1*ec8f24b7SThomas Gleixner# SPDX-License-Identifier: GPL-2.0-only 2face4374SRoman Zippelconfig DEFCONFIG_LIST 3face4374SRoman Zippel string 4b2670eacSPaolo 'Blaisorblade' Giarrusso depends on !UML 5face4374SRoman Zippel option defconfig_list 647f38ae0SRob Landley default "/lib/modules/$(shell,uname -r)/.config" 7face4374SRoman Zippel default "/etc/kernel-config" 847f38ae0SRob Landley default "/boot/config-$(shell,uname -r)" 9104daea1SMasahiro Yamada default ARCH_DEFCONFIG 10104daea1SMasahiro Yamada default "arch/$(ARCH)/defconfig" 11face4374SRoman Zippel 12a4353898SMasahiro Yamadaconfig CC_IS_GCC 13a4353898SMasahiro Yamada def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc) 14a4353898SMasahiro Yamada 15a4353898SMasahiro Yamadaconfig GCC_VERSION 16a4353898SMasahiro Yamada int 17fa7295abSMasahiro Yamada default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC 18a4353898SMasahiro Yamada default 0 19a4353898SMasahiro Yamada 20469cb737SMasahiro Yamadaconfig CC_IS_CLANG 21469cb737SMasahiro Yamada def_bool $(success,$(CC) --version | head -n 1 | grep -q clang) 22469cb737SMasahiro Yamada 23469cb737SMasahiro Yamadaconfig CLANG_VERSION 24469cb737SMasahiro Yamada int 25469cb737SMasahiro Yamada default $(shell,$(srctree)/scripts/clang-version.sh $(CC)) 26469cb737SMasahiro Yamada 27e9666d10SMasahiro Yamadaconfig CC_HAS_ASM_GOTO 28e9666d10SMasahiro Yamada def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC)) 29e9666d10SMasahiro Yamada 30b303c6dfSMasahiro Yamadaconfig CC_HAS_WARN_MAYBE_UNINITIALIZED 31b303c6dfSMasahiro Yamada def_bool $(cc-option,-Wmaybe-uninitialized) 32b303c6dfSMasahiro Yamada help 33b303c6dfSMasahiro Yamada GCC >= 4.7 supports this option. 34b303c6dfSMasahiro Yamada 35b303c6dfSMasahiro Yamadaconfig CC_DISABLE_WARN_MAYBE_UNINITIALIZED 36b303c6dfSMasahiro Yamada bool 37b303c6dfSMasahiro Yamada depends on CC_HAS_WARN_MAYBE_UNINITIALIZED 38b303c6dfSMasahiro Yamada default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9 39b303c6dfSMasahiro Yamada help 40b303c6dfSMasahiro Yamada GCC's -Wmaybe-uninitialized is not reliable by definition. 41b303c6dfSMasahiro Yamada Lots of false positive warnings are produced in some cases. 42b303c6dfSMasahiro Yamada 43b303c6dfSMasahiro Yamada If this option is enabled, -Wno-maybe-uninitialzed is passed 44b303c6dfSMasahiro Yamada to the compiler to suppress maybe-uninitialized warnings. 45b303c6dfSMasahiro Yamada 46b99b87f7SPeter Oberparleiterconfig CONSTRUCTORS 47b99b87f7SPeter Oberparleiter bool 48b99b87f7SPeter Oberparleiter depends on !UML 49b99b87f7SPeter Oberparleiter 50e360adbeSPeter Zijlstraconfig IRQ_WORK 51e360adbeSPeter Zijlstra bool 52e360adbeSPeter Zijlstra 531dbdc6f1SDavid Daneyconfig BUILDTIME_EXTABLE_SORT 541dbdc6f1SDavid Daney bool 551dbdc6f1SDavid Daney 56c65eacbeSAndy Lutomirskiconfig THREAD_INFO_IN_TASK 57c65eacbeSAndy Lutomirski bool 58c65eacbeSAndy Lutomirski help 59c65eacbeSAndy Lutomirski Select this to move thread_info off the stack into task_struct. To 60c65eacbeSAndy Lutomirski make this work, an arch will need to remove all thread_info fields 61c65eacbeSAndy Lutomirski except flags and fix any runtime bugs. 62c65eacbeSAndy Lutomirski 63c6c314a6SAndy Lutomirski One subtle change that will be needed is to use try_get_task_stack() 64c6c314a6SAndy Lutomirski and put_task_stack() in save_thread_stack_tsk() and get_wchan(). 65c6c314a6SAndy Lutomirski 66ff0cfc66SAl Boldimenu "General setup" 671da177e4SLinus Torvalds 681da177e4SLinus Torvaldsconfig BROKEN 691da177e4SLinus Torvalds bool 701da177e4SLinus Torvalds 711da177e4SLinus Torvaldsconfig BROKEN_ON_SMP 721da177e4SLinus Torvalds bool 731da177e4SLinus Torvalds depends on BROKEN || !SMP 741da177e4SLinus Torvalds default y 751da177e4SLinus Torvalds 761da177e4SLinus Torvaldsconfig INIT_ENV_ARG_LIMIT 771da177e4SLinus Torvalds int 78dd673bcaSAdrian Bunk default 32 if !UML 79dd673bcaSAdrian Bunk default 128 if UML 801da177e4SLinus Torvalds help 8134ad92c2SRandy Dunlap Maximum of each of the number of arguments and environment 8234ad92c2SRandy Dunlap variables passed to init from the kernel command line. 831da177e4SLinus Torvalds 844bb16672SJiri Slabyconfig COMPILE_TEST 854bb16672SJiri Slaby bool "Compile also drivers which will not load" 86bc083a64SRichard Weinberger depends on !UML 874bb16672SJiri Slaby default n 884bb16672SJiri Slaby help 894bb16672SJiri Slaby Some drivers can be compiled on a different platform than they are 904bb16672SJiri Slaby intended to be run on. Despite they cannot be loaded there (or even 914bb16672SJiri Slaby when they load they cannot be used due to missing HW support), 924bb16672SJiri Slaby developers still, opposing to distributors, might want to build such 934bb16672SJiri Slaby drivers to compile-test them. 944bb16672SJiri Slaby 954bb16672SJiri Slaby If you are a developer and want to build everything available, say Y 964bb16672SJiri Slaby here. If you are a user/distributor, say N here to exclude useless 974bb16672SJiri Slaby drivers to be distributed. 984bb16672SJiri Slaby 991da177e4SLinus Torvaldsconfig LOCALVERSION 1001da177e4SLinus Torvalds string "Local version - append to kernel release" 1011da177e4SLinus Torvalds help 1021da177e4SLinus Torvalds Append an extra string to the end of your kernel version. 1031da177e4SLinus Torvalds This will show up when you type uname, for example. 1041da177e4SLinus Torvalds The string you set here will be appended after the contents of 1051da177e4SLinus Torvalds any files with a filename matching localversion* in your 1061da177e4SLinus Torvalds object and source tree, in that order. Your total string can 1071da177e4SLinus Torvalds be a maximum of 64 characters. 1081da177e4SLinus Torvalds 109aaebf433SRyan Andersonconfig LOCALVERSION_AUTO 110aaebf433SRyan Anderson bool "Automatically append version information to the version string" 111aaebf433SRyan Anderson default y 112ac3339baSAlexey Dobriyan depends on !COMPILE_TEST 113aaebf433SRyan Anderson help 114aaebf433SRyan Anderson This will try to automatically determine if the current tree is a 1156e5a5420SRobert P. J. Day release tree by looking for git tags that belong to the current 1166e5a5420SRobert P. J. Day top of tree revision. 117aaebf433SRyan Anderson 118aaebf433SRyan Anderson A string of the format -gxxxxxxxx will be added to the localversion 1196e5a5420SRobert P. J. Day if a git-based tree is found. The string generated by this will be 120aaebf433SRyan Anderson appended after any matching localversion* files, and after the value 1216e5a5420SRobert P. J. Day set in CONFIG_LOCALVERSION. 122aaebf433SRyan Anderson 1236e5a5420SRobert P. J. Day (The actual string used here is the first eight characters produced 1246e5a5420SRobert P. J. Day by running the command: 1256e5a5420SRobert P. J. Day 1266e5a5420SRobert P. J. Day $ git rev-parse --verify HEAD 1276e5a5420SRobert P. J. Day 1286e5a5420SRobert P. J. Day which is done within the script "scripts/setlocalversion".) 129aaebf433SRyan Anderson 1309afb719eSLaura Abbottconfig BUILD_SALT 1319afb719eSLaura Abbott string "Build ID Salt" 1329afb719eSLaura Abbott default "" 1339afb719eSLaura Abbott help 1349afb719eSLaura Abbott The build ID is used to link binaries and their debug info. Setting 1359afb719eSLaura Abbott this option will use the value in the calculation of the build id. 1369afb719eSLaura Abbott This is mostly useful for distributions which want to ensure the 1379afb719eSLaura Abbott build is unique between builds. It's safe to leave the default. 1389afb719eSLaura Abbott 1392e9f3bddSH. Peter Anvinconfig HAVE_KERNEL_GZIP 1402e9f3bddSH. Peter Anvin bool 1412e9f3bddSH. Peter Anvin 1422e9f3bddSH. Peter Anvinconfig HAVE_KERNEL_BZIP2 1432e9f3bddSH. Peter Anvin bool 1442e9f3bddSH. Peter Anvin 1452e9f3bddSH. Peter Anvinconfig HAVE_KERNEL_LZMA 1462e9f3bddSH. Peter Anvin bool 1472e9f3bddSH. Peter Anvin 1483ebe1243SLasse Collinconfig HAVE_KERNEL_XZ 1493ebe1243SLasse Collin bool 1503ebe1243SLasse Collin 1517dd65febSAlbin Tonnerreconfig HAVE_KERNEL_LZO 1527dd65febSAlbin Tonnerre bool 1537dd65febSAlbin Tonnerre 154e76e1fdfSKyungsik Leeconfig HAVE_KERNEL_LZ4 155e76e1fdfSKyungsik Lee bool 156e76e1fdfSKyungsik Lee 157f16466afSVasily Gorbikconfig HAVE_KERNEL_UNCOMPRESSED 158f16466afSVasily Gorbik bool 159f16466afSVasily Gorbik 16030d65dbfSAlain Knaffchoice 16130d65dbfSAlain Knaff prompt "Kernel compression mode" 16230d65dbfSAlain Knaff default KERNEL_GZIP 163f16466afSVasily Gorbik depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED 16430d65dbfSAlain Knaff help 16530d65dbfSAlain Knaff The linux kernel is a kind of self-extracting executable. 16630d65dbfSAlain Knaff Several compression algorithms are available, which differ 16730d65dbfSAlain Knaff in efficiency, compression and decompression speed. 16830d65dbfSAlain Knaff Compression speed is only relevant when building a kernel. 16930d65dbfSAlain Knaff Decompression speed is relevant at each boot. 17030d65dbfSAlain Knaff 17130d65dbfSAlain Knaff If you have any problems with bzip2 or lzma compressed 17230d65dbfSAlain Knaff kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older 17330d65dbfSAlain Knaff version of this functionality (bzip2 only), for 2.4, was 17430d65dbfSAlain Knaff supplied by Christian Ludwig) 17530d65dbfSAlain Knaff 17630d65dbfSAlain Knaff High compression options are mostly useful for users, who 17730d65dbfSAlain Knaff are low on disk space (embedded systems), but for whom ram 17830d65dbfSAlain Knaff size matters less. 17930d65dbfSAlain Knaff 18030d65dbfSAlain Knaff If in doubt, select 'gzip' 18130d65dbfSAlain Knaff 18230d65dbfSAlain Knaffconfig KERNEL_GZIP 18330d65dbfSAlain Knaff bool "Gzip" 1842e9f3bddSH. Peter Anvin depends on HAVE_KERNEL_GZIP 18530d65dbfSAlain Knaff help 1867dd65febSAlbin Tonnerre The old and tried gzip compression. It provides a good balance 1877dd65febSAlbin Tonnerre between compression ratio and decompression speed. 18830d65dbfSAlain Knaff 18930d65dbfSAlain Knaffconfig KERNEL_BZIP2 19030d65dbfSAlain Knaff bool "Bzip2" 1912e9f3bddSH. Peter Anvin depends on HAVE_KERNEL_BZIP2 19230d65dbfSAlain Knaff help 19330d65dbfSAlain Knaff Its compression ratio and speed is intermediate. 1940a4dd35cSRandy Dunlap Decompression speed is slowest among the choices. The kernel 1952e9f3bddSH. Peter Anvin size is about 10% smaller with bzip2, in comparison to gzip. 1962e9f3bddSH. Peter Anvin Bzip2 uses a large amount of memory. For modern kernels you 1972e9f3bddSH. Peter Anvin will need at least 8MB RAM or more for booting. 19830d65dbfSAlain Knaff 19930d65dbfSAlain Knaffconfig KERNEL_LZMA 20030d65dbfSAlain Knaff bool "LZMA" 2012e9f3bddSH. Peter Anvin depends on HAVE_KERNEL_LZMA 20230d65dbfSAlain Knaff help 2030a4dd35cSRandy Dunlap This compression algorithm's ratio is best. Decompression speed 2040a4dd35cSRandy Dunlap is between gzip and bzip2. Compression is slowest. 2050a4dd35cSRandy Dunlap The kernel size is about 33% smaller with LZMA in comparison to gzip. 20630d65dbfSAlain Knaff 2073ebe1243SLasse Collinconfig KERNEL_XZ 2083ebe1243SLasse Collin bool "XZ" 2093ebe1243SLasse Collin depends on HAVE_KERNEL_XZ 2103ebe1243SLasse Collin help 2113ebe1243SLasse Collin XZ uses the LZMA2 algorithm and instruction set specific 2123ebe1243SLasse Collin BCJ filters which can improve compression ratio of executable 2133ebe1243SLasse Collin code. The size of the kernel is about 30% smaller with XZ in 2143ebe1243SLasse Collin comparison to gzip. On architectures for which there is a BCJ 2153ebe1243SLasse Collin filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ 2163ebe1243SLasse Collin will create a few percent smaller kernel than plain LZMA. 2173ebe1243SLasse Collin 2183ebe1243SLasse Collin The speed is about the same as with LZMA: The decompression 2193ebe1243SLasse Collin speed of XZ is better than that of bzip2 but worse than gzip 2203ebe1243SLasse Collin and LZO. Compression is slow. 2213ebe1243SLasse Collin 2227dd65febSAlbin Tonnerreconfig KERNEL_LZO 2237dd65febSAlbin Tonnerre bool "LZO" 2247dd65febSAlbin Tonnerre depends on HAVE_KERNEL_LZO 2257dd65febSAlbin Tonnerre help 2260a4dd35cSRandy Dunlap Its compression ratio is the poorest among the choices. The kernel 227681b3049SStephan Sperber size is about 10% bigger than gzip; however its speed 2287dd65febSAlbin Tonnerre (both compression and decompression) is the fastest. 2297dd65febSAlbin Tonnerre 230e76e1fdfSKyungsik Leeconfig KERNEL_LZ4 231e76e1fdfSKyungsik Lee bool "LZ4" 232e76e1fdfSKyungsik Lee depends on HAVE_KERNEL_LZ4 233e76e1fdfSKyungsik Lee help 234e76e1fdfSKyungsik Lee LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding. 235e76e1fdfSKyungsik Lee A preliminary version of LZ4 de/compression tool is available at 236e76e1fdfSKyungsik Lee <https://code.google.com/p/lz4/>. 237e76e1fdfSKyungsik Lee 238e76e1fdfSKyungsik Lee Its compression ratio is worse than LZO. The size of the kernel 239e76e1fdfSKyungsik Lee is about 8% bigger than LZO. But the decompression speed is 240e76e1fdfSKyungsik Lee faster than LZO. 241e76e1fdfSKyungsik Lee 242f16466afSVasily Gorbikconfig KERNEL_UNCOMPRESSED 243f16466afSVasily Gorbik bool "None" 244f16466afSVasily Gorbik depends on HAVE_KERNEL_UNCOMPRESSED 245f16466afSVasily Gorbik help 246f16466afSVasily Gorbik Produce uncompressed kernel image. This option is usually not what 247f16466afSVasily Gorbik you want. It is useful for debugging the kernel in slow simulation 248f16466afSVasily Gorbik environments, where decompressing and moving the kernel is awfully 249f16466afSVasily Gorbik slow. This option allows early boot code to skip the decompressor 250f16466afSVasily Gorbik and jump right at uncompressed kernel image. 251f16466afSVasily Gorbik 25230d65dbfSAlain Knaffendchoice 25330d65dbfSAlain Knaff 254bd5dc17bSJosh Triplettconfig DEFAULT_HOSTNAME 255bd5dc17bSJosh Triplett string "Default hostname" 256bd5dc17bSJosh Triplett default "(none)" 257bd5dc17bSJosh Triplett help 258bd5dc17bSJosh Triplett This option determines the default system hostname before userspace 259bd5dc17bSJosh Triplett calls sethostname(2). The kernel traditionally uses "(none)" here, 260bd5dc17bSJosh Triplett but you may wish to use a different default here to make a minimal 261bd5dc17bSJosh Triplett system more usable with less configuration. 262bd5dc17bSJosh Triplett 26317c46a6aSChristoph Hellwig# 26417c46a6aSChristoph Hellwig# For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can 26517c46a6aSChristoph Hellwig# add proper SWAP support to them, in which case this can be remove. 