2507SArd Biesheuvel reference without having been initialized. 490f7dd2507SArd Biesheuvel 491c61f13eaSKees Cookconfig GCC_PLUGIN_STRUCTLEAK_VERBOSE 492c61f13eaSKees Cook bool "Report forcefully initialized variables" 493c61f13eaSKees Cook depends on GCC_PLUGIN_STRUCTLEAK 494c61f13eaSKees Cook depends on !COMPILE_TEST 495c61f13eaSKees Cook help 496c61f13eaSKees Cook This option will cause a warning to be printed each time the 497c61f13eaSKees Cook structleak plugin finds a variable it thinks needs to be 498c61f13eaSKees Cook initialized. Since not all existing initializers are detected 499c61f13eaSKees Cook by the plugin, this can produce false positive warnings. 500c61f13eaSKees Cook 501313dd1b6SKees Cookconfig GCC_PLUGIN_RANDSTRUCT 502313dd1b6SKees Cook bool "Randomize layout of sensitive kernel structures" 503313dd1b6SKees Cook depends on GCC_PLUGINS 504313dd1b6SKees Cook select MODVERSIONS if MODULES 505313dd1b6SKees Cook help 5069225331bSKees Cook If you say Y here, the layouts of structures that are entirely 5079225331bSKees Cook function pointers (and have not been manually annotated with 5089225331bSKees Cook __no_randomize_layout), or structures that have been explicitly 5099225331bSKees Cook marked with __randomize_layout, will be randomized at compile-time. 5109225331bSKees Cook This can introduce the requirement of an additional information 5119225331bSKees Cook exposure vulnerability for exploits targeting these structure 5129225331bSKees Cook types. 513313dd1b6SKees Cook 514313dd1b6SKees Cook Enabling this feature will introduce some performance impact, 515313dd1b6SKees Cook slightly increase memory usage, and prevent the use of forensic 516313dd1b6SKees Cook tools like Volatility against the system (unless the kernel 517313dd1b6SKees Cook source tree isn't cleaned after kernel installation). 518313dd1b6SKees Cook 519313dd1b6SKees Cook The seed used for compilation is located at 520313dd1b6SKees Cook scripts/gcc-plgins/randomize_layout_seed.h. It remains after 521313dd1b6SKees Cook a make clean to allow for external modules to be compiled with 522313dd1b6SKees Cook the existing seed and will be removed by a make mrproper or 523313dd1b6SKees Cook make distclean. 524313dd1b6SKees Cook 525313dd1b6SKees Cook Note that the implementation requires gcc 4.7 or newer. 526313dd1b6SKees Cook 527313dd1b6SKees Cook This plugin was ported from grsecurity/PaX. More information at: 528313dd1b6SKees Cook * https://grsecurity.net/ 529313dd1b6SKees Cook * https://pax.grsecurity.net/ 530313dd1b6SKees Cook 531313dd1b6SKees Cookconfig GCC_PLUGIN_RANDSTRUCT_PERFORMANCE 532313dd1b6SKees Cook bool "Use cacheline-aware structure randomization" 533313dd1b6SKees Cook depends on GCC_PLUGIN_RANDSTRUCT 534313dd1b6SKees Cook depends on !COMPILE_TEST 535313dd1b6SKees Cook help 536313dd1b6SKees Cook If you say Y here, the RANDSTRUCT randomization will make a 537313dd1b6SKees Cook best effort at restricting randomization to cacheline-sized 538313dd1b6SKees Cook groups of elements. It will further not randomize bitfields 539313dd1b6SKees Cook in structures. This reduces the performance hit of RANDSTRUCT 540313dd1b6SKees Cook at the cost of weakened randomization. 