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)"
9 default "arch/$(SRCARCH)/configs/$(KBUILD_DEFCONFIG)"
12 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
16 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
20 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
24 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
27 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
29 config CC_HAS_ASM_GOTO
30 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
32 config TOOLS_SUPPORT_RELR
33 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
35 config CC_HAS_ASM_INLINE
36 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
38 config CC_HAS_WARN_MAYBE_UNINITIALIZED
39 def_bool $(cc-option,-Wmaybe-uninitialized)
41 GCC >= 4.7 supports this option.
43 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
45 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
46 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
48 GCC's -Wmaybe-uninitialized is not reliable by definition.
49 Lots of false positive warnings are produced in some cases.
51 If this option is enabled, -Wno-maybe-uninitialzed is passed
52 to the compiler to suppress maybe-uninitialized warnings.
61 config BUILDTIME_TABLE_SORT
64 config THREAD_INFO_IN_TASK
67 Select this to move thread_info off the stack into task_struct. To
68 make this work, an arch will need to remove all thread_info fields
69 except flags and fix any runtime bugs.
71 One subtle change that will be needed is to use try_get_task_stack()
72 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
81 depends on BROKEN || !SMP
84 config INIT_ENV_ARG_LIMIT
89 Maximum of each of the number of arguments and environment
90 variables passed to init from the kernel command line.
93 bool "Compile also drivers which will not load"
97 Some drivers can be compiled on a different platform than they are
98 intended to be run on. Despite they cannot be loaded there (or even
99 when they load they cannot be used due to missing HW support),
100 developers still, opposing to distributors, might want to build such
101 drivers to compile-test them.
103 If you are a developer and want to build everything available, say Y
104 here. If you are a user/distributor, say N here to exclude useless
105 drivers to be distributed.
107 config UAPI_HEADER_TEST
108 bool "Compile test UAPI headers"
109 depends on HEADERS_INSTALL && CC_CAN_LINK
111 Compile test headers exported to user-space to ensure they are
112 self-contained, i.e. compilable as standalone units.
114 If you are a developer or tester and want to ensure the exported
115 headers are self-contained, say Y here. Otherwise, choose N.
118 string "Local version - append to kernel release"
120 Append an extra string to the end of your kernel version.
121 This will show up when you type uname, for example.
122 The string you set here will be appended after the contents of
123 any files with a filename matching localversion* in your
124 object and source tree, in that order. Your total string can
125 be a maximum of 64 characters.
127 config LOCALVERSION_AUTO
128 bool "Automatically append version information to the version string"
130 depends on !COMPILE_TEST
132 This will try to automatically determine if the current tree is a
133 release tree by looking for git tags that belong to the current
134 top of tree revision.
136 A string of the format -gxxxxxxxx will be added to the localversion
137 if a git-based tree is found. The string generated by this will be
138 appended after any matching localversion* files, and after the value
139 set in CONFIG_LOCALVERSION.
141 (The actual string used here is the first eight characters produced
142 by running the command:
144 $ git rev-parse --verify HEAD
146 which is done within the script "scripts/setlocalversion".)
149 string "Build ID Salt"
152 The build ID is used to link binaries and their debug info. Setting
153 this option will use the value in the calculation of the build id.
154 This is mostly useful for distributions which want to ensure the
155 build is unique between builds. It's safe to leave the default.
157 config HAVE_KERNEL_GZIP
160 config HAVE_KERNEL_BZIP2
163 config HAVE_KERNEL_LZMA
166 config HAVE_KERNEL_XZ
169 config HAVE_KERNEL_LZO
172 config HAVE_KERNEL_LZ4
175 config HAVE_KERNEL_UNCOMPRESSED
179 prompt "Kernel compression mode"
181 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
183 The linux kernel is a kind of self-extracting executable.
184 Several compression algorithms are available, which differ
185 in efficiency, compression and decompression speed.
186 Compression speed is only relevant when building a kernel.
187 Decompression speed is relevant at each boot.
189 If you have any problems with bzip2 or lzma compressed
190 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
191 version of this functionality (bzip2 only), for 2.4, was
192 supplied by Christian Ludwig)
194 High compression options are mostly useful for users, who
195 are low on disk space (embedded systems), but for whom ram
198 If in doubt, select 'gzip'
202 depends on HAVE_KERNEL_GZIP
204 The old and tried gzip compression. It provides a good balance
205 between compression ratio and decompression speed.
209 depends on HAVE_KERNEL_BZIP2
211 Its compression ratio and speed is intermediate.
212 Decompression speed is slowest among the choices. The kernel
213 size is about 10% smaller with bzip2, in comparison to gzip.
214 Bzip2 uses a large amount of memory. For modern kernels you
215 will need at least 8MB RAM or more for booting.
219 depends on HAVE_KERNEL_LZMA
221 This compression algorithm's ratio is best. Decompression speed
222 is between gzip and bzip2. Compression is slowest.
223 The kernel size is about 33% smaller with LZMA in comparison to gzip.
227 depends on HAVE_KERNEL_XZ
229 XZ uses the LZMA2 algorithm and instruction set specific
230 BCJ filters which can improve compression ratio of executable
231 code. The size of the kernel is about 30% smaller with XZ in
232 comparison to gzip. On architectures for which there is a BCJ
233 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
234 will create a few percent smaller kernel than plain LZMA.
236 The speed is about the same as with LZMA: The decompression
237 speed of XZ is better than that of bzip2 but worse than gzip
238 and LZO. Compression is slow.
242 depends on HAVE_KERNEL_LZO
244 Its compression ratio is the poorest among the choices. The kernel
245 size is about 10% bigger than gzip; however its speed
246 (both compression and decompression) is the fastest.
250 depends on HAVE_KERNEL_LZ4
252 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
253 A preliminary version of LZ4 de/compression tool is available at
254 <https://code.google.com/p/lz4/>.
