1fb32e03fSMathieu Desnoyers# 2fb32e03fSMathieu Desnoyers# General architecture dependent options 3fb32e03fSMathieu Desnoyers# 4125e5645SMathieu Desnoyers 5692f66f2SHari Bathiniconfig CRASH_CORE 6692f66f2SHari Bathini bool 7692f66f2SHari Bathini 82965faa5SDave Youngconfig KEXEC_CORE 9692f66f2SHari Bathini select CRASH_CORE 102965faa5SDave Young bool 112965faa5SDave Young 12467d2782SThiago Jung Bauermannconfig HAVE_IMA_KEXEC 13467d2782SThiago Jung Bauermann bool 14467d2782SThiago Jung Bauermann 15125e5645SMathieu Desnoyersconfig OPROFILE 16b309a294SRobert Richter tristate "OProfile system profiling" 17125e5645SMathieu Desnoyers depends on PROFILING 18125e5645SMathieu Desnoyers depends on HAVE_OPROFILE 19d69d59f4SIngo Molnar select RING_BUFFER 209a5963ebSChristian Borntraeger select RING_BUFFER_ALLOW_SWAP 21125e5645SMathieu Desnoyers help 22125e5645SMathieu Desnoyers OProfile is a profiling system capable of profiling the 23125e5645SMathieu Desnoyers whole system, include the kernel, kernel modules, libraries, 24125e5645SMathieu Desnoyers and applications. 25125e5645SMathieu Desnoyers 26125e5645SMathieu Desnoyers If unsure, say N. 27125e5645SMathieu Desnoyers 284d4036e0SJason Yehconfig OPROFILE_EVENT_MULTIPLEX 294d4036e0SJason Yeh bool "OProfile multiplexing support (EXPERIMENTAL)" 304d4036e0SJason Yeh default n 314d4036e0SJason Yeh depends on OPROFILE && X86 324d4036e0SJason Yeh help 334d4036e0SJason Yeh The number of hardware counters is limited. The multiplexing 344d4036e0SJason Yeh feature enables OProfile to gather more events than counters 354d4036e0SJason Yeh are provided by the hardware. This is realized by switching 369332ef9dSMasahiro Yamada between events at a user specified time interval. 374d4036e0SJason Yeh 384d4036e0SJason Yeh If unsure, say N. 394d4036e0SJason Yeh 40125e5645SMathieu Desnoyersconfig HAVE_OPROFILE 419ba16087SJan Beulich bool 42125e5645SMathieu Desnoyers 43dcfce4a0SRobert Richterconfig OPROFILE_NMI_TIMER 44dcfce4a0SRobert Richter def_bool y 45af9feebeSAnton Blanchard depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64 46dcfce4a0SRobert Richter 47125e5645SMathieu Desnoyersconfig KPROBES 48125e5645SMathieu Desnoyers bool "Kprobes" 4905ed160eSMasami Hiramatsu depends on MODULES 50125e5645SMathieu Desnoyers depends on HAVE_KPROBES 5105ed160eSMasami Hiramatsu select KALLSYMS 52125e5645SMathieu Desnoyers help 53125e5645SMathieu Desnoyers Kprobes allows you to trap at almost any kernel address and 54125e5645SMathieu Desnoyers execute a callback function. register_kprobe() establishes 55125e5645SMathieu Desnoyers a probepoint and specifies the callback. Kprobes is useful 56125e5645SMathieu Desnoyers for kernel debugging, non-intrusive instrumentation and testing. 57125e5645SMathieu Desnoyers If in doubt, say "N". 58125e5645SMathieu Desnoyers 5945f81b1cSSteven Rostedtconfig JUMP_LABEL 60c5905afbSIngo Molnar bool "Optimize very unlikely/likely branches" 6145f81b1cSSteven Rostedt depends on HAVE_ARCH_JUMP_LABEL 6245f81b1cSSteven Rostedt help 63c5905afbSIngo Molnar This option enables a transparent branch optimization that 64c5905afbSIngo Molnar makes certain almost-always-true or almost-always-false branch 65c5905afbSIngo Molnar conditions even cheaper to execute within the kernel. 6645f81b1cSSteven Rostedt 67c5905afbSIngo Molnar Certain performance-sensitive kernel code, such as trace points, 68c5905afbSIngo Molnar scheduler functionality, networking code and KVM have such 69c5905afbSIngo Molnar branches and include support for this optimization technique. 70c5905afbSIngo Molnar 71c5905afbSIngo Molnar If it is detected that the compiler has support for "asm goto", 72c5905afbSIngo Molnar the kernel will compile such branches with just a nop 73c5905afbSIngo Molnar instruction. When the condition flag is toggled to true, the 74c5905afbSIngo Molnar nop will be converted to a jump instruction to execute the 75c5905afbSIngo Molnar conditional block of instructions. 76c5905afbSIngo Molnar 77c5905afbSIngo Molnar This technique lowers overhead and stress on the branch prediction 78c5905afbSIngo Molnar of the processor and generally makes the kernel faster. The update 79c5905afbSIngo Molnar of the condition is slower, but those are always very rare. 80c5905afbSIngo Molnar 81c5905afbSIngo Molnar ( On 32-bit x86, the necessary options added to the compiler 82c5905afbSIngo Molnar flags may increase the size of the kernel slightly. ) 8345f81b1cSSteven Rostedt 841987c947SPeter Zijlstraconfig STATIC_KEYS_SELFTEST 851987c947SPeter Zijlstra bool "Static key selftest" 861987c947SPeter Zijlstra depends on JUMP_LABEL 871987c947SPeter Zijlstra help 881987c947SPeter Zijlstra Boot time self-test of the branch patching code. 891987c947SPeter Zijlstra 90afd66255SMasami Hiramatsuconfig OPTPROBES 915cc718b9SMasami Hiramatsu def_bool y 925cc718b9SMasami Hiramatsu depends on KPROBES && HAVE_OPTPROBES 93afd66255SMasami Hiramatsu depends on !PREEMPT 94afd66255SMasami Hiramatsu 95e7dbfe34SMasami Hiramatsuconfig KPROBES_ON_FTRACE 96e7dbfe34SMasami Hiramatsu def_bool y 97e7dbfe34SMasami Hiramatsu depends on KPROBES && HAVE_KPROBES_ON_FTRACE 98e7dbfe34SMasami Hiramatsu depends on DYNAMIC_FTRACE_WITH_REGS 99e7dbfe34SMasami Hiramatsu help 100e7dbfe34SMasami Hiramatsu If function tracer is enabled and the arch supports full 101e7dbfe34SMasami Hiramatsu passing of pt_regs to function tracing, then kprobes can 102e7dbfe34SMasami Hiramatsu optimize on top of function tracing. 103e7dbfe34SMasami Hiramatsu 1042b144498SSrikar Dronamrajuconfig UPROBES 10509294e31SDavid A. Long def_bool n 106e8f4aa60SAllen Pais depends on ARCH_SUPPORTS_UPROBES 1072b144498SSrikar Dronamraju help 1087b2d81d4SIngo Molnar Uprobes is the user-space counterpart to kprobes: they 1097b2d81d4SIngo Molnar enable instrumentation applications (such as 'perf probe') 1107b2d81d4SIngo Molnar to establish unintrusive probes in user-space binaries and 1117b2d81d4SIngo Molnar libraries, by executing handler functions when the probes 1127b2d81d4SIngo Molnar are hit by user-space applications. 