2055 lines · plain
1# SPDX-License-Identifier: GPL-2.0-only2config CC_VERSION_TEXT3 string4 default "$(CC_VERSION_TEXT)"5 help6 This is used in unclear ways:7 8 - Re-run Kconfig when the compiler is updated9 The 'default' property references the environment variable,10 CC_VERSION_TEXT so it is recorded in include/config/auto.conf.cmd.11 When the compiler is updated, Kconfig will be invoked.12 13 - Ensure full rebuild when the compiler is updated14 include/linux/compiler-version.h contains this option in the comment15 line so fixdep adds include/config/CC_VERSION_TEXT into the16 auto-generated dependency. When the compiler is updated, syncconfig17 will touch it and then every file will be rebuilt.18 19config CC_IS_GCC20 def_bool $(success,test "$(cc-name)" = GCC)21 22config GCC_VERSION23 int24 default $(cc-version) if CC_IS_GCC25 default 026 27config CC_IS_CLANG28 def_bool $(success,test "$(cc-name)" = Clang)29 30config CLANG_VERSION31 int32 default $(cc-version) if CC_IS_CLANG33 default 034 35config AS_IS_GNU36 def_bool $(success,test "$(as-name)" = GNU)37 38config AS_IS_LLVM39 def_bool $(success,test "$(as-name)" = LLVM)40 41config AS_VERSION42 int43 # Use clang version if this is the integrated assembler44 default CLANG_VERSION if AS_IS_LLVM45 default $(as-version)46 47config LD_IS_BFD48 def_bool $(success,test "$(ld-name)" = BFD)49 50config LD_VERSION51 int52 default $(ld-version) if LD_IS_BFD53 default 054 55config LD_IS_LLD56 def_bool $(success,test "$(ld-name)" = LLD)57 58config LLD_VERSION59 int60 default $(ld-version) if LD_IS_LLD61 default 062 63config RUSTC_VERSION64 int65 default $(rustc-version)66 help67 It does not depend on `RUST` since that one may need to use the version68 in a `depends on`.69 70config RUST_IS_AVAILABLE71 def_bool $(success,$(srctree)/scripts/rust_is_available.sh)72 help73 This shows whether a suitable Rust toolchain is available (found).74 75 Please see Documentation/rust/quick-start.rst for instructions on how76 to satisfy the build requirements of Rust support.77 78 In particular, the Makefile target 'rustavailable' is useful to check79 why the Rust toolchain is not being detected.80 81config RUSTC_LLVM_VERSION82 int83 default $(rustc-llvm-version)84 85config CC_CAN_LINK86 bool87 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(USERCFLAGS) $(USERLDFLAGS) $(m64-flag)) if 64BIT88 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(USERCFLAGS) $(USERLDFLAGS) $(m32-flag))89 90config CC_CAN_LINK_STATIC91 bool92 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(USERCFLAGS) $(USERLDFLAGS) $(m64-flag) -static) if 64BIT93 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(USERCFLAGS) $(USERLDFLAGS) $(m32-flag) -static)94 95# Fixed in GCC 14, 13.3, 12.4 and 11.596# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=11392197config GCC_ASM_GOTO_OUTPUT_BROKEN98 bool99 depends on CC_IS_GCC100 default y if GCC_VERSION < 110500101 default y if GCC_VERSION >= 120000 && GCC_VERSION < 120400102 default y if GCC_VERSION >= 130000 && GCC_VERSION < 130300103 104config CC_HAS_ASM_GOTO_OUTPUT105 def_bool y106 depends on !GCC_ASM_GOTO_OUTPUT_BROKEN107 depends on $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)108 109config CC_HAS_ASM_GOTO_TIED_OUTPUT110 depends on CC_HAS_ASM_GOTO_OUTPUT111 # Detect buggy gcc and clang, fixed in gcc-11 clang-14.112 def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)113 114config TOOLS_SUPPORT_RELR115 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)116 117config CC_HAS_ASM_INLINE118 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)119 120config CC_HAS_NO_PROFILE_FN_ATTR121 def_bool $(success,echo '__attribute__((no_profile_instrument_function)) int x();' | $(CC) -x c - -c -o /dev/null -Werror)122 123config PAHOLE_VERSION124 int125 default $(shell,$(srctree)/scripts/pahole-version.sh $(PAHOLE))126 127config CONSTRUCTORS128 bool129 130config IRQ_WORK131 def_bool y if SMP132 133config BUILDTIME_TABLE_SORT134 bool135 136config THREAD_INFO_IN_TASK137 bool138 help139 Select this to move thread_info off the stack into task_struct. To140 make this work, an arch will need to remove all thread_info fields141 except flags and fix any runtime bugs.142 143 One subtle change that will be needed is to use try_get_task_stack()144 and put_task_stack() in save_thread_stack_tsk() and get_wchan().145 146menu "General setup"147 148config BROKEN149 bool150 151config BROKEN_ON_SMP152 bool153 depends on BROKEN || !SMP154 default y155 156config INIT_ENV_ARG_LIMIT157 int158 default 32 if !UML159 default 128 if UML160 help161 Maximum of each of the number of arguments and environment162 variables passed to init from the kernel command line.163 164config COMPILE_TEST165 bool "Compile also drivers which will not load"166 depends on HAS_IOMEM167 help168 Some drivers can be compiled on a different platform than they are169 intended to be run on. Despite they cannot be loaded there (or even170 when they load they cannot be used due to missing HW support),171 developers still, opposing to distributors, might want to build such172 drivers to compile-test them.173 174 If you are a developer and want to build everything available, say Y175 here. If you are a user/distributor, say N here to exclude useless176 drivers to be distributed.177 178config WERROR179 bool "Compile the kernel with warnings as errors"180 default COMPILE_TEST181 help182 A kernel build should not cause any compiler warnings, and this183 enables the '-Werror' (for C) and '-Dwarnings' (for Rust) flags184 to enforce that rule by default. Certain warnings from other tools185 such as the linker may be upgraded to errors with this option as186 well.187 188 However, if you have a new (or very old) compiler or linker with odd189 and unusual warnings, or you have some architecture with problems,190 you may need to disable this config option in order to191 successfully build the kernel.192 193 If in doubt, say Y.194 195config UAPI_HEADER_TEST196 bool "Compile test UAPI headers"197 depends on HEADERS_INSTALL && CC_CAN_LINK198 help199 Compile test headers exported to user-space to ensure they are200 self-contained, i.e. compilable as standalone units.201 202 If you are a developer or tester and want to ensure the exported203 headers are self-contained, say Y here. Otherwise, choose N.204 205config LOCALVERSION206 string "Local version - append to kernel release"207 help208 Append an extra string to the end of your kernel version.209 This will show up when you type uname, for example.210 The string you set here will be appended after the contents of211 any files with a filename matching localversion* in your212 object and source tree, in that order. Your total string can213 be a maximum of 64 characters.214 215config LOCALVERSION_AUTO216 bool "Automatically append version information to the version string"217 default y218 depends on !COMPILE_TEST219 help220 This will try to automatically determine if the current tree is a221 release tree by looking for git tags that belong to the current222 top of tree revision.223 224 A string of the format -gxxxxxxxx will be added to the localversion225 if a git-based tree is found. The string generated by this will be226 appended after any matching localversion* files, and after the value227 set in CONFIG_LOCALVERSION.228 229 (The actual string used here is the first 12 characters produced230 by running the command:231 232 $ git rev-parse --verify HEAD233 234 which is done within the script "scripts/setlocalversion".)235 236config BUILD_SALT237 string "Build ID Salt"238 default ""239 help240 The build ID is used to link binaries and their debug info. Setting241 this option will use the value in the calculation of the build id.242 This is mostly useful for distributions which want to ensure the243 build is unique between builds. It's safe to leave the default.244 245config HAVE_KERNEL_GZIP246 bool247 248config HAVE_KERNEL_BZIP2249 bool250 251config HAVE_KERNEL_LZMA252 bool253 254config HAVE_KERNEL_XZ255 bool256 257config HAVE_KERNEL_LZO258 bool259 260config HAVE_KERNEL_LZ4261 bool262 263config HAVE_KERNEL_ZSTD264 bool265 266config HAVE_KERNEL_UNCOMPRESSED267 bool268 269choice270 prompt "Kernel compression mode"271 default KERNEL_GZIP272 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_ZSTD || HAVE_KERNEL_UNCOMPRESSED273 help274 The linux kernel is a kind of self-extracting executable.275 Several compression algorithms are available, which differ276 in efficiency, compression and decompression speed.277 Compression speed is only relevant when building a kernel.278 Decompression speed is relevant at each boot.279 280 If you have any problems with bzip2 