1fb32e03fSMathieu Desnoyers# 2fb32e03fSMathieu Desnoyers# General architecture dependent options 3fb32e03fSMathieu Desnoyers# 4125e5645SMathieu Desnoyers 52965faa5SDave Youngconfig KEXEC_CORE 62965faa5SDave Young bool 72965faa5SDave Young 8467d2782SThiago Jung Bauermannconfig HAVE_IMA_KEXEC 9467d2782SThiago Jung Bauermann bool 10467d2782SThiago Jung Bauermann 11125e5645SMathieu Desnoyersconfig OPROFILE 12b309a294SRobert Richter tristate "OProfile system profiling" 13125e5645SMathieu Desnoyers depends on PROFILING 14125e5645SMathieu Desnoyers depends on HAVE_OPROFILE 15d69d59f4SIngo Molnar select RING_BUFFER 169a5963ebSChristian Borntraeger select RING_BUFFER_ALLOW_SWAP 17125e5645SMathieu Desnoyers help 18125e5645SMathieu Desnoyers OProfile is a profiling system capable of profiling the 19125e5645SMathieu Desnoyers whole system, include the kernel, kernel modules, libraries, 20125e5645SMathieu Desnoyers and applications. 21125e5645SMathieu Desnoyers 22125e5645SMathieu Desnoyers If unsure, say N. 23125e5645SMathieu Desnoyers 244d4036e0SJason Yehconfig OPROFILE_EVENT_MULTIPLEX 254d4036e0SJason Yeh bool "OProfile multiplexing support (EXPERIMENTAL)" 264d4036e0SJason Yeh default n 274d4036e0SJason Yeh depends on OPROFILE && X86 284d4036e0SJason Yeh help 294d4036e0SJason Yeh The number of hardware counters is limited. The multiplexing 304d4036e0SJason Yeh feature enables OProfile to gather more events than counters 314d4036e0SJason Yeh are provided by the hardware. This is realized by switching 324d4036e0SJason Yeh between events at an user specified time interval. 334d4036e0SJason Yeh 344d4036e0SJason Yeh If unsure, say N. 354d4036e0SJason Yeh 36125e5645SMathieu Desnoyersconfig HAVE_OPROFILE 379ba16087SJan Beulich bool 38125e5645SMathieu Desnoyers 39dcfce4a0SRobert Richterconfig OPROFILE_NMI_TIMER 40dcfce4a0SRobert Richter def_bool y 41af9feebeSAnton Blanchard depends on PERF_EVENTS && HAVE_PERF_EVENTS_NMI && !PPC64 42dcfce4a0SRobert Richter 43125e5645SMathieu Desnoyersconfig KPROBES 44125e5645SMathieu Desnoyers bool "Kprobes" 4505ed160eSMasami Hiramatsu depends on MODULES 46125e5645SMathieu Desnoyers depends on HAVE_KPROBES 4705ed160eSMasami Hiramatsu select KALLSYMS 48125e5645SMathieu Desnoyers help 49125e5645SMathieu Desnoyers Kprobes allows you to trap at almost any kernel address and 50125e5645SMathieu Desnoyers execute a callback function. register_kprobe() establishes 51125e5645SMathieu Desnoyers a probepoint and specifies the callback. Kprobes is useful 52125e5645SMathieu Desnoyers for kernel debugging, non-intrusive instrumentation and testing. 53125e5645SMathieu Desnoyers If in doubt, say "N". 54125e5645SMathieu Desnoyers 5545f81b1cSSteven Rostedtconfig JUMP_LABEL 56c5905afbSIngo Molnar bool "Optimize very unlikely/likely branches" 5745f81b1cSSteven Rostedt depends on HAVE_ARCH_JUMP_LABEL 5845f81b1cSSteven Rostedt help 59c5905afbSIngo Molnar This option enables a transparent branch optimization that 60c5905afbSIngo Molnar makes certain almost-always-true or almost-always-false branch 61c5905afbSIngo Molnar conditions even cheaper to execute within the kernel. 6245f81b1cSSteven Rostedt 63c5905afbSIngo Molnar Certain performance-sensitive kernel code, such as trace points, 64c5905afbSIngo Molnar scheduler functionality, networking code and KVM have such 65c5905afbSIngo Molnar branches and include support for this optimization technique. 66c5905afbSIngo Molnar 67c5905afbSIngo Molnar If it is detected that the compiler has support for "asm goto", 68c5905afbSIngo Molnar the kernel will compile such branches with just a nop 69c5905afbSIngo Molnar instruction. When the condition flag is toggled to true, the 70c5905afbSIngo Molnar nop will be converted to a jump instruction to execute the 71c5905afbSIngo Molnar conditional block of instructions. 72c5905afbSIngo Molnar 73c5905afbSIngo Molnar This technique lowers overhead and stress on the branch prediction 74c5905afbSIngo Molnar of the processor and generally makes the kernel faster. The update 75c5905afbSIngo Molnar of the condition is slower, but those are always very rare. 76c5905afbSIngo Molnar 77c5905afbSIngo Molnar ( On 32-bit x86, the necessary options added to the compiler 78c5905afbSIngo Molnar flags may increase the size of the kernel slightly. ) 7945f81b1cSSteven Rostedt 801987c947SPeter Zijlstraconfig STATIC_KEYS_SELFTEST 811987c947SPeter Zijlstra bool "Static key selftest" 821987c947SPeter Zijlstra depends on JUMP_LABEL 831987c947SPeter Zijlstra help 841987c947SPeter Zijlstra Boot time self-test of the branch patching code. 851987c947SPeter Zijlstra 86afd66255SMasami Hiramatsuconfig OPTPROBES 875cc718b9SMasami Hiramatsu def_bool y 885cc718b9SMasami Hiramatsu depends on KPROBES && HAVE_OPTPROBES 89afd66255SMasami Hiramatsu depends on !PREEMPT 90afd66255SMasami Hiramatsu 91e7dbfe34SMasami Hiramatsuconfig KPROBES_ON_FTRACE 92e7dbfe34SMasami Hiramatsu def_bool y 93e7dbfe34SMasami Hiramatsu depends on KPROBES && HAVE_KPROBES_ON_FTRACE 94e7dbfe34SMasami Hiramatsu depends on DYNAMIC_FTRACE_WITH_REGS 95e7dbfe34SMasami Hiramatsu help 96e7dbfe34SMasami Hiramatsu If function tracer is enabled and the arch supports full 97e7dbfe34SMasami Hiramatsu passing of pt_regs to function tracing, then kprobes can 98e7dbfe34SMasami Hiramatsu optimize on top of function tracing. 