1 # SPDX-License-Identifier: GPL-2.0-only
4 default "$(CC_VERSION_TEXT)"
6 This is used in unclear ways:
8 - Re-run Kconfig when the compiler is updated
9 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.
13 - Ensure full rebuild when the compiler is updated
14 include/linux/compiler-version.h contains this option in the comment
15 line so fixdep adds include/config/CC_VERSION_TEXT into the
16 auto-generated dependency. When the compiler is updated, syncconfig
17 will touch it and then every file will be rebuilt.
20 def_bool $(success,test "$(cc-name)" = GCC)
24 default $(cc-version) if CC_IS_GCC
28 def_bool $(success,test "$(cc-name)" = Clang)
32 default $(cc-version) if CC_IS_CLANG
36 def_bool $(success,test "$(as-name)" = GNU)
39 def_bool $(success,test "$(as-name)" = LLVM)
43 # Use clang version if this is the integrated assembler
44 default CLANG_VERSION if AS_IS_LLVM
48 def_bool $(success,test "$(ld-name)" = BFD)
52 default $(ld-version) if LD_IS_BFD
56 def_bool $(success,test "$(ld-name)" = LLD)
60 default $(ld-version) if LD_IS_LLD
65 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
66 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
68 config CC_CAN_LINK_STATIC
70 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
71 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
73 config CC_HAS_ASM_GOTO
74 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
76 config CC_HAS_ASM_GOTO_OUTPUT
77 depends on CC_HAS_ASM_GOTO
78 def_bool $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
80 config TOOLS_SUPPORT_RELR
81 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
83 config CC_HAS_ASM_INLINE
84 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
86 config CC_HAS_NO_PROFILE_FN_ATTR
87 def_bool $(success,echo '__attribute__((no_profile_instrument_function)) int x();' | $(CC) -x c - -c -o /dev/null -Werror)
95 config BUILDTIME_TABLE_SORT
98 config THREAD_INFO_IN_TASK
101 Select this to move thread_info off the stack into task_struct. To
102 make this work, an arch will need to remove all thread_info fields
103 except flags and fix any runtime bugs.
105 One subtle change that will be needed is to use try_get_task_stack()
106 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
115 depends on BROKEN || !SMP
118 config INIT_ENV_ARG_LIMIT
123 Maximum of each of the number of arguments and environment
124 variables passed to init from the kernel command line.
127 bool "Compile also drivers which will not load"
130 Some drivers can be compiled on a different platform than they are
131 intended to be run on. Despite they cannot be loaded there (or even
132 when they load they cannot be used due to missing HW support),
133 developers still, opposing to distributors, might want to build such
134 drivers to compile-test them.
136 If you are a developer and want to build everything available, say Y
137 here. If you are a user/distributor, say N here to exclude useless
138 drivers to be distributed.
140 config UAPI_HEADER_TEST
141 bool "Compile test UAPI headers"
142 depends on HEADERS_INSTALL && CC_CAN_LINK
144 Compile test headers exported to user-space to ensure they are
145 self-contained, i.e. compilable as standalone units.
147 If you are a developer or tester and want to ensure the exported
148 headers are self-contained, say Y here. Otherwise, choose N.
151 string "Local version - append to kernel release"
153 Append an extra string to the end of your kernel version.
154 This will show up when you type uname, for example.
155 The string you set here will be appended after the contents of
156 any files with a filename matching localversion* in your
157 object and source tree, in that order. Your total string can
158 be a maximum of 64 characters.
160 config LOCALVERSION_AUTO
161 bool "Automatically append version information to the version string"
163 depends on !COMPILE_TEST
165 This will try to automatically determine if the current tree is a
166 release tree by looking for git tags that belong to the current
167 top of tree revision.
169 A string of the format -gxxxxxxxx will be added to the localversion
170 if a git-based tree is found. The string generated by this will be
171 appended after any matching localversion* files, and after the value
172 set in CONFIG_LOCALVERSION.
174 (The actual string used here is the first eight characters produced
175 by running the command:
177 $ git rev-parse --verify HEAD
179 which is done within the script "scripts/setlocalversion".)
182 string "Build ID Salt"
185 The build ID is used to link binaries and their debug info. Setting
186 this option will use the value in the calculation of the build id.
187 This is mostly useful for distributions which want to ensure the
188 build is unique between builds. It's safe to leave the default.
190 config HAVE_KERNEL_GZIP
193 config HAVE_KERNEL_BZIP2
196 config HAVE_KERNEL_LZMA
199 config HAVE_KERNEL_XZ
202 config HAVE_KERNEL_LZO
205 config HAVE_KERNEL_LZ4
208 config HAVE_KERNEL_ZSTD
211 config HAVE_KERNEL_UNCOMPRESSED
215 prompt "Kernel compression mode"
217 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_UNCOMPRESSED
219 The linux kernel is a kind of self-extracting executable.
220 Several compression algorithms are available, which differ
221 in efficiency, compression and decompression speed.
222 Compression speed is only relevant when building a kernel.
223 Decompression speed is relevant at each boot.
225 If you have any problems with bzip2 or lzma compressed
227 version of this functionality (bzip2 only), for 2.4, was
228 supplied by Christian Ludwig)
230 High compression options are mostly useful for users, who
231 are low on disk space (embedded systems), but for whom ram
234 If in doubt, select 'gzip'
238 depends on HAVE_KERNEL_GZIP
240 The old and tried gzip compression. It provides a good balance
241 between compression ratio and decompression speed.
245 depends on HAVE_KERNEL_BZIP2
247 Its compression ratio and speed is intermediate.
248 Decompression speed is slowest among the choices. The kernel
249 size is about 10% smaller with bzip2, in comparison to gzip.
250 Bzip2 uses a large amount of memory. For modern kernels you
251 will need at least 8MB RAM or more for booting.
255 depends on HAVE_KERNEL_LZMA
257 This compression algorithm's ratio is best. Decompression speed
258 is between gzip and bzip2. Compression is slowest.
259 The kernel size is about 33% smaller with LZMA in comparison to gzip.
263 depends on HAVE_KERNEL_XZ
265 XZ uses the LZMA2 algorithm and instruction set specific
266 BCJ filters which can improve compression ratio of executable
267 code. The size of the kernel is about 30% smaller with XZ in
268 comparison to gzip. On architectures for which there is a BCJ
269 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
270 will create a few percent smaller kernel than plain LZMA.
272 The speed is about the same as with LZMA: The decompression
273 speed of XZ is better than that of bzip2 but worse than gzip
274 and LZO. Compression is slow.
278 depends on HAVE_KERNEL_LZO
280 Its compression ratio is the poorest among the choices. The kernel
281 size is about 10% bigger than gzip; however its speed
282 (both compression and decompression) is the fastest.
286 depends on HAVE_KERNEL_LZ4
288 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
289 A preliminary version of LZ4 de/compression tool is available at
290 <https://code.google.com/p/lz4/>.
292 Its compression ratio is worse than LZO. The size of the kernel
293 is about 8% bigger than LZO. But the decompression speed is
298 depends on HAVE_KERNEL_ZSTD
300 ZSTD is a compression algorithm targeting intermediate compression
301 with fast decompression speed. It will compress better than GZIP and
302 decompress around the same speed as LZO, but slower than LZ4. You
303 will need at least 192 KB RAM or more for booting. The zstd command
304 line tool is required for compression.
306 config KERNEL_UNCOMPRESSED
308 depends on HAVE_KERNEL_UNCOMPRESSED
310 Produce uncompressed kernel image. This option is usually not what
311 you want. It is useful for debugging the kernel in slow simulation
312 environments, where decompressing and moving the kernel is awfully
313 slow. This option allows early boot code to skip the decompressor
314 and jump right at uncompressed kernel image.
319 string "Default init path"
322 This option determines the default init for the system if no init=
323 option is passed on the kernel command line. If the requested path is
324 not present, we will still then move on to attempting further
325 locations (e.g. /sbin/init, etc). If this is empty, we will just use
326 the fallback list when init= is not passed.