26617c46a6aSChristoph Hellwig# 26717c46a6aSChristoph Hellwigconfig ARCH_NO_SWAP 26817c46a6aSChristoph Hellwig bool 26917c46a6aSChristoph Hellwig 2701da177e4SLinus Torvaldsconfig SWAP 2711da177e4SLinus Torvalds bool "Support for paging of anonymous memory (swap)" 27217c46a6aSChristoph Hellwig depends on MMU && BLOCK && !ARCH_NO_SWAP 2731da177e4SLinus Torvalds default y 2741da177e4SLinus Torvalds help 2751da177e4SLinus Torvalds This option allows you to choose whether you want to have support 2761da177e4SLinus Torvalds for so called swap devices or swap files in your kernel that are 2771da177e4SLinus Torvalds used to provide more virtual memory than the actual RAM present 2781da177e4SLinus Torvalds in your computer. If unsure say Y. 2791da177e4SLinus Torvalds 2801da177e4SLinus Torvaldsconfig SYSVIPC 2811da177e4SLinus Torvalds bool "System V IPC" 2821da177e4SLinus Torvalds ---help--- 2831da177e4SLinus Torvalds Inter Process Communication is a suite of library functions and 2841da177e4SLinus Torvalds system calls which let processes (running programs) synchronize and 2851da177e4SLinus Torvalds exchange information. It is generally considered to be a good thing, 2861da177e4SLinus Torvalds and some programs won't run unless you say Y here. In particular, if 2871da177e4SLinus Torvalds you want to run the DOS emulator dosemu under Linux (read the 2881da177e4SLinus Torvalds DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>), 2891da177e4SLinus Torvalds you'll need to say Y here. 2901da177e4SLinus Torvalds 2911da177e4SLinus Torvalds You can find documentation about IPC with "info ipc" and also in 2921da177e4SLinus Torvalds section 6.4 of the Linux Programmer's Guide, available from 2931da177e4SLinus Torvalds <http://www.tldp.org/guides.html>. 2941da177e4SLinus Torvalds 295a5494dcdSEric W. Biedermanconfig SYSVIPC_SYSCTL 296a5494dcdSEric W. Biederman bool 297a5494dcdSEric W. Biederman depends on SYSVIPC 298a5494dcdSEric W. Biederman depends on SYSCTL 299a5494dcdSEric W. Biederman default y 300a5494dcdSEric W. Biederman 3011da177e4SLinus Torvaldsconfig POSIX_MQUEUE 3021da177e4SLinus Torvalds bool "POSIX Message Queues" 30319c92399SKees Cook depends on NET 3041da177e4SLinus Torvalds ---help--- 3051da177e4SLinus Torvalds POSIX variant of message queues is a part of IPC. In POSIX message 3061da177e4SLinus Torvalds queues every message has a priority which decides about succession 3071da177e4SLinus Torvalds of receiving it by a process. If you want to compile and run 3081da177e4SLinus Torvalds programs written e.g. for Solaris with use of its POSIX message 309b0e37650SRobert P. J. Day queues (functions mq_*) say Y here. 3101da177e4SLinus Torvalds 3111da177e4SLinus Torvalds POSIX message queues are visible as a filesystem called 'mqueue' 3121da177e4SLinus Torvalds and can be mounted somewhere if you want to do filesystem 3131da177e4SLinus Torvalds operations on message queues. 3141da177e4SLinus Torvalds 3151da177e4SLinus Torvalds If unsure, say Y. 3161da177e4SLinus Torvalds 317bdc8e5f8SSerge E. Hallynconfig POSIX_MQUEUE_SYSCTL 318bdc8e5f8SSerge E. Hallyn bool 319bdc8e5f8SSerge E. Hallyn depends on POSIX_MQUEUE 320bdc8e5f8SSerge E. Hallyn depends on SYSCTL 321bdc8e5f8SSerge E. Hallyn default y 322bdc8e5f8SSerge E. Hallyn 323226b4ccdSKonstantin Khlebnikovconfig CROSS_MEMORY_ATTACH 324226b4ccdSKonstantin Khlebnikov bool "Enable process_vm_readv/writev syscalls" 325226b4ccdSKonstantin Khlebnikov depends on MMU 326226b4ccdSKonstantin Khlebnikov default y 327226b4ccdSKonstantin Khlebnikov help 328226b4ccdSKonstantin Khlebnikov Enabling this option adds the system calls process_vm_readv and 329226b4ccdSKonstantin Khlebnikov process_vm_writev which allow a process with the correct privileges 330a2a368d9SGeert Uytterhoeven to directly read from or write to another process' address space. 331226b4ccdSKonstantin Khlebnikov See the man page for more details. 332226b4ccdSKonstantin Khlebnikov 33369369a70SJosh Triplettconfig USELIB 33469369a70SJosh Triplett bool "uselib syscall" 335b2113a41SRiku Voipio def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION 33669369a70SJosh Triplett help 33769369a70SJosh Triplett This option enables the uselib syscall, a system call used in the 33869369a70SJosh Triplett dynamic linker from libc5 and earlier. glibc does not use this 33969369a70SJosh Triplett system call. If you intend to run programs built on libc5 or 34069369a70SJosh Triplett earlier, you may need to enable this syscall. Current systems 34169369a70SJosh Triplett running glibc can safely disable this. 34269369a70SJosh Triplett 3431da177e4SLinus Torvaldsconfig AUDIT 3441da177e4SLinus Torvalds bool "Auditing support" 345804a6a49SChris Wright depends on NET 3461da177e4SLinus Torvalds help 3471da177e4SLinus Torvalds Enable auditing infrastructure that can be used with another 3481da177e4SLinus Torvalds kernel subsystem, such as SELinux (which requires this for 349cb74ed27SPaul Moore logging of avc messages output). System call auditing is included 350cb74ed27SPaul Moore on architectures which support it. 3511da177e4SLinus Torvalds 3527a017721SAKASHI Takahiroconfig HAVE_ARCH_AUDITSYSCALL 3537a017721SAKASHI Takahiro bool 3547a017721SAKASHI Takahiro 3551da177e4SLinus Torvaldsconfig AUDITSYSCALL 356cb74ed27SPaul Moore def_bool y 3577a017721SAKASHI Takahiro depends on AUDIT && HAVE_ARCH_AUDITSYSCALL 35828a3a7ebSEric Paris select FSNOTIFY 35974c3cbe3SAl Viro 360d9817ebeSThomas Gleixnersource "kernel/irq/Kconfig" 361764e0da1SThomas Gleixnersource "kernel/time/Kconfig" 36287a4c375SChristoph Hellwigsource "kernel/Kconfig.preempt" 363d9817ebeSThomas Gleixner 364391dc69cSFrederic Weisbeckermenu "CPU/Task time and stats accounting" 365391dc69cSFrederic Weisbecker 366abf917cdSFrederic Weisbeckerconfig VIRT_CPU_ACCOUNTING 367abf917cdSFrederic Weisbecker bool 368abf917cdSFrederic Weisbecker 369fdf9c356SFrederic Weisbeckerchoice 370fdf9c356SFrederic Weisbecker prompt "Cputime accounting" 371fdf9c356SFrederic Weisbecker default TICK_CPU_ACCOUNTING if !PPC64 37202fc8d37SStephen Rothwell default VIRT_CPU_ACCOUNTING_NATIVE if PPC64 373fdf9c356SFrederic Weisbecker 374fdf9c356SFrederic Weisbecker# Kind of a stub config for the pure tick based cputime accounting 375fdf9c356SFrederic Weisbeckerconfig TICK_CPU_ACCOUNTING 376fdf9c356SFrederic Weisbecker bool "Simple tick based cputime accounting" 377c58b0df1SFrederic Weisbecker depends on !S390 && !NO_HZ_FULL 378fdf9c356SFrederic Weisbecker help 379fdf9c356SFrederic Weisbecker This is the basic tick based cputime accounting that maintains 380fdf9c356SFrederic Weisbecker statistics about user, system and idle time spent on per jiffies 381fdf9c356SFrederic Weisbecker granularity. 382fdf9c356SFrederic Weisbecker 383fdf9c356SFrederic Weisbecker If unsure, say Y. 384fdf9c356SFrederic Weisbecker 385abf917cdSFrederic Weisbeckerconfig VIRT_CPU_ACCOUNTING_NATIVE 386391dc69cSFrederic Weisbecker bool "Deterministic task and CPU time accounting" 387c58b0df1SFrederic Weisbecker depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL 388abf917cdSFrederic Weisbecker select VIRT_CPU_ACCOUNTING 389391dc69cSFrederic Weisbecker help 390391dc69cSFrederic Weisbecker Select this option to enable more accurate task and CPU time 391391dc69cSFrederic Weisbecker accounting. This is done by reading a CPU counter on each 392391dc69cSFrederic Weisbecker kernel entry and exit and on transitions within the kernel 393391dc69cSFrederic Weisbecker between system, softirq and hardirq state, so there is a 394391dc69cSFrederic Weisbecker small performance impact. In the case of s390 or IBM POWER > 5, 395391dc69cSFrederic Weisbecker this also enables accounting of stolen time on logically-partitioned 396391dc69cSFrederic Weisbecker systems. 397391dc69cSFrederic Weisbecker 398abf917cdSFrederic Weisbeckerconfig VIRT_CPU_ACCOUNTING_GEN 399abf917cdSFrederic Weisbecker bool "Full dynticks CPU time accounting" 400ff3fb254SKevin Hilman depends on HAVE_CONTEXT_TRACKING 401554b0004SKevin Hilman depends on HAVE_VIRT_CPU_ACCOUNTING_GEN 402041a1574SArnd Bergmann depends on GENERIC_CLOCKEVENTS 403abf917cdSFrederic Weisbecker select VIRT_CPU_ACCOUNTING 404abf917cdSFrederic Weisbecker select CONTEXT_TRACKING 405abf917cdSFrederic Weisbecker help 406abf917cdSFrederic Weisbecker Select this option to enable task and CPU time accounting on full 407abf917cdSFrederic Weisbecker dynticks systems. This accounting is implemented by watching every 408abf917cdSFrederic Weisbecker kernel-user boundaries using the context tracking subsystem. 409abf917cdSFrederic Weisbecker The accounting is thus performed at the expense of some significant 410abf917cdSFrederic Weisbecker overhead. 411abf917cdSFrederic Weisbecker 412abf917cdSFrederic Weisbecker For now this is only useful if you are working on the full 413abf917cdSFrederic Weisbecker dynticks subsystem development. 414abf917cdSFrederic Weisbecker 415abf917cdSFrederic Weisbecker If unsure, say N. 416abf917cdSFrederic Weisbecker 417b58c3584SRik van Rielendchoice 418b58c3584SRik van Riel 419fdf9c356SFrederic Weisbeckerconfig IRQ_TIME_ACCOUNTING 420fdf9c356SFrederic Weisbecker bool "Fine granularity task level IRQ time accounting" 421b58c3584SRik van Riel depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE 422fdf9c356SFrederic Weisbecker help 423fdf9c356SFrederic Weisbecker Select this option to enable fine granularity task irq time 424fdf9c356SFrederic Weisbecker accounting. This is done by reading a timestamp on each 425fdf9c356SFrederic Weisbecker transitions between softirq and hardirq state, so there can be a 426fdf9c356SFrederic Weisbecker small performance impact. 427fdf9c356SFrederic Weisbecker 428fdf9c356SFrederic Weisbecker If in doubt, say N here. 429fdf9c356SFrederic Weisbecker 43011d4afd4SVincent Guittotconfig HAVE_SCHED_AVG_IRQ 43111d4afd4SVincent Guittot def_bool y 43211d4afd4SVincent Guittot depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING 43311d4afd4SVincent Guittot depends on SMP 43411d4afd4SVincent Guittot 435391dc69cSFrederic Weisbeckerconfig BSD_PROCESS_ACCT 436391dc69cSFrederic Weisbecker bool "BSD Process Accounting" 4372813893fSIulia Manda depends on MULTIUSER 438391dc69cSFrederic Weisbecker help 439391dc69cSFrederic Weisbecker If you say Y here, a user level program will be able to instruct the 440391dc69cSFrederic Weisbecker kernel (via a special system call) to write process accounting 441391dc69cSFrederic Weisbecker information to a file: whenever a process exits, information about 442391dc69cSFrederic Weisbecker that process will be appended to the file by the kernel. The 443391dc69cSFrederic Weisbecker information includes things such as creation time, owning user, 444391dc69cSFrederic Weisbecker command name, memory usage, controlling terminal etc. (the complete 445391dc69cSFrederic Weisbecker list is in the struct acct in <file:include/linux/acct.h>). It is 446391dc69cSFrederic Weisbecker up to the user level program to do useful things with this 447391dc69cSFrederic Weisbecker information. This is generally a good idea, so say Y. 448391dc69cSFrederic Weisbecker 449391dc69cSFrederic Weisbeckerconfig BSD_PROCESS_ACCT_V3 450391dc69cSFrederic Weisbecker bool "BSD Process Accounting version 3 file format" 451391dc69cSFrederic Weisbecker depends on BSD_PROCESS_ACCT 452391dc69cSFrederic Weisbecker default n 453391dc69cSFrederic Weisbecker help 454391dc69cSFrederic Weisbecker If you say Y here, the process accounting information is written 455391dc69cSFrederic Weisbecker in a new file format that also logs the process IDs of each 4563903bf94SRandy Dunlap process and its parent. Note that this file format is incompatible 457391dc69cSFrederic Weisbecker with previous v0/v1/v2 file formats, so you will need updated tools 458391dc69cSFrederic Weisbecker for processing it. A preliminary version of these tools is available 459391dc69cSFrederic Weisbecker at <http://www.gnu.org/software/acct/>. 460391dc69cSFrederic Weisbecker 461391dc69cSFrederic Weisbeckerconfig TASKSTATS 46219c92399SKees Cook bool "Export task/process statistics through netlink" 463391dc69cSFrederic Weisbecker depends on NET 4642813893fSIulia Manda depends on MULTIUSER 465391dc69cSFrederic Weisbecker default n 466391dc69cSFrederic Weisbecker help 467391dc69cSFrederic Weisbecker Export selected statistics for tasks/processes through the 468391dc69cSFrederic Weisbecker generic netlink interface. Unlike BSD process accounting, the 469391dc69cSFrederic Weisbecker statistics are available during the lifetime of tasks/processes as 470391dc69cSFrederic Weisbecker responses to commands. Like BSD accounting, they are sent to user 471391dc69cSFrederic Weisbecker space on task exit. 472391dc69cSFrederic Weisbecker 473391dc69cSFrederic Weisbecker Say N if unsure. 474391dc69cSFrederic Weisbecker 475391dc69cSFrederic Weisbeckerconfig TASK_DELAY_ACCT 47619c92399SKees Cook bool "Enable per-task delay accounting" 477391dc69cSFrederic Weisbecker depends on TASKSTATS 478f6db8347SNaveen N. Rao select SCHED_INFO 479391dc69cSFrederic Weisbecker help 480391dc69cSFrederic Weisbecker Collect information on time spent by a task waiting for system 481391dc69cSFrederic Weisbecker resources like cpu, synchronous block I/O completion and swapping 482391dc69cSFrederic Weisbecker in pages. Such statistics can help in setting a task's priorities 483391dc69cSFrederic Weisbecker relative to other tasks for cpu, io, rss limits etc. 484391dc69cSFrederic Weisbecker 485391dc69cSFrederic Weisbecker Say N if unsure. 486391dc69cSFrederic Weisbecker 487391dc69cSFrederic Weisbeckerconfig TASK_XACCT 48819c92399SKees Cook bool "Enable extended accounting over taskstats" 489391dc69cSFrederic Weisbecker depends on TASKSTATS 490391dc69cSFrederic Weisbecker help 491391dc69cSFrederic Weisbecker Collect extended task accounting data and send the data 492391dc69cSFrederic Weisbecker to userland for processing over the taskstats interface. 493391dc69cSFrederic Weisbecker 494391dc69cSFrederic Weisbecker Say N if unsure. 495391dc69cSFrederic Weisbecker 496391dc69cSFrederic Weisbeckerconfig TASK_IO_ACCOUNTING 49719c92399SKees Cook bool "Enable per-task storage I/O accounting" 498391dc69cSFrederic Weisbecker depends on TASK_XACCT 499391dc69cSFrederic Weisbecker help 500391dc69cSFrederic Weisbecker Collect information on the number of bytes of storage I/O which this 501391dc69cSFrederic Weisbecker task has caused. 502391dc69cSFrederic Weisbecker 503391dc69cSFrederic Weisbecker Say N if unsure. 504391dc69cSFrederic Weisbecker 505eb414681SJohannes Weinerconfig PSI 506eb414681SJohannes Weiner bool "Pressure stall information tracking" 507eb414681SJohannes Weiner help 508eb414681SJohannes Weiner Collect metrics that indicate how overcommitted the CPU, memory, 509eb414681SJohannes Weiner and IO capacity are in the system. 510eb414681SJohannes Weiner 511eb414681SJohannes Weiner If you say Y here, the kernel will create /proc/pressure/ with the 512eb414681SJohannes Weiner pressure statistics files cpu, memory, and io. These will indicate 513eb414681SJohannes Weiner the share of walltime in which some or all tasks in the system are 514eb414681SJohannes Weiner delayed due to contention of the respective resource. 515eb414681SJohannes Weiner 5162ce7135aSJohannes Weiner In kernels with cgroup support, cgroups (cgroup2 only) will 5172ce7135aSJohannes Weiner have cpu.pressure, memory.pressure, and io.pressure files, 5182ce7135aSJohannes Weiner which aggregate pressure stalls for the grouped tasks only. 5192ce7135aSJohannes Weiner 520eb414681SJohannes Weiner For more details see Documentation/accounting/psi.txt. 521eb414681SJohannes Weiner 522eb414681SJohannes Weiner Say N if unsure. 