1137b2d81d4SIngo Molnar 1147b2d81d4SIngo Molnar ( These probes come in the form of single-byte breakpoints, 1157b2d81d4SIngo Molnar managed by the kernel and kept transparent to the probed 1167b2d81d4SIngo Molnar application. ) 1172b144498SSrikar Dronamraju 118c19fa94aSJames Hoganconfig HAVE_64BIT_ALIGNED_ACCESS 119c19fa94aSJames Hogan def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 120c19fa94aSJames Hogan help 121c19fa94aSJames Hogan Some architectures require 64 bit accesses to be 64 bit 122c19fa94aSJames Hogan aligned, which also requires structs containing 64 bit values 123c19fa94aSJames Hogan to be 64 bit aligned too. This includes some 32 bit 124c19fa94aSJames Hogan architectures which can do 64 bit accesses, as well as 64 bit 125c19fa94aSJames Hogan architectures without unaligned access. 126c19fa94aSJames Hogan 127c19fa94aSJames Hogan This symbol should be selected by an architecture if 64 bit 128c19fa94aSJames Hogan accesses are required to be 64 bit aligned in this way even 129c19fa94aSJames Hogan though it is not a 64 bit architecture. 130c19fa94aSJames Hogan 131c19fa94aSJames Hogan See Documentation/unaligned-memory-access.txt for more 132c19fa94aSJames Hogan information on the topic of unaligned memory accesses. 133c19fa94aSJames Hogan 13458340a07SJohannes Bergconfig HAVE_EFFICIENT_UNALIGNED_ACCESS 1359ba16087SJan Beulich bool 13658340a07SJohannes Berg help 13758340a07SJohannes Berg Some architectures are unable to perform unaligned accesses 13858340a07SJohannes Berg without the use of get_unaligned/put_unaligned. Others are 13958340a07SJohannes Berg unable to perform such accesses efficiently (e.g. trap on 14058340a07SJohannes Berg unaligned access and require fixing it up in the exception 14158340a07SJohannes Berg handler.) 14258340a07SJohannes Berg 14358340a07SJohannes Berg This symbol should be selected by an architecture if it can 14458340a07SJohannes Berg perform unaligned accesses efficiently to allow different 14558340a07SJohannes Berg code paths to be selected for these cases. Some network 14658340a07SJohannes Berg drivers, for example, could opt to not fix up alignment 14758340a07SJohannes Berg problems with received packets if doing so would not help 14858340a07SJohannes Berg much. 14958340a07SJohannes Berg 15058340a07SJohannes Berg See Documentation/unaligned-memory-access.txt for more 15158340a07SJohannes Berg information on the topic of unaligned memory accesses. 15258340a07SJohannes Berg 153cf66bb93SDavid Woodhouseconfig ARCH_USE_BUILTIN_BSWAP 154cf66bb93SDavid Woodhouse bool 155cf66bb93SDavid Woodhouse help 156cf66bb93SDavid Woodhouse Modern versions of GCC (since 4.4) have builtin functions 157cf66bb93SDavid Woodhouse for handling byte-swapping. Using these, instead of the old 158cf66bb93SDavid Woodhouse inline assembler that the architecture code provides in the 159cf66bb93SDavid Woodhouse __arch_bswapXX() macros, allows the compiler to see what's 160cf66bb93SDavid Woodhouse happening and offers more opportunity for optimisation. In 161cf66bb93SDavid Woodhouse particular, the compiler will be able to combine the byteswap 162cf66bb93SDavid Woodhouse with a nearby load or store and use load-and-swap or 163cf66bb93SDavid Woodhouse store-and-swap instructions if the architecture has them. It 164cf66bb93SDavid Woodhouse should almost *never* result in code which is worse than the 165cf66bb93SDavid Woodhouse hand-coded assembler in <asm/swab.h>. But just in case it 166cf66bb93SDavid Woodhouse does, the use of the builtins is optional. 167cf66bb93SDavid Woodhouse 168cf66bb93SDavid Woodhouse Any architecture with load-and-swap or store-and-swap 169cf66bb93SDavid Woodhouse instructions should set this. And it shouldn't hurt to set it 170cf66bb93SDavid Woodhouse on architectures that don't have such instructions. 171cf66bb93SDavid Woodhouse 1729edddaa2SAnanth N Mavinakayanahalliconfig KRETPROBES 1739edddaa2SAnanth N Mavinakayanahalli def_bool y 1749edddaa2SAnanth N Mavinakayanahalli depends on KPROBES && HAVE_KRETPROBES 1759edddaa2SAnanth N Mavinakayanahalli 1767c68af6eSAvi Kivityconfig USER_RETURN_NOTIFIER 1777c68af6eSAvi Kivity bool 1787c68af6eSAvi Kivity depends on HAVE_USER_RETURN_NOTIFIER 1797c68af6eSAvi Kivity help 1807c68af6eSAvi Kivity Provide a kernel-internal notification when a cpu is about to 1817c68af6eSAvi Kivity switch to user mode. 1827c68af6eSAvi Kivity 18328b2ee20SRik van Rielconfig HAVE_IOREMAP_PROT 1849ba16087SJan Beulich bool 18528b2ee20SRik van Riel 186125e5645SMathieu Desnoyersconfig HAVE_KPROBES 1879ba16087SJan Beulich bool 1889edddaa2SAnanth N Mavinakayanahalli 1899edddaa2SAnanth N Mavinakayanahalliconfig HAVE_KRETPROBES 1909ba16087SJan Beulich bool 19174bc7ceeSArthur Kepner 192afd66255SMasami Hiramatsuconfig HAVE_OPTPROBES 193afd66255SMasami Hiramatsu bool 194d314d74cSCong Wang 195e7dbfe34SMasami Hiramatsuconfig HAVE_KPROBES_ON_FTRACE 196e7dbfe34SMasami Hiramatsu bool 197e7dbfe34SMasami Hiramatsu 19842a0bb3fSPetr Mladekconfig HAVE_NMI 19942a0bb3fSPetr Mladek bool 20042a0bb3fSPetr Mladek 2011f5a4ad9SRoland McGrath# 2021f5a4ad9SRoland McGrath# An arch should select this if it provides all these things: 2031f5a4ad9SRoland McGrath# 2041f5a4ad9SRoland McGrath# task_pt_regs() in asm/processor.h or asm/ptrace.h 2051f5a4ad9SRoland McGrath# arch_has_single_step() if there is hardware single-step support 2061f5a4ad9SRoland McGrath# arch_has_block_step() if there is hardware block-step support 2071f5a4ad9SRoland McGrath# asm/syscall.h supplying asm-generic/syscall.h interface 2081f5a4ad9SRoland McGrath# linux/regset.h user_regset interfaces 2091f5a4ad9SRoland McGrath# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 2101f5a4ad9SRoland McGrath# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 2111f5a4ad9SRoland McGrath# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 2121f5a4ad9SRoland McGrath# signal delivery calls tracehook_signal_handler() 2131f5a4ad9SRoland McGrath# 2141f5a4ad9SRoland McGrathconfig HAVE_ARCH_TRACEHOOK 2159ba16087SJan Beulich bool 2161f5a4ad9SRoland McGrath 217c64be2bbSMarek Szyprowskiconfig HAVE_DMA_CONTIGUOUS 218c64be2bbSMarek Szyprowski bool 219c64be2bbSMarek Szyprowski 22029d5e047SThomas Gleixnerconfig GENERIC_SMP_IDLE_THREAD 22129d5e047SThomas Gleixner bool 22229d5e047SThomas Gleixner 223485cf5daSKevin Hilmanconfig GENERIC_IDLE_POLL_SETUP 224485cf5daSKevin Hilman bool 225485cf5daSKevin Hilman 2266974f0c4SDaniel Micayconfig ARCH_HAS_FORTIFY_SOURCE 2276974f0c4SDaniel Micay bool 2286974f0c4SDaniel Micay help 2296974f0c4SDaniel Micay An architecture should select this when it can successfully 2306974f0c4SDaniel Micay build and run with CONFIG_FORTIFY_SOURCE. 