541313dd1b6SKees Cook 54219952a92SKees Cookconfig HAVE_CC_STACKPROTECTOR 54319952a92SKees Cook bool 54419952a92SKees Cook help 54519952a92SKees Cook An arch should select this symbol if: 54619952a92SKees Cook - its compiler supports the -fstack-protector option 54719952a92SKees Cook - it has implemented a stack canary (e.g. __stack_chk_guard) 54819952a92SKees Cook 5498779657dSKees Cookchoice 5508779657dSKees Cook prompt "Stack Protector buffer overflow detection" 5518779657dSKees Cook depends on HAVE_CC_STACKPROTECTOR 55244c6dc94SKees Cook default CC_STACKPROTECTOR_AUTO 5538779657dSKees Cook help 5548779657dSKees Cook This option turns on the "stack-protector" GCC feature. This 55519952a92SKees Cook feature puts, at the beginning of functions, a canary value on 55619952a92SKees Cook the stack just before the return address, and validates 55719952a92SKees Cook the value just before actually returning. Stack based buffer 55819952a92SKees Cook overflows (that need to overwrite this return address) now also 55919952a92SKees Cook overwrite the canary, which gets detected and the attack is then 56019952a92SKees Cook neutralized via a kernel panic. 56119952a92SKees Cook 5628779657dSKees Cookconfig CC_STACKPROTECTOR_NONE 5638779657dSKees Cook bool "None" 5648779657dSKees Cook help 5658779657dSKees Cook Disable "stack-protector" GCC feature. 5668779657dSKees Cook 5678779657dSKees Cookconfig CC_STACKPROTECTOR_REGULAR 5688779657dSKees Cook bool "Regular" 5698779657dSKees Cook help 5708779657dSKees Cook Functions will have the stack-protector canary logic added if they 5718779657dSKees Cook have an 8-byte or larger character array on the stack. 5728779657dSKees Cook 57319952a92SKees Cook This feature requires gcc version 4.2 or above, or a distribution 5748779657dSKees Cook gcc with the feature backported ("-fstack-protector"). 5758779657dSKees Cook 5768779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 5778779657dSKees Cook about 3% of all kernel functions, which increases kernel code size 5788779657dSKees Cook by about 0.3%. 5798779657dSKees Cook 5808779657dSKees Cookconfig CC_STACKPROTECTOR_STRONG 5818779657dSKees Cook bool "Strong" 5828779657dSKees Cook help 5838779657dSKees Cook Functions will have the stack-protector canary logic added in any 5848779657dSKees Cook of the following conditions: 5858779657dSKees Cook 5868779657dSKees Cook - local variable's address used as part of the right hand side of an 5878779657dSKees Cook assignment or function argument 5888779657dSKees Cook - local variable is an array (or union containing an array), 5898779657dSKees Cook regardless of array type or length 5908779657dSKees Cook - uses register local variables 5918779657dSKees Cook 5928779657dSKees Cook This feature requires gcc version 4.9 or above, or a distribution 5938779657dSKees Cook gcc with the feature backported ("-fstack-protector-strong"). 5948779657dSKees Cook 5958779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 5968779657dSKees Cook about 20% of all kernel functions, which increases the kernel code 5978779657dSKees Cook size by about 2%. 5988779657dSKees Cook 59944c6dc94SKees Cookconfig CC_STACKPROTECTOR_AUTO 60044c6dc94SKees Cook bool "Automatic" 60144c6dc94SKees Cook help 60244c6dc94SKees Cook If the compiler supports it, the best available stack-protector 60344c6dc94SKees Cook option will be chosen. 60444c6dc94SKees Cook 6058779657dSKees Cookendchoice 60619952a92SKees Cook 607b67067f1SNicholas Pigginconfig LD_DEAD_CODE_DATA_ELIMINATION 608b67067f1SNicholas Piggin bool 609b67067f1SNicholas Piggin help 610b67067f1SNicholas Piggin Select this if the architecture wants to do dead code and 611b67067f1SNicholas Piggin data elimination with the linker by compiling with 612b67067f1SNicholas Piggin -ffunction-sections -fdata-sections and linking with 613b67067f1SNicholas Piggin --gc-sections. 