or lzma compressed281 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older282 version of this functionality (bzip2 only), for 2.4, was283 supplied by Christian Ludwig)284 285 High compression options are mostly useful for users, who286 are low on disk space (embedded systems), but for whom ram287 size matters less.288 289 If in doubt, select 'gzip'290 291config KERNEL_GZIP292 bool "Gzip"293 depends on HAVE_KERNEL_GZIP294 help295 The old and tried gzip compression. It provides a good balance296 between compression ratio and decompression speed.297 298config KERNEL_BZIP2299 bool "Bzip2"300 depends on HAVE_KERNEL_BZIP2301 help302 Its compression ratio and speed is intermediate.303 Decompression speed is slowest among the choices. The kernel304 size is about 10% smaller with bzip2, in comparison to gzip.305 Bzip2 uses a large amount of memory. For modern kernels you306 will need at least 8MB RAM or more for booting.307 308config KERNEL_LZMA309 bool "LZMA"310 depends on HAVE_KERNEL_LZMA311 help312 This compression algorithm's ratio is best. Decompression speed313 is between gzip and bzip2. Compression is slowest.314 The kernel size is about 33% smaller with LZMA in comparison to gzip.315 316config KERNEL_XZ317 bool "XZ"318 depends on HAVE_KERNEL_XZ319 help320 XZ uses the LZMA2 algorithm and instruction set specific321 BCJ filters which can improve compression ratio of executable322 code. The size of the kernel is about 30% smaller with XZ in323 comparison to gzip. On architectures for which there is a BCJ324 filter (i386, x86_64, ARM, ARM64, RISC-V, big endian PowerPC,325 and SPARC), XZ will create a few percent smaller kernel than326 plain LZMA.327 328 The speed is about the same as with LZMA: The decompression329 speed of XZ is better than that of bzip2 but worse than gzip330 and LZO. Compression is slow.331 332config KERNEL_LZO333 bool "LZO"334 depends on HAVE_KERNEL_LZO335 help336 Its compression ratio is the poorest among the choices. The kernel337 size is about 10% bigger than gzip; however its speed338 (both compression and decompression) is the fastest.339 340config KERNEL_LZ4341 bool "LZ4"342 depends on HAVE_KERNEL_LZ4343 help344 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.345 A preliminary version of LZ4 de/compression tool is available at346 <https://code.google.com/p/lz4/>.347 348 Its compression ratio is worse than LZO. The size of the kernel349 is about 8% bigger than LZO. But the decompression speed is350 faster than LZO.351 352config KERNEL_ZSTD353 bool "ZSTD"354 depends on HAVE_KERNEL_ZSTD355 help356 ZSTD is a compression algorithm targeting intermediate compression357 with fast decompression speed. It will compress better than GZIP and358 decompress around the same speed as LZO, but slower than LZ4. You359 will need at least 192 KB RAM or more for booting. The zstd command360 line tool is required for compression.361 362config KERNEL_UNCOMPRESSED363 bool "None"364 depends on HAVE_KERNEL_UNCOMPRESSED365 help366 Produce uncompressed kernel image. This option is usually not what367 you want. It is useful for debugging the kernel in slow simulation368 environments, where decompressing and moving the kernel is awfully369 slow. This option allows early boot code to skip the decompressor370 and jump right at uncompressed kernel image.371 372endchoice373 374config DEFAULT_INIT375 string "Default init path"376 default ""377 help378 This option determines the default init for the system if no init=379 option is passed on the kernel command line. If the requested path is380 not present, we will still then move on to attempting further381 locations (e.g. /sbin/init, etc). If this is empty, we will just use382 the fallback list when init= is not passed.383 384config DEFAULT_HOSTNAME385 string "Default hostname"386 default "(none)"387 help388 This option determines the default system hostname before userspace389 calls sethostname(2). The kernel traditionally uses "(none)" here,390 but you may wish to use a different default here to make a minimal391 system more usable with less configuration.392 393config SYSVIPC394 bool "System V IPC"395 help396 Inter Process Communication is a suite of library functions and397 system calls which let processes (running programs) synchronize and398 exchange information. It is generally considered to be a good thing,399 and some programs won't run unless you say Y here. In particular, if400 you want to run the DOS emulator dosemu under Linux (read the401 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),402 you'll need to say Y here.403 404 You can find documentation about IPC with "info ipc" and also in405 section 6.4 of the Linux Programmer's Guide, available from406 <http://www.tldp.org/guides.html>.407 408config SYSVIPC_SYSCTL409 bool410 depends on SYSVIPC411 depends on SYSCTL412 default y413 414config SYSVIPC_COMPAT415 def_bool y416 depends on COMPAT && SYSVIPC417 418config POSIX_MQUEUE419 bool "POSIX Message Queues"420 depends on NET421 help422 POSIX variant of message queues is a part of IPC. In POSIX message423 queues every message has a priority which decides about succession424 of receiving it by a process. If you want to compile and run425 programs written e.g. for Solaris with use of its POSIX message426 queues (functions mq_*) say Y here.427 428 POSIX message queues are visible as a filesystem called 'mqueue'429 and can be mounted somewhere if you want to do filesystem430 operations on message queues.431 432 If unsure, say Y.433 434config POSIX_MQUEUE_SYSCTL435 bool436 depends on POSIX_MQUEUE437 depends on SYSCTL438 default y439 440config WATCH_QUEUE441 bool "General notification queue"442 default n443 help444 445 This is a general notification queue for the kernel to pass events to446 userspace by splicing them into pipes. It can be used in conjunction447 with watches for key/keyring change notifications and device448 notifications.449 450 See Documentation/core-api/watch_queue.rst451 452config CROSS_MEMORY_ATTACH453 bool "Enable process_vm_readv/writev syscalls"454 depends on MMU455 default y456 help457 Enabling this option adds the system calls process_vm_readv and458 process_vm_writev which allow a process with the correct privileges459 to directly read from or write to another process' address space.460 See the man page for more details.461 462config USELIB463 bool "uselib syscall (for libc5 and earlier)"464 default ALPHA || M68K || SPARC465 help466 This option enables the uselib syscall, a system call used in the467 dynamic linker from libc5 and earlier. glibc does not use this468 system call. If you intend to run programs built on libc5 or469 earlier, you may need to enable this syscall. Current systems470 running glibc can safely disable this.471 472config AUDIT473 bool "Auditing support"474 depends on NET475 help476 Enable auditing infrastructure that can be used with another477 kernel subsystem, such as SELinux (which requires this for478 logging of avc messages output). System call auditing is included479 on architectures which support it.480 481config HAVE_ARCH_AUDITSYSCALL482 bool483 484config AUDITSYSCALL485 def_bool y486 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL487 select FSNOTIFY488 489source "kernel/irq/Kconfig"490source "kernel/time/Kconfig"491source "kernel/bpf/Kconfig"492source "kernel/Kconfig.preempt"493 494menu "CPU/Task time and stats accounting"495 496config VIRT_CPU_ACCOUNTING497 bool498 499choice500 prompt "Cputime accounting"501 default TICK_CPU_ACCOUNTING502 503# Kind of a stub config for the pure tick based cputime accounting504config TICK_CPU_ACCOUNTING505 bool "Simple tick based cputime accounting"506 depends on !S390 && !NO_HZ_FULL507 help508 This is the basic tick based cputime accounting that maintains509 statistics about user, system and idle time spent on per jiffies510 granularity.511 512 If unsure, say Y.513 514config VIRT_CPU_ACCOUNTING_NATIVE515 bool "Deterministic task and CPU time accounting"516 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL517 select VIRT_CPU_ACCOUNTING518 help519 Select this option to enable more accurate task and CPU time520 accounting. This is done by reading a CPU counter on each521 kernel entry and exit and on transitions within the kernel522 between system, softirq and hardirq state, so there is a523 small performance impact. In the case of s390 or IBM POWER > 5,524 this also enables accounting of stolen time on logically-partitioned525 systems.526 527config VIRT_CPU_ACCOUNTING_GEN528 bool "Full dynticks CPU time accounting"529 depends on HAVE_CONTEXT_TRACKING_USER530 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN531 depends on