523eb414681SJohannes Weiner 524e0c27447SJohannes Weinerconfig PSI_DEFAULT_DISABLED 525e0c27447SJohannes Weiner bool "Require boot parameter to enable pressure stall information tracking" 526e0c27447SJohannes Weiner default n 527e0c27447SJohannes Weiner depends on PSI 528e0c27447SJohannes Weiner help 529e0c27447SJohannes Weiner If set, pressure stall information tracking will be disabled 530428a1cb4SBaruch Siach per default but can be enabled through passing psi=1 on the 531428a1cb4SBaruch Siach kernel commandline during boot. 532e0c27447SJohannes Weiner 5337b2489d3SJohannes Weiner This feature adds some code to the task wakeup and sleep 5347b2489d3SJohannes Weiner paths of the scheduler. The overhead is too low to affect 5357b2489d3SJohannes Weiner common scheduling-intense workloads in practice (such as 5367b2489d3SJohannes Weiner webservers, memcache), but it does show up in artificial 5377b2489d3SJohannes Weiner scheduler stress tests, such as hackbench. 5387b2489d3SJohannes Weiner 5397b2489d3SJohannes Weiner If you are paranoid and not sure what the kernel will be 5407b2489d3SJohannes Weiner used for, say Y. 5417b2489d3SJohannes Weiner 5427b2489d3SJohannes Weiner Say N if unsure. 5437b2489d3SJohannes Weiner 544391dc69cSFrederic Weisbeckerendmenu # "CPU/Task time and stats accounting" 545391dc69cSFrederic Weisbecker 5465c4991e2SFrederic Weisbeckerconfig CPU_ISOLATION 5475c4991e2SFrederic Weisbecker bool "CPU isolation" 548414a2dc1SGeert Uytterhoeven depends on SMP || COMPILE_TEST 5492c43838cSFrederic Weisbecker default y 5505c4991e2SFrederic Weisbecker help 5515c4991e2SFrederic Weisbecker Make sure that CPUs running critical tasks are not disturbed by 5525c4991e2SFrederic Weisbecker any source of "noise" such as unbound workqueues, timers, kthreads... 5532c43838cSFrederic Weisbecker Unbound jobs get offloaded to housekeeping CPUs. This is driven by 5542c43838cSFrederic Weisbecker the "isolcpus=" boot parameter. 5552c43838cSFrederic Weisbecker 5562c43838cSFrederic Weisbecker Say Y if unsure. 5575c4991e2SFrederic Weisbecker 5580af92d46SPaul E. McKenneysource "kernel/rcu/Kconfig" 559c903ff83SMike Travis 560de5b56baSVivek Goyalconfig BUILD_BIN2C 561de5b56baSVivek Goyal bool 562de5b56baSVivek Goyal default n 563de5b56baSVivek Goyal 5641da177e4SLinus Torvaldsconfig IKCONFIG 565f2443ab6SRoss Biro tristate "Kernel .config support" 5661da177e4SLinus Torvalds ---help--- 5671da177e4SLinus Torvalds This option enables the complete Linux kernel ".config" file 5681da177e4SLinus Torvalds contents to be saved in the kernel. It provides documentation 5691da177e4SLinus Torvalds of which kernel options are used in a running kernel or in an 5701da177e4SLinus Torvalds on-disk kernel. This information can be extracted from the kernel 5711da177e4SLinus Torvalds image file with the script scripts/extract-ikconfig and used as 5721da177e4SLinus Torvalds input to rebuild the current kernel or to build another kernel. 5731da177e4SLinus Torvalds It can also be extracted from a running kernel by reading 5741da177e4SLinus Torvalds /proc/config.gz if enabled (below). 5751da177e4SLinus Torvalds 5761da177e4SLinus Torvaldsconfig IKCONFIG_PROC 5771da177e4SLinus Torvalds bool "Enable access to .config through /proc/config.gz" 5781da177e4SLinus Torvalds depends on IKCONFIG && PROC_FS 5791da177e4SLinus Torvalds ---help--- 5801da177e4SLinus Torvalds This option enables access to the kernel configuration file 5811da177e4SLinus Torvalds through /proc/config.gz. 5821da177e4SLinus Torvalds 58343d8ce9dSJoel Fernandes (Google)config IKHEADERS_PROC 58443d8ce9dSJoel Fernandes (Google) tristate "Enable kernel header artifacts through /proc/kheaders.tar.xz" 58543d8ce9dSJoel Fernandes (Google) depends on PROC_FS 58643d8ce9dSJoel Fernandes (Google) help 58743d8ce9dSJoel Fernandes (Google) This option enables access to the kernel header and other artifacts that 58843d8ce9dSJoel Fernandes (Google) are generated during the build process. These can be used to build eBPF 58943d8ce9dSJoel Fernandes (Google) tracing programs, or similar programs. If you build the headers as a 59043d8ce9dSJoel Fernandes (Google) module, a module called kheaders.ko is built which can be loaded on-demand 59143d8ce9dSJoel Fernandes (Google) to get access to the headers. 59243d8ce9dSJoel Fernandes (Google) 593794543a2SAlistair John Strachanconfig LOG_BUF_SHIFT 594794543a2SAlistair John Strachan int "Kernel log buffer size (16 => 64KB, 17 => 128KB)" 595fb39f98dSIngo Molnar range 12 25 596f17a32e9SAdrian Bunk default 17 597361e9dfbSJosh Triplett depends on PRINTK 598794543a2SAlistair John Strachan help 59923b2899fSLuis R. Rodriguez Select the minimal kernel log buffer size as a power of 2. 60023b2899fSLuis R. Rodriguez The final size is affected by LOG_CPU_MAX_BUF_SHIFT config 60123b2899fSLuis R. Rodriguez parameter, see below. Any higher size also might be forced 60223b2899fSLuis R. Rodriguez by "log_buf_len" boot parameter. 60323b2899fSLuis R. Rodriguez 604f17a32e9SAdrian Bunk Examples: 605f17a32e9SAdrian Bunk 17 => 128 KB 606f17a32e9SAdrian Bunk 16 => 64 KB 607f17a32e9SAdrian Bunk 15 => 32 KB 608f17a32e9SAdrian Bunk 14 => 16 KB 609794543a2SAlistair John Strachan 13 => 8 KB 610794543a2SAlistair John Strachan 12 => 4 KB 611794543a2SAlistair John Strachan 61223b2899fSLuis R. Rodriguezconfig LOG_CPU_MAX_BUF_SHIFT 61323b2899fSLuis R. Rodriguez int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)" 6142240a31dSGeert Uytterhoeven depends on SMP 61523b2899fSLuis R. Rodriguez range 0 21 61623b2899fSLuis R. Rodriguez default 12 if !BASE_SMALL 61723b2899fSLuis R. Rodriguez default 0 if BASE_SMALL 618361e9dfbSJosh Triplett depends on PRINTK 61923b2899fSLuis R. Rodriguez help 62023b2899fSLuis R. Rodriguez This option allows to increase the default ring buffer size 62123b2899fSLuis R. Rodriguez according to the number of CPUs. The value defines the contribution 62223b2899fSLuis R. Rodriguez of each CPU as a power of 2. The used space is typically only few 62323b2899fSLuis R. Rodriguez lines however it might be much more when problems are reported, 62423b2899fSLuis R. Rodriguez e.g. backtraces. 62523b2899fSLuis R. Rodriguez 62623b2899fSLuis R. Rodriguez The increased size means that a new buffer has to be allocated and 62723b2899fSLuis R. Rodriguez the original static one is unused. It makes sense only on systems 62823b2899fSLuis R. Rodriguez with more CPUs. Therefore this value is used only when the sum of 62923b2899fSLuis R. Rodriguez contributions is greater than the half of the default kernel ring 63023b2899fSLuis R. Rodriguez buffer as defined by LOG_BUF_SHIFT. The default values are set 63123b2899fSLuis R. Rodriguez so that more than 64 CPUs are needed to trigger the allocation. 63223b2899fSLuis R. Rodriguez 63323b2899fSLuis R. Rodriguez Also this option is ignored when "log_buf_len" kernel parameter is 63423b2899fSLuis R. Rodriguez used as it forces an exact (power of two) size of the ring buffer. 63523b2899fSLuis R. Rodriguez 63623b2899fSLuis R. Rodriguez The number of possible CPUs is used for this computation ignoring 6375e0d8d59SGeert Uytterhoeven hotplugging making the computation optimal for the worst case 6385e0d8d59SGeert Uytterhoeven scenario while allowing a simple algorithm to be used from bootup. 63923b2899fSLuis R. Rodriguez 64023b2899fSLuis R. Rodriguez Examples shift values and their meaning: 64123b2899fSLuis R. Rodriguez 17 => 128 KB for each CPU 64223b2899fSLuis R. Rodriguez 16 => 64 KB for each CPU 64323b2899fSLuis R. Rodriguez 15 => 32 KB for each CPU 64423b2899fSLuis R. Rodriguez 14 => 16 KB for each CPU 64523b2899fSLuis R. Rodriguez 13 => 8 KB for each CPU 64623b2899fSLuis R. Rodriguez 12 => 4 KB for each CPU 64723b2899fSLuis R. Rodriguez 648f92bac3bSSergey Senozhatskyconfig PRINTK_SAFE_LOG_BUF_SHIFT 649f92bac3bSSergey Senozhatsky int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)" 650427934b8SPetr Mladek range 10 21 651427934b8SPetr Mladek default 13 652f92bac3bSSergey Senozhatsky depends on PRINTK 653427934b8SPetr Mladek help 654f92bac3bSSergey Senozhatsky Select the size of an alternate printk per-CPU buffer where messages 655f92bac3bSSergey Senozhatsky printed from usafe contexts are temporary stored. One example would 656f92bac3bSSergey Senozhatsky be NMI messages, another one - printk recursion. The messages are 657f92bac3bSSergey Senozhatsky copied to the main log buffer in a safe context to avoid a deadlock. 658f92bac3bSSergey Senozhatsky The value defines the size as a power of 2. 659427934b8SPetr Mladek 660f92bac3bSSergey Senozhatsky Those messages are rare and limited. The largest one is when 661427934b8SPetr Mladek a backtrace is printed. It usually fits into 4KB. Select 662427934b8SPetr Mladek 8KB if you want to be on the safe side. 663427934b8SPetr Mladek 664427934b8SPetr Mladek Examples: 665427934b8SPetr Mladek 17 => 128 KB for each CPU 666427934b8SPetr Mladek 16 => 64 KB for each CPU 667427934b8SPetr Mladek 15 => 32 KB for each CPU 668427934b8SPetr Mladek 14 => 16 KB for each CPU 669427934b8SPetr Mladek 13 => 8 KB for each CPU 670427934b8SPetr Mladek 12 => 4 KB for each CPU 671427934b8SPetr Mladek 6725cdc38f9SKAMEZAWA Hiroyuki# 6735cdc38f9SKAMEZAWA Hiroyuki# Architectures with an unreliable sched_clock() should select this: 6745cdc38f9SKAMEZAWA Hiroyuki# 6755cdc38f9SKAMEZAWA Hiroyukiconfig HAVE_UNSTABLE_SCHED_CLOCK 6765cdc38f9SKAMEZAWA Hiroyuki bool 6775cdc38f9SKAMEZAWA Hiroyuki 67838ff87f7SStephen Boydconfig GENERIC_SCHED_CLOCK 67938ff87f7SStephen Boyd bool 68038ff87f7SStephen Boyd 681be3a7284SAndrea Arcangeli# 682be3a7284SAndrea Arcangeli# For architectures that want to enable the support for NUMA-affine scheduler 683be3a7284SAndrea Arcangeli# balancing logic: 684be3a7284SAndrea Arcangeli# 685be3a7284SAndrea Arcangeliconfig ARCH_SUPPORTS_NUMA_BALANCING 686be3a7284SAndrea Arcangeli bool 687be3a7284SAndrea Arcangeli 688be5e610cSPeter Zijlstra# 68972b252aeSMel Gorman# For architectures that prefer to flush all TLBs after a number of pages 69072b252aeSMel Gorman# are unmapped instead of sending one IPI per page to flush. The architecture 69172b252aeSMel Gorman# must provide guarantees on what happens if a clean TLB cache entry is 69272b252aeSMel Gorman# written after the unmap. Details are in mm/rmap.c near the check for 69372b252aeSMel Gorman# should_defer_flush. The architecture should also consider if the full flush 69472b252aeSMel Gorman# and the refill costs are offset by the savings of sending fewer IPIs. 69572b252aeSMel Gormanconfig ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH 69672b252aeSMel Gorman bool 69772b252aeSMel Gorman 69872b252aeSMel Gorman# 699be5e610cSPeter Zijlstra# For architectures that know their GCC __int128 support is sound 700be5e610cSPeter Zijlstra# 701be5e610cSPeter Zijlstraconfig ARCH_SUPPORTS_INT128 702be5e610cSPeter Zijlstra bool 703be5e610cSPeter Zijlstra 704be3a7284SAndrea Arcangeli# For architectures that (ab)use NUMA to represent different memory regions 705be3a7284SAndrea Arcangeli# all cpu-local but of different latencies, such as SuperH. 706be3a7284SAndrea Arcangeli# 707be3a7284SAndrea Arcangeliconfig ARCH_WANT_NUMA_VARIABLE_LOCALITY 708be3a7284SAndrea Arcangeli bool 709be3a7284SAndrea Arcangeli 710be3a7284SAndrea Arcangeliconfig NUMA_BALANCING 711be3a7284SAndrea Arcangeli bool "Memory placement aware NUMA scheduler" 712be3a7284SAndrea Arcangeli depends on ARCH_SUPPORTS_NUMA_BALANCING 713be3a7284SAndrea Arcangeli depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY 714be3a7284SAndrea Arcangeli depends on SMP && NUMA && MIGRATION 715be3a7284SAndrea Arcangeli help 716be3a7284SAndrea Arcangeli This option adds support for automatic NUMA aware memory/task placement. 717be3a7284SAndrea Arcangeli The mechanism is quite primitive and is based on migrating memory when 7186d56a410SPaul Gortmaker it has references to the node the task is running on. 719be3a7284SAndrea Arcangeli 720be3a7284SAndrea Arcangeli This system will be inactive on UMA systems. 721be3a7284SAndrea Arcangeli 7226f7c97e8SAneesh Kumar K.Vconfig NUMA_BALANCING_DEFAULT_ENABLED 7236f7c97e8SAneesh Kumar K.V bool "Automatically enable NUMA aware memory/task placement" 7246f7c97e8SAneesh Kumar K.V default y 7256f7c97e8SAneesh Kumar K.V depends on NUMA_BALANCING 7266f7c97e8SAneesh Kumar K.V help 7276f7c97e8SAneesh Kumar K.V If set, automatic NUMA balancing will be enabled if running on a NUMA 7286f7c97e8SAneesh Kumar K.V machine. 7296f7c97e8SAneesh Kumar K.V 73023964d2dSLi Zefanmenuconfig CGROUPS 7316341e62bSChristoph Jaeger bool "Control Group support" 7322bd59d48STejun Heo select KERNFS 733ddbcc7e8SPaul Menage help 73423964d2dSLi Zefan This option adds support for grouping sets of processes together, for 7355cdc38f9SKAMEZAWA Hiroyuki use with process control subsystems such as Cpusets, CFS, memory 7365cdc38f9SKAMEZAWA Hiroyuki controls or device isolation. 7375cdc38f9SKAMEZAWA Hiroyuki See 7385cdc38f9SKAMEZAWA Hiroyuki - Documentation/scheduler/sched-design-CFS.txt (CFS) 7399991a9c8Sseokhoon.yoon - Documentation/cgroup-v1/ (features for grouping, isolation 74045ce80fbSLi Zefan and resource control) 741ddbcc7e8SPaul Menage 742ddbcc7e8SPaul Menage Say N if unsure. 743ddbcc7e8SPaul Menage 74423964d2dSLi Zefanif CGROUPS 74523964d2dSLi Zefan 7463e32cb2eSJohannes Weinerconfig PAGE_COUNTER 7473e32cb2eSJohannes Weiner bool 7483e32cb2eSJohannes Weiner 749c255a458SAndrew Mortonconfig MEMCG 750a0166ec4SJohannes Weiner bool "Memory controller" 7513e32cb2eSJohannes Weiner select PAGE_COUNTER 75279bd9814STejun Heo select EVENTFD 75300f0b825SBalbir Singh help 754a0166ec4SJohannes Weiner Provides control over the memory footprint of tasks in a cgroup. 75500f0b825SBalbir Singh 756c255a458SAndrew Mortonconfig MEMCG_SWAP 757a0166ec4SJohannes Weiner bool "Swap controller" 758c255a458SAndrew Morton depends on MEMCG && SWAP 759c077719bSKAMEZAWA Hiroyuki help 760a0166ec4SJohannes Weiner Provides control over the swap space consumed by tasks in a cgroup. 761a0166ec4SJohannes Weiner 762c255a458SAndrew Mortonconfig MEMCG_SWAP_ENABLED 763a0166ec4SJohannes Weiner bool "Swap controller enabled by default" 764c255a458SAndrew Morton depends on MEMCG_SWAP 765a42c390cSMichal Hocko default y 766a42c390cSMichal Hocko help 767a42c390cSMichal Hocko Memory Resource Controller Swap Extension comes with its price in 768a42c390cSMichal Hocko a bigger memory consumption. General purpose distribution kernels 76943d547f9SJim Cromie which want to enable the feature but keep it disabled by default 77007555ac1SMichal Hocko and let the user enable it by swapaccount=1 boot command line 771a42c390cSMichal Hocko parameter should have this option unselected. 772a42c390cSMichal Hocko For those who want to have the feature enabled by default should 773a42c390cSMichal Hocko select this option (if, for some reason, they need to disable it 77400a66d29SWANG Cong then swapaccount=0 does the trick). 775c077719bSKAMEZAWA Hiroyuki 77684c07d11SKirill Tkhaiconfig MEMCG_KMEM 77784c07d11SKirill Tkhai bool 77884c07d11SKirill Tkhai depends on MEMCG && !SLOB 77984c07d11SKirill Tkhai default y 78084c07d11SKirill Tkhai 7816bf024e6SJohannes Weinerconfig BLK_CGROUP 7826bf024e6SJohannes Weiner bool "IO controller" 7836bf024e6SJohannes Weiner depends on BLOCK 7842bc64a20SAneesh Kumar K.V default n 7856bf024e6SJohannes Weiner ---help--- 7866bf024e6SJohannes Weiner Generic block IO controller cgroup interface. This is the common 7876bf024e6SJohannes Weiner cgroup interface which should be used by various IO controlling 7886bf024e6SJohannes Weiner policies. 