614b67067f1SNicholas Piggin 615b67067f1SNicholas Piggin This requires that the arch annotates or otherwise protects 616b67067f1SNicholas Piggin its external entry points from being discarded. Linker scripts 617b67067f1SNicholas Piggin must also merge .text.*, .data.*, and .bss.* correctly into 6180f4c4af0SNicholas Piggin output sections. Care must be taken not to pull in unrelated 6190f4c4af0SNicholas Piggin sections (e.g., '.text.init'). Typically '.' in section names 6200f4c4af0SNicholas Piggin is used to distinguish them from label names / C identifiers. 621b67067f1SNicholas Piggin 6220f60a8efSKees Cookconfig HAVE_ARCH_WITHIN_STACK_FRAMES 6230f60a8efSKees Cook bool 6240f60a8efSKees Cook help 6250f60a8efSKees Cook An architecture should select this if it can walk the kernel stack 6260f60a8efSKees Cook frames to determine if an object is part of either the arguments 6270f60a8efSKees Cook or local variables (i.e. that it excludes saved return addresses, 6280f60a8efSKees Cook and similar) by implementing an inline arch_within_stack_frames(), 6290f60a8efSKees Cook which is used by CONFIG_HARDENED_USERCOPY. 6300f60a8efSKees Cook 63191d1aa43SFrederic Weisbeckerconfig HAVE_CONTEXT_TRACKING 6322b1d5024SFrederic Weisbecker bool 6332b1d5024SFrederic Weisbecker help 63491d1aa43SFrederic Weisbecker Provide kernel/user boundaries probes necessary for subsystems 63591d1aa43SFrederic Weisbecker that need it, such as userspace RCU extended quiescent state. 63691d1aa43SFrederic Weisbecker Syscalls need to be wrapped inside user_exit()-user_enter() through 63791d1aa43SFrederic Weisbecker the slow path using TIF_NOHZ flag. Exceptions handlers must be 63891d1aa43SFrederic Weisbecker wrapped as well. Irqs are already protected inside 63991d1aa43SFrederic Weisbecker rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on 64091d1aa43SFrederic Weisbecker irq exit still need to be protected. 6412b1d5024SFrederic Weisbecker 642b952741cSFrederic Weisbeckerconfig HAVE_VIRT_CPU_ACCOUNTING 643b952741cSFrederic Weisbecker bool 644b952741cSFrederic Weisbecker 64540565b5aSStanislaw Gruszkaconfig ARCH_HAS_SCALED_CPUTIME 64640565b5aSStanislaw Gruszka bool 64740565b5aSStanislaw Gruszka 648554b0004SKevin Hilmanconfig HAVE_VIRT_CPU_ACCOUNTING_GEN 649554b0004SKevin Hilman bool 650554b0004SKevin Hilman default y if 64BIT 651554b0004SKevin Hilman help 652554b0004SKevin Hilman With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 653554b0004SKevin Hilman Before enabling this option, arch code must be audited 654554b0004SKevin Hilman to ensure there are no races in concurrent read/write of 655554b0004SKevin Hilman cputime_t. For example, reading/writing 64-bit cputime_t on 656554b0004SKevin Hilman some 32-bit arches may require multiple accesses, so proper 657554b0004SKevin Hilman locking is needed to protect against concurrent accesses. 658554b0004SKevin Hilman 659554b0004SKevin Hilman 660fdf9c356SFrederic Weisbeckerconfig HAVE_IRQ_TIME_ACCOUNTING 661fdf9c356SFrederic Weisbecker bool 662fdf9c356SFrederic Weisbecker help 663fdf9c356SFrederic Weisbecker Archs need to ensure they use a high enough resolution clock to 664fdf9c356SFrederic Weisbecker support irq time accounting and then call enable_sched_clock_irqtime(). 