2316974f0c4SDaniel Micay 232d2852a22SDaniel Borkmann# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 233d2852a22SDaniel Borkmannconfig ARCH_HAS_SET_MEMORY 234d2852a22SDaniel Borkmann bool 235d2852a22SDaniel Borkmann 236a6359d1eSThomas Gleixner# Select if arch init_task initializer is different to init/init_task.c 237a6359d1eSThomas Gleixnerconfig ARCH_INIT_TASK 238a4a2eb49SThomas Gleixner bool 239a4a2eb49SThomas Gleixner 240f5e10287SThomas Gleixner# Select if arch has its private alloc_task_struct() function 241f5e10287SThomas Gleixnerconfig ARCH_TASK_STRUCT_ALLOCATOR 242f5e10287SThomas Gleixner bool 243f5e10287SThomas Gleixner 244b235beeaSLinus Torvalds# Select if arch has its private alloc_thread_stack() function 245b235beeaSLinus Torvaldsconfig ARCH_THREAD_STACK_ALLOCATOR 246f5e10287SThomas Gleixner bool 247f5e10287SThomas Gleixner 2485aaeb5c0SIngo Molnar# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 2495aaeb5c0SIngo Molnarconfig ARCH_WANTS_DYNAMIC_TASK_STRUCT 2505aaeb5c0SIngo Molnar bool 2515aaeb5c0SIngo Molnar 252f850c30cSHeiko Carstensconfig HAVE_REGS_AND_STACK_ACCESS_API 253f850c30cSHeiko Carstens bool 254e01292b1SHeiko Carstens help 255e01292b1SHeiko Carstens This symbol should be selected by an architecure if it supports 256e01292b1SHeiko Carstens the API needed to access registers and stack entries from pt_regs, 257e01292b1SHeiko Carstens declared in asm/ptrace.h 258e01292b1SHeiko Carstens For example the kprobes-based event tracer needs this API. 259f850c30cSHeiko Carstens 2609483a578SDavid Brownellconfig HAVE_CLK 2619ba16087SJan Beulich bool 2629483a578SDavid Brownell help 2639483a578SDavid Brownell The <linux/clk.h> calls support software clock gating and 2649483a578SDavid Brownell thus are a key power management tool on many systems. 2659483a578SDavid Brownell 2665ee00bd4SJoerg Roedelconfig HAVE_DMA_API_DEBUG 2675ee00bd4SJoerg Roedel bool 26836cd3c9fSHeiko Carstens 26962a038d3SK.Prasadconfig HAVE_HW_BREAKPOINT 27062a038d3SK.Prasad bool 27199e8c5a3SFrederic Weisbecker depends on PERF_EVENTS 27262a038d3SK.Prasad 2730102752eSFrederic Weisbeckerconfig HAVE_MIXED_BREAKPOINTS_REGS 2740102752eSFrederic Weisbecker bool 2750102752eSFrederic Weisbecker depends on HAVE_HW_BREAKPOINT 2760102752eSFrederic Weisbecker help 2770102752eSFrederic Weisbecker Depending on the arch implementation of hardware breakpoints, 2780102752eSFrederic Weisbecker some of them have separate registers for data and instruction 2790102752eSFrederic Weisbecker breakpoints addresses, others have mixed registers to store 2800102752eSFrederic Weisbecker them but define the access type in a control register. 2810102752eSFrederic Weisbecker Select this option if your arch implements breakpoints under the 2820102752eSFrederic Weisbecker latter fashion. 2830102752eSFrederic Weisbecker 2847c68af6eSAvi Kivityconfig HAVE_USER_RETURN_NOTIFIER 2857c68af6eSAvi Kivity bool 286a1922ed6SIngo Molnar 287c01d4323SFrederic Weisbeckerconfig HAVE_PERF_EVENTS_NMI 288c01d4323SFrederic Weisbecker bool 28923637d47SFrederic Weisbecker help 29023637d47SFrederic Weisbecker System hardware can generate an NMI using the perf event 29123637d47SFrederic Weisbecker subsystem. Also has support for calculating CPU cycle events 29223637d47SFrederic Weisbecker to determine how many clock cycles in a given period. 293c01d4323SFrederic Weisbecker 29405a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_PERF 29505a4a952SNicholas Piggin bool 29605a4a952SNicholas Piggin depends on HAVE_PERF_EVENTS_NMI 29705a4a952SNicholas Piggin help 29805a4a952SNicholas Piggin The arch chooses to use the generic perf-NMI-based hardlockup 29905a4a952SNicholas Piggin detector. Must define HAVE_PERF_EVENTS_NMI. 30005a4a952SNicholas Piggin 30105a4a952SNicholas Pigginconfig HAVE_NMI_WATCHDOG 30205a4a952SNicholas Piggin depends on HAVE_NMI 30305a4a952SNicholas Piggin bool 30405a4a952SNicholas Piggin help 30505a4a952SNicholas Piggin The arch provides a low level NMI watchdog. It provides 30605a4a952SNicholas Piggin asm/nmi.h, and defines its own arch_touch_nmi_watchdog(). 30705a4a952SNicholas Piggin 30805a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_ARCH 30905a4a952SNicholas Piggin bool 31005a4a952SNicholas Piggin select HAVE_NMI_WATCHDOG 31105a4a952SNicholas Piggin help 31205a4a952SNicholas Piggin The arch chooses to provide its own hardlockup detector, which is 31305a4a952SNicholas Piggin a superset of the HAVE_NMI_WATCHDOG. It also conforms to config 31405a4a952SNicholas Piggin interfaces and parameters provided by hardlockup detector subsystem. 31505a4a952SNicholas Piggin 316c5e63197SJiri Olsaconfig HAVE_PERF_REGS 317c5e63197SJiri Olsa bool 318c5e63197SJiri Olsa help 319c5e63197SJiri Olsa Support selective register dumps for perf events. This includes 320c5e63197SJiri Olsa bit-mapping of each registers and a unique architecture id. 321c5e63197SJiri Olsa 322c5ebcedbSJiri Olsaconfig HAVE_PERF_USER_STACK_DUMP 323c5ebcedbSJiri Olsa bool 324c5ebcedbSJiri Olsa help 325c5ebcedbSJiri Olsa Support user stack dumps for perf event samples. This needs 326c5ebcedbSJiri Olsa access to the user stack pointer which is not unified across 327c5ebcedbSJiri Olsa architectures. 328c5ebcedbSJiri Olsa 329bf5438fcSJason Baronconfig HAVE_ARCH_JUMP_LABEL 330bf5438fcSJason Baron bool 331bf5438fcSJason Baron 33226723911SPeter Zijlstraconfig HAVE_RCU_TABLE_FREE 33326723911SPeter Zijlstra bool 33426723911SPeter Zijlstra 335df013ffbSHuang Yingconfig ARCH_HAVE_NMI_SAFE_CMPXCHG 336df013ffbSHuang Ying bool 337df013ffbSHuang Ying 33843570fd2SHeiko Carstensconfig HAVE_ALIGNED_STRUCT_PAGE 33943570fd2SHeiko Carstens bool 34043570fd2SHeiko Carstens help 34143570fd2SHeiko Carstens This makes sure that struct pages are double word aligned and that 34243570fd2SHeiko Carstens e.g. the SLUB allocator can perform double word atomic operations 34343570fd2SHeiko Carstens on a struct page for better performance. However selecting this 34443570fd2SHeiko Carstens might increase the size of a struct page by a word. 34543570fd2SHeiko Carstens 3464156153cSHeiko Carstensconfig HAVE_CMPXCHG_LOCAL 3474156153cSHeiko Carstens bool 3484156153cSHeiko Carstens 3492565409fSHeiko Carstensconfig HAVE_CMPXCHG_DOUBLE 3502565409fSHeiko Carstens bool 3512565409fSHeiko Carstens 35277e58496SPaul E. McKenneyconfig ARCH_WEAK_RELEASE_ACQUIRE 35377e58496SPaul E. McKenney bool 35477e58496SPaul E. McKenney 355c1d7e01dSWill Deaconconfig ARCH_WANT_IPC_PARSE_VERSION 356c1d7e01dSWill Deacon bool 357c1d7e01dSWill Deacon 358c1d7e01dSWill