328 config DEFAULT_HOSTNAME
329 string "Default hostname"
332 This option determines the default system hostname before userspace
333 calls sethostname(2). The kernel traditionally uses "(none)" here,
334 but you may wish to use a different default here to make a minimal
335 system more usable with less configuration.
338 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
339 # add proper SWAP support to them, in which case this can be remove.
345 bool "Support for paging of anonymous memory (swap)"
346 depends on MMU && BLOCK && !ARCH_NO_SWAP
349 This option allows you to choose whether you want to have support
350 for so called swap devices or swap files in your kernel that are
351 used to provide more virtual memory than the actual RAM present
352 in your computer. If unsure say Y.
357 Inter Process Communication is a suite of library functions and
358 system calls which let processes (running programs) synchronize and
359 exchange information. It is generally considered to be a good thing,
360 and some programs won't run unless you say Y here. In particular, if
361 you want to run the DOS emulator dosemu under Linux (read the
362 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
363 you'll need to say Y here.
365 You can find documentation about IPC with "info ipc" and also in
366 section 6.4 of the Linux Programmer's Guide, available from
367 <http://www.tldp.org/guides.html>.
369 config SYSVIPC_SYSCTL
376 bool "POSIX Message Queues"
379 POSIX variant of message queues is a part of IPC. In POSIX message
380 queues every message has a priority which decides about succession
381 of receiving it by a process. If you want to compile and run
382 programs written e.g. for Solaris with use of its POSIX message
383 queues (functions mq_*) say Y here.
385 POSIX message queues are visible as a filesystem called 'mqueue'
386 and can be mounted somewhere if you want to do filesystem
387 operations on message queues.
391 config POSIX_MQUEUE_SYSCTL
393 depends on POSIX_MQUEUE
398 bool "General notification queue"
402 This is a general notification queue for the kernel to pass events to
403 userspace by splicing them into pipes. It can be used in conjunction
404 with watches for key/keyring change notifications and device
407 See Documentation/watch_queue.rst
409 config CROSS_MEMORY_ATTACH
410 bool "Enable process_vm_readv/writev syscalls"
414 Enabling this option adds the system calls process_vm_readv and
415 process_vm_writev which allow a process with the correct privileges
416 to directly read from or write to another process' address space.
417 See the man page for more details.
420 bool "uselib syscall"
421 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
423 This option enables the uselib syscall, a system call used in the
424 dynamic linker from libc5 and earlier. glibc does not use this
425 system call. If you intend to run programs built on libc5 or
426 earlier, you may need to enable this syscall. Current systems
427 running glibc can safely disable this.
430 bool "Auditing support"
433 Enable auditing infrastructure that can be used with another
434 kernel subsystem, such as SELinux (which requires this for
435 logging of avc messages output). System call auditing is included
436 on architectures which support it.
438 config HAVE_ARCH_AUDITSYSCALL
443 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
446 source "kernel/irq/Kconfig"
447 source "kernel/time/Kconfig"
448 source "kernel/bpf/Kconfig"
449 source "kernel/Kconfig.preempt"
451 menu "CPU/Task time and stats accounting"
453 config VIRT_CPU_ACCOUNTING
457 prompt "Cputime accounting"
458 default TICK_CPU_ACCOUNTING if !PPC64
459 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
461 # Kind of a stub config for the pure tick based cputime accounting
462 config TICK_CPU_ACCOUNTING
463 bool "Simple tick based cputime accounting"
464 depends on !S390 && !NO_HZ_FULL
466 This is the basic tick based cputime accounting that maintains
467 statistics about user, system and idle time spent on per jiffies
472 config VIRT_CPU_ACCOUNTING_NATIVE
473 bool "Deterministic task and CPU time accounting"
474 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
475 select VIRT_CPU_ACCOUNTING
477 Select this option to enable more accurate task and CPU time
478 accounting. This is done by reading a CPU counter on each
479 kernel entry and exit and on transitions within the kernel
480 between system, softirq and hardirq state, so there is a
481 small performance impact. In the case of s390 or IBM POWER > 5,
482 this also enables accounting of stolen time on logically-partitioned
485 config VIRT_CPU_ACCOUNTING_GEN
486 bool "Full dynticks CPU time accounting"
487 depends on HAVE_CONTEXT_TRACKING
488 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
489 depends on GENERIC_CLOCKEVENTS
490 select VIRT_CPU_ACCOUNTING
491 select CONTEXT_TRACKING
493 Select this option to enable task and CPU time accounting on full
494 dynticks systems. This accounting is implemented by watching every
495 kernel-user boundaries using the context tracking subsystem.
496 The accounting is thus performed at the expense of some significant
499 For now this is only useful if you are working on the full
500 dynticks subsystem development.
506 config IRQ_TIME_ACCOUNTING
507 bool "Fine granularity task level IRQ time accounting"
508 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
510 Select this option to enable fine granularity task irq time
511 accounting. This is done by reading a timestamp on each
512 transitions between softirq and hardirq state, so there can be a
513 small performance impact.
515 If in doubt, say N here.
517 config HAVE_SCHED_AVG_IRQ
519 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
522 config SCHED_THERMAL_PRESSURE
524 default y if ARM && ARM_CPU_TOPOLOGY
527 depends on CPU_FREQ_THERMAL
529 Select this option to enable thermal pressure accounting in the
530 scheduler. Thermal pressure is the value conveyed to the scheduler
531 that reflects the reduction in CPU compute capacity resulted from
532 thermal throttling. Thermal throttling occurs when the performance of
533 a CPU is capped due to high operating temperatures.
535 If selected, the scheduler will be able to balance tasks accordingly,
536 i.e. put less load on throttled CPUs than on non/less throttled ones.
538 This requires the architecture to implement
539 arch_set_thermal_pressure() and arch_scale_thermal_pressure().
541 config BSD_PROCESS_ACCT
542 bool "BSD Process Accounting"
545 If you say Y here, a user level program will be able to instruct the
546 kernel (via a special system call) to write process accounting
547 information to a file: whenever a process exits, information about
548 that process will be appended to the file by the kernel. The
549 information includes things such as creation time, owning user,
550 command name, memory usage, controlling terminal etc. (the complete
551 list is in the struct acct in <file:include/linux/acct.h>). It is
552 up to the user level program to do useful things with this
553 information. This is generally a good idea, so say Y.
555 config BSD_PROCESS_ACCT_V3
556 bool "BSD Process Accounting version 3 file format"
557 depends on BSD_PROCESS_ACCT
560 If you say Y here, the process accounting information is written
561 in a new file format that also logs the process IDs of each
562 process and its parent. Note that this file format is incompatible
563 with previous v0/v1/v2 file formats, so you will need updated tools
564 for processing it. A preliminary version of these tools is available
565 at <http://www.gnu.org/software/acct/>.
568 bool "Export task/process statistics through netlink"
573 Export selected statistics for tasks/processes through the
574 generic netlink interface. Unlike BSD process accounting, the
575 statistics are available during the lifetime of tasks/processes as
576 responses to commands. Like BSD accounting, they are sent to user
581 config TASK_DELAY_ACCT
582 bool "Enable per-task delay accounting"
586 Collect information on time spent by a task waiting for system
587 resources like cpu, synchronous block I/O completion and swapping
588 in pages. Such statistics can help in setting a task's priorities
589 relative to other tasks for cpu, io, rss limits etc.
594 bool "Enable extended accounting over taskstats"
597 Collect extended task accounting data and send the data
598 to userland for processing over the taskstats interface.
602 config TASK_IO_ACCOUNTING
603 bool "Enable per-task storage I/O accounting"
604 depends on TASK_XACCT
606 Collect information on the number of bytes of storage I/O which this
612 bool "Pressure stall information tracking"
614 Collect metrics that indicate how overcommitted the CPU, memory,
615 and IO capacity are in the system.
617 If you say Y here, the kernel will create /proc/pressure/ with the
618 pressure statistics files cpu, memory, and io. These will indicate
619 the share of walltime in which some or all tasks in the system are
620 delayed due to contention of the respective resource.
