1b2441318SGreg Kroah-Hartman# SPDX-License-Identifier: GPL-2.0 2fb32e03fSMathieu Desnoyers# 3fb32e03fSMathieu Desnoyers# General architecture dependent options 4fb32e03fSMathieu Desnoyers# 5125e5645SMathieu Desnoyers 61572497cSChristoph Hellwig# 71572497cSChristoph Hellwig# Note: arch/$(SRCARCH)/Kconfig needs to be included first so that it can 81572497cSChristoph Hellwig# override the default values in this file. 91572497cSChristoph Hellwig# 101572497cSChristoph Hellwigsource "arch/$(SRCARCH)/Kconfig" 111572497cSChristoph Hellwig 12fe42754bSSean Christophersonconfig ARCH_CONFIGURES_CPU_MITIGATIONS 13fe42754bSSean Christopherson bool 14fe42754bSSean Christopherson 15fe42754bSSean Christophersonif !ARCH_CONFIGURES_CPU_MITIGATIONS 16fe42754bSSean Christophersonconfig CPU_MITIGATIONS 17fe42754bSSean Christopherson def_bool y 18fe42754bSSean Christophersonendif 19fe42754bSSean Christopherson 20de6c85bfSChristoph Hellwig# 21de6c85bfSChristoph Hellwig# Selected by architectures that need custom DMA operations for e.g. legacy 22de6c85bfSChristoph Hellwig# IOMMUs not handled by dma-iommu. Drivers must never select this symbol. 23de6c85bfSChristoph Hellwig# 24de6c85bfSChristoph Hellwigconfig ARCH_HAS_DMA_OPS 25de6c85bfSChristoph Hellwig depends on HAS_DMA 26de6c85bfSChristoph Hellwig select DMA_OPS_HELPERS 27de6c85bfSChristoph Hellwig bool 28de6c85bfSChristoph Hellwig 2922471e13SRandy Dunlapmenu "General architecture-dependent options" 3022471e13SRandy Dunlap 31da32b581SCatalin Marinasconfig ARCH_HAS_SUBPAGE_FAULTS 32da32b581SCatalin Marinas bool 33da32b581SCatalin Marinas help 34da32b581SCatalin Marinas Select if the architecture can check permissions at sub-page 35da32b581SCatalin Marinas granularity (e.g. arm64 MTE). The probe_user_*() functions 36da32b581SCatalin Marinas must be implemented. 37da32b581SCatalin Marinas 3805736e4aSThomas Gleixnerconfig HOTPLUG_SMT 3905736e4aSThomas Gleixner bool 4005736e4aSThomas Gleixner 4138253464SMichael Ellermanconfig SMT_NUM_THREADS_DYNAMIC 4238253464SMichael Ellerman bool 4338253464SMichael Ellerman 446f062123SThomas Gleixner# Selected by HOTPLUG_CORE_SYNC_DEAD or HOTPLUG_CORE_SYNC_FULL 456f062123SThomas Gleixnerconfig HOTPLUG_CORE_SYNC 466f062123SThomas Gleixner bool 476f062123SThomas Gleixner 486f062123SThomas Gleixner# Basic CPU dead synchronization selected by architecture 496f062123SThomas Gleixnerconfig HOTPLUG_CORE_SYNC_DEAD 506f062123SThomas Gleixner bool 516f062123SThomas Gleixner select HOTPLUG_CORE_SYNC 526f062123SThomas Gleixner 536f062123SThomas Gleixner# Full CPU synchronization with alive state selected by architecture 546f062123SThomas Gleixnerconfig HOTPLUG_CORE_SYNC_FULL 556f062123SThomas Gleixner bool 566f062123SThomas Gleixner select HOTPLUG_CORE_SYNC_DEAD if HOTPLUG_CPU 576f062123SThomas Gleixner select HOTPLUG_CORE_SYNC 586f062123SThomas Gleixner 59a631be92SThomas Gleixnerconfig HOTPLUG_SPLIT_STARTUP 60a631be92SThomas Gleixner bool 61a631be92SThomas Gleixner select HOTPLUG_CORE_SYNC_FULL 62a631be92SThomas Gleixner 6318415f33SThomas Gleixnerconfig HOTPLUG_PARALLEL 6418415f33SThomas Gleixner bool 6518415f33SThomas Gleixner select HOTPLUG_SPLIT_STARTUP 6618415f33SThomas Gleixner 67*a70e9f64SJinjie Ruanconfig GENERIC_IRQ_ENTRY 68*a70e9f64SJinjie Ruan bool 69*a70e9f64SJinjie Ruan 70*a70e9f64SJinjie Ruanconfig GENERIC_SYSCALL 71*a70e9f64SJinjie Ruan bool 72*a70e9f64SJinjie Ruan depends on GENERIC_IRQ_ENTRY 73*a70e9f64SJinjie Ruan 74142781e1SThomas Gleixnerconfig GENERIC_ENTRY 75142781e1SThomas Gleixner bool 76*a70e9f64SJinjie Ruan select GENERIC_IRQ_ENTRY 77*a70e9f64SJinjie Ruan select GENERIC_SYSCALL 78142781e1SThomas Gleixner 79125e5645SMathieu Desnoyersconfig KPROBES 80125e5645SMathieu Desnoyers bool "Kprobes" 81125e5645SMathieu Desnoyers depends on HAVE_KPROBES 8205ed160eSMasami Hiramatsu select KALLSYMS 837582b7beSMike Rapoport (IBM) select EXECMEM 84900da4d2SPaul E. McKenney select NEED_TASKS_RCU 85125e5645SMathieu Desnoyers help 86125e5645SMathieu Desnoyers Kprobes allows you to trap at almost any kernel address and 87125e5645SMathieu Desnoyers execute a callback function. register_kprobe() establishes 88125e5645SMathieu Desnoyers a probepoint and specifies the callback. Kprobes is useful 89125e5645SMathieu Desnoyers for kernel debugging, non-intrusive instrumentation and testing. 90125e5645SMathieu Desnoyers If in doubt, say "N". 91125e5645SMathieu Desnoyers 9245f81b1cSSteven Rostedtconfig JUMP_LABEL 93c5905afbSIngo Molnar bool "Optimize very unlikely/likely branches" 9445f81b1cSSteven Rostedt depends on HAVE_ARCH_JUMP_LABEL 954ab7674fSJosh Poimboeuf select OBJTOOL if HAVE_JUMP_LABEL_HACK 9645f81b1cSSteven Rostedt help 97c5905afbSIngo Molnar This option enables a transparent branch optimization that 98c5905afbSIngo Molnar makes certain almost-always-true or almost-always-false branch 99c5905afbSIngo Molnar conditions even cheaper to execute within the kernel. 10045f81b1cSSteven Rostedt 101c5905afbSIngo Molnar Certain performance-sensitive kernel code, such as trace points, 102c5905afbSIngo Molnar scheduler functionality, networking code and KVM have such 103c5905afbSIngo Molnar branches and include support for this optimization technique. 104c5905afbSIngo Molnar 105c5905afbSIngo Molnar If it is detected that the compiler has support for "asm goto", 106c5905afbSIngo Molnar the kernel will compile such branches with just a nop 107c5905afbSIngo Molnar instruction. When the condition flag is toggled to true, the 108c5905afbSIngo Molnar nop will be converted to a jump instruction to execute the 109c5905afbSIngo Molnar conditional block of instructions. 110c5905afbSIngo Molnar 111c5905afbSIngo Molnar This technique lowers overhead and stress on the branch prediction 112c5905afbSIngo Molnar of the processor and generally makes the kernel faster. The update 113c5905afbSIngo Molnar of the condition is slower, but those are always very rare. 114c5905afbSIngo Molnar 115c5905afbSIngo Molnar ( On 32-bit x86, the necessary options added to the compiler 116c5905afbSIngo Molnar flags may increase the size of the kernel slightly. ) 11745f81b1cSSteven Rostedt 1181987c947SPeter Zijlstraconfig STATIC_KEYS_SELFTEST 1191987c947SPeter Zijlstra bool "Static key selftest" 1201987c947SPeter Zijlstra depends on JUMP_LABEL 1211987c947SPeter Zijlstra help 1221987c947SPeter Zijlstra Boot time self-test of the branch patching code. 1231987c947SPeter Zijlstra 124f03c4129SPeter Zijlstraconfig STATIC_CALL_SELFTEST 125f03c4129SPeter Zijlstra bool "Static call selftest" 126f03c4129SPeter Zijlstra depends on HAVE_STATIC_CALL 127f03c4129SPeter Zijlstra help 128f03c4129SPeter Zijlstra Boot time self-test of the call patching code. 129f03c4129SPeter Zijlstra 130afd66255SMasami Hiramatsuconfig OPTPROBES 1315cc718b9SMasami Hiramatsu def_bool y 1325cc718b9SMasami Hiramatsu depends on KPROBES && HAVE_OPTPROBES 133900da4d2SPaul E. McKenney select NEED_TASKS_RCU 134afd66255SMasami Hiramatsu 135e7dbfe34SMasami Hiramatsuconfig KPROBES_ON_FTRACE 136e7dbfe34SMasami Hiramatsu def_bool y 137e7dbfe34SMasami Hiramatsu depends on KPROBES && HAVE_KPROBES_ON_FTRACE 138e7dbfe34SMasami Hiramatsu depends on DYNAMIC_FTRACE_WITH_REGS 139e7dbfe34SMasami Hiramatsu help 140e7dbfe34SMasami Hiramatsu If function tracer is enabled and the arch supports full 141e7dbfe34SMasami Hiramatsu passing of pt_regs to function tracing, then kprobes can 142e7dbfe34SMasami Hiramatsu optimize on top of function tracing. 143e7dbfe34SMasami Hiramatsu 1442b144498SSrikar Dronamrajuconfig UPROBES 14509294e31SDavid A. Long def_bool n 146e8f4aa60SAllen Pais depends on ARCH_SUPPORTS_UPROBES 14787195a1eSAndrii Nakryiko select TASKS_TRACE_RCU 1482b144498SSrikar Dronamraju help 1497b2d81d4SIngo Molnar Uprobes is the user-space counterpart to kprobes: they 1507b2d81d4SIngo Molnar enable instrumentation applications (such as 'perf probe') 1517b2d81d4SIngo Molnar to establish unintrusive probes in user-space binaries and 1527b2d81d4SIngo Molnar libraries, by executing handler functions when the probes 1537b2d81d4SIngo Molnar are hit by user-space applications. 1547b2d81d4SIngo Molnar 1557b2d81d4SIngo Molnar ( These probes come in the form of single-byte breakpoints, 1567b2d81d4SIngo Molnar managed by the kernel and kept transparent to the probed 1577b2d81d4SIngo Molnar application. ) 1582b144498SSrikar Dronamraju 159adab66b7SSteven Rostedt (VMware)config HAVE_64BIT_ALIGNED_ACCESS 160adab66b7SSteven Rostedt (VMware) def_bool 64BIT && !HAVE_EFFICIENT_UNALIGNED_ACCESS 161adab66b7SSteven Rostedt (VMware) help 162adab66b7SSteven Rostedt (VMware) Some architectures require 64 bit accesses to be 64 bit 163adab66b7SSteven Rostedt (VMware) aligned, which also requires structs containing 64 bit values 164adab66b7SSteven Rostedt (VMware) to be 64 bit aligned too. This includes some 32 bit 165adab66b7SSteven Rostedt (VMware) architectures which can do 64 bit accesses, as well as 64 bit 166adab66b7SSteven Rostedt (VMware) architectures without unaligned access. 167adab66b7SSteven Rostedt (VMware) 168adab66b7SSteven Rostedt (VMware) This symbol should be selected by an architecture if 64 bit 169adab66b7SSteven Rostedt (VMware) accesses are required to be 64 bit aligned in this way even 170adab66b7SSteven Rostedt (VMware) though it is not a 64 bit architecture. 171adab66b7SSteven Rostedt (VMware) 172ba1a297dSLukas Bulwahn See Documentation/core-api/unaligned-memory-access.rst for 173ba1a297dSLukas Bulwahn more information on the topic of unaligned memory accesses. 174adab66b7SSteven Rostedt (VMware) 17558340a07SJohannes Bergconfig HAVE_EFFICIENT_UNALIGNED_ACCESS 1769ba16087SJan Beulich bool 17758340a07SJohannes Berg help 17858340a07SJohannes Berg Some architectures are unable to perform unaligned accesses 17958340a07SJohannes Berg without the use of get_unaligned/put_unaligned. Others are 18058340a07SJohannes Berg unable to perform such accesses efficiently (e.g. trap on 18158340a07SJohannes Berg unaligned access and require fixing it up in the exception 18258340a07SJohannes Berg handler.) 18358340a07SJohannes Berg 18458340a07SJohannes Berg This symbol should be selected by an architecture if it can 18558340a07SJohannes Berg perform unaligned accesses efficiently to allow different 18658340a07SJohannes Berg code paths to be selected for these cases. Some network 18758340a07SJohannes Berg drivers, for example, could opt to not fix up alignment 18858340a07SJohannes Berg problems with received packets if doing so would not help 18958340a07SJohannes Berg much. 19058340a07SJohannes Berg 191c9b54d6fSMauro Carvalho Chehab See Documentation/core-api/unaligned-memory-access.rst for more 19258340a07SJohannes Berg information on the topic of unaligned memory accesses. 19358340a07SJohannes Berg 194cf66bb93SDavid Woodhouseconfig ARCH_USE_BUILTIN_BSWAP 195cf66bb93SDavid Woodhouse bool 196cf66bb93SDavid Woodhouse help 197cf66bb93SDavid Woodhouse Modern versions of GCC (since 4.4) have builtin functions 198cf66bb93SDavid Woodhouse for handling byte-swapping. Using these, instead of the old 199cf66bb93SDavid Woodhouse inline assembler that the architecture code provides in the 200cf66bb93SDavid Woodhouse __arch_bswapXX() macros, allows the compiler to see what's 201cf66bb93SDavid Woodhouse happening and offers more opportunity for optimisation. In 202cf66bb93SDavid Woodhouse particular, the compiler will be able to combine the byteswap 203cf66bb93SDavid Woodhouse with a nearby load or store and use load-and-swap or 204cf66bb93SDavid Woodhouse store-and-swap instructions if the architecture has them. It 205cf66bb93SDavid Woodhouse should almost *never* result in code which is worse than the 206cf66bb93SDavid Woodhouse hand-coded assembler in <asm/swab.h>. But just in case it 207cf66bb93SDavid Woodhouse does, the use of the builtins is optional. 208cf66bb93SDavid Woodhouse 209cf66bb93SDavid Woodhouse Any architecture with load-and-swap or store-and-swap 210cf66bb93SDavid Woodhouse instructions should set this. And it shouldn't hurt to set it 211cf66bb93SDavid Woodhouse on architectures that don't have such instructions. 