99e7dbfe34SMasami Hiramatsu 1002b144498SSrikar Dronamrajuconfig UPROBES 10109294e31SDavid A. Long def_bool n 102e8f4aa60SAllen Pais depends on ARCH_SUPPORTS_UPROBES 1032b144498SSrikar Dronamraju help 1047b2d81d4SIngo Molnar Uprobes is the user-space counterpart to kprobes: they 1057b2d81d4SIngo Molnar enable instrumentation applications (such as 'perf probe') 1067b2d81d4SIngo Molnar to establish unintrusive probes in user-space binaries and 1077b2d81d4SIngo Molnar libraries, by executing handler functions when the probes 1087b2d81d4SIngo Molnar are hit by user-space applications. 1097b2d81d4SIngo Molnar 1107b2d81d4SIngo Molnar ( These probes come in the form of single-byte breakpoints, 1117b2d81d4SIngo Molnar managed by the kernel and kept transparent to the probed 1127b2d81d4SIngo Molnar application. ) 1132b144498SSrikar Dronamraju 114c19fa94aSJames Hoganconfig HAVE_64BIT_ALIGNED_ACCESS 115c19fa94aSJames Hogan def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 116c19fa94aSJames Hogan help 117c19fa94aSJames Hogan Some architectures require 64 bit accesses to be 64 bit 118c19fa94aSJames Hogan aligned, which also requires structs containing 64 bit values 119c19fa94aSJames Hogan to be 64 bit aligned too. This includes some 32 bit 120c19fa94aSJames Hogan architectures which can do 64 bit accesses, as well as 64 bit 121c19fa94aSJames Hogan architectures without unaligned access. 122c19fa94aSJames Hogan 123c19fa94aSJames Hogan This symbol should be selected by an architecture if 64 bit 124c19fa94aSJames Hogan accesses are required to be 64 bit aligned in this way even 125c19fa94aSJames Hogan though it is not a 64 bit architecture. 126c19fa94aSJames Hogan 127c19fa94aSJames Hogan See Documentation/unaligned-memory-access.txt for more 128c19fa94aSJames Hogan information on the topic of unaligned memory accesses. 129c19fa94aSJames Hogan 13058340a07SJohannes Bergconfig HAVE_EFFICIENT_UNALIGNED_ACCESS 1319ba16087SJan Beulich bool 13258340a07SJohannes Berg help 13358340a07SJohannes Berg Some architectures are unable to perform unaligned accesses 13458340a07SJohannes Berg without the use of get_unaligned/put_unaligned. Others are 13558340a07SJohannes Berg unable to perform such accesses efficiently (e.g. trap on 13658340a07SJohannes Berg unaligned access and require fixing it up in the exception 13758340a07SJohannes Berg handler.) 13858340a07SJohannes Berg 13958340a07SJohannes Berg This symbol should be selected by an architecture if it can 14058340a07SJohannes Berg perform unaligned accesses efficiently to allow different 14158340a07SJohannes Berg code paths to be selected for these cases. Some network 14258340a07SJohannes Berg drivers, for example, could opt to not fix up alignment 14358340a07SJohannes Berg problems with received packets if doing so would not help 14458340a07SJohannes Berg much. 14558340a07SJohannes Berg 14658340a07SJohannes Berg See Documentation/unaligned-memory-access.txt for more 14758340a07SJohannes Berg information on the topic of unaligned memory accesses. 14858340a07SJohannes Berg 149cf66bb93SDavid Woodhouseconfig ARCH_USE_BUILTIN_BSWAP 150cf66bb93SDavid Woodhouse bool 151cf66bb93SDavid Woodhouse help 152cf66bb93SDavid Woodhouse Modern versions of GCC (since 4.4) have builtin functions 153cf66bb93SDavid Woodhouse for handling byte-swapping. Using these, instead of the old 154cf66bb93SDavid Woodhouse inline assembler that the architecture code provides in the 155cf66bb93SDavid Woodhouse __arch_bswapXX() macros, allows the compiler to see what's 156cf66bb93SDavid Woodhouse happening and offers more opportunity for optimisation. In 157cf66bb93SDavid Woodhouse particular, the compiler will be able to combine the byteswap 158cf66bb93SDavid Woodhouse with a nearby load or store and use load-and-swap or 159cf66bb93SDavid Woodhouse store-and-swap instructions if the architecture has them. It 160cf66bb93SDavid Woodhouse should almost *never* result in code which is worse than the 161cf66bb93SDavid Woodhouse hand-coded assembler in <asm/swab.h>. But just in case it 162cf66bb93SDavid Woodhouse does, the use of the builtins is optional. 163cf66bb93SDavid Woodhouse 164cf66bb93SDavid Woodhouse Any architecture with load-and-swap or store-and-swap 165cf66bb93SDavid Woodhouse instructions should set this. And it shouldn't hurt to set it 166cf66bb93SDavid Woodhouse on architectures that don't have such instructions. 167cf66bb93SDavid Woodhouse 1689edddaa2SAnanth N Mavinakayanahalliconfig KRETPROBES 1699edddaa2SAnanth N Mavinakayanahalli def_bool y 1709edddaa2SAnanth N Mavinakayanahalli depends on KPROBES && HAVE_KRETPROBES 1719edddaa2SAnanth N Mavinakayanahalli 1727c68af6eSAvi Kivityconfig USER_RETURN_NOTIFIER 1737c68af6eSAvi Kivity bool 1747c68af6eSAvi Kivity depends on HAVE_USER_RETURN_NOTIFIER 1757c68af6eSAvi Kivity help 1767c68af6eSAvi Kivity Provide a kernel-internal notification when a cpu is about to 1777c68af6eSAvi Kivity switch to user mode. 1787c68af6eSAvi Kivity 17928b2ee20SRik van Rielconfig HAVE_IOREMAP_PROT 1809ba16087SJan Beulich bool 18128b2ee20SRik van Riel 182125e5645SMathieu Desnoyersconfig HAVE_KPROBES 1839ba16087SJan Beulich bool 1849edddaa2SAnanth N Mavinakayanahalli 1859edddaa2SAnanth N Mavinakayanahalliconfig HAVE_KRETPROBES 1869ba16087SJan Beulich bool 18774bc7ceeSArthur Kepner 188afd66255SMasami Hiramatsuconfig HAVE_OPTPROBES 189afd66255SMasami Hiramatsu bool 190d314d74cSCong Wang 191e7dbfe34SMasami Hiramatsuconfig HAVE_KPROBES_ON_FTRACE 192e7dbfe34SMasami Hiramatsu bool 193e7dbfe34SMasami Hiramatsu 19442a0bb3fSPetr Mladekconfig HAVE_NMI 19542a0bb3fSPetr Mladek bool 19642a0bb3fSPetr Mladek 197d314d74cSCong Wangconfig HAVE_NMI_WATCHDOG 19842a0bb3fSPetr Mladek depends