622 In kernels with cgroup support, cgroups (cgroup2 only) will
623 have cpu.pressure, memory.pressure, and io.pressure files,
624 which aggregate pressure stalls for the grouped tasks only.
626 For more details see Documentation/accounting/psi.rst.
630 config PSI_DEFAULT_DISABLED
631 bool "Require boot parameter to enable pressure stall information tracking"
635 If set, pressure stall information tracking will be disabled
636 per default but can be enabled through passing psi=1 on the
637 kernel commandline during boot.
639 This feature adds some code to the task wakeup and sleep
640 paths of the scheduler. The overhead is too low to affect
641 common scheduling-intense workloads in practice (such as
642 webservers, memcache), but it does show up in artificial
643 scheduler stress tests, such as hackbench.
645 If you are paranoid and not sure what the kernel will be
650 endmenu # "CPU/Task time and stats accounting"
654 depends on SMP || COMPILE_TEST
657 Make sure that CPUs running critical tasks are not disturbed by
658 any source of "noise" such as unbound workqueues, timers, kthreads...
659 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
660 the "isolcpus=" boot parameter.
664 source "kernel/rcu/Kconfig"
671 tristate "Kernel .config support"
673 This option enables the complete Linux kernel ".config" file
674 contents to be saved in the kernel. It provides documentation
675 of which kernel options are used in a running kernel or in an
676 on-disk kernel. This information can be extracted from the kernel
677 image file with the script scripts/extract-ikconfig and used as
678 input to rebuild the current kernel or to build another kernel.
679 It can also be extracted from a running kernel by reading
680 /proc/config.gz if enabled (below).
683 bool "Enable access to .config through /proc/config.gz"
684 depends on IKCONFIG && PROC_FS
686 This option enables access to the kernel configuration file
687 through /proc/config.gz.
690 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
693 This option enables access to the in-kernel headers that are generated during
694 the build process. These can be used to build eBPF tracing programs,
695 or similar programs. If you build the headers as a module, a module called
696 kheaders.ko is built which can be loaded on-demand to get access to headers.
699 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
700 range 12 25 if !H8300
705 Select the minimal kernel log buffer size as a power of 2.
706 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
707 parameter, see below. Any higher size also might be forced
708 by "log_buf_len" boot parameter.
718 config LOG_CPU_MAX_BUF_SHIFT
719 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
722 default 12 if !BASE_SMALL
723 default 0 if BASE_SMALL
726 This option allows to increase the default ring buffer size
727 according to the number of CPUs. The value defines the contribution
728 of each CPU as a power of 2. The used space is typically only few
729 lines however it might be much more when problems are reported,
732 The increased size means that a new buffer has to be allocated and
733 the original static one is unused. It makes sense only on systems
734 with more CPUs. Therefore this value is used only when the sum of
735 contributions is greater than the half of the default kernel ring
736 buffer as defined by LOG_BUF_SHIFT. The default values are set
737 so that more than 16 CPUs are needed to trigger the allocation.
739 Also this option is ignored when "log_buf_len" kernel parameter is
740 used as it forces an exact (power of two) size of the ring buffer.
742 The number of possible CPUs is used for this computation ignoring
743 hotplugging making the computation optimal for the worst case
744 scenario while allowing a simple algorithm to be used from bootup.
746 Examples shift values and their meaning:
747 17 => 128 KB for each CPU
748 16 => 64 KB for each CPU
749 15 => 32 KB for each CPU
750 14 => 16 KB for each CPU
751 13 => 8 KB for each CPU
752 12 => 4 KB for each CPU
754 config PRINTK_SAFE_LOG_BUF_SHIFT
755 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
760 Select the size of an alternate printk per-CPU buffer where messages
761 printed from usafe contexts are temporary stored. One example would
762 be NMI messages, another one - printk recursion. The messages are
763 copied to the main log buffer in a safe context to avoid a deadlock.
764 The value defines the size as a power of 2.
766 Those messages are rare and limited. The largest one is when
767 a backtrace is printed. It usually fits into 4KB. Select
768 8KB if you want to be on the safe side.
771 17 => 128 KB for each CPU
772 16 => 64 KB for each CPU
773 15 => 32 KB for each CPU
774 14 => 16 KB for each CPU
775 13 => 8 KB for each CPU
776 12 => 4 KB for each CPU
779 bool "Printk indexing debugfs interface"
780 depends on PRINTK && DEBUG_FS
782 Add support for indexing of all printk formats known at compile time
783 at <debugfs>/printk/index/<module>.
785 This can be used as part of maintaining daemons which monitor
786 /dev/kmsg, as it permits auditing the printk formats present in a
787 kernel, allowing detection of cases where monitored printks are
788 changed or no longer present.
790 There is no additional runtime cost to printk with this enabled.
793 # Architectures with an unreliable sched_clock() should select this:
795 config HAVE_UNSTABLE_SCHED_CLOCK
798 config GENERIC_SCHED_CLOCK
801 menu "Scheduler features"
804 bool "Enable utilization clamping for RT/FAIR tasks"
805 depends on CPU_FREQ_GOV_SCHEDUTIL
807 This feature enables the scheduler to track the clamped utilization
808 of each CPU based on RUNNABLE tasks scheduled on that CPU.
810 With this option, the user can specify the min and max CPU
811 utilization allowed for RUNNABLE tasks. The max utilization defines
812 the maximum frequency a task should use while the min utilization
813 defines the minimum frequency it should use.
815 Both min and max utilization clamp values are hints to the scheduler,
816 aiming at improving its frequency selection policy, but they do not
817 enforce or grant any specific bandwidth for tasks.
821 config UCLAMP_BUCKETS_COUNT
822 int "Number of supported utilization clamp buckets"
825 depends on UCLAMP_TASK
827 Defines the number of clamp buckets to use. The range of each bucket
828 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
829 number of clamp buckets the finer their granularity and the higher
830 the precision of clamping aggregation and tracking at run-time.
832 For example, with the minimum configuration value we will have 5
833 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
834 be refcounted in the [20..39]% bucket and will set the bucket clamp
835 effective value to 25%.
836 If a second 30% boosted task should be co-scheduled on the same CPU,
837 that task will be refcounted in the same bucket of the first task and
838 it will boost the bucket clamp effective value to 30%.
839 The clamp effective value of a bucket is reset to its nominal value
840 (20% in the example above) when there are no more tasks refcounted in
843 An additional boost/capping margin can be added to some tasks. In the
844 example above the 25% task will be boosted to 30% until it exits the
845 CPU. If that should be considered not acceptable on certain systems,
846 it's always possible to reduce the margin by increasing the number of
847 clamp buckets to trade off used memory for run-time tracking
850 If in doubt, use the default value.
855 # For architectures that want to enable the support for NUMA-affine scheduler
858 config ARCH_SUPPORTS_NUMA_BALANCING
862 # For architectures that prefer to flush all TLBs after a number of pages
863 # are unmapped instead of sending one IPI per page to flush. The architecture
864 # must provide guarantees on what happens if a clean TLB cache entry is
865 # written after the unmap. Details are in mm/rmap.c near the check for
866 # should_defer_flush. The architecture should also consider if the full flush
867 # and the refill costs are offset by the savings of sending fewer IPIs.
868 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
872 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
875 # For architectures that know their GCC __int128 support is sound
877 config ARCH_SUPPORTS_INT128
880 # For architectures that (ab)use NUMA to represent different memory regions
881 # all cpu-local but of different latencies, such as SuperH.
883 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
886 config NUMA_BALANCING
887 bool "Memory placement aware NUMA scheduler"
888 depends on ARCH_SUPPORTS_NUMA_BALANCING
889 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
890 depends on SMP && NUMA && MIGRATION
892 This option adds support for automatic NUMA aware memory/task placement.
893 The mechanism is quite primitive and is based on migrating memory when
894 it has references to the node the task is running on.