212cf66bb93SDavid Woodhouse 2139edddaa2SAnanth N Mavinakayanahalliconfig KRETPROBES 2149edddaa2SAnanth N Mavinakayanahalli def_bool y 21573f9b911SMasami Hiramatsu depends on KPROBES && (HAVE_KRETPROBES || HAVE_RETHOOK) 21673f9b911SMasami Hiramatsu 21773f9b911SMasami Hiramatsuconfig KRETPROBE_ON_RETHOOK 21873f9b911SMasami Hiramatsu def_bool y 21973f9b911SMasami Hiramatsu depends on HAVE_RETHOOK 22073f9b911SMasami Hiramatsu depends on KRETPROBES 22173f9b911SMasami Hiramatsu select RETHOOK 2229edddaa2SAnanth N Mavinakayanahalli 2237c68af6eSAvi Kivityconfig USER_RETURN_NOTIFIER 2247c68af6eSAvi Kivity bool 2257c68af6eSAvi Kivity depends on HAVE_USER_RETURN_NOTIFIER 2267c68af6eSAvi Kivity help 2277c68af6eSAvi Kivity Provide a kernel-internal notification when a cpu is about to 2287c68af6eSAvi Kivity switch to user mode. 2297c68af6eSAvi Kivity 23028b2ee20SRik van Rielconfig HAVE_IOREMAP_PROT 2319ba16087SJan Beulich bool 23228b2ee20SRik van Riel 233125e5645SMathieu Desnoyersconfig HAVE_KPROBES 2349ba16087SJan Beulich bool 2359edddaa2SAnanth N Mavinakayanahalli 2369edddaa2SAnanth N Mavinakayanahalliconfig HAVE_KRETPROBES 2379ba16087SJan Beulich bool 23874bc7ceeSArthur Kepner 239afd66255SMasami Hiramatsuconfig HAVE_OPTPROBES 240afd66255SMasami Hiramatsu bool 241d314d74cSCong Wang 242e7dbfe34SMasami Hiramatsuconfig HAVE_KPROBES_ON_FTRACE 243e7dbfe34SMasami Hiramatsu bool 244e7dbfe34SMasami Hiramatsu 2451f6d3a8fSMasami Hiramatsuconfig ARCH_CORRECT_STACKTRACE_ON_KRETPROBE 2461f6d3a8fSMasami Hiramatsu bool 2471f6d3a8fSMasami Hiramatsu help 2481f6d3a8fSMasami Hiramatsu Since kretprobes modifies return address on the stack, the 2491f6d3a8fSMasami Hiramatsu stacktrace may see the kretprobe trampoline address instead 2501f6d3a8fSMasami Hiramatsu of correct one. If the architecture stacktrace code and 2511f6d3a8fSMasami Hiramatsu unwinder can adjust such entries, select this configuration. 2521f6d3a8fSMasami Hiramatsu 253540adea3SMasami Hiramatsuconfig HAVE_FUNCTION_ERROR_INJECTION 2549802d865SJosef Bacik bool 2559802d865SJosef Bacik 25642a0bb3fSPetr Mladekconfig HAVE_NMI 25742a0bb3fSPetr Mladek bool 25842a0bb3fSPetr Mladek 259a257caccSChristophe Leroyconfig HAVE_FUNCTION_DESCRIPTORS 260a257caccSChristophe Leroy bool 261a257caccSChristophe Leroy 2624aae683fSMasahiro Yamadaconfig TRACE_IRQFLAGS_SUPPORT 2634aae683fSMasahiro Yamada bool 2644aae683fSMasahiro Yamada 2654510bffbSMark Rutlandconfig TRACE_IRQFLAGS_NMI_SUPPORT 2664510bffbSMark Rutland bool 2674510bffbSMark Rutland 2681f5a4ad9SRoland McGrath# 2691f5a4ad9SRoland McGrath# An arch should select this if it provides all these things: 2701f5a4ad9SRoland McGrath# 2711f5a4ad9SRoland McGrath# task_pt_regs() in asm/processor.h or asm/ptrace.h 2721f5a4ad9SRoland McGrath# arch_has_single_step() if there is hardware single-step support 2731f5a4ad9SRoland McGrath# arch_has_block_step() if there is hardware block-step support 2741f5a4ad9SRoland McGrath# asm/syscall.h supplying asm-generic/syscall.h interface 2751f5a4ad9SRoland McGrath# linux/regset.h user_regset interfaces 2761f5a4ad9SRoland McGrath# CORE_DUMP_USE_REGSET #define'd in linux/elf.h 277153474baSEric W. Biederman# TIF_SYSCALL_TRACE calls ptrace_report_syscall_{entry,exit} 27803248addSEric W. Biederman# TIF_NOTIFY_RESUME calls resume_user_mode_work() 2791f5a4ad9SRoland McGrath# 2801f5a4ad9SRoland McGrathconfig HAVE_ARCH_TRACEHOOK 2819ba16087SJan Beulich bool 2821f5a4ad9SRoland McGrath 283c64be2bbSMarek Szyprowskiconfig HAVE_DMA_CONTIGUOUS 284c64be2bbSMarek Szyprowski bool 285c64be2bbSMarek Szyprowski 28629d5e047SThomas Gleixnerconfig GENERIC_SMP_IDLE_THREAD 28729d5e047SThomas Gleixner bool 28829d5e047SThomas Gleixner 289485cf5daSKevin Hilmanconfig GENERIC_IDLE_POLL_SETUP 290485cf5daSKevin Hilman bool 291485cf5daSKevin Hilman 2926974f0c4SDaniel Micayconfig ARCH_HAS_FORTIFY_SOURCE 2936974f0c4SDaniel Micay bool 2946974f0c4SDaniel Micay help 2956974f0c4SDaniel Micay An architecture should select this when it can successfully 2966974f0c4SDaniel Micay build and run with CONFIG_FORTIFY_SOURCE. 2976974f0c4SDaniel Micay 298d8ae8a37SChristoph Hellwig# 299d8ae8a37SChristoph Hellwig# Select if the arch provides a historic keepinit alias for the retain_initrd 300d8ae8a37SChristoph Hellwig# command line option 301d8ae8a37SChristoph Hellwig# 302d8ae8a37SChristoph Hellwigconfig ARCH_HAS_KEEPINITRD 303d8ae8a37SChristoph Hellwig bool 304d8ae8a37SChristoph Hellwig 305d2852a22SDaniel Borkmann# Select if arch has all set_memory_ro/rw/x/nx() functions in asm/cacheflush.h 306d2852a22SDaniel Borkmannconfig ARCH_HAS_SET_MEMORY 307d2852a22SDaniel Borkmann bool 308d2852a22SDaniel Borkmann 309d253ca0cSRick Edgecombe# Select if arch has all set_direct_map_invalid/default() functions 310d253ca0cSRick Edgecombeconfig ARCH_HAS_SET_DIRECT_MAP 311d253ca0cSRick Edgecombe bool 312d253ca0cSRick Edgecombe 313c30700dbSChristoph Hellwig# 314fa7e2247SChristoph Hellwig# Select if the architecture provides the arch_dma_set_uncached symbol to 315a86ecfa6SColin Ian King# either provide an uncached segment alias for a DMA allocation, or 316fa7e2247SChristoph Hellwig# to remap the page tables in place. 317c30700dbSChristoph Hellwig# 318fa7e2247SChristoph Hellwigconfig ARCH_HAS_DMA_SET_UNCACHED 319c30700dbSChristoph Hellwig bool 320c30700dbSChristoph Hellwig 321999a5d12SChristoph Hellwig# 322999a5d12SChristoph Hellwig# Select if the architectures provides the arch_dma_clear_uncached symbol 323999a5d12SChristoph Hellwig# to undo an in-place page table remap for uncached access. 324999a5d12SChristoph Hellwig# 325999a5d12SChristoph Hellwigconfig ARCH_HAS_DMA_CLEAR_UNCACHED 326f5e10287SThomas Gleixner bool 327f5e10287SThomas Gleixner 3287725acaaSThomas Gleixnerconfig ARCH_HAS_CPU_FINALIZE_INIT 3297725acaaSThomas Gleixner bool 3307725acaaSThomas Gleixner 3318f23f5dbSJason Gunthorpe# The architecture has a per-task state that includes the mm's PASID 3328f23f5dbSJason Gunthorpeconfig ARCH_HAS_CPU_PASID 3338f23f5dbSJason Gunthorpe bool 3348f23f5dbSJason Gunthorpe select IOMMU_MM_DATA 3358f23f5dbSJason Gunthorpe 3365905429aSKees Cookconfig HAVE_ARCH_THREAD_STRUCT_WHITELIST 3375905429aSKees Cook bool 3385905429aSKees Cook help 3395905429aSKees Cook An architecture should select this to provide hardened usercopy 3405905429aSKees Cook knowledge about what region of the thread_struct should be 3415905429aSKees Cook whitelisted for copying to userspace. Normally this is only the 3425905429aSKees Cook FPU registers. Specifically, arch_thread_struct_whitelist() 3435905429aSKees Cook should be implemented. Without this, the entire thread_struct 3445905429aSKees Cook field in task_struct will be left whitelisted. 3455905429aSKees Cook 3465aaeb5c0SIngo Molnar# Select if arch wants to size task_struct dynamically via arch_task_struct_size: 3475aaeb5c0SIngo Molnarconfig ARCH_WANTS_DYNAMIC_TASK_STRUCT 3485aaeb5c0SIngo Molnar bool 3495aaeb5c0SIngo Molnar 35051c2ee6dSNick Desaulniersconfig ARCH_WANTS_NO_INSTR 35151c2ee6dSNick Desaulniers bool 35251c2ee6dSNick Desaulniers help 35351c2ee6dSNick Desaulniers An architecture should select this if the noinstr macro is being used on 35451c2ee6dSNick Desaulniers functions to denote that the toolchain should avoid instrumenting such 35551c2ee6dSNick Desaulniers functions and is required for correctness. 35651c2ee6dSNick Desaulniers 357942fa985SYury Norovconfig ARCH_32BIT_OFF_T 358942fa985SYury Norov bool 359942fa985SYury Norov depends on !64BIT 360942fa985SYury Norov help 361942fa985SYury Norov All new 32-bit architectures should have 64-bit off_t type on 362942fa985SYury Norov userspace side which corresponds to the loff_t kernel type. This 363942fa985SYury Norov is the requirement for modern ABIs. Some existing architectures 364942fa985SYury Norov still support 32-bit off_t. This option is enabled for all such 365942fa985SYury Norov architectures explicitly. 366942fa985SYury Norov 36796c0a6a7SHeiko Carstens# Selected by 64 bit architectures which have a 32 bit f_tinode in struct ustat 36896c0a6a7SHeiko Carstensconfig ARCH_32BIT_USTAT_F_TINODE 36996c0a6a7SHeiko Carstens bool 37096c0a6a7SHeiko Carstens 3712ff2b7ecSMasahiro Yamadaconfig HAVE_ASM_MODVERSIONS 3722ff2b7ecSMasahiro Yamada bool 3732ff2b7ecSMasahiro Yamada help 374a86ecfa6SColin Ian King This symbol should be selected by an architecture if it provides 3752ff2b7ecSMasahiro Yamada <asm/asm-prototypes.h> to support the module versioning for symbols 3762ff2b7ecSMasahiro Yamada exported from assembly code. 3772ff2b7ecSMasahiro Yamada 378f850c30cSHeiko Carstensconfig HAVE_REGS_AND_STACK_ACCESS_API 379f850c30cSHeiko Carstens bool 380e01292b1SHeiko Carstens help 381a86ecfa6SColin Ian King This symbol should be selected by an architecture if it supports 382e01292b1SHeiko Carstens the API needed to access registers and stack entries from pt_regs, 383e01292b1SHeiko Carstens declared in asm/ptrace.h 384e01292b1SHeiko Carstens For example the kprobes-based event tracer needs this API. 385f850c30cSHeiko Carstens 386d7822b1eSMathieu Desnoyersconfig HAVE_RSEQ 387d7822b1eSMathieu Desnoyers bool 388d7822b1eSMathieu Desnoyers depends on HAVE_REGS_AND_STACK_ACCESS_API 389d7822b1eSMathieu Desnoyers help 390d7822b1eSMathieu Desnoyers This symbol should be selected by an architecture if it 391d7822b1eSMathieu Desnoyers supports an implementation of restartable sequences. 392d7822b1eSMathieu Desnoyers 3932f7ab126SMiguel Ojedaconfig HAVE_RUST 3942f7ab126SMiguel Ojeda bool 3952f7ab126SMiguel Ojeda help 3962f7ab126SMiguel Ojeda This symbol should be selected by an architecture if it 3972f7ab126SMiguel Ojeda supports Rust. 3982f7ab126SMiguel Ojeda 3993c88ee19SMasami Hiramatsuconfig HAVE_FUNCTION_ARG_ACCESS_API 4003c88ee19SMasami Hiramatsu bool 4013c88ee19SMasami Hiramatsu help 402a86ecfa6SColin Ian King This symbol should be selected by an architecture if it supports 4033c88ee19SMasami Hiramatsu the API needed to access function arguments from pt_regs, 4043c88ee19SMasami Hiramatsu declared in asm/ptrace.h 4053c88ee19SMasami Hiramatsu 40662a038d3SK.Prasadconfig HAVE_HW_BREAKPOINT 40762a038d3SK.Prasad bool 40899e8c5a3SFrederic Weisbecker depends on PERF_EVENTS 40962a038d3SK.Prasad 4100102752eSFrederic Weisbeckerconfig HAVE_MIXED_BREAKPOINTS_REGS 4110102752eSFrederic Weisbecker bool 4120102752eSFrederic Weisbecker depends on HAVE_HW_BREAKPOINT 4130102752eSFrederic Weisbecker help 4140102752eSFrederic Weisbecker Depending on the arch implementation of hardware breakpoints, 4150102752eSFrederic Weisbecker some of them have separate registers for data and instruction 4160102752eSFrederic Weisbecker breakpoints addresses, others have mixed registers to store 4170102752eSFrederic Weisbecker them but define the access type in a control register. 4180102752eSFrederic Weisbecker Select this option if your arch implements breakpoints under the 4190102752eSFrederic Weisbecker latter fashion. 4200102752eSFrederic Weisbecker 4217c68af6eSAvi Kivityconfig HAVE_USER_RETURN_NOTIFIER 4227c68af6eSAvi Kivity bool 423a1922ed6SIngo Molnar 424c01d4323SFrederic Weisbeckerconfig HAVE_PERF_EVENTS_NMI 425c01d4323SFrederic Weisbecker bool 42623637d47SFrederic Weisbecker help 42723637d47SFrederic Weisbecker System hardware can generate an NMI using the perf event 42823637d47SFrederic Weisbecker subsystem. Also has support for calculating CPU cycle events 42923637d47SFrederic Weisbecker to determine how many clock cycles in a given period. 430c01d4323SFrederic Weisbecker 43105a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_PERF 43205a4a952SNicholas Piggin bool 43305a4a952SNicholas Piggin depends on HAVE_PERF_EVENTS_NMI 43405a4a952SNicholas Piggin help 43505a4a952SNicholas Piggin The arch chooses to use the generic perf-NMI-based hardlockup 43605a4a952SNicholas Piggin detector. Must define HAVE_PERF_EVENTS_NMI. 43705a4a952SNicholas Piggin 43805a4a952SNicholas Pigginconfig HAVE_HARDLOCKUP_DETECTOR_ARCH 43905a4a952SNicholas Piggin bool 44005a4a952SNicholas Piggin help 4411356d0b9SPetr Mladek The arch provides its own hardlockup detector implementation instead 4421356d0b9SPetr Mladek of the generic ones. 4431356d0b9SPetr Mladek 4441356d0b9SPetr Mladek It uses the same command line parameters, and sysctl interface, 4451356d0b9SPetr Mladek as the generic hardlockup detectors. 44605a4a952SNicholas Piggin 447c5e63197SJiri Olsaconfig HAVE_PERF_REGS 448c5e63197SJiri Olsa bool 449c5e63197SJiri Olsa help 450c5e63197SJiri Olsa Support selective register dumps for perf events. This includes 451c5e63197SJiri Olsa bit-mapping of each registers and a unique architecture id. 452c5e63197SJiri Olsa 453c5ebcedbSJiri Olsaconfig HAVE_PERF_USER_STACK_DUMP 454c5ebcedbSJiri Olsa bool 455c5ebcedbSJiri Olsa help 456c5ebcedbSJiri Olsa Support user stack dumps for perf event samples. This needs 457c5ebcedbSJiri Olsa access to the user stack pointer which is not unified across 458c5ebcedbSJiri Olsa architectures. 