Deaconconfig ARCH_WANT_COMPAT_IPC_PARSE_VERSION 359c1d7e01dSWill Deacon bool 360c1d7e01dSWill Deacon 36148b25c43SChris Metcalfconfig ARCH_WANT_OLD_COMPAT_IPC 362c1d7e01dSWill Deacon select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 36348b25c43SChris Metcalf bool 36448b25c43SChris Metcalf 365e2cfabdfSWill Drewryconfig HAVE_ARCH_SECCOMP_FILTER 366e2cfabdfSWill Drewry bool 367e2cfabdfSWill Drewry help 368fb0fadf9SWill Drewry An arch should select this symbol if it provides all of these things: 369bb6ea430SWill Drewry - syscall_get_arch() 370bb6ea430SWill Drewry - syscall_get_arguments() 371bb6ea430SWill Drewry - syscall_rollback() 372bb6ea430SWill Drewry - syscall_set_return_value() 373fb0fadf9SWill Drewry - SIGSYS siginfo_t support 374fb0fadf9SWill Drewry - secure_computing is called from a ptrace_event()-safe context 375fb0fadf9SWill Drewry - secure_computing return value is checked and a return value of -1 376fb0fadf9SWill Drewry results in the system call being skipped immediately. 37748dc92b9SKees Cook - seccomp syscall wired up 378e2cfabdfSWill Drewry 379e2cfabdfSWill Drewryconfig SECCOMP_FILTER 380e2cfabdfSWill Drewry def_bool y 381e2cfabdfSWill Drewry depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 382e2cfabdfSWill Drewry help 383e2cfabdfSWill Drewry Enable tasks to build secure computing environments defined 384e2cfabdfSWill Drewry in terms of Berkeley Packet Filter programs which implement 385e2cfabdfSWill Drewry task-defined system call filtering polices. 386e2cfabdfSWill Drewry 387e2cfabdfSWill Drewry See Documentation/prctl/seccomp_filter.txt for details. 388e2cfabdfSWill Drewry 3896b90bd4bSEmese Revfyconfig HAVE_GCC_PLUGINS 3906b90bd4bSEmese Revfy bool 3916b90bd4bSEmese Revfy help 3926b90bd4bSEmese Revfy An arch should select this symbol if it supports building with 3936b90bd4bSEmese Revfy GCC plugins. 3946b90bd4bSEmese Revfy 3956b90bd4bSEmese Revfymenuconfig GCC_PLUGINS 3966b90bd4bSEmese Revfy bool "GCC plugins" 3976b90bd4bSEmese Revfy depends on HAVE_GCC_PLUGINS 398a519167eSKees Cook depends on !COMPILE_TEST 3996b90bd4bSEmese Revfy help 4006b90bd4bSEmese Revfy GCC plugins are loadable modules that provide extra features to the 4016b90bd4bSEmese Revfy compiler. They are useful for runtime instrumentation and static analysis. 4026b90bd4bSEmese Revfy 4036b90bd4bSEmese Revfy See Documentation/gcc-plugins.txt for details. 4046b90bd4bSEmese Revfy 4050dae776cSEmese Revfyconfig GCC_PLUGIN_CYC_COMPLEXITY 406215e2aa6SKees Cook bool "Compute the cyclomatic complexity of a function" if EXPERT 4070dae776cSEmese Revfy depends on GCC_PLUGINS 408215e2aa6SKees Cook depends on !COMPILE_TEST 4090dae776cSEmese Revfy help 4100dae776cSEmese Revfy The complexity M of a function's control flow graph is defined as: 4110dae776cSEmese Revfy M = E - N + 2P 4120dae776cSEmese Revfy where 4130dae776cSEmese Revfy 4140dae776cSEmese Revfy E = the number of edges 4150dae776cSEmese Revfy N = the number of nodes 4160dae776cSEmese Revfy P = the number of connected components (exit nodes). 4170dae776cSEmese Revfy 418215e2aa6SKees Cook Enabling this plugin reports the complexity to stderr during the 419215e2aa6SKees Cook build. It mainly serves as a simple example of how to create a 420215e2aa6SKees Cook gcc plugin for the kernel. 421215e2aa6SKees Cook 422543c37cbSEmese Revfyconfig GCC_PLUGIN_SANCOV 423543c37cbSEmese Revfy bool 424543c37cbSEmese Revfy depends on GCC_PLUGINS 425543c37cbSEmese Revfy help 426543c37cbSEmese Revfy This plugin inserts a __sanitizer_cov_trace_pc() call at the start of 427543c37cbSEmese Revfy basic blocks. It supports all gcc versions with plugin support (from 428543c37cbSEmese Revfy gcc-4.5 on). It is based on the commit "Add fuzzing coverage support" 429543c37cbSEmese Revfy by Dmitry Vyukov <dvyukov@google.com>. 430543c37cbSEmese Revfy 43138addce8SEmese Revfyconfig GCC_PLUGIN_LATENT_ENTROPY 43238addce8SEmese Revfy bool "Generate some entropy during boot and runtime" 43338addce8SEmese Revfy depends on GCC_PLUGINS 43438addce8SEmese Revfy help 43538addce8SEmese Revfy By saying Y here the kernel will instrument some kernel code to 43638addce8SEmese Revfy extract some entropy from both original and artificially created 43738addce8SEmese Revfy program state. This will help especially embedded systems where 43838addce8SEmese Revfy there is little 'natural' source of entropy normally. The cost 43938addce8SEmese Revfy is some slowdown of the boot process (about 0.5%) and fork and 44038addce8SEmese Revfy irq processing. 44138addce8SEmese Revfy 44238addce8SEmese Revfy Note that entropy extracted this way is not cryptographically 44338addce8SEmese Revfy secure! 44438addce8SEmese Revfy 44538addce8SEmese Revfy This plugin was ported from grsecurity/PaX. More information at: 44638addce8SEmese Revfy * https://grsecurity.net/ 44738addce8SEmese Revfy * https://pax.grsecurity.net/ 44838addce8SEmese Revfy 449c61f13eaSKees Cookconfig GCC_PLUGIN_STRUCTLEAK 450c61f13eaSKees Cook bool "Force initialization of variables containing userspace addresses" 451c61f13eaSKees Cook depends on GCC_PLUGINS 452c61f13eaSKees Cook help 453f136e090SJean Delvare This plugin zero-initializes any structures containing a 454c61f13eaSKees Cook __user attribute. This can prevent some classes of information 455c61f13eaSKees Cook exposures. 456c61f13eaSKees Cook 457c61f13eaSKees Cook This plugin was ported from grsecurity/PaX. More information at: 458c61f13eaSKees Cook * https://grsecurity.net/ 459c61f13eaSKees Cook * https://pax.grsecurity.net/ 460c61f13eaSKees Cook 461c61f13eaSKees Cookconfig GCC_PLUGIN_STRUCTLEAK_VERBOSE 462c61f13eaSKees Cook bool "Report forcefully initialized variables" 463c61f13eaSKees Cook depends on GCC_PLUGIN_STRUCTLEAK 464c61f13eaSKees Cook depends on !COMPILE_TEST 465c61f13eaSKees Cook help 466c61f13eaSKees Cook This option will cause a warning to be printed each time the 467c61f13eaSKees Cook structleak plugin finds a variable it thinks needs to be 468c61f13eaSKees Cook initialized. Since not all existing initializers are detected 469c61f13eaSKees Cook by the plugin, this can produce false positive warnings. 