1 # SPDX-License-Identifier: GPL-2.0-only
6 default "/lib/modules/$(shell,uname -r)/.config"
7 default "/etc/kernel-config"
8 default "/boot/config-$(shell,uname -r)"
10 default "arch/$(ARCH)/defconfig"
13 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
17 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
21 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
25 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
28 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
30 config CC_HAS_ASM_GOTO
31 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
33 config TOOLS_SUPPORT_RELR
34 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
36 config CC_HAS_ASM_INLINE
37 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
39 config CC_HAS_WARN_MAYBE_UNINITIALIZED
40 def_bool $(cc-option,-Wmaybe-uninitialized)
42 GCC >= 4.7 supports this option.
44 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
46 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
47 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
49 GCC's -Wmaybe-uninitialized is not reliable by definition.
50 Lots of false positive warnings are produced in some cases.
52 If this option is enabled, -Wno-maybe-uninitialzed is passed
53 to the compiler to suppress maybe-uninitialized warnings.
62 config BUILDTIME_TABLE_SORT
65 config THREAD_INFO_IN_TASK
68 Select this to move thread_info off the stack into task_struct. To
69 make this work, an arch will need to remove all thread_info fields
70 except flags and fix any runtime bugs.
72 One subtle change that will be needed is to use try_get_task_stack()
73 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
82 depends on BROKEN || !SMP
85 config INIT_ENV_ARG_LIMIT
90 Maximum of each of the number of arguments and environment
91 variables passed to init from the kernel command line.
94 bool "Compile also drivers which will not load"
98 Some drivers can be compiled on a different platform than they are
99 intended to be run on. Despite they cannot be loaded there (or even
100 when they load they cannot be used due to missing HW support),
101 developers still, opposing to distributors, might want to build such
102 drivers to compile-test them.
104 If you are a developer and want to build everything available, say Y
105 here. If you are a user/distributor, say N here to exclude useless
106 drivers to be distributed.
108 config UAPI_HEADER_TEST
109 bool "Compile test UAPI headers"
110 depends on HEADERS_INSTALL && CC_CAN_LINK
112 Compile test headers exported to user-space to ensure they are
113 self-contained, i.e. compilable as standalone units.
115 If you are a developer or tester and want to ensure the exported
116 headers are self-contained, say Y here. Otherwise, choose N.
119 string "Local version - append to kernel release"
121 Append an extra string to the end of your kernel version.
122 This will show up when you type uname, for example.
123 The string you set here will be appended after the contents of
124 any files with a filename matching localversion* in your
125 object and source tree, in that order. Your total string can
126 be a maximum of 64 characters.
128 config LOCALVERSION_AUTO
129 bool "Automatically append version information to the version string"
131 depends on !COMPILE_TEST
133 This will try to automatically determine if the current tree is a
134 release tree by looking for git tags that belong to the current
135 top of tree revision.
137 A string of the format -gxxxxxxxx will be added to the localversion
138 if a git-based tree is found. The string generated by this will be
139 appended after any matching localversion* files, and after the value
140 set in CONFIG_LOCALVERSION.
142 (The actual string used here is the first eight characters produced
143 by running the command:
145 $ git rev-parse --verify HEAD
147 which is done within the script "scripts/setlocalversion".)
150 string "Build ID Salt"
153 The build ID is used to link binaries and their debug info. Setting
154 this option will use the value in the calculation of the build id.
155 This is mostly useful for distributions which want to ensure the
156 build is unique between builds. It's safe to leave the default.
158 config HAVE_KERNEL_GZIP
161 config HAVE_KERNEL_BZIP2
164 config HAVE_KERNEL_LZMA
167 config HAVE_KERNEL_XZ
170 config HAVE_KERNEL_LZO
173 config HAVE_KERNEL_LZ4
176 config HAVE_KERNEL_UNCOMPRESSED
180 prompt "Kernel compression mode"
182 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
184 The linux kernel is a kind of self-extracting executable.
185 Several compression algorithms are available, which differ
186 in efficiency, compression and decompression speed.
187 Compression speed is only relevant when building a kernel.
188 Decompression speed is relevant at each boot.
190 If you have any problems with bzip2 or lzma compressed
191 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
192 version of this functionality (bzip2 only), for 2.4, was
193 supplied by Christian Ludwig)
195 High compression options are mostly useful for users, who
196 are low on disk space (embedded systems), but for whom ram
199 If in doubt, select 'gzip'
203 depends on HAVE_KERNEL_GZIP
205 The old and tried gzip compression. It provides a good balance
206 between compression ratio and decompression speed.
210 depends on HAVE_KERNEL_BZIP2
212 Its compression ratio and speed is intermediate.
213 Decompression speed is slowest among the choices. The kernel
214 size is about 10% smaller with bzip2, in comparison to gzip.
215 Bzip2 uses a large amount of memory. For modern kernels you
216 will need at least 8MB RAM or more for booting.
220 depends on HAVE_KERNEL_LZMA
222 This compression algorithm's ratio is best. Decompression speed
223 is between gzip and bzip2. Compression is slowest.
224 The kernel size is about 33% smaller with LZMA in comparison to gzip.
228 depends on HAVE_KERNEL_XZ
230 XZ uses the LZMA2 algorithm and instruction set specific
231 BCJ filters which can improve compression ratio of executable
232 code. The size of the kernel is about 30% smaller with XZ in
233 comparison to gzip. On architectures for which there is a BCJ
234 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
235 will create a few percent smaller kernel than plain LZMA.
237 The speed is about the same as with LZMA: The decompression
238 speed of XZ is better than that of bzip2 but worse than gzip
239 and LZO. Compression is slow.
243 depends on HAVE_KERNEL_LZO
245 Its compression ratio is the poorest among the choices. The kernel
246 size is about 10% bigger than gzip; however its speed
247 (both compression and decompression) is the fastest.
251 depends on HAVE_KERNEL_LZ4
253 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
254 A preliminary version of LZ4 de/compression tool is available at
255 <https://code.google.com/p/lz4/>.
257 Its compression ratio is worse than LZO. The size of the kernel
258 is about 8% bigger than LZO. But the decompression speed is
261 config KERNEL_UNCOMPRESSED
263 depends on HAVE_KERNEL_UNCOMPRESSED
265 Produce uncompressed kernel image. This option is usually not what
266 you want. It is useful for debugging the kernel in slow simulation
267 environments, where decompressing and moving the kernel is awfully
268 slow. This option allows early boot code to skip the decompressor
269 and jump right at uncompressed kernel image.
273 config DEFAULT_HOSTNAME
274 string "Default hostname"
277 This option determines the default system hostname before userspace
278 calls sethostname(2). The kernel traditionally uses "(none)" here,
279 but you may wish to use a different default here to make a minimal
280 system more usable with less configuration.
283 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
284 # add proper SWAP support to them, in which case this can be remove.
290 bool "Support for paging of anonymous memory (swap)"
291 depends on MMU && BLOCK && !ARCH_NO_SWAP
294 This option allows you to choose whether you want to have support
295 for so called swap devices or swap files in your kernel that are
296 used to provide more virtual memory than the actual RAM present
297 in your computer. If unsure say Y.
302 Inter Process Communication is a suite of library functions and
303 system calls which let processes (running programs) synchronize and
304 exchange information. It is generally considered to be a good thing,
305 and some programs won't run unless you say Y here. In particular, if
306 you want to run the DOS emulator dosemu under Linux (read the
307 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
308 you'll need to say Y here.
310 You can find documentation about IPC with "info ipc" and also in
311 section 6.4 of the Linux Programmer's Guide, available from
312 <http://www.tldp.org/guides.html>.