GENERIC_CLOCKEVENTS532 select VIRT_CPU_ACCOUNTING533 select CONTEXT_TRACKING_USER534 help535 Select this option to enable task and CPU time accounting on full536 dynticks systems. This accounting is implemented by watching every537 kernel-user boundaries using the context tracking subsystem.538 The accounting is thus performed at the expense of some significant539 overhead.540 541 For now this is only useful if you are working on the full542 dynticks subsystem development.543 544 If unsure, say N.545 546endchoice547 548config IRQ_TIME_ACCOUNTING549 bool "Fine granularity task level IRQ time accounting"550 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE551 help552 Select this option to enable fine granularity task irq time553 accounting. This is done by reading a timestamp on each554 transitions between softirq and hardirq state, so there can be a555 small performance impact.556 557 If in doubt, say N here.558 559config HAVE_SCHED_AVG_IRQ560 def_bool y561 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING562 depends on SMP563 564config SCHED_HW_PRESSURE565 bool566 default y if ARM && ARM_CPU_TOPOLOGY567 default y if ARM64568 depends on SMP569 depends on CPU_FREQ_THERMAL570 help571 Select this option to enable HW pressure accounting in the572 scheduler. HW pressure is the value conveyed to the scheduler573 that reflects the reduction in CPU compute capacity resulted from574 HW throttling. HW throttling occurs when the performance of575 a CPU is capped due to high operating temperatures as an example.576 577 If selected, the scheduler will be able to balance tasks accordingly,578 i.e. put less load on throttled CPUs than on non/less throttled ones.579 580 This requires the architecture to implement581 arch_update_hw_pressure() and arch_scale_thermal_pressure().582 583config BSD_PROCESS_ACCT584 bool "BSD Process Accounting"585 depends on MULTIUSER586 help587 If you say Y here, a user level program will be able to instruct the588 kernel (via a special system call) to write process accounting589 information to a file: whenever a process exits, information about590 that process will be appended to the file by the kernel. The591 information includes things such as creation time, owning user,592 command name, memory usage, controlling terminal etc. (the complete593 list is in the struct acct in <file:include/linux/acct.h>). It is594 up to the user level program to do useful things with this595 information. This is generally a good idea, so say Y.596 597config BSD_PROCESS_ACCT_V3598 bool "BSD Process Accounting version 3 file format"599 depends on BSD_PROCESS_ACCT600 default n601 help602 If you say Y here, the process accounting information is written603 in a new file format that also logs the process IDs of each604 process and its parent. Note that this file format is incompatible605 with previous v0/v1/v2 file formats, so you will need updated tools606 for processing it. A preliminary version of these tools is available607 at <http://www.gnu.org/software/acct/>.608 609config TASKSTATS610 bool "Export task/process statistics through netlink"611 depends on NET612 depends on MULTIUSER613 default n614 help615 Export selected statistics for tasks/processes through the616 generic netlink interface. Unlike BSD process accounting, the617 statistics are available during the lifetime of tasks/processes as618 responses to commands. Like BSD accounting, they are sent to user619 space on task exit.620 621 Say N if unsure.622 623config TASK_DELAY_ACCT624 bool "Enable per-task delay accounting"625 depends on TASKSTATS626 select SCHED_INFO627 help628 Collect information on time spent by a task waiting for system629 resources like cpu, synchronous block I/O completion and swapping630 in pages. Such statistics can help in setting a task's priorities631 relative to other tasks for cpu, io, rss limits etc.632 633 Say N if unsure.634 635config TASK_XACCT636 bool "Enable extended accounting over taskstats"637 depends on TASKSTATS638 help639 Collect extended task accounting data and send the data640 to userland for processing over the taskstats interface.641 642 Say N if unsure.643 644config TASK_IO_ACCOUNTING645 bool "Enable per-task storage I/O accounting"646 depends on TASK_XACCT647 help648 Collect information on the number of bytes of storage I/O which this649 task has caused.650 651 Say N if unsure.652 653config PSI654 bool "Pressure stall information tracking"655 select KERNFS656 help657 Collect metrics that indicate how overcommitted the CPU, memory,658 and IO capacity are in the system.659 660 If you say Y here, the kernel will create /proc/pressure/ with the661 pressure statistics files cpu, memory, and io. These will indicate662 the share of walltime in which some or all tasks in the system are663 delayed due to contention of the respective resource.664 665 In kernels with cgroup support, cgroups (cgroup2 only) will666 have cpu.pressure, memory.pressure, and io.pressure files,667 which aggregate pressure stalls for the grouped tasks only.668 669 For more details see Documentation/accounting/psi.rst.670 671 Say N if unsure.672 673config PSI_DEFAULT_DISABLED674 bool "Require boot parameter to enable pressure stall information tracking"675 default n676 depends on PSI677 help678 If set, pressure stall information tracking will be disabled679 per default but can be enabled through passing psi=1 on the680 kernel commandline during boot.681 682 This feature adds some code to the task wakeup and sleep683 paths of the scheduler. The overhead is too low to affect684 common scheduling-intense workloads in practice (such as685 webservers, memcache), but it does show up in artificial686 scheduler stress tests, such as hackbench.687 688 If you are paranoid and not sure what the kernel will be689 used for, say Y.690 691 Say N if unsure.692 693endmenu # "CPU/Task time and stats accounting"694 695config CPU_ISOLATION696 bool "CPU isolation"697 depends on SMP || COMPILE_TEST698 default y699 help700 Make sure that CPUs running critical tasks are not disturbed by701 any source of "noise" such as unbound workqueues, timers, kthreads...702 Unbound jobs get offloaded to housekeeping CPUs. This is driven by703 the "isolcpus=" boot parameter.704 705 Say Y if unsure.706 707source "kernel/rcu/Kconfig"708 709config IKCONFIG710 tristate "Kernel .config support"711 help712 This option enables the complete Linux kernel ".config" file713 contents to be saved in the kernel. It provides documentation714 of which kernel options are used in a running kernel or in an715 on-disk kernel. This information can be extracted from the kernel716 image file with the script scripts/extract-ikconfig and used as717 input to rebuild the current kernel or to build another kernel.718 It can also be extracted from a running kernel by reading719 /proc/config.gz if enabled (below).720 721config IKCONFIG_PROC722 bool "Enable access to .config through /proc/config.gz"723 depends on IKCONFIG && PROC_FS724 help725 This option enables access to the kernel configuration file726 through /proc/config.gz.727 728config IKHEADERS729 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"730 depends on SYSFS731 help732 This option enables access to the in-kernel headers that are generated during733 the build process. These can be used to build eBPF tracing programs,734 or similar programs. If you build the headers as a module, a module called735 kheaders.ko is built which can be loaded on-demand to get access to headers.736 737config LOG_BUF_SHIFT738 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"739 range 12 25740 default 17741 depends on PRINTK742 help743 Select the minimal kernel log buffer size as a power of 2.744 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config745 parameter, see below. Any higher size also might be forced746 by "log_buf_len" boot parameter.747 748 Examples:749 17 => 128 KB750 16 => 64 KB751 15 => 32 KB752 14 => 16 KB753 13 => 8 KB754 12 => 4 KB755 756config LOG_CPU_MAX_BUF_SHIFT757 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"758 depends on SMP759 range 0 21760 default 0 if BASE_SMALL761 default 12762 depends on PRINTK763 help764 This option allows to increase the default ring buffer size765 according to the number of CPUs. The value defines the contribution766 of each CPU as a power of 2. The used space is typically only few767 lines however it might be much more when problems are reported,768 e.g. backtraces.769 770 The increased size means that a new buffer has to be allocated and771 the original static one is unused. It makes sense only on systems772 with more CPUs. Therefore this value is used only when the sum of773 