459c5ebcedbSJiri Olsa 460bf5438fcSJason Baronconfig HAVE_ARCH_JUMP_LABEL 461bf5438fcSJason Baron bool 462bf5438fcSJason Baron 46350ff18abSArd Biesheuvelconfig HAVE_ARCH_JUMP_LABEL_RELATIVE 46450ff18abSArd Biesheuvel bool 46550ff18abSArd Biesheuvel 4660d6e24d4SPeter Zijlstraconfig MMU_GATHER_TABLE_FREE 4670d6e24d4SPeter Zijlstra bool 4680d6e24d4SPeter Zijlstra 469ff2e6d72SPeter Zijlstraconfig MMU_GATHER_RCU_TABLE_FREE 47026723911SPeter Zijlstra bool 4710d6e24d4SPeter Zijlstra select MMU_GATHER_TABLE_FREE 47226723911SPeter Zijlstra 4733af4bd03SPeter Zijlstraconfig MMU_GATHER_PAGE_SIZE 474ed6a7935SPeter Zijlstra bool 475ed6a7935SPeter Zijlstra 47627796d03SPeter Zijlstraconfig MMU_GATHER_NO_RANGE 47727796d03SPeter Zijlstra bool 4781e9fdf21SPeter Zijlstra select MMU_GATHER_MERGE_VMAS 4791e9fdf21SPeter Zijlstra 4801e9fdf21SPeter Zijlstraconfig MMU_GATHER_NO_FLUSH_CACHE 4811e9fdf21SPeter Zijlstra bool 4821e9fdf21SPeter Zijlstra 4831e9fdf21SPeter Zijlstraconfig MMU_GATHER_MERGE_VMAS 4841e9fdf21SPeter Zijlstra bool 48527796d03SPeter Zijlstra 486580a586cSPeter Zijlstraconfig MMU_GATHER_NO_GATHER 487952a31c9SMartin Schwidefsky bool 4880d6e24d4SPeter Zijlstra depends on MMU_GATHER_TABLE_FREE 489952a31c9SMartin Schwidefsky 490d53c3dfbSNicholas Pigginconfig ARCH_WANT_IRQS_OFF_ACTIVATE_MM 491d53c3dfbSNicholas Piggin bool 492d53c3dfbSNicholas Piggin help 493d53c3dfbSNicholas Piggin Temporary select until all architectures can be converted to have 494d53c3dfbSNicholas Piggin irqs disabled over activate_mm. Architectures that do IPI based TLB 495d53c3dfbSNicholas Piggin shootdowns should enable this. 496d53c3dfbSNicholas Piggin 49788e3009bSNicholas Piggin# Use normal mm refcounting for MMU_LAZY_TLB kernel thread references. 49888e3009bSNicholas Piggin# MMU_LAZY_TLB_REFCOUNT=n can improve the scalability of context switching 49988e3009bSNicholas Piggin# to/from kernel threads when the same mm is running on a lot of CPUs (a large 50088e3009bSNicholas Piggin# multi-threaded application), by reducing contention on the mm refcount. 50188e3009bSNicholas Piggin# 50288e3009bSNicholas Piggin# This can be disabled if the architecture ensures no CPUs are using an mm as a 50388e3009bSNicholas Piggin# "lazy tlb" beyond its final refcount (i.e., by the time __mmdrop frees the mm 50488e3009bSNicholas Piggin# or its kernel page tables). This could be arranged by arch_exit_mmap(), or 50588e3009bSNicholas Piggin# final exit(2) TLB flush, for example. 50688e3009bSNicholas Piggin# 50788e3009bSNicholas Piggin# To implement this, an arch *must*: 50888e3009bSNicholas Piggin# Ensure the _lazy_tlb variants of mmgrab/mmdrop are used when manipulating 50988e3009bSNicholas Piggin# the lazy tlb reference of a kthread's ->active_mm (non-arch code has been 51088e3009bSNicholas Piggin# converted already). 51188e3009bSNicholas Pigginconfig MMU_LAZY_TLB_REFCOUNT 51288e3009bSNicholas Piggin def_bool y 5132655421aSNicholas Piggin depends on !MMU_LAZY_TLB_SHOOTDOWN 5142655421aSNicholas Piggin 5152655421aSNicholas Piggin# This option allows MMU_LAZY_TLB_REFCOUNT=n. It ensures no CPUs are using an 5162655421aSNicholas Piggin# mm as a lazy tlb beyond its last reference count, by shooting down these 5172655421aSNicholas Piggin# users before the mm is deallocated. __mmdrop() first IPIs all CPUs that may 5182655421aSNicholas Piggin# be using the mm as a lazy tlb, so that they may switch themselves to using 5192655421aSNicholas Piggin# init_mm for their active mm. mm_cpumask(mm) is used to determine which CPUs 5202655421aSNicholas Piggin# may be using mm as a lazy tlb mm. 5212655421aSNicholas Piggin# 5222655421aSNicholas Piggin# To implement this, an arch *must*: 5232655421aSNicholas Piggin# - At the time of the final mmdrop of the mm, ensure mm_cpumask(mm) contains 5242655421aSNicholas Piggin# at least all possible CPUs in which the mm is lazy. 5252655421aSNicholas Piggin# - It must meet the requirements for MMU_LAZY_TLB_REFCOUNT=n (see above). 5262655421aSNicholas Pigginconfig MMU_LAZY_TLB_SHOOTDOWN 5272655421aSNicholas Piggin bool 52888e3009bSNicholas Piggin 529df013ffbSHuang Yingconfig ARCH_HAVE_NMI_SAFE_CMPXCHG 530df013ffbSHuang Ying bool 531df013ffbSHuang Ying 532a9c3475dSVignesh Balasubramanianconfig ARCH_HAVE_EXTRA_ELF_NOTES 533a9c3475dSVignesh Balasubramanian bool 534a9c3475dSVignesh Balasubramanian help 535a9c3475dSVignesh Balasubramanian An architecture should select this in order to enable adding an 536a9c3475dSVignesh Balasubramanian arch-specific ELF note section to core files. It must provide two 537a9c3475dSVignesh Balasubramanian functions: elf_coredump_extra_notes_size() and 538a9c3475dSVignesh Balasubramanian elf_coredump_extra_notes_write() which are invoked by the ELF core 539a9c3475dSVignesh Balasubramanian dumper. 540a9c3475dSVignesh Balasubramanian 5412e83b879SPaul E. McKenneyconfig ARCH_HAS_NMI_SAFE_THIS_CPU_OPS 5422e83b879SPaul E. McKenney bool 5432e83b879SPaul E. McKenney 54443570fd2SHeiko Carstensconfig HAVE_ALIGNED_STRUCT_PAGE 54543570fd2SHeiko Carstens bool 54643570fd2SHeiko Carstens help 54743570fd2SHeiko Carstens This makes sure that struct pages are double word aligned and that 54843570fd2SHeiko Carstens e.g. the SLUB allocator can perform double word atomic operations 54943570fd2SHeiko Carstens on a struct page for better performance. However selecting this 55043570fd2SHeiko Carstens might increase the size of a struct page by a word. 55143570fd2SHeiko Carstens 5524156153cSHeiko Carstensconfig HAVE_CMPXCHG_LOCAL 5534156153cSHeiko Carstens bool 5544156153cSHeiko Carstens 5552565409fSHeiko Carstensconfig HAVE_CMPXCHG_DOUBLE 5562565409fSHeiko Carstens bool 5572565409fSHeiko Carstens 55877e58496SPaul E. McKenneyconfig ARCH_WEAK_RELEASE_ACQUIRE 55977e58496SPaul E. McKenney bool 56077e58496SPaul E. McKenney 561c1d7e01dSWill Deaconconfig ARCH_WANT_IPC_PARSE_VERSION 562c1d7e01dSWill Deacon bool 563c1d7e01dSWill Deacon 564c1d7e01dSWill Deaconconfig ARCH_WANT_COMPAT_IPC_PARSE_VERSION 565c1d7e01dSWill Deacon bool 566c1d7e01dSWill Deacon 56748b25c43SChris Metcalfconfig ARCH_WANT_OLD_COMPAT_IPC 568c1d7e01dSWill Deacon select ARCH_WANT_COMPAT_IPC_PARSE_VERSION 56948b25c43SChris Metcalf bool 57048b25c43SChris Metcalf 571282a181bSYiFei Zhuconfig HAVE_ARCH_SECCOMP 572e2cfabdfSWill Drewry bool 573e2cfabdfSWill Drewry help 574282a181bSYiFei Zhu An arch should select this symbol to support seccomp mode 1 (the fixed 575282a181bSYiFei Zhu syscall policy), and must provide an overrides for __NR_seccomp_sigreturn, 576282a181bSYiFei Zhu and compat syscalls if the asm-generic/seccomp.h defaults need adjustment: 577282a181bSYiFei Zhu - __NR_seccomp_read_32 578282a181bSYiFei Zhu - __NR_seccomp_write_32 579282a181bSYiFei Zhu - __NR_seccomp_exit_32 580282a181bSYiFei Zhu - __NR_seccomp_sigreturn_32 581282a181bSYiFei Zhu 582282a181bSYiFei Zhuconfig HAVE_ARCH_SECCOMP_FILTER 583282a181bSYiFei Zhu bool 584282a181bSYiFei Zhu select HAVE_ARCH_SECCOMP 585282a181bSYiFei Zhu help 586fb0fadf9SWill Drewry An arch should select this symbol if it provides all of these things: 587282a181bSYiFei Zhu - all the requirements for HAVE_ARCH_SECCOMP 588bb6ea430SWill Drewry - syscall_get_arch() 589bb6ea430SWill Drewry - syscall_get_arguments() 590bb6ea430SWill Drewry - syscall_rollback() 591bb6ea430SWill Drewry - syscall_set_return_value() 592fb0fadf9SWill Drewry - SIGSYS siginfo_t support 593fb0fadf9SWill Drewry - secure_computing is called from a ptrace_event()-safe context 594fb0fadf9SWill Drewry - secure_computing return value is checked and a return value of -1 595fb0fadf9SWill Drewry results in the system call being skipped immediately. 59648dc92b9SKees Cook - seccomp syscall wired up 5970d8315ddSYiFei Zhu - if !HAVE_SPARSE_SYSCALL_NR, have SECCOMP_ARCH_NATIVE, 5980d8315ddSYiFei Zhu SECCOMP_ARCH_NATIVE_NR, SECCOMP_ARCH_NATIVE_NAME defined. If 5990d8315ddSYiFei Zhu COMPAT is supported, have the SECCOMP_ARCH_COMPAT* defines too. 600e2cfabdfSWill Drewry 601282a181bSYiFei Zhuconfig SECCOMP 602282a181bSYiFei Zhu prompt "Enable seccomp to safely execute untrusted bytecode" 603282a181bSYiFei Zhu def_bool y 604282a181bSYiFei Zhu depends on HAVE_ARCH_SECCOMP 605282a181bSYiFei Zhu help 606282a181bSYiFei Zhu This kernel feature is useful for number crunching applications 607282a181bSYiFei Zhu that may need to handle untrusted bytecode during their 608282a181bSYiFei Zhu execution. By using pipes or other transports made available 609282a181bSYiFei Zhu to the process as file descriptors supporting the read/write 610282a181bSYiFei Zhu syscalls, it's possible to isolate those applications in their 611282a181bSYiFei Zhu own address space using seccomp. Once seccomp is enabled via 612282a181bSYiFei Zhu prctl(PR_SET_SECCOMP) or the seccomp() syscall, it cannot be 613282a181bSYiFei Zhu disabled and the task is only allowed to execute a few safe 614282a181bSYiFei Zhu syscalls defined by each seccomp mode. 615282a181bSYiFei Zhu 616282a181bSYiFei Zhu If unsure, say Y. 617282a181bSYiFei Zhu 618e2cfabdfSWill Drewryconfig SECCOMP_FILTER 619e2cfabdfSWill Drewry def_bool y 620e2cfabdfSWill Drewry depends on HAVE_ARCH_SECCOMP_FILTER && SECCOMP && NET 621e2cfabdfSWill Drewry help 622e2cfabdfSWill Drewry Enable tasks to build secure computing environments defined 623e2cfabdfSWill Drewry in terms of Berkeley Packet Filter programs which implement 624e2cfabdfSWill Drewry task-defined system call filtering polices. 625e2cfabdfSWill Drewry 6265fb94e9cSMauro Carvalho Chehab See Documentation/userspace-api/seccomp_filter.rst for details. 627e2cfabdfSWill Drewry 6280d8315ddSYiFei Zhuconfig SECCOMP_CACHE_DEBUG 6290d8315ddSYiFei Zhu bool "Show seccomp filter cache status in /proc/pid/seccomp_cache" 6300d8315ddSYiFei Zhu depends on SECCOMP_FILTER && !HAVE_SPARSE_SYSCALL_NR 6310d8315ddSYiFei Zhu depends on PROC_FS 6320d8315ddSYiFei Zhu help 6330d8315ddSYiFei Zhu This enables the /proc/pid/seccomp_cache interface to monitor 6340d8315ddSYiFei Zhu seccomp cache data. The file format is subject to change. Reading 6350d8315ddSYiFei Zhu the file requires CAP_SYS_ADMIN. 6360d8315ddSYiFei Zhu 6370d8315ddSYiFei Zhu This option is for debugging only. Enabling presents the risk that 6380d8315ddSYiFei Zhu an adversary may be able to infer the seccomp filter logic. 6390d8315ddSYiFei Zhu 6400d8315ddSYiFei Zhu If unsure, say N. 6410d8315ddSYiFei Zhu 642afaef01cSAlexander Popovconfig HAVE_ARCH_STACKLEAK 643afaef01cSAlexander Popov bool 644afaef01cSAlexander Popov help 645afaef01cSAlexander Popov An architecture should select this if it has the code which 646afaef01cSAlexander Popov fills the used part of the kernel stack with the STACKLEAK_POISON 647afaef01cSAlexander Popov value before returning from system calls. 648afaef01cSAlexander Popov 649d148eac0SMasahiro Yamadaconfig HAVE_STACKPROTECTOR 65019952a92SKees Cook bool 65119952a92SKees Cook help 65219952a92SKees Cook An arch should select this symbol if: 65319952a92SKees Cook - it has implemented a stack canary (e.g. __stack_chk_guard) 65419952a92SKees Cook 655050e9baaSLinus Torvaldsconfig STACKPROTECTOR 6562a61f474SMasahiro Yamada bool "Stack Protector buffer overflow detection" 657d148eac0SMasahiro Yamada depends on HAVE_STACKPROTECTOR 6582a61f474SMasahiro Yamada depends on $(cc-option,-fstack-protector) 6592a61f474SMasahiro Yamada default y 6608779657dSKees Cook help 6618779657dSKees Cook This option turns on the "stack-protector" GCC feature. This 66219952a92SKees Cook feature puts, at the beginning of functions, a canary value on 66319952a92SKees Cook the stack just before the return address, and validates 66419952a92SKees Cook the value just before actually returning. Stack based buffer 66519952a92SKees Cook overflows (that need to overwrite this return address) now also 66619952a92SKees Cook overwrite the canary, which gets detected and the attack is then 66719952a92SKees Cook neutralized via a kernel panic. 66819952a92SKees Cook 6698779657dSKees Cook Functions will have the stack-protector canary logic added if they 6708779657dSKees Cook have an 8-byte or larger character array on the stack. 