314 config SYSVIPC_SYSCTL
321 bool "POSIX Message Queues"
324 POSIX variant of message queues is a part of IPC. In POSIX message
325 queues every message has a priority which decides about succession
326 of receiving it by a process. If you want to compile and run
327 programs written e.g. for Solaris with use of its POSIX message
328 queues (functions mq_*) say Y here.
330 POSIX message queues are visible as a filesystem called 'mqueue'
331 and can be mounted somewhere if you want to do filesystem
332 operations on message queues.
336 config POSIX_MQUEUE_SYSCTL
338 depends on POSIX_MQUEUE
342 config CROSS_MEMORY_ATTACH
343 bool "Enable process_vm_readv/writev syscalls"
347 Enabling this option adds the system calls process_vm_readv and
348 process_vm_writev which allow a process with the correct privileges
349 to directly read from or write to another process' address space.
350 See the man page for more details.
353 bool "uselib syscall"
354 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
356 This option enables the uselib syscall, a system call used in the
357 dynamic linker from libc5 and earlier. glibc does not use this
358 system call. If you intend to run programs built on libc5 or
359 earlier, you may need to enable this syscall. Current systems
360 running glibc can safely disable this.
363 bool "Auditing support"
366 Enable auditing infrastructure that can be used with another
367 kernel subsystem, such as SELinux (which requires this for
368 logging of avc messages output). System call auditing is included
369 on architectures which support it.
371 config HAVE_ARCH_AUDITSYSCALL
376 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
379 source "kernel/irq/Kconfig"
380 source "kernel/time/Kconfig"
381 source "kernel/Kconfig.preempt"
383 menu "CPU/Task time and stats accounting"
385 config VIRT_CPU_ACCOUNTING
389 prompt "Cputime accounting"
390 default TICK_CPU_ACCOUNTING if !PPC64
391 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
393 # Kind of a stub config for the pure tick based cputime accounting
394 config TICK_CPU_ACCOUNTING
395 bool "Simple tick based cputime accounting"
396 depends on !S390 && !NO_HZ_FULL
398 This is the basic tick based cputime accounting that maintains
399 statistics about user, system and idle time spent on per jiffies
404 config VIRT_CPU_ACCOUNTING_NATIVE
405 bool "Deterministic task and CPU time accounting"
406 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
407 select VIRT_CPU_ACCOUNTING
409 Select this option to enable more accurate task and CPU time
410 accounting. This is done by reading a CPU counter on each
411 kernel entry and exit and on transitions within the kernel
412 between system, softirq and hardirq state, so there is a
413 small performance impact. In the case of s390 or IBM POWER > 5,
414 this also enables accounting of stolen time on logically-partitioned
417 config VIRT_CPU_ACCOUNTING_GEN
418 bool "Full dynticks CPU time accounting"
419 depends on HAVE_CONTEXT_TRACKING
420 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
421 depends on GENERIC_CLOCKEVENTS
422 select VIRT_CPU_ACCOUNTING
423 select CONTEXT_TRACKING
425 Select this option to enable task and CPU time accounting on full
426 dynticks systems. This accounting is implemented by watching every
427 kernel-user boundaries using the context tracking subsystem.
428 The accounting is thus performed at the expense of some significant
431 For now this is only useful if you are working on the full
432 dynticks subsystem development.
438 config IRQ_TIME_ACCOUNTING
439 bool "Fine granularity task level IRQ time accounting"
440 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
442 Select this option to enable fine granularity task irq time
443 accounting. This is done by reading a timestamp on each
444 transitions between softirq and hardirq state, so there can be a
445 small performance impact.
447 If in doubt, say N here.
449 config HAVE_SCHED_AVG_IRQ
451 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
454 config BSD_PROCESS_ACCT
455 bool "BSD Process Accounting"
458 If you say Y here, a user level program will be able to instruct the
459 kernel (via a special system call) to write process accounting
460 information to a file: whenever a process exits, information about
461 that process will be appended to the file by the kernel. The
462 information includes things such as creation time, owning user,
463 command name, memory usage, controlling terminal etc. (the complete
464 list is in the struct acct in <file:include/linux/acct.h>). It is
465 up to the user level program to do useful things with this
466 information. This is generally a good idea, so say Y.
468 config BSD_PROCESS_ACCT_V3
469 bool "BSD Process Accounting version 3 file format"
470 depends on BSD_PROCESS_ACCT
473 If you say Y here, the process accounting information is written
474 in a new file format that also logs the process IDs of each
475 process and its parent. Note that this file format is incompatible
476 with previous v0/v1/v2 file formats, so you will need updated tools
477 for processing it. A preliminary version of these tools is available
478 at <http://www.gnu.org/software/acct/>.
481 bool "Export task/process statistics through netlink"
486 Export selected statistics for tasks/processes through the
487 generic netlink interface. Unlike BSD process accounting, the
488 statistics are available during the lifetime of tasks/processes as
489 responses to commands. Like BSD accounting, they are sent to user
494 config TASK_DELAY_ACCT
495 bool "Enable per-task delay accounting"
499 Collect information on time spent by a task waiting for system
500 resources like cpu, synchronous block I/O completion and swapping
501 in pages. Such statistics can help in setting a task's priorities
502 relative to other tasks for cpu, io, rss limits etc.
507 bool "Enable extended accounting over taskstats"
510 Collect extended task accounting data and send the data
511 to userland for processing over the taskstats interface.
515 config TASK_IO_ACCOUNTING
516 bool "Enable per-task storage I/O accounting"
517 depends on TASK_XACCT
519 Collect information on the number of bytes of storage I/O which this
525 bool "Pressure stall information tracking"
527 Collect metrics that indicate how overcommitted the CPU, memory,
528 and IO capacity are in the system.
530 If you say Y here, the kernel will create /proc/pressure/ with the
531 pressure statistics files cpu, memory, and io. These will indicate
532 the share of walltime in which some or all tasks in the system are
533 delayed due to contention of the respective resource.
535 In kernels with cgroup support, cgroups (cgroup2 only) will
536 have cpu.pressure, memory.pressure, and io.pressure files,
537 which aggregate pressure stalls for the grouped tasks only.
539 For more details see Documentation/accounting/psi.rst.
543 config PSI_DEFAULT_DISABLED
544 bool "Require boot parameter to enable pressure stall information tracking"
548 If set, pressure stall information tracking will be disabled
549 per default but can be enabled through passing psi=1 on the
550 kernel commandline during boot.
552 This feature adds some code to the task wakeup and sleep
553 paths of the scheduler. The overhead is too low to affect
554 common scheduling-intense workloads in practice (such as
555 webservers, memcache), but it does show up in artificial
556 scheduler stress tests, such as hackbench.
558 If you are paranoid and not sure what the kernel will be
563 endmenu # "CPU/Task time and stats accounting"
567 depends on SMP || COMPILE_TEST
570 Make sure that CPUs running critical tasks are not disturbed by
571 any source of "noise" such as unbound workqueues, timers, kthreads...
572 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
573 the "isolcpus=" boot parameter.
577 source "kernel/rcu/Kconfig"
584 tristate "Kernel .config support"
586 This option enables the complete Linux kernel ".config" file
587 contents to be saved in the kernel. It provides documentation
588 of which kernel options are used in a running kernel or in an
589 on-disk kernel. This information can be extracted from the kernel
590 image file with the script scripts/extract-ikconfig and used as
591 input to rebuild the current kernel or to build another kernel.
592 It can also be extracted from a running kernel by reading
593 /proc/config.gz if enabled (below).
596 bool "Enable access to .config through /proc/config.gz"
597 depends on IKCONFIG && PROC_FS
599 This option enables access to the kernel configuration file
600 through /proc/config.gz.