896 This system will be inactive on UMA systems.
898 config NUMA_BALANCING_DEFAULT_ENABLED
899 bool "Automatically enable NUMA aware memory/task placement"
901 depends on NUMA_BALANCING
903 If set, automatic NUMA balancing will be enabled if running on a NUMA
907 bool "Control Group support"
910 This option adds support for grouping sets of processes together, for
911 use with process control subsystems such as Cpusets, CFS, memory
912 controls or device isolation.
914 - Documentation/scheduler/sched-design-CFS.rst (CFS)
915 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
916 and resource control)
926 bool "Memory controller"
930 Provides control over the memory footprint of tasks in a cgroup.
934 depends on MEMCG && SWAP
939 depends on MEMCG && !SLOB
947 Generic block IO controller cgroup interface. This is the common
948 cgroup interface which should be used by various IO controlling
951 Currently, CFQ IO scheduler uses it to recognize task groups and
952 control disk bandwidth allocation (proportional time slice allocation)
953 to such task groups. It is also used by bio throttling logic in
954 block layer to implement upper limit in IO rates on a device.
956 This option only enables generic Block IO controller infrastructure.
957 One needs to also enable actual IO controlling logic/policy. For
958 enabling proportional weight division of disk bandwidth in CFQ, set
959 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
960 CONFIG_BLK_DEV_THROTTLING=y.
962 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
964 config CGROUP_WRITEBACK
966 depends on MEMCG && BLK_CGROUP
969 menuconfig CGROUP_SCHED
970 bool "CPU controller"
973 This feature lets CPU scheduler recognize task groups and control CPU
974 bandwidth allocation to such task groups. It uses cgroups to group
978 config FAIR_GROUP_SCHED
979 bool "Group scheduling for SCHED_OTHER"
980 depends on CGROUP_SCHED
984 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
985 depends on FAIR_GROUP_SCHED
988 This option allows users to define CPU bandwidth rates (limits) for
989 tasks running within the fair group scheduler. Groups with no limit
990 set are considered to be unconstrained and will run with no
992 See Documentation/scheduler/sched-bwc.rst for more information.
994 config RT_GROUP_SCHED
995 bool "Group scheduling for SCHED_RR/FIFO"
996 depends on CGROUP_SCHED
999 This feature lets you explicitly allocate real CPU bandwidth
1000 to task groups. If enabled, it will also make it impossible to
1001 schedule realtime tasks for non-root users until you allocate
1002 realtime bandwidth for them.
1003 See Documentation/scheduler/sched-rt-group.rst for more information.
1007 config UCLAMP_TASK_GROUP
1008 bool "Utilization clamping per group of tasks"
1009 depends on CGROUP_SCHED
1010 depends on UCLAMP_TASK
1013 This feature enables the scheduler to track the clamped utilization
1014 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
1016 When this option is enabled, the user can specify a min and max
1017 CPU bandwidth which is allowed for each single task in a group.
1018 The max bandwidth allows to clamp the maximum frequency a task
1019 can use, while the min bandwidth allows to define a minimum
1020 frequency a task will always use.
1022 When task group based utilization clamping is enabled, an eventually
1023 specified task-specific clamp value is constrained by the cgroup
1024 specified clamp value. Both minimum and maximum task clamping cannot
1025 be bigger than the corresponding clamping defined at task group level.
1030 bool "PIDs controller"
1032 Provides enforcement of process number limits in the scope of a
1033 cgroup. Any attempt to fork more processes than is allowed in the
1034 cgroup will fail. PIDs are fundamentally a global resource because it
1035 is fairly trivial to reach PID exhaustion before you reach even a
1036 conservative kmemcg limit. As a result, it is possible to grind a
1037 system to halt without being limited by other cgroup policies. The
1038 PIDs controller is designed to stop this from happening.
1040 It should be noted that organisational operations (such as attaching
1041 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1042 since the PIDs limit only affects a process's ability to fork, not to
1046 bool "RDMA controller"
1048 Provides enforcement of RDMA resources defined by IB stack.
1049 It is fairly easy for consumers to exhaust RDMA resources, which
1050 can result into resource unavailability to other consumers.
1051 RDMA controller is designed to stop this from happening.
1052 Attaching processes with active RDMA resources to the cgroup
1053 hierarchy is allowed even if can cross the hierarchy's limit.
1055 config CGROUP_FREEZER
1056 bool "Freezer controller"
1058 Provides a way to freeze and unfreeze all tasks in a
1061 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1062 controller includes important in-kernel memory consumers per default.
1064 If you're using cgroup2, say N.
1066 config CGROUP_HUGETLB
1067 bool "HugeTLB controller"
1068 depends on HUGETLB_PAGE
1072 Provides a cgroup controller for HugeTLB pages.
1073 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1074 The limit is enforced during page fault. Since HugeTLB doesn't
1075 support page reclaim, enforcing the limit at page fault time implies
1076 that, the application will get SIGBUS signal if it tries to access
1077 HugeTLB pages beyond its limit. This requires the application to know
1078 beforehand how much HugeTLB pages it would require for its use. The
1079 control group is tracked in the third page lru pointer. This means
1080 that we cannot use the controller with huge page less than 3 pages.
1083 bool "Cpuset controller"
1086 This option will let you create and manage CPUSETs which
1087 allow dynamically partitioning a system into sets of CPUs and
1088 Memory Nodes and assigning tasks to run only within those sets.
1089 This is primarily useful on large SMP or NUMA systems.
1093 config PROC_PID_CPUSET
1094 bool "Include legacy /proc/<pid>/cpuset file"
1098 config CGROUP_DEVICE
1099 bool "Device controller"
1101 Provides a cgroup controller implementing whitelists for
1102 devices which a process in the cgroup can mknod or open.
1104 config CGROUP_CPUACCT
1105 bool "Simple CPU accounting controller"
1107 Provides a simple controller for monitoring the
1108 total CPU consumed by the tasks in a cgroup.
1111 bool "Perf controller"
1112 depends on PERF_EVENTS
1114 This option extends the perf per-cpu mode to restrict monitoring
1115 to threads which belong to the cgroup specified and run on the
1116 designated cpu. Or this can be used to have cgroup ID in samples
1117 so that it can monitor performance events among cgroups.
1122 bool "Support for eBPF programs attached to cgroups"
1123 depends on BPF_SYSCALL
1124 select SOCK_CGROUP_DATA
1126 Allow attaching eBPF programs to a cgroup using the bpf(2)
1127 syscall command BPF_PROG_ATTACH.
1129 In which context these programs are accessed depends on the type
1130 of attachment. For instance, programs that are attached using
1131 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1135 bool "Misc resource controller"
1138 Provides a controller for miscellaneous resources on a host.
1140 Miscellaneous scalar resources are the resources on the host system
1141 which cannot be abstracted like the other cgroups. This controller
1142 tracks and limits the miscellaneous resources used by a process
1143 attached to a cgroup hierarchy.
1145 For more information, please check misc cgroup section in
1146 /Documentation/admin-guide/cgroup-v2.rst.
1149 bool "Debug controller"
1151 depends on DEBUG_KERNEL
1153 This option enables a simple controller that exports
1154 debugging information about the cgroups framework. This
1155 controller is for control cgroup debugging only. Its
1156 interfaces are not stable.
1160 config SOCK_CGROUP_DATA
1166 menuconfig NAMESPACES
1167 bool "Namespaces support" if EXPERT
1168 depends on MULTIUSER
1171 Provides the way to make tasks work with different objects using
1172 the same id. For example same IPC id may refer to different objects
1173 or same user id or pid may refer to different tasks when used in
1174 different namespaces.
1179 bool "UTS namespace"
1182 In this namespace tasks see different info provided with the
1186 bool "TIME namespace"
1187 depends on GENERIC_VDSO_TIME_NS
1190 In this namespace boottime and monotonic clocks can be set.
1191 The time will keep going with the same pace.
1194 bool "IPC namespace"
1195 depends on (SYSVIPC || POSIX_MQUEUE)
1198 In this namespace tasks work with IPC ids which correspond to
1199 different IPC objects in different namespaces.
1202 bool "User namespace"
1205 This allows containers, i.e. vservers, to use user namespaces
1206 to provide different user info for different servers.