470c61f13eaSKees Cook 471313dd1b6SKees Cookconfig GCC_PLUGIN_RANDSTRUCT 472313dd1b6SKees Cook bool "Randomize layout of sensitive kernel structures" 473313dd1b6SKees Cook depends on GCC_PLUGINS 474313dd1b6SKees Cook select MODVERSIONS if MODULES 475313dd1b6SKees Cook help 4769225331bSKees Cook If you say Y here, the layouts of structures that are entirely 4779225331bSKees Cook function pointers (and have not been manually annotated with 4789225331bSKees Cook __no_randomize_layout), or structures that have been explicitly 4799225331bSKees Cook marked with __randomize_layout, will be randomized at compile-time. 4809225331bSKees Cook This can introduce the requirement of an additional information 4819225331bSKees Cook exposure vulnerability for exploits targeting these structure 4829225331bSKees Cook types. 483313dd1b6SKees Cook 484313dd1b6SKees Cook Enabling this feature will introduce some performance impact, 485313dd1b6SKees Cook slightly increase memory usage, and prevent the use of forensic 486313dd1b6SKees Cook tools like Volatility against the system (unless the kernel 487313dd1b6SKees Cook source tree isn't cleaned after kernel installation). 488313dd1b6SKees Cook 489313dd1b6SKees Cook The seed used for compilation is located at 490313dd1b6SKees Cook scripts/gcc-plgins/randomize_layout_seed.h. It remains after 491313dd1b6SKees Cook a make clean to allow for external modules to be compiled with 492313dd1b6SKees Cook the existing seed and will be removed by a make mrproper or 493313dd1b6SKees Cook make distclean. 494313dd1b6SKees Cook 495313dd1b6SKees Cook Note that the implementation requires gcc 4.7 or newer. 496313dd1b6SKees Cook 497313dd1b6SKees Cook This plugin was ported from grsecurity/PaX. More information at: 498313dd1b6SKees Cook * https://grsecurity.net/ 499313dd1b6SKees Cook * https://pax.grsecurity.net/ 500313dd1b6SKees Cook 501313dd1b6SKees Cookconfig GCC_PLUGIN_RANDSTRUCT_PERFORMANCE 502313dd1b6SKees Cook bool "Use cacheline-aware structure randomization" 503313dd1b6SKees Cook depends on GCC_PLUGIN_RANDSTRUCT 504313dd1b6SKees Cook depends on !COMPILE_TEST 505313dd1b6SKees Cook help 506313dd1b6SKees Cook If you say Y here, the RANDSTRUCT randomization will make a 507313dd1b6SKees Cook best effort at restricting randomization to cacheline-sized 508313dd1b6SKees Cook groups of elements. It will further not randomize bitfields 509313dd1b6SKees Cook in structures. This reduces the performance hit of RANDSTRUCT 510313dd1b6SKees Cook at the cost of weakened randomization. 511313dd1b6SKees Cook 51219952a92SKees Cookconfig HAVE_CC_STACKPROTECTOR 51319952a92SKees Cook bool 51419952a92SKees Cook help 51519952a92SKees Cook An arch should select this symbol if: 51619952a92SKees Cook - its compiler supports the -fstack-protector option 51719952a92SKees Cook - it has implemented a stack canary (e.g. __stack_chk_guard) 51819952a92SKees Cook 51919952a92SKees Cookconfig CC_STACKPROTECTOR 5208779657dSKees Cook def_bool n 52119952a92SKees Cook help 5228779657dSKees Cook Set when a stack-protector mode is enabled, so that the build 5238779657dSKees Cook can enable kernel-side support for the GCC feature. 5248779657dSKees Cook 5258779657dSKees Cookchoice 5268779657dSKees Cook prompt "Stack Protector buffer overflow detection" 5278779657dSKees Cook depends on HAVE_CC_STACKPROTECTOR 5288779657dSKees Cook default CC_STACKPROTECTOR_NONE 5298779657dSKees Cook help 5308779657dSKees Cook This option turns on the "stack-protector" GCC feature. This 53119952a92SKees Cook feature puts, at the beginning of functions, a canary value on 53219952a92SKees Cook the stack just before the return address, and validates 53319952a92SKees Cook the value just before actually returning. Stack based buffer 53419952a92SKees Cook overflows (that need to overwrite this return address) now also 53519952a92SKees Cook overwrite the canary, which gets detected and the attack is then 53619952a92SKees Cook neutralized via a kernel panic. 53719952a92SKees Cook 5388779657dSKees Cookconfig CC_STACKPROTECTOR_NONE 5398779657dSKees Cook bool "None" 5408779657dSKees Cook help 5418779657dSKees Cook Disable "stack-protector" GCC feature. 5428779657dSKees Cook 5438779657dSKees Cookconfig CC_STACKPROTECTOR_REGULAR 5448779657dSKees Cook bool "Regular" 5458779657dSKees Cook select CC_STACKPROTECTOR 5468779657dSKees Cook help 5478779657dSKees Cook Functions will have the stack-protector canary logic added if they 5488779657dSKees Cook have an 8-byte or larger character array on the stack. 5498779657dSKees Cook 55019952a92SKees Cook This feature requires gcc version 4.2 or above, or a distribution 5518779657dSKees Cook gcc with the feature backported ("-fstack-protector"). 5528779657dSKees Cook 5538779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 5548779657dSKees Cook about 3% of all kernel functions, which increases kernel code size 5558779657dSKees Cook by about 0.3%. 5568779657dSKees Cook 5578779657dSKees Cookconfig CC_STACKPROTECTOR_STRONG 5588779657dSKees Cook bool "Strong" 5598779657dSKees Cook select CC_STACKPROTECTOR 5608779657dSKees Cook help 5618779657dSKees Cook Functions will have the stack-protector canary logic added in any 5628779657dSKees Cook of the following conditions: 5638779657dSKees Cook 5648779657dSKees Cook - local variable's address used as part of the right hand side of an 5658779657dSKees Cook assignment or function argument 5668779657dSKees Cook - local variable is an array (or union containing an array), 5678779657dSKees Cook regardless of array type or length 5688779657dSKees Cook - uses register local variables 5698779657dSKees Cook 5708779657dSKees Cook This feature requires gcc version 4.9 or above, or a distribution 5718779657dSKees Cook gcc with the feature backported ("-fstack-protector-strong"). 5728779657dSKees Cook 5738779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 5748779657dSKees Cook about 20% of all kernel functions, which increases the kernel code 5758779657dSKees Cook size by about 2%. 5768779657dSKees Cook 5778779657dSKees Cookendchoice 57819952a92SKees Cook 579a5967db9SStephen Rothwellconfig THIN_ARCHIVES 580799c4341SNicholas Piggin def_bool y 581a5967db9SStephen Rothwell help 582a5967db9SStephen Rothwell Select this if the architecture wants to use thin archives 583a5967db9SStephen Rothwell instead of ld -r to create the built-in.o files. 584a5967db9SStephen Rothwell 585b67067f1SNicholas Pigginconfig LD_DEAD_CODE_DATA_ELIMINATION 586b67067f1SNicholas Piggin bool 587b67067f1SNicholas Piggin help 588b67067f1SNicholas Piggin Select this if the architecture wants to do dead code and 589b67067f1SNicholas Piggin data elimination with the linker by compiling with 590b67067f1SNicholas Piggin -ffunction-sections -fdata-sections and linking with 591b67067f1SNicholas Piggin --gc-sections. 