6718779657dSKees Cook 67219952a92SKees Cook This feature requires gcc version 4.2 or above, or a distribution 6738779657dSKees Cook gcc with the feature backported ("-fstack-protector"). 6748779657dSKees Cook 6758779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 6768779657dSKees Cook about 3% of all kernel functions, which increases kernel code size 6778779657dSKees Cook by about 0.3%. 6788779657dSKees Cook 679050e9baaSLinus Torvaldsconfig STACKPROTECTOR_STRONG 6802a61f474SMasahiro Yamada bool "Strong Stack Protector" 681050e9baaSLinus Torvalds depends on STACKPROTECTOR 6822a61f474SMasahiro Yamada depends on $(cc-option,-fstack-protector-strong) 6832a61f474SMasahiro Yamada default y 6848779657dSKees Cook help 6858779657dSKees Cook Functions will have the stack-protector canary logic added in any 6868779657dSKees Cook of the following conditions: 6878779657dSKees Cook 6888779657dSKees Cook - local variable's address used as part of the right hand side of an 6898779657dSKees Cook assignment or function argument 6908779657dSKees Cook - local variable is an array (or union containing an array), 6918779657dSKees Cook regardless of array type or length 6928779657dSKees Cook - uses register local variables 6938779657dSKees Cook 6948779657dSKees Cook This feature requires gcc version 4.9 or above, or a distribution 6958779657dSKees Cook gcc with the feature backported ("-fstack-protector-strong"). 6968779657dSKees Cook 6978779657dSKees Cook On an x86 "defconfig" build, this feature adds canary checks to 6988779657dSKees Cook about 20% of all kernel functions, which increases the kernel code 6998779657dSKees Cook size by about 2%. 7008779657dSKees Cook 701d08b9f0cSSami Tolvanenconfig ARCH_SUPPORTS_SHADOW_CALL_STACK 702d08b9f0cSSami Tolvanen bool 703d08b9f0cSSami Tolvanen help 704afcf5441SDan Li An architecture should select this if it supports the compiler's 705afcf5441SDan Li Shadow Call Stack and implements runtime support for shadow stack 706aa7a65aeSWill Deacon switching. 707d08b9f0cSSami Tolvanen 708d08b9f0cSSami Tolvanenconfig SHADOW_CALL_STACK 709afcf5441SDan Li bool "Shadow Call Stack" 710afcf5441SDan Li depends on ARCH_SUPPORTS_SHADOW_CALL_STACK 71138792972SArd Biesheuvel depends on DYNAMIC_FTRACE_WITH_ARGS || DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER 7126f9dc684SSamuel Holland depends on MMU 713d08b9f0cSSami Tolvanen help 714afcf5441SDan Li This option enables the compiler's Shadow Call Stack, which 715afcf5441SDan Li uses a shadow stack to protect function return addresses from 716afcf5441SDan Li being overwritten by an attacker. More information can be found 717afcf5441SDan Li in the compiler's documentation: 718d08b9f0cSSami Tolvanen 719afcf5441SDan Li - Clang: https://clang.llvm.org/docs/ShadowCallStack.html 720afcf5441SDan Li - GCC: https://gcc.gnu.org/onlinedocs/gcc/Instrumentation-Options.html#Instrumentation-Options 721d08b9f0cSSami Tolvanen 722d08b9f0cSSami Tolvanen Note that security guarantees in the kernel differ from the 723d08b9f0cSSami Tolvanen ones documented for user space. The kernel must store addresses 724d08b9f0cSSami Tolvanen of shadow stacks in memory, which means an attacker capable of 725d08b9f0cSSami Tolvanen reading and writing arbitrary memory may be able to locate them 726d08b9f0cSSami Tolvanen and hijack control flow by modifying the stacks. 727d08b9f0cSSami Tolvanen 7289beccca0SArd Biesheuvelconfig DYNAMIC_SCS 7299beccca0SArd Biesheuvel bool 7309beccca0SArd Biesheuvel help 7319beccca0SArd Biesheuvel Set by the arch code if it relies on code patching to insert the 7329beccca0SArd Biesheuvel shadow call stack push and pop instructions rather than on the 7339beccca0SArd Biesheuvel compiler. 7349beccca0SArd Biesheuvel 735dc5723b0SSami Tolvanenconfig LTO 736dc5723b0SSami Tolvanen bool 737dc5723b0SSami Tolvanen help 738dc5723b0SSami Tolvanen Selected if the kernel will be built using the compiler's LTO feature. 739dc5723b0SSami Tolvanen 740dc5723b0SSami Tolvanenconfig LTO_CLANG 741dc5723b0SSami Tolvanen bool 742dc5723b0SSami Tolvanen select LTO 743dc5723b0SSami Tolvanen help 744dc5723b0SSami Tolvanen Selected if the kernel will be built using Clang's LTO feature. 745dc5723b0SSami Tolvanen 746dc5723b0SSami Tolvanenconfig ARCH_SUPPORTS_LTO_CLANG 747dc5723b0SSami Tolvanen bool 748dc5723b0SSami Tolvanen help 749dc5723b0SSami Tolvanen An architecture should select this option if it supports: 750dc5723b0SSami Tolvanen - compiling with Clang, 751dc5723b0SSami Tolvanen - compiling inline assembly with Clang's integrated assembler, 752dc5723b0SSami Tolvanen - and linking with LLD. 753dc5723b0SSami Tolvanen 754dc5723b0SSami Tolvanenconfig ARCH_SUPPORTS_LTO_CLANG_THIN 755dc5723b0SSami Tolvanen bool 756dc5723b0SSami Tolvanen help 757dc5723b0SSami Tolvanen An architecture should select this option if it can support Clang's 758dc5723b0SSami Tolvanen ThinLTO mode. 759dc5723b0SSami Tolvanen 760dc5723b0SSami Tolvanenconfig HAS_LTO_CLANG 761dc5723b0SSami Tolvanen def_bool y 7621e68a8afSNathan Chancellor depends on CC_IS_CLANG && LD_IS_LLD && AS_IS_LLVM 763dc5723b0SSami Tolvanen depends on $(success,$(NM) --help | head -n 1 | grep -qi llvm) 764dc5723b0SSami Tolvanen depends on $(success,$(AR) --help | head -n 1 | grep -qi llvm) 765dc5723b0SSami Tolvanen depends on ARCH_SUPPORTS_LTO_CLANG 766dc5723b0SSami Tolvanen depends on !FTRACE_MCOUNT_USE_RECORDMCOUNT 767349fde59SJakob Koschel # https://github.com/ClangBuiltLinux/linux/issues/1721 768349fde59SJakob Koschel depends on (!KASAN || KASAN_HW_TAGS || CLANG_VERSION >= 170000) || !DEBUG_INFO 769349fde59SJakob Koschel depends on (!KCOV || CLANG_VERSION >= 170000) || !DEBUG_INFO 770dc5723b0SSami Tolvanen depends on !GCOV_KERNEL 771dc5723b0SSami Tolvanen help 772dc5723b0SSami Tolvanen The compiler and Kconfig options support building with Clang's 773dc5723b0SSami Tolvanen LTO. 774dc5723b0SSami Tolvanen 775dc5723b0SSami Tolvanenchoice 776dc5723b0SSami Tolvanen prompt "Link Time Optimization (LTO)" 777dc5723b0SSami Tolvanen default LTO_NONE 778dc5723b0SSami Tolvanen help 779dc5723b0SSami Tolvanen This option enables Link Time Optimization (LTO), which allows the 780dc5723b0SSami Tolvanen compiler to optimize binaries globally. 781dc5723b0SSami Tolvanen 782dc5723b0SSami Tolvanen If unsure, select LTO_NONE. Note that LTO is very resource-intensive 783dc5723b0SSami Tolvanen so it's disabled by default. 784dc5723b0SSami Tolvanen 785dc5723b0SSami Tolvanenconfig LTO_NONE 786dc5723b0SSami Tolvanen bool "None" 787dc5723b0SSami Tolvanen help 788dc5723b0SSami Tolvanen Build the kernel normally, without Link Time Optimization (LTO). 789dc5723b0SSami Tolvanen 790dc5723b0SSami Tolvanenconfig LTO_CLANG_FULL 791dc5723b0SSami Tolvanen bool "Clang Full LTO (EXPERIMENTAL)" 792dc5723b0SSami Tolvanen depends on HAS_LTO_CLANG 793dc5723b0SSami Tolvanen depends on !COMPILE_TEST 794dc5723b0SSami Tolvanen select LTO_CLANG 795dc5723b0SSami Tolvanen help 796dc5723b0SSami Tolvanen This option enables Clang's full Link Time Optimization (LTO), which 797dc5723b0SSami Tolvanen allows the compiler to optimize the kernel globally. If you enable 798dc5723b0SSami Tolvanen this option, the compiler generates LLVM bitcode instead of ELF 799dc5723b0SSami Tolvanen object files, and the actual compilation from bitcode happens at 800dc5723b0SSami Tolvanen the LTO link step, which may take several minutes depending on the 801dc5723b0SSami Tolvanen kernel configuration. More information can be found from LLVM's 802dc5723b0SSami Tolvanen documentation: 803dc5723b0SSami Tolvanen 804dc5723b0SSami Tolvanen https://llvm.org/docs/LinkTimeOptimization.html 805dc5723b0SSami Tolvanen 806dc5723b0SSami Tolvanen During link time, this option can use a large amount of RAM, and 807dc5723b0SSami Tolvanen may take much longer than the ThinLTO option. 808dc5723b0SSami Tolvanen 809dc5723b0SSami Tolvanenconfig LTO_CLANG_THIN 810dc5723b0SSami Tolvanen bool "Clang ThinLTO (EXPERIMENTAL)" 811dc5723b0SSami Tolvanen depends on HAS_LTO_CLANG && ARCH_SUPPORTS_LTO_CLANG_THIN 812dc5723b0SSami Tolvanen select LTO_CLANG 813dc5723b0SSami Tolvanen help 814dc5723b0SSami Tolvanen This option enables Clang's ThinLTO, which allows for parallel 815dc5723b0SSami Tolvanen optimization and faster incremental compiles compared to the 816dc5723b0SSami Tolvanen CONFIG_LTO_CLANG_FULL option. More information can be found 817dc5723b0SSami Tolvanen from Clang's documentation: 818dc5723b0SSami Tolvanen 819dc5723b0SSami Tolvanen https://clang.llvm.org/docs/ThinLTO.html 820dc5723b0SSami Tolvanen 821dc5723b0SSami Tolvanen If unsure, say Y. 822dc5723b0SSami Tolvanenendchoice 823dc5723b0SSami Tolvanen 824315ad878SRong Xuconfig ARCH_SUPPORTS_AUTOFDO_CLANG 825315ad878SRong Xu bool 826315ad878SRong Xu 827315ad878SRong Xuconfig AUTOFDO_CLANG 828315ad878SRong Xu bool "Enable Clang's AutoFDO build (EXPERIMENTAL)" 829315ad878SRong Xu depends on ARCH_SUPPORTS_AUTOFDO_CLANG 830315ad878SRong Xu depends on CC_IS_CLANG && CLANG_VERSION >= 170000 831315ad878SRong Xu help 832315ad878SRong Xu This option enables Clang’s AutoFDO build. When 833315ad878SRong Xu an AutoFDO profile is specified in variable 834315ad878SRong Xu CLANG_AUTOFDO_PROFILE during the build process, 835315ad878SRong Xu Clang uses the profile to optimize the kernel. 836315ad878SRong Xu 837315ad878SRong Xu If no profile is specified, AutoFDO options are 838315ad878SRong Xu still passed to Clang to facilitate the collection 839315ad878SRong Xu of perf data for creating an AutoFDO profile in 840315ad878SRong Xu subsequent builds. 841315ad878SRong Xu 842315ad878SRong Xu If unsure, say N. 843315ad878SRong Xu 844d5dc9583SRong Xuconfig ARCH_SUPPORTS_PROPELLER_CLANG 845d5dc9583SRong Xu bool 846d5dc9583SRong Xu 847d5dc9583SRong Xuconfig PROPELLER_CLANG 848d5dc9583SRong Xu bool "Enable Clang's Propeller build" 849d5dc9583SRong Xu depends on ARCH_SUPPORTS_PROPELLER_CLANG 850d5dc9583SRong Xu depends on CC_IS_CLANG && CLANG_VERSION >= 190000 851d5dc9583SRong Xu help 852d5dc9583SRong Xu This option enables Clang’s Propeller build. When the Propeller 853d5dc9583SRong Xu profiles is specified in variable CLANG_PROPELLER_PROFILE_PREFIX 854d5dc9583SRong Xu during the build process, Clang uses the profiles to optimize 855d5dc9583SRong Xu the kernel. 856d5dc9583SRong Xu 857d5dc9583SRong Xu If no profile is specified, Propeller options are still passed 858d5dc9583SRong Xu to Clang to facilitate the collection of perf data for creating 859d5dc9583SRong Xu the Propeller profiles in subsequent builds. 860d5dc9583SRong Xu 861d5dc9583SRong Xu If unsure, say N. 862d5dc9583SRong Xu 863cf68fffbSSami Tolvanenconfig ARCH_SUPPORTS_CFI_CLANG 864cf68fffbSSami Tolvanen bool 865cf68fffbSSami Tolvanen help 866cf68fffbSSami Tolvanen An architecture should select this option if it can support Clang's 867cf68fffbSSami Tolvanen Control-Flow Integrity (CFI) checking. 