contributions is greater than the half of the default kernel ring774 buffer as defined by LOG_BUF_SHIFT. The default values are set775 so that more than 16 CPUs are needed to trigger the allocation.776 777 Also this option is ignored when "log_buf_len" kernel parameter is778 used as it forces an exact (power of two) size of the ring buffer.779 780 The number of possible CPUs is used for this computation ignoring781 hotplugging making the computation optimal for the worst case782 scenario while allowing a simple algorithm to be used from bootup.783 784 Examples shift values and their meaning:785 17 => 128 KB for each CPU786 16 => 64 KB for each CPU787 15 => 32 KB for each CPU788 14 => 16 KB for each CPU789 13 => 8 KB for each CPU790 12 => 4 KB for each CPU791 792config PRINTK_INDEX793 bool "Printk indexing debugfs interface"794 depends on PRINTK && DEBUG_FS795 help796 Add support for indexing of all printk formats known at compile time797 at <debugfs>/printk/index/<module>.798 799 This can be used as part of maintaining daemons which monitor800 /dev/kmsg, as it permits auditing the printk formats present in a801 kernel, allowing detection of cases where monitored printks are802 changed or no longer present.803 804 There is no additional runtime cost to printk with this enabled.805 806#807# Architectures with an unreliable sched_clock() should select this:808#809config HAVE_UNSTABLE_SCHED_CLOCK810 bool811 812config GENERIC_SCHED_CLOCK813 bool814 815menu "Scheduler features"816 817config UCLAMP_TASK818 bool "Enable utilization clamping for RT/FAIR tasks"819 depends on CPU_FREQ_GOV_SCHEDUTIL820 help821 This feature enables the scheduler to track the clamped utilization822 of each CPU based on RUNNABLE tasks scheduled on that CPU.823 824 With this option, the user can specify the min and max CPU825 utilization allowed for RUNNABLE tasks. The max utilization defines826 the maximum frequency a task should use while the min utilization827 defines the minimum frequency it should use.828 829 Both min and max utilization clamp values are hints to the scheduler,830 aiming at improving its frequency selection policy, but they do not831 enforce or grant any specific bandwidth for tasks.832 833 If in doubt, say N.834 835config UCLAMP_BUCKETS_COUNT836 int "Number of supported utilization clamp buckets"837 range 5 20838 default 5839 depends on UCLAMP_TASK840 help841 Defines the number of clamp buckets to use. The range of each bucket842 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the843 number of clamp buckets the finer their granularity and the higher844 the precision of clamping aggregation and tracking at run-time.845 846 For example, with the minimum configuration value we will have 5847 clamp buckets tracking 20% utilization each. A 25% boosted tasks will848 be refcounted in the [20..39]% bucket and will set the bucket clamp849 effective value to 25%.850 If a second 30% boosted task should be co-scheduled on the same CPU,851 that task will be refcounted in the same bucket of the first task and852 it will boost the bucket clamp effective value to 30%.853 The clamp effective value of a bucket is reset to its nominal value854 (20% in the example above) when there are no more tasks refcounted in855 that bucket.856 857 An additional boost/capping margin can be added to some tasks. In the858 example above the 25% task will be boosted to 30% until it exits the859 CPU. If that should be considered not acceptable on certain systems,860 it's always possible to reduce the margin by increasing the number of861 clamp buckets to trade off used memory for run-time tracking862 precision.863 864 If in doubt, use the default value.865 866endmenu867 868#869# For architectures that want to enable the support for NUMA-affine scheduler870# balancing logic:871#872config ARCH_SUPPORTS_NUMA_BALANCING873 bool874 875#876# For architectures that prefer to flush all TLBs after a number of pages877# are unmapped instead of sending one IPI per page to flush. The architecture878# must provide guarantees on what happens if a clean TLB cache entry is879# written after the unmap. Details are in mm/rmap.c near the check for880# should_defer_flush. The architecture should also consider if the full flush881# and the refill costs are offset by the savings of sending fewer IPIs.882config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH883 bool884 885config CC_HAS_INT128886 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT887 888config CC_IMPLICIT_FALLTHROUGH889 string890 default "-Wimplicit-fallthrough=5" if CC_IS_GCC && $(cc-option,-Wimplicit-fallthrough=5)891 default "-Wimplicit-fallthrough" if CC_IS_CLANG && $(cc-option,-Wunreachable-code-fallthrough)892 893# Currently, disable gcc-10+ array-bounds globally.894# It's still broken in gcc-13, so no upper bound yet.895config GCC10_NO_ARRAY_BOUNDS896 def_bool y897 898config CC_NO_ARRAY_BOUNDS899 bool900 default y if CC_IS_GCC && GCC_VERSION >= 90000 && GCC10_NO_ARRAY_BOUNDS901 902# Currently, disable -Wstringop-overflow for GCC globally.903config GCC_NO_STRINGOP_OVERFLOW904 def_bool y905 906config CC_NO_STRINGOP_OVERFLOW907 bool908 default y if CC_IS_GCC && GCC_NO_STRINGOP_OVERFLOW909 910config CC_STRINGOP_OVERFLOW911 bool912 default y if CC_IS_GCC && !CC_NO_STRINGOP_OVERFLOW913 914#915# For architectures that know their GCC __int128 support is sound916#917config ARCH_SUPPORTS_INT128918 bool919 920# For architectures that (ab)use NUMA to represent different memory regions921# all cpu-local but of different latencies, such as SuperH.922#923config ARCH_WANT_NUMA_VARIABLE_LOCALITY924 bool925 926config NUMA_BALANCING927 bool "Memory placement aware NUMA scheduler"928 depends on ARCH_SUPPORTS_NUMA_BALANCING929 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY930 depends on SMP && NUMA && MIGRATION && !PREEMPT_RT931 help932 This option adds support for automatic NUMA aware memory/task placement.933 The mechanism is quite primitive and is based on migrating memory when934 it has references to the node the task is running on.935 936 This system will be inactive on UMA systems.937 938config NUMA_BALANCING_DEFAULT_ENABLED939 bool "Automatically enable NUMA aware memory/task placement"940 default y941 depends on NUMA_BALANCING942 help943 If set, automatic NUMA balancing will be enabled if running on a NUMA944 machine.945 946config SLAB_OBJ_EXT947 bool948 949menuconfig CGROUPS950 bool "Control Group support"951 select KERNFS952 help953 This option adds support for grouping sets of processes together, for954 use with process control subsystems such as Cpusets, CFS, memory955 controls or device isolation.956 See957 - Documentation/scheduler/sched-design-CFS.rst (CFS)958 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation959 and resource control)960 961 Say N if unsure.962 963if CGROUPS964 965config PAGE_COUNTER966 bool967 968config CGROUP_FAVOR_DYNMODS969 bool "Favor dynamic modification latency reduction by default"970 help971 This option enables the "favordynmods" mount option by default972 which reduces the latencies of dynamic cgroup modifications such973 as task migrations and controller on/offs at the cost of making974 hot path operations such as forks and exits more expensive.975 976 Say N if unsure.977 978config MEMCG979 bool "Memory controller"980 select PAGE_COUNTER981 select EVENTFD982 select SLAB_OBJ_EXT983 help984 Provides control over the memory footprint of tasks in a cgroup.985 986config MEMCG_V1987 bool "Legacy cgroup v1 memory controller"988 depends on MEMCG989 default n990 help991 Legacy cgroup v1 memory controller which has been deprecated by992 cgroup v2 implementation. The v1 is there for legacy applications993 which haven't migrated to the new cgroup v2 interface yet. If you994 do not have any such application then you are completely fine leaving995 this option disabled.996 997 Please note that feature set of the legacy memory controller is likely998 going to shrink due to deprecation process. New deployments with v1999 controller are highly discouraged.1000 1001 Say N if unsure.1002 1003config BLK_CGROUP1004 bool "IO controller"1005 depends on BLOCK1006 default n1007 help1008 Generic block IO controller cgroup interface. This is the common1009 cgroup interface which should be used by various IO controlling1010 policies.1011 1012 Currently, CFQ IO scheduler uses it to recognize task groups and1013 control disk bandwidth allocation (proportional time slice allocation)1014 to such task groups. It is also used by bio throttling logic in1015 block layer to implement upper limit in IO rates on a device.1016 1017 This option only enables generic Block IO controller