7892bc64a20SAneesh Kumar K.V 7906bf024e6SJohannes Weiner Currently, CFQ IO scheduler uses it to recognize task groups and 7916bf024e6SJohannes Weiner control disk bandwidth allocation (proportional time slice allocation) 7926bf024e6SJohannes Weiner to such task groups. It is also used by bio throttling logic in 7936bf024e6SJohannes Weiner block layer to implement upper limit in IO rates on a device. 794e5d1367fSStephane Eranian 7956bf024e6SJohannes Weiner This option only enables generic Block IO controller infrastructure. 7966bf024e6SJohannes Weiner One needs to also enable actual IO controlling logic/policy. For 7976bf024e6SJohannes Weiner enabling proportional weight division of disk bandwidth in CFQ, set 7986bf024e6SJohannes Weiner CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set 7996bf024e6SJohannes Weiner CONFIG_BLK_DEV_THROTTLING=y. 8006bf024e6SJohannes Weiner 8019991a9c8Sseokhoon.yoon See Documentation/cgroup-v1/blkio-controller.txt for more information. 8026bf024e6SJohannes Weiner 8036bf024e6SJohannes Weinerconfig DEBUG_BLK_CGROUP 8046bf024e6SJohannes Weiner bool "IO controller debugging" 8056bf024e6SJohannes Weiner depends on BLK_CGROUP 8066bf024e6SJohannes Weiner default n 8076bf024e6SJohannes Weiner ---help--- 8086bf024e6SJohannes Weiner Enable some debugging help. Currently it exports additional stat 8096bf024e6SJohannes Weiner files in a cgroup which can be useful for debugging. 8106bf024e6SJohannes Weiner 8116bf024e6SJohannes Weinerconfig CGROUP_WRITEBACK 8126bf024e6SJohannes Weiner bool 8136bf024e6SJohannes Weiner depends on MEMCG && BLK_CGROUP 8146bf024e6SJohannes Weiner default y 815e5d1367fSStephane Eranian 8167c941438SDhaval Gianimenuconfig CGROUP_SCHED 817a0166ec4SJohannes Weiner bool "CPU controller" 8187c941438SDhaval Giani default n 8197c941438SDhaval Giani help 8207c941438SDhaval Giani This feature lets CPU scheduler recognize task groups and control CPU 8217c941438SDhaval Giani bandwidth allocation to such task groups. It uses cgroups to group 8227c941438SDhaval Giani tasks. 8237c941438SDhaval Giani 8247c941438SDhaval Gianiif CGROUP_SCHED 8257c941438SDhaval Gianiconfig FAIR_GROUP_SCHED 8267c941438SDhaval Giani bool "Group scheduling for SCHED_OTHER" 8277c941438SDhaval Giani depends on CGROUP_SCHED 8287c941438SDhaval Giani default CGROUP_SCHED 8297c941438SDhaval Giani 830ab84d31eSPaul Turnerconfig CFS_BANDWIDTH 831ab84d31eSPaul Turner bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED" 832ab84d31eSPaul Turner depends on FAIR_GROUP_SCHED 833ab84d31eSPaul Turner default n 834ab84d31eSPaul Turner help 835ab84d31eSPaul Turner This option allows users to define CPU bandwidth rates (limits) for 836ab84d31eSPaul Turner tasks running within the fair group scheduler. Groups with no limit 837ab84d31eSPaul Turner set are considered to be unconstrained and will run with no 838ab84d31eSPaul Turner restriction. 839cd33d880SSebastian Andrzej Siewior See Documentation/scheduler/sched-bwc.txt for more information. 840ab84d31eSPaul Turner 8417c941438SDhaval Gianiconfig RT_GROUP_SCHED 8427c941438SDhaval Giani bool "Group scheduling for SCHED_RR/FIFO" 8437c941438SDhaval Giani depends on CGROUP_SCHED 8447c941438SDhaval Giani default n 8457c941438SDhaval Giani help 8467c941438SDhaval Giani This feature lets you explicitly allocate real CPU bandwidth 84732bd7eb5SLi Zefan to task groups. If enabled, it will also make it impossible to 8487c941438SDhaval Giani schedule realtime tasks for non-root users until you allocate 8497c941438SDhaval Giani realtime bandwidth for them. 8507c941438SDhaval Giani See Documentation/scheduler/sched-rt-group.txt for more information. 8517c941438SDhaval Giani 8527c941438SDhaval Gianiendif #CGROUP_SCHED 8537c941438SDhaval Giani 8546bf024e6SJohannes Weinerconfig CGROUP_PIDS 8556bf024e6SJohannes Weiner bool "PIDs controller" 8566bf024e6SJohannes Weiner help 8576bf024e6SJohannes Weiner Provides enforcement of process number limits in the scope of a 8586bf024e6SJohannes Weiner cgroup. Any attempt to fork more processes than is allowed in the 8596bf024e6SJohannes Weiner cgroup will fail. PIDs are fundamentally a global resource because it 8606bf024e6SJohannes Weiner is fairly trivial to reach PID exhaustion before you reach even a 8616bf024e6SJohannes Weiner conservative kmemcg limit. As a result, it is possible to grind a 8626bf024e6SJohannes Weiner system to halt without being limited by other cgroup policies. The 8636cc578dfSParav Pandit PIDs controller is designed to stop this from happening. 8646bf024e6SJohannes Weiner 8656bf024e6SJohannes Weiner It should be noted that organisational operations (such as attaching 86698076833SJonathan Neuschäfer to a cgroup hierarchy) will *not* be blocked by the PIDs controller, 8676bf024e6SJohannes Weiner since the PIDs limit only affects a process's ability to fork, not to 8686bf024e6SJohannes Weiner attach to a cgroup. 8696bf024e6SJohannes Weiner 87039d3e758SParav Panditconfig CGROUP_RDMA 87139d3e758SParav Pandit bool "RDMA controller" 87239d3e758SParav Pandit help 87339d3e758SParav Pandit Provides enforcement of RDMA resources defined by IB stack. 87439d3e758SParav Pandit It is fairly easy for consumers to exhaust RDMA resources, which 87539d3e758SParav Pandit can result into resource unavailability to other consumers. 87639d3e758SParav Pandit RDMA controller is designed to stop this from happening. 87739d3e758SParav Pandit Attaching processes with active RDMA resources to the cgroup 87839d3e758SParav Pandit hierarchy is allowed even if can cross the hierarchy's limit. 87939d3e758SParav Pandit 8806bf024e6SJohannes Weinerconfig CGROUP_FREEZER 8816bf024e6SJohannes Weiner bool "Freezer controller" 8826bf024e6SJohannes Weiner help 8836bf024e6SJohannes Weiner Provides a way to freeze and unfreeze all tasks in a 8846bf024e6SJohannes Weiner cgroup. 8856bf024e6SJohannes Weiner 886489c2a20SJohannes Weiner This option affects the ORIGINAL cgroup interface. The cgroup2 memory 887489c2a20SJohannes Weiner controller includes important in-kernel memory consumers per default. 888489c2a20SJohannes Weiner 889489c2a20SJohannes Weiner If you're using cgroup2, say N. 890489c2a20SJohannes Weiner 8916bf024e6SJohannes Weinerconfig CGROUP_HUGETLB 8926bf024e6SJohannes Weiner bool "HugeTLB controller" 8936bf024e6SJohannes Weiner depends on HUGETLB_PAGE 8946bf024e6SJohannes Weiner select PAGE_COUNTER 895afc24d49SVivek Goyal default n 8966bf024e6SJohannes Weiner help 8976bf024e6SJohannes Weiner Provides a cgroup controller for HugeTLB pages. 8986bf024e6SJohannes Weiner When you enable this, you can put a per cgroup limit on HugeTLB usage. 8996bf024e6SJohannes Weiner The limit is enforced during page fault. Since HugeTLB doesn't 9006bf024e6SJohannes Weiner support page reclaim, enforcing the limit at page fault time implies 9016bf024e6SJohannes Weiner that, the application will get SIGBUS signal if it tries to access 9026bf024e6SJohannes Weiner HugeTLB pages beyond its limit. This requires the application to know 9036bf024e6SJohannes Weiner beforehand how much HugeTLB pages it would require for its use. The 9046bf024e6SJohannes Weiner control group is tracked in the third page lru pointer. This means 9056bf024e6SJohannes Weiner that we cannot use the controller with huge page less than 3 pages. 906afc24d49SVivek Goyal 9076bf024e6SJohannes Weinerconfig CPUSETS 9086bf024e6SJohannes Weiner bool "Cpuset controller" 909e1d4eeecSNicolas Pitre depends on SMP 9106bf024e6SJohannes Weiner help 9116bf024e6SJohannes Weiner This option will let you create and manage CPUSETs which 9126bf024e6SJohannes Weiner allow dynamically partitioning a system into sets of CPUs and 9136bf024e6SJohannes Weiner Memory Nodes and assigning tasks to run only within those sets. 9146bf024e6SJohannes Weiner This is primarily useful on large SMP or NUMA systems. 915afc24d49SVivek Goyal 9166bf024e6SJohannes Weiner Say N if unsure. 917afc24d49SVivek Goyal 9186bf024e6SJohannes Weinerconfig PROC_PID_CPUSET 9196bf024e6SJohannes Weiner bool "Include legacy /proc/<pid>/cpuset file" 9206bf024e6SJohannes Weiner depends on CPUSETS 92189e9b9e0STejun Heo default y 92289e9b9e0STejun Heo 9236bf024e6SJohannes Weinerconfig CGROUP_DEVICE 9246bf024e6SJohannes Weiner bool "Device controller" 9256bf024e6SJohannes Weiner help 9266bf024e6SJohannes Weiner Provides a cgroup controller implementing whitelists for 9276bf024e6SJohannes Weiner devices which a process in the cgroup can mknod or open. 9286bf024e6SJohannes Weiner 9296bf024e6SJohannes Weinerconfig CGROUP_CPUACCT 9306bf024e6SJohannes Weiner bool "Simple CPU accounting controller" 9316bf024e6SJohannes Weiner help 9326bf024e6SJohannes Weiner Provides a simple controller for monitoring the 9336bf024e6SJohannes Weiner total CPU consumed by the tasks in a cgroup. 9346bf024e6SJohannes Weiner 9356bf024e6SJohannes Weinerconfig CGROUP_PERF 9366bf024e6SJohannes Weiner bool "Perf controller" 9376bf024e6SJohannes Weiner depends on PERF_EVENTS 9386bf024e6SJohannes Weiner help 9396bf024e6SJohannes Weiner This option extends the perf per-cpu mode to restrict monitoring 9406bf024e6SJohannes Weiner to threads which belong to the cgroup specified and run on the 9416bf024e6SJohannes Weiner designated cpu. 9426bf024e6SJohannes Weiner 9436bf024e6SJohannes Weiner Say N if unsure. 9446bf024e6SJohannes Weiner 94530070984SDaniel Mackconfig CGROUP_BPF 94630070984SDaniel Mack bool "Support for eBPF programs attached to cgroups" 947483c4933SAndy Lutomirski depends on BPF_SYSCALL 948483c4933SAndy Lutomirski select SOCK_CGROUP_DATA 94930070984SDaniel Mack help 95030070984SDaniel Mack Allow attaching eBPF programs to a cgroup using the bpf(2) 95130070984SDaniel Mack syscall command BPF_PROG_ATTACH. 95230070984SDaniel Mack 95330070984SDaniel Mack In which context these programs are accessed depends on the type 95430070984SDaniel Mack of attachment. For instance, programs that are attached using 95530070984SDaniel Mack BPF_CGROUP_INET_INGRESS will be executed on the ingress path of 95630070984SDaniel Mack inet sockets. 95730070984SDaniel Mack 9586bf024e6SJohannes Weinerconfig CGROUP_DEBUG 95923b0be48SWaiman Long bool "Debug controller" 9606bf024e6SJohannes Weiner default n 96123b0be48SWaiman Long depends on DEBUG_KERNEL 9626bf024e6SJohannes Weiner help 9636bf024e6SJohannes Weiner This option enables a simple controller that exports 96423b0be48SWaiman Long debugging information about the cgroups framework. This 96523b0be48SWaiman Long controller is for control cgroup debugging only. Its 96623b0be48SWaiman Long interfaces are not stable. 9676bf024e6SJohannes Weiner 9686bf024e6SJohannes Weiner Say N. 9696bf024e6SJohannes Weiner 97073b35147SArnd Bergmannconfig SOCK_CGROUP_DATA 97173b35147SArnd Bergmann bool 97273b35147SArnd Bergmann default n 97373b35147SArnd Bergmann 97423964d2dSLi Zefanendif # CGROUPS 975c077719bSKAMEZAWA Hiroyuki 9768dd2a82cSDaniel Lezcanomenuconfig NAMESPACES 9776a108a14SDavid Rientjes bool "Namespaces support" if EXPERT 9782813893fSIulia Manda depends on MULTIUSER 9796a108a14SDavid Rientjes default !EXPERT 980c5289a69SPavel Emelyanov help 981c5289a69SPavel Emelyanov Provides the way to make tasks work with different objects using 982c5289a69SPavel Emelyanov the same id. For example same IPC id may refer to different objects 983c5289a69SPavel Emelyanov or same user id or pid may refer to different tasks when used in 984c5289a69SPavel Emelyanov different namespaces. 985c5289a69SPavel Emelyanov 9868dd2a82cSDaniel Lezcanoif NAMESPACES 9878dd2a82cSDaniel Lezcano 98858bfdd6dSPavel Emelyanovconfig UTS_NS 98958bfdd6dSPavel Emelyanov bool "UTS namespace" 99017a6d441SDaniel Lezcano default y 99158bfdd6dSPavel Emelyanov help 99258bfdd6dSPavel Emelyanov In this namespace tasks see different info provided with the 99358bfdd6dSPavel Emelyanov uname() system call 99458bfdd6dSPavel Emelyanov 995ae5e1b22SPavel Emelyanovconfig IPC_NS 996ae5e1b22SPavel Emelyanov bool "IPC namespace" 9978dd2a82cSDaniel Lezcano depends on (SYSVIPC || POSIX_MQUEUE) 99817a6d441SDaniel Lezcano default y 999ae5e1b22SPavel Emelyanov help 1000ae5e1b22SPavel Emelyanov In this namespace tasks work with IPC ids which correspond to 1001614b84cfSSerge E. Hallyn different IPC objects in different namespaces. 1002ae5e1b22SPavel Emelyanov 1003aee16ce7SPavel Emelyanovconfig USER_NS 100419c92399SKees Cook bool "User namespace" 10055673a94cSEric W. Biederman default n 1006aee16ce7SPavel Emelyanov help 1007aee16ce7SPavel Emelyanov This allows containers, i.e. vservers, to use user namespaces 1008aee16ce7SPavel Emelyanov to provide different user info for different servers. 1009e11f0ae3SEric W. Biederman 1010e11f0ae3SEric W. Biederman When user namespaces are enabled in the kernel it is 1011d886f4e4SJohannes Weiner recommended that the MEMCG option also be enabled and that 1012d886f4e4SJohannes Weiner user-space use the memory control groups to limit the amount 1013d886f4e4SJohannes Weiner of memory a memory unprivileged users can use. 1014e11f0ae3SEric W. Biederman 1015aee16ce7SPavel Emelyanov If unsure, say N. 1016aee16ce7SPavel Emelyanov 101774bd59bbSPavel Emelyanovconfig PID_NS 10189bd38c2cSDaniel Lezcano bool "PID Namespaces" 101917a6d441SDaniel Lezcano default y 102074bd59bbSPavel Emelyanov help 102112d2b8f9SHeikki Orsila Support process id namespaces. This allows having multiple 1022692105b8SMatt LaPlante processes with the same pid as long as they are in different 102374bd59bbSPavel Emelyanov pid namespaces. This is a building block of containers. 102474bd59bbSPavel Emelyanov 1025d6eb633fSMatt Helsleyconfig NET_NS 1026d6eb633fSMatt Helsley bool "Network namespace" 10278dd2a82cSDaniel Lezcano depends on NET 102817a6d441SDaniel Lezcano default y 1029d6eb633fSMatt Helsley help 1030d6eb633fSMatt Helsley Allow user space to create what appear to be multiple instances 1031d6eb633fSMatt Helsley of the network stack. 1032d6eb633fSMatt Helsley 10338dd2a82cSDaniel Lezcanoendif # NAMESPACES 10348dd2a82cSDaniel Lezcano 10355cb366bbSAdrian Reberconfig CHECKPOINT_RESTORE 10365cb366bbSAdrian Reber bool "Checkpoint/restore support" 10375cb366bbSAdrian Reber select PROC_CHILDREN 10385cb366bbSAdrian Reber default n 10395cb366bbSAdrian Reber help 10405cb366bbSAdrian Reber Enables additional kernel features in a sake of checkpoint/restore. 10415cb366bbSAdrian Reber In particular it adds auxiliary prctl codes to setup process text, 10425cb366bbSAdrian Reber data and heap segment sizes, and a few additional /proc filesystem 10435cb366bbSAdrian Reber entries. 10445cb366bbSAdrian Reber 10455cb366bbSAdrian Reber If unsure, say N here. 10465cb366bbSAdrian Reber 10475091faa4SMike Galbraithconfig SCHED_AUTOGROUP 10485091faa4SMike Galbraith bool "Automatic process group scheduling" 10495091faa4SMike Galbraith select CGROUPS 10505091faa4SMike Galbraith select CGROUP_SCHED 10515091faa4SMike Galbraith select FAIR_GROUP_SCHED 10525091faa4SMike Galbraith help 10535091faa4SMike Galbraith This option optimizes the scheduler for common desktop workloads by 10545091faa4SMike Galbraith automatically creating and populating task groups. This separation 10555091faa4SMike Galbraith of workloads isolates aggressive CPU burners (like build jobs) from 10565091faa4SMike Galbraith desktop applications. Task group autogeneration is currently based 10575091faa4SMike Galbraith upon task session. 