1208 When user namespaces are enabled in the kernel it is
1209 recommended that the MEMCG option also be enabled and that
1210 user-space use the memory control groups to limit the amount
1211 of memory a memory unprivileged users can use.
1216 bool "PID Namespaces"
1219 Support process id namespaces. This allows having multiple
1220 processes with the same pid as long as they are in different
1221 pid namespaces. This is a building block of containers.
1224 bool "Network namespace"
1228 Allow user space to create what appear to be multiple instances
1229 of the network stack.
1233 config CHECKPOINT_RESTORE
1234 bool "Checkpoint/restore support"
1235 select PROC_CHILDREN
1239 Enables additional kernel features in a sake of checkpoint/restore.
1240 In particular it adds auxiliary prctl codes to setup process text,
1241 data and heap segment sizes, and a few additional /proc filesystem
1244 If unsure, say N here.
1246 config SCHED_AUTOGROUP
1247 bool "Automatic process group scheduling"
1250 select FAIR_GROUP_SCHED
1252 This option optimizes the scheduler for common desktop workloads by
1253 automatically creating and populating task groups. This separation
1254 of workloads isolates aggressive CPU burners (like build jobs) from
1255 desktop applications. Task group autogeneration is currently based
1258 config SYSFS_DEPRECATED
1259 bool "Enable deprecated sysfs features to support old userspace tools"
1263 This option adds code that switches the layout of the "block" class
1264 devices, to not show up in /sys/class/block/, but only in
1267 This switch is only active when the sysfs.deprecated=1 boot option is
1268 passed or the SYSFS_DEPRECATED_V2 option is set.
1270 This option allows new kernels to run on old distributions and tools,
1271 which might get confused by /sys/class/block/. Since 2007/2008 all
1272 major distributions and tools handle this just fine.
1274 Recent distributions and userspace tools after 2009/2010 depend on
1275 the existence of /sys/class/block/, and will not work with this
1278 Only if you are using a new kernel on an old distribution, you might
1281 config SYSFS_DEPRECATED_V2
1282 bool "Enable deprecated sysfs features by default"
1285 depends on SYSFS_DEPRECATED
1287 Enable deprecated sysfs by default.
1289 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1292 Only if you are using a new kernel on an old distribution, you might
1293 need to say Y here. Even then, odds are you would not need it
1294 enabled, you can always pass the boot option if absolutely necessary.
1297 bool "Kernel->user space relay support (formerly relayfs)"
1300 This option enables support for relay interface support in
1301 certain file systems (such as debugfs).
1302 It is designed to provide an efficient mechanism for tools and
1303 facilities to relay large amounts of data from kernel space to
1308 config BLK_DEV_INITRD
1309 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1311 The initial RAM filesystem is a ramfs which is loaded by the
1312 boot loader (loadlin or lilo) and that is mounted as root
1313 before the normal boot procedure. It is typically used to
1314 load modules needed to mount the "real" root file system,
1315 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1317 If RAM disk support (BLK_DEV_RAM) is also included, this
1318 also enables initial RAM disk (initrd) support and adds
1319 15 Kbytes (more on some other architectures) to the kernel size.
1325 source "usr/Kconfig"
1330 bool "Boot config support"
1331 select BLK_DEV_INITRD
1333 Extra boot config allows system admin to pass a config file as
1334 complemental extension of kernel cmdline when booting.
1335 The boot config file must be attached at the end of initramfs
1336 with checksum, size and magic word.
1337 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1342 prompt "Compiler optimization level"
1343 default CC_OPTIMIZE_FOR_PERFORMANCE
1345 config CC_OPTIMIZE_FOR_PERFORMANCE
1346 bool "Optimize for performance (-O2)"
1348 This is the default optimization level for the kernel, building
1349 with the "-O2" compiler flag for best performance and most
1350 helpful compile-time warnings.
1352 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1353 bool "Optimize more for performance (-O3)"
1356 Choosing this option will pass "-O3" to your compiler to optimize
1357 the kernel yet more for performance.
1359 config CC_OPTIMIZE_FOR_SIZE
1360 bool "Optimize for size (-Os)"
1362 Choosing this option will pass "-Os" to your compiler resulting
1363 in a smaller kernel.
1367 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1370 This requires that the arch annotates or otherwise protects
1371 its external entry points from being discarded. Linker scripts
1372 must also merge .text.*, .data.*, and .bss.* correctly into
1373 output sections. Care must be taken not to pull in unrelated
1374 sections (e.g., '.text.init'). Typically '.' in section names
1375 is used to distinguish them from label names / C identifiers.
1377 config LD_DEAD_CODE_DATA_ELIMINATION
1378 bool "Dead code and data elimination (EXPERIMENTAL)"
1379 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1381 depends on $(cc-option,-ffunction-sections -fdata-sections)
1382 depends on $(ld-option,--gc-sections)
1384 Enable this if you want to do dead code and data elimination with
1385 the linker by compiling with -ffunction-sections -fdata-sections,
1386 and linking with --gc-sections.
1388 This can reduce on disk and in-memory size of the kernel
1389 code and static data, particularly for small configs and
1390 on small systems. This has the possibility of introducing
1391 silently broken kernel if the required annotations are not
1392 present. This option is not well tested yet, so use at your
1395 config LD_ORPHAN_WARN
1397 depends on ARCH_WANT_LD_ORPHAN_WARN
1398 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1399 depends on $(ld-option,--orphan-handling=warn)
1407 config SYSCTL_EXCEPTION_TRACE
1410 Enable support for /proc/sys/debug/exception-trace.
1412 config SYSCTL_ARCH_UNALIGN_NO_WARN
1415 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1416 Allows arch to define/use @no_unaligned_warning to possibly warn
1417 about unaligned access emulation going on under the hood.
1419 config SYSCTL_ARCH_UNALIGN_ALLOW
1422 Enable support for /proc/sys/kernel/unaligned-trap
1423 Allows arches to define/use @unaligned_enabled to runtime toggle
1424 the unaligned access emulation.
1425 see arch/parisc/kernel/unaligned.c for reference
1427 config HAVE_PCSPKR_PLATFORM
1430 # interpreter that classic socket filters depend on
1435 bool "Configure standard kernel features (expert users)"
1436 # Unhide debug options, to make the on-by-default options visible
1439 This option allows certain base kernel options and settings
1440 to be disabled or tweaked. This is for specialized
1441 environments which can tolerate a "non-standard" kernel.
1442 Only use this if you really know what you are doing.
1445 bool "Enable 16-bit UID system calls" if EXPERT
1446 depends on HAVE_UID16 && MULTIUSER
1449 This enables the legacy 16-bit UID syscall wrappers.
1452 bool "Multiple users, groups and capabilities support" if EXPERT
1455 This option enables support for non-root users, groups and
1458 If you say N here, all processes will run with UID 0, GID 0, and all
1459 possible capabilities. Saying N here also compiles out support for
1460 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1463 If unsure, say Y here.
1465 config SGETMASK_SYSCALL
1466 bool "sgetmask/ssetmask syscalls support" if EXPERT
1467 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1469 sys_sgetmask and sys_ssetmask are obsolete system calls
1470 no longer supported in libc but still enabled by default in some
1473 If unsure, leave the default option here.
1475 config SYSFS_SYSCALL
1476 bool "Sysfs syscall support" if EXPERT
1479 sys_sysfs is an obsolete system call no longer supported in libc.
1480 Note that disabling this option is more secure but might break
1481 compatibility with some systems.
1483 If unsure say Y here.
1486 bool "open by fhandle syscalls" if EXPERT
1490 If you say Y here, a user level program will be able to map
1491 file names to handle and then later use the handle for
1492 different file system operations. This is useful in implementing
1493 userspace file servers, which now track files using handles instead
1494 of names. The handle would remain the same even if file names
1495 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1499 bool "Posix Clocks & timers" if EXPERT
1502 This includes native support for POSIX timers to the kernel.
1503 Some embedded systems have no use for them and therefore they
1504 can be configured out to reduce the size of the kernel image.