665fdf9c356SFrederic Weisbecker 66615626062SGerald Schaeferconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE 66715626062SGerald Schaefer bool 66815626062SGerald Schaefer 669a00cc7d9SMatthew Wilcoxconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 670a00cc7d9SMatthew Wilcox bool 671a00cc7d9SMatthew Wilcox 6720ddab1d2SToshi Kaniconfig HAVE_ARCH_HUGE_VMAP 6730ddab1d2SToshi Kani bool 6740ddab1d2SToshi Kani 6750f8975ecSPavel Emelyanovconfig HAVE_ARCH_SOFT_DIRTY 6760f8975ecSPavel Emelyanov bool 6770f8975ecSPavel Emelyanov 678786d35d4SDavid Howellsconfig HAVE_MOD_ARCH_SPECIFIC 679786d35d4SDavid Howells bool 680786d35d4SDavid Howells help 681786d35d4SDavid Howells The arch uses struct mod_arch_specific to store data. Many arches 682786d35d4SDavid Howells just need a simple module loader without arch specific data - those 683786d35d4SDavid Howells should not enable this. 684786d35d4SDavid Howells 685786d35d4SDavid Howellsconfig MODULES_USE_ELF_RELA 686786d35d4SDavid Howells bool 687786d35d4SDavid Howells help 688786d35d4SDavid Howells Modules only use ELF RELA relocations. Modules with ELF REL 689786d35d4SDavid Howells relocations will give an error. 690786d35d4SDavid Howells 691786d35d4SDavid Howellsconfig MODULES_USE_ELF_REL 692786d35d4SDavid Howells bool 693786d35d4SDavid Howells help 694786d35d4SDavid Howells Modules only use ELF REL relocations. Modules with ELF RELA 695786d35d4SDavid Howells relocations will give an error. 696786d35d4SDavid Howells 697b92021b0SRusty Russellconfig HAVE_UNDERSCORE_SYMBOL_PREFIX 698b92021b0SRusty Russell bool 699b92021b0SRusty Russell help 700b92021b0SRusty Russell Some architectures generate an _ in front of C symbols; things like 701b92021b0SRusty Russell module loading and assembly files need to know about this. 702b92021b0SRusty Russell 703cc1f0274SFrederic Weisbeckerconfig HAVE_IRQ_EXIT_ON_IRQ_STACK 704cc1f0274SFrederic Weisbecker bool 705cc1f0274SFrederic Weisbecker help 706cc1f0274SFrederic Weisbecker Architecture doesn't only execute the irq handler on the irq stack 707cc1f0274SFrederic Weisbecker but also irq_exit(). This way we can process softirqs on this irq 708cc1f0274SFrederic Weisbecker stack instead of switching to a new one when we call __do_softirq() 709cc1f0274SFrederic Weisbecker in the end of an hardirq. 710cc1f0274SFrederic Weisbecker This spares a stack switch and improves cache usage on softirq 711cc1f0274SFrederic Weisbecker processing. 712cc1f0274SFrederic Weisbecker 713235a8f02SKirill A. Shutemovconfig PGTABLE_LEVELS 714235a8f02SKirill A. Shutemov int 715235a8f02SKirill A. Shutemov default 2 716235a8f02SKirill A. Shutemov 7172b68f6caSKees Cookconfig ARCH_HAS_ELF_RANDOMIZE 7182b68f6caSKees Cook bool 7192b68f6caSKees Cook help 7202b68f6caSKees Cook An architecture supports choosing randomized locations for 7212b68f6caSKees Cook stack, mmap, brk, and ET_DYN. Defined functions: 7222b68f6caSKees Cook - arch_mmap_rnd() 723204db6edSKees Cook - arch_randomize_brk() 7242b68f6caSKees Cook 725d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_BITS 726d07e2259SDaniel Cashman bool 727d07e2259SDaniel Cashman help 728d07e2259SDaniel Cashman An arch should select this symbol if it supports setting a variable 729d07e2259SDaniel Cashman number of bits for use in establishing the base address for mmap 730d07e2259SDaniel Cashman allocations, has MMU enabled and provides values for both: 731d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MIN 732d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MAX 733d07e2259SDaniel Cashman 7345f56a5dfSJiri Slabyconfig HAVE_EXIT_THREAD 7355f56a5dfSJiri Slaby bool 7365f56a5dfSJiri Slaby help 7375f56a5dfSJiri Slaby An architecture implements exit_thread. 