10585091faa4SMike Galbraith 10597af37becSDaniel Lezcanoconfig SYSFS_DEPRECATED 10605d6a4ea5SFerenc Wagner bool "Enable deprecated sysfs features to support old userspace tools" 10617af37becSDaniel Lezcano depends on SYSFS 10627af37becSDaniel Lezcano default n 10637af37becSDaniel Lezcano help 10647af37becSDaniel Lezcano This option adds code that switches the layout of the "block" class 10657af37becSDaniel Lezcano devices, to not show up in /sys/class/block/, but only in 10667af37becSDaniel Lezcano /sys/block/. 10677af37becSDaniel Lezcano 10687af37becSDaniel Lezcano This switch is only active when the sysfs.deprecated=1 boot option is 10697af37becSDaniel Lezcano passed or the SYSFS_DEPRECATED_V2 option is set. 10707af37becSDaniel Lezcano 10717af37becSDaniel Lezcano This option allows new kernels to run on old distributions and tools, 10727af37becSDaniel Lezcano which might get confused by /sys/class/block/. Since 2007/2008 all 10737af37becSDaniel Lezcano major distributions and tools handle this just fine. 10747af37becSDaniel Lezcano 10757af37becSDaniel Lezcano Recent distributions and userspace tools after 2009/2010 depend on 10767af37becSDaniel Lezcano the existence of /sys/class/block/, and will not work with this 10777af37becSDaniel Lezcano option enabled. 10787af37becSDaniel Lezcano 10797af37becSDaniel Lezcano Only if you are using a new kernel on an old distribution, you might 10807af37becSDaniel Lezcano need to say Y here. 10817af37becSDaniel Lezcano 10827af37becSDaniel Lezcanoconfig SYSFS_DEPRECATED_V2 10835d6a4ea5SFerenc Wagner bool "Enable deprecated sysfs features by default" 10847af37becSDaniel Lezcano default n 10857af37becSDaniel Lezcano depends on SYSFS 10867af37becSDaniel Lezcano depends on SYSFS_DEPRECATED 10877af37becSDaniel Lezcano help 10887af37becSDaniel Lezcano Enable deprecated sysfs by default. 10897af37becSDaniel Lezcano 10907af37becSDaniel Lezcano See the CONFIG_SYSFS_DEPRECATED option for more details about this 10917af37becSDaniel Lezcano option. 10927af37becSDaniel Lezcano 10937af37becSDaniel Lezcano Only if you are using a new kernel on an old distribution, you might 10947af37becSDaniel Lezcano need to say Y here. Even then, odds are you would not need it 10957af37becSDaniel Lezcano enabled, you can always pass the boot option if absolutely necessary. 10967af37becSDaniel Lezcano 10977af37becSDaniel Lezcanoconfig RELAY 10987af37becSDaniel Lezcano bool "Kernel->user space relay support (formerly relayfs)" 109926b5679eSPeter Zijlstra select IRQ_WORK 11007af37becSDaniel Lezcano help 11017af37becSDaniel Lezcano This option enables support for relay interface support in 11027af37becSDaniel Lezcano certain file systems (such as debugfs). 11037af37becSDaniel Lezcano It is designed to provide an efficient mechanism for tools and 11047af37becSDaniel Lezcano facilities to relay large amounts of data from kernel space to 11057af37becSDaniel Lezcano user space. 11067af37becSDaniel Lezcano 11077af37becSDaniel Lezcano If unsure, say N. 11087af37becSDaniel Lezcano 1109f991633dSDimitri Gorokhovikconfig BLK_DEV_INITRD 1110f991633dSDimitri Gorokhovik bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support" 1111f991633dSDimitri Gorokhovik help 1112f991633dSDimitri Gorokhovik The initial RAM filesystem is a ramfs which is loaded by the 1113f991633dSDimitri Gorokhovik boot loader (loadlin or lilo) and that is mounted as root 1114f991633dSDimitri Gorokhovik before the normal boot procedure. It is typically used to 1115f991633dSDimitri Gorokhovik load modules needed to mount the "real" root file system, 11168c27ceffSMauro Carvalho Chehab etc. See <file:Documentation/admin-guide/initrd.rst> for details. 1117f991633dSDimitri Gorokhovik 1118f991633dSDimitri Gorokhovik If RAM disk support (BLK_DEV_RAM) is also included, this 1119f991633dSDimitri Gorokhovik also enables initial RAM disk (initrd) support and adds 1120f991633dSDimitri Gorokhovik 15 Kbytes (more on some other architectures) to the kernel size. 1121f991633dSDimitri Gorokhovik 1122f991633dSDimitri Gorokhovik If unsure say Y. 1123f991633dSDimitri Gorokhovik 1124c33df4eaSJean-Paul Samanif BLK_DEV_INITRD 1125c33df4eaSJean-Paul Saman 1126dbec4866SSam Ravnborgsource "usr/Kconfig" 1127dbec4866SSam Ravnborg 1128c33df4eaSJean-Paul Samanendif 1129c33df4eaSJean-Paul Saman 1130877417e6SArnd Bergmannchoice 1131877417e6SArnd Bergmann prompt "Compiler optimization level" 11322cc3ce24SUlf Magnusson default CC_OPTIMIZE_FOR_PERFORMANCE 1133877417e6SArnd Bergmann 1134877417e6SArnd Bergmannconfig CC_OPTIMIZE_FOR_PERFORMANCE 1135877417e6SArnd Bergmann bool "Optimize for performance" 1136877417e6SArnd Bergmann help 1137877417e6SArnd Bergmann This is the default optimization level for the kernel, building 1138877417e6SArnd Bergmann with the "-O2" compiler flag for best performance and most 1139877417e6SArnd Bergmann helpful compile-time warnings. 1140877417e6SArnd Bergmann 1141c45b4f1fSLinus Torvaldsconfig CC_OPTIMIZE_FOR_SIZE 114296fffeb4SIngo Molnar bool "Optimize for size" 1143b303c6dfSMasahiro Yamada imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives 1144c45b4f1fSLinus Torvalds help 114531a4af7fSMasahiro Yamada Enabling this option will pass "-Os" instead of "-O2" to 114631a4af7fSMasahiro Yamada your compiler resulting in a smaller kernel. 1147c45b4f1fSLinus Torvalds 11483a55fb0dSKirill Smelkov If unsure, say N. 1149c45b4f1fSLinus Torvalds 1150877417e6SArnd Bergmannendchoice 1151877417e6SArnd Bergmann 11525d20ee31SNicholas Pigginconfig HAVE_LD_DEAD_CODE_DATA_ELIMINATION 11535d20ee31SNicholas Piggin bool 11545d20ee31SNicholas Piggin help 11555d20ee31SNicholas Piggin This requires that the arch annotates or otherwise protects 11565d20ee31SNicholas Piggin its external entry points from being discarded. Linker scripts 11575d20ee31SNicholas Piggin must also merge .text.*, .data.*, and .bss.* correctly into 11585d20ee31SNicholas Piggin output sections. Care must be taken not to pull in unrelated 11595d20ee31SNicholas Piggin sections (e.g., '.text.init'). Typically '.' in section names 11605d20ee31SNicholas Piggin is used to distinguish them from label names / C identifiers. 11615d20ee31SNicholas Piggin 11625d20ee31SNicholas Pigginconfig LD_DEAD_CODE_DATA_ELIMINATION 11635d20ee31SNicholas Piggin bool "Dead code and data elimination (EXPERIMENTAL)" 11645d20ee31SNicholas Piggin depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION 11655d20ee31SNicholas Piggin depends on EXPERT 116616fd20aaSPaul Burton depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800) 1167e85d1d65SMasahiro Yamada depends on $(cc-option,-ffunction-sections -fdata-sections) 1168e85d1d65SMasahiro Yamada depends on $(ld-option,--gc-sections) 11695d20ee31SNicholas Piggin help 11708b9d2712SMasahiro Yamada Enable this if you want to do dead code and data elimination with 11718b9d2712SMasahiro Yamada the linker by compiling with -ffunction-sections -fdata-sections, 11728b9d2712SMasahiro Yamada and linking with --gc-sections. 11735d20ee31SNicholas Piggin 11745d20ee31SNicholas Piggin This can reduce on disk and in-memory size of the kernel 11755d20ee31SNicholas Piggin code and static data, particularly for small configs and 11765d20ee31SNicholas Piggin on small systems. This has the possibility of introducing 11775d20ee31SNicholas Piggin silently broken kernel if the required annotations are not 11785d20ee31SNicholas Piggin present. This option is not well tested yet, so use at your 11795d20ee31SNicholas Piggin own risk. 11805d20ee31SNicholas Piggin 11810847062aSRandy Dunlapconfig SYSCTL 11820847062aSRandy Dunlap bool 11830847062aSRandy Dunlap 1184657a5209SMike Frysingerconfig HAVE_UID16 1185657a5209SMike Frysinger bool 1186657a5209SMike Frysinger 1187657a5209SMike Frysingerconfig SYSCTL_EXCEPTION_TRACE 1188657a5209SMike Frysinger bool 1189657a5209SMike Frysinger help 1190657a5209SMike Frysinger Enable support for /proc/sys/debug/exception-trace. 1191657a5209SMike Frysinger 1192657a5209SMike Frysingerconfig SYSCTL_ARCH_UNALIGN_NO_WARN 1193657a5209SMike Frysinger bool 1194657a5209SMike Frysinger help 1195657a5209SMike Frysinger Enable support for /proc/sys/kernel/ignore-unaligned-usertrap 1196657a5209SMike Frysinger Allows arch to define/use @no_unaligned_warning to possibly warn 1197657a5209SMike Frysinger about unaligned access emulation going on under the hood. 1198657a5209SMike Frysinger 1199657a5209SMike Frysingerconfig SYSCTL_ARCH_UNALIGN_ALLOW 1200657a5209SMike Frysinger bool 1201657a5209SMike Frysinger help 1202657a5209SMike Frysinger Enable support for /proc/sys/kernel/unaligned-trap 1203657a5209SMike Frysinger Allows arches to define/use @unaligned_enabled to runtime toggle 1204657a5209SMike Frysinger the unaligned access emulation. 1205657a5209SMike Frysinger see arch/parisc/kernel/unaligned.c for reference 1206657a5209SMike Frysinger 1207657a5209SMike Frysingerconfig HAVE_PCSPKR_PLATFORM 1208657a5209SMike Frysinger bool 1209657a5209SMike Frysinger 1210f89b7755SAlexei Starovoitov# interpreter that classic socket filters depend on 1211f89b7755SAlexei Starovoitovconfig BPF 1212f89b7755SAlexei Starovoitov bool 1213f89b7755SAlexei Starovoitov 12146a108a14SDavid Rientjesmenuconfig EXPERT 12156a108a14SDavid Rientjes bool "Configure standard kernel features (expert users)" 1216f505c553SJosh Triplett # Unhide debug options, to make the on-by-default options visible 1217f505c553SJosh Triplett select DEBUG_KERNEL 12181da177e4SLinus Torvalds help 12191da177e4SLinus Torvalds This option allows certain base kernel options and settings 12201da177e4SLinus Torvalds to be disabled or tweaked. This is for specialized 12211da177e4SLinus Torvalds environments which can tolerate a "non-standard" kernel. 12221da177e4SLinus Torvalds Only use this if you really know what you are doing. 12231da177e4SLinus Torvalds 1224ae81f9e3SChuck Ebbertconfig UID16 12256a108a14SDavid Rientjes bool "Enable 16-bit UID system calls" if EXPERT 12262813893fSIulia Manda depends on HAVE_UID16 && MULTIUSER 1227ae81f9e3SChuck Ebbert default y 1228ae81f9e3SChuck Ebbert help 1229ae81f9e3SChuck Ebbert This enables the legacy 16-bit UID syscall wrappers. 1230ae81f9e3SChuck Ebbert 12312813893fSIulia Mandaconfig MULTIUSER 12322813893fSIulia Manda bool "Multiple users, groups and capabilities support" if EXPERT 12332813893fSIulia Manda default y 12342813893fSIulia Manda help 12352813893fSIulia Manda This option enables support for non-root users, groups and 12362813893fSIulia Manda capabilities. 12372813893fSIulia Manda 12382813893fSIulia Manda If you say N here, all processes will run with UID 0, GID 0, and all 12392813893fSIulia Manda possible capabilities. Saying N here also compiles out support for 12402813893fSIulia Manda system calls related to UIDs, GIDs, and capabilities, such as setuid, 12412813893fSIulia Manda setgid, and capset. 12422813893fSIulia Manda 12432813893fSIulia Manda If unsure, say Y here. 12442813893fSIulia Manda 1245f6187769SFabian Frederickconfig SGETMASK_SYSCALL 1246f6187769SFabian Frederick bool "sgetmask/ssetmask syscalls support" if EXPERT 1247a687a533SArnd Bergmann def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH 1248f6187769SFabian Frederick ---help--- 1249f6187769SFabian Frederick sys_sgetmask and sys_ssetmask are obsolete system calls 1250f6187769SFabian Frederick no longer supported in libc but still enabled by default in some 1251f6187769SFabian Frederick architectures. 1252f6187769SFabian Frederick 1253f6187769SFabian Frederick If unsure, leave the default option here. 1254f6187769SFabian Frederick 12556af9f7bfSFabian Frederickconfig SYSFS_SYSCALL 12566af9f7bfSFabian Frederick bool "Sysfs syscall support" if EXPERT 12576af9f7bfSFabian Frederick default y 12586af9f7bfSFabian Frederick ---help--- 12596af9f7bfSFabian Frederick sys_sysfs is an obsolete system call no longer supported in libc. 12606af9f7bfSFabian Frederick Note that disabling this option is more secure but might break 12616af9f7bfSFabian Frederick compatibility with some systems. 12626af9f7bfSFabian Frederick 12636af9f7bfSFabian Frederick If unsure say Y here. 12646af9f7bfSFabian Frederick 1265b89a8171SEric W. Biedermanconfig SYSCTL_SYSCALL 12666a108a14SDavid Rientjes bool "Sysctl syscall support" if EXPERT 126726a7034bSEric W. Biederman depends on PROC_SYSCTL 1268c736de60SWANG Cong default n 1269b89a8171SEric W. Biederman select SYSCTL 1270b89a8171SEric W. Biederman ---help--- 127113bb7e37SEric W. Biederman sys_sysctl uses binary paths that have been found challenging 127213bb7e37SEric W. Biederman to properly maintain and use. The interface in /proc/sys 127313bb7e37SEric W. Biederman using paths with ascii names is now the primary path to this 127413bb7e37SEric W. Biederman information. 1275b89a8171SEric W. Biederman 127613bb7e37SEric W. Biederman Almost nothing using the binary sysctl interface so if you are 127713bb7e37SEric W. Biederman trying to save some space it is probably safe to disable this, 127813bb7e37SEric W. Biederman making your kernel marginally smaller. 1279b89a8171SEric W. Biederman 1280c736de60SWANG Cong If unsure say N here. 1281ae81f9e3SChuck Ebbert 1282d1b069f5SRandy Dunlapconfig FHANDLE 1283d1b069f5SRandy Dunlap bool "open by fhandle syscalls" if EXPERT 1284d1b069f5SRandy Dunlap select EXPORTFS 1285d1b069f5SRandy Dunlap default y 1286d1b069f5SRandy Dunlap help 1287d1b069f5SRandy Dunlap If you say Y here, a user level program will be able to map 1288d1b069f5SRandy Dunlap file names to handle and then later use the handle for 1289d1b069f5SRandy Dunlap different file system operations. This is useful in implementing 1290d1b069f5SRandy Dunlap userspace file servers, which now track files using handles instead 1291d1b069f5SRandy Dunlap of names. The handle would remain the same even if file names 1292d1b069f5SRandy Dunlap get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2) 1293d1b069f5SRandy Dunlap syscalls. 1294d1b069f5SRandy Dunlap 1295baa73d9eSNicolas Pitreconfig POSIX_TIMERS 1296baa73d9eSNicolas Pitre bool "Posix Clocks & timers" if EXPERT 1297baa73d9eSNicolas Pitre default y 1298baa73d9eSNicolas Pitre help 1299baa73d9eSNicolas Pitre This includes native support for POSIX timers to the kernel. 1300baa73d9eSNicolas Pitre Some embedded systems have no use for them and therefore they 1301baa73d9eSNicolas Pitre can be configured out to reduce the size of the kernel image. 1302baa73d9eSNicolas Pitre 1303baa73d9eSNicolas Pitre When this option is disabled, the following syscalls won't be 1304baa73d9eSNicolas Pitre available: timer_create, timer_gettime: timer_getoverrun, 1305baa73d9eSNicolas Pitre timer_settime, timer_delete, clock_adjtime, getitimer, 1306baa73d9eSNicolas Pitre setitimer, alarm. Furthermore, the clock_settime, clock_gettime, 1307baa73d9eSNicolas Pitre clock_getres and clock_nanosleep syscalls will be limited to 1308baa73d9eSNicolas Pitre CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only. 1309baa73d9eSNicolas Pitre 1310baa73d9eSNicolas Pitre If unsure say y. 1311baa73d9eSNicolas Pitre 1312d59745ceSMatt Mackallconfig PRINTK 1313d59745ceSMatt Mackall default y 13146a108a14SDavid Rientjes bool "Enable support for printk" if EXPERT 131574876a98SFrederic Weisbecker select IRQ_WORK 1316d59745ceSMatt Mackall help 1317d59745ceSMatt Mackall This option enables normal printk support. Removing it 1318d59745ceSMatt Mackall eliminates most of the message strings from the kernel image 1319d59745ceSMatt Mackall and makes the kernel more or less silent. As this makes it 1320d59745ceSMatt Mackall very difficult to diagnose system problems, saying N here is 1321d59745ceSMatt Mackall strongly discouraged. 