256 Its compression ratio is worse than LZO. The size of the kernel
257 is about 8% bigger than LZO. But the decompression speed is
260 config KERNEL_UNCOMPRESSED
262 depends on HAVE_KERNEL_UNCOMPRESSED
264 Produce uncompressed kernel image. This option is usually not what
265 you want. It is useful for debugging the kernel in slow simulation
266 environments, where decompressing and moving the kernel is awfully
267 slow. This option allows early boot code to skip the decompressor
268 and jump right at uncompressed kernel image.
272 config DEFAULT_HOSTNAME
273 string "Default hostname"
276 This option determines the default system hostname before userspace
277 calls sethostname(2). The kernel traditionally uses "(none)" here,
278 but you may wish to use a different default here to make a minimal
279 system more usable with less configuration.
282 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
283 # add proper SWAP support to them, in which case this can be remove.
289 bool "Support for paging of anonymous memory (swap)"
290 depends on MMU && BLOCK && !ARCH_NO_SWAP
293 This option allows you to choose whether you want to have support
294 for so called swap devices or swap files in your kernel that are
295 used to provide more virtual memory than the actual RAM present
296 in your computer. If unsure say Y.
301 Inter Process Communication is a suite of library functions and
302 system calls which let processes (running programs) synchronize and
303 exchange information. It is generally considered to be a good thing,
304 and some programs won't run unless you say Y here. In particular, if
305 you want to run the DOS emulator dosemu under Linux (read the
306 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
307 you'll need to say Y here.
309 You can find documentation about IPC with "info ipc" and also in
310 section 6.4 of the Linux Programmer's Guide, available from
311 <http://www.tldp.org/guides.html>.
313 config SYSVIPC_SYSCTL
320 bool "POSIX Message Queues"
323 POSIX variant of message queues is a part of IPC. In POSIX message
324 queues every message has a priority which decides about succession
325 of receiving it by a process. If you want to compile and run
326 programs written e.g. for Solaris with use of its POSIX message
327 queues (functions mq_*) say Y here.
329 POSIX message queues are visible as a filesystem called 'mqueue'
330 and can be mounted somewhere if you want to do filesystem
331 operations on message queues.
335 config POSIX_MQUEUE_SYSCTL
337 depends on POSIX_MQUEUE
341 config CROSS_MEMORY_ATTACH
342 bool "Enable process_vm_readv/writev syscalls"
346 Enabling this option adds the system calls process_vm_readv and
347 process_vm_writev which allow a process with the correct privileges
348 to directly read from or write to another process' address space.
349 See the man page for more details.
352 bool "uselib syscall"
353 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
355 This option enables the uselib syscall, a system call used in the
356 dynamic linker from libc5 and earlier. glibc does not use this
357 system call. If you intend to run programs built on libc5 or
358 earlier, you may need to enable this syscall. Current systems
359 running glibc can safely disable this.
362 bool "Auditing support"
365 Enable auditing infrastructure that can be used with another
366 kernel subsystem, such as SELinux (which requires this for
367 logging of avc messages output). System call auditing is included
368 on architectures which support it.
370 config HAVE_ARCH_AUDITSYSCALL
375 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
378 source "kernel/irq/Kconfig"
379 source "kernel/time/Kconfig"
380 source "kernel/Kconfig.preempt"
382 menu "CPU/Task time and stats accounting"
384 config VIRT_CPU_ACCOUNTING
388 prompt "Cputime accounting"
389 default TICK_CPU_ACCOUNTING if !PPC64
390 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
392 # Kind of a stub config for the pure tick based cputime accounting
393 config TICK_CPU_ACCOUNTING
394 bool "Simple tick based cputime accounting"
395 depends on !S390 && !NO_HZ_FULL
397 This is the basic tick based cputime accounting that maintains
398 statistics about user, system and idle time spent on per jiffies
403 config VIRT_CPU_ACCOUNTING_NATIVE
404 bool "Deterministic task and CPU time accounting"
405 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
406 select VIRT_CPU_ACCOUNTING
408 Select this option to enable more accurate task and CPU time
409 accounting. This is done by reading a CPU counter on each
410 kernel entry and exit and on transitions within the kernel
411 between system, softirq and hardirq state, so there is a
412 small performance impact. In the case of s390 or IBM POWER > 5,
413 this also enables accounting of stolen time on logically-partitioned
416 config VIRT_CPU_ACCOUNTING_GEN
417 bool "Full dynticks CPU time accounting"
418 depends on HAVE_CONTEXT_TRACKING
419 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
420 depends on GENERIC_CLOCKEVENTS
421 select VIRT_CPU_ACCOUNTING
422 select CONTEXT_TRACKING
424 Select this option to enable task and CPU time accounting on full
425 dynticks systems. This accounting is implemented by watching every
426 kernel-user boundaries using the context tracking subsystem.
427 The accounting is thus performed at the expense of some significant
430 For now this is only useful if you are working on the full
431 dynticks subsystem development.
437 config IRQ_TIME_ACCOUNTING
438 bool "Fine granularity task level IRQ time accounting"
439 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
441 Select this option to enable fine granularity task irq time
442 accounting. This is done by reading a timestamp on each
443 transitions between softirq and hardirq state, so there can be a
444 small performance impact.
446 If in doubt, say N here.
448 config HAVE_SCHED_AVG_IRQ
450 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
453 config SCHED_THERMAL_PRESSURE
454 bool "Enable periodic averaging of thermal pressure"
457 config BSD_PROCESS_ACCT
458 bool "BSD Process Accounting"
461 If you say Y here, a user level program will be able to instruct the
462 kernel (via a special system call) to write process accounting
463 information to a file: whenever a process exits, information about
464 that process will be appended to the file by the kernel. The
465 information includes things such as creation time, owning user,
466 command name, memory usage, controlling terminal etc. (the complete
467 list is in the struct acct in <file:include/linux/acct.h>). It is
468 up to the user level program to do useful things with this
469 information. This is generally a good idea, so say Y.