7385f56a5dfSJiri Slaby 739d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MIN 740d07e2259SDaniel Cashman int 741d07e2259SDaniel Cashman 742d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MAX 743d07e2259SDaniel Cashman int 744d07e2259SDaniel Cashman 745d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_DEFAULT 746d07e2259SDaniel Cashman int 747d07e2259SDaniel Cashman 748d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS 749d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address" if EXPERT 750d07e2259SDaniel Cashman range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 751d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 752d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_MIN 753d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_BITS 754d07e2259SDaniel Cashman help 755d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 756d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 757d07e2259SDaniel Cashman resulting from mmap allocations. This value will be bounded 758d07e2259SDaniel Cashman by the architecture's minimum and maximum supported values. 759d07e2259SDaniel Cashman 760d07e2259SDaniel Cashman This value can be changed after boot using the 761d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_bits tunable 762d07e2259SDaniel Cashman 763d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_COMPAT_BITS 764d07e2259SDaniel Cashman bool 765d07e2259SDaniel Cashman help 766d07e2259SDaniel Cashman An arch should select this symbol if it supports running applications 767d07e2259SDaniel Cashman in compatibility mode, supports setting a variable number of bits for 768d07e2259SDaniel Cashman use in establishing the base address for mmap allocations, has MMU 769d07e2259SDaniel Cashman enabled and provides values for both: 770d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MIN 771d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MAX 772d07e2259SDaniel Cashman 773d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MIN 774d07e2259SDaniel Cashman int 775d07e2259SDaniel Cashman 776d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MAX 777d07e2259SDaniel Cashman int 778d07e2259SDaniel Cashman 779d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 780d07e2259SDaniel Cashman int 781d07e2259SDaniel Cashman 782d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS 783d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 784d07e2259SDaniel Cashman range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 785d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 786d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_MIN 787d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 788d07e2259SDaniel Cashman help 789d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 790d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 791d07e2259SDaniel Cashman resulting from mmap allocations for compatible applications This 792d07e2259SDaniel Cashman value will be bounded by the architecture's minimum and maximum 793d07e2259SDaniel Cashman supported values. 794d07e2259SDaniel Cashman 795d07e2259SDaniel Cashman This value can be changed after boot using the 796d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_compat_bits tunable 797d07e2259SDaniel Cashman 7981b028f78SDmitry Safonovconfig HAVE_ARCH_COMPAT_MMAP_BASES 7991b028f78SDmitry Safonov bool 8001b028f78SDmitry Safonov help 8011b028f78SDmitry Safonov This allows 64bit applications to invoke 32-bit mmap() syscall 8021b028f78SDmitry Safonov and vice-versa 32-bit applications to call 64-bit mmap(). 