592b67067f1SNicholas Piggin 593b67067f1SNicholas Piggin This requires that the arch annotates or otherwise protects 594b67067f1SNicholas Piggin its external entry points from being discarded. Linker scripts 595b67067f1SNicholas Piggin must also merge .text.*, .data.*, and .bss.* correctly into 5960f4c4af0SNicholas Piggin output sections. Care must be taken not to pull in unrelated 5970f4c4af0SNicholas Piggin sections (e.g., '.text.init'). Typically '.' in section names 5980f4c4af0SNicholas Piggin is used to distinguish them from label names / C identifiers. 599b67067f1SNicholas Piggin 6000f60a8efSKees Cookconfig HAVE_ARCH_WITHIN_STACK_FRAMES 6010f60a8efSKees Cook bool 6020f60a8efSKees Cook help 6030f60a8efSKees Cook An architecture should select this if it can walk the kernel stack 6040f60a8efSKees Cook frames to determine if an object is part of either the arguments 6050f60a8efSKees Cook or local variables (i.e. that it excludes saved return addresses, 6060f60a8efSKees Cook and similar) by implementing an inline arch_within_stack_frames(), 6070f60a8efSKees Cook which is used by CONFIG_HARDENED_USERCOPY. 6080f60a8efSKees Cook 60991d1aa43SFrederic Weisbeckerconfig HAVE_CONTEXT_TRACKING 6102b1d5024SFrederic Weisbecker bool 6112b1d5024SFrederic Weisbecker help 61291d1aa43SFrederic Weisbecker Provide kernel/user boundaries probes necessary for subsystems 61391d1aa43SFrederic Weisbecker that need it, such as userspace RCU extended quiescent state. 61491d1aa43SFrederic Weisbecker Syscalls need to be wrapped inside user_exit()-user_enter() through 61591d1aa43SFrederic Weisbecker the slow path using TIF_NOHZ flag. Exceptions handlers must be 61691d1aa43SFrederic Weisbecker wrapped as well. Irqs are already protected inside 61791d1aa43SFrederic Weisbecker rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on 61891d1aa43SFrederic Weisbecker irq exit still need to be protected. 6192b1d5024SFrederic Weisbecker 620b952741cSFrederic Weisbeckerconfig HAVE_VIRT_CPU_ACCOUNTING 621b952741cSFrederic Weisbecker bool 622b952741cSFrederic Weisbecker 62340565b5aSStanislaw Gruszkaconfig ARCH_HAS_SCALED_CPUTIME 62440565b5aSStanislaw Gruszka bool 62540565b5aSStanislaw Gruszka 626554b0004SKevin Hilmanconfig HAVE_VIRT_CPU_ACCOUNTING_GEN 627554b0004SKevin Hilman bool 628554b0004SKevin Hilman default y if 64BIT 629554b0004SKevin Hilman help 630554b0004SKevin Hilman With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 631554b0004SKevin Hilman Before enabling this option, arch code must be audited 632554b0004SKevin Hilman to ensure there are no races in concurrent read/write of 633554b0004SKevin Hilman cputime_t. For example, reading/writing 64-bit cputime_t on 634554b0004SKevin Hilman some 32-bit arches may require multiple accesses, so proper 635554b0004SKevin Hilman locking is needed to protect against concurrent accesses. 636554b0004SKevin Hilman 637554b0004SKevin Hilman 638fdf9c356SFrederic Weisbeckerconfig HAVE_IRQ_TIME_ACCOUNTING 639fdf9c356SFrederic Weisbecker bool 640fdf9c356SFrederic Weisbecker help 641fdf9c356SFrederic Weisbecker Archs need to ensure they use a high enough resolution clock to 642fdf9c356SFrederic Weisbecker support irq time accounting and then call enable_sched_clock_irqtime(). 643fdf9c356SFrederic Weisbecker 64415626062SGerald Schaeferconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE 64515626062SGerald Schaefer bool 64615626062SGerald Schaefer 647a00cc7d9SMatthew Wilcoxconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 648a00cc7d9SMatthew Wilcox bool 649a00cc7d9SMatthew Wilcox 6500ddab1d2SToshi Kaniconfig HAVE_ARCH_HUGE_VMAP 6510ddab1d2SToshi Kani bool 6520ddab1d2SToshi Kani 6530f8975ecSPavel Emelyanovconfig HAVE_ARCH_SOFT_DIRTY 6540f8975ecSPavel Emelyanov bool 6550f8975ecSPavel Emelyanov 656786d35d4SDavid Howellsconfig HAVE_MOD_ARCH_SPECIFIC 657786d35d4SDavid Howells bool 658786d35d4SDavid Howells help 659786d35d4SDavid Howells The arch uses struct mod_arch_specific to store data. Many arches 660786d35d4SDavid Howells just need a simple module loader without arch specific data - those 661786d35d4SDavid Howells should not enable this. 662786d35d4SDavid Howells 663786d35d4SDavid Howellsconfig MODULES_USE_ELF_RELA 664786d35d4SDavid Howells bool 665786d35d4SDavid Howells help 666786d35d4SDavid Howells Modules only use ELF RELA relocations. Modules with ELF REL 667786d35d4SDavid Howells relocations will give an error. 668786d35d4SDavid Howells 669786d35d4SDavid Howellsconfig MODULES_USE_ELF_REL 670786d35d4SDavid Howells bool 671786d35d4SDavid Howells help 672786d35d4SDavid Howells Modules only use ELF REL relocations. Modules with ELF RELA 673786d35d4SDavid Howells relocations will give an error. 674786d35d4SDavid Howells 675b92021b0SRusty Russellconfig HAVE_UNDERSCORE_SYMBOL_PREFIX 676b92021b0SRusty Russell bool 677b92021b0SRusty Russell help 678b92021b0SRusty Russell Some architectures generate an _ in front of C symbols; things like 679b92021b0SRusty Russell module loading and assembly files need to know about this. 680b92021b0SRusty Russell 681cc1f0274SFrederic Weisbeckerconfig HAVE_IRQ_EXIT_ON_IRQ_STACK 682cc1f0274SFrederic Weisbecker bool 683cc1f0274SFrederic Weisbecker help 684cc1f0274SFrederic Weisbecker Architecture doesn't only execute the irq handler on the irq stack 685cc1f0274SFrederic Weisbecker but also irq_exit(). This way we can process softirqs on this irq 686cc1f0274SFrederic Weisbecker stack instead of switching to a new one when we call __do_softirq() 687cc1f0274SFrederic Weisbecker in the end of an hardirq. 688cc1f0274SFrederic Weisbecker This spares a stack switch and improves cache usage on softirq 689cc1f0274SFrederic Weisbecker processing. 