868cf68fffbSSami Tolvanen 86989245600SSami Tolvanenconfig ARCH_USES_CFI_TRAPS 87089245600SSami Tolvanen bool 87189245600SSami Tolvanen 872cf68fffbSSami Tolvanenconfig CFI_CLANG 873cf68fffbSSami Tolvanen bool "Use Clang's Control Flow Integrity (CFI)" 87489245600SSami Tolvanen depends on ARCH_SUPPORTS_CFI_CLANG 87589245600SSami Tolvanen depends on $(cc-option,-fsanitize=kcfi) 876cf68fffbSSami Tolvanen help 877c4ca2276SLiu Song This option enables Clang's forward-edge Control Flow Integrity 878cf68fffbSSami Tolvanen (CFI) checking, where the compiler injects a runtime check to each 879cf68fffbSSami Tolvanen indirect function call to ensure the target is a valid function with 880cf68fffbSSami Tolvanen the correct static type. This restricts possible call targets and 881cf68fffbSSami Tolvanen makes it more difficult for an attacker to exploit bugs that allow 882cf68fffbSSami Tolvanen the modification of stored function pointers. More information can be 883cf68fffbSSami Tolvanen found from Clang's documentation: 884cf68fffbSSami Tolvanen 885cf68fffbSSami Tolvanen https://clang.llvm.org/docs/ControlFlowIntegrity.html 886cf68fffbSSami Tolvanen 887ce4a2620SAlice Ryhlconfig CFI_ICALL_NORMALIZE_INTEGERS 888ce4a2620SAlice Ryhl bool "Normalize CFI tags for integers" 889ce4a2620SAlice Ryhl depends on CFI_CLANG 8908b8ca9c2SAlice Ryhl depends on HAVE_CFI_ICALL_NORMALIZE_INTEGERS_CLANG 891ce4a2620SAlice Ryhl help 892ce4a2620SAlice Ryhl This option normalizes the CFI tags for integer types so that all 893ce4a2620SAlice Ryhl integer types of the same size and signedness receive the same CFI 894ce4a2620SAlice Ryhl tag. 895ce4a2620SAlice Ryhl 896ce4a2620SAlice Ryhl The option is separate from CONFIG_RUST because it affects the ABI. 897ce4a2620SAlice Ryhl When working with build systems that care about the ABI, it is 898ce4a2620SAlice Ryhl convenient to be able to turn on this flag first, before Rust is 899ce4a2620SAlice Ryhl turned on. 900ce4a2620SAlice Ryhl 901ce4a2620SAlice Ryhl This option is necessary for using CFI with Rust. If unsure, say N. 902ce4a2620SAlice Ryhl 9038b8ca9c2SAlice Ryhlconfig HAVE_CFI_ICALL_NORMALIZE_INTEGERS_CLANG 9048b8ca9c2SAlice Ryhl def_bool y 9054c66f830SAlice Ryhl depends on $(cc-option,-fsanitize=kcfi -fsanitize-cfi-icall-experimental-normalize-integers) 9068b8ca9c2SAlice Ryhl # With GCOV/KASAN we need this fix: https://github.com/llvm/llvm-project/pull/104826 9072313ab74SAlice Ryhl depends on CLANG_VERSION >= 190103 || (!GCOV_KERNEL && !KASAN_GENERIC && !KASAN_SW_TAGS) 9084c66f830SAlice Ryhl 9098b8ca9c2SAlice Ryhlconfig HAVE_CFI_ICALL_NORMALIZE_INTEGERS_RUSTC 9108b8ca9c2SAlice Ryhl def_bool y 9118b8ca9c2SAlice Ryhl depends on HAVE_CFI_ICALL_NORMALIZE_INTEGERS_CLANG 9128b8ca9c2SAlice Ryhl depends on RUSTC_VERSION >= 107900 9138b8ca9c2SAlice Ryhl # With GCOV/KASAN we need this fix: https://github.com/rust-lang/rust/pull/129373 9142313ab74SAlice Ryhl depends on (RUSTC_LLVM_VERSION >= 190103 && RUSTC_VERSION >= 108200) || \ 9158b8ca9c2SAlice Ryhl (!GCOV_KERNEL && !KASAN_GENERIC && !KASAN_SW_TAGS) 9164c66f830SAlice Ryhl 917cf68fffbSSami Tolvanenconfig CFI_PERMISSIVE 918cf68fffbSSami Tolvanen bool "Use CFI in permissive mode" 919cf68fffbSSami Tolvanen depends on CFI_CLANG 920cf68fffbSSami Tolvanen help 921cf68fffbSSami Tolvanen When selected, Control Flow Integrity (CFI) violations result in a 922cf68fffbSSami Tolvanen warning instead of a kernel panic. This option should only be used 923cf68fffbSSami Tolvanen for finding indirect call type mismatches during development. 924cf68fffbSSami Tolvanen 925cf68fffbSSami Tolvanen If unsure, say N. 926cf68fffbSSami Tolvanen 9270f60a8efSKees Cookconfig HAVE_ARCH_WITHIN_STACK_FRAMES 9280f60a8efSKees Cook bool 9290f60a8efSKees Cook help 9300f60a8efSKees Cook An architecture should select this if it can walk the kernel stack 9310f60a8efSKees Cook frames to determine if an object is part of either the arguments 9320f60a8efSKees Cook or local variables (i.e. that it excludes saved return addresses, 9330f60a8efSKees Cook and similar) by implementing an inline arch_within_stack_frames(), 9340f60a8efSKees Cook which is used by CONFIG_HARDENED_USERCOPY. 9350f60a8efSKees Cook 93624a9c541SFrederic Weisbeckerconfig HAVE_CONTEXT_TRACKING_USER 9372b1d5024SFrederic Weisbecker bool 9382b1d5024SFrederic Weisbecker help 93991d1aa43SFrederic Weisbecker Provide kernel/user boundaries probes necessary for subsystems 94091d1aa43SFrederic Weisbecker that need it, such as userspace RCU extended quiescent state. 941490f561bSFrederic Weisbecker Syscalls need to be wrapped inside user_exit()-user_enter(), either 942490f561bSFrederic Weisbecker optimized behind static key or through the slow path using TIF_NOHZ 943490f561bSFrederic Weisbecker flag. Exceptions handlers must be wrapped as well. Irqs are already 9446f0e6c15SFrederic Weisbecker protected inside ct_irq_enter/ct_irq_exit() but preemption or signal 945490f561bSFrederic Weisbecker handling on irq exit still need to be protected. 946490f561bSFrederic Weisbecker 94724a9c541SFrederic Weisbeckerconfig HAVE_CONTEXT_TRACKING_USER_OFFSTACK 94883c2da2eSFrederic Weisbecker bool 94983c2da2eSFrederic Weisbecker help 95083c2da2eSFrederic Weisbecker Architecture neither relies on exception_enter()/exception_exit() 95183c2da2eSFrederic Weisbecker nor on schedule_user(). Also preempt_schedule_notrace() and 95283c2da2eSFrederic Weisbecker preempt_schedule_irq() can't be called in a preemptible section 953d65d411cSValentin Schneider while context tracking is CT_STATE_USER. This feature reflects a sane 95483c2da2eSFrederic Weisbecker entry implementation where the following requirements are met on 95583c2da2eSFrederic Weisbecker critical entry code, ie: before user_exit() or after user_enter(): 95683c2da2eSFrederic Weisbecker 95783c2da2eSFrederic Weisbecker - Critical entry code isn't preemptible (or better yet: 95883c2da2eSFrederic Weisbecker not interruptible). 959493c1822SFrederic Weisbecker - No use of RCU read side critical sections, unless ct_nmi_enter() 96083c2da2eSFrederic Weisbecker got called. 96183c2da2eSFrederic Weisbecker - No use of instrumentation, unless instrumentation_begin() got 96283c2da2eSFrederic Weisbecker called. 96383c2da2eSFrederic Weisbecker 964490f561bSFrederic Weisbeckerconfig HAVE_TIF_NOHZ 965490f561bSFrederic Weisbecker bool 966490f561bSFrederic Weisbecker help 967490f561bSFrederic Weisbecker Arch relies on TIF_NOHZ and syscall slow path to implement context 968490f561bSFrederic Weisbecker tracking calls to user_enter()/user_exit(). 9692b1d5024SFrederic Weisbecker 970b952741cSFrederic Weisbeckerconfig HAVE_VIRT_CPU_ACCOUNTING 971b952741cSFrederic Weisbecker bool 972b952741cSFrederic Weisbecker 9732b91ec9fSFrederic Weisbeckerconfig HAVE_VIRT_CPU_ACCOUNTING_IDLE 9742b91ec9fSFrederic Weisbecker bool 9752b91ec9fSFrederic Weisbecker help 9762b91ec9fSFrederic Weisbecker Architecture has its own way to account idle CPU time and therefore 9772b91ec9fSFrederic Weisbecker doesn't implement vtime_account_idle(). 9782b91ec9fSFrederic Weisbecker 97940565b5aSStanislaw Gruszkaconfig ARCH_HAS_SCALED_CPUTIME 98040565b5aSStanislaw Gruszka bool 98140565b5aSStanislaw Gruszka 982554b0004SKevin Hilmanconfig HAVE_VIRT_CPU_ACCOUNTING_GEN 983554b0004SKevin Hilman bool 984554b0004SKevin Hilman default y if 64BIT 985554b0004SKevin Hilman help 986554b0004SKevin Hilman With VIRT_CPU_ACCOUNTING_GEN, cputime_t becomes 64-bit. 987554b0004SKevin Hilman Before enabling this option, arch code must be audited 988554b0004SKevin Hilman to ensure there are no races in concurrent read/write of 989554b0004SKevin Hilman cputime_t. For example, reading/writing 64-bit cputime_t on 990554b0004SKevin Hilman some 32-bit arches may require multiple accesses, so proper 991554b0004SKevin Hilman locking is needed to protect against concurrent accesses. 992554b0004SKevin Hilman 993fdf9c356SFrederic Weisbeckerconfig HAVE_IRQ_TIME_ACCOUNTING 994fdf9c356SFrederic Weisbecker bool 995fdf9c356SFrederic Weisbecker help 996fdf9c356SFrederic Weisbecker Archs need to ensure they use a high enough resolution clock to 997fdf9c356SFrederic Weisbecker support irq time accounting and then call enable_sched_clock_irqtime(). 998fdf9c356SFrederic Weisbecker 999c49dd340SKalesh Singhconfig HAVE_MOVE_PUD 1000c49dd340SKalesh Singh bool 1001c49dd340SKalesh Singh help 1002c49dd340SKalesh Singh Architectures that select this are able to move page tables at the 1003c49dd340SKalesh Singh PUD level. If there are only 3 page table levels, the move effectively 1004c49dd340SKalesh Singh happens at the PGD level. 1005c49dd340SKalesh Singh 10062c91bd4aSJoel Fernandes (Google)config HAVE_MOVE_PMD 10072c91bd4aSJoel Fernandes (Google) bool 10082c91bd4aSJoel Fernandes (Google) help 10092c91bd4aSJoel Fernandes (Google) Archs that select this are able to move page tables at the PMD level. 10102c91bd4aSJoel Fernandes (Google) 101115626062SGerald Schaeferconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE 101215626062SGerald Schaefer bool 101315626062SGerald Schaefer 1014a00cc7d9SMatthew Wilcoxconfig HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD 1015a00cc7d9SMatthew Wilcox bool 1016a00cc7d9SMatthew Wilcox 10170ddab1d2SToshi Kaniconfig HAVE_ARCH_HUGE_VMAP 10180ddab1d2SToshi Kani bool 10190ddab1d2SToshi Kani 1020121e6f32SNicholas Piggin# 1021121e6f32SNicholas Piggin# Archs that select this would be capable of PMD-sized vmaps (i.e., 1022559089e0SSong Liu# arch_vmap_pmd_supported() returns true). The VM_ALLOW_HUGE_VMAP flag 1023559089e0SSong Liu# must be used to enable allocations to use hugepages. 1024121e6f32SNicholas Piggin# 1025121e6f32SNicholas Pigginconfig HAVE_ARCH_HUGE_VMALLOC 1026121e6f32SNicholas Piggin depends on HAVE_ARCH_HUGE_VMAP 1027121e6f32SNicholas Piggin bool 1028121e6f32SNicholas Piggin 10293876d4a3SAlexandre Ghiticonfig ARCH_WANT_HUGE_PMD_SHARE 10303876d4a3SAlexandre Ghiti bool 10313876d4a3SAlexandre Ghiti 10322f0584f3SRick Edgecombe# Archs that want to use pmd_mkwrite on kernel memory need it defined even 10332f0584f3SRick Edgecombe# if there are no userspace memory management features that use it 10342f0584f3SRick Edgecombeconfig ARCH_WANT_KERNEL_PMD_MKWRITE 10352f0584f3SRick Edgecombe bool 10362f0584f3SRick Edgecombe 10372f0584f3SRick Edgecombeconfig ARCH_WANT_PMD_MKWRITE 10382f0584f3SRick Edgecombe def_bool TRANSPARENT_HUGEPAGE || ARCH_WANT_KERNEL_PMD_MKWRITE 10392f0584f3SRick Edgecombe 10400f8975ecSPavel Emelyanovconfig HAVE_ARCH_SOFT_DIRTY 10410f8975ecSPavel Emelyanov bool 10420f8975ecSPavel Emelyanov 1043786d35d4SDavid Howellsconfig HAVE_MOD_ARCH_SPECIFIC 1044786d35d4SDavid Howells bool 1045786d35d4SDavid Howells help 1046786d35d4SDavid Howells The arch uses struct mod_arch_specific to store data. Many arches 1047786d35d4SDavid Howells just need a simple module loader without arch specific data - those 1048786d35d4SDavid Howells should not enable this. 1049786d35d4SDavid Howells 1050786d35d4SDavid Howellsconfig MODULES_USE_ELF_RELA 1051786d35d4SDavid Howells bool 1052786d35d4SDavid Howells help 1053786d35d4SDavid Howells Modules only use ELF RELA relocations. Modules with ELF REL 1054786d35d4SDavid Howells relocations will give an error. 1055786d35d4SDavid Howells 1056786d35d4SDavid Howellsconfig MODULES_USE_ELF_REL 1057786d35d4SDavid Howells bool 1058786d35d4SDavid Howells help 1059786d35d4SDavid Howells Modules only use ELF REL relocations. Modules with ELF RELA 1060786d35d4SDavid Howells relocations will give an error. 1061786d35d4SDavid Howells 106201dc0386SChristophe Leroyconfig ARCH_WANTS_MODULES_DATA_IN_VMALLOC 106301dc0386SChristophe Leroy bool 106401dc0386SChristophe Leroy help 106501dc0386SChristophe Leroy For architectures like powerpc/32 which have constraints on module 106601dc0386SChristophe Leroy allocation and need to allocate module data outside of module area. 106701dc0386SChristophe Leroy 1068223b5e57SMike Rapoport (IBM)config ARCH_WANTS_EXECMEM_LATE 1069223b5e57SMike Rapoport (IBM) bool 1070223b5e57SMike Rapoport (IBM) help 1071223b5e57SMike Rapoport (IBM) For architectures that do not allocate executable memory early on 1072223b5e57SMike Rapoport (IBM) boot, but rather require its initialization late when there is 1073223b5e57SMike Rapoport (IBM) enough entropy for module space randomization, for instance 1074223b5e57SMike Rapoport (IBM) arm64. 1075223b5e57SMike Rapoport (IBM) 10762e45474aSMike Rapoport (Microsoft)config ARCH_HAS_EXECMEM_ROX 10772e45474aSMike Rapoport (Microsoft) bool 10782e45474aSMike Rapoport (Microsoft) depends on MMU && !HIGHMEM 10792e45474aSMike Rapoport (Microsoft) help 10802e45474aSMike Rapoport (Microsoft) For architectures that support allocations of executable memory 10812e45474aSMike Rapoport (Microsoft) with read-only execute permissions. Architecture must implement 10822e45474aSMike Rapoport (Microsoft) execmem_fill_trapping_insns() callback to enable this. 10832e45474aSMike Rapoport (Microsoft) 1084cc1f0274SFrederic Weisbeckerconfig HAVE_IRQ_EXIT_ON_IRQ_STACK 1085cc1f0274SFrederic Weisbecker bool 1086cc1f0274SFrederic Weisbecker help 1087cc1f0274SFrederic Weisbecker Architecture doesn't only execute the irq handler on the irq stack 1088cc1f0274SFrederic Weisbecker but also irq_exit(). This way we can process softirqs on this irq 1089cc1f0274SFrederic Weisbecker stack instead of switching to a new one when we call __do_softirq() 1090cc1f0274SFrederic Weisbecker in the end of an hardirq. 1091cc1f0274SFrederic Weisbecker This spares a stack switch and improves cache usage on softirq 1092cc1f0274SFrederic Weisbecker processing. 1093cc1f0274SFrederic Weisbecker 1094cd1a41ceSThomas Gleixnerconfig HAVE_SOFTIRQ_ON_OWN_STACK 1095cd1a41ceSThomas Gleixner bool 1096cd1a41ceSThomas Gleixner help 1097cd1a41ceSThomas Gleixner Architecture provides a function to run __do_softirq() on a 1098c226bc3cSColin Ian King separate stack. 