1322d59745ceSMatt Mackall 132342a0bb3fSPetr Mladekconfig PRINTK_NMI 132442a0bb3fSPetr Mladek def_bool y 132542a0bb3fSPetr Mladek depends on PRINTK 132642a0bb3fSPetr Mladek depends on HAVE_NMI 132742a0bb3fSPetr Mladek 1328c8538a7aSMatt Mackallconfig BUG 13296a108a14SDavid Rientjes bool "BUG() support" if EXPERT 1330c8538a7aSMatt Mackall default y 1331c8538a7aSMatt Mackall help 1332c8538a7aSMatt Mackall Disabling this option eliminates support for BUG and WARN, reducing 1333c8538a7aSMatt Mackall the size of your kernel image and potentially quietly ignoring 1334c8538a7aSMatt Mackall numerous fatal conditions. You should only consider disabling this 1335c8538a7aSMatt Mackall option for embedded systems with no facilities for reporting errors. 1336c8538a7aSMatt Mackall Just say Y. 1337c8538a7aSMatt Mackall 1338708e9a79SMatt Mackallconfig ELF_CORE 1339046d662fSAlex Kelly depends on COREDUMP 1340708e9a79SMatt Mackall default y 13416a108a14SDavid Rientjes bool "Enable ELF core dumps" if EXPERT 1342708e9a79SMatt Mackall help 1343708e9a79SMatt Mackall Enable support for generating core dumps. Disabling saves about 4k. 1344708e9a79SMatt Mackall 13458761f1abSRalf Baechle 1346e5e1d3cbSStas Sergeevconfig PCSPKR_PLATFORM 13476a108a14SDavid Rientjes bool "Enable PC-Speaker support" if EXPERT 13488761f1abSRalf Baechle depends on HAVE_PCSPKR_PLATFORM 134915f304b6SRalf Baechle select I8253_LOCK 1350e5e1d3cbSStas Sergeev default y 1351e5e1d3cbSStas Sergeev help 1352e5e1d3cbSStas Sergeev This option allows to disable the internal PC-Speaker 1353e5e1d3cbSStas Sergeev support, saving some memory. 1354e5e1d3cbSStas Sergeev 13551da177e4SLinus Torvaldsconfig BASE_FULL 13561da177e4SLinus Torvalds default y 13576a108a14SDavid Rientjes bool "Enable full-sized data structures for core" if EXPERT 13581da177e4SLinus Torvalds help 13591da177e4SLinus Torvalds Disabling this option reduces the size of miscellaneous core 13601da177e4SLinus Torvalds kernel data structures. This saves memory on small machines, 13611da177e4SLinus Torvalds but may reduce performance. 13621da177e4SLinus Torvalds 13631da177e4SLinus Torvaldsconfig FUTEX 13646a108a14SDavid Rientjes bool "Enable futex support" if EXPERT 13651da177e4SLinus Torvalds default y 1366bc2eecd7SNicolas Pitre imply RT_MUTEXES 13671da177e4SLinus Torvalds help 13681da177e4SLinus Torvalds Disabling this option will cause the kernel to be built without 13691da177e4SLinus Torvalds support for "fast userspace mutexes". The resulting kernel may not 13701da177e4SLinus Torvalds run glibc-based applications correctly. 13711da177e4SLinus Torvalds 1372bc2eecd7SNicolas Pitreconfig FUTEX_PI 1373bc2eecd7SNicolas Pitre bool 1374bc2eecd7SNicolas Pitre depends on FUTEX && RT_MUTEXES 1375bc2eecd7SNicolas Pitre default y 1376bc2eecd7SNicolas Pitre 137703b8c7b6SHeiko Carstensconfig HAVE_FUTEX_CMPXCHG 137803b8c7b6SHeiko Carstens bool 137962b4d204SJosh Triplett depends on FUTEX 138003b8c7b6SHeiko Carstens help 138103b8c7b6SHeiko Carstens Architectures should select this if futex_atomic_cmpxchg_inatomic() 138203b8c7b6SHeiko Carstens is implemented and always working. This removes a couple of runtime 138303b8c7b6SHeiko Carstens checks. 138403b8c7b6SHeiko Carstens 13851da177e4SLinus Torvaldsconfig EPOLL 13866a108a14SDavid Rientjes bool "Enable eventpoll support" if EXPERT 13871da177e4SLinus Torvalds default y 13881da177e4SLinus Torvalds help 13891da177e4SLinus Torvalds Disabling this option will cause the kernel to be built without 13901da177e4SLinus Torvalds support for epoll family of system calls. 13911da177e4SLinus Torvalds 1392fba2afaaSDavide Libenziconfig SIGNALFD 13936a108a14SDavid Rientjes bool "Enable signalfd() system call" if EXPERT 1394fba2afaaSDavide Libenzi default y 1395fba2afaaSDavide Libenzi help 1396fba2afaaSDavide Libenzi Enable the signalfd() system call that allows to receive signals 1397fba2afaaSDavide Libenzi on a file descriptor. 1398fba2afaaSDavide Libenzi 1399fba2afaaSDavide Libenzi If unsure, say Y. 1400fba2afaaSDavide Libenzi 1401b215e283SDavide Libenziconfig TIMERFD 14026a108a14SDavid Rientjes bool "Enable timerfd() system call" if EXPERT 1403b215e283SDavide Libenzi default y 1404b215e283SDavide Libenzi help 1405b215e283SDavide Libenzi Enable the timerfd() system call that allows to receive timer 1406b215e283SDavide Libenzi events on a file descriptor. 1407b215e283SDavide Libenzi 1408b215e283SDavide Libenzi If unsure, say Y. 1409b215e283SDavide Libenzi 1410e1ad7468SDavide Libenziconfig EVENTFD 14116a108a14SDavid Rientjes bool "Enable eventfd() system call" if EXPERT 1412e1ad7468SDavide Libenzi default y 1413e1ad7468SDavide Libenzi help 1414e1ad7468SDavide Libenzi Enable the eventfd() system call that allows to receive both 1415e1ad7468SDavide Libenzi kernel notification (ie. KAIO) or userspace notifications. 1416e1ad7468SDavide Libenzi 1417e1ad7468SDavide Libenzi If unsure, say Y. 1418e1ad7468SDavide Libenzi 14191da177e4SLinus Torvaldsconfig SHMEM 14206a108a14SDavid Rientjes bool "Use full shmem filesystem" if EXPERT 14211da177e4SLinus Torvalds default y 14221da177e4SLinus Torvalds depends on MMU 14231da177e4SLinus Torvalds help 14241da177e4SLinus Torvalds The shmem is an internal filesystem used to manage shared memory. 14251da177e4SLinus Torvalds It is backed by swap and manages resource limits. It is also exported 14261da177e4SLinus Torvalds to userspace as tmpfs if TMPFS is enabled. Disabling this 14271da177e4SLinus Torvalds option replaces shmem and tmpfs with the much simpler ramfs code, 14281da177e4SLinus Torvalds which may be appropriate on small systems without swap. 14291da177e4SLinus Torvalds 1430ebf3f09cSThomas Petazzoniconfig AIO 14316a108a14SDavid Rientjes bool "Enable AIO support" if EXPERT 1432ebf3f09cSThomas Petazzoni default y 1433ebf3f09cSThomas Petazzoni help 1434ebf3f09cSThomas Petazzoni This option enables POSIX asynchronous I/O which may by used 1435ebf3f09cSThomas Petazzoni by some high performance threaded applications. Disabling 1436ebf3f09cSThomas Petazzoni this option saves about 7k. 1437ebf3f09cSThomas Petazzoni 14382b188cc1SJens Axboeconfig IO_URING 14392b188cc1SJens Axboe bool "Enable IO uring support" if EXPERT 14402b188cc1SJens Axboe select ANON_INODES 14412b188cc1SJens Axboe default y 14422b188cc1SJens Axboe help 14432b188cc1SJens Axboe This option enables support for the io_uring interface, enabling 14442b188cc1SJens Axboe applications to submit and complete IO through submission and 14452b188cc1SJens Axboe completion rings that are shared between the kernel and application. 14462b188cc1SJens Axboe 1447d3ac21caSJosh Triplettconfig ADVISE_SYSCALLS 1448d3ac21caSJosh Triplett bool "Enable madvise/fadvise syscalls" if EXPERT 1449d3ac21caSJosh Triplett default y 1450d3ac21caSJosh Triplett help 1451d3ac21caSJosh Triplett This option enables the madvise and fadvise syscalls, used by 1452d3ac21caSJosh Triplett applications to advise the kernel about their future memory or file 1453d3ac21caSJosh Triplett usage, improving performance. If building an embedded system where no 1454d3ac21caSJosh Triplett applications use these syscalls, you can disable this option to save 1455d3ac21caSJosh Triplett space. 1456d3ac21caSJosh Triplett 14575b25b13aSMathieu Desnoyersconfig MEMBARRIER 14585b25b13aSMathieu Desnoyers bool "Enable membarrier() system call" if EXPERT 14595b25b13aSMathieu Desnoyers default y 14605b25b13aSMathieu Desnoyers help 14615b25b13aSMathieu Desnoyers Enable the membarrier() system call that allows issuing memory 14625b25b13aSMathieu Desnoyers barriers across all running threads, which can be used to distribute 14635b25b13aSMathieu Desnoyers the cost of user-space memory barriers asymmetrically by transforming 14645b25b13aSMathieu Desnoyers pairs of memory barriers into pairs consisting of membarrier() and a 14655b25b13aSMathieu Desnoyers compiler barrier. 14665b25b13aSMathieu Desnoyers 14675b25b13aSMathieu Desnoyers If unsure, say Y. 14685b25b13aSMathieu Desnoyers 1469d1b069f5SRandy Dunlapconfig KALLSYMS 1470d1b069f5SRandy Dunlap bool "Load all symbols for debugging/ksymoops" if EXPERT 1471d1b069f5SRandy Dunlap default y 1472d1b069f5SRandy Dunlap help 1473d1b069f5SRandy Dunlap Say Y here to let the kernel print out symbolic crash information and 1474d1b069f5SRandy Dunlap symbolic stack backtraces. This increases the size of the kernel 1475d1b069f5SRandy Dunlap somewhat, as all symbols have to be loaded into the kernel image. 1476d1b069f5SRandy Dunlap 1477d1b069f5SRandy Dunlapconfig KALLSYMS_ALL 1478d1b069f5SRandy Dunlap bool "Include all symbols in kallsyms" 1479d1b069f5SRandy Dunlap depends on DEBUG_KERNEL && KALLSYMS 1480d1b069f5SRandy Dunlap help 1481d1b069f5SRandy Dunlap Normally kallsyms only contains the symbols of functions for nicer 1482d1b069f5SRandy Dunlap OOPS messages and backtraces (i.e., symbols from the text and inittext 1483d1b069f5SRandy Dunlap sections). This is sufficient for most cases. And only in very rare 1484d1b069f5SRandy Dunlap cases (e.g., when a debugger is used) all symbols are required (e.g., 1485d1b069f5SRandy Dunlap names of variables from the data sections, etc). 1486d1b069f5SRandy Dunlap 1487d1b069f5SRandy Dunlap This option makes sure that all symbols are loaded into the kernel 1488d1b069f5SRandy Dunlap image (i.e., symbols from all sections) in cost of increased kernel 1489d1b069f5SRandy Dunlap size (depending on the kernel configuration, it may be 300KiB or 1490d1b069f5SRandy Dunlap something like this). 1491d1b069f5SRandy Dunlap 1492d1b069f5SRandy Dunlap Say N unless you really need all symbols. 1493d1b069f5SRandy Dunlap 1494d1b069f5SRandy Dunlapconfig KALLSYMS_ABSOLUTE_PERCPU 1495d1b069f5SRandy Dunlap bool 1496d1b069f5SRandy Dunlap depends on KALLSYMS 1497d1b069f5SRandy Dunlap default X86_64 && SMP 1498d1b069f5SRandy Dunlap 1499d1b069f5SRandy Dunlapconfig KALLSYMS_BASE_RELATIVE 1500d1b069f5SRandy Dunlap bool 1501d1b069f5SRandy Dunlap depends on KALLSYMS 1502a687a533SArnd Bergmann default !IA64 1503d1b069f5SRandy Dunlap help 1504d1b069f5SRandy Dunlap Instead of emitting them as absolute values in the native word size, 1505d1b069f5SRandy Dunlap emit the symbol references in the kallsyms table as 32-bit entries, 1506d1b069f5SRandy Dunlap each containing a relative value in the range [base, base + U32_MAX] 1507d1b069f5SRandy Dunlap or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either 1508d1b069f5SRandy Dunlap an absolute value in the range [0, S32_MAX] or a relative value in the 1509d1b069f5SRandy Dunlap range [base, base + S32_MAX], where base is the lowest relative symbol 1510d1b069f5SRandy Dunlap address encountered in the image. 1511d1b069f5SRandy Dunlap 1512d1b069f5SRandy Dunlap On 64-bit builds, this reduces the size of the address table by 50%, 1513d1b069f5SRandy Dunlap but more importantly, it results in entries whose values are build 1514d1b069f5SRandy Dunlap time constants, and no relocation pass is required at runtime to fix 1515d1b069f5SRandy Dunlap up the entries based on the runtime load address of the kernel. 1516d1b069f5SRandy Dunlap 1517d1b069f5SRandy Dunlap# end of the "standard kernel features (expert users)" menu 1518d1b069f5SRandy Dunlap 1519d1b069f5SRandy Dunlap# syscall, maps, verifier 1520d1b069f5SRandy Dunlapconfig BPF_SYSCALL 1521d1b069f5SRandy Dunlap bool "Enable bpf() system call" 1522d1b069f5SRandy Dunlap select BPF 1523bae77c5eSSong Liu select IRQ_WORK 1524d1b069f5SRandy Dunlap default n 1525d1b069f5SRandy Dunlap help 1526d1b069f5SRandy Dunlap Enable the bpf() system call that allows to manipulate eBPF 1527d1b069f5SRandy Dunlap programs and maps via file descriptors. 1528d1b069f5SRandy Dunlap 1529290af866SAlexei Starovoitovconfig BPF_JIT_ALWAYS_ON 1530290af866SAlexei Starovoitov bool "Permanently enable BPF JIT and remove BPF interpreter" 1531290af866SAlexei Starovoitov depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT 1532290af866SAlexei Starovoitov help 1533290af866SAlexei Starovoitov Enables BPF JIT and removes BPF interpreter to avoid 1534290af866SAlexei Starovoitov speculative execution of BPF instructions by the interpreter 1535290af866SAlexei Starovoitov 1536d1b069f5SRandy Dunlapconfig USERFAULTFD 1537d1b069f5SRandy Dunlap bool "Enable userfaultfd() system call" 1538d1b069f5SRandy Dunlap depends on MMU 1539d1b069f5SRandy Dunlap help 1540d1b069f5SRandy Dunlap Enable the userfaultfd() system call that allows to intercept and 1541d1b069f5SRandy Dunlap handle page faults in userland. 1542d1b069f5SRandy Dunlap 15433ccfebedSMathieu Desnoyersconfig ARCH_HAS_MEMBARRIER_CALLBACKS 15443ccfebedSMathieu Desnoyers bool 15453ccfebedSMathieu Desnoyers 154670216e18SMathieu Desnoyersconfig ARCH_HAS_MEMBARRIER_SYNC_CORE 154770216e18SMathieu Desnoyers bool 154870216e18SMathieu Desnoyers 1549d7822b1eSMathieu Desnoyersconfig RSEQ 1550d7822b1eSMathieu Desnoyers bool "Enable rseq() system call" if EXPERT 1551d7822b1eSMathieu Desnoyers default y 1552d7822b1eSMathieu Desnoyers depends on HAVE_RSEQ 1553d7822b1eSMathieu Desnoyers select MEMBARRIER 1554d7822b1eSMathieu Desnoyers help 1555d7822b1eSMathieu Desnoyers Enable the restartable sequences system call. It provides a 1556d7822b1eSMathieu Desnoyers user-space cache for the current CPU number value, which 1557d7822b1eSMathieu Desnoyers speeds up getting the current CPU number from user-space, 1558d7822b1eSMathieu Desnoyers as well as an ABI to speed up user-space operations on 1559d7822b1eSMathieu Desnoyers per-CPU data. 1560d7822b1eSMathieu Desnoyers 1561d7822b1eSMathieu Desnoyers If unsure, say Y. 1562d7822b1eSMathieu Desnoyers 1563d7822b1eSMathieu Desnoyersconfig DEBUG_RSEQ 1564d7822b1eSMathieu Desnoyers default n 1565d7822b1eSMathieu Desnoyers bool "Enabled debugging of rseq() system call" if EXPERT 1566d7822b1eSMathieu Desnoyers depends on RSEQ && DEBUG_KERNEL 1567d7822b1eSMathieu Desnoyers help 1568d7822b1eSMathieu Desnoyers Enable extra debugging checks for the rseq system call. 1569d7822b1eSMathieu Desnoyers 1570d7822b1eSMathieu Desnoyers If unsure, say N. 1571d7822b1eSMathieu Desnoyers 15726befe5f6SRandy Dunlapconfig EMBEDDED 15736befe5f6SRandy Dunlap bool "Embedded system" 15745d2acfc7SJosh Triplett option allnoconfig_y 15756befe5f6SRandy Dunlap select EXPERT 15766befe5f6SRandy Dunlap help 15776befe5f6SRandy Dunlap This option should be enabled if compiling the kernel for 15786befe5f6SRandy Dunlap an embedded system so certain expert options are available 15796befe5f6SRandy Dunlap for configuration. 15806befe5f6SRandy Dunlap 1581cdd6c482SIngo Molnarconfig HAVE_PERF_EVENTS 15820793a61dSThomas Gleixner bool 1583018df72dSMike Frysinger help 1584018df72dSMike Frysinger See tools/perf/design.txt for details. 