690cc1f0274SFrederic Weisbecker 691235a8f02SKirill A. Shutemovconfig PGTABLE_LEVELS 692235a8f02SKirill A. Shutemov int 693235a8f02SKirill A. Shutemov default 2 694235a8f02SKirill A. Shutemov 6952b68f6caSKees Cookconfig ARCH_HAS_ELF_RANDOMIZE 6962b68f6caSKees Cook bool 6972b68f6caSKees Cook help 6982b68f6caSKees Cook An architecture supports choosing randomized locations for 6992b68f6caSKees Cook stack, mmap, brk, and ET_DYN. Defined functions: 7002b68f6caSKees Cook - arch_mmap_rnd() 701204db6edSKees Cook - arch_randomize_brk() 7022b68f6caSKees Cook 703d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_BITS 704d07e2259SDaniel Cashman bool 705d07e2259SDaniel Cashman help 706d07e2259SDaniel Cashman An arch should select this symbol if it supports setting a variable 707d07e2259SDaniel Cashman number of bits for use in establishing the base address for mmap 708d07e2259SDaniel Cashman allocations, has MMU enabled and provides values for both: 709d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MIN 710d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MAX 711d07e2259SDaniel Cashman 7125f56a5dfSJiri Slabyconfig HAVE_EXIT_THREAD 7135f56a5dfSJiri Slaby bool 7145f56a5dfSJiri Slaby help 7155f56a5dfSJiri Slaby An architecture implements exit_thread. 7165f56a5dfSJiri Slaby 717d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MIN 718d07e2259SDaniel Cashman int 719d07e2259SDaniel Cashman 720d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MAX 721d07e2259SDaniel Cashman int 722d07e2259SDaniel Cashman 723d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_DEFAULT 724d07e2259SDaniel Cashman int 725d07e2259SDaniel Cashman 726d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS 727d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address" if EXPERT 728d07e2259SDaniel Cashman range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 729d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 730d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_MIN 731d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_BITS 732d07e2259SDaniel Cashman help 733d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 734d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 735d07e2259SDaniel Cashman resulting from mmap allocations. This value will be bounded 736d07e2259SDaniel Cashman by the architecture's minimum and maximum supported values. 737d07e2259SDaniel Cashman 738d07e2259SDaniel Cashman This value can be changed after boot using the 739d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_bits tunable 740d07e2259SDaniel Cashman 741d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_COMPAT_BITS 742d07e2259SDaniel Cashman bool 743d07e2259SDaniel Cashman help 744d07e2259SDaniel Cashman An arch should select this symbol if it supports running applications 745d07e2259SDaniel Cashman in compatibility mode, supports setting a variable number of bits for 746d07e2259SDaniel Cashman use in establishing the base address for mmap allocations, has MMU 747d07e2259SDaniel Cashman enabled and provides values for both: 748d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MIN 749d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MAX 750d07e2259SDaniel Cashman 751d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MIN 752d07e2259SDaniel Cashman int 753d07e2259SDaniel Cashman 754d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MAX 755d07e2259SDaniel Cashman int 756d07e2259SDaniel Cashman 757d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 758d07e2259SDaniel Cashman int 759d07e2259SDaniel Cashman 760d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS 761d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 762d07e2259SDaniel Cashman range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 763d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 764d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_MIN 765d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 766d07e2259SDaniel Cashman help 767d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 768d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 769d07e2259SDaniel Cashman resulting from mmap allocations for compatible applications This 770d07e2259SDaniel Cashman value will be bounded by the architecture's minimum and maximum 771d07e2259SDaniel Cashman supported values. 772d07e2259SDaniel Cashman 773d07e2259SDaniel Cashman This value can be changed after boot using the 774d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_compat_bits tunable 775d07e2259SDaniel Cashman 7761b028f78SDmitry Safonovconfig HAVE_ARCH_COMPAT_MMAP_BASES 7771b028f78SDmitry Safonov bool 7781b028f78SDmitry Safonov help 7791b028f78SDmitry Safonov This allows 64bit applications to invoke 32-bit mmap() syscall 7801b028f78SDmitry Safonov and vice-versa 32-bit applications to call 64-bit mmap(). 7811b028f78SDmitry Safonov Required for applications doing different bitness syscalls. 7821b028f78SDmitry Safonov 7833033f14aSJosh Triplettconfig HAVE_COPY_THREAD_TLS 7843033f14aSJosh Triplett bool 7853033f14aSJosh Triplett help 7863033f14aSJosh Triplett Architecture provides copy_thread_tls to accept tls argument via 7873033f14aSJosh Triplett normal C parameter passing, rather than extracting the syscall 7883033f14aSJosh Triplett argument from pt_regs. 7893033f14aSJosh Triplett 790b9ab5ebbSJosh Poimboeufconfig HAVE_STACK_VALIDATION 791b9ab5ebbSJosh Poimboeuf bool 792b9ab5ebbSJosh Poimboeuf help 793b9ab5ebbSJosh Poimboeuf Architecture supports the 'objtool check' host tool command, which 794b9ab5ebbSJosh Poimboeuf performs compile-time stack metadata validation. 795b9ab5ebbSJosh Poimboeuf 796af085d90SJosh Poimboeufconfig HAVE_RELIABLE_STACKTRACE 797af085d90SJosh Poimboeuf bool 798af085d90SJosh Poimboeuf help 799af085d90SJosh Poimboeuf Architecture has a save_stack_trace_tsk_reliable() function which 800af085d90SJosh Poimboeuf only returns a stack trace if it can guarantee the trace is reliable. 801af085d90SJosh Poimboeuf 802468a9428SGeorge Spelvinconfig HAVE_ARCH_HASH 803468a9428SGeorge Spelvin bool 804468a9428SGeorge Spelvin default n 805468a9428SGeorge Spelvin help 806468a9428SGeorge Spelvin If this is set, the architecture provides an <asm/hash.h> 807468a9428SGeorge Spelvin file which provides platform-specific implementations of some 808468a9428SGeorge Spelvin functions in <linux/hash.h> or fs/namei.c. 809468a9428SGeorge Spelvin 8103a495511SWilliam Breathitt Grayconfig ISA_BUS_API 8113a495511SWilliam Breathitt Gray def_bool ISA 8123a495511SWilliam Breathitt Gray 813d2125043SAl Viro# 814d2125043SAl Viro# ABI hall of shame 815d2125043SAl Viro# 816d2125043SAl Viroconfig CLONE_BACKWARDS 817d2125043SAl Viro bool 818d2125043SAl Viro help 819d2125043SAl Viro Architecture has tls passed as the 4th argument of clone(2), 820d2125043SAl Viro not the 5th one. 821d2125043SAl Viro 822d2125043SAl Viroconfig CLONE_BACKWARDS2 823d2125043SAl Viro bool 824d2125043SAl Viro help 825d2125043SAl Viro Architecture has the first two arguments of clone(2) swapped. 