1099cd1a41ceSThomas Gleixner 11008cbb2b50SSebastian Andrzej Siewiorconfig SOFTIRQ_ON_OWN_STACK 11018cbb2b50SSebastian Andrzej Siewior def_bool HAVE_SOFTIRQ_ON_OWN_STACK && !PREEMPT_RT 11028cbb2b50SSebastian Andrzej Siewior 110312700c17SArnd Bergmannconfig ALTERNATE_USER_ADDRESS_SPACE 110412700c17SArnd Bergmann bool 110512700c17SArnd Bergmann help 110612700c17SArnd Bergmann Architectures set this when the CPU uses separate address 110712700c17SArnd Bergmann spaces for kernel and user space pointers. In this case, the 110812700c17SArnd Bergmann access_ok() check on a __user pointer is skipped. 110912700c17SArnd Bergmann 1110235a8f02SKirill A. Shutemovconfig PGTABLE_LEVELS 1111235a8f02SKirill A. Shutemov int 1112235a8f02SKirill A. Shutemov default 2 1113235a8f02SKirill A. Shutemov 11142b68f6caSKees Cookconfig ARCH_HAS_ELF_RANDOMIZE 11152b68f6caSKees Cook bool 11162b68f6caSKees Cook help 11172b68f6caSKees Cook An architecture supports choosing randomized locations for 11182b68f6caSKees Cook stack, mmap, brk, and ET_DYN. Defined functions: 11192b68f6caSKees Cook - arch_mmap_rnd() 1120204db6edSKees Cook - arch_randomize_brk() 11212b68f6caSKees Cook 1122d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_BITS 1123d07e2259SDaniel Cashman bool 1124d07e2259SDaniel Cashman help 1125d07e2259SDaniel Cashman An arch should select this symbol if it supports setting a variable 1126d07e2259SDaniel Cashman number of bits for use in establishing the base address for mmap 1127d07e2259SDaniel Cashman allocations, has MMU enabled and provides values for both: 1128d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MIN 1129d07e2259SDaniel Cashman - ARCH_MMAP_RND_BITS_MAX 1130d07e2259SDaniel Cashman 11315f56a5dfSJiri Slabyconfig HAVE_EXIT_THREAD 11325f56a5dfSJiri Slaby bool 11335f56a5dfSJiri Slaby help 11345f56a5dfSJiri Slaby An architecture implements exit_thread. 11355f56a5dfSJiri Slaby 1136d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MIN 1137d07e2259SDaniel Cashman int 1138d07e2259SDaniel Cashman 1139d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_MAX 1140d07e2259SDaniel Cashman int 1141d07e2259SDaniel Cashman 1142d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS_DEFAULT 1143d07e2259SDaniel Cashman int 1144d07e2259SDaniel Cashman 1145d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_BITS 1146d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address" if EXPERT 1147d07e2259SDaniel Cashman range ARCH_MMAP_RND_BITS_MIN ARCH_MMAP_RND_BITS_MAX 1148d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_DEFAULT if ARCH_MMAP_RND_BITS_DEFAULT 1149d07e2259SDaniel Cashman default ARCH_MMAP_RND_BITS_MIN 1150d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_BITS 1151d07e2259SDaniel Cashman help 1152d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 1153d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 1154d07e2259SDaniel Cashman resulting from mmap allocations. This value will be bounded 1155d07e2259SDaniel Cashman by the architecture's minimum and maximum supported values. 1156d07e2259SDaniel Cashman 1157d07e2259SDaniel Cashman This value can be changed after boot using the 1158d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_bits tunable 1159d07e2259SDaniel Cashman 1160d07e2259SDaniel Cashmanconfig HAVE_ARCH_MMAP_RND_COMPAT_BITS 1161d07e2259SDaniel Cashman bool 1162d07e2259SDaniel Cashman help 1163d07e2259SDaniel Cashman An arch should select this symbol if it supports running applications 1164d07e2259SDaniel Cashman in compatibility mode, supports setting a variable number of bits for 1165d07e2259SDaniel Cashman use in establishing the base address for mmap allocations, has MMU 1166d07e2259SDaniel Cashman enabled and provides values for both: 1167d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MIN 1168d07e2259SDaniel Cashman - ARCH_MMAP_RND_COMPAT_BITS_MAX 1169d07e2259SDaniel Cashman 1170d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MIN 1171d07e2259SDaniel Cashman int 1172d07e2259SDaniel Cashman 1173d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_MAX 1174d07e2259SDaniel Cashman int 1175d07e2259SDaniel Cashman 1176d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 1177d07e2259SDaniel Cashman int 1178d07e2259SDaniel Cashman 1179d07e2259SDaniel Cashmanconfig ARCH_MMAP_RND_COMPAT_BITS 1180d07e2259SDaniel Cashman int "Number of bits to use for ASLR of mmap base address for compatible applications" if EXPERT 1181d07e2259SDaniel Cashman range ARCH_MMAP_RND_COMPAT_BITS_MIN ARCH_MMAP_RND_COMPAT_BITS_MAX 1182d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_DEFAULT if ARCH_MMAP_RND_COMPAT_BITS_DEFAULT 1183d07e2259SDaniel Cashman default ARCH_MMAP_RND_COMPAT_BITS_MIN 1184d07e2259SDaniel Cashman depends on HAVE_ARCH_MMAP_RND_COMPAT_BITS 1185d07e2259SDaniel Cashman help 1186d07e2259SDaniel Cashman This value can be used to select the number of bits to use to 1187d07e2259SDaniel Cashman determine the random offset to the base address of vma regions 1188d07e2259SDaniel Cashman resulting from mmap allocations for compatible applications This 1189d07e2259SDaniel Cashman value will be bounded by the architecture's minimum and maximum 1190d07e2259SDaniel Cashman supported values. 1191d07e2259SDaniel Cashman 1192d07e2259SDaniel Cashman This value can be changed after boot using the 1193d07e2259SDaniel Cashman /proc/sys/vm/mmap_rnd_compat_bits tunable 1194d07e2259SDaniel Cashman 11951b028f78SDmitry Safonovconfig HAVE_ARCH_COMPAT_MMAP_BASES 11961b028f78SDmitry Safonov bool 11971b028f78SDmitry Safonov help 11981b028f78SDmitry Safonov This allows 64bit applications to invoke 32-bit mmap() syscall 11991b028f78SDmitry Safonov and vice-versa 32-bit applications to call 64-bit mmap(). 12001b028f78SDmitry Safonov Required for applications doing different bitness syscalls. 12011b028f78SDmitry Safonov 1202ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_4KB 1203ba89f9c8SArnd Bergmann bool 1204ba89f9c8SArnd Bergmann 1205ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_8KB 1206ba89f9c8SArnd Bergmann bool 1207ba89f9c8SArnd Bergmann 1208ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_16KB 1209ba89f9c8SArnd Bergmann bool 1210ba89f9c8SArnd Bergmann 1211ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_32KB 1212ba89f9c8SArnd Bergmann bool 1213ba89f9c8SArnd Bergmann 1214ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_64KB 1215ba89f9c8SArnd Bergmann bool 1216ba89f9c8SArnd Bergmann 1217ba89f9c8SArnd Bergmannconfig HAVE_PAGE_SIZE_256KB 1218ba89f9c8SArnd Bergmann bool 1219ba89f9c8SArnd Bergmann 1220ba89f9c8SArnd Bergmannchoice 1221ba89f9c8SArnd Bergmann prompt "MMU page size" 1222ba89f9c8SArnd Bergmann 1223ba89f9c8SArnd Bergmannconfig PAGE_SIZE_4KB 1224ba89f9c8SArnd Bergmann bool "4KiB pages" 1225ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_4KB 1226ba89f9c8SArnd Bergmann help 1227ba89f9c8SArnd Bergmann This option select the standard 4KiB Linux page size and the only 1228ba89f9c8SArnd Bergmann available option on many architectures. Using 4KiB page size will 1229ba89f9c8SArnd Bergmann minimize memory consumption and is therefore recommended for low 1230ba89f9c8SArnd Bergmann memory systems. 1231ba89f9c8SArnd Bergmann Some software that is written for x86 systems makes incorrect 1232ba89f9c8SArnd Bergmann assumptions about the page size and only runs on 4KiB pages. 1233ba89f9c8SArnd Bergmann 1234ba89f9c8SArnd Bergmannconfig PAGE_SIZE_8KB 1235ba89f9c8SArnd Bergmann bool "8KiB pages" 1236ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_8KB 1237ba89f9c8SArnd Bergmann help 1238ba89f9c8SArnd Bergmann This option is the only supported page size on a few older 1239ba89f9c8SArnd Bergmann processors, and can be slightly faster than 4KiB pages. 1240ba89f9c8SArnd Bergmann 1241ba89f9c8SArnd Bergmannconfig PAGE_SIZE_16KB 1242ba89f9c8SArnd Bergmann bool "16KiB pages" 1243ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_16KB 1244ba89f9c8SArnd Bergmann help 1245ba89f9c8SArnd Bergmann This option is usually a good compromise between memory 1246ba89f9c8SArnd Bergmann consumption and performance for typical desktop and server 1247ba89f9c8SArnd Bergmann workloads, often saving a level of page table lookups compared 1248ba89f9c8SArnd Bergmann to 4KB pages as well as reducing TLB pressure and overhead of 1249ba89f9c8SArnd Bergmann per-page operations in the kernel at the expense of a larger 1250ba89f9c8SArnd Bergmann page cache. 1251ba89f9c8SArnd Bergmann 1252ba89f9c8SArnd Bergmannconfig PAGE_SIZE_32KB 1253ba89f9c8SArnd Bergmann bool "32KiB pages" 1254ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_32KB 1255ba89f9c8SArnd Bergmann help 1256ba89f9c8SArnd Bergmann Using 32KiB page size will result in slightly higher performance 1257ba89f9c8SArnd Bergmann kernel at the price of higher memory consumption compared to 1258ba89f9c8SArnd Bergmann 16KiB pages. This option is available only on cnMIPS cores. 1259ba89f9c8SArnd Bergmann Note that you will need a suitable Linux distribution to 1260ba89f9c8SArnd Bergmann support this. 1261ba89f9c8SArnd Bergmann 1262ba89f9c8SArnd Bergmannconfig PAGE_SIZE_64KB 1263ba89f9c8SArnd Bergmann bool "64KiB pages" 1264ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_64KB 1265ba89f9c8SArnd Bergmann help 1266ba89f9c8SArnd Bergmann Using 64KiB page size will result in slightly higher performance 1267ba89f9c8SArnd Bergmann kernel at the price of much higher memory consumption compared to 1268ba89f9c8SArnd Bergmann 4KiB or 16KiB pages. 1269ba89f9c8SArnd Bergmann This is not suitable for general-purpose workloads but the 1270ba89f9c8SArnd Bergmann better performance may be worth the cost for certain types of 1271ba89f9c8SArnd Bergmann supercomputing or database applications that work mostly with 1272ba89f9c8SArnd Bergmann large in-memory data rather than small files. 1273ba89f9c8SArnd Bergmann 1274ba89f9c8SArnd Bergmannconfig PAGE_SIZE_256KB 1275ba89f9c8SArnd Bergmann bool "256KiB pages" 1276ba89f9c8SArnd Bergmann depends on HAVE_PAGE_SIZE_256KB 1277ba89f9c8SArnd Bergmann help 1278ba89f9c8SArnd Bergmann 256KiB pages have little practical value due to their extreme 1279ba89f9c8SArnd Bergmann memory usage. The kernel will only be able to run applications 1280ba89f9c8SArnd Bergmann that have been compiled with '-zmax-page-size' set to 256KiB 1281ba89f9c8SArnd Bergmann (the default is 64KiB or 4KiB on most architectures). 1282ba89f9c8SArnd Bergmann 1283ba89f9c8SArnd Bergmannendchoice 1284ba89f9c8SArnd Bergmann 12851f0e290cSGuenter Roeckconfig PAGE_SIZE_LESS_THAN_64KB 12861f0e290cSGuenter Roeck def_bool y 12871f0e290cSGuenter Roeck depends on !PAGE_SIZE_64KB 1288e4bbd20dSNathan Chancellor depends on PAGE_SIZE_LESS_THAN_256KB 1289e4bbd20dSNathan Chancellor 1290e4bbd20dSNathan Chancellorconfig PAGE_SIZE_LESS_THAN_256KB 1291e4bbd20dSNathan Chancellor def_bool y 12921f0e290cSGuenter Roeck depends on !PAGE_SIZE_256KB 12931f0e290cSGuenter Roeck 1294ba89f9c8SArnd Bergmannconfig PAGE_SHIFT 1295ba89f9c8SArnd Bergmann int 1296ba89f9c8SArnd Bergmann default 12 if PAGE_SIZE_4KB 1297ba89f9c8SArnd Bergmann default 13 if PAGE_SIZE_8KB 1298ba89f9c8SArnd Bergmann default 14 if PAGE_SIZE_16KB 1299ba89f9c8SArnd Bergmann default 15 if PAGE_SIZE_32KB 1300ba89f9c8SArnd Bergmann default 16 if PAGE_SIZE_64KB 1301ba89f9c8SArnd Bergmann default 18 if PAGE_SIZE_256KB 1302ba89f9c8SArnd Bergmann 130367f3977fSAlexandre Ghiti# This allows to use a set of generic functions to determine mmap base 130467f3977fSAlexandre Ghiti# address by giving priority to top-down scheme only if the process 130567f3977fSAlexandre Ghiti# is not in legacy mode (compat task, unlimited stack size or 130667f3977fSAlexandre Ghiti# sysctl_legacy_va_layout). 130767f3977fSAlexandre Ghiti# Architecture that selects this option can provide its own version of: 130867f3977fSAlexandre Ghiti# - STACK_RND_MASK 130967f3977fSAlexandre Ghiticonfig ARCH_WANT_DEFAULT_TOPDOWN_MMAP_LAYOUT 131067f3977fSAlexandre Ghiti bool 131167f3977fSAlexandre Ghiti depends on MMU 1312e7142bf5SAlexandre Ghiti select ARCH_HAS_ELF_RANDOMIZE 131367f3977fSAlexandre Ghiti 131403f16cd0SJosh Poimboeufconfig HAVE_OBJTOOL 131503f16cd0SJosh Poimboeuf bool 131603f16cd0SJosh Poimboeuf 13174ab7674fSJosh Poimboeufconfig HAVE_JUMP_LABEL_HACK 13184ab7674fSJosh Poimboeuf bool 13194ab7674fSJosh Poimboeuf 132022102f45SJosh Poimboeufconfig HAVE_NOINSTR_HACK 132122102f45SJosh Poimboeuf bool 132222102f45SJosh Poimboeuf 1323489e355bSJosh Poimboeufconfig HAVE_NOINSTR_VALIDATION 1324489e355bSJosh Poimboeuf bool 1325489e355bSJosh Poimboeuf 13265f3da8c0SJosh Poimboeufconfig HAVE_UACCESS_VALIDATION 13275f3da8c0SJosh Poimboeuf bool 13285f3da8c0SJosh Poimboeuf select OBJTOOL 13295f3da8c0SJosh Poimboeuf 1330b9ab5ebbSJosh Poimboeufconfig HAVE_STACK_VALIDATION 1331b9ab5ebbSJosh Poimboeuf bool 1332b9ab5ebbSJosh Poimboeuf help 133303f16cd0SJosh Poimboeuf Architecture supports objtool compile-time frame pointer rule 133403f16cd0SJosh Poimboeuf validation. 