1506 When this option is disabled, the following syscalls won't be
1507 available: timer_create, timer_gettime: timer_getoverrun,
1508 timer_settime, timer_delete, clock_adjtime, getitimer,
1509 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1510 clock_getres and clock_nanosleep syscalls will be limited to
1511 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1517 bool "Enable support for printk" if EXPERT
1520 This option enables normal printk support. Removing it
1521 eliminates most of the message strings from the kernel image
1522 and makes the kernel more or less silent. As this makes it
1523 very difficult to diagnose system problems, saying N here is
1524 strongly discouraged.
1532 bool "BUG() support" if EXPERT
1535 Disabling this option eliminates support for BUG and WARN, reducing
1536 the size of your kernel image and potentially quietly ignoring
1537 numerous fatal conditions. You should only consider disabling this
1538 option for embedded systems with no facilities for reporting errors.
1544 bool "Enable ELF core dumps" if EXPERT
1546 Enable support for generating core dumps. Disabling saves about 4k.
1549 config PCSPKR_PLATFORM
1550 bool "Enable PC-Speaker support" if EXPERT
1551 depends on HAVE_PCSPKR_PLATFORM
1555 This option allows to disable the internal PC-Speaker
1556 support, saving some memory.
1560 bool "Enable full-sized data structures for core" if EXPERT
1562 Disabling this option reduces the size of miscellaneous core
1563 kernel data structures. This saves memory on small machines,
1564 but may reduce performance.
1567 bool "Enable futex support" if EXPERT
1571 Disabling this option will cause the kernel to be built without
1572 support for "fast userspace mutexes". The resulting kernel may not
1573 run glibc-based applications correctly.
1577 depends on FUTEX && RT_MUTEXES
1580 config HAVE_FUTEX_CMPXCHG
1584 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1585 is implemented and always working. This removes a couple of runtime
1589 bool "Enable eventpoll support" if EXPERT
1592 Disabling this option will cause the kernel to be built without
1593 support for epoll family of system calls.
1596 bool "Enable signalfd() system call" if EXPERT
1599 Enable the signalfd() system call that allows to receive signals
1600 on a file descriptor.
1605 bool "Enable timerfd() system call" if EXPERT
1608 Enable the timerfd() system call that allows to receive timer
1609 events on a file descriptor.
1614 bool "Enable eventfd() system call" if EXPERT
1617 Enable the eventfd() system call that allows to receive both
1618 kernel notification (ie. KAIO) or userspace notifications.
1623 bool "Use full shmem filesystem" if EXPERT
1627 The shmem is an internal filesystem used to manage shared memory.
1628 It is backed by swap and manages resource limits. It is also exported
1629 to userspace as tmpfs if TMPFS is enabled. Disabling this
1630 option replaces shmem and tmpfs with the much simpler ramfs code,
1631 which may be appropriate on small systems without swap.
1634 bool "Enable AIO support" if EXPERT
1637 This option enables POSIX asynchronous I/O which may by used
1638 by some high performance threaded applications. Disabling
1639 this option saves about 7k.
1642 bool "Enable IO uring support" if EXPERT
1646 This option enables support for the io_uring interface, enabling
1647 applications to submit and complete IO through submission and
1648 completion rings that are shared between the kernel and application.
1650 config ADVISE_SYSCALLS
1651 bool "Enable madvise/fadvise syscalls" if EXPERT
1654 This option enables the madvise and fadvise syscalls, used by
1655 applications to advise the kernel about their future memory or file
1656 usage, improving performance. If building an embedded system where no
1657 applications use these syscalls, you can disable this option to save
1660 config HAVE_ARCH_USERFAULTFD_WP
1663 Arch has userfaultfd write protection support
1665 config HAVE_ARCH_USERFAULTFD_MINOR
1668 Arch has userfaultfd minor fault support
1671 bool "Enable membarrier() system call" if EXPERT
1674 Enable the membarrier() system call that allows issuing memory
1675 barriers across all running threads, which can be used to distribute
1676 the cost of user-space memory barriers asymmetrically by transforming
1677 pairs of memory barriers into pairs consisting of membarrier() and a
1683 bool "Load all symbols for debugging/ksymoops" if EXPERT
1686 Say Y here to let the kernel print out symbolic crash information and
1687 symbolic stack backtraces. This increases the size of the kernel
1688 somewhat, as all symbols have to be loaded into the kernel image.
1691 bool "Include all symbols in kallsyms"
1692 depends on DEBUG_KERNEL && KALLSYMS
1694 Normally kallsyms only contains the symbols of functions for nicer
1695 OOPS messages and backtraces (i.e., symbols from the text and inittext
1696 sections). This is sufficient for most cases. And only in very rare
1697 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1698 names of variables from the data sections, etc).
1700 This option makes sure that all symbols are loaded into the kernel
1701 image (i.e., symbols from all sections) in cost of increased kernel
1702 size (depending on the kernel configuration, it may be 300KiB or
1703 something like this).
1705 Say N unless you really need all symbols.
1707 config KALLSYMS_ABSOLUTE_PERCPU
1710 default X86_64 && SMP
1712 config KALLSYMS_BASE_RELATIVE
1717 Instead of emitting them as absolute values in the native word size,
1718 emit the symbol references in the kallsyms table as 32-bit entries,
1719 each containing a relative value in the range [base, base + U32_MAX]
1720 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1721 an absolute value in the range [0, S32_MAX] or a relative value in the
1722 range [base, base + S32_MAX], where base is the lowest relative symbol
1723 address encountered in the image.
1725 On 64-bit builds, this reduces the size of the address table by 50%,
1726 but more importantly, it results in entries whose values are build
1727 time constants, and no relocation pass is required at runtime to fix
1728 up the entries based on the runtime load address of the kernel.
1730 # end of the "standard kernel features (expert users)" menu
1732 # syscall, maps, verifier
1735 bool "Enable userfaultfd() system call"
1738 Enable the userfaultfd() system call that allows to intercept and
1739 handle page faults in userland.
1741 config ARCH_HAS_MEMBARRIER_CALLBACKS
1744 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1748 bool "Enable kcmp() system call" if EXPERT
1750 Enable the kernel resource comparison system call. It provides
1751 user-space with the ability to compare two processes to see if they
1752 share a common resource, such as a file descriptor or even virtual
1758 bool "Enable rseq() system call" if EXPERT
1760 depends on HAVE_RSEQ
1763 Enable the restartable sequences system call. It provides a
1764 user-space cache for the current CPU number value, which
1765 speeds up getting the current CPU number from user-space,
1766 as well as an ABI to speed up user-space operations on
1773 bool "Enabled debugging of rseq() system call" if EXPERT
1774 depends on RSEQ && DEBUG_KERNEL
1776 Enable extra debugging checks for the rseq system call.
1781 bool "Embedded system"
1784 This option should be enabled if compiling the kernel for
1785 an embedded system so certain expert options are available
1788 config HAVE_PERF_EVENTS
1791 See tools/perf/design.txt for details.
1793 config PERF_USE_VMALLOC
1796 See tools/perf/design.txt for details
1799 bool "PC/104 support" if EXPERT
1801 Expose PC/104 form factor device drivers and options available for
1802 selection and configuration. Enable this option if your target
1803 machine has a PC/104 bus.
1805 menu "Kernel Performance Events And Counters"
1808 bool "Kernel performance events and counters"
1809 default y if PROFILING
1810 depends on HAVE_PERF_EVENTS
1814 Enable kernel support for various performance events provided
1815 by software and hardware.
1817 Software events are supported either built-in or via the
1818 use of generic tracepoints.
1820 Most modern CPUs support performance events via performance
1821 counter registers. These registers count the number of certain
1822 types of hw events: such as instructions executed, cachemisses
1823 suffered, or branches mis-predicted - without slowing down the
1824 kernel or applications. These registers can also trigger interrupts
1825 when a threshold number of events have passed - and can thus be
1826 used to profile the code that runs on that CPU.