471 config BSD_PROCESS_ACCT_V3
472 bool "BSD Process Accounting version 3 file format"
473 depends on BSD_PROCESS_ACCT
476 If you say Y here, the process accounting information is written
477 in a new file format that also logs the process IDs of each
478 process and its parent. Note that this file format is incompatible
479 with previous v0/v1/v2 file formats, so you will need updated tools
480 for processing it. A preliminary version of these tools is available
481 at <http://www.gnu.org/software/acct/>.
484 bool "Export task/process statistics through netlink"
489 Export selected statistics for tasks/processes through the
490 generic netlink interface. Unlike BSD process accounting, the
491 statistics are available during the lifetime of tasks/processes as
492 responses to commands. Like BSD accounting, they are sent to user
497 config TASK_DELAY_ACCT
498 bool "Enable per-task delay accounting"
502 Collect information on time spent by a task waiting for system
503 resources like cpu, synchronous block I/O completion and swapping
504 in pages. Such statistics can help in setting a task's priorities
505 relative to other tasks for cpu, io, rss limits etc.
510 bool "Enable extended accounting over taskstats"
513 Collect extended task accounting data and send the data
514 to userland for processing over the taskstats interface.
518 config TASK_IO_ACCOUNTING
519 bool "Enable per-task storage I/O accounting"
520 depends on TASK_XACCT
522 Collect information on the number of bytes of storage I/O which this
528 bool "Pressure stall information tracking"
530 Collect metrics that indicate how overcommitted the CPU, memory,
531 and IO capacity are in the system.
533 If you say Y here, the kernel will create /proc/pressure/ with the
534 pressure statistics files cpu, memory, and io. These will indicate
535 the share of walltime in which some or all tasks in the system are
536 delayed due to contention of the respective resource.
538 In kernels with cgroup support, cgroups (cgroup2 only) will
539 have cpu.pressure, memory.pressure, and io.pressure files,
540 which aggregate pressure stalls for the grouped tasks only.
542 For more details see Documentation/accounting/psi.rst.
546 config PSI_DEFAULT_DISABLED
547 bool "Require boot parameter to enable pressure stall information tracking"
551 If set, pressure stall information tracking will be disabled
552 per default but can be enabled through passing psi=1 on the
553 kernel commandline during boot.
555 This feature adds some code to the task wakeup and sleep
556 paths of the scheduler. The overhead is too low to affect
557 common scheduling-intense workloads in practice (such as
558 webservers, memcache), but it does show up in artificial
559 scheduler stress tests, such as hackbench.
561 If you are paranoid and not sure what the kernel will be
566 endmenu # "CPU/Task time and stats accounting"
570 depends on SMP || COMPILE_TEST
573 Make sure that CPUs running critical tasks are not disturbed by
574 any source of "noise" such as unbound workqueues, timers, kthreads...
575 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
576 the "isolcpus=" boot parameter.
580 source "kernel/rcu/Kconfig"
587 tristate "Kernel .config support"
589 This option enables the complete Linux kernel ".config" file
590 contents to be saved in the kernel. It provides documentation
591 of which kernel options are used in a running kernel or in an
592 on-disk kernel. This information can be extracted from the kernel
593 image file with the script scripts/extract-ikconfig and used as
594 input to rebuild the current kernel or to build another kernel.
595 It can also be extracted from a running kernel by reading
596 /proc/config.gz if enabled (below).
599 bool "Enable access to .config through /proc/config.gz"
600 depends on IKCONFIG && PROC_FS
602 This option enables access to the kernel configuration file
603 through /proc/config.gz.
606 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
609 This option enables access to the in-kernel headers that are generated during
610 the build process. These can be used to build eBPF tracing programs,
611 or similar programs. If you build the headers as a module, a module called
612 kheaders.ko is built which can be loaded on-demand to get access to headers.
615 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
620 Select the minimal kernel log buffer size as a power of 2.
621 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
622 parameter, see below. Any higher size also might be forced
623 by "log_buf_len" boot parameter.
633 config LOG_CPU_MAX_BUF_SHIFT
634 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
637 default 12 if !BASE_SMALL
638 default 0 if BASE_SMALL
641 This option allows to increase the default ring buffer size
642 according to the number of CPUs. The value defines the contribution
643 of each CPU as a power of 2. The used space is typically only few
644 lines however it might be much more when problems are reported,
647 The increased size means that a new buffer has to be allocated and
648 the original static one is unused. It makes sense only on systems
649 with more CPUs. Therefore this value is used only when the sum of
650 contributions is greater than the half of the default kernel ring
651 buffer as defined by LOG_BUF_SHIFT. The default values are set
652 so that more than 64 CPUs are needed to trigger the allocation.
654 Also this option is ignored when "log_buf_len" kernel parameter is
655 used as it forces an exact (power of two) size of the ring buffer.
657 The number of possible CPUs is used for this computation ignoring
658 hotplugging making the computation optimal for the worst case
659 scenario while allowing a simple algorithm to be used from bootup.
661 Examples shift values and their meaning:
662 17 => 128 KB for each CPU
663 16 => 64 KB for each CPU
664 15 => 32 KB for each CPU
665 14 => 16 KB for each CPU
666 13 => 8 KB for each CPU
667 12 => 4 KB for each CPU
669 config PRINTK_SAFE_LOG_BUF_SHIFT
670 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
675 Select the size of an alternate printk per-CPU buffer where messages
676 printed from usafe contexts are temporary stored. One example would
677 be NMI messages, another one - printk recursion. The messages are
678 copied to the main log buffer in a safe context to avoid a deadlock.
679 The value defines the size as a power of 2.
681 Those messages are rare and limited. The largest one is when
682 a backtrace is printed. It usually fits into 4KB. Select
683 8KB if you want to be on the safe side.