15850793a61dSThomas Gleixner 1586906010b2SPeter Zijlstraconfig PERF_USE_VMALLOC 1587906010b2SPeter Zijlstra bool 1588906010b2SPeter Zijlstra help 1589906010b2SPeter Zijlstra See tools/perf/design.txt for details 1590906010b2SPeter Zijlstra 1591ad90a3deSWilliam Breathitt Grayconfig PC104 1592424529fbSWilliam Breathitt Gray bool "PC/104 support" if EXPERT 1593ad90a3deSWilliam Breathitt Gray help 1594ad90a3deSWilliam Breathitt Gray Expose PC/104 form factor device drivers and options available for 1595ad90a3deSWilliam Breathitt Gray selection and configuration. Enable this option if your target 1596ad90a3deSWilliam Breathitt Gray machine has a PC/104 bus. 1597ad90a3deSWilliam Breathitt Gray 159857c0c15bSIngo Molnarmenu "Kernel Performance Events And Counters" 15990793a61dSThomas Gleixner 1600cdd6c482SIngo Molnarconfig PERF_EVENTS 160157c0c15bSIngo Molnar bool "Kernel performance events and counters" 1602392d65a9SRobert Richter default y if PROFILING 1603cdd6c482SIngo Molnar depends on HAVE_PERF_EVENTS 1604e360adbeSPeter Zijlstra select IRQ_WORK 160583fe27eaSPranith Kumar select SRCU 16060793a61dSThomas Gleixner help 160757c0c15bSIngo Molnar Enable kernel support for various performance events provided 160857c0c15bSIngo Molnar by software and hardware. 16090793a61dSThomas Gleixner 1610dd77038dSThadeu Lima de Souza Cascardo Software events are supported either built-in or via the 161157c0c15bSIngo Molnar use of generic tracepoints. 161257c0c15bSIngo Molnar 161357c0c15bSIngo Molnar Most modern CPUs support performance events via performance 161457c0c15bSIngo Molnar counter registers. These registers count the number of certain 16150793a61dSThomas Gleixner types of hw events: such as instructions executed, cachemisses 16160793a61dSThomas Gleixner suffered, or branches mis-predicted - without slowing down the 16170793a61dSThomas Gleixner kernel or applications. These registers can also trigger interrupts 16180793a61dSThomas Gleixner when a threshold number of events have passed - and can thus be 16190793a61dSThomas Gleixner used to profile the code that runs on that CPU. 16200793a61dSThomas Gleixner 162157c0c15bSIngo Molnar The Linux Performance Event subsystem provides an abstraction of 1622dd77038dSThadeu Lima de Souza Cascardo these software and hardware event capabilities, available via a 162357c0c15bSIngo Molnar system call and used by the "perf" utility in tools/perf/. It 16240793a61dSThomas Gleixner provides per task and per CPU counters, and it provides event 16250793a61dSThomas Gleixner capabilities on top of those. 16260793a61dSThomas Gleixner 16270793a61dSThomas Gleixner Say Y if unsure. 16280793a61dSThomas Gleixner 1629906010b2SPeter Zijlstraconfig DEBUG_PERF_USE_VMALLOC 1630906010b2SPeter Zijlstra default n 1631906010b2SPeter Zijlstra bool "Debug: use vmalloc to back perf mmap() buffers" 1632cb307113SMichael Ellerman depends on PERF_EVENTS && DEBUG_KERNEL && !PPC 1633906010b2SPeter Zijlstra select PERF_USE_VMALLOC 1634906010b2SPeter Zijlstra help 1635906010b2SPeter Zijlstra Use vmalloc memory to back perf mmap() buffers. 1636906010b2SPeter Zijlstra 1637906010b2SPeter Zijlstra Mostly useful for debugging the vmalloc code on platforms 1638906010b2SPeter Zijlstra that don't require it. 1639906010b2SPeter Zijlstra 1640906010b2SPeter Zijlstra Say N if unsure. 1641906010b2SPeter Zijlstra 16420793a61dSThomas Gleixnerendmenu 16430793a61dSThomas Gleixner 1644f8891e5eSChristoph Lameterconfig VM_EVENT_COUNTERS 1645f8891e5eSChristoph Lameter default y 16466a108a14SDavid Rientjes bool "Enable VM event counters for /proc/vmstat" if EXPERT 1647f8891e5eSChristoph Lameter help 16482aea4fb6SPaul Jackson VM event counters are needed for event counts to be shown. 16492aea4fb6SPaul Jackson This option allows the disabling of the VM event counters 16506a108a14SDavid Rientjes on EXPERT systems. /proc/vmstat will only show page counts 16512aea4fb6SPaul Jackson if VM event counters are disabled. 1652f8891e5eSChristoph Lameter 165341ecc55bSChristoph Lameterconfig SLUB_DEBUG 165441ecc55bSChristoph Lameter default y 16556a108a14SDavid Rientjes bool "Enable SLUB debugging support" if EXPERT 1656f6acb635SChristoph Lameter depends on SLUB && SYSFS 165741ecc55bSChristoph Lameter help 165841ecc55bSChristoph Lameter SLUB has extensive debug support features. Disabling these can 165941ecc55bSChristoph Lameter result in significant savings in code size. This also disables 166041ecc55bSChristoph Lameter SLUB sysfs support. /sys/slab will not exist and there will be 166141ecc55bSChristoph Lameter no support for cache validation etc. 166241ecc55bSChristoph Lameter 16631663f26dSTejun Heoconfig SLUB_MEMCG_SYSFS_ON 16641663f26dSTejun Heo default n 16651663f26dSTejun Heo bool "Enable memcg SLUB sysfs support by default" if EXPERT 16661663f26dSTejun Heo depends on SLUB && SYSFS && MEMCG 16671663f26dSTejun Heo help 16681663f26dSTejun Heo SLUB creates a directory under /sys/kernel/slab for each 16691663f26dSTejun Heo allocation cache to host info and debug files. If memory 16701663f26dSTejun Heo cgroup is enabled, each cache can have per memory cgroup 16711663f26dSTejun Heo caches. SLUB can create the same sysfs directories for these 16721663f26dSTejun Heo caches under /sys/kernel/slab/CACHE/cgroup but it can lead 16731663f26dSTejun Heo to a very high number of debug files being created. This is 16741663f26dSTejun Heo controlled by slub_memcg_sysfs boot parameter and this 16751663f26dSTejun Heo config option determines the parameter's default value. 16761663f26dSTejun Heo 1677b943c460SRandy Dunlapconfig COMPAT_BRK 1678b943c460SRandy Dunlap bool "Disable heap randomization" 1679b943c460SRandy Dunlap default y 1680b943c460SRandy Dunlap help 1681b943c460SRandy Dunlap Randomizing heap placement makes heap exploits harder, but it 1682b943c460SRandy Dunlap also breaks ancient binaries (including anything libc5 based). 1683b943c460SRandy Dunlap This option changes the bootup default to heap randomization 1684692105b8SMatt LaPlante disabled, and can be overridden at runtime by setting 1685b943c460SRandy Dunlap /proc/sys/kernel/randomize_va_space to 2. 1686b943c460SRandy Dunlap 1687b943c460SRandy Dunlap On non-ancient distros (post-2000 ones) N is usually a safe choice. 1688b943c460SRandy Dunlap 168981819f0fSChristoph Lameterchoice 169081819f0fSChristoph Lameter prompt "Choose SLAB allocator" 1691a0acd820SChristoph Lameter default SLUB 169281819f0fSChristoph Lameter help 169381819f0fSChristoph Lameter This option allows to select a slab allocator. 169481819f0fSChristoph Lameter 169581819f0fSChristoph Lameterconfig SLAB 169681819f0fSChristoph Lameter bool "SLAB" 169704385fc5SKees Cook select HAVE_HARDENED_USERCOPY_ALLOCATOR 169881819f0fSChristoph Lameter help 169981819f0fSChristoph Lameter The regular slab allocator that is established and known to work 170034013886SChristoph Lameter well in all environments. It organizes cache hot objects in 170102f56210SSimon Arlott per cpu and per node queues. 170281819f0fSChristoph Lameter 170381819f0fSChristoph Lameterconfig SLUB 170481819f0fSChristoph Lameter bool "SLUB (Unqueued Allocator)" 1705ed18adc1SKees Cook select HAVE_HARDENED_USERCOPY_ALLOCATOR 170681819f0fSChristoph Lameter help 170781819f0fSChristoph Lameter SLUB is a slab allocator that minimizes cache line usage 170881819f0fSChristoph Lameter instead of managing queues of cached objects (SLAB approach). 170981819f0fSChristoph Lameter Per cpu caching is realized using slabs of objects instead 171081819f0fSChristoph Lameter of queues of objects. SLUB can use memory efficiently 171102f56210SSimon Arlott and has enhanced diagnostics. SLUB is the default choice for 171202f56210SSimon Arlott a slab allocator. 171381819f0fSChristoph Lameter 171481819f0fSChristoph Lameterconfig SLOB 17156a108a14SDavid Rientjes depends on EXPERT 171681819f0fSChristoph Lameter bool "SLOB (Simple Allocator)" 171781819f0fSChristoph Lameter help 171837291458SMatt Mackall SLOB replaces the stock allocator with a drastically simpler 171937291458SMatt Mackall allocator. SLOB is generally more space efficient but 172037291458SMatt Mackall does not perform as well on large systems. 172181819f0fSChristoph Lameter 172281819f0fSChristoph Lameterendchoice 172381819f0fSChristoph Lameter 17247660a6fdSKees Cookconfig SLAB_MERGE_DEFAULT 17257660a6fdSKees Cook bool "Allow slab caches to be merged" 17267660a6fdSKees Cook default y 17277660a6fdSKees Cook help 17287660a6fdSKees Cook For reduced kernel memory fragmentation, slab caches can be 17297660a6fdSKees Cook merged when they share the same size and other characteristics. 17307660a6fdSKees Cook This carries a risk of kernel heap overflows being able to 17317660a6fdSKees Cook overwrite objects from merged caches (and more easily control 17327660a6fdSKees Cook cache layout), which makes such heap attacks easier to exploit 17337660a6fdSKees Cook by attackers. By keeping caches unmerged, these kinds of exploits 17347660a6fdSKees Cook can usually only damage objects in the same cache. To disable 17357660a6fdSKees Cook merging at runtime, "slab_nomerge" can be passed on the kernel 17367660a6fdSKees Cook command line. 17377660a6fdSKees Cook 1738c7ce4f60SThomas Garnierconfig SLAB_FREELIST_RANDOM 1739c7ce4f60SThomas Garnier default n 1740210e7a43SThomas Garnier depends on SLAB || SLUB 1741c7ce4f60SThomas Garnier bool "SLAB freelist randomization" 1742c7ce4f60SThomas Garnier help 1743210e7a43SThomas Garnier Randomizes the freelist order used on creating new pages. This 1744c7ce4f60SThomas Garnier security feature reduces the predictability of the kernel slab 1745c7ce4f60SThomas Garnier allocator against heap overflows. 1746c7ce4f60SThomas Garnier 17472482ddecSKees Cookconfig SLAB_FREELIST_HARDENED 17482482ddecSKees Cook bool "Harden slab freelist metadata" 17492482ddecSKees Cook depends on SLUB 17502482ddecSKees Cook help 17512482ddecSKees Cook Many kernel heap attacks try to target slab cache metadata and 17522482ddecSKees Cook other infrastructure. This options makes minor performance 17532482ddecSKees Cook sacrifies to harden the kernel slab allocator against common 17542482ddecSKees Cook freelist exploit methods. 17552482ddecSKees Cook 1756e900a918SDan Williamsconfig SHUFFLE_PAGE_ALLOCATOR 1757e900a918SDan Williams bool "Page allocator randomization" 1758e900a918SDan Williams default SLAB_FREELIST_RANDOM && ACPI_NUMA 1759e900a918SDan Williams help 1760e900a918SDan Williams Randomization of the page allocator improves the average 1761e900a918SDan Williams utilization of a direct-mapped memory-side-cache. See section 1762e900a918SDan Williams 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI 1763e900a918SDan Williams 6.2a specification for an example of how a platform advertises 1764e900a918SDan Williams the presence of a memory-side-cache. There are also incidental 1765e900a918SDan Williams security benefits as it reduces the predictability of page 1766e900a918SDan Williams allocations to compliment SLAB_FREELIST_RANDOM, but the 1767e900a918SDan Williams default granularity of shuffling on the "MAX_ORDER - 1" i.e, 1768e900a918SDan Williams 10th order of pages is selected based on cache utilization 1769e900a918SDan Williams benefits on x86. 1770e900a918SDan Williams 1771e900a918SDan Williams While the randomization improves cache utilization it may 1772e900a918SDan Williams negatively impact workloads on platforms without a cache. For 1773e900a918SDan Williams this reason, by default, the randomization is enabled only 1774e900a918SDan Williams after runtime detection of a direct-mapped memory-side-cache. 1775e900a918SDan Williams Otherwise, the randomization may be force enabled with the 1776e900a918SDan Williams 'page_alloc.shuffle' kernel command line parameter. 1777e900a918SDan Williams 1778e900a918SDan Williams Say Y if unsure. 1779e900a918SDan Williams 1780345c905dSJoonsoo Kimconfig SLUB_CPU_PARTIAL 1781345c905dSJoonsoo Kim default y 1782b39ffbf8SUwe Kleine-König depends on SLUB && SMP 1783345c905dSJoonsoo Kim bool "SLUB per cpu partial cache" 1784345c905dSJoonsoo Kim help 1785345c905dSJoonsoo Kim Per cpu partial caches accellerate objects allocation and freeing 1786345c905dSJoonsoo Kim that is local to a processor at the price of more indeterminism 1787345c905dSJoonsoo Kim in the latency of the free. On overflow these caches will be cleared 1788345c905dSJoonsoo Kim which requires the taking of locks that may cause latency spikes. 1789345c905dSJoonsoo Kim Typically one would choose no for a realtime system. 1790345c905dSJoonsoo Kim 1791ea637639SJie Zhangconfig MMAP_ALLOW_UNINITIALIZED 1792ea637639SJie Zhang bool "Allow mmapped anonymous memory to be uninitialized" 17936a108a14SDavid Rientjes depends on EXPERT && !MMU 1794ea637639SJie Zhang default n 1795ea637639SJie Zhang help 1796ea637639SJie Zhang Normally, and according to the Linux spec, anonymous memory obtained 17973903bf94SRandy Dunlap from mmap() has its contents cleared before it is passed to 1798ea637639SJie Zhang userspace. Enabling this config option allows you to request that 1799ea637639SJie Zhang mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus 1800ea637639SJie Zhang providing a huge performance boost. If this option is not enabled, 1801ea637639SJie Zhang then the flag will be ignored. 1802ea637639SJie Zhang 1803ea637639SJie Zhang This is taken advantage of by uClibc's malloc(), and also by 1804ea637639SJie Zhang ELF-FDPIC binfmt's brk and stack allocator. 1805ea637639SJie Zhang 1806ea637639SJie Zhang Because of the obvious security issues, this option should only be 1807ea637639SJie Zhang enabled on embedded devices where you control what is run in 1808ea637639SJie Zhang userspace. Since that isn't generally a problem on no-MMU systems, 1809ea637639SJie Zhang it is normally safe to say Y here. 1810ea637639SJie Zhang 1811ea637639SJie Zhang See Documentation/nommu-mmap.txt for more information. 1812ea637639SJie Zhang 1813091f6e26SDavid Howellsconfig SYSTEM_DATA_VERIFICATION 1814091f6e26SDavid Howells def_bool n 1815091f6e26SDavid Howells select SYSTEM_TRUSTED_KEYRING 1816091f6e26SDavid Howells select KEYS 1817091f6e26SDavid Howells select CRYPTO 1818d43de6c7SDavid Howells select CRYPTO_RSA 1819091f6e26SDavid Howells select ASYMMETRIC_KEY_TYPE 1820091f6e26SDavid Howells select ASYMMETRIC_PUBLIC_KEY_SUBTYPE 1821091f6e26SDavid Howells select ASN1 1822091f6e26SDavid Howells select OID_REGISTRY 1823091f6e26SDavid Howells select X509_CERTIFICATE_PARSER 1824091f6e26SDavid Howells select PKCS7_MESSAGE_PARSER 182582c04ff8SPeter Foley help 1826091f6e26SDavid Howells Provide PKCS#7 message verification using the contents of the system 1827091f6e26SDavid Howells trusted keyring to provide public keys. This then can be used for 1828091f6e26SDavid Howells module verification, kexec image verification and firmware blob 1829091f6e26SDavid Howells verification. 183082c04ff8SPeter Foley 1831125e5645SMathieu Desnoyersconfig PROFILING 1832b309a294SRobert Richter bool "Profiling support" 1833125e5645SMathieu Desnoyers help 1834125e5645SMathieu Desnoyers Say Y here to enable the extended profiling support mechanisms used 1835125e5645SMathieu Desnoyers by profilers such as OProfile. 1836125e5645SMathieu Desnoyers 18375f87f112SIngo Molnar# 18385f87f112SIngo Molnar# Place an empty function call at each tracepoint site. Can be 18395f87f112SIngo Molnar# dynamically changed for a probe function. 