1828 The Linux Performance Event subsystem provides an abstraction of
1829 these software and hardware event capabilities, available via a
1830 system call and used by the "perf" utility in tools/perf/. It
1831 provides per task and per CPU counters, and it provides event
1832 capabilities on top of those.
1836 config DEBUG_PERF_USE_VMALLOC
1838 bool "Debug: use vmalloc to back perf mmap() buffers"
1839 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1840 select PERF_USE_VMALLOC
1842 Use vmalloc memory to back perf mmap() buffers.
1844 Mostly useful for debugging the vmalloc code on platforms
1845 that don't require it.
1851 config VM_EVENT_COUNTERS
1853 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1855 VM event counters are needed for event counts to be shown.
1856 This option allows the disabling of the VM event counters
1857 on EXPERT systems. /proc/vmstat will only show page counts
1858 if VM event counters are disabled.
1862 bool "Enable SLUB debugging support" if EXPERT
1863 depends on SLUB && SYSFS
1864 select STACKDEPOT if STACKTRACE_SUPPORT
1866 SLUB has extensive debug support features. Disabling these can
1867 result in significant savings in code size. This also disables
1868 SLUB sysfs support. /sys/slab will not exist and there will be
1869 no support for cache validation etc.
1872 bool "Disable heap randomization"
1875 Randomizing heap placement makes heap exploits harder, but it
1876 also breaks ancient binaries (including anything libc5 based).
1877 This option changes the bootup default to heap randomization
1878 disabled, and can be overridden at runtime by setting
1879 /proc/sys/kernel/randomize_va_space to 2.
1881 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1884 prompt "Choose SLAB allocator"
1887 This option allows to select a slab allocator.
1891 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1893 The regular slab allocator that is established and known to work
1894 well in all environments. It organizes cache hot objects in
1895 per cpu and per node queues.
1898 bool "SLUB (Unqueued Allocator)"
1899 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1901 SLUB is a slab allocator that minimizes cache line usage
1902 instead of managing queues of cached objects (SLAB approach).
1903 Per cpu caching is realized using slabs of objects instead
1904 of queues of objects. SLUB can use memory efficiently
1905 and has enhanced diagnostics. SLUB is the default choice for
1910 bool "SLOB (Simple Allocator)"
1912 SLOB replaces the stock allocator with a drastically simpler
1913 allocator. SLOB is generally more space efficient but
1914 does not perform as well on large systems.
1918 config SLAB_MERGE_DEFAULT
1919 bool "Allow slab caches to be merged"
1922 For reduced kernel memory fragmentation, slab caches can be
1923 merged when they share the same size and other characteristics.
1924 This carries a risk of kernel heap overflows being able to
1925 overwrite objects from merged caches (and more easily control
1926 cache layout), which makes such heap attacks easier to exploit
1927 by attackers. By keeping caches unmerged, these kinds of exploits
1928 can usually only damage objects in the same cache. To disable
1929 merging at runtime, "slab_nomerge" can be passed on the kernel
1932 config SLAB_FREELIST_RANDOM
1933 bool "Randomize slab freelist"
1934 depends on SLAB || SLUB
1936 Randomizes the freelist order used on creating new pages. This
1937 security feature reduces the predictability of the kernel slab
1938 allocator against heap overflows.
1940 config SLAB_FREELIST_HARDENED
1941 bool "Harden slab freelist metadata"
1942 depends on SLAB || SLUB
1944 Many kernel heap attacks try to target slab cache metadata and
1945 other infrastructure. This options makes minor performance
1946 sacrifices to harden the kernel slab allocator against common
1947 freelist exploit methods. Some slab implementations have more
1948 sanity-checking than others. This option is most effective with
1951 config SHUFFLE_PAGE_ALLOCATOR
1952 bool "Page allocator randomization"
1953 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1955 Randomization of the page allocator improves the average
1956 utilization of a direct-mapped memory-side-cache. See section
1957 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1958 6.2a specification for an example of how a platform advertises
1959 the presence of a memory-side-cache. There are also incidental
1960 security benefits as it reduces the predictability of page
1961 allocations to compliment SLAB_FREELIST_RANDOM, but the
1962 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1963 10th order of pages is selected based on cache utilization
1966 While the randomization improves cache utilization it may
1967 negatively impact workloads on platforms without a cache. For
1968 this reason, by default, the randomization is enabled only
1969 after runtime detection of a direct-mapped memory-side-cache.
1970 Otherwise, the randomization may be force enabled with the
1971 'page_alloc.shuffle' kernel command line parameter.
1975 config SLUB_CPU_PARTIAL
1977 depends on SLUB && SMP
1978 bool "SLUB per cpu partial cache"
1980 Per cpu partial caches accelerate objects allocation and freeing
1981 that is local to a processor at the price of more indeterminism
1982 in the latency of the free. On overflow these caches will be cleared
1983 which requires the taking of locks that may cause latency spikes.
1984 Typically one would choose no for a realtime system.
1986 config MMAP_ALLOW_UNINITIALIZED
1987 bool "Allow mmapped anonymous memory to be uninitialized"
1988 depends on EXPERT && !MMU
1991 Normally, and according to the Linux spec, anonymous memory obtained
1992 from mmap() has its contents cleared before it is passed to
1993 userspace. Enabling this config option allows you to request that
1994 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1995 providing a huge performance boost. If this option is not enabled,
1996 then the flag will be ignored.
1998 This is taken advantage of by uClibc's malloc(), and also by
1999 ELF-FDPIC binfmt's brk and stack allocator.
2001 Because of the obvious security issues, this option should only be
2002 enabled on embedded devices where you control what is run in
2003 userspace. Since that isn't generally a problem on no-MMU systems,
2004 it is normally safe to say Y here.
2006 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2008 config SYSTEM_DATA_VERIFICATION
2010 select SYSTEM_TRUSTED_KEYRING
2014 select ASYMMETRIC_KEY_TYPE
2015 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2018 select X509_CERTIFICATE_PARSER
2019 select PKCS7_MESSAGE_PARSER
2021 Provide PKCS#7 message verification using the contents of the system
2022 trusted keyring to provide public keys. This then can be used for
2023 module verification, kexec image verification and firmware blob
2027 bool "Profiling support"
2029 Say Y here to enable the extended profiling support mechanisms used
2033 # Place an empty function call at each tracepoint site. Can be
2034 # dynamically changed for a probe function.
2039 endmenu # General setup
2041 source "arch/Kconfig"
2048 default 0 if BASE_FULL
2049 default 1 if !BASE_FULL
2051 config MODULE_SIG_FORMAT
2053 select SYSTEM_DATA_VERIFICATION
2056 bool "Enable loadable module support"
2059 Kernel modules are small pieces of compiled code which can
2060 be inserted in the running kernel, rather than being
2061 permanently built into the kernel. You use the "modprobe"
2062 tool to add (and sometimes remove) them. If you say Y here,
2063 many parts of the kernel can be built as modules (by
2064 answering M instead of Y where indicated): this is most
2065 useful for infrequently used options which are not required
2066 for booting. For more information, see the man pages for
2067 modprobe, lsmod, modinfo, insmod and rmmod.
2069 If you say Y here, you will need to run "make
2070 modules_install" to put the modules under /lib/modules/
2071 where modprobe can find them (you may need to be root to do
2078 config MODULE_FORCE_LOAD
2079 bool "Forced module loading"
2082 Allow loading of modules without version information (ie. modprobe
2083 --force). Forced module loading sets the 'F' (forced) taint flag and
2084 is usually a really bad idea.
2086 config MODULE_UNLOAD
2087 bool "Module unloading"
2089 Without this option you will not be able to unload any
2090 modules (note that some modules may not be unloadable
2091 anyway), which makes your kernel smaller, faster
2092 and simpler. If unsure, say Y.
2094 config MODULE_FORCE_UNLOAD
2095 bool "Forced module unloading"
2096 depends on MODULE_UNLOAD
2098 This option allows you to force a module to unload, even if the
2099 kernel believes it is unsafe: the kernel will remove the module
2100 without waiting for anyone to stop using it (using the -f option to
2101 rmmod). This is mainly for kernel developers and desperate users.