603 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
606 This option enables access to the in-kernel headers that are generated during
607 the build process. These can be used to build eBPF tracing programs,
608 or similar programs. If you build the headers as a module, a module called
609 kheaders.ko is built which can be loaded on-demand to get access to headers.
612 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
617 Select the minimal kernel log buffer size as a power of 2.
618 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
619 parameter, see below. Any higher size also might be forced
620 by "log_buf_len" boot parameter.
630 config LOG_CPU_MAX_BUF_SHIFT
631 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
634 default 12 if !BASE_SMALL
635 default 0 if BASE_SMALL
638 This option allows to increase the default ring buffer size
639 according to the number of CPUs. The value defines the contribution
640 of each CPU as a power of 2. The used space is typically only few
641 lines however it might be much more when problems are reported,
644 The increased size means that a new buffer has to be allocated and
645 the original static one is unused. It makes sense only on systems
646 with more CPUs. Therefore this value is used only when the sum of
647 contributions is greater than the half of the default kernel ring
648 buffer as defined by LOG_BUF_SHIFT. The default values are set
649 so that more than 64 CPUs are needed to trigger the allocation.
651 Also this option is ignored when "log_buf_len" kernel parameter is
652 used as it forces an exact (power of two) size of the ring buffer.
654 The number of possible CPUs is used for this computation ignoring
655 hotplugging making the computation optimal for the worst case
656 scenario while allowing a simple algorithm to be used from bootup.
658 Examples shift values and their meaning:
659 17 => 128 KB for each CPU
660 16 => 64 KB for each CPU
661 15 => 32 KB for each CPU
662 14 => 16 KB for each CPU
663 13 => 8 KB for each CPU
664 12 => 4 KB for each CPU
666 config PRINTK_SAFE_LOG_BUF_SHIFT
667 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
672 Select the size of an alternate printk per-CPU buffer where messages
673 printed from usafe contexts are temporary stored. One example would
674 be NMI messages, another one - printk recursion. The messages are
675 copied to the main log buffer in a safe context to avoid a deadlock.
676 The value defines the size as a power of 2.
678 Those messages are rare and limited. The largest one is when
679 a backtrace is printed. It usually fits into 4KB. Select
680 8KB if you want to be on the safe side.
683 17 => 128 KB for each CPU
684 16 => 64 KB for each CPU
685 15 => 32 KB for each CPU
686 14 => 16 KB for each CPU
687 13 => 8 KB for each CPU
688 12 => 4 KB for each CPU
691 # Architectures with an unreliable sched_clock() should select this:
693 config HAVE_UNSTABLE_SCHED_CLOCK
696 config GENERIC_SCHED_CLOCK
699 menu "Scheduler features"
702 bool "Enable utilization clamping for RT/FAIR tasks"
703 depends on CPU_FREQ_GOV_SCHEDUTIL
705 This feature enables the scheduler to track the clamped utilization
706 of each CPU based on RUNNABLE tasks scheduled on that CPU.
708 With this option, the user can specify the min and max CPU
709 utilization allowed for RUNNABLE tasks. The max utilization defines
710 the maximum frequency a task should use while the min utilization
711 defines the minimum frequency it should use.
713 Both min and max utilization clamp values are hints to the scheduler,
714 aiming at improving its frequency selection policy, but they do not
715 enforce or grant any specific bandwidth for tasks.
719 config UCLAMP_BUCKETS_COUNT
720 int "Number of supported utilization clamp buckets"
723 depends on UCLAMP_TASK
725 Defines the number of clamp buckets to use. The range of each bucket
726 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
727 number of clamp buckets the finer their granularity and the higher
728 the precision of clamping aggregation and tracking at run-time.
730 For example, with the minimum configuration value we will have 5
731 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
732 be refcounted in the [20..39]% bucket and will set the bucket clamp
733 effective value to 25%.
734 If a second 30% boosted task should be co-scheduled on the same CPU,
735 that task will be refcounted in the same bucket of the first task and
736 it will boost the bucket clamp effective value to 30%.
737 The clamp effective value of a bucket is reset to its nominal value
738 (20% in the example above) when there are no more tasks refcounted in
741 An additional boost/capping margin can be added to some tasks. In the
742 example above the 25% task will be boosted to 30% until it exits the
743 CPU. If that should be considered not acceptable on certain systems,
744 it's always possible to reduce the margin by increasing the number of
745 clamp buckets to trade off used memory for run-time tracking
748 If in doubt, use the default value.
753 # For architectures that want to enable the support for NUMA-affine scheduler
756 config ARCH_SUPPORTS_NUMA_BALANCING
760 # For architectures that prefer to flush all TLBs after a number of pages
761 # are unmapped instead of sending one IPI per page to flush. The architecture
762 # must provide guarantees on what happens if a clean TLB cache entry is
763 # written after the unmap. Details are in mm/rmap.c near the check for
764 # should_defer_flush. The architecture should also consider if the full flush
765 # and the refill costs are offset by the savings of sending fewer IPIs.
766 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
770 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
773 # For architectures that know their GCC __int128 support is sound
775 config ARCH_SUPPORTS_INT128
778 # For architectures that (ab)use NUMA to represent different memory regions
779 # all cpu-local but of different latencies, such as SuperH.
781 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
784 config NUMA_BALANCING
785 bool "Memory placement aware NUMA scheduler"
786 depends on ARCH_SUPPORTS_NUMA_BALANCING
787 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
788 depends on SMP && NUMA && MIGRATION
790 This option adds support for automatic NUMA aware memory/task placement.
791 The mechanism is quite primitive and is based on migrating memory when
792 it has references to the node the task is running on.
794 This system will be inactive on UMA systems.
796 config NUMA_BALANCING_DEFAULT_ENABLED
797 bool "Automatically enable NUMA aware memory/task placement"
799 depends on NUMA_BALANCING
801 If set, automatic NUMA balancing will be enabled if running on a NUMA
805 bool "Control Group support"
808 This option adds support for grouping sets of processes together, for
809 use with process control subsystems such as Cpusets, CFS, memory
810 controls or device isolation.
812 - Documentation/scheduler/sched-design-CFS.rst (CFS)
813 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
814 and resource control)
824 bool "Memory controller"
828 Provides control over the memory footprint of tasks in a cgroup.
831 bool "Swap controller"
832 depends on MEMCG && SWAP
834 Provides control over the swap space consumed by tasks in a cgroup.
836 config MEMCG_SWAP_ENABLED
837 bool "Swap controller enabled by default"
838 depends on MEMCG_SWAP
841 Memory Resource Controller Swap Extension comes with its price in
842 a bigger memory consumption. General purpose distribution kernels
843 which want to enable the feature but keep it disabled by default
844 and let the user enable it by swapaccount=1 boot command line
845 parameter should have this option unselected.
846 For those who want to have the feature enabled by default should
847 select this option (if, for some reason, they need to disable it
848 then swapaccount=0 does the trick).
852 depends on MEMCG && !SLOB
860 Generic block IO controller cgroup interface. This is the common
861 cgroup interface which should be used by various IO controlling
864 Currently, CFQ IO scheduler uses it to recognize task groups and
865 control disk bandwidth allocation (proportional time slice allocation)
866 to such task groups. It is also used by bio throttling logic in
867 block layer to implement upper limit in IO rates on a device.