infrastructure.1018 One needs to also enable actual IO controlling logic/policy. For1019 enabling proportional weight division of disk bandwidth in CFQ, set1020 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set1021 CONFIG_BLK_DEV_THROTTLING=y.1022 1023 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.1024 1025config CGROUP_WRITEBACK1026 bool1027 depends on MEMCG && BLK_CGROUP1028 default y1029 1030menuconfig CGROUP_SCHED1031 bool "CPU controller"1032 default n1033 help1034 This feature lets CPU scheduler recognize task groups and control CPU1035 bandwidth allocation to such task groups. It uses cgroups to group1036 tasks.1037 1038if CGROUP_SCHED1039config GROUP_SCHED_WEIGHT1040 def_bool n1041 1042config FAIR_GROUP_SCHED1043 bool "Group scheduling for SCHED_OTHER"1044 depends on CGROUP_SCHED1045 select GROUP_SCHED_WEIGHT1046 default CGROUP_SCHED1047 1048config CFS_BANDWIDTH1049 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"1050 depends on FAIR_GROUP_SCHED1051 default n1052 help1053 This option allows users to define CPU bandwidth rates (limits) for1054 tasks running within the fair group scheduler. Groups with no limit1055 set are considered to be unconstrained and will run with no1056 restriction.1057 See Documentation/scheduler/sched-bwc.rst for more information.1058 1059config RT_GROUP_SCHED1060 bool "Group scheduling for SCHED_RR/FIFO"1061 depends on CGROUP_SCHED1062 default n1063 help1064 This feature lets you explicitly allocate real CPU bandwidth1065 to task groups. If enabled, it will also make it impossible to1066 schedule realtime tasks for non-root users until you allocate1067 realtime bandwidth for them.1068 See Documentation/scheduler/sched-rt-group.rst for more information.1069 1070config EXT_GROUP_SCHED1071 bool1072 depends on SCHED_CLASS_EXT && CGROUP_SCHED1073 select GROUP_SCHED_WEIGHT1074 default y1075 1076endif #CGROUP_SCHED1077 1078config SCHED_MM_CID1079 def_bool y1080 depends on SMP && RSEQ1081 1082config UCLAMP_TASK_GROUP1083 bool "Utilization clamping per group of tasks"1084 depends on CGROUP_SCHED1085 depends on UCLAMP_TASK1086 default n1087 help1088 This feature enables the scheduler to track the clamped utilization1089 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.1090 1091 When this option is enabled, the user can specify a min and max1092 CPU bandwidth which is allowed for each single task in a group.1093 The max bandwidth allows to clamp the maximum frequency a task1094 can use, while the min bandwidth allows to define a minimum1095 frequency a task will always use.1096 1097 When task group based utilization clamping is enabled, an eventually1098 specified task-specific clamp value is constrained by the cgroup1099 specified clamp value. Both minimum and maximum task clamping cannot1100 be bigger than the corresponding clamping defined at task group level.1101 1102 If in doubt, say N.1103 1104config CGROUP_PIDS1105 bool "PIDs controller"1106 help1107 Provides enforcement of process number limits in the scope of a1108 cgroup. Any attempt to fork more processes than is allowed in the1109 cgroup will fail. PIDs are fundamentally a global resource because it1110 is fairly trivial to reach PID exhaustion before you reach even a1111 conservative kmemcg limit. As a result, it is possible to grind a1112 system to halt without being limited by other cgroup policies. The1113 PIDs controller is designed to stop this from happening.1114 1115 It should be noted that organisational operations (such as attaching1116 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,1117 since the PIDs limit only affects a process's ability to fork, not to1118 attach to a cgroup.1119 1120config CGROUP_RDMA1121 bool "RDMA controller"1122 help1123 Provides enforcement of RDMA resources defined by IB stack.1124 It is fairly easy for consumers to exhaust RDMA resources, which1125 can result into resource unavailability to other consumers.1126 RDMA controller is designed to stop this from happening.1127 Attaching processes with active RDMA resources to the cgroup1128 hierarchy is allowed even if can cross the hierarchy's limit.1129 1130config CGROUP_FREEZER1131 bool "Freezer controller"1132 help1133 Provides a way to freeze and unfreeze all tasks in a1134 cgroup.1135 1136 This option affects the ORIGINAL cgroup interface. The cgroup2 memory1137 controller includes important in-kernel memory consumers per default.1138 1139 If you're using cgroup2, say N.1140 1141config CGROUP_HUGETLB1142 bool "HugeTLB controller"1143 depends on HUGETLB_PAGE1144 select PAGE_COUNTER1145 default n1146 help1147 Provides a cgroup controller for HugeTLB pages.1148 When you enable this, you can put a per cgroup limit on HugeTLB usage.1149 The limit is enforced during page fault. Since HugeTLB doesn't1150 support page reclaim, enforcing the limit at page fault time implies1151 that, the application will get SIGBUS signal if it tries to access1152 HugeTLB pages beyond its limit. This requires the application to know1153 beforehand how much HugeTLB pages it would require for its use. The1154 control group is tracked in the third page lru pointer. This means1155 that we cannot use the controller with huge page less than 3 pages.1156 1157config CPUSETS1158 bool "Cpuset controller"1159 depends on SMP1160 help1161 This option will let you create and manage CPUSETs which1162 allow dynamically partitioning a system into sets of CPUs and1163 Memory Nodes and assigning tasks to run only within those sets.1164 This is primarily useful on large SMP or NUMA systems.1165 1166 Say N if unsure.1167 1168config CPUSETS_V11169 bool "Legacy cgroup v1 cpusets controller"1170 depends on CPUSETS1171 default n1172 help1173 Legacy cgroup v1 cpusets controller which has been deprecated by1174 cgroup v2 implementation. The v1 is there for legacy applications1175 which haven't migrated to the new cgroup v2 interface yet. If you1176 do not have any such application then you are completely fine leaving1177 this option disabled.1178 1179 Say N if unsure.1180 1181config PROC_PID_CPUSET1182 bool "Include legacy /proc/<pid>/cpuset file"1183 depends on CPUSETS1184 default y1185 1186config CGROUP_DEVICE1187 bool "Device controller"1188 help1189 Provides a cgroup controller implementing whitelists for1190 devices which a process in the cgroup can mknod or open.1191 1192config CGROUP_CPUACCT1193 bool "Simple CPU accounting controller"1194 help1195 Provides a simple controller for monitoring the1196 total CPU consumed by the tasks in a cgroup.1197 1198config CGROUP_PERF1199 bool "Perf controller"1200 depends on PERF_EVENTS1201 help1202 This option extends the perf per-cpu mode to restrict monitoring1203 to threads which belong to the cgroup specified and run on the1204 designated cpu. Or this can be used to have cgroup ID in samples1205 so that it can monitor performance events among cgroups.1206 1207 Say N if unsure.1208 1209config CGROUP_BPF1210 bool "Support for eBPF programs attached to cgroups"1211 depends on BPF_SYSCALL1212 select SOCK_CGROUP_DATA1213 help1214 Allow attaching eBPF programs to a cgroup using the bpf(2)1215 syscall command BPF_PROG_ATTACH.1216 1217 In which context these programs are accessed depends on the type1218 of attachment. For instance, programs that are attached using1219 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of1220 inet sockets.1221 1222config CGROUP_MISC1223 bool "Misc resource controller"1224 default n1225 help1226 Provides a controller for miscellaneous resources on a host.1227 1228 Miscellaneous scalar resources are the resources on the host system1229 which cannot be abstracted like the other cgroups. This controller1230 tracks and limits the miscellaneous resources used by a process1231 attached to a cgroup hierarchy.1232 1233 For more information, please check misc cgroup section in1234 /Documentation/admin-guide/cgroup-v2.rst.1235 1236config CGROUP_DEBUG1237 bool "Debug controller"1238 default n1239 depends on DEBUG_KERNEL1240 help1241 This option enables a simple controller that exports1242 debugging information about the cgroups framework. This1243 controller is for control cgroup debugging only. Its1244 interfaces are not stable.1245 1246 Say N.1247 1248config SOCK_CGROUP_DATA1249 bool1250 default n1251 1252endif # CGROUPS1253 1254menuconfig NAMESPACES1255 bool "Namespaces support" if EXPERT1256 depends on MULTIUSER1257 default !EXPERT1258 help1259 Provides the way to make tasks work with different objects using1260 the same id. For example same IPC id may refer to different objects1261 or same user id or pid may refer to different tasks when used in1262 different namespaces.1263 