8031b028f78SDmitry Safonov Required for applications doing different bitness syscalls. 8041b028f78SDmitry Safonov 8053033f14aSJosh Triplettconfig HAVE_COPY_THREAD_TLS 8063033f14aSJosh Triplett bool 8073033f14aSJosh Triplett help 8083033f14aSJosh Triplett Architecture provides copy_thread_tls to accept tls argument via 8093033f14aSJosh Triplett normal C parameter passing, rather than extracting the syscall 8103033f14aSJosh Triplett argument from pt_regs. 8113033f14aSJosh Triplett 812b9ab5ebbSJosh Poimboeufconfig HAVE_STACK_VALIDATION 813b9ab5ebbSJosh Poimboeuf bool 814b9ab5ebbSJosh Poimboeuf help 815b9ab5ebbSJosh Poimboeuf Architecture supports the 'objtool check' host tool command, which 816b9ab5ebbSJosh Poimboeuf performs compile-time stack metadata validation. 817b9ab5ebbSJosh Poimboeuf 818af085d90SJosh Poimboeufconfig HAVE_RELIABLE_STACKTRACE 819af085d90SJosh Poimboeuf bool 820af085d90SJosh Poimboeuf help 821af085d90SJosh Poimboeuf Architecture has a save_stack_trace_tsk_reliable() function which 822af085d90SJosh Poimboeuf only returns a stack trace if it can guarantee the trace is reliable. 823af085d90SJosh Poimboeuf 824468a9428SGeorge Spelvinconfig HAVE_ARCH_HASH 825468a9428SGeorge Spelvin bool 826468a9428SGeorge Spelvin default n 827468a9428SGeorge Spelvin help 828468a9428SGeorge Spelvin If this is set, the architecture provides an <asm/hash.h> 829468a9428SGeorge Spelvin file which provides platform-specific implementations of some 830468a9428SGeorge Spelvin functions in <linux/hash.h> or fs/namei.c. 831468a9428SGeorge Spelvin 8323a495511SWilliam Breathitt Grayconfig ISA_BUS_API 8333a495511SWilliam Breathitt Gray def_bool ISA 8343a495511SWilliam Breathitt Gray 835d2125043SAl Viro# 836d2125043SAl Viro# ABI hall of shame 837d2125043SAl Viro# 838d2125043SAl Viroconfig CLONE_BACKWARDS 839d2125043SAl Viro bool 840d2125043SAl Viro help 841d2125043SAl Viro Architecture has tls passed as the 4th argument of clone(2), 842d2125043SAl Viro not the 5th one. 843d2125043SAl Viro 844d2125043SAl Viroconfig CLONE_BACKWARDS2 845d2125043SAl Viro bool 846d2125043SAl Viro help 847d2125043SAl Viro Architecture has the first two arguments of clone(2) swapped. 848d2125043SAl Viro 849dfa9771aSMichal Simekconfig CLONE_BACKWARDS3 850dfa9771aSMichal Simek bool 851dfa9771aSMichal Simek help 852dfa9771aSMichal Simek Architecture has tls passed as the 3rd argument of clone(2), 853dfa9771aSMichal Simek not the 5th one. 854dfa9771aSMichal Simek 855eaca6eaeSAl Viroconfig ODD_RT_SIGACTION 856eaca6eaeSAl Viro bool 857eaca6eaeSAl Viro help 858eaca6eaeSAl Viro Architecture has unusual rt_sigaction(2) arguments 859eaca6eaeSAl Viro 8600a0e8cdfSAl Viroconfig OLD_SIGSUSPEND 8610a0e8cdfSAl Viro bool 8620a0e8cdfSAl Viro help 8630a0e8cdfSAl Viro Architecture has old sigsuspend(2) syscall, of one-argument variety 8640a0e8cdfSAl Viro 8650a0e8cdfSAl Viroconfig OLD_SIGSUSPEND3 8660a0e8cdfSAl Viro bool 8670a0e8cdfSAl Viro help 8680a0e8cdfSAl Viro Even weirder antique ABI - three-argument sigsuspend(2) 8690a0e8cdfSAl Viro 870495dfbf7SAl Viroconfig OLD_SIGACTION 871495dfbf7SAl Viro bool 872495dfbf7SAl Viro help 873495dfbf7SAl Viro Architecture has old sigaction(2) syscall. Nope, not the same 874495dfbf7SAl Viro as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 875495dfbf7SAl Viro but fairly different variant of sigaction(2), thanks to OSF/1 876495dfbf7SAl Viro compatibility... 