2105 bool "Module versioning support"
2107 Usually, you have to use modules compiled with your kernel.
2108 Saying Y here makes it sometimes possible to use modules
2109 compiled for different kernels, by adding enough information
2110 to the modules to (hopefully) spot any changes which would
2111 make them incompatible with the kernel you are running. If
2114 config ASM_MODVERSIONS
2116 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2118 This enables module versioning for exported symbols also from
2119 assembly. This can be enabled only when the target architecture
2122 config MODULE_REL_CRCS
2124 depends on MODVERSIONS
2126 config MODULE_SRCVERSION_ALL
2127 bool "Source checksum for all modules"
2129 Modules which contain a MODULE_VERSION get an extra "srcversion"
2130 field inserted into their modinfo section, which contains a
2131 sum of the source files which made it. This helps maintainers
2132 see exactly which source was used to build a module (since
2133 others sometimes change the module source without updating
2134 the version). With this option, such a "srcversion" field
2135 will be created for all modules. If unsure, say N.
2138 bool "Module signature verification"
2139 select MODULE_SIG_FORMAT
2141 Check modules for valid signatures upon load: the signature
2142 is simply appended to the module. For more information see
2143 <file:Documentation/admin-guide/module-signing.rst>.
2145 Note that this option adds the OpenSSL development packages as a
2146 kernel build dependency so that the signing tool can use its crypto
2149 You should enable this option if you wish to use either
2150 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2151 another LSM - otherwise unsigned modules will be loadable regardless
2152 of the lockdown policy.
2154 !!!WARNING!!! If you enable this option, you MUST make sure that the
2155 module DOES NOT get stripped after being signed. This includes the
2156 debuginfo strip done by some packagers (such as rpmbuild) and
2157 inclusion into an initramfs that wants the module size reduced.
2159 config MODULE_SIG_FORCE
2160 bool "Require modules to be validly signed"
2161 depends on MODULE_SIG
2163 Reject unsigned modules or signed modules for which we don't have a
2164 key. Without this, such modules will simply taint the kernel.
2166 config MODULE_SIG_ALL
2167 bool "Automatically sign all modules"
2169 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2171 Sign all modules during make modules_install. Without this option,
2172 modules must be signed manually, using the scripts/sign-file tool.
2174 comment "Do not forget to sign required modules with scripts/sign-file"
2175 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2178 prompt "Which hash algorithm should modules be signed with?"
2179 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2181 This determines which sort of hashing algorithm will be used during
2182 signature generation. This algorithm _must_ be built into the kernel
2183 directly so that signature verification can take place. It is not
2184 possible to load a signed module containing the algorithm to check
2185 the signature on that module.
2187 config MODULE_SIG_SHA1
2188 bool "Sign modules with SHA-1"
2191 config MODULE_SIG_SHA224
2192 bool "Sign modules with SHA-224"
2193 select CRYPTO_SHA256
2195 config MODULE_SIG_SHA256
2196 bool "Sign modules with SHA-256"
2197 select CRYPTO_SHA256
2199 config MODULE_SIG_SHA384
2200 bool "Sign modules with SHA-384"
2201 select CRYPTO_SHA512
2203 config MODULE_SIG_SHA512
2204 bool "Sign modules with SHA-512"
2205 select CRYPTO_SHA512
2209 config MODULE_SIG_HASH
2211 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2212 default "sha1" if MODULE_SIG_SHA1
2213 default "sha224" if MODULE_SIG_SHA224
2214 default "sha256" if MODULE_SIG_SHA256
2215 default "sha384" if MODULE_SIG_SHA384
2216 default "sha512" if MODULE_SIG_SHA512
2219 prompt "Module compression mode"
2221 This option allows you to choose the algorithm which will be used to
2222 compress modules when 'make modules_install' is run. (or, you can
2223 choose to not compress modules at all.)
2225 External modules will also be compressed in the same way during the
2228 For modules inside an initrd or initramfs, it's more efficient to
2229 compress the whole initrd or initramfs instead.
2231 This is fully compatible with signed modules.
2233 Please note that the tool used to load modules needs to support the
2234 corresponding algorithm. module-init-tools MAY support gzip, and kmod
2235 MAY support gzip, xz and zstd.
2237 Your build system needs to provide the appropriate compression tool
2238 to compress the modules.
2240 If in doubt, select 'None'.
2242 config MODULE_COMPRESS_NONE
2245 Do not compress modules. The installed modules are suffixed
2248 config MODULE_COMPRESS_GZIP
2251 Compress modules with GZIP. The installed modules are suffixed
2254 config MODULE_COMPRESS_XZ
2257 Compress modules with XZ. The installed modules are suffixed
2260 config MODULE_COMPRESS_ZSTD
2263 Compress modules with ZSTD. The installed modules are suffixed
2268 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2269 bool "Allow loading of modules with missing namespace imports"
2271 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2272 a namespace. A module that makes use of a symbol exported with such a
2273 namespace is required to import the namespace via MODULE_IMPORT_NS().
2274 There is no technical reason to enforce correct namespace imports,
2275 but it creates consistency between symbols defining namespaces and
2276 users importing namespaces they make use of. This option relaxes this
2277 requirement and lifts the enforcement when loading a module.
2281 config MODPROBE_PATH
2282 string "Path to modprobe binary"
2283 default "/sbin/modprobe"
2285 When kernel code requests a module, it does so by calling
2286 the "modprobe" userspace utility. This option allows you to
2287 set the path where that binary is found. This can be changed
2288 at runtime via the sysctl file
2289 /proc/sys/kernel/modprobe. Setting this to the empty string
2290 removes the kernel's ability to request modules (but
2291 userspace can still load modules explicitly).
2293 config TRIM_UNUSED_KSYMS
2294 bool "Trim unused exported kernel symbols" if EXPERT
2295 depends on !COMPILE_TEST
2297 The kernel and some modules make many symbols available for
2298 other modules to use via EXPORT_SYMBOL() and variants. Depending
2299 on the set of modules being selected in your kernel configuration,
2300 many of those exported symbols might never be used.
2302 This option allows for unused exported symbols to be dropped from
2303 the build. In turn, this provides the compiler more opportunities
2304 (especially when using LTO) for optimizing the code and reducing
2305 binary size. This might have some security advantages as well.
2307 If unsure, or if you need to build out-of-tree modules, say N.
2309 config UNUSED_KSYMS_WHITELIST
2310 string "Whitelist of symbols to keep in ksymtab"
2311 depends on TRIM_UNUSED_KSYMS
2313 By default, all unused exported symbols will be un-exported from the
2314 build when TRIM_UNUSED_KSYMS is selected.
2316 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2317 exported at all times, even in absence of in-tree users. The value to
2318 set here is the path to a text file containing the list of symbols,
2319 one per line. The path can be absolute, or relative to the kernel
2324 config MODULES_TREE_LOOKUP
2326 depends on PERF_EVENTS || TRACING || CFI_CLANG
2328 config INIT_ALL_POSSIBLE
2331 Back when each arch used to define their own cpu_online_mask and
2332 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2333 with all 1s, and others with all 0s. When they were centralised,
2334 it was better to provide this option than to break all the archs
2335 and have several arch maintainers pursuing me down dark alleys.
2337 source "block/Kconfig"
2339 config PREEMPT_NOTIFIERS
2349 Build a simple ASN.1 grammar compiler that produces a bytecode output
2350 that can be interpreted by the ASN.1 stream decoder and used to
2351 inform it as to what tags are to be expected in a stream and what
2352 functions to call on what tags.
2354 source "kernel/Kconfig.locks"
2356 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2359 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2362 # It may be useful for an architecture to override the definitions of the
2363 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2364 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2365 # different calling convention for syscalls. They can also override the
2366 # macros for not-implemented syscalls in kernel/sys_ni.c and
2367 # kernel/time/posix-stubs.c. All these overrides need to be available in
2368 # <asm/syscall_wrapper.h>.
2369 config ARCH_HAS_SYSCALL_WRAPPER
projects / linux.git / blob