826d2125043SAl Viro 827dfa9771aSMichal Simekconfig CLONE_BACKWARDS3 828dfa9771aSMichal Simek bool 829dfa9771aSMichal Simek help 830dfa9771aSMichal Simek Architecture has tls passed as the 3rd argument of clone(2), 831dfa9771aSMichal Simek not the 5th one. 832dfa9771aSMichal Simek 833eaca6eaeSAl Viroconfig ODD_RT_SIGACTION 834eaca6eaeSAl Viro bool 835eaca6eaeSAl Viro help 836eaca6eaeSAl Viro Architecture has unusual rt_sigaction(2) arguments 837eaca6eaeSAl Viro 8380a0e8cdfSAl Viroconfig OLD_SIGSUSPEND 8390a0e8cdfSAl Viro bool 8400a0e8cdfSAl Viro help 8410a0e8cdfSAl Viro Architecture has old sigsuspend(2) syscall, of one-argument variety 8420a0e8cdfSAl Viro 8430a0e8cdfSAl Viroconfig OLD_SIGSUSPEND3 8440a0e8cdfSAl Viro bool 8450a0e8cdfSAl Viro help 8460a0e8cdfSAl Viro Even weirder antique ABI - three-argument sigsuspend(2) 8470a0e8cdfSAl Viro 848495dfbf7SAl Viroconfig OLD_SIGACTION 849495dfbf7SAl Viro bool 850495dfbf7SAl Viro help 851495dfbf7SAl Viro Architecture has old sigaction(2) syscall. Nope, not the same 852495dfbf7SAl Viro as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 853495dfbf7SAl Viro but fairly different variant of sigaction(2), thanks to OSF/1 854495dfbf7SAl Viro compatibility... 855495dfbf7SAl Viro 856495dfbf7SAl Viroconfig COMPAT_OLD_SIGACTION 857495dfbf7SAl Viro bool 858495dfbf7SAl Viro 8590d4a619bSChristoph Hellwigconfig ARCH_NO_COHERENT_DMA_MMAP 8600d4a619bSChristoph Hellwig bool 8610d4a619bSChristoph Hellwig 862fff7fb0bSZhaoxiu Zengconfig CPU_NO_EFFICIENT_FFS 863fff7fb0bSZhaoxiu Zeng def_bool n 864fff7fb0bSZhaoxiu Zeng 865ba14a194SAndy Lutomirskiconfig HAVE_ARCH_VMAP_STACK 866ba14a194SAndy Lutomirski def_bool n 867ba14a194SAndy Lutomirski help 868ba14a194SAndy Lutomirski An arch should select this symbol if it can support kernel stacks 869ba14a194SAndy Lutomirski in vmalloc space. This means: 870ba14a194SAndy Lutomirski 871ba14a194SAndy Lutomirski - vmalloc space must be large enough to hold many kernel stacks. 872ba14a194SAndy Lutomirski This may rule out many 32-bit architectures. 873ba14a194SAndy Lutomirski 874ba14a194SAndy Lutomirski - Stacks in vmalloc space need to work reliably. For example, if 875ba14a194SAndy Lutomirski vmap page tables are created on demand, either this mechanism 876ba14a194SAndy Lutomirski needs to work while the stack points to a virtual address with 877ba14a194SAndy Lutomirski unpopulated page tables or arch code (switch_to() and switch_mm(), 878ba14a194SAndy Lutomirski most likely) needs to ensure that the stack's page table entries 879ba14a194SAndy Lutomirski are populated before running on a possibly unpopulated stack. 880ba14a194SAndy Lutomirski 881ba14a194SAndy Lutomirski - If the stack overflows into a guard page, something reasonable 882ba14a194SAndy Lutomirski should happen. The definition of "reasonable" is flexible, but 883ba14a194SAndy Lutomirski instantly rebooting without logging anything would be unfriendly. 884ba14a194SAndy Lutomirski 885ba14a194SAndy Lutomirskiconfig VMAP_STACK 886ba14a194SAndy Lutomirski default y 887ba14a194SAndy Lutomirski bool "Use a virtually-mapped stack" 888ba14a194SAndy Lutomirski depends on HAVE_ARCH_VMAP_STACK && !KASAN 889ba14a194SAndy Lutomirski ---help--- 890ba14a194SAndy Lutomirski Enable this if you want the use virtually-mapped kernel stacks 891ba14a194SAndy Lutomirski with guard pages. This causes kernel stack overflows to be 892ba14a194SAndy Lutomirski caught immediately rather than causing difficult-to-diagnose 893ba14a194SAndy Lutomirski corruption. 894ba14a194SAndy Lutomirski 895ba14a194SAndy Lutomirski This is presently incompatible with KASAN because KASAN expects 896ba14a194SAndy Lutomirski the stack to map directly to the KASAN shadow map using a formula 897ba14a194SAndy Lutomirski that is incorrect if the stack is in vmalloc space. 898ba14a194SAndy Lutomirski 899ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX 900ad21fc4fSLaura Abbott def_bool n 901ad21fc4fSLaura Abbott 902ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 903ad21fc4fSLaura Abbott def_bool n 904ad21fc4fSLaura Abbott 905ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_KERNEL_RWX 906ad21fc4fSLaura Abbott def_bool n 907ad21fc4fSLaura Abbott 9080f5bf6d0SLaura Abbottconfig STRICT_KERNEL_RWX 909ad21fc4fSLaura Abbott bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 910ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_KERNEL_RWX 911ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 912ad21fc4fSLaura Abbott help 913ad21fc4fSLaura Abbott If this is set, kernel text and rodata memory will be made read-only, 914ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 915ad21fc4fSLaura Abbott protection against certain security exploits (e.g. executing the heap 916ad21fc4fSLaura Abbott or modifying text) 917ad21fc4fSLaura Abbott 918ad21fc4fSLaura Abbott These features are considered standard security practice these days. 919ad21fc4fSLaura Abbott You should say Y here in almost all cases. 920ad21fc4fSLaura Abbott 921ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_MODULE_RWX 922ad21fc4fSLaura Abbott def_bool n 923ad21fc4fSLaura Abbott 9240f5bf6d0SLaura Abbottconfig STRICT_MODULE_RWX 925ad21fc4fSLaura Abbott bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 926ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 927ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 928ad21fc4fSLaura Abbott help 929ad21fc4fSLaura Abbott If this is set, module text and rodata memory will be made read-only, 930ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 931ad21fc4fSLaura Abbott protection against certain security exploits (e.g. writing to text) 932ad21fc4fSLaura Abbott 9331a8b6d76SMao Wenanconfig ARCH_WANT_RELAX_ORDER 9341a8b6d76SMao Wenan bool 9351a8b6d76SMao Wenan 936fd25d19fSKees Cookconfig REFCOUNT_FULL 937fd25d19fSKees Cook bool "Perform full reference count validation at the expense of speed" 938fd25d19fSKees Cook help 939fd25d19fSKees Cook Enabling this switches the refcounting infrastructure from a fast 940fd25d19fSKees Cook unchecked atomic_t implementation to a fully state checked 941fd25d19fSKees Cook implementation, which can be (slightly) slower but provides protections 942fd25d19fSKees Cook against various use-after-free conditions that can be used in 943fd25d19fSKees Cook security flaw exploits. 944fd25d19fSKees Cook 9452521f2c2SPeter Oberparleitersource "kernel/gcov/Kconfig" 946