877495dfbf7SAl Viro 878495dfbf7SAl Viroconfig COMPAT_OLD_SIGACTION 879495dfbf7SAl Viro bool 880495dfbf7SAl Viro 881d4703ddaSDeepa Dinamaniconfig 64BIT_TIME 882d4703ddaSDeepa Dinamani def_bool ARCH_HAS_64BIT_TIME 883d4703ddaSDeepa Dinamani help 884d4703ddaSDeepa Dinamani This should be selected by all architectures that need to support 885d4703ddaSDeepa Dinamani new system calls with a 64-bit time_t. This is relevant on all 32-bit 886d4703ddaSDeepa Dinamani architectures, and 64-bit architectures as part of compat syscall 887d4703ddaSDeepa Dinamani handling. 888d4703ddaSDeepa Dinamani 88917435e5fSDeepa Dinamaniconfig COMPAT_32BIT_TIME 89017435e5fSDeepa Dinamani def_bool (!64BIT && 64BIT_TIME) || COMPAT 89117435e5fSDeepa Dinamani help 89217435e5fSDeepa Dinamani This enables 32 bit time_t support in addition to 64 bit time_t support. 89317435e5fSDeepa Dinamani This is relevant on all 32-bit architectures, and 64-bit architectures 89417435e5fSDeepa Dinamani as part of compat syscall handling. 89517435e5fSDeepa Dinamani 8960d4a619bSChristoph Hellwigconfig ARCH_NO_COHERENT_DMA_MMAP 8970d4a619bSChristoph Hellwig bool 8980d4a619bSChristoph Hellwig 899fff7fb0bSZhaoxiu Zengconfig CPU_NO_EFFICIENT_FFS 900fff7fb0bSZhaoxiu Zeng def_bool n 901fff7fb0bSZhaoxiu Zeng 902ba14a194SAndy Lutomirskiconfig HAVE_ARCH_VMAP_STACK 903ba14a194SAndy Lutomirski def_bool n 904ba14a194SAndy Lutomirski help 905ba14a194SAndy Lutomirski An arch should select this symbol if it can support kernel stacks 906ba14a194SAndy Lutomirski in vmalloc space. This means: 907ba14a194SAndy Lutomirski 908ba14a194SAndy Lutomirski - vmalloc space must be large enough to hold many kernel stacks. 909ba14a194SAndy Lutomirski This may rule out many 32-bit architectures. 910ba14a194SAndy Lutomirski 911ba14a194SAndy Lutomirski - Stacks in vmalloc space need to work reliably. For example, if 912ba14a194SAndy Lutomirski vmap page tables are created on demand, either this mechanism 913ba14a194SAndy Lutomirski needs to work while the stack points to a virtual address with 914ba14a194SAndy Lutomirski unpopulated page tables or arch code (switch_to() and switch_mm(), 915ba14a194SAndy Lutomirski most likely) needs to ensure that the stack's page table entries 916ba14a194SAndy Lutomirski are populated before running on a possibly unpopulated stack. 917ba14a194SAndy Lutomirski 918ba14a194SAndy Lutomirski - If the stack overflows into a guard page, something reasonable 919ba14a194SAndy Lutomirski should happen. The definition of "reasonable" is flexible, but 920ba14a194SAndy Lutomirski instantly rebooting without logging anything would be unfriendly. 921ba14a194SAndy Lutomirski 922ba14a194SAndy Lutomirskiconfig VMAP_STACK 923ba14a194SAndy Lutomirski default y 924ba14a194SAndy Lutomirski bool "Use a virtually-mapped stack" 925ba14a194SAndy Lutomirski depends on HAVE_ARCH_VMAP_STACK && !KASAN 926ba14a194SAndy Lutomirski ---help--- 927ba14a194SAndy Lutomirski Enable this if you want the use virtually-mapped kernel stacks 928ba14a194SAndy Lutomirski with guard pages. This causes kernel stack overflows to be 929ba14a194SAndy Lutomirski caught immediately rather than causing difficult-to-diagnose 930ba14a194SAndy Lutomirski corruption. 931ba14a194SAndy Lutomirski 932ba14a194SAndy Lutomirski This is presently incompatible with KASAN because KASAN expects 933ba14a194SAndy Lutomirski the stack to map directly to the KASAN shadow map using a formula 934ba14a194SAndy Lutomirski that is incorrect if the stack is in vmalloc space. 