686 17 => 128 KB for each CPU
687 16 => 64 KB for each CPU
688 15 => 32 KB for each CPU
689 14 => 16 KB for each CPU
690 13 => 8 KB for each CPU
691 12 => 4 KB for each CPU
694 # Architectures with an unreliable sched_clock() should select this:
696 config HAVE_UNSTABLE_SCHED_CLOCK
699 config GENERIC_SCHED_CLOCK
702 menu "Scheduler features"
705 bool "Enable utilization clamping for RT/FAIR tasks"
706 depends on CPU_FREQ_GOV_SCHEDUTIL
708 This feature enables the scheduler to track the clamped utilization
709 of each CPU based on RUNNABLE tasks scheduled on that CPU.
711 With this option, the user can specify the min and max CPU
712 utilization allowed for RUNNABLE tasks. The max utilization defines
713 the maximum frequency a task should use while the min utilization
714 defines the minimum frequency it should use.
716 Both min and max utilization clamp values are hints to the scheduler,
717 aiming at improving its frequency selection policy, but they do not
718 enforce or grant any specific bandwidth for tasks.
722 config UCLAMP_BUCKETS_COUNT
723 int "Number of supported utilization clamp buckets"
726 depends on UCLAMP_TASK
728 Defines the number of clamp buckets to use. The range of each bucket
729 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
730 number of clamp buckets the finer their granularity and the higher
731 the precision of clamping aggregation and tracking at run-time.
733 For example, with the minimum configuration value we will have 5
734 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
735 be refcounted in the [20..39]% bucket and will set the bucket clamp
736 effective value to 25%.
737 If a second 30% boosted task should be co-scheduled on the same CPU,
738 that task will be refcounted in the same bucket of the first task and
739 it will boost the bucket clamp effective value to 30%.
740 The clamp effective value of a bucket is reset to its nominal value
741 (20% in the example above) when there are no more tasks refcounted in
744 An additional boost/capping margin can be added to some tasks. In the
745 example above the 25% task will be boosted to 30% until it exits the
746 CPU. If that should be considered not acceptable on certain systems,
747 it's always possible to reduce the margin by increasing the number of
748 clamp buckets to trade off used memory for run-time tracking
751 If in doubt, use the default value.
756 # For architectures that want to enable the support for NUMA-affine scheduler
759 config ARCH_SUPPORTS_NUMA_BALANCING
763 # For architectures that prefer to flush all TLBs after a number of pages
764 # are unmapped instead of sending one IPI per page to flush. The architecture
765 # must provide guarantees on what happens if a clean TLB cache entry is
766 # written after the unmap. Details are in mm/rmap.c near the check for
767 # should_defer_flush. The architecture should also consider if the full flush
768 # and the refill costs are offset by the savings of sending fewer IPIs.
769 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
773 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
776 # For architectures that know their GCC __int128 support is sound
778 config ARCH_SUPPORTS_INT128
781 # For architectures that (ab)use NUMA to represent different memory regions
782 # all cpu-local but of different latencies, such as SuperH.
784 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
787 config NUMA_BALANCING
788 bool "Memory placement aware NUMA scheduler"
789 depends on ARCH_SUPPORTS_NUMA_BALANCING
790 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
791 depends on SMP && NUMA && MIGRATION
793 This option adds support for automatic NUMA aware memory/task placement.
794 The mechanism is quite primitive and is based on migrating memory when
795 it has references to the node the task is running on.
797 This system will be inactive on UMA systems.
799 config NUMA_BALANCING_DEFAULT_ENABLED
800 bool "Automatically enable NUMA aware memory/task placement"
802 depends on NUMA_BALANCING
804 If set, automatic NUMA balancing will be enabled if running on a NUMA
808 bool "Control Group support"
811 This option adds support for grouping sets of processes together, for
812 use with process control subsystems such as Cpusets, CFS, memory
813 controls or device isolation.
815 - Documentation/scheduler/sched-design-CFS.rst (CFS)
816 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
817 and resource control)
827 bool "Memory controller"
831 Provides control over the memory footprint of tasks in a cgroup.
834 bool "Swap controller"
835 depends on MEMCG && SWAP
837 Provides control over the swap space consumed by tasks in a cgroup.
839 config MEMCG_SWAP_ENABLED
840 bool "Swap controller enabled by default"
841 depends on MEMCG_SWAP
844 Memory Resource Controller Swap Extension comes with its price in
845 a bigger memory consumption. General purpose distribution kernels
846 which want to enable the feature but keep it disabled by default
847 and let the user enable it by swapaccount=1 boot command line
848 parameter should have this option unselected.
849 For those who want to have the feature enabled by default should
850 select this option (if, for some reason, they need to disable it
851 then swapaccount=0 does the trick).
855 depends on MEMCG && !SLOB
863 Generic block IO controller cgroup interface. This is the common
864 cgroup interface which should be used by various IO controlling
867 Currently, CFQ IO scheduler uses it to recognize task groups and
868 control disk bandwidth allocation (proportional time slice allocation)
869 to such task groups. It is also used by bio throttling logic in
870 block layer to implement upper limit in IO rates on a device.
872 This option only enables generic Block IO controller infrastructure.
873 One needs to also enable actual IO controlling logic/policy. For
874 enabling proportional weight division of disk bandwidth in CFQ, set
875 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
876 CONFIG_BLK_DEV_THROTTLING=y.
878 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
880 config CGROUP_WRITEBACK
882 depends on MEMCG && BLK_CGROUP
885 menuconfig CGROUP_SCHED
886 bool "CPU controller"
889 This feature lets CPU scheduler recognize task groups and control CPU
890 bandwidth allocation to such task groups. It uses cgroups to group
894 config FAIR_GROUP_SCHED
895 bool "Group scheduling for SCHED_OTHER"
896 depends on CGROUP_SCHED
900 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
901 depends on FAIR_GROUP_SCHED
904 This option allows users to define CPU bandwidth rates (limits) for
905 tasks running within the fair group scheduler. Groups with no limit
906 set are considered to be unconstrained and will run with no
908 See Documentation/scheduler/sched-bwc.rst for more information.
910 config RT_GROUP_SCHED
911 bool "Group scheduling for SCHED_RR/FIFO"
912 depends on CGROUP_SCHED
915 This feature lets you explicitly allocate real CPU bandwidth
916 to task groups. If enabled, it will also make it impossible to
917 schedule realtime tasks for non-root users until you allocate
918 realtime bandwidth for them.