869 This option only enables generic Block IO controller infrastructure.
870 One needs to also enable actual IO controlling logic/policy. For
871 enabling proportional weight division of disk bandwidth in CFQ, set
872 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
873 CONFIG_BLK_DEV_THROTTLING=y.
875 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
877 config CGROUP_WRITEBACK
879 depends on MEMCG && BLK_CGROUP
882 menuconfig CGROUP_SCHED
883 bool "CPU controller"
886 This feature lets CPU scheduler recognize task groups and control CPU
887 bandwidth allocation to such task groups. It uses cgroups to group
891 config FAIR_GROUP_SCHED
892 bool "Group scheduling for SCHED_OTHER"
893 depends on CGROUP_SCHED
897 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
898 depends on FAIR_GROUP_SCHED
901 This option allows users to define CPU bandwidth rates (limits) for
902 tasks running within the fair group scheduler. Groups with no limit
903 set are considered to be unconstrained and will run with no
905 See Documentation/scheduler/sched-bwc.rst for more information.
907 config RT_GROUP_SCHED
908 bool "Group scheduling for SCHED_RR/FIFO"
909 depends on CGROUP_SCHED
912 This feature lets you explicitly allocate real CPU bandwidth
913 to task groups. If enabled, it will also make it impossible to
914 schedule realtime tasks for non-root users until you allocate
915 realtime bandwidth for them.
916 See Documentation/scheduler/sched-rt-group.rst for more information.
920 config UCLAMP_TASK_GROUP
921 bool "Utilization clamping per group of tasks"
922 depends on CGROUP_SCHED
923 depends on UCLAMP_TASK
926 This feature enables the scheduler to track the clamped utilization
927 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
929 When this option is enabled, the user can specify a min and max
930 CPU bandwidth which is allowed for each single task in a group.
931 The max bandwidth allows to clamp the maximum frequency a task
932 can use, while the min bandwidth allows to define a minimum
933 frequency a task will always use.
935 When task group based utilization clamping is enabled, an eventually
936 specified task-specific clamp value is constrained by the cgroup
937 specified clamp value. Both minimum and maximum task clamping cannot
938 be bigger than the corresponding clamping defined at task group level.
943 bool "PIDs controller"
945 Provides enforcement of process number limits in the scope of a
946 cgroup. Any attempt to fork more processes than is allowed in the
947 cgroup will fail. PIDs are fundamentally a global resource because it
948 is fairly trivial to reach PID exhaustion before you reach even a
949 conservative kmemcg limit. As a result, it is possible to grind a
950 system to halt without being limited by other cgroup policies. The
951 PIDs controller is designed to stop this from happening.
953 It should be noted that organisational operations (such as attaching
954 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
955 since the PIDs limit only affects a process's ability to fork, not to
959 bool "RDMA controller"
961 Provides enforcement of RDMA resources defined by IB stack.
962 It is fairly easy for consumers to exhaust RDMA resources, which
963 can result into resource unavailability to other consumers.
964 RDMA controller is designed to stop this from happening.
965 Attaching processes with active RDMA resources to the cgroup
966 hierarchy is allowed even if can cross the hierarchy's limit.
968 config CGROUP_FREEZER
969 bool "Freezer controller"
971 Provides a way to freeze and unfreeze all tasks in a
974 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
975 controller includes important in-kernel memory consumers per default.
977 If you're using cgroup2, say N.
979 config CGROUP_HUGETLB
980 bool "HugeTLB controller"
981 depends on HUGETLB_PAGE
985 Provides a cgroup controller for HugeTLB pages.
986 When you enable this, you can put a per cgroup limit on HugeTLB usage.
987 The limit is enforced during page fault. Since HugeTLB doesn't
988 support page reclaim, enforcing the limit at page fault time implies
989 that, the application will get SIGBUS signal if it tries to access
990 HugeTLB pages beyond its limit. This requires the application to know
991 beforehand how much HugeTLB pages it would require for its use. The
992 control group is tracked in the third page lru pointer. This means
993 that we cannot use the controller with huge page less than 3 pages.
996 bool "Cpuset controller"
999 This option will let you create and manage CPUSETs which
1000 allow dynamically partitioning a system into sets of CPUs and
1001 Memory Nodes and assigning tasks to run only within those sets.
1002 This is primarily useful on large SMP or NUMA systems.
1006 config PROC_PID_CPUSET
1007 bool "Include legacy /proc/<pid>/cpuset file"
1011 config CGROUP_DEVICE
1012 bool "Device controller"
1014 Provides a cgroup controller implementing whitelists for
1015 devices which a process in the cgroup can mknod or open.
1017 config CGROUP_CPUACCT
1018 bool "Simple CPU accounting controller"
1020 Provides a simple controller for monitoring the
1021 total CPU consumed by the tasks in a cgroup.
1024 bool "Perf controller"
1025 depends on PERF_EVENTS
1027 This option extends the perf per-cpu mode to restrict monitoring
1028 to threads which belong to the cgroup specified and run on the
1034 bool "Support for eBPF programs attached to cgroups"
1035 depends on BPF_SYSCALL
1036 select SOCK_CGROUP_DATA
1038 Allow attaching eBPF programs to a cgroup using the bpf(2)
1039 syscall command BPF_PROG_ATTACH.
1041 In which context these programs are accessed depends on the type
1042 of attachment. For instance, programs that are attached using
1043 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1047 bool "Debug controller"
1049 depends on DEBUG_KERNEL
1051 This option enables a simple controller that exports
1052 debugging information about the cgroups framework. This
1053 controller is for control cgroup debugging only. Its
1054 interfaces are not stable.
1058 config SOCK_CGROUP_DATA
1064 menuconfig NAMESPACES
1065 bool "Namespaces support" if EXPERT
1066 depends on MULTIUSER
1069 Provides the way to make tasks work with different objects using
1070 the same id. For example same IPC id may refer to different objects
1071 or same user id or pid may refer to different tasks when used in
1072 different namespaces.
1077 bool "UTS namespace"
1080 In this namespace tasks see different info provided with the
1084 bool "TIME namespace"
1085 depends on GENERIC_VDSO_TIME_NS
1088 In this namespace boottime and monotonic clocks can be set.
1089 The time will keep going with the same pace.
1092 bool "IPC namespace"
1093 depends on (SYSVIPC || POSIX_MQUEUE)
1096 In this namespace tasks work with IPC ids which correspond to
1097 different IPC objects in different namespaces.
1100 bool "User namespace"
1103 This allows containers, i.e. vservers, to use user namespaces
1104 to provide different user info for different servers.
1106 When user namespaces are enabled in the kernel it is
1107 recommended that the MEMCG option also be enabled and that
1108 user-space use the memory control groups to limit the amount
1109 of memory a memory unprivileged users can use.
1114 bool "PID Namespaces"
1117 Support process id namespaces. This allows having multiple
1118 processes with the same pid as long as they are in different
1119 pid namespaces. This is a building block of containers.
1122 bool "Network namespace"
1126 Allow user space to create what appear to be multiple instances
1127 of the network stack.
1131 config CHECKPOINT_RESTORE
1132 bool "Checkpoint/restore support"
1133 select PROC_CHILDREN
1136 Enables additional kernel features in a sake of checkpoint/restore.
1137 In particular it adds auxiliary prctl codes to setup process text,
1138 data and heap segment sizes, and a few additional /proc filesystem
1141 If unsure, say N here.
1143 config SCHED_AUTOGROUP
1144 bool "Automatic process group scheduling"
1147 select FAIR_GROUP_SCHED
1149 This option optimizes the scheduler for common desktop workloads by
1150 automatically creating and populating task groups. This separation
1151 of workloads isolates aggressive CPU burners (like build jobs) from
1152 desktop applications. Task group autogeneration is currently based
1155 config SYSFS_DEPRECATED
1156 bool "Enable deprecated sysfs features to support old userspace tools"
1160 This option adds code that switches the layout of the "block" class
1161 devices, to not show up in /sys/class/block/, but only in
1164 This switch is only active when the sysfs.deprecated=1 boot option is
1165 passed or the SYSFS_DEPRECATED_V2 option is set.