1264if NAMESPACES1265 1266config UTS_NS1267 bool "UTS namespace"1268 default y1269 help1270 In this namespace tasks see different info provided with the1271 uname() system call1272 1273config TIME_NS1274 bool "TIME namespace"1275 depends on GENERIC_VDSO_TIME_NS1276 default y1277 help1278 In this namespace boottime and monotonic clocks can be set.1279 The time will keep going with the same pace.1280 1281config IPC_NS1282 bool "IPC namespace"1283 depends on (SYSVIPC || POSIX_MQUEUE)1284 default y1285 help1286 In this namespace tasks work with IPC ids which correspond to1287 different IPC objects in different namespaces.1288 1289config USER_NS1290 bool "User namespace"1291 default n1292 help1293 This allows containers, i.e. vservers, to use user namespaces1294 to provide different user info for different servers.1295 1296 When user namespaces are enabled in the kernel it is1297 recommended that the MEMCG option also be enabled and that1298 user-space use the memory control groups to limit the amount1299 of memory a memory unprivileged users can use.1300 1301 If unsure, say N.1302 1303config PID_NS1304 bool "PID Namespaces"1305 default y1306 help1307 Support process id namespaces. This allows having multiple1308 processes with the same pid as long as they are in different1309 pid namespaces. This is a building block of containers.1310 1311config NET_NS1312 bool "Network namespace"1313 depends on NET1314 default y1315 help1316 Allow user space to create what appear to be multiple instances1317 of the network stack.1318 1319endif # NAMESPACES1320 1321config CHECKPOINT_RESTORE1322 bool "Checkpoint/restore support"1323 depends on PROC_FS1324 select PROC_CHILDREN1325 select KCMP1326 default n1327 help1328 Enables additional kernel features in a sake of checkpoint/restore.1329 In particular it adds auxiliary prctl codes to setup process text,1330 data and heap segment sizes, and a few additional /proc filesystem1331 entries.1332 1333 If unsure, say N here.1334 1335config SCHED_AUTOGROUP1336 bool "Automatic process group scheduling"1337 select CGROUPS1338 select CGROUP_SCHED1339 select FAIR_GROUP_SCHED1340 help1341 This option optimizes the scheduler for common desktop workloads by1342 automatically creating and populating task groups. This separation1343 of workloads isolates aggressive CPU burners (like build jobs) from1344 desktop applications. Task group autogeneration is currently based1345 upon task session.1346 1347config RELAY1348 bool "Kernel->user space relay support (formerly relayfs)"1349 select IRQ_WORK1350 help1351 This option enables support for relay interface support in1352 certain file systems (such as debugfs).1353 It is designed to provide an efficient mechanism for tools and1354 facilities to relay large amounts of data from kernel space to1355 user space.1356 1357 If unsure, say N.1358 1359config BLK_DEV_INITRD1360 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"1361 help1362 The initial RAM filesystem is a ramfs which is loaded by the1363 boot loader (loadlin or lilo) and that is mounted as root1364 before the normal boot procedure. It is typically used to1365 load modules needed to mount the "real" root file system,1366 etc. See <file:Documentation/admin-guide/initrd.rst> for details.1367 1368 If RAM disk support (BLK_DEV_RAM) is also included, this1369 also enables initial RAM disk (initrd) support and adds1370 15 Kbytes (more on some other architectures) to the kernel size.1371 1372 If unsure say Y.1373 1374if BLK_DEV_INITRD1375 1376source "usr/Kconfig"1377 1378endif1379 1380config BOOT_CONFIG1381 bool "Boot config support"1382 select BLK_DEV_INITRD if !BOOT_CONFIG_EMBED1383 help1384 Extra boot config allows system admin to pass a config file as1385 complemental extension of kernel cmdline when booting.1386 The boot config file must be attached at the end of initramfs1387 with checksum, size and magic word.1388 See <file:Documentation/admin-guide/bootconfig.rst> for details.1389 1390 If unsure, say Y.1391 1392config BOOT_CONFIG_FORCE1393 bool "Force unconditional bootconfig processing"1394 depends on BOOT_CONFIG1395 default y if BOOT_CONFIG_EMBED1396 help1397 With this Kconfig option set, BOOT_CONFIG processing is carried1398 out even when the "bootconfig" kernel-boot parameter is omitted.1399 In fact, with this Kconfig option set, there is no way to1400 make the kernel ignore the BOOT_CONFIG-supplied kernel-boot1401 parameters.1402 1403 If unsure, say N.1404 1405config BOOT_CONFIG_EMBED1406 bool "Embed bootconfig file in the kernel"1407 depends on BOOT_CONFIG1408 help1409 Embed a bootconfig file given by BOOT_CONFIG_EMBED_FILE in the1410 kernel. Usually, the bootconfig file is loaded with the initrd1411 image. But if the system doesn't support initrd, this option will1412 help you by embedding a bootconfig file while building the kernel.1413 1414 If unsure, say N.1415 1416config BOOT_CONFIG_EMBED_FILE1417 string "Embedded bootconfig file path"1418 depends on BOOT_CONFIG_EMBED1419 help1420 Specify a bootconfig file which will be embedded to the kernel.1421 This bootconfig will be used if there is no initrd or no other1422 bootconfig in the initrd.1423 1424config INITRAMFS_PRESERVE_MTIME1425 bool "Preserve cpio archive mtimes in initramfs"1426 default y1427 help1428 Each entry in an initramfs cpio archive carries an mtime value. When1429 enabled, extracted cpio items take this mtime, with directory mtime1430 setting deferred until after creation of any child entries.1431 1432 If unsure, say Y.1433 1434choice1435 prompt "Compiler optimization level"1436 default CC_OPTIMIZE_FOR_PERFORMANCE1437 1438config CC_OPTIMIZE_FOR_PERFORMANCE1439 bool "Optimize for performance (-O2)"1440 help1441 This is the default optimization level for the kernel, building1442 with the "-O2" compiler flag for best performance and most1443 helpful compile-time warnings.1444 1445config CC_OPTIMIZE_FOR_SIZE1446 bool "Optimize for size (-Os)"1447 help1448 Choosing this option will pass "-Os" to your compiler resulting1449 in a smaller kernel.1450 1451endchoice1452 1453config HAVE_LD_DEAD_CODE_DATA_ELIMINATION1454 bool1455 help1456 This requires that the arch annotates or otherwise protects1457 its external entry points from being discarded. Linker scripts1458 must also merge .text.*, .data.*, and .bss.* correctly into1459 output sections. Care must be taken not to pull in unrelated1460 sections (e.g., '.text.init'). Typically '.' in section names1461 is used to distinguish them from label names / C identifiers.1462 1463config LD_DEAD_CODE_DATA_ELIMINATION1464 bool "Dead code and data elimination (EXPERIMENTAL)"1465 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION1466 depends on EXPERT1467 depends on $(cc-option,-ffunction-sections -fdata-sections)1468 depends on $(ld-option,--gc-sections)1469 help1470 Enable this if you want to do dead code and data elimination with1471 the linker by compiling with -ffunction-sections -fdata-sections,1472 and linking with --gc-sections.1473 1474 This can reduce on disk and in-memory size of the kernel1475 code and static data, particularly for small configs and1476 on small systems. This has the possibility of introducing1477 silently broken kernel if the required annotations are not1478 present. This option is not well tested yet, so use at your1479 own risk.1480 1481config LD_ORPHAN_WARN1482 def_bool y1483 depends on ARCH_WANT_LD_ORPHAN_WARN1484 depends on $(ld-option,--orphan-handling=warn)1485 depends on $(ld-option,--orphan-handling=error)1486 1487config LD_ORPHAN_WARN_LEVEL1488 string1489 depends on LD_ORPHAN_WARN1490 default "error" if WERROR1491 default "warn"1492 1493config SYSCTL1494 bool1495 1496config HAVE_UID161497 bool1498 1499config SYSCTL_EXCEPTION_TRACE1500 bool1501 help1502 Enable support for /proc/sys/debug/exception-trace.1503 1504config SYSCTL_ARCH_UNALIGN_NO_WARN1505 bool1506 help1507 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap1508 Allows arch to define/use @no_unaligned_warning to possibly warn1509 about unaligned access emulation going on under the hood.1510 1511config SYSCTL_ARCH_UNALIGN_ALLOW1512 bool1513 help1514 Enable support for /proc/sys/kernel/unaligned-trap1515 Allows arches to define/use @unaligned_enabled to runtime toggle1516 the unaligned access emulation.1517 see arch/parisc/kernel/unaligned.c for reference1518 1519config HAVE_PCSPKR_PLATFORM1520 bool1521 1522menuconfig EXPERT1523 bool "Configure standard kernel features (expert users)"1524 # Unhide debug options, to make the on-by-default options visible1525 select DEBUG_KERNEL1526 help1527 This option allows certain base kernel options and settings1528 