919 See Documentation/scheduler/sched-rt-group.rst for more information.
923 config UCLAMP_TASK_GROUP
924 bool "Utilization clamping per group of tasks"
925 depends on CGROUP_SCHED
926 depends on UCLAMP_TASK
929 This feature enables the scheduler to track the clamped utilization
930 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
932 When this option is enabled, the user can specify a min and max
933 CPU bandwidth which is allowed for each single task in a group.
934 The max bandwidth allows to clamp the maximum frequency a task
935 can use, while the min bandwidth allows to define a minimum
936 frequency a task will always use.
938 When task group based utilization clamping is enabled, an eventually
939 specified task-specific clamp value is constrained by the cgroup
940 specified clamp value. Both minimum and maximum task clamping cannot
941 be bigger than the corresponding clamping defined at task group level.
946 bool "PIDs controller"
948 Provides enforcement of process number limits in the scope of a
949 cgroup. Any attempt to fork more processes than is allowed in the
950 cgroup will fail. PIDs are fundamentally a global resource because it
951 is fairly trivial to reach PID exhaustion before you reach even a
952 conservative kmemcg limit. As a result, it is possible to grind a
953 system to halt without being limited by other cgroup policies. The
954 PIDs controller is designed to stop this from happening.
956 It should be noted that organisational operations (such as attaching
957 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
958 since the PIDs limit only affects a process's ability to fork, not to
962 bool "RDMA controller"
964 Provides enforcement of RDMA resources defined by IB stack.
965 It is fairly easy for consumers to exhaust RDMA resources, which
966 can result into resource unavailability to other consumers.
967 RDMA controller is designed to stop this from happening.
968 Attaching processes with active RDMA resources to the cgroup
969 hierarchy is allowed even if can cross the hierarchy's limit.
971 config CGROUP_FREEZER
972 bool "Freezer controller"
974 Provides a way to freeze and unfreeze all tasks in a
977 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
978 controller includes important in-kernel memory consumers per default.
980 If you're using cgroup2, say N.
982 config CGROUP_HUGETLB
983 bool "HugeTLB controller"
984 depends on HUGETLB_PAGE
988 Provides a cgroup controller for HugeTLB pages.
989 When you enable this, you can put a per cgroup limit on HugeTLB usage.
990 The limit is enforced during page fault. Since HugeTLB doesn't
991 support page reclaim, enforcing the limit at page fault time implies
992 that, the application will get SIGBUS signal if it tries to access
993 HugeTLB pages beyond its limit. This requires the application to know
994 beforehand how much HugeTLB pages it would require for its use. The
995 control group is tracked in the third page lru pointer. This means
996 that we cannot use the controller with huge page less than 3 pages.
999 bool "Cpuset controller"
1002 This option will let you create and manage CPUSETs which
1003 allow dynamically partitioning a system into sets of CPUs and
1004 Memory Nodes and assigning tasks to run only within those sets.
1005 This is primarily useful on large SMP or NUMA systems.
1009 config PROC_PID_CPUSET
1010 bool "Include legacy /proc/<pid>/cpuset file"
1014 config CGROUP_DEVICE
1015 bool "Device controller"
1017 Provides a cgroup controller implementing whitelists for
1018 devices which a process in the cgroup can mknod or open.
1020 config CGROUP_CPUACCT
1021 bool "Simple CPU accounting controller"
1023 Provides a simple controller for monitoring the
1024 total CPU consumed by the tasks in a cgroup.
1027 bool "Perf controller"
1028 depends on PERF_EVENTS
1030 This option extends the perf per-cpu mode to restrict monitoring
1031 to threads which belong to the cgroup specified and run on the
1037 bool "Support for eBPF programs attached to cgroups"
1038 depends on BPF_SYSCALL
1039 select SOCK_CGROUP_DATA
1041 Allow attaching eBPF programs to a cgroup using the bpf(2)
1042 syscall command BPF_PROG_ATTACH.
1044 In which context these programs are accessed depends on the type
1045 of attachment. For instance, programs that are attached using
1046 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1050 bool "Debug controller"
1052 depends on DEBUG_KERNEL
1054 This option enables a simple controller that exports
1055 debugging information about the cgroups framework. This
1056 controller is for control cgroup debugging only. Its
1057 interfaces are not stable.
1061 config SOCK_CGROUP_DATA
1067 menuconfig NAMESPACES
1068 bool "Namespaces support" if EXPERT
1069 depends on MULTIUSER
1072 Provides the way to make tasks work with different objects using
1073 the same id. For example same IPC id may refer to different objects
1074 or same user id or pid may refer to different tasks when used in
1075 different namespaces.
1080 bool "UTS namespace"
1083 In this namespace tasks see different info provided with the
1087 bool "TIME namespace"
1088 depends on GENERIC_VDSO_TIME_NS
1091 In this namespace boottime and monotonic clocks can be set.
1092 The time will keep going with the same pace.
1095 bool "IPC namespace"
1096 depends on (SYSVIPC || POSIX_MQUEUE)
1099 In this namespace tasks work with IPC ids which correspond to
1100 different IPC objects in different namespaces.
1103 bool "User namespace"
1106 This allows containers, i.e. vservers, to use user namespaces
1107 to provide different user info for different servers.
1109 When user namespaces are enabled in the kernel it is
1110 recommended that the MEMCG option also be enabled and that
1111 user-space use the memory control groups to limit the amount
1112 of memory a memory unprivileged users can use.
1117 bool "PID Namespaces"
1120 Support process id namespaces. This allows having multiple
1121 processes with the same pid as long as they are in different
1122 pid namespaces. This is a building block of containers.