1167 This option allows new kernels to run on old distributions and tools,
1168 which might get confused by /sys/class/block/. Since 2007/2008 all
1169 major distributions and tools handle this just fine.
1171 Recent distributions and userspace tools after 2009/2010 depend on
1172 the existence of /sys/class/block/, and will not work with this
1175 Only if you are using a new kernel on an old distribution, you might
1178 config SYSFS_DEPRECATED_V2
1179 bool "Enable deprecated sysfs features by default"
1182 depends on SYSFS_DEPRECATED
1184 Enable deprecated sysfs by default.
1186 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1189 Only if you are using a new kernel on an old distribution, you might
1190 need to say Y here. Even then, odds are you would not need it
1191 enabled, you can always pass the boot option if absolutely necessary.
1194 bool "Kernel->user space relay support (formerly relayfs)"
1197 This option enables support for relay interface support in
1198 certain file systems (such as debugfs).
1199 It is designed to provide an efficient mechanism for tools and
1200 facilities to relay large amounts of data from kernel space to
1205 config BLK_DEV_INITRD
1206 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1208 The initial RAM filesystem is a ramfs which is loaded by the
1209 boot loader (loadlin or lilo) and that is mounted as root
1210 before the normal boot procedure. It is typically used to
1211 load modules needed to mount the "real" root file system,
1212 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1214 If RAM disk support (BLK_DEV_RAM) is also included, this
1215 also enables initial RAM disk (initrd) support and adds
1216 15 Kbytes (more on some other architectures) to the kernel size.
1222 source "usr/Kconfig"
1227 bool "Boot config support"
1228 select BLK_DEV_INITRD
1230 Extra boot config allows system admin to pass a config file as
1231 complemental extension of kernel cmdline when booting.
1232 The boot config file must be attached at the end of initramfs
1233 with checksum, size and magic word.
1234 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1239 prompt "Compiler optimization level"
1240 default CC_OPTIMIZE_FOR_PERFORMANCE
1242 config CC_OPTIMIZE_FOR_PERFORMANCE
1243 bool "Optimize for performance (-O2)"
1245 This is the default optimization level for the kernel, building
1246 with the "-O2" compiler flag for best performance and most
1247 helpful compile-time warnings.
1249 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1250 bool "Optimize more for performance (-O3)"
1252 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1254 Choosing this option will pass "-O3" to your compiler to optimize
1255 the kernel yet more for performance.
1257 config CC_OPTIMIZE_FOR_SIZE
1258 bool "Optimize for size (-Os)"
1259 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1261 Choosing this option will pass "-Os" to your compiler resulting
1262 in a smaller kernel.
1266 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1269 This requires that the arch annotates or otherwise protects
1270 its external entry points from being discarded. Linker scripts
1271 must also merge .text.*, .data.*, and .bss.* correctly into
1272 output sections. Care must be taken not to pull in unrelated
1273 sections (e.g., '.text.init'). Typically '.' in section names
1274 is used to distinguish them from label names / C identifiers.
1276 config LD_DEAD_CODE_DATA_ELIMINATION
1277 bool "Dead code and data elimination (EXPERIMENTAL)"
1278 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1280 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1281 depends on $(cc-option,-ffunction-sections -fdata-sections)
1282 depends on $(ld-option,--gc-sections)
1284 Enable this if you want to do dead code and data elimination with
1285 the linker by compiling with -ffunction-sections -fdata-sections,
1286 and linking with --gc-sections.
1288 This can reduce on disk and in-memory size of the kernel
1289 code and static data, particularly for small configs and
1290 on small systems. This has the possibility of introducing
1291 silently broken kernel if the required annotations are not
1292 present. This option is not well tested yet, so use at your
1301 config SYSCTL_EXCEPTION_TRACE
1304 Enable support for /proc/sys/debug/exception-trace.
1306 config SYSCTL_ARCH_UNALIGN_NO_WARN
1309 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1310 Allows arch to define/use @no_unaligned_warning to possibly warn
1311 about unaligned access emulation going on under the hood.
1313 config SYSCTL_ARCH_UNALIGN_ALLOW
1316 Enable support for /proc/sys/kernel/unaligned-trap
1317 Allows arches to define/use @unaligned_enabled to runtime toggle
1318 the unaligned access emulation.
1319 see arch/parisc/kernel/unaligned.c for reference
1321 config HAVE_PCSPKR_PLATFORM
1324 # interpreter that classic socket filters depend on
1329 bool "Configure standard kernel features (expert users)"
1330 # Unhide debug options, to make the on-by-default options visible
1333 This option allows certain base kernel options and settings
1334 to be disabled or tweaked. This is for specialized
1335 environments which can tolerate a "non-standard" kernel.
1336 Only use this if you really know what you are doing.
1339 bool "Enable 16-bit UID system calls" if EXPERT
1340 depends on HAVE_UID16 && MULTIUSER
1343 This enables the legacy 16-bit UID syscall wrappers.
1346 bool "Multiple users, groups and capabilities support" if EXPERT
1349 This option enables support for non-root users, groups and
1352 If you say N here, all processes will run with UID 0, GID 0, and all
1353 possible capabilities. Saying N here also compiles out support for
1354 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1357 If unsure, say Y here.
1359 config SGETMASK_SYSCALL
1360 bool "sgetmask/ssetmask syscalls support" if EXPERT
1361 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1363 sys_sgetmask and sys_ssetmask are obsolete system calls
1364 no longer supported in libc but still enabled by default in some
1367 If unsure, leave the default option here.
1369 config SYSFS_SYSCALL
1370 bool "Sysfs syscall support" if EXPERT
1373 sys_sysfs is an obsolete system call no longer supported in libc.
1374 Note that disabling this option is more secure but might break
1375 compatibility with some systems.
1377 If unsure say Y here.
1380 bool "open by fhandle syscalls" if EXPERT
1384 If you say Y here, a user level program will be able to map
1385 file names to handle and then later use the handle for
1386 different file system operations. This is useful in implementing
1387 userspace file servers, which now track files using handles instead
1388 of names. The handle would remain the same even if file names
1389 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1393 bool "Posix Clocks & timers" if EXPERT
1396 This includes native support for POSIX timers to the kernel.
1397 Some embedded systems have no use for them and therefore they
1398 can be configured out to reduce the size of the kernel image.
1400 When this option is disabled, the following syscalls won't be
1401 available: timer_create, timer_gettime: timer_getoverrun,
1402 timer_settime, timer_delete, clock_adjtime, getitimer,
1403 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1404 clock_getres and clock_nanosleep syscalls will be limited to
1405 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1411 bool "Enable support for printk" if EXPERT
1414 This option enables normal printk support. Removing it
1415 eliminates most of the message strings from the kernel image
1416 and makes the kernel more or less silent. As this makes it
1417 very difficult to diagnose system problems, saying N here is
1418 strongly discouraged.
1426 bool "BUG() support" if EXPERT
1429 Disabling this option eliminates support for BUG and WARN, reducing
1430 the size of your kernel image and potentially quietly ignoring
1431 numerous fatal conditions. You should only consider disabling this
1432 option for embedded systems with no facilities for reporting errors.
1438 bool "Enable ELF core dumps" if EXPERT
1440 Enable support for generating core dumps. Disabling saves about 4k.
1443 config PCSPKR_PLATFORM
1444 bool "Enable PC-Speaker support" if EXPERT
1445 depends on HAVE_PCSPKR_PLATFORM
1449 This option allows to disable the internal PC-Speaker
1450 support, saving some memory.
1454 bool "Enable full-sized data structures for core" if EXPERT
1456 Disabling this option reduces the size of miscellaneous core
1457 kernel data structures. This saves memory on small machines,
1458 but may reduce performance.