on HAVE_NMI 199d314d74cSCong Wang bool 2001f5a4ad9SRoland McGrath# 2011f5a4ad9SRoland McGrath# An arch should select this if it provides all these things: 2021f5a4ad9SRoland McGrath# 2031f5a4ad9SRoland McGrath# task_pt_regs() in asm/processor.h or asm/ptrace.h 2041f5a4ad9SRoland McGrath# arch_has_single_step() if there is hardware single-step support 2051f5a4ad9SRoland McGrath# arch_has_block_step() if there is hardware block-step support 2061f5a4ad9SRoland McGrath# asm/syscall.h supplying asm-generic/syscall.h interface 2071f5a4ad9SRoland McGrath# linux/regset.h user_regset interfaces 2081f5a4ad9SRoland McGrath# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 2091f5a4ad9SRoland McGrath# TIF_SYSCALL_TRACE calls tracehook_report_syscall_{entry,exit} 2101f5a4ad9SRoland McGrath# TIF_NOTIFY_RESUME calls tracehook_notify_resume() 2111f5a4ad9SRoland McGrath# signal delivery calls tracehook_signal_handler() 2121f5a4ad9SRoland McGrath# 2131f5a4ad9SRoland McGrathconfig HAVE_ARCH_TRACEHOOK 2149ba16087SJan Beulich bool 2151f5a4ad9SRoland McGrath 216c64be2bbSMarek Szyprowskiconfig HAVE_DMA_CONTIGUOUS 217c64be2bbSMarek Szyprowski bool 218c64be2bbSMarek Szyprowski 21929d5e047SThomas Gleixnerconfig GENERIC_SMP_IDLE_THREAD 22029d5e047SThomas Gleixner bool 22129d5e047SThomas Gleixner 222485cf5daSKevin Hilmanconfig GENERIC_IDLE_POLL_SETUP 223485cf5daSKevin Hilman bool 224485cf5daSKevin Hilman 225a6359d1eSThomas Gleixner# Select if arch init_task initializer is different to init/init_task.c 226a6359d1eSThomas Gleixnerconfig ARCH_INIT_TASK 227a4a2eb49SThomas Gleixner bool 228a4a2eb49SThomas Gleixner 229f5e10287SThomas Gleixner# Select if arch has its private alloc_task_struct() function 230f5e10287SThomas Gleixnerconfig ARCH_TASK_STRUCT_ALLOCATOR 231f5e10287SThomas Gleixner bool 232f5e10287SThomas Gleixner 233b235beeaSLinus Torvalds# Select if arch has its private alloc_thread_stack() function 234b235beeaSLinus Torvaldsconfig ARCH_THREAD_STACK_ALLOCATOR 235f5e10287SThomas Gleixner bool 236f5e10287SThomas Gleixner 2375aaeb5c0SIngo Molnar# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 2385aaeb5c0SIngo Molnarconfig ARCH_WANTS_DYNAMIC_TASK_STRUCT 2395aaeb5c0SIngo Molnar bool 2405aaeb5c0SIngo Molnar 241f850c30cSHeiko Carstensconfig HAVE_REGS_AND_STACK_ACCESS_API 242f850c30cSHeiko Carstens bool 243e01292b1SHeiko Carstens help 244e01292b1SHeiko Carstens This symbol should be selected by an architecure if it supports 245e01292b1SHeiko Carstens the API needed to access registers and stack entries from pt_regs, 246e01292b1SHeiko Carstens declared in asm/ptrace.h 247e01292b1SHeiko Carstens For example the kprobes-based event tracer needs this API. 248f850c30cSHeiko Carstens 2499483a578SDavid Brownellconfig HAVE_CLK 2509ba16087SJan Beulich bool 2519483a578SDavid Brownell help 2529483a578SDavid Brownell The <linux/clk.h> calls support software clock gating and 2539483a578SDavid Brownell thus are a key power management tool on many systems. 2549483a578SDavid Brownell 2555ee00bd4SJoerg Roedelconfig HAVE_DMA_API_DEBUG 2565ee00bd4SJoerg Roedel bool 25736cd3c9fSHeiko Carstens 25862a038d3SK.Prasadconfig HAVE_HW_BREAKPOINT 25962a038d3SK.Prasad bool 26099e8c5a3SFrederic Weisbecker depends on PERF_EVENTS 26162a038d3SK.Prasad 2620102752eSFrederic Weisbeckerconfig HAVE_MIXED_BREAKPOINTS_REGS 2630102752eSFrederic Weisbecker bool 2640102752eSFrederic Weisbecker depends on HAVE_HW_BREAKPOINT 2650102752eSFrederic Weisbecker help 2660102752eSFrederic Weisbecker Depending on the arch implementation of hardware breakpoints, 2670102752eSFrederic Weisbecker some of them have separate registers for data and instruction 2680102752eSFrederic Weisbecker breakpoints addresses, others have mixed registers to store 2690102752eSFrederic Weisbecker them but define the access type in a control register. 2700102752eSFrederic Weisbecker Select this option if your arch implements breakpoints under the 2710102752eSFrederic Weisbecker latter fashion. 2720102752eSFrederic Weisbecker 2737c68af6eSAvi Kivityconfig HAVE_USER_RETURN_NOTIFIER 2747c68af6eSAvi Kivity bool 275a1922ed6SIngo Molnar 276c01d4323SFrederic Weisbeckerconfig HAVE_PERF_EVENTS_NMI 277c01d4323SFrederic Weisbecker bool 27823637d47SFrederic Weisbecker help 27923637d47SFrederic Weisbecker System hardware can generate an NMI using the perf event 28023637d47SFrederic Weisbecker subsystem. Also has support for calculating CPU cycle events 28123637d47SFrederic Weisbecker to determine how many clock cycles in a given period. 282c01d4323SFrederic Weisbecker 283c5e63197SJiri Olsaconfig HAVE_PERF_REGS 284c5e63197SJiri Olsa bool 285c5e63197SJiri Olsa help 286c5e63197SJiri Olsa Support selective register dumps for perf events. This includes 287c5e63197SJiri Olsa bit-mapping of each registers and a unique architecture id. 288c5e63197SJiri Olsa 289c5ebcedbSJiri Olsaconfig HAVE_PERF_USER_STACK_DUMP 290c5ebcedbSJiri Olsa bool 291c5ebcedbSJiri Olsa help 292c5ebcedbSJiri Olsa Support user stack dumps for perf event samples. This needs 293c5ebcedbSJiri Olsa access to the user stack pointer which is not unified across 294c5ebcedbSJiri Olsa architectures. 