1125 bool "Network namespace"
1129 Allow user space to create what appear to be multiple instances
1130 of the network stack.
1134 config CHECKPOINT_RESTORE
1135 bool "Checkpoint/restore support"
1136 select PROC_CHILDREN
1139 Enables additional kernel features in a sake of checkpoint/restore.
1140 In particular it adds auxiliary prctl codes to setup process text,
1141 data and heap segment sizes, and a few additional /proc filesystem
1144 If unsure, say N here.
1146 config SCHED_AUTOGROUP
1147 bool "Automatic process group scheduling"
1150 select FAIR_GROUP_SCHED
1152 This option optimizes the scheduler for common desktop workloads by
1153 automatically creating and populating task groups. This separation
1154 of workloads isolates aggressive CPU burners (like build jobs) from
1155 desktop applications. Task group autogeneration is currently based
1158 config SYSFS_DEPRECATED
1159 bool "Enable deprecated sysfs features to support old userspace tools"
1163 This option adds code that switches the layout of the "block" class
1164 devices, to not show up in /sys/class/block/, but only in
1167 This switch is only active when the sysfs.deprecated=1 boot option is
1168 passed or the SYSFS_DEPRECATED_V2 option is set.
1170 This option allows new kernels to run on old distributions and tools,
1171 which might get confused by /sys/class/block/. Since 2007/2008 all
1172 major distributions and tools handle this just fine.
1174 Recent distributions and userspace tools after 2009/2010 depend on
1175 the existence of /sys/class/block/, and will not work with this
1178 Only if you are using a new kernel on an old distribution, you might
1181 config SYSFS_DEPRECATED_V2
1182 bool "Enable deprecated sysfs features by default"
1185 depends on SYSFS_DEPRECATED
1187 Enable deprecated sysfs by default.
1189 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1192 Only if you are using a new kernel on an old distribution, you might
1193 need to say Y here. Even then, odds are you would not need it
1194 enabled, you can always pass the boot option if absolutely necessary.
1197 bool "Kernel->user space relay support (formerly relayfs)"
1200 This option enables support for relay interface support in
1201 certain file systems (such as debugfs).
1202 It is designed to provide an efficient mechanism for tools and
1203 facilities to relay large amounts of data from kernel space to
1208 config BLK_DEV_INITRD
1209 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1211 The initial RAM filesystem is a ramfs which is loaded by the
1212 boot loader (loadlin or lilo) and that is mounted as root
1213 before the normal boot procedure. It is typically used to
1214 load modules needed to mount the "real" root file system,
1215 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1217 If RAM disk support (BLK_DEV_RAM) is also included, this
1218 also enables initial RAM disk (initrd) support and adds
1219 15 Kbytes (more on some other architectures) to the kernel size.
1225 source "usr/Kconfig"
1230 bool "Boot config support"
1231 select BLK_DEV_INITRD
1233 Extra boot config allows system admin to pass a config file as
1234 complemental extension of kernel cmdline when booting.
1235 The boot config file must be attached at the end of initramfs
1236 with checksum, size and magic word.
1237 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1242 prompt "Compiler optimization level"
1243 default CC_OPTIMIZE_FOR_PERFORMANCE
1245 config CC_OPTIMIZE_FOR_PERFORMANCE
1246 bool "Optimize for performance (-O2)"
1248 This is the default optimization level for the kernel, building
1249 with the "-O2" compiler flag for best performance and most
1250 helpful compile-time warnings.
1252 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1253 bool "Optimize more for performance (-O3)"
1255 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1257 Choosing this option will pass "-O3" to your compiler to optimize
1258 the kernel yet more for performance.
1260 config CC_OPTIMIZE_FOR_SIZE
1261 bool "Optimize for size (-Os)"
1262 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1264 Choosing this option will pass "-Os" to your compiler resulting
1265 in a smaller kernel.
1269 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1272 This requires that the arch annotates or otherwise protects
1273 its external entry points from being discarded. Linker scripts
1274 must also merge .text.*, .data.*, and .bss.* correctly into
1275 output sections. Care must be taken not to pull in unrelated
1276 sections (e.g., '.text.init'). Typically '.' in section names
1277 is used to distinguish them from label names / C identifiers.
1279 config LD_DEAD_CODE_DATA_ELIMINATION
1280 bool "Dead code and data elimination (EXPERIMENTAL)"
1281 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1283 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1284 depends on $(cc-option,-ffunction-sections -fdata-sections)
1285 depends on $(ld-option,--gc-sections)
1287 Enable this if you want to do dead code and data elimination with
1288 the linker by compiling with -ffunction-sections -fdata-sections,
1289 and linking with --gc-sections.
1291 This can reduce on disk and in-memory size of the kernel
1292 code and static data, particularly for small configs and
1293 on small systems. This has the possibility of introducing
1294 silently broken kernel if the required annotations are not
1295 present. This option is not well tested yet, so use at your
1304 config SYSCTL_EXCEPTION_TRACE
1307 Enable support for /proc/sys/debug/exception-trace.
1309 config SYSCTL_ARCH_UNALIGN_NO_WARN
1312 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1313 Allows arch to define/use @no_unaligned_warning to possibly warn
1314 about unaligned access emulation going on under the hood.
1316 config SYSCTL_ARCH_UNALIGN_ALLOW
1319 Enable support for /proc/sys/kernel/unaligned-trap
1320 Allows arches to define/use @unaligned_enabled to runtime toggle
1321 the unaligned access emulation.
1322 see arch/parisc/kernel/unaligned.c for reference
1324 config HAVE_PCSPKR_PLATFORM
1327 # interpreter that classic socket filters depend on
1332 bool "Configure standard kernel features (expert users)"
1333 # Unhide debug options, to make the on-by-default options visible
1336 This option allows certain base kernel options and settings
1337 to be disabled or tweaked. This is for specialized
1338 environments which can tolerate a "non-standard" kernel.