1335b9ab5ebbSJosh Poimboeuf 1336af085d90SJosh Poimboeufconfig HAVE_RELIABLE_STACKTRACE 1337af085d90SJosh Poimboeuf bool 1338af085d90SJosh Poimboeuf help 1339140d7e88SMiroslav Benes Architecture has either save_stack_trace_tsk_reliable() or 1340140d7e88SMiroslav Benes arch_stack_walk_reliable() function which only returns a stack trace 1341140d7e88SMiroslav Benes if it can guarantee the trace is reliable. 1342af085d90SJosh Poimboeuf 1343468a9428SGeorge Spelvinconfig HAVE_ARCH_HASH 1344468a9428SGeorge Spelvin bool 1345468a9428SGeorge Spelvin default n 1346468a9428SGeorge Spelvin help 1347468a9428SGeorge Spelvin If this is set, the architecture provides an <asm/hash.h> 1348468a9428SGeorge Spelvin file which provides platform-specific implementations of some 1349468a9428SGeorge Spelvin functions in <linux/hash.h> or fs/namei.c. 1350468a9428SGeorge Spelvin 1351666047feSFinn Thainconfig HAVE_ARCH_NVRAM_OPS 1352666047feSFinn Thain bool 1353666047feSFinn Thain 13543a495511SWilliam Breathitt Grayconfig ISA_BUS_API 13553a495511SWilliam Breathitt Gray def_bool ISA 13563a495511SWilliam Breathitt Gray 1357d2125043SAl Viro# 1358d2125043SAl Viro# ABI hall of shame 1359d2125043SAl Viro# 1360d2125043SAl Viroconfig CLONE_BACKWARDS 1361d2125043SAl Viro bool 1362d2125043SAl Viro help 1363d2125043SAl Viro Architecture has tls passed as the 4th argument of clone(2), 1364d2125043SAl Viro not the 5th one. 1365d2125043SAl Viro 1366d2125043SAl Viroconfig CLONE_BACKWARDS2 1367d2125043SAl Viro bool 1368d2125043SAl Viro help 1369d2125043SAl Viro Architecture has the first two arguments of clone(2) swapped. 1370d2125043SAl Viro 1371dfa9771aSMichal Simekconfig CLONE_BACKWARDS3 1372dfa9771aSMichal Simek bool 1373dfa9771aSMichal Simek help 1374dfa9771aSMichal Simek Architecture has tls passed as the 3rd argument of clone(2), 1375dfa9771aSMichal Simek not the 5th one. 1376dfa9771aSMichal Simek 1377eaca6eaeSAl Viroconfig ODD_RT_SIGACTION 1378eaca6eaeSAl Viro bool 1379eaca6eaeSAl Viro help 1380eaca6eaeSAl Viro Architecture has unusual rt_sigaction(2) arguments 1381eaca6eaeSAl Viro 13820a0e8cdfSAl Viroconfig OLD_SIGSUSPEND 13830a0e8cdfSAl Viro bool 13840a0e8cdfSAl Viro help 13850a0e8cdfSAl Viro Architecture has old sigsuspend(2) syscall, of one-argument variety 13860a0e8cdfSAl Viro 13870a0e8cdfSAl Viroconfig OLD_SIGSUSPEND3 13880a0e8cdfSAl Viro bool 13890a0e8cdfSAl Viro help 13900a0e8cdfSAl Viro Even weirder antique ABI - three-argument sigsuspend(2) 13910a0e8cdfSAl Viro 1392495dfbf7SAl Viroconfig OLD_SIGACTION 1393495dfbf7SAl Viro bool 1394495dfbf7SAl Viro help 1395495dfbf7SAl Viro Architecture has old sigaction(2) syscall. Nope, not the same 1396495dfbf7SAl Viro as OLD_SIGSUSPEND | OLD_SIGSUSPEND3 - alpha has sigsuspend(2), 1397495dfbf7SAl Viro but fairly different variant of sigaction(2), thanks to OSF/1 1398495dfbf7SAl Viro compatibility... 1399495dfbf7SAl Viro 1400495dfbf7SAl Viroconfig COMPAT_OLD_SIGACTION 1401495dfbf7SAl Viro bool 1402495dfbf7SAl Viro 140317435e5fSDeepa Dinamaniconfig COMPAT_32BIT_TIME 1404942437c9SArnd Bergmann bool "Provide system calls for 32-bit time_t" 1405942437c9SArnd Bergmann default !64BIT || COMPAT 140617435e5fSDeepa Dinamani help 140717435e5fSDeepa Dinamani This enables 32 bit time_t support in addition to 64 bit time_t support. 140817435e5fSDeepa Dinamani This is relevant on all 32-bit architectures, and 64-bit architectures 140917435e5fSDeepa Dinamani as part of compat syscall handling. 141017435e5fSDeepa Dinamani 141187a4c375SChristoph Hellwigconfig ARCH_NO_PREEMPT 141287a4c375SChristoph Hellwig bool 141387a4c375SChristoph Hellwig 1414a50a3f4bSThomas Gleixnerconfig ARCH_SUPPORTS_RT 1415a50a3f4bSThomas Gleixner bool 1416a50a3f4bSThomas Gleixner 1417fff7fb0bSZhaoxiu Zengconfig CPU_NO_EFFICIENT_FFS 1418fff7fb0bSZhaoxiu Zeng def_bool n 1419fff7fb0bSZhaoxiu Zeng 1420ba14a194SAndy Lutomirskiconfig HAVE_ARCH_VMAP_STACK 1421ba14a194SAndy Lutomirski def_bool n 1422ba14a194SAndy Lutomirski help 1423ba14a194SAndy Lutomirski An arch should select this symbol if it can support kernel stacks 1424ba14a194SAndy Lutomirski in vmalloc space. This means: 1425ba14a194SAndy Lutomirski 1426ba14a194SAndy Lutomirski - vmalloc space must be large enough to hold many kernel stacks. 1427ba14a194SAndy Lutomirski This may rule out many 32-bit architectures. 1428ba14a194SAndy Lutomirski 1429ba14a194SAndy Lutomirski - Stacks in vmalloc space need to work reliably. For example, if 1430ba14a194SAndy Lutomirski vmap page tables are created on demand, either this mechanism 1431ba14a194SAndy Lutomirski needs to work while the stack points to a virtual address with 1432ba14a194SAndy Lutomirski unpopulated page tables or arch code (switch_to() and switch_mm(), 1433ba14a194SAndy Lutomirski most likely) needs to ensure that the stack's page table entries 1434ba14a194SAndy Lutomirski are populated before running on a possibly unpopulated stack. 1435ba14a194SAndy Lutomirski 1436ba14a194SAndy Lutomirski - If the stack overflows into a guard page, something reasonable 1437ba14a194SAndy Lutomirski should happen. The definition of "reasonable" is flexible, but 1438ba14a194SAndy Lutomirski instantly rebooting without logging anything would be unfriendly. 1439ba14a194SAndy Lutomirski 1440ba14a194SAndy Lutomirskiconfig VMAP_STACK 1441ba14a194SAndy Lutomirski default y 1442ba14a194SAndy Lutomirski bool "Use a virtually-mapped stack" 1443eafb149eSDaniel Axtens depends on HAVE_ARCH_VMAP_STACK 144438dd767dSAndrey Konovalov depends on !KASAN || KASAN_HW_TAGS || KASAN_VMALLOC 1445a7f7f624SMasahiro Yamada help 1446ba14a194SAndy Lutomirski Enable this if you want the use virtually-mapped kernel stacks 1447ba14a194SAndy Lutomirski with guard pages. This causes kernel stack overflows to be 1448ba14a194SAndy Lutomirski caught immediately rather than causing difficult-to-diagnose 1449ba14a194SAndy Lutomirski corruption. 1450ba14a194SAndy Lutomirski 145138dd767dSAndrey Konovalov To use this with software KASAN modes, the architecture must support 145238dd767dSAndrey Konovalov backing virtual mappings with real shadow memory, and KASAN_VMALLOC 145338dd767dSAndrey Konovalov must be enabled. 1454ba14a194SAndy Lutomirski 145539218ff4SKees Cookconfig HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 145639218ff4SKees Cook def_bool n 145739218ff4SKees Cook help 145839218ff4SKees Cook An arch should select this symbol if it can support kernel stack 145939218ff4SKees Cook offset randomization with calls to add_random_kstack_offset() 146039218ff4SKees Cook during syscall entry and choose_random_kstack_offset() during 146139218ff4SKees Cook syscall exit. Careful removal of -fstack-protector-strong and 146239218ff4SKees Cook -fstack-protector should also be applied to the entry code and 146339218ff4SKees Cook closely examined, as the artificial stack bump looks like an array 146439218ff4SKees Cook to the compiler, so it will attempt to add canary checks regardless 146539218ff4SKees Cook of the static branch state. 146639218ff4SKees Cook 14678cb37a59SMarco Elverconfig RANDOMIZE_KSTACK_OFFSET 14688cb37a59SMarco Elver bool "Support for randomizing kernel stack offset on syscall entry" if EXPERT 14698cb37a59SMarco Elver default y 147039218ff4SKees Cook depends on HAVE_ARCH_RANDOMIZE_KSTACK_OFFSET 1471efa90c11SMarco Elver depends on INIT_STACK_NONE || !CC_IS_CLANG || CLANG_VERSION >= 140000 147239218ff4SKees Cook help 147339218ff4SKees Cook The kernel stack offset can be randomized (after pt_regs) by 147439218ff4SKees Cook roughly 5 bits of entropy, frustrating memory corruption 147539218ff4SKees Cook attacks that depend on stack address determinism or 14768cb37a59SMarco Elver cross-syscall address exposures. 14778cb37a59SMarco Elver 14788cb37a59SMarco Elver The feature is controlled via the "randomize_kstack_offset=on/off" 14798cb37a59SMarco Elver kernel boot param, and if turned off has zero overhead due to its use 14808cb37a59SMarco Elver of static branches (see JUMP_LABEL). 14818cb37a59SMarco Elver 14828cb37a59SMarco Elver If unsure, say Y. 14838cb37a59SMarco Elver 14848cb37a59SMarco Elverconfig RANDOMIZE_KSTACK_OFFSET_DEFAULT 14858cb37a59SMarco Elver bool "Default state of kernel stack offset randomization" 14868cb37a59SMarco Elver depends on RANDOMIZE_KSTACK_OFFSET 14878cb37a59SMarco Elver help 14888cb37a59SMarco Elver Kernel stack offset randomization is controlled by kernel boot param 14898cb37a59SMarco Elver "randomize_kstack_offset=on/off", and this config chooses the default 14908cb37a59SMarco Elver boot state. 149139218ff4SKees Cook 1492ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX 1493ad21fc4fSLaura Abbott def_bool n 1494ad21fc4fSLaura Abbott 1495ad21fc4fSLaura Abbottconfig ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1496ad21fc4fSLaura Abbott def_bool n 1497ad21fc4fSLaura Abbott 1498ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_KERNEL_RWX 1499ad21fc4fSLaura Abbott def_bool n 1500ad21fc4fSLaura Abbott 15010f5bf6d0SLaura Abbottconfig STRICT_KERNEL_RWX 1502ad21fc4fSLaura Abbott bool "Make kernel text and rodata read-only" if ARCH_OPTIONAL_KERNEL_RWX 1503ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_KERNEL_RWX 1504ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1505ad21fc4fSLaura Abbott help 1506ad21fc4fSLaura Abbott If this is set, kernel text and rodata memory will be made read-only, 1507ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 1508ad21fc4fSLaura Abbott protection against certain security exploits (e.g. executing the heap 1509ad21fc4fSLaura Abbott or modifying text) 1510ad21fc4fSLaura Abbott 1511ad21fc4fSLaura Abbott These features are considered standard security practice these days. 1512ad21fc4fSLaura Abbott You should say Y here in almost all cases. 1513ad21fc4fSLaura Abbott 1514ad21fc4fSLaura Abbottconfig ARCH_HAS_STRICT_MODULE_RWX 1515ad21fc4fSLaura Abbott def_bool n 1516ad21fc4fSLaura Abbott 15170f5bf6d0SLaura Abbottconfig STRICT_MODULE_RWX 1518ad21fc4fSLaura Abbott bool "Set loadable kernel module data as NX and text as RO" if ARCH_OPTIONAL_KERNEL_RWX 1519ad21fc4fSLaura Abbott depends on ARCH_HAS_STRICT_MODULE_RWX && MODULES 1520ad21fc4fSLaura Abbott default !ARCH_OPTIONAL_KERNEL_RWX || ARCH_OPTIONAL_KERNEL_RWX_DEFAULT 1521ad21fc4fSLaura Abbott help 1522ad21fc4fSLaura Abbott If this is set, module text and rodata memory will be made read-only, 1523ad21fc4fSLaura Abbott and non-text memory will be made non-executable. This provides 1524ad21fc4fSLaura Abbott protection against certain security exploits (e.g. writing to text) 1525ad21fc4fSLaura Abbott 1526ea8c64acSChristoph Hellwig# select if the architecture provides an asm/dma-direct.h header 1527ea8c64acSChristoph Hellwigconfig ARCH_HAS_PHYS_TO_DMA 1528ea8c64acSChristoph Hellwig bool 1529ea8c64acSChristoph Hellwig 1530bff70402SJames Morseconfig ARCH_HAS_CPU_RESCTRL 1531bff70402SJames Morse bool 1532bff70402SJames Morse help 1533bff70402SJames Morse An architecture selects this option to indicate that the necessary 1534bff70402SJames Morse hooks are provided to support the common memory system usage 1535bff70402SJames Morse monitoring and control interfaces provided by the 'resctrl' 1536bff70402SJames Morse filesystem (see RESCTRL_FS). 1537bff70402SJames Morse 153804f264d3SPaul Burtonconfig HAVE_ARCH_COMPILER_H 153904f264d3SPaul Burton bool 154004f264d3SPaul Burton help 154104f264d3SPaul Burton An architecture can select this if it provides an 154204f264d3SPaul Burton asm/compiler.h header that should be included after 154304f264d3SPaul Burton linux/compiler-*.h in order to override macro definitions that those 154404f264d3SPaul Burton headers generally provide. 154504f264d3SPaul Burton 1546e419ddeaSGreg Ungererconfig HAVE_ARCH_LIBGCC_H 1547e419ddeaSGreg Ungerer bool 1548e419ddeaSGreg Ungerer help 1549e419ddeaSGreg Ungerer An architecture can select this if it provides an 1550e419ddeaSGreg Ungerer asm/libgcc.h header that should be included after 1551e419ddeaSGreg Ungerer linux/libgcc.h in order to override macro definitions that 1552e419ddeaSGreg Ungerer header generally provides. 