295c5ebcedbSJiri Olsa 296bf5438fcSJason Baronconfig HAVE_ARCH_JUMP_LABEL 297bf5438fcSJason Baron bool 298bf5438fcSJason Baron 29926723911SPeter Zijlstraconfig HAVE_RCU_TABLE_FREE 30026723911SPeter Zijlstra bool 30126723911SPeter Zijlstra 302df013ffbSHuang Yingconfig ARCH_HAVE_NMI_SAFE_CMPXCHG 303df013ffbSHuang Ying bool 304df013ffbSHuang Ying 30543570fd2SHeiko Carstensconfig HAVE_ALIGNED_STRUCT_PAGE 30643570fd2SHeiko Carstens bool 30743570fd2SHeiko Carstens help 30843570fd2SHeiko Carstens This makes sure that struct pages are double word aligned and that 30943570fd2SHeiko Carstens e.g. the SLUB allocator can perform double word atomic operations 31043570fd2SHeiko Carstens on a struct page for better performance. However selecting this 31143570fd2SHeiko Carstens might increase the size of a struct page by a word. 31243570fd2SHeiko Carstens 3134156153cSHeiko Carstensconfig HAVE_CMPXCHG_LOCAL 3144156153cSHeiko Carstens bool 3154156153cSHeiko Carstens 3162565409fSHeiko Carstensconfig HAVE_CMPXCHG_DOUBLE 3172565409fSHeiko Carstens bool 3182565409fSHeiko Carstens 319c1d7e01dSWill Deaconconfig ARCH_WANT_IPC_PARSE_VERSION 320c1d7e01dSWill Deacon bool 321c1d7e01dSWill Deacon 322c1d7e01dSWill Deaconconfig ARCH_WANT_COMPAT_IPC_PARSE_VERSION 323c1d7e01dSWill Deacon bool 324c1d7e01dSWill Deacon 32548b25c43SChris Metcalfconfig ARCH_WANT_OLD_COMPAT_IPC 326c1d7e01dSWill Deacon select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 32748b25c43SChris Metcalf bool 32848b25c43SChris Metcalf 329e2cfabdfSWill Drewryconfig HAVE_ARCH_SECCOMP_FILTER 330e2cfabdfSWill Drewry bool 331e2cfabdfSWill Drewry help 332fb0fadf9SWill Drewry An arch should select this symbol if it provides all of these things: 333bb6ea430SWill Drewry - syscall_get_arch() 334bb6ea430SWill Drewry - syscall_get_arguments() 335bb6ea430SWill Drewry - syscall_rollback() 336bb6ea430SWill Drewry - syscall_set_return_value() 337fb0fadf9SWill Drewry - SIGSYS siginfo_t support 338fb0fadf9SWill Drewry - secure_computing is called from a ptrace_event()-safe context 339fb0fadf9SWill Drewry - secure_computing return value is checked and a return value of -1 340fb0fadf9SWill Drewry results in the system call being skipped immediately. 34148dc92b9SKees Cook - seccomp syscall wired up 342e2cfabdfSWill Drewry 343e2cfabdfSWill Drewryconfig SECCOMP_FILTER 344e2cfabdfSWill Drewry def_bool y 345e2cfabdfSWill Drewry depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 346e2cfabdfSWill Drewry help 347e2cfabdfSWill Drewry Enable tasks to build secure computing environments defined 348e2cfabdfSWill Drewry in terms of Berkeley Packet Filter programs which implement 349e2cfabdfSWill Drewry task-defined system call filtering polices. 350e2cfabdfSWill Drewry 351e2cfabdfSWill Drewry See Documentation/prctl/seccomp_filter.txt for details. 352e2cfabdfSWill Drewry 3536b90bd4bSEmese Revfyconfig HAVE_GCC_PLUGINS 3546b90bd4bSEmese Revfy bool 3556b90bd4bSEmese Revfy help 3566b90bd4bSEmese Revfy An arch should select this symbol if it supports building with 3576b90bd4bSEmese Revfy GCC plugins. 3586b90bd4bSEmese Revfy 3596b90bd4bSEmese Revfymenuconfig GCC_PLUGINS 3606b90bd4bSEmese Revfy bool "GCC plugins" 3616b90bd4bSEmese Revfy depends on HAVE_GCC_PLUGINS 362a519167eSKees Cook depends on !COMPILE_TEST 3636b90bd4bSEmese Revfy help 3646b90bd4bSEmese Revfy GCC plugins are loadable modules that provide extra features to the 3656b90bd4bSEmese Revfy compiler. They are useful for runtime instrumentation and static analysis. 3666b90bd4bSEmese Revfy 3676b90bd4bSEmese Revfy See Documentation/gcc-plugins.txt for details. 3686b90bd4bSEmese Revfy 3690dae776cSEmese Revfyconfig GCC_PLUGIN_CYC_COMPLEXITY 370215e2aa6SKees Cook bool "Compute the cyclomatic complexity of a function" if EXPERT 3710dae776cSEmese Revfy depends on GCC_PLUGINS 372215e2aa6SKees Cook depends on !COMPILE_TEST 3730dae776cSEmese Revfy help 3740dae776cSEmese Revfy The complexity M of a function's control flow graph is defined as: 3750dae776cSEmese Revfy M = E - N + 2P 3760dae776cSEmese Revfy where 3770dae776cSEmese Revfy 3780dae776cSEmese Revfy E = the number of edges 3790dae776cSEmese Revfy N = the number of nodes 3800dae776cSEmese Revfy P = the number of connected components (exit nodes). 3810dae776cSEmese Revfy 382215e2aa6SKees Cook Enabling this plugin reports the complexity to stderr during the 383215e2aa6SKees Cook build. It mainly serves as a simple example of how to create a 384215e2aa6SKees Cook gcc plugin for the kernel. 385215e2aa6SKees Cook 386543c37cbSEmese Revfyconfig GCC_PLUGIN_SANCOV 387543c37cbSEmese Revfy bool 388543c37cbSEmese Revfy depends on GCC_PLUGINS 389543c37cbSEmese Revfy help 390543c37cbSEmese Revfy This plugin inserts a __sanitizer_cov_trace_pc() call at the start of 391543c37cbSEmese Revfy basic blocks. It supports all gcc versions with plugin support (from 392543c37cbSEmese Revfy gcc-4.5 on). It is based on the commit "Add fuzzing coverage support" 393543c37cbSEmese Revfy by Dmitry Vyukov <dvyukov@google.com>. 394543c37cbSEmese Revfy 39538addce8SEmese Revfyconfig GCC_PLUGIN_LATENT_ENTROPY 39638addce8SEmese Revfy bool "Generate some entropy during boot and runtime" 39738addce8SEmese Revfy depends on GCC_PLUGINS 39838addce8SEmese Revfy help 39938addce8SEmese Revfy By saying Y here the kernel will instrument some kernel code to 40038addce8SEmese Revfy extract some entropy from both original and artificially created 40138addce8SEmese Revfy program state. This will help especially embedded systems where 40238addce8SEmese Revfy there is little 'natural' source of entropy normally. The cost 40338addce8SEmese Revfy is some slowdown of the boot process (about 0.5%) and fork and 40438addce8SEmese Revfy irq processing. 40538addce8SEmese Revfy 40638addce8SEmese Revfy Note that entropy extracted this way is not cryptographically 40738addce8SEmese Revfy secure! 