1339 Only use this if you really know what you are doing.
1342 bool "Enable 16-bit UID system calls" if EXPERT
1343 depends on HAVE_UID16 && MULTIUSER
1346 This enables the legacy 16-bit UID syscall wrappers.
1349 bool "Multiple users, groups and capabilities support" if EXPERT
1352 This option enables support for non-root users, groups and
1355 If you say N here, all processes will run with UID 0, GID 0, and all
1356 possible capabilities. Saying N here also compiles out support for
1357 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1360 If unsure, say Y here.
1362 config SGETMASK_SYSCALL
1363 bool "sgetmask/ssetmask syscalls support" if EXPERT
1364 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1366 sys_sgetmask and sys_ssetmask are obsolete system calls
1367 no longer supported in libc but still enabled by default in some
1370 If unsure, leave the default option here.
1372 config SYSFS_SYSCALL
1373 bool "Sysfs syscall support" if EXPERT
1376 sys_sysfs is an obsolete system call no longer supported in libc.
1377 Note that disabling this option is more secure but might break
1378 compatibility with some systems.
1380 If unsure say Y here.
1383 bool "open by fhandle syscalls" if EXPERT
1387 If you say Y here, a user level program will be able to map
1388 file names to handle and then later use the handle for
1389 different file system operations. This is useful in implementing
1390 userspace file servers, which now track files using handles instead
1391 of names. The handle would remain the same even if file names
1392 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1396 bool "Posix Clocks & timers" if EXPERT
1399 This includes native support for POSIX timers to the kernel.
1400 Some embedded systems have no use for them and therefore they
1401 can be configured out to reduce the size of the kernel image.
1403 When this option is disabled, the following syscalls won't be
1404 available: timer_create, timer_gettime: timer_getoverrun,
1405 timer_settime, timer_delete, clock_adjtime, getitimer,
1406 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1407 clock_getres and clock_nanosleep syscalls will be limited to
1408 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1414 bool "Enable support for printk" if EXPERT
1417 This option enables normal printk support. Removing it
1418 eliminates most of the message strings from the kernel image
1419 and makes the kernel more or less silent. As this makes it
1420 very difficult to diagnose system problems, saying N here is
1421 strongly discouraged.
1429 bool "BUG() support" if EXPERT
1432 Disabling this option eliminates support for BUG and WARN, reducing
1433 the size of your kernel image and potentially quietly ignoring
1434 numerous fatal conditions. You should only consider disabling this
1435 option for embedded systems with no facilities for reporting errors.
1441 bool "Enable ELF core dumps" if EXPERT
1443 Enable support for generating core dumps. Disabling saves about 4k.
1446 config PCSPKR_PLATFORM
1447 bool "Enable PC-Speaker support" if EXPERT
1448 depends on HAVE_PCSPKR_PLATFORM
1452 This option allows to disable the internal PC-Speaker
1453 support, saving some memory.
1457 bool "Enable full-sized data structures for core" if EXPERT
1459 Disabling this option reduces the size of miscellaneous core
1460 kernel data structures. This saves memory on small machines,
1461 but may reduce performance.
1464 bool "Enable futex support" if EXPERT
1468 Disabling this option will cause the kernel to be built without
1469 support for "fast userspace mutexes". The resulting kernel may not
1470 run glibc-based applications correctly.
1474 depends on FUTEX && RT_MUTEXES
1477 config HAVE_FUTEX_CMPXCHG
1481 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1482 is implemented and always working. This removes a couple of runtime
1486 bool "Enable eventpoll support" if EXPERT
1489 Disabling this option will cause the kernel to be built without
1490 support for epoll family of system calls.
1493 bool "Enable signalfd() system call" if EXPERT
1496 Enable the signalfd() system call that allows to receive signals
1497 on a file descriptor.
1502 bool "Enable timerfd() system call" if EXPERT
1505 Enable the timerfd() system call that allows to receive timer
1506 events on a file descriptor.
1511 bool "Enable eventfd() system call" if EXPERT
1514 Enable the eventfd() system call that allows to receive both
1515 kernel notification (ie. KAIO) or userspace notifications.
1520 bool "Use full shmem filesystem" if EXPERT
1524 The shmem is an internal filesystem used to manage shared memory.
1525 It is backed by swap and manages resource limits. It is also exported
1526 to userspace as tmpfs if TMPFS is enabled. Disabling this
1527 option replaces shmem and tmpfs with the much simpler ramfs code,
1528 which may be appropriate on small systems without swap.
1531 bool "Enable AIO support" if EXPERT
1534 This option enables POSIX asynchronous I/O which may by used
1535 by some high performance threaded applications. Disabling
1536 this option saves about 7k.
1539 bool "Enable IO uring support" if EXPERT
1544 This option enables support for the io_uring interface, enabling
1545 applications to submit and complete IO through submission and
1546 completion rings that are shared between the kernel and application.
1548 config ADVISE_SYSCALLS
1549 bool "Enable madvise/fadvise syscalls" if EXPERT
1552 This option enables the madvise and fadvise syscalls, used by
1553 applications to advise the kernel about their future memory or file
1554 usage, improving performance. If building an embedded system where no
1555 applications use these syscalls, you can disable this option to save
1559 bool "Enable membarrier() system call" if EXPERT
1562 Enable the membarrier() system call that allows issuing memory
1563 barriers across all running threads, which can be used to distribute
1564 the cost of user-space memory barriers asymmetrically by transforming
1565 pairs of memory barriers into pairs consisting of membarrier() and a
1571 bool "Load all symbols for debugging/ksymoops" if EXPERT
1574 Say Y here to let the kernel print out symbolic crash information and
1575 symbolic stack backtraces. This increases the size of the kernel
1576 somewhat, as all symbols have to be loaded into the kernel image.