1553e419ddeaSGreg Ungerer 1554271ca788SArd Biesheuvelconfig HAVE_ARCH_PREL32_RELOCATIONS 1555271ca788SArd Biesheuvel bool 1556271ca788SArd Biesheuvel help 1557271ca788SArd Biesheuvel May be selected by an architecture if it supports place-relative 1558271ca788SArd Biesheuvel 32-bit relocations, both in the toolchain and in the module loader, 1559271ca788SArd Biesheuvel in which case relative references can be used in special sections 1560271ca788SArd Biesheuvel for PCI fixup, initcalls etc which are only half the size on 64 bit 1561271ca788SArd Biesheuvel architectures, and don't require runtime relocation on relocatable 1562271ca788SArd Biesheuvel kernels. 1563271ca788SArd Biesheuvel 1564ce9084baSArd Biesheuvelconfig ARCH_USE_MEMREMAP_PROT 1565ce9084baSArd Biesheuvel bool 1566ce9084baSArd Biesheuvel 1567fb346fd9SWaiman Longconfig LOCK_EVENT_COUNTS 1568fb346fd9SWaiman Long bool "Locking event counts collection" 1569fb346fd9SWaiman Long depends on DEBUG_FS 1570a7f7f624SMasahiro Yamada help 1571fb346fd9SWaiman Long Enable light-weight counting of various locking related events 1572fb346fd9SWaiman Long in the system with minimal performance impact. This reduces 1573fb346fd9SWaiman Long the chance of application behavior change because of timing 1574fb346fd9SWaiman Long differences. The counts are reported via debugfs. 1575fb346fd9SWaiman Long 15765cf896fbSPeter Collingbourne# Select if the architecture has support for applying RELR relocations. 15775cf896fbSPeter Collingbourneconfig ARCH_HAS_RELR 15785cf896fbSPeter Collingbourne bool 15795cf896fbSPeter Collingbourne 15805cf896fbSPeter Collingbourneconfig RELR 15815cf896fbSPeter Collingbourne bool "Use RELR relocation packing" 15825cf896fbSPeter Collingbourne depends on ARCH_HAS_RELR && TOOLS_SUPPORT_RELR 15835cf896fbSPeter Collingbourne default y 15845cf896fbSPeter Collingbourne help 15855cf896fbSPeter Collingbourne Store the kernel's dynamic relocations in the RELR relocation packing 15865cf896fbSPeter Collingbourne format. Requires a compatible linker (LLD supports this feature), as 15875cf896fbSPeter Collingbourne well as compatible NM and OBJCOPY utilities (llvm-nm and llvm-objcopy 15885cf896fbSPeter Collingbourne are compatible). 15895cf896fbSPeter Collingbourne 15900c9c1d56SThiago Jung Bauermannconfig ARCH_HAS_MEM_ENCRYPT 15910c9c1d56SThiago Jung Bauermann bool 15920c9c1d56SThiago Jung Bauermann 159346b49b12STom Lendackyconfig ARCH_HAS_CC_PLATFORM 159446b49b12STom Lendacky bool 159546b49b12STom Lendacky 15960e242208SHassan Naveedconfig HAVE_SPARSE_SYSCALL_NR 15970e242208SHassan Naveed bool 15980e242208SHassan Naveed help 15990e242208SHassan Naveed An architecture should select this if its syscall numbering is sparse 16000e242208SHassan Naveed to save space. For example, MIPS architecture has a syscall array with 16010e242208SHassan Naveed entries at 4000, 5000 and 6000 locations. This option turns on syscall 16020e242208SHassan Naveed related optimizations for a given architecture. 16030e242208SHassan Naveed 1604365841e1SThomas Weißschuhconfig ARCH_HAS_VDSO_ARCH_DATA 1605365841e1SThomas Weißschuh depends on GENERIC_VDSO_DATA_STORE 1606365841e1SThomas Weißschuh bool 1607365841e1SThomas Weißschuh 1608a812eee0SNam Caoconfig ARCH_HAS_VDSO_TIME_DATA 1609d60d7de3SSven Schnelle bool 1610d60d7de3SSven Schnelle 1611115284d8SJosh Poimboeufconfig HAVE_STATIC_CALL 1612115284d8SJosh Poimboeuf bool 1613115284d8SJosh Poimboeuf 16149183c3f9SJosh Poimboeufconfig HAVE_STATIC_CALL_INLINE 16159183c3f9SJosh Poimboeuf bool 16169183c3f9SJosh Poimboeuf depends on HAVE_STATIC_CALL 161703f16cd0SJosh Poimboeuf select OBJTOOL 16189183c3f9SJosh Poimboeuf 16196ef869e0SMichal Hockoconfig HAVE_PREEMPT_DYNAMIC 16206ef869e0SMichal Hocko bool 162199cf983cSMark Rutland 162299cf983cSMark Rutlandconfig HAVE_PREEMPT_DYNAMIC_CALL 162399cf983cSMark Rutland bool 16246ef869e0SMichal Hocko depends on HAVE_STATIC_CALL 162599cf983cSMark Rutland select HAVE_PREEMPT_DYNAMIC 16266ef869e0SMichal Hocko help 162799cf983cSMark Rutland An architecture should select this if it can handle the preemption 162899cf983cSMark Rutland model being selected at boot time using static calls. 162999cf983cSMark Rutland 163099cf983cSMark Rutland Where an architecture selects HAVE_STATIC_CALL_INLINE, any call to a 163199cf983cSMark Rutland preemption function will be patched directly. 163299cf983cSMark Rutland 163399cf983cSMark Rutland Where an architecture does not select HAVE_STATIC_CALL_INLINE, any 163499cf983cSMark Rutland call to a preemption function will go through a trampoline, and the 163599cf983cSMark Rutland trampoline will be patched. 163699cf983cSMark Rutland 163799cf983cSMark Rutland It is strongly advised to support inline static call to avoid any 163899cf983cSMark Rutland overhead. 163999cf983cSMark Rutland 164099cf983cSMark Rutlandconfig HAVE_PREEMPT_DYNAMIC_KEY 164199cf983cSMark Rutland bool 1642a0a12c3eSNick Desaulniers depends on HAVE_ARCH_JUMP_LABEL 164399cf983cSMark Rutland select HAVE_PREEMPT_DYNAMIC 164499cf983cSMark Rutland help 164599cf983cSMark Rutland An architecture should select this if it can handle the preemption 164699cf983cSMark Rutland model being selected at boot time using static keys. 164799cf983cSMark Rutland 164899cf983cSMark Rutland Each preemption function will be given an early return based on a 164999cf983cSMark Rutland static key. This should have slightly lower overhead than non-inline 165099cf983cSMark Rutland static calls, as this effectively inlines each trampoline into the 165199cf983cSMark Rutland start of its callee. This may avoid redundant work, and may 165299cf983cSMark Rutland integrate better with CFI schemes. 165399cf983cSMark Rutland 165499cf983cSMark Rutland This will have greater overhead than using inline static calls as 165599cf983cSMark Rutland the call to the preemption function cannot be entirely elided. 16566ef869e0SMichal Hocko 165759612b24SNathan Chancellorconfig ARCH_WANT_LD_ORPHAN_WARN 165859612b24SNathan Chancellor bool 165959612b24SNathan Chancellor help 166059612b24SNathan Chancellor An arch should select this symbol once all linker sections are explicitly 166159612b24SNathan Chancellor included, size-asserted, or discarded in the linker scripts. This is 166259612b24SNathan Chancellor important because we never want expected sections to be placed heuristically 166359612b24SNathan Chancellor by the linker, since the locations of such sections can change between linker 166459612b24SNathan Chancellor versions. 166559612b24SNathan Chancellor 16664f5b0c17SMike Rapoportconfig HAVE_ARCH_PFN_VALID 16674f5b0c17SMike Rapoport bool 16684f5b0c17SMike Rapoport 16695d6ad668SMike Rapoportconfig ARCH_SUPPORTS_DEBUG_PAGEALLOC 16705d6ad668SMike Rapoport bool 16715d6ad668SMike Rapoport 1672df4e817bSPasha Tatashinconfig ARCH_SUPPORTS_PAGE_TABLE_CHECK 1673df4e817bSPasha Tatashin bool 1674df4e817bSPasha Tatashin 16752ca408d9SBrian Gerstconfig ARCH_SPLIT_ARG64 16762ca408d9SBrian Gerst bool 16772ca408d9SBrian Gerst help 16782ca408d9SBrian Gerst If a 32-bit architecture requires 64-bit arguments to be split into 16792ca408d9SBrian Gerst pairs of 32-bit arguments, select this option. 16802ca408d9SBrian Gerst 16817facdc42SAl Viroconfig ARCH_HAS_ELFCORE_COMPAT 16827facdc42SAl Viro bool 16837facdc42SAl Viro 168458e106e7SBalbir Singhconfig ARCH_HAS_PARANOID_L1D_FLUSH 168558e106e7SBalbir Singh bool 168658e106e7SBalbir Singh 1687d593d64fSPrasad Sodagudiconfig ARCH_HAVE_TRACE_MMIO_ACCESS 1688d593d64fSPrasad Sodagudi bool 1689d593d64fSPrasad Sodagudi 16901bdda24cSThomas Gleixnerconfig DYNAMIC_SIGFRAME 16911bdda24cSThomas Gleixner bool 16921bdda24cSThomas Gleixner 169350468e43SJarkko Sakkinen# Select, if arch has a named attribute group bound to NUMA device nodes. 169450468e43SJarkko Sakkinenconfig HAVE_ARCH_NODE_DEV_GROUP 169550468e43SJarkko Sakkinen bool 169650468e43SJarkko Sakkinen 169771ce1ab5SKinsey Hoconfig ARCH_HAS_HW_PTE_YOUNG 169871ce1ab5SKinsey Ho bool 169971ce1ab5SKinsey Ho help 170071ce1ab5SKinsey Ho Architectures that select this option are capable of setting the 170171ce1ab5SKinsey Ho accessed bit in PTE entries when using them as part of linear address 170271ce1ab5SKinsey Ho translations. Architectures that require runtime check should select 170371ce1ab5SKinsey Ho this option and override arch_has_hw_pte_young(). 170471ce1ab5SKinsey Ho 1705eed9a328SYu Zhaoconfig ARCH_HAS_NONLEAF_PMD_YOUNG 1706eed9a328SYu Zhao bool 1707eed9a328SYu Zhao help 1708eed9a328SYu Zhao Architectures that select this option are capable of setting the 1709eed9a328SYu Zhao accessed bit in non-leaf PMD entries when using them as part of linear 1710eed9a328SYu Zhao address translations. Page table walkers that clear the accessed bit 1711eed9a328SYu Zhao may use this capability to reduce their search space. 1712eed9a328SYu Zhao 17136cbd1d6dSSamuel Hollandconfig ARCH_HAS_KERNEL_FPU_SUPPORT 17146cbd1d6dSSamuel Holland bool 17156cbd1d6dSSamuel Holland help 17166cbd1d6dSSamuel Holland Architectures that select this option can run floating-point code in 17176cbd1d6dSSamuel Holland the kernel, as described in Documentation/core-api/floating-point.rst. 17186cbd1d6dSSamuel Holland 17199b400d17SArd Biesheuvelconfig ARCH_VMLINUX_NEEDS_RELOCS 17209b400d17SArd Biesheuvel bool 17219b400d17SArd Biesheuvel help 17229b400d17SArd Biesheuvel Whether the architecture needs vmlinux to be built with static 17239b400d17SArd Biesheuvel relocations preserved. This is used by some architectures to 17249b400d17SArd Biesheuvel construct bespoke relocation tables for KASLR. 17259b400d17SArd Biesheuvel 17262521f2c2SPeter Oberparleitersource "kernel/gcov/Kconfig" 172745332b1bSMasahiro Yamada 172845332b1bSMasahiro Yamadasource "scripts/gcc-plugins/Kconfig" 1729fa1b5d09SLinus Torvalds 1730d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT_4B 1731d49a0626SPeter Zijlstra bool 1732d49a0626SPeter Zijlstra 1733d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT_8B 1734d49a0626SPeter Zijlstra bool 1735d49a0626SPeter Zijlstra 1736d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT_16B 1737d49a0626SPeter Zijlstra bool 1738d49a0626SPeter Zijlstra 1739d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT_32B 1740d49a0626SPeter Zijlstra bool 1741d49a0626SPeter Zijlstra 1742d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT_64B 1743d49a0626SPeter Zijlstra bool 1744d49a0626SPeter Zijlstra 1745d49a0626SPeter Zijlstraconfig FUNCTION_ALIGNMENT 1746d49a0626SPeter Zijlstra int 1747d49a0626SPeter Zijlstra default 64 if FUNCTION_ALIGNMENT_64B 1748d49a0626SPeter Zijlstra default 32 if FUNCTION_ALIGNMENT_32B 1749d49a0626SPeter Zijlstra default 16 if FUNCTION_ALIGNMENT_16B 1750d49a0626SPeter Zijlstra default 8 if FUNCTION_ALIGNMENT_8B 1751d49a0626SPeter Zijlstra default 4 if FUNCTION_ALIGNMENT_4B 1752d49a0626SPeter Zijlstra default 0 1753d49a0626SPeter Zijlstra 17545270316cSPetr Pavluconfig CC_HAS_MIN_FUNCTION_ALIGNMENT 17555270316cSPetr Pavlu # Detect availability of the GCC option -fmin-function-alignment which 17565270316cSPetr Pavlu # guarantees minimal alignment for all functions, unlike 17575270316cSPetr Pavlu # -falign-functions which the compiler ignores for cold functions. 17585270316cSPetr Pavlu def_bool $(cc-option, -fmin-function-alignment=8) 17595270316cSPetr Pavlu 17605270316cSPetr Pavluconfig CC_HAS_SANE_FUNCTION_ALIGNMENT 17615270316cSPetr Pavlu # Set if the guaranteed alignment with -fmin-function-alignment is 17625270316cSPetr Pavlu # available or extra care is required in the kernel. Clang provides 17635270316cSPetr Pavlu # strict alignment always, even with -falign-functions. 17645270316cSPetr Pavlu def_bool CC_HAS_MIN_FUNCTION_ALIGNMENT || CC_IS_CLANG 17655270316cSPetr Pavlu 1766a88d970cSPaul E. McKenneyconfig ARCH_NEED_CMPXCHG_1_EMU 1767a88d970cSPaul E. McKenney bool 1768a88d970cSPaul E. McKenney 17691198c9c6SNaveen N Raoconfig ARCH_WANTS_PRE_LINK_VMLINUX 17701198c9c6SNaveen N Rao bool 17711198c9c6SNaveen N Rao help 17721198c9c6SNaveen N Rao An architecture can select this if it provides arch/<arch>/tools/Makefile 17731198c9c6SNaveen N Rao with .arch.vmlinux.o target to be linked into vmlinux. 17741198c9c6SNaveen N Rao 177522471e13SRandy Dunlapendmenu 1776