40838addce8SEmese Revfy 40938addce8SEmese Revfy This plugin was ported from grsecurity/PaX. More information at: 41038addce8SEmese Revfy * https://grsecurity.net/ 41138addce8SEmese Revfy * https://pax.grsecurity.net/ 41238addce8SEmese Revfy 41319952a92SKees Cookconfig HAVE_CC_STACKPROTECTOR 41419952a92SKees Cook bool 41519952a92SKees Cook help 41619952a92SKees Cook An arch should select this symbol if: 41719952a92SKees Cook - its compiler supports the -fstack-protector option 41819952a92SKees Cook - it has implemented a stack canary (e.g. __stack_chk_guard) 41919952a92SKees Cook 42019952a92SKees Cookconfig CC_STACKPROTECTOR 4218779657dSKees Cook def_bool n 42219952a92SKees Cook help 4238779657dSKees Cook Set when a stack-protector mode is enabled, so that the build 4248779657dSKees Cook can enable kernel-side support for the GCC feature. 4258779657dSKees Cook 4268779657dSKees Cookchoice 4278779657dSKees Cook prompt "Stack Protector buffer overflow detection" 4288779657dSKees Cook depends on HAVE_CC_STACKPROTECTOR 4298779657dSKees Cook default CC_STACKPROTECTOR_NONE 4308779657dSKees Cook help 4318779657dSKees Cook This option turns on the "stack-protector" GCC feature. This 43219952a92SKees Cook feature puts, at the beginning of functions, a canary value on 43319952a92SKees Cook the stack just before the return address, and validates 43419952a92SKees Cook the value just before actually returning. Stack based buffer 43519952a92SKees Cook overflows (that need to overwrite this return address) now also 43619952a92SKees Cook overwrite the canary, which gets detected and the attack is then 43719952a92SKees Cook neutralized via a kernel panic. 43819952a92SKees Cook 4398779657dSKees Cookconfig CC_STACKPROTECTOR_NONE 4408779657dSKees Cook bool "None" 4418779657dSKees Cook help 4428779657dSKees Cook Disable "stack-protector" GCC feature. 4438779657dSKees Cook 4448779657dSKees Cookconfig CC_STACKPROTECTOR_REGULAR 4458779657dSKees Cook bool "Regular" 4468779657dSKees Cook select CC_STACKPROTECTOR 4478779657dSKees Cook help 4488779657dSKees Cook Functions will have the stack-protector canary logic added if they 4498779657dSKees Cook have an 8-byte or larger character array on the stack. 4508779657dSKees Cook 45119952a92SKees Cook This feature requires gcc version 4.2 or above, or a distribution 4528779657dSKees Cook gcc with the feature backported ("-fstack-protector"). 4538779657dSKees Cook 4548779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 4558779657dSKees Cook about 3% of all kernel functions, which increases kernel code size 4568779657dSKees Cook by about 0.3%. 4578779657dSKees Cook 4588779657dSKees Cookconfig CC_STACKPROTECTOR_STRONG 4598779657dSKees Cook bool "Strong" 4608779657dSKees Cook select CC_STACKPROTECTOR 4618779657dSKees Cook help 4628779657dSKees Cook Functions will have the stack-protector canary logic added in any 4638779657dSKees Cook of the following conditions: 4648779657dSKees Cook 4658779657dSKees Cook - local variable's address used as part of the right hand side of an 4668779657dSKees Cook assignment or function argument 4678779657dSKees Cook - local variable is an array (or union containing an array), 4688779657dSKees Cook regardless of array type or length 4698779657dSKees Cook - uses register local variables 4708779657dSKees Cook 4718779657dSKees Cook This feature requires gcc version 4.9 or above, or a distribution 4728779657dSKees Cook gcc with the feature backported ("-fstack-protector-strong"). 4738779657dSKees Cook 4748779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 4758779657dSKees Cook about 20% of all kernel functions, which increases the kernel code 4768779657dSKees Cook size by about 2%. 4778779657dSKees Cook 4788779657dSKees Cookendchoice 47919952a92SKees Cook 480a5967db9SStephen Rothwellconfig THIN_ARCHIVES 481a5967db9SStephen Rothwell bool 482a5967db9SStephen Rothwell help 483a5967db9SStephen Rothwell Select this if the architecture wants to use thin archives 484a5967db9SStephen Rothwell instead of ld -r to create the built-in.o files. 485a5967db9SStephen Rothwell 486b67067f1SNicholas Pigginconfig LD_DEAD_CODE_DATA_ELIMINATION 487b67067f1SNicholas Piggin bool 488b67067f1SNicholas Piggin help 489b67067f1SNicholas Piggin Select this if the architecture wants to do dead code and 490b67067f1SNicholas Piggin data elimination with the linker by compiling with 491b67067f1SNicholas Piggin -ffunction-sections -fdata-sections and linking with 492b67067f1SNicholas Piggin --gc-sections. 493b67067f1SNicholas Piggin 494b67067f1SNicholas Piggin This requires that the arch annotates or otherwise protects 495b67067f1SNicholas Piggin its external entry points from being discarded. Linker scripts 496b67067f1SNicholas Piggin must also merge .text.*, .data.*, and .bss.* correctly into 4970f4c4af0SNicholas Piggin output sections. Care must be taken not to pull in unrelated 4980f4c4af0SNicholas Piggin sections (e.g., '.text.init'). Typically '.' in section names 4990f4c4af0SNicholas Piggin is used to distinguish them from label names / C identifiers. 500b67067f1SNicholas Piggin 5010f60a8efSKees Cookconfig HAVE_ARCH_WITHIN_STACK_FRAMES 5020f60a8efSKees Cook bool 5030f60a8efSKees Cook help 5040f60a8efSKees Cook An architecture should select this if it can walk the kernel stack 5050f60a8efSKees Cook frames to determine if an object is part of either the arguments 5060f60a8efSKees Cook or local variables (i.e. that it excludes saved return addresses, 5070f60a8efSKees Cook and similar) by implementing an inline arch_within_stack_frames(), 5080f60a8efSKees Cook which is used by CONFIG_HARDENED_USERCOPY. 