1461 bool "Enable futex support" if EXPERT
1465 Disabling this option will cause the kernel to be built without
1466 support for "fast userspace mutexes". The resulting kernel may not
1467 run glibc-based applications correctly.
1471 depends on FUTEX && RT_MUTEXES
1474 config HAVE_FUTEX_CMPXCHG
1478 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1479 is implemented and always working. This removes a couple of runtime
1483 bool "Enable eventpoll support" if EXPERT
1486 Disabling this option will cause the kernel to be built without
1487 support for epoll family of system calls.
1490 bool "Enable signalfd() system call" if EXPERT
1493 Enable the signalfd() system call that allows to receive signals
1494 on a file descriptor.
1499 bool "Enable timerfd() system call" if EXPERT
1502 Enable the timerfd() system call that allows to receive timer
1503 events on a file descriptor.
1508 bool "Enable eventfd() system call" if EXPERT
1511 Enable the eventfd() system call that allows to receive both
1512 kernel notification (ie. KAIO) or userspace notifications.
1517 bool "Use full shmem filesystem" if EXPERT
1521 The shmem is an internal filesystem used to manage shared memory.
1522 It is backed by swap and manages resource limits. It is also exported
1523 to userspace as tmpfs if TMPFS is enabled. Disabling this
1524 option replaces shmem and tmpfs with the much simpler ramfs code,
1525 which may be appropriate on small systems without swap.
1528 bool "Enable AIO support" if EXPERT
1531 This option enables POSIX asynchronous I/O which may by used
1532 by some high performance threaded applications. Disabling
1533 this option saves about 7k.
1536 bool "Enable IO uring support" if EXPERT
1541 This option enables support for the io_uring interface, enabling
1542 applications to submit and complete IO through submission and
1543 completion rings that are shared between the kernel and application.
1545 config ADVISE_SYSCALLS
1546 bool "Enable madvise/fadvise syscalls" if EXPERT
1549 This option enables the madvise and fadvise syscalls, used by
1550 applications to advise the kernel about their future memory or file
1551 usage, improving performance. If building an embedded system where no
1552 applications use these syscalls, you can disable this option to save
1556 bool "Enable membarrier() system call" if EXPERT
1559 Enable the membarrier() system call that allows issuing memory
1560 barriers across all running threads, which can be used to distribute
1561 the cost of user-space memory barriers asymmetrically by transforming
1562 pairs of memory barriers into pairs consisting of membarrier() and a
1568 bool "Load all symbols for debugging/ksymoops" if EXPERT
1571 Say Y here to let the kernel print out symbolic crash information and
1572 symbolic stack backtraces. This increases the size of the kernel
1573 somewhat, as all symbols have to be loaded into the kernel image.
1576 bool "Include all symbols in kallsyms"
1577 depends on DEBUG_KERNEL && KALLSYMS
1579 Normally kallsyms only contains the symbols of functions for nicer
1580 OOPS messages and backtraces (i.e., symbols from the text and inittext
1581 sections). This is sufficient for most cases. And only in very rare
1582 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1583 names of variables from the data sections, etc).
1585 This option makes sure that all symbols are loaded into the kernel
1586 image (i.e., symbols from all sections) in cost of increased kernel
1587 size (depending on the kernel configuration, it may be 300KiB or
1588 something like this).
1590 Say N unless you really need all symbols.
1592 config KALLSYMS_ABSOLUTE_PERCPU
1595 default X86_64 && SMP
1597 config KALLSYMS_BASE_RELATIVE
1602 Instead of emitting them as absolute values in the native word size,
1603 emit the symbol references in the kallsyms table as 32-bit entries,
1604 each containing a relative value in the range [base, base + U32_MAX]
1605 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1606 an absolute value in the range [0, S32_MAX] or a relative value in the
1607 range [base, base + S32_MAX], where base is the lowest relative symbol
1608 address encountered in the image.
1610 On 64-bit builds, this reduces the size of the address table by 50%,
1611 but more importantly, it results in entries whose values are build
1612 time constants, and no relocation pass is required at runtime to fix
1613 up the entries based on the runtime load address of the kernel.
1615 # end of the "standard kernel features (expert users)" menu
1617 # syscall, maps, verifier
1620 bool "LSM Instrumentation with BPF"
1621 depends on BPF_EVENTS
1622 depends on BPF_SYSCALL
1626 Enables instrumentation of the security hooks with eBPF programs for
1627 implementing dynamic MAC and Audit Policies.
1629 If you are unsure how to answer this question, answer N.
1632 bool "Enable bpf() system call"
1637 Enable the bpf() system call that allows to manipulate eBPF
1638 programs and maps via file descriptors.
1640 config ARCH_WANT_DEFAULT_BPF_JIT
1643 config BPF_JIT_ALWAYS_ON
1644 bool "Permanently enable BPF JIT and remove BPF interpreter"
1645 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1647 Enables BPF JIT and removes BPF interpreter to avoid
1648 speculative execution of BPF instructions by the interpreter
1650 config BPF_JIT_DEFAULT_ON
1651 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1652 depends on HAVE_EBPF_JIT && BPF_JIT
1655 bool "Enable userfaultfd() system call"
1658 Enable the userfaultfd() system call that allows to intercept and
1659 handle page faults in userland.
1661 config ARCH_HAS_MEMBARRIER_CALLBACKS
1664 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1668 bool "Enable rseq() system call" if EXPERT
1670 depends on HAVE_RSEQ
1673 Enable the restartable sequences system call. It provides a
1674 user-space cache for the current CPU number value, which
1675 speeds up getting the current CPU number from user-space,
1676 as well as an ABI to speed up user-space operations on
1683 bool "Enabled debugging of rseq() system call" if EXPERT
1684 depends on RSEQ && DEBUG_KERNEL
1686 Enable extra debugging checks for the rseq system call.
1691 bool "Embedded system"
1692 option allnoconfig_y
1695 This option should be enabled if compiling the kernel for
1696 an embedded system so certain expert options are available
1699 config HAVE_PERF_EVENTS
1702 See tools/perf/design.txt for details.
1704 config PERF_USE_VMALLOC
1707 See tools/perf/design.txt for details
1710 bool "PC/104 support" if EXPERT
1712 Expose PC/104 form factor device drivers and options available for
1713 selection and configuration. Enable this option if your target
1714 machine has a PC/104 bus.
1716 menu "Kernel Performance Events And Counters"
1719 bool "Kernel performance events and counters"
1720 default y if PROFILING
1721 depends on HAVE_PERF_EVENTS
1725 Enable kernel support for various performance events provided
1726 by software and hardware.
1728 Software events are supported either built-in or via the
1729 use of generic tracepoints.
1731 Most modern CPUs support performance events via performance
1732 counter registers. These registers count the number of certain
1733 types of hw events: such as instructions executed, cachemisses
1734 suffered, or branches mis-predicted - without slowing down the
1735 kernel or applications. These registers can also trigger interrupts
1736 when a threshold number of events have passed - and can thus be
1737 used to profile the code that runs on that CPU.
1739 The Linux Performance Event subsystem provides an abstraction of
1740 these software and hardware event capabilities, available via a
1741 system call and used by the "perf" utility in tools/perf/. It
1742 provides per task and per CPU counters, and it provides event
1743 capabilities on top of those.
1747 config DEBUG_PERF_USE_VMALLOC
1749 bool "Debug: use vmalloc to back perf mmap() buffers"
1750 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1751 select PERF_USE_VMALLOC
1753 Use vmalloc memory to back perf mmap() buffers.
1755 Mostly useful for debugging the vmalloc code on platforms
1756 that don't require it.
1762 config VM_EVENT_COUNTERS
1764 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1766 VM event counters are needed for event counts to be shown.