18405f87f112SIngo Molnar# 184197e1c18eSMathieu Desnoyersconfig TRACEPOINTS 18425f87f112SIngo Molnar bool 184397e1c18eSMathieu Desnoyers 18441da177e4SLinus Torvaldsendmenu # General setup 18451da177e4SLinus Torvalds 18461572497cSChristoph Hellwigsource "arch/Kconfig" 18471572497cSChristoph Hellwig 1848ae81f9e3SChuck Ebbertconfig RT_MUTEXES 18496341e62bSChristoph Jaeger bool 1850ae81f9e3SChuck Ebbert 18511da177e4SLinus Torvaldsconfig BASE_SMALL 18521da177e4SLinus Torvalds int 18531da177e4SLinus Torvalds default 0 if BASE_FULL 18541da177e4SLinus Torvalds default 1 if !BASE_FULL 18551da177e4SLinus Torvalds 185666da5733SJan Engelhardtmenuconfig MODULES 18571da177e4SLinus Torvalds bool "Enable loadable module support" 185811097a03SYann E. MORIN option modules 18591da177e4SLinus Torvalds help 18601da177e4SLinus Torvalds Kernel modules are small pieces of compiled code which can 18611da177e4SLinus Torvalds be inserted in the running kernel, rather than being 18621da177e4SLinus Torvalds permanently built into the kernel. You use the "modprobe" 18631da177e4SLinus Torvalds tool to add (and sometimes remove) them. If you say Y here, 18641da177e4SLinus Torvalds many parts of the kernel can be built as modules (by 18651da177e4SLinus Torvalds answering M instead of Y where indicated): this is most 18661da177e4SLinus Torvalds useful for infrequently used options which are not required 18671da177e4SLinus Torvalds for booting. For more information, see the man pages for 18681da177e4SLinus Torvalds modprobe, lsmod, modinfo, insmod and rmmod. 18691da177e4SLinus Torvalds 18701da177e4SLinus Torvalds If you say Y here, you will need to run "make 18711da177e4SLinus Torvalds modules_install" to put the modules under /lib/modules/ 18721da177e4SLinus Torvalds where modprobe can find them (you may need to be root to do 18731da177e4SLinus Torvalds this). 18741da177e4SLinus Torvalds 18751da177e4SLinus Torvalds If unsure, say Y. 18761da177e4SLinus Torvalds 18770b0de144SRobert P. J. Dayif MODULES 18780b0de144SRobert P. J. Day 1879826e4506SLinus Torvaldsconfig MODULE_FORCE_LOAD 1880826e4506SLinus Torvalds bool "Forced module loading" 1881826e4506SLinus Torvalds default n 1882826e4506SLinus Torvalds help 188391e37a79SRusty Russell Allow loading of modules without version information (ie. modprobe 188491e37a79SRusty Russell --force). Forced module loading sets the 'F' (forced) taint flag and 188591e37a79SRusty Russell is usually a really bad idea. 1886826e4506SLinus Torvalds 18871da177e4SLinus Torvaldsconfig MODULE_UNLOAD 18881da177e4SLinus Torvalds bool "Module unloading" 18891da177e4SLinus Torvalds help 18901da177e4SLinus Torvalds Without this option you will not be able to unload any 18911da177e4SLinus Torvalds modules (note that some modules may not be unloadable 1892f7f5b675SDenys Vlasenko anyway), which makes your kernel smaller, faster 1893f7f5b675SDenys Vlasenko and simpler. If unsure, say Y. 18941da177e4SLinus Torvalds 18951da177e4SLinus Torvaldsconfig MODULE_FORCE_UNLOAD 18961da177e4SLinus Torvalds bool "Forced module unloading" 189719c92399SKees Cook depends on MODULE_UNLOAD 18981da177e4SLinus Torvalds help 18991da177e4SLinus Torvalds This option allows you to force a module to unload, even if the 19001da177e4SLinus Torvalds kernel believes it is unsafe: the kernel will remove the module 19011da177e4SLinus Torvalds without waiting for anyone to stop using it (using the -f option to 19021da177e4SLinus Torvalds rmmod). This is mainly for kernel developers and desperate users. 19031da177e4SLinus Torvalds If unsure, say N. 19041da177e4SLinus Torvalds 19051da177e4SLinus Torvaldsconfig MODVERSIONS 19060d541643SSam Ravnborg bool "Module versioning support" 19071da177e4SLinus Torvalds help 19081da177e4SLinus Torvalds Usually, you have to use modules compiled with your kernel. 19091da177e4SLinus Torvalds Saying Y here makes it sometimes possible to use modules 19101da177e4SLinus Torvalds compiled for different kernels, by adding enough information 19111da177e4SLinus Torvalds to the modules to (hopefully) spot any changes which would 19121da177e4SLinus Torvalds make them incompatible with the kernel you are running. If 19131da177e4SLinus Torvalds unsure, say N. 19141da177e4SLinus Torvalds 191556067812SArd Biesheuvelconfig MODULE_REL_CRCS 191656067812SArd Biesheuvel bool 191756067812SArd Biesheuvel depends on MODVERSIONS 191856067812SArd Biesheuvel 19191da177e4SLinus Torvaldsconfig MODULE_SRCVERSION_ALL 19201da177e4SLinus Torvalds bool "Source checksum for all modules" 19211da177e4SLinus Torvalds help 19221da177e4SLinus Torvalds Modules which contain a MODULE_VERSION get an extra "srcversion" 19231da177e4SLinus Torvalds field inserted into their modinfo section, which contains a 19241da177e4SLinus Torvalds sum of the source files which made it. This helps maintainers 19251da177e4SLinus Torvalds see exactly which source was used to build a module (since 19261da177e4SLinus Torvalds others sometimes change the module source without updating 19271da177e4SLinus Torvalds the version). With this option, such a "srcversion" field 19281da177e4SLinus Torvalds will be created for all modules. If unsure, say N. 19291da177e4SLinus Torvalds 1930106a4ee2SRusty Russellconfig MODULE_SIG 1931106a4ee2SRusty Russell bool "Module signature verification" 1932106a4ee2SRusty Russell depends on MODULES 1933091f6e26SDavid Howells select SYSTEM_DATA_VERIFICATION 1934106a4ee2SRusty Russell help 1935106a4ee2SRusty Russell Check modules for valid signatures upon load: the signature 1936106a4ee2SRusty Russell is simply appended to the module. For more information see 1937cbdc8217SNathan Chancellor <file:Documentation/admin-guide/module-signing.rst>. 1938106a4ee2SRusty Russell 1939228c37ffSDavid Howells Note that this option adds the OpenSSL development packages as a 1940228c37ffSDavid Howells kernel build dependency so that the signing tool can use its crypto 1941228c37ffSDavid Howells library. 1942228c37ffSDavid Howells 1943ea0b6dcfSDavid Howells !!!WARNING!!! If you enable this option, you MUST make sure that the 1944ea0b6dcfSDavid Howells module DOES NOT get stripped after being signed. This includes the 1945ea0b6dcfSDavid Howells debuginfo strip done by some packagers (such as rpmbuild) and 1946ea0b6dcfSDavid Howells inclusion into an initramfs that wants the module size reduced. 1947ea0b6dcfSDavid Howells 1948106a4ee2SRusty Russellconfig MODULE_SIG_FORCE 1949106a4ee2SRusty Russell bool "Require modules to be validly signed" 1950106a4ee2SRusty Russell depends on MODULE_SIG 1951106a4ee2SRusty Russell help 1952106a4ee2SRusty Russell Reject unsigned modules or signed modules for which we don't have a 1953106a4ee2SRusty Russell key. Without this, such modules will simply taint the kernel. 1954ea0b6dcfSDavid Howells 1955d9d8d7edSMichal Marekconfig MODULE_SIG_ALL 1956d9d8d7edSMichal Marek bool "Automatically sign all modules" 1957d9d8d7edSMichal Marek default y 1958d9d8d7edSMichal Marek depends on MODULE_SIG 1959d9d8d7edSMichal Marek help 1960d9d8d7edSMichal Marek Sign all modules during make modules_install. Without this option, 1961d9d8d7edSMichal Marek modules must be signed manually, using the scripts/sign-file tool. 1962d9d8d7edSMichal Marek 1963d9d8d7edSMichal Marekcomment "Do not forget to sign required modules with scripts/sign-file" 1964d9d8d7edSMichal Marek depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL 1965d9d8d7edSMichal Marek 1966ea0b6dcfSDavid Howellschoice 1967ea0b6dcfSDavid Howells prompt "Which hash algorithm should modules be signed with?" 1968ea0b6dcfSDavid Howells depends on MODULE_SIG 1969ea0b6dcfSDavid Howells help 1970ea0b6dcfSDavid Howells This determines which sort of hashing algorithm will be used during 1971ea0b6dcfSDavid Howells signature generation. This algorithm _must_ be built into the kernel 1972ea0b6dcfSDavid Howells directly so that signature verification can take place. It is not 1973ea0b6dcfSDavid Howells possible to load a signed module containing the algorithm to check 1974ea0b6dcfSDavid Howells the signature on that module. 1975ea0b6dcfSDavid Howells 1976ea0b6dcfSDavid Howellsconfig MODULE_SIG_SHA1 1977ea0b6dcfSDavid Howells bool "Sign modules with SHA-1" 1978ea0b6dcfSDavid Howells select CRYPTO_SHA1 1979ea0b6dcfSDavid Howells 1980ea0b6dcfSDavid Howellsconfig MODULE_SIG_SHA224 1981ea0b6dcfSDavid Howells bool "Sign modules with SHA-224" 1982ea0b6dcfSDavid Howells select CRYPTO_SHA256 1983ea0b6dcfSDavid Howells 1984ea0b6dcfSDavid Howellsconfig MODULE_SIG_SHA256 1985ea0b6dcfSDavid Howells bool "Sign modules with SHA-256" 1986ea0b6dcfSDavid Howells select CRYPTO_SHA256 1987ea0b6dcfSDavid Howells 1988ea0b6dcfSDavid Howellsconfig MODULE_SIG_SHA384 1989ea0b6dcfSDavid Howells bool "Sign modules with SHA-384" 1990ea0b6dcfSDavid Howells select CRYPTO_SHA512 1991ea0b6dcfSDavid Howells 1992ea0b6dcfSDavid Howellsconfig MODULE_SIG_SHA512 1993ea0b6dcfSDavid Howells bool "Sign modules with SHA-512" 1994ea0b6dcfSDavid Howells select CRYPTO_SHA512 1995ea0b6dcfSDavid Howells 1996ea0b6dcfSDavid Howellsendchoice 1997ea0b6dcfSDavid Howells 199822753674SMichal Marekconfig MODULE_SIG_HASH 199922753674SMichal Marek string 200022753674SMichal Marek depends on MODULE_SIG 200122753674SMichal Marek default "sha1" if MODULE_SIG_SHA1 200222753674SMichal Marek default "sha224" if MODULE_SIG_SHA224 200322753674SMichal Marek default "sha256" if MODULE_SIG_SHA256 200422753674SMichal Marek default "sha384" if MODULE_SIG_SHA384 200522753674SMichal Marek default "sha512" if MODULE_SIG_SHA512 200622753674SMichal Marek 2007beb50df3SBertrand Jacquinconfig MODULE_COMPRESS 2008beb50df3SBertrand Jacquin bool "Compress modules on installation" 2009beb50df3SBertrand Jacquin depends on MODULES 2010beb50df3SBertrand Jacquin help 2011beb50df3SBertrand Jacquin 2012b6c09b51SRusty Russell Compresses kernel modules when 'make modules_install' is run; gzip or 2013b6c09b51SRusty Russell xz depending on "Compression algorithm" below. 2014beb50df3SBertrand Jacquin 2015b6c09b51SRusty Russell module-init-tools MAY support gzip, and kmod MAY support gzip and xz. 2016beb50df3SBertrand Jacquin 2017b6c09b51SRusty Russell Out-of-tree kernel modules installed using Kbuild will also be 2018b6c09b51SRusty Russell compressed upon installation. 2019beb50df3SBertrand Jacquin 2020b6c09b51SRusty Russell Note: for modules inside an initrd or initramfs, it's more efficient 2021b6c09b51SRusty Russell to compress the whole initrd or initramfs instead. 2022beb50df3SBertrand Jacquin 2023b6c09b51SRusty Russell Note: This is fully compatible with signed modules. 2024b6c09b51SRusty Russell 2025b6c09b51SRusty Russell If in doubt, say N. 2026beb50df3SBertrand Jacquin 2027beb50df3SBertrand Jacquinchoice 2028beb50df3SBertrand Jacquin prompt "Compression algorithm" 2029beb50df3SBertrand Jacquin depends on MODULE_COMPRESS 2030beb50df3SBertrand Jacquin default MODULE_COMPRESS_GZIP 2031beb50df3SBertrand Jacquin help 2032beb50df3SBertrand Jacquin This determines which sort of compression will be used during 2033beb50df3SBertrand Jacquin 'make modules_install'. 2034beb50df3SBertrand Jacquin 2035beb50df3SBertrand Jacquin GZIP (default) and XZ are supported. 2036beb50df3SBertrand Jacquin 2037beb50df3SBertrand Jacquinconfig MODULE_COMPRESS_GZIP 2038beb50df3SBertrand Jacquin bool "GZIP" 2039beb50df3SBertrand Jacquin 2040beb50df3SBertrand Jacquinconfig MODULE_COMPRESS_XZ 2041beb50df3SBertrand Jacquin bool "XZ" 2042beb50df3SBertrand Jacquin 2043beb50df3SBertrand Jacquinendchoice 2044beb50df3SBertrand Jacquin 2045dbacb0efSNicolas Pitreconfig TRIM_UNUSED_KSYMS 2046dbacb0efSNicolas Pitre bool "Trim unused exported kernel symbols" 2047dbacb0efSNicolas Pitre depends on MODULES && !UNUSED_SYMBOLS 2048dbacb0efSNicolas Pitre help 2049dbacb0efSNicolas Pitre The kernel and some modules make many symbols available for 2050dbacb0efSNicolas Pitre other modules to use via EXPORT_SYMBOL() and variants. Depending 2051dbacb0efSNicolas Pitre on the set of modules being selected in your kernel configuration, 2052dbacb0efSNicolas Pitre many of those exported symbols might never be used. 2053dbacb0efSNicolas Pitre 2054dbacb0efSNicolas Pitre This option allows for unused exported symbols to be dropped from 2055dbacb0efSNicolas Pitre the build. In turn, this provides the compiler more opportunities 2056dbacb0efSNicolas Pitre (especially when using LTO) for optimizing the code and reducing 2057dbacb0efSNicolas Pitre binary size. This might have some security advantages as well. 2058dbacb0efSNicolas Pitre 2059f1cb637eSValdis Kletnieks If unsure, or if you need to build out-of-tree modules, say N. 2060dbacb0efSNicolas Pitre 20610b0de144SRobert P. J. Dayendif # MODULES 20620b0de144SRobert P. J. Day 20636c9692e2SPeter Zijlstraconfig MODULES_TREE_LOOKUP 20646c9692e2SPeter Zijlstra def_bool y 20656c9692e2SPeter Zijlstra depends on PERF_EVENTS || TRACING 20666c9692e2SPeter Zijlstra 206798a79d6aSRusty Russellconfig INIT_ALL_POSSIBLE 206898a79d6aSRusty Russell bool 206998a79d6aSRusty Russell help 20705f054e31SRusty Russell Back when each arch used to define their own cpu_online_mask and 20715f054e31SRusty Russell cpu_possible_mask, some of them chose to initialize cpu_possible_mask 207298a79d6aSRusty Russell with all 1s, and others with all 0s. When they were centralised, 207398a79d6aSRusty Russell it was better to provide this option than to break all the archs 2074692105b8SMatt LaPlante and have several arch maintainers pursuing me down dark alleys. 207598a79d6aSRusty Russell 20763a65dfe8SJens Axboesource "block/Kconfig" 2077e98c3202SAvi Kivity 2078e98c3202SAvi Kivityconfig PREEMPT_NOTIFIERS 2079e98c3202SAvi Kivity bool 2080e260be67SPaul E. McKenney 208116295becSSteffen Klassertconfig PADATA 208216295becSSteffen Klassert depends on SMP 208316295becSSteffen Klassert bool 208416295becSSteffen Klassert 20854520c6a4SDavid Howellsconfig ASN1 20864520c6a4SDavid Howells tristate 20874520c6a4SDavid Howells help 20884520c6a4SDavid Howells Build a simple ASN.1 grammar compiler that produces a bytecode output 20894520c6a4SDavid Howells that can be interpreted by the ASN.1 stream decoder and used to 20904520c6a4SDavid Howells inform it as to what tags are to be expected in a stream and what 20914520c6a4SDavid Howells functions to call on what tags. 20924520c6a4SDavid Howells 20936beb0009SThomas Gleixnersource "kernel/Kconfig.locks" 2094e61938a9SMathieu Desnoyers 2095e61938a9SMathieu Desnoyersconfig ARCH_HAS_SYNC_CORE_BEFORE_USERMODE 2096e61938a9SMathieu Desnoyers bool 20971bd21c6cSDominik Brodowski 20981bd21c6cSDominik Brodowski# It may be useful for an architecture to override the definitions of the 20997303e30eSDominik Brodowski# SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h> 21007303e30eSDominik Brodowski# and the COMPAT_ variants in <linux/compat.h>, in particular to use a 21017303e30eSDominik Brodowski# different calling convention for syscalls. They can also override the 21027303e30eSDominik Brodowski# macros for not-implemented syscalls in kernel/sys_ni.c and 21037303e30eSDominik Brodowski# kernel/time/posix-stubs.c. All these overrides need to be available in 21047303e30eSDominik Brodowski# <asm/syscall_wrapper.h>. 21051bd21c6cSDominik Brodowskiconfig ARCH_HAS_SYSCALL_WRAPPER 21061bd21c6cSDominik Brodowski def_bool n 2107