935ba14a194SAndy Lutomirski 936ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX 937ad21fc4fSLaura Abbott def_bool n 938ad21fc4fSLaura Abbott 939ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 940ad21fc4fSLaura Abbott def_bool n 941ad21fc4fSLaura Abbott 942ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_KERNEL_RWX 943ad21fc4fSLaura Abbott def_bool n 944ad21fc4fSLaura Abbott 9450f5bf6d0SLaura Abbottconfig STRICT_KERNEL_RWX 946ad21fc4fSLaura Abbott bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 947ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_KERNEL_RWX 948ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 949ad21fc4fSLaura Abbott help 950ad21fc4fSLaura Abbott If this is set, kernel text and rodata memory will be made read-only, 951ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 952ad21fc4fSLaura Abbott protection against certain security exploits (e.g. executing the heap 953ad21fc4fSLaura Abbott or modifying text) 954ad21fc4fSLaura Abbott 955ad21fc4fSLaura Abbott These features are considered standard security practice these days. 956ad21fc4fSLaura Abbott You should say Y here in almost all cases. 957ad21fc4fSLaura Abbott 958ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_MODULE_RWX 959ad21fc4fSLaura Abbott def_bool n 960ad21fc4fSLaura Abbott 9610f5bf6d0SLaura Abbottconfig STRICT_MODULE_RWX 962ad21fc4fSLaura Abbott bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 963ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 964ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 965ad21fc4fSLaura Abbott help 966ad21fc4fSLaura Abbott If this is set, module text and rodata memory will be made read-only, 967ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 968ad21fc4fSLaura Abbott protection against certain security exploits (e.g. writing to text) 969ad21fc4fSLaura Abbott 970ea8c64acSChristoph Hellwig# select if the architecture provides an asm/dma-direct.h header 971ea8c64acSChristoph Hellwigconfig ARCH_HAS_PHYS_TO_DMA 972ea8c64acSChristoph Hellwig bool 973ea8c64acSChristoph Hellwig 9747a46ec0eSKees Cookconfig ARCH_HAS_REFCOUNT 9757a46ec0eSKees Cook bool 9767a46ec0eSKees Cook help 9777a46ec0eSKees Cook An architecture selects this when it has implemented refcount_t 9787a46ec0eSKees Cook using open coded assembly primitives that provide an optimized 9797a46ec0eSKees Cook refcount_t implementation, possibly at the expense of some full 9807a46ec0eSKees Cook refcount state checks of CONFIG_REFCOUNT_FULL=y. 9817a46ec0eSKees Cook 9827a46ec0eSKees Cook The refcount overflow check behavior, however, must be retained. 9837a46ec0eSKees Cook Catching overflows is the primary security concern for protecting 9847a46ec0eSKees Cook against bugs in reference counts. 9857a46ec0eSKees Cook 986fd25d19fSKees Cookconfig REFCOUNT_FULL 987fd25d19fSKees Cook bool "Perform full reference count validation at the expense of speed" 988fd25d19fSKees Cook help 989fd25d19fSKees Cook Enabling this switches the refcounting infrastructure from a fast 990fd25d19fSKees Cook unchecked atomic_t implementation to a fully state checked 991fd25d19fSKees Cook implementation, which can be (slightly) slower but provides protections 992fd25d19fSKees Cook against various use-after-free conditions that can be used in 993fd25d19fSKees Cook security flaw exploits. 994fd25d19fSKees Cook 9952521f2c2SPeter Oberparleitersource "kernel/gcov/Kconfig" 996