5090f60a8efSKees Cook 51091d1aa43SFrederic Weisbeckerconfig HAVE_CONTEXT_TRACKING 5112b1d5024SFrederic Weisbecker bool 5122b1d5024SFrederic Weisbecker help 51391d1aa43SFrederic Weisbecker Provide kernel/user boundaries probes necessary for subsystems 51491d1aa43SFrederic Weisbecker that need it, such as userspace RCU extended quiescent state. 51591d1aa43SFrederic Weisbecker Syscalls need to be wrapped inside user_exit()-user_enter() through 51691d1aa43SFrederic Weisbecker the slow path using TIF_NOHZ flag. Exceptions handlers must be 51791d1aa43SFrederic Weisbecker wrapped as well. Irqs are already protected inside 51891d1aa43SFrederic Weisbecker rcu_irq_enter/rcu_irq_exit() but preemption or signal handling on 51991d1aa43SFrederic Weisbecker irq exit still need to be protected. 5202b1d5024SFrederic Weisbecker 521b952741cSFrederic Weisbeckerconfig HAVE_VIRT_CPU_ACCOUNTING 522b952741cSFrederic Weisbecker bool 523b952741cSFrederic Weisbecker 52440565b5aSStanislaw Gruszkaconfig ARCH_HAS_SCALED_CPUTIME 52540565b5aSStanislaw Gruszka bool 52640565b5aSStanislaw Gruszka 527554b0004SKevin Hilmanconfig HAVE_VIRT_CPU_ACCOUNTING_GEN 528554b0004SKevin Hilman bool 529554b0004SKevin Hilman default y if 64BIT 530554b0004SKevin Hilman help 531554b0004SKevin Hilman With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 532554b0004SKevin Hilman Before enabling this option, arch code must be audited 533554b0004SKevin Hilman to ensure there are no races in concurrent read/write of 534554b0004SKevin Hilman cputime_t. For example, reading/writing 64-bit cputime_t on 535554b0004SKevin Hilman some 32-bit arches may require multiple accesses, so proper 536554b0004SKevin Hilman locking is needed to protect against concurrent accesses. 537554b0004SKevin Hilman 538554b0004SKevin Hilman 539fdf9c356SFrederic Weisbeckerconfig HAVE_IRQ_TIME_ACCOUNTING 540fdf9c356SFrederic Weisbecker bool 541fdf9c356SFrederic Weisbecker help 542fdf9c356SFrederic Weisbecker Archs need to ensure they use a high enough resolution clock to 543fdf9c356SFrederic Weisbecker support irq time accounting and then call enable_sched_clock_irqtime(). 544fdf9c356SFrederic Weisbecker 54515626062SGerald Schaeferconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE 54615626062SGerald Schaefer bool 54715626062SGerald Schaefer 5480ddab1d2SToshi Kaniconfig HAVE_ARCH_HUGE_VMAP 5490ddab1d2SToshi Kani bool 5500ddab1d2SToshi Kani 5510f8975ecSPavel Emelyanovconfig HAVE_ARCH_SOFT_DIRTY 5520f8975ecSPavel Emelyanov bool 5530f8975ecSPavel Emelyanov 554786d35d4SDavid Howellsconfig HAVE_MOD_ARCH_SPECIFIC 555786d35d4SDavid Howells bool 556786d35d4SDavid Howells help 557786d35d4SDavid Howells The arch uses struct mod_arch_specific to store data. Many arches 558786d35d4SDavid Howells just need a simple module loader without arch specific data - those 559786d35d4SDavid Howells should not enable this. 560786d35d4SDavid Howells 561786d35d4SDavid Howellsconfig MODULES_USE_ELF_RELA 562786d35d4SDavid Howells bool 563786d35d4SDavid Howells help 564786d35d4SDavid Howells Modules only use ELF RELA relocations. Modules with ELF REL 565786d35d4SDavid Howells relocations will give an error. 566786d35d4SDavid Howells 567786d35d4SDavid Howellsconfig MODULES_USE_ELF_REL 568786d35d4SDavid Howells bool 569786d35d4SDavid Howells help 570786d35d4SDavid Howells Modules only use ELF REL relocations. Modules with ELF RELA 571786d35d4SDavid Howells relocations will give an error. 572786d35d4SDavid Howells 573b92021b0SRusty Russellconfig HAVE_UNDERSCORE_SYMBOL_PREFIX 574b92021b0SRusty Russell bool 575b92021b0SRusty Russell help 576b92021b0SRusty Russell Some architectures generate an _ in front of C symbols; things like 577b92021b0SRusty Russell module loading and assembly files need to know about this. 578b92021b0SRusty Russell 579cc1f0274SFrederic Weisbeckerconfig HAVE_IRQ_EXIT_ON_IRQ_STACK 580cc1f0274SFrederic Weisbecker bool 581cc1f0274SFrederic Weisbecker help 582cc1f0274SFrederic Weisbecker Architecture doesn't only execute the irq handler on the irq stack 583cc1f0274SFrederic Weisbecker but also irq_exit(). This way we can process softirqs on this irq 584cc1f0274SFrederic Weisbecker stack instead of switching to a new one when we call __do_softirq() 585cc1f0274SFrederic Weisbecker in the end of an hardirq. 586cc1f0274SFrederic Weisbecker This spares a stack switch and improves cache usage on softirq 587cc1f0274SFrederic Weisbecker processing. 588cc1f0274SFrederic Weisbecker 589235a8f02SKirill A. Shutemovconfig PGTABLE_LEVELS 590235a8f02SKirill A. Shutemov int 591235a8f02SKirill A. Shutemov default 2 592235a8f02SKirill A. Shutemov 5932b68f6caSKees Cookconfig ARCH_HAS_ELF_RANDOMIZE 5942b68f6caSKees Cook bool 5952b68f6caSKees Cook help 5962b68f6caSKees Cook An architecture supports choosing randomized locations for 5972b68f6caSKees Cook stack, mmap, brk, and ET_DYN. Defined functions: 5982b68f6caSKees Cook - arch_mmap_rnd() 599204db6edSKees Cook - arch_randomize_brk() 6002b68f6caSKees Cook 601d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_BITS 602d07e2259SDaniel Cashman bool 603d07e2259SDaniel Cashman help 604d07e2259SDaniel Cashman An arch should select this symbol if it supports setting a variable 605d07e2259SDaniel Cashman number of bits for use in establishing the base address for mmap 606d07e2259SDaniel Cashman allocations, has MMU enabled and provides values for both: 607d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MIN 608d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MAX 609d07e2259SDaniel Cashman 6105f56a5dfSJiri Slabyconfig HAVE_EXIT_THREAD 6115f56a5dfSJiri Slaby bool 6125f56a5dfSJiri Slaby help 6135f56a5dfSJiri Slaby An architecture implements exit_thread. 