to be disabled or tweaked. This is for specialized1529 environments which can tolerate a "non-standard" kernel.1530 Only use this if you really know what you are doing.1531 1532config UID161533 bool "Enable 16-bit UID system calls" if EXPERT1534 depends on HAVE_UID16 && MULTIUSER1535 default y1536 help1537 This enables the legacy 16-bit UID syscall wrappers.1538 1539config MULTIUSER1540 bool "Multiple users, groups and capabilities support" if EXPERT1541 default y1542 help1543 This option enables support for non-root users, groups and1544 capabilities.1545 1546 If you say N here, all processes will run with UID 0, GID 0, and all1547 possible capabilities. Saying N here also compiles out support for1548 system calls related to UIDs, GIDs, and capabilities, such as setuid,1549 setgid, and capset.1550 1551 If unsure, say Y here.1552 1553config SGETMASK_SYSCALL1554 bool "sgetmask/ssetmask syscalls support" if EXPERT1555 default PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH1556 help1557 sys_sgetmask and sys_ssetmask are obsolete system calls1558 no longer supported in libc but still enabled by default in some1559 architectures.1560 1561 If unsure, leave the default option here.1562 1563config SYSFS_SYSCALL1564 bool "Sysfs syscall support" if EXPERT1565 default y1566 help1567 sys_sysfs is an obsolete system call no longer supported in libc.1568 Note that disabling this option is more secure but might break1569 compatibility with some systems.1570 1571 If unsure say Y here.1572 1573config FHANDLE1574 bool "open by fhandle syscalls" if EXPERT1575 select EXPORTFS1576 default y1577 help1578 If you say Y here, a user level program will be able to map1579 file names to handle and then later use the handle for1580 different file system operations. This is useful in implementing1581 userspace file servers, which now track files using handles instead1582 of names. The handle would remain the same even if file names1583 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)1584 syscalls.1585 1586config POSIX_TIMERS1587 bool "Posix Clocks & timers" if EXPERT1588 default y1589 help1590 This includes native support for POSIX timers to the kernel.1591 Some embedded systems have no use for them and therefore they1592 can be configured out to reduce the size of the kernel image.1593 1594 When this option is disabled, the following syscalls won't be1595 available: timer_create, timer_gettime: timer_getoverrun,1596 timer_settime, timer_delete, clock_adjtime, getitimer,1597 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,1598 clock_getres and clock_nanosleep syscalls will be limited to1599 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.1600 1601 If unsure say y.1602 1603config PRINTK1604 default y1605 bool "Enable support for printk" if EXPERT1606 select IRQ_WORK1607 help1608 This option enables normal printk support. Removing it1609 eliminates most of the message strings from the kernel image1610 and makes the kernel more or less silent. As this makes it1611 very difficult to diagnose system problems, saying N here is1612 strongly discouraged.1613 1614config BUG1615 bool "BUG() support" if EXPERT1616 default y1617 help1618 Disabling this option eliminates support for BUG and WARN, reducing1619 the size of your kernel image and potentially quietly ignoring1620 numerous fatal conditions. You should only consider disabling this1621 option for embedded systems with no facilities for reporting errors.1622 Just say Y.1623 1624config ELF_CORE1625 depends on COREDUMP1626 default y1627 bool "Enable ELF core dumps" if EXPERT1628 help1629 Enable support for generating core dumps. Disabling saves about 4k.1630 1631 1632config PCSPKR_PLATFORM1633 bool "Enable PC-Speaker support" if EXPERT1634 depends on HAVE_PCSPKR_PLATFORM1635 select I8253_LOCK1636 default y1637 help1638 This option allows to disable the internal PC-Speaker1639 support, saving some memory.1640 1641config BASE_SMALL1642 bool "Enable smaller-sized data structures for core" if EXPERT1643 help1644 Enabling this option reduces the size of miscellaneous core1645 kernel data structures. This saves memory on small machines,1646 but may reduce performance.1647 1648config FUTEX1649 bool "Enable futex support" if EXPERT1650 depends on !(SPARC32 && SMP)1651 default y1652 imply RT_MUTEXES1653 help1654 Disabling this option will cause the kernel to be built without1655 support for "fast userspace mutexes". The resulting kernel may not1656 run glibc-based applications correctly.1657 1658config FUTEX_PI1659 bool1660 depends on FUTEX && RT_MUTEXES1661 default y1662 1663config EPOLL1664 bool "Enable eventpoll support" if EXPERT1665 default y1666 help1667 Disabling this option will cause the kernel to be built without1668 support for epoll family of system calls.1669 1670config SIGNALFD1671 bool "Enable signalfd() system call" if EXPERT1672 default y1673 help1674 Enable the signalfd() system call that allows to receive signals1675 on a file descriptor.1676 1677 If unsure, say Y.1678 1679config TIMERFD1680 bool "Enable timerfd() system call" if EXPERT1681 default y1682 help1683 Enable the timerfd() system call that allows to receive timer1684 events on a file descriptor.1685 1686 If unsure, say Y.1687 1688config EVENTFD1689 bool "Enable eventfd() system call" if EXPERT1690 default y1691 help1692 Enable the eventfd() system call that allows to receive both1693 kernel notification (ie. KAIO) or userspace notifications.1694 1695 If unsure, say Y.1696 1697config SHMEM1698 bool "Use full shmem filesystem" if EXPERT1699 default y1700 depends on MMU1701 help1702 The shmem is an internal filesystem used to manage shared memory.1703 It is backed by swap and manages resource limits. It is also exported1704 to userspace as tmpfs if TMPFS is enabled. Disabling this1705 option replaces shmem and tmpfs with the much simpler ramfs code,1706 which may be appropriate on small systems without swap.1707 1708config AIO1709 bool "Enable AIO support" if EXPERT1710 default y1711 help1712 This option enables POSIX asynchronous I/O which may by used1713 by some high performance threaded applications. Disabling1714 this option saves about 7k.1715 1716config IO_URING1717 bool "Enable IO uring support" if EXPERT1718 select IO_WQ1719 default y1720 help1721 This option enables support for the io_uring interface, enabling1722 applications to submit and complete IO through submission and1723 completion rings that are shared between the kernel and application.1724 1725config GCOV_PROFILE_URING1726 bool "Enable GCOV profiling on the io_uring subsystem"1727 depends on GCOV_KERNEL1728 help1729 Enable GCOV profiling on the io_uring subsystem, to facilitate1730 code coverage testing.1731 1732 If unsure, say N.1733 1734 Note that this will have a negative impact on the performance of1735 the io_uring subsystem, hence this should only be enabled for1736 specific test purposes.1737 1738config ADVISE_SYSCALLS1739 bool "Enable madvise/fadvise syscalls" if EXPERT1740 default y1741 help1742 This option enables the madvise and fadvise syscalls, used by1743 applications to advise the kernel about their future memory or file1744 usage, improving performance. If building an embedded system where no1745 applications use these syscalls, you can disable this option to save1746 space.1747 1748config MEMBARRIER1749 bool "Enable membarrier() system call" if EXPERT1750 default y1751 help1752 Enable the membarrier() system call that allows issuing memory1753 barriers across all running threads, which can be used to distribute1754 the cost of user-space memory barriers asymmetrically by transforming1755 pairs of memory barriers into pairs consisting of membarrier() and a1756 compiler barrier.1757 1758 If unsure, say Y.1759 1760config KCMP1761 bool "Enable kcmp() system call" if EXPERT1762 help1763 Enable the kernel resource comparison system call. It provides1764 user-space with the ability to compare two processes to see if they1765 share a common resource, such as a file descriptor or even virtual1766 memory space.1767 1768 If unsure, say N.1769 1770config RSEQ1771 bool "Enable rseq() system call" if EXPERT1772 default y1773 depends on HAVE_RSEQ1774 select MEMBARRIER1775 help1776 Enable the restartable sequences system call. It provides a1777 user-space cache for the current CPU number value, which1778 speeds up getting the current CPU number from user-space,1779 as well as an ABI to speed up user-space operations on1780 per-CPU data.1781 1782 If unsure, say Y.1783 1784config DEBUG_RSEQ1785 default n1786 bool "Enable debugging of rseq() system call" if EXPERT1787 depends on RSEQ && DEBUG_KERNEL1788 help1789 Enable