1767 This option allows the disabling of the VM event counters
1768 on EXPERT systems. /proc/vmstat will only show page counts
1769 if VM event counters are disabled.
1773 bool "Enable SLUB debugging support" if EXPERT
1774 depends on SLUB && SYSFS
1776 SLUB has extensive debug support features. Disabling these can
1777 result in significant savings in code size. This also disables
1778 SLUB sysfs support. /sys/slab will not exist and there will be
1779 no support for cache validation etc.
1781 config SLUB_MEMCG_SYSFS_ON
1783 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1784 depends on SLUB && SYSFS && MEMCG
1786 SLUB creates a directory under /sys/kernel/slab for each
1787 allocation cache to host info and debug files. If memory
1788 cgroup is enabled, each cache can have per memory cgroup
1789 caches. SLUB can create the same sysfs directories for these
1790 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1791 to a very high number of debug files being created. This is
1792 controlled by slub_memcg_sysfs boot parameter and this
1793 config option determines the parameter's default value.
1796 bool "Disable heap randomization"
1799 Randomizing heap placement makes heap exploits harder, but it
1800 also breaks ancient binaries (including anything libc5 based).
1801 This option changes the bootup default to heap randomization
1802 disabled, and can be overridden at runtime by setting
1803 /proc/sys/kernel/randomize_va_space to 2.
1805 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1808 prompt "Choose SLAB allocator"
1811 This option allows to select a slab allocator.
1815 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1817 The regular slab allocator that is established and known to work
1818 well in all environments. It organizes cache hot objects in
1819 per cpu and per node queues.
1822 bool "SLUB (Unqueued Allocator)"
1823 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1825 SLUB is a slab allocator that minimizes cache line usage
1826 instead of managing queues of cached objects (SLAB approach).
1827 Per cpu caching is realized using slabs of objects instead
1828 of queues of objects. SLUB can use memory efficiently
1829 and has enhanced diagnostics. SLUB is the default choice for
1834 bool "SLOB (Simple Allocator)"
1836 SLOB replaces the stock allocator with a drastically simpler
1837 allocator. SLOB is generally more space efficient but
1838 does not perform as well on large systems.
1842 config SLAB_MERGE_DEFAULT
1843 bool "Allow slab caches to be merged"
1846 For reduced kernel memory fragmentation, slab caches can be
1847 merged when they share the same size and other characteristics.
1848 This carries a risk of kernel heap overflows being able to
1849 overwrite objects from merged caches (and more easily control
1850 cache layout), which makes such heap attacks easier to exploit
1851 by attackers. By keeping caches unmerged, these kinds of exploits
1852 can usually only damage objects in the same cache. To disable
1853 merging at runtime, "slab_nomerge" can be passed on the kernel
1856 config SLAB_FREELIST_RANDOM
1858 depends on SLAB || SLUB
1859 bool "SLAB freelist randomization"
1861 Randomizes the freelist order used on creating new pages. This
1862 security feature reduces the predictability of the kernel slab
1863 allocator against heap overflows.
1865 config SLAB_FREELIST_HARDENED
1866 bool "Harden slab freelist metadata"
1869 Many kernel heap attacks try to target slab cache metadata and
1870 other infrastructure. This options makes minor performance
1871 sacrifices to harden the kernel slab allocator against common
1872 freelist exploit methods.
1874 config SHUFFLE_PAGE_ALLOCATOR
1875 bool "Page allocator randomization"
1876 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1878 Randomization of the page allocator improves the average
1879 utilization of a direct-mapped memory-side-cache. See section
1880 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1881 6.2a specification for an example of how a platform advertises
1882 the presence of a memory-side-cache. There are also incidental
1883 security benefits as it reduces the predictability of page
1884 allocations to compliment SLAB_FREELIST_RANDOM, but the
1885 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1886 10th order of pages is selected based on cache utilization
1889 While the randomization improves cache utilization it may
1890 negatively impact workloads on platforms without a cache. For
1891 this reason, by default, the randomization is enabled only
1892 after runtime detection of a direct-mapped memory-side-cache.
1893 Otherwise, the randomization may be force enabled with the
1894 'page_alloc.shuffle' kernel command line parameter.
1898 config SLUB_CPU_PARTIAL
1900 depends on SLUB && SMP
1901 bool "SLUB per cpu partial cache"
1903 Per cpu partial caches accelerate objects allocation and freeing
1904 that is local to a processor at the price of more indeterminism
1905 in the latency of the free. On overflow these caches will be cleared
1906 which requires the taking of locks that may cause latency spikes.
1907 Typically one would choose no for a realtime system.
1909 config MMAP_ALLOW_UNINITIALIZED
1910 bool "Allow mmapped anonymous memory to be uninitialized"
1911 depends on EXPERT && !MMU
1914 Normally, and according to the Linux spec, anonymous memory obtained
1915 from mmap() has its contents cleared before it is passed to
1916 userspace. Enabling this config option allows you to request that
1917 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1918 providing a huge performance boost. If this option is not enabled,
1919 then the flag will be ignored.
1921 This is taken advantage of by uClibc's malloc(), and also by
1922 ELF-FDPIC binfmt's brk and stack allocator.
1924 Because of the obvious security issues, this option should only be
1925 enabled on embedded devices where you control what is run in
1926 userspace. Since that isn't generally a problem on no-MMU systems,
1927 it is normally safe to say Y here.
1929 See Documentation/nommu-mmap.txt for more information.
1931 config SYSTEM_DATA_VERIFICATION
1933 select SYSTEM_TRUSTED_KEYRING
1937 select ASYMMETRIC_KEY_TYPE
1938 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1941 select X509_CERTIFICATE_PARSER
1942 select PKCS7_MESSAGE_PARSER
1944 Provide PKCS#7 message verification using the contents of the system
1945 trusted keyring to provide public keys. This then can be used for
1946 module verification, kexec image verification and firmware blob
1950 bool "Profiling support"
1952 Say Y here to enable the extended profiling support mechanisms used
1953 by profilers such as OProfile.
1956 # Place an empty function call at each tracepoint site. Can be
1957 # dynamically changed for a probe function.
1962 endmenu # General setup
1964 source "arch/Kconfig"
1971 default 0 if BASE_FULL
1972 default 1 if !BASE_FULL
1974 config MODULE_SIG_FORMAT
1976 select SYSTEM_DATA_VERIFICATION
1979 bool "Enable loadable module support"
1982 Kernel modules are small pieces of compiled code which can
1983 be inserted in the running kernel, rather than being
1984 permanently built into the kernel. You use the "modprobe"
1985 tool to add (and sometimes remove) them. If you say Y here,
1986 many parts of the kernel can be built as modules (by
1987 answering M instead of Y where indicated): this is most
1988 useful for infrequently used options which are not required
1989 for booting. For more information, see the man pages for
1990 modprobe, lsmod, modinfo, insmod and rmmod.
1992 If you say Y here, you will need to run "make
1993 modules_install" to put the modules under /lib/modules/
1994 where modprobe can find them (you may need to be root to do
2001 config MODULE_FORCE_LOAD
2002 bool "Forced module loading"
2005 Allow loading of modules without version information (ie. modprobe
2006 --force). Forced module loading sets the 'F' (forced) taint flag and
2007 is usually a really bad idea.
2009 config MODULE_UNLOAD
2010 bool "Module unloading"
2012 Without this option you will not be able to unload any
2013 modules (note that some modules may not be unloadable
2014 anyway), which makes your kernel smaller, faster
2015 and simpler. If unsure, say Y.