6145f56a5dfSJiri Slaby 615d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MIN 616d07e2259SDaniel Cashman int 617d07e2259SDaniel Cashman 618d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MAX 619d07e2259SDaniel Cashman int 620d07e2259SDaniel Cashman 621d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_DEFAULT 622d07e2259SDaniel Cashman int 623d07e2259SDaniel Cashman 624d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS 625d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address" if EXPERT 626d07e2259SDaniel Cashman range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 627d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 628d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_MIN 629d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_BITS 630d07e2259SDaniel Cashman help 631d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 632d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 633d07e2259SDaniel Cashman resulting from mmap allocations. This value will be bounded 634d07e2259SDaniel Cashman by the architecture's minimum and maximum supported values. 635d07e2259SDaniel Cashman 636d07e2259SDaniel Cashman This value can be changed after boot using the 637d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_bits tunable 638d07e2259SDaniel Cashman 639d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_COMPAT_BITS 640d07e2259SDaniel Cashman bool 641d07e2259SDaniel Cashman help 642d07e2259SDaniel Cashman An arch should select this symbol if it supports running applications 643d07e2259SDaniel Cashman in compatibility mode, supports setting a variable number of bits for 644d07e2259SDaniel Cashman use in establishing the base address for mmap allocations, has MMU 645d07e2259SDaniel Cashman enabled and provides values for both: 646d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MIN 647d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MAX 648d07e2259SDaniel Cashman 649d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MIN 650d07e2259SDaniel Cashman int 651d07e2259SDaniel Cashman 652d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MAX 653d07e2259SDaniel Cashman int 654d07e2259SDaniel Cashman 655d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 656d07e2259SDaniel Cashman int 657d07e2259SDaniel Cashman 658d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS 659d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 660d07e2259SDaniel Cashman range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 661d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 662d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_MIN 663d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 664d07e2259SDaniel Cashman help 665d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 666d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 667d07e2259SDaniel Cashman resulting from mmap allocations for compatible applications This 668d07e2259SDaniel Cashman value will be bounded by the architecture's minimum and maximum 669d07e2259SDaniel Cashman supported values. 670d07e2259SDaniel Cashman 671d07e2259SDaniel Cashman This value can be changed after boot using the 672d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_compat_bits tunable 673d07e2259SDaniel Cashman 6743033f14aSJosh Triplettconfig HAVE_COPY_THREAD_TLS 6753033f14aSJosh Triplett bool 6763033f14aSJosh Triplett help 6773033f14aSJosh Triplett Architecture provides copy_thread_tls to accept tls argument via 6783033f14aSJosh Triplett normal C parameter passing, rather than extracting the syscall 6793033f14aSJosh Triplett argument from pt_regs. 6803033f14aSJosh Triplett 681b9ab5ebbSJosh Poimboeufconfig HAVE_STACK_VALIDATION 682b9ab5ebbSJosh Poimboeuf bool 683b9ab5ebbSJosh Poimboeuf help 684b9ab5ebbSJosh Poimboeuf Architecture supports the 'objtool check' host tool command, which 685b9ab5ebbSJosh Poimboeuf performs compile-time stack metadata validation. 686b9ab5ebbSJosh Poimboeuf 687468a9428SGeorge Spelvinconfig HAVE_ARCH_HASH 688468a9428SGeorge Spelvin bool 689468a9428SGeorge Spelvin default n 690468a9428SGeorge Spelvin help 691468a9428SGeorge Spelvin If this is set, the architecture provides an <asm/hash.h> 692468a9428SGeorge Spelvin file which provides platform-specific implementations of some 693468a9428SGeorge Spelvin functions in <linux/hash.h> or fs/namei.c. 694468a9428SGeorge Spelvin 6953a495511SWilliam Breathitt Grayconfig ISA_BUS_API 6963a495511SWilliam Breathitt Gray def_bool ISA 6973a495511SWilliam Breathitt Gray 698d2125043SAl Viro# 699d2125043SAl Viro# ABI hall of shame 700d2125043SAl Viro# 701d2125043SAl Viroconfig CLONE_BACKWARDS 702d2125043SAl Viro bool 703d2125043SAl Viro help 704d2125043SAl Viro Architecture has tls passed as the 4th argument of clone(2), 705d2125043SAl Viro not the 5th one. 706d2125043SAl Viro 707d2125043SAl Viroconfig CLONE_BACKWARDS2 708d2125043SAl Viro bool 709d2125043SAl Viro help 710d2125043SAl Viro Architecture has the first two arguments of clone(2) swapped. 