extra debugging checks for the rseq system call.1790 1791 If unsure, say N.1792 1793config CACHESTAT_SYSCALL1794 bool "Enable cachestat() system call" if EXPERT1795 default y1796 help1797 Enable the cachestat system call, which queries the page cache1798 statistics of a file (number of cached pages, dirty pages,1799 pages marked for writeback, (recently) evicted pages).1800 1801 If unsure say Y here.1802 1803config PC1041804 bool "PC/104 support" if EXPERT1805 help1806 Expose PC/104 form factor device drivers and options available for1807 selection and configuration. Enable this option if your target1808 machine has a PC/104 bus.1809 1810config KALLSYMS1811 bool "Load all symbols for debugging/ksymoops" if EXPERT1812 default y1813 help1814 Say Y here to let the kernel print out symbolic crash information and1815 symbolic stack backtraces. This increases the size of the kernel1816 somewhat, as all symbols have to be loaded into the kernel image.1817 1818config KALLSYMS_SELFTEST1819 bool "Test the basic functions and performance of kallsyms"1820 depends on KALLSYMS1821 default n1822 help1823 Test the basic functions and performance of some interfaces, such as1824 kallsyms_lookup_name. It also calculates the compression rate of the1825 kallsyms compression algorithm for the current symbol set.1826 1827 Start self-test automatically after system startup. Suggest executing1828 "dmesg | grep kallsyms_selftest" to collect test results. "finish" is1829 displayed in the last line, indicating that the test is complete.1830 1831config KALLSYMS_ALL1832 bool "Include all symbols in kallsyms"1833 depends on DEBUG_KERNEL && KALLSYMS1834 help1835 Normally kallsyms only contains the symbols of functions for nicer1836 OOPS messages and backtraces (i.e., symbols from the text and inittext1837 sections). This is sufficient for most cases. And only if you want to1838 enable kernel live patching, or other less common use cases (e.g.,1839 when a debugger is used) all symbols are required (i.e., names of1840 variables from the data sections, etc).1841 1842 This option makes sure that all symbols are loaded into the kernel1843 image (i.e., symbols from all sections) in cost of increased kernel1844 size (depending on the kernel configuration, it may be 300KiB or1845 something like this).1846 1847 Say N unless you really need all symbols, or kernel live patching.1848 1849config KALLSYMS_ABSOLUTE_PERCPU1850 bool1851 depends on KALLSYMS1852 default X86_64 && SMP1853 1854# end of the "standard kernel features (expert users)" menu1855 1856config ARCH_HAS_MEMBARRIER_CALLBACKS1857 bool1858 1859config ARCH_HAS_MEMBARRIER_SYNC_CORE1860 bool1861 1862config HAVE_PERF_EVENTS1863 bool1864 help1865 See tools/perf/design.txt for details.1866 1867config GUEST_PERF_EVENTS1868 bool1869 depends on HAVE_PERF_EVENTS1870 1871config PERF_USE_VMALLOC1872 bool1873 help1874 See tools/perf/design.txt for details1875 1876menu "Kernel Performance Events And Counters"1877 1878config PERF_EVENTS1879 bool "Kernel performance events and counters"1880 default y if PROFILING1881 depends on HAVE_PERF_EVENTS1882 select IRQ_WORK1883 help1884 Enable kernel support for various performance events provided1885 by software and hardware.1886 1887 Software events are supported either built-in or via the1888 use of generic tracepoints.1889 1890 Most modern CPUs support performance events via performance1891 counter registers. These registers count the number of certain1892 types of hw events: such as instructions executed, cachemisses1893 suffered, or branches mis-predicted - without slowing down the1894 kernel or applications. These registers can also trigger interrupts1895 when a threshold number of events have passed - and can thus be1896 used to profile the code that runs on that CPU.1897 1898 The Linux Performance Event subsystem provides an abstraction of1899 these software and hardware event capabilities, available via a1900 system call and used by the "perf" utility in tools/perf/. It1901 provides per task and per CPU counters, and it provides event1902 capabilities on top of those.1903 1904 Say Y if unsure.1905 1906config DEBUG_PERF_USE_VMALLOC1907 default n1908 bool "Debug: use vmalloc to back perf mmap() buffers"1909 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC1910 select PERF_USE_VMALLOC1911 help1912 Use vmalloc memory to back perf mmap() buffers.1913 1914 Mostly useful for debugging the vmalloc code on platforms1915 that don't require it.1916 1917 Say N if unsure.1918 1919endmenu1920 1921config SYSTEM_DATA_VERIFICATION1922 def_bool n1923 select SYSTEM_TRUSTED_KEYRING1924 select KEYS1925 select CRYPTO1926 select CRYPTO_RSA1927 select ASYMMETRIC_KEY_TYPE1928 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE1929 select ASN11930 select OID_REGISTRY1931 select X509_CERTIFICATE_PARSER1932 select PKCS7_MESSAGE_PARSER1933 help1934 Provide PKCS#7 message verification using the contents of the system1935 trusted keyring to provide public keys. This then can be used for1936 module verification, kexec image verification and firmware blob1937 verification.1938 1939config PROFILING1940 bool "Profiling support"1941 help1942 Say Y here to enable the extended profiling support mechanisms used1943 by profilers.1944 1945config RUST1946 bool "Rust support"1947 depends on HAVE_RUST1948 depends on RUST_IS_AVAILABLE1949 depends on !MODVERSIONS1950 depends on !GCC_PLUGIN_RANDSTRUCT1951 depends on !RANDSTRUCT1952 depends on !DEBUG_INFO_BTF || PAHOLE_HAS_LANG_EXCLUDE1953 depends on !CFI_CLANG || HAVE_CFI_ICALL_NORMALIZE_INTEGERS_RUSTC1954 select CFI_ICALL_NORMALIZE_INTEGERS if CFI_CLANG1955 depends on !CALL_PADDING || RUSTC_VERSION >= 1081001956 depends on !KASAN_SW_TAGS1957 depends on !(MITIGATION_RETHUNK && KASAN) || RUSTC_VERSION >= 1083001958 help1959 Enables Rust support in the kernel.1960 1961 This allows other Rust-related options, like drivers written in Rust,1962 to be selected.1963 1964 It is also required to be able to load external kernel modules1965 written in Rust.1966 1967 See Documentation/rust/ for more information.1968 1969 If unsure, say N.1970 1971config RUSTC_VERSION_TEXT1972 string1973 depends on RUST1974 default "$(RUSTC_VERSION_TEXT)"1975 help1976 See `CC_VERSION_TEXT`.1977 1978config BINDGEN_VERSION_TEXT1979 string1980 depends on RUST1981 # The dummy parameter `workaround-for-0.69.0` is required to support 0.69.01982 # (https://github.com/rust-lang/rust-bindgen/pull/2678). It can be removed when1983 # the minimum version is upgraded past that (0.69.1 already fixed the issue).1984 default "$(shell,$(BINDGEN) --version workaround-for-0.69.0 2>/dev/null)"1985 1986#1987# Place an empty function call at each tracepoint site. Can be1988# dynamically changed for a probe function.1989#1990config TRACEPOINTS1991 bool1992 1993source "kernel/Kconfig.kexec"1994 1995endmenu # General setup1996 1997source "arch/Kconfig"1998 1999config RT_MUTEXES2000 bool2001 default y if PREEMPT_RT2002 2003config MODULE_SIG_FORMAT2004 def_bool n2005 select SYSTEM_DATA_VERIFICATION2006 2007source "kernel/module/Kconfig"2008 2009config INIT_ALL_POSSIBLE2010 bool2011 help2012 Back when each arch used to define their own cpu_online_mask and2013 cpu_possible_mask, some of them chose to initialize cpu_possible_mask2014 with all 1s, and others with all 0s. When they were centralised,2015 it was better to provide this option than to break all the archs2016 and have several arch maintainers pursuing me down dark alleys.2017 2018source "block/Kconfig"2019 2020config PREEMPT_NOTIFIERS2021 bool2022 2023config PADATA2024 depends on SMP2025 bool2026 2027config ASN12028 tristate2029 help2030 Build a simple ASN.1 grammar compiler that produces a bytecode output2031 that can be interpreted by the ASN.1 stream decoder and used to2032 inform it as to what tags are to be expected in a stream and what2033 functions to call on what tags.2034 2035source "kernel/Kconfig.locks"2036 2037config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE2038 bool2039 2040config ARCH_HAS_PREPARE_SYNC_CORE_CMD2041 bool2042 2043config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE2044 bool2045 2046# It may be useful for an architecture to override the definitions of the2047# SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>2048# and the COMPAT_ variants in <linux/compat.h>, in particular to use a2049# different calling convention for syscalls. They can also override the2050# macros for not-implemented syscalls in kernel/sys_ni.c and2051# kernel/time/posix-stubs.c. All these overrides need to be available in2052# <asm/syscall_wrapper.h>.2053config ARCH_HAS_SYSCALL_WRAPPER2054 def_bool n2055