2017 config MODULE_FORCE_UNLOAD
2018 bool "Forced module unloading"
2019 depends on MODULE_UNLOAD
2021 This option allows you to force a module to unload, even if the
2022 kernel believes it is unsafe: the kernel will remove the module
2023 without waiting for anyone to stop using it (using the -f option to
2024 rmmod). This is mainly for kernel developers and desperate users.
2028 bool "Module versioning support"
2030 Usually, you have to use modules compiled with your kernel.
2031 Saying Y here makes it sometimes possible to use modules
2032 compiled for different kernels, by adding enough information
2033 to the modules to (hopefully) spot any changes which would
2034 make them incompatible with the kernel you are running. If
2037 config ASM_MODVERSIONS
2039 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2041 This enables module versioning for exported symbols also from
2042 assembly. This can be enabled only when the target architecture
2045 config MODULE_REL_CRCS
2047 depends on MODVERSIONS
2049 config MODULE_SRCVERSION_ALL
2050 bool "Source checksum for all modules"
2052 Modules which contain a MODULE_VERSION get an extra "srcversion"
2053 field inserted into their modinfo section, which contains a
2054 sum of the source files which made it. This helps maintainers
2055 see exactly which source was used to build a module (since
2056 others sometimes change the module source without updating
2057 the version). With this option, such a "srcversion" field
2058 will be created for all modules. If unsure, say N.
2061 bool "Module signature verification"
2062 select MODULE_SIG_FORMAT
2064 Check modules for valid signatures upon load: the signature
2065 is simply appended to the module. For more information see
2066 <file:Documentation/admin-guide/module-signing.rst>.
2068 Note that this option adds the OpenSSL development packages as a
2069 kernel build dependency so that the signing tool can use its crypto
2072 You should enable this option if you wish to use either
2073 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2074 another LSM - otherwise unsigned modules will be loadable regardless
2075 of the lockdown policy.
2077 !!!WARNING!!! If you enable this option, you MUST make sure that the
2078 module DOES NOT get stripped after being signed. This includes the
2079 debuginfo strip done by some packagers (such as rpmbuild) and
2080 inclusion into an initramfs that wants the module size reduced.
2082 config MODULE_SIG_FORCE
2083 bool "Require modules to be validly signed"
2084 depends on MODULE_SIG
2086 Reject unsigned modules or signed modules for which we don't have a
2087 key. Without this, such modules will simply taint the kernel.
2089 config MODULE_SIG_ALL
2090 bool "Automatically sign all modules"
2092 depends on MODULE_SIG
2094 Sign all modules during make modules_install. Without this option,
2095 modules must be signed manually, using the scripts/sign-file tool.
2097 comment "Do not forget to sign required modules with scripts/sign-file"
2098 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2101 prompt "Which hash algorithm should modules be signed with?"
2102 depends on MODULE_SIG
2104 This determines which sort of hashing algorithm will be used during
2105 signature generation. This algorithm _must_ be built into the kernel
2106 directly so that signature verification can take place. It is not
2107 possible to load a signed module containing the algorithm to check
2108 the signature on that module.
2110 config MODULE_SIG_SHA1
2111 bool "Sign modules with SHA-1"
2114 config MODULE_SIG_SHA224
2115 bool "Sign modules with SHA-224"
2116 select CRYPTO_SHA256
2118 config MODULE_SIG_SHA256
2119 bool "Sign modules with SHA-256"
2120 select CRYPTO_SHA256
2122 config MODULE_SIG_SHA384
2123 bool "Sign modules with SHA-384"
2124 select CRYPTO_SHA512
2126 config MODULE_SIG_SHA512
2127 bool "Sign modules with SHA-512"
2128 select CRYPTO_SHA512
2132 config MODULE_SIG_HASH
2134 depends on MODULE_SIG
2135 default "sha1" if MODULE_SIG_SHA1
2136 default "sha224" if MODULE_SIG_SHA224
2137 default "sha256" if MODULE_SIG_SHA256
2138 default "sha384" if MODULE_SIG_SHA384
2139 default "sha512" if MODULE_SIG_SHA512
2141 config MODULE_COMPRESS
2142 bool "Compress modules on installation"
2145 Compresses kernel modules when 'make modules_install' is run; gzip or
2146 xz depending on "Compression algorithm" below.
2148 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2150 Out-of-tree kernel modules installed using Kbuild will also be
2151 compressed upon installation.
2153 Note: for modules inside an initrd or initramfs, it's more efficient
2154 to compress the whole initrd or initramfs instead.
2156 Note: This is fully compatible with signed modules.
2161 prompt "Compression algorithm"
2162 depends on MODULE_COMPRESS
2163 default MODULE_COMPRESS_GZIP
2165 This determines which sort of compression will be used during
2166 'make modules_install'.
2168 GZIP (default) and XZ are supported.
2170 config MODULE_COMPRESS_GZIP
2173 config MODULE_COMPRESS_XZ
2178 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2179 bool "Allow loading of modules with missing namespace imports"
2181 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2182 a namespace. A module that makes use of a symbol exported with such a
2183 namespace is required to import the namespace via MODULE_IMPORT_NS().
2184 There is no technical reason to enforce correct namespace imports,
2185 but it creates consistency between symbols defining namespaces and
2186 users importing namespaces they make use of. This option relaxes this
2187 requirement and lifts the enforcement when loading a module.
2191 config UNUSED_SYMBOLS
2192 bool "Enable unused/obsolete exported symbols"
2195 Unused but exported symbols make the kernel needlessly bigger. For
2196 that reason most of these unused exports will soon be removed. This
2197 option is provided temporarily to provide a transition period in case
2198 some external kernel module needs one of these symbols anyway. If you
2199 encounter such a case in your module, consider if you are actually
2200 using the right API. (rationale: since nobody in the kernel is using
2201 this in a module, there is a pretty good chance it's actually the
2202 wrong interface to use). If you really need the symbol, please send a
2203 mail to the linux kernel mailing list mentioning the symbol and why
2204 you really need it, and what the merge plan to the mainline kernel for
2207 config TRIM_UNUSED_KSYMS
2208 bool "Trim unused exported kernel symbols"
2209 depends on !UNUSED_SYMBOLS
2211 The kernel and some modules make many symbols available for
2212 other modules to use via EXPORT_SYMBOL() and variants. Depending
2213 on the set of modules being selected in your kernel configuration,
2214 many of those exported symbols might never be used.
2216 This option allows for unused exported symbols to be dropped from
2217 the build. In turn, this provides the compiler more opportunities
2218 (especially when using LTO) for optimizing the code and reducing
2219 binary size. This might have some security advantages as well.
2221 If unsure, or if you need to build out-of-tree modules, say N.
2225 config MODULES_TREE_LOOKUP
2227 depends on PERF_EVENTS || TRACING
2229 config INIT_ALL_POSSIBLE
2232 Back when each arch used to define their own cpu_online_mask and
2233 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2234 with all 1s, and others with all 0s. When they were centralised,
2235 it was better to provide this option than to break all the archs
2236 and have several arch maintainers pursuing me down dark alleys.
2238 source "block/Kconfig"
2240 config PREEMPT_NOTIFIERS
2250 Build a simple ASN.1 grammar compiler that produces a bytecode output
2251 that can be interpreted by the ASN.1 stream decoder and used to
2252 inform it as to what tags are to be expected in a stream and what
2253 functions to call on what tags.
2255 source "kernel/Kconfig.locks"
2257 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2260 # It may be useful for an architecture to override the definitions of the
2261 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2262 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2263 # different calling convention for syscalls. They can also override the
2264 # macros for not-implemented syscalls in kernel/sys_ni.c and
2265 # kernel/time/posix-stubs.c. All these overrides need to be available in
2266 # <asm/syscall_wrapper.h>.
2267 config ARCH_HAS_SYSCALL_WRAPPER