711d2125043SAl Viro 712dfa9771aSMichal Simekconfig CLONE_BACKWARDS3 713dfa9771aSMichal Simek bool 714dfa9771aSMichal Simek help 715dfa9771aSMichal Simek Architecture has tls passed as the 3rd argument of clone(2), 716dfa9771aSMichal Simek not the 5th one. 717dfa9771aSMichal Simek 718eaca6eaeSAl Viroconfig ODD_RT_SIGACTION 719eaca6eaeSAl Viro bool 720eaca6eaeSAl Viro help 721eaca6eaeSAl Viro Architecture has unusual rt_sigaction(2) arguments 722eaca6eaeSAl Viro 7230a0e8cdfSAl Viroconfig OLD_SIGSUSPEND 7240a0e8cdfSAl Viro bool 7250a0e8cdfSAl Viro help 7260a0e8cdfSAl Viro Architecture has old sigsuspend(2) syscall, of one-argument variety 7270a0e8cdfSAl Viro 7280a0e8cdfSAl Viroconfig OLD_SIGSUSPEND3 7290a0e8cdfSAl Viro bool 7300a0e8cdfSAl Viro help 7310a0e8cdfSAl Viro Even weirder antique ABI - three-argument sigsuspend(2) 7320a0e8cdfSAl Viro 733495dfbf7SAl Viroconfig OLD_SIGACTION 734495dfbf7SAl Viro bool 735495dfbf7SAl Viro help 736495dfbf7SAl Viro Architecture has old sigaction(2) syscall. Nope, not the same 737495dfbf7SAl Viro as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 738495dfbf7SAl Viro but fairly different variant of sigaction(2), thanks to OSF/1 739495dfbf7SAl Viro compatibility... 740495dfbf7SAl Viro 741495dfbf7SAl Viroconfig COMPAT_OLD_SIGACTION 742495dfbf7SAl Viro bool 743495dfbf7SAl Viro 7440d4a619bSChristoph Hellwigconfig ARCH_NO_COHERENT_DMA_MMAP 7450d4a619bSChristoph Hellwig bool 7460d4a619bSChristoph Hellwig 747fff7fb0bSZhaoxiu Zengconfig CPU_NO_EFFICIENT_FFS 748fff7fb0bSZhaoxiu Zeng def_bool n 749fff7fb0bSZhaoxiu Zeng 750ba14a194SAndy Lutomirskiconfig HAVE_ARCH_VMAP_STACK 751ba14a194SAndy Lutomirski def_bool n 752ba14a194SAndy Lutomirski help 753ba14a194SAndy Lutomirski An arch should select this symbol if it can support kernel stacks 754ba14a194SAndy Lutomirski in vmalloc space. This means: 755ba14a194SAndy Lutomirski 756ba14a194SAndy Lutomirski - vmalloc space must be large enough to hold many kernel stacks. 757ba14a194SAndy Lutomirski This may rule out many 32-bit architectures. 758ba14a194SAndy Lutomirski 759ba14a194SAndy Lutomirski - Stacks in vmalloc space need to work reliably. For example, if 760ba14a194SAndy Lutomirski vmap page tables are created on demand, either this mechanism 761ba14a194SAndy Lutomirski needs to work while the stack points to a virtual address with 762ba14a194SAndy Lutomirski unpopulated page tables or arch code (switch_to() and switch_mm(), 763ba14a194SAndy Lutomirski most likely) needs to ensure that the stack's page table entries 764ba14a194SAndy Lutomirski are populated before running on a possibly unpopulated stack. 765ba14a194SAndy Lutomirski 766ba14a194SAndy Lutomirski - If the stack overflows into a guard page, something reasonable 767ba14a194SAndy Lutomirski should happen. The definition of "reasonable" is flexible, but 768ba14a194SAndy Lutomirski instantly rebooting without logging anything would be unfriendly. 769ba14a194SAndy Lutomirski 770ba14a194SAndy Lutomirskiconfig VMAP_STACK 771ba14a194SAndy Lutomirski default y 772ba14a194SAndy Lutomirski bool "Use a virtually-mapped stack" 773ba14a194SAndy Lutomirski depends on HAVE_ARCH_VMAP_STACK && !KASAN 774ba14a194SAndy Lutomirski ---help--- 775ba14a194SAndy Lutomirski Enable this if you want the use virtually-mapped kernel stacks 776ba14a194SAndy Lutomirski with guard pages. This causes kernel stack overflows to be 777ba14a194SAndy Lutomirski caught immediately rather than causing difficult-to-diagnose 778ba14a194SAndy Lutomirski corruption. 779ba14a194SAndy Lutomirski 780ba14a194SAndy Lutomirski This is presently incompatible with KASAN because KASAN expects 781ba14a194SAndy Lutomirski the stack to map directly to the KASAN shadow map using a formula 782ba14a194SAndy Lutomirski that is incorrect if the stack is in vmalloc space. 783ba14a194SAndy Lutomirski 784ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX 785ad21fc4fSLaura Abbott def_bool n 786ad21fc4fSLaura Abbott 787ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 788ad21fc4fSLaura Abbott def_bool n 789ad21fc4fSLaura Abbott 790ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_KERNEL_RWX 791ad21fc4fSLaura Abbott def_bool n 792ad21fc4fSLaura Abbott 7930f5bf6d0SLaura Abbottconfig STRICT_KERNEL_RWX 794ad21fc4fSLaura Abbott bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 795ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_KERNEL_RWX 796ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 797ad21fc4fSLaura Abbott help 798ad21fc4fSLaura Abbott If this is set, kernel text and rodata memory will be made read-only, 799ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 800ad21fc4fSLaura Abbott protection against certain security exploits (e.g. executing the heap 801ad21fc4fSLaura Abbott or modifying text) 802ad21fc4fSLaura Abbott 803ad21fc4fSLaura Abbott These features are considered standard security practice these days. 804ad21fc4fSLaura Abbott You should say Y here in almost all cases. 805ad21fc4fSLaura Abbott 806ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_MODULE_RWX 807ad21fc4fSLaura Abbott def_bool n 808ad21fc4fSLaura Abbott 8090f5bf6d0SLaura Abbottconfig STRICT_MODULE_RWX 810ad21fc4fSLaura Abbott bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 811ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 812ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 813ad21fc4fSLaura Abbott help 814ad21fc4fSLaura Abbott If this is set, module text and rodata memory will be made read-only, 815ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 816ad21fc4fSLaura Abbott protection against certain security exploits (e.g. writing to text) 817ad21fc4fSLaura Abbott 8182521f2c2SPeter Oberparleitersource "kernel/gcov/Kconfig" 819