1579 bool "Include all symbols in kallsyms"
1580 depends on DEBUG_KERNEL && KALLSYMS
1582 Normally kallsyms only contains the symbols of functions for nicer
1583 OOPS messages and backtraces (i.e., symbols from the text and inittext
1584 sections). This is sufficient for most cases. And only in very rare
1585 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1586 names of variables from the data sections, etc).
1588 This option makes sure that all symbols are loaded into the kernel
1589 image (i.e., symbols from all sections) in cost of increased kernel
1590 size (depending on the kernel configuration, it may be 300KiB or
1591 something like this).
1593 Say N unless you really need all symbols.
1595 config KALLSYMS_ABSOLUTE_PERCPU
1598 default X86_64 && SMP
1600 config KALLSYMS_BASE_RELATIVE
1605 Instead of emitting them as absolute values in the native word size,
1606 emit the symbol references in the kallsyms table as 32-bit entries,
1607 each containing a relative value in the range [base, base + U32_MAX]
1608 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1609 an absolute value in the range [0, S32_MAX] or a relative value in the
1610 range [base, base + S32_MAX], where base is the lowest relative symbol
1611 address encountered in the image.
1613 On 64-bit builds, this reduces the size of the address table by 50%,
1614 but more importantly, it results in entries whose values are build
1615 time constants, and no relocation pass is required at runtime to fix
1616 up the entries based on the runtime load address of the kernel.
1618 # end of the "standard kernel features (expert users)" menu
1620 # syscall, maps, verifier
1623 bool "LSM Instrumentation with BPF"
1624 depends on BPF_EVENTS
1625 depends on BPF_SYSCALL
1629 Enables instrumentation of the security hooks with eBPF programs for
1630 implementing dynamic MAC and Audit Policies.
1632 If you are unsure how to answer this question, answer N.
1635 bool "Enable bpf() system call"
1640 Enable the bpf() system call that allows to manipulate eBPF
1641 programs and maps via file descriptors.
1643 config ARCH_WANT_DEFAULT_BPF_JIT
1646 config BPF_JIT_ALWAYS_ON
1647 bool "Permanently enable BPF JIT and remove BPF interpreter"
1648 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1650 Enables BPF JIT and removes BPF interpreter to avoid
1651 speculative execution of BPF instructions by the interpreter
1653 config BPF_JIT_DEFAULT_ON
1654 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1655 depends on HAVE_EBPF_JIT && BPF_JIT
1658 bool "Enable userfaultfd() system call"
1661 Enable the userfaultfd() system call that allows to intercept and
1662 handle page faults in userland.
1664 config ARCH_HAS_MEMBARRIER_CALLBACKS
1667 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1671 bool "Enable rseq() system call" if EXPERT
1673 depends on HAVE_RSEQ
1676 Enable the restartable sequences system call. It provides a
1677 user-space cache for the current CPU number value, which
1678 speeds up getting the current CPU number from user-space,
1679 as well as an ABI to speed up user-space operations on
1686 bool "Enabled debugging of rseq() system call" if EXPERT
1687 depends on RSEQ && DEBUG_KERNEL
1689 Enable extra debugging checks for the rseq system call.
1694 bool "Embedded system"
1695 option allnoconfig_y
1698 This option should be enabled if compiling the kernel for
1699 an embedded system so certain expert options are available
1702 config HAVE_PERF_EVENTS
1705 See tools/perf/design.txt for details.
1707 config PERF_USE_VMALLOC
1710 See tools/perf/design.txt for details
1713 bool "PC/104 support" if EXPERT
1715 Expose PC/104 form factor device drivers and options available for
1716 selection and configuration. Enable this option if your target
1717 machine has a PC/104 bus.
1719 menu "Kernel Performance Events And Counters"
1722 bool "Kernel performance events and counters"
1723 default y if PROFILING
1724 depends on HAVE_PERF_EVENTS
1728 Enable kernel support for various performance events provided
1729 by software and hardware.
1731 Software events are supported either built-in or via the
1732 use of generic tracepoints.
1734 Most modern CPUs support performance events via performance
1735 counter registers. These registers count the number of certain
1736 types of hw events: such as instructions executed, cachemisses
1737 suffered, or branches mis-predicted - without slowing down the
1738 kernel or applications. These registers can also trigger interrupts
1739 when a threshold number of events have passed - and can thus be
1740 used to profile the code that runs on that CPU.
1742 The Linux Performance Event subsystem provides an abstraction of
1743 these software and hardware event capabilities, available via a
1744 system call and used by the "perf" utility in tools/perf/. It
1745 provides per task and per CPU counters, and it provides event
1746 capabilities on top of those.
1750 config DEBUG_PERF_USE_VMALLOC
1752 bool "Debug: use vmalloc to back perf mmap() buffers"
1753 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1754 select PERF_USE_VMALLOC
1756 Use vmalloc memory to back perf mmap() buffers.
1758 Mostly useful for debugging the vmalloc code on platforms
1759 that don't require it.
1765 config VM_EVENT_COUNTERS
1767 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1769 VM event counters are needed for event counts to be shown.
1770 This option allows the disabling of the VM event counters
1771 on EXPERT systems. /proc/vmstat will only show page counts
1772 if VM event counters are disabled.
1776 bool "Enable SLUB debugging support" if EXPERT
1777 depends on SLUB && SYSFS
1779 SLUB has extensive debug support features. Disabling these can
1780 result in significant savings in code size. This also disables
1781 SLUB sysfs support. /sys/slab will not exist and there will be
1782 no support for cache validation etc.
1784 config SLUB_MEMCG_SYSFS_ON
1786 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1787 depends on SLUB && SYSFS && MEMCG
1789 SLUB creates a directory under /sys/kernel/slab for each
1790 allocation cache to host info and debug files. If memory
1791 cgroup is enabled, each cache can have per memory cgroup
1792 caches. SLUB can create the same sysfs directories for these
1793 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1794 to a very high number of debug files being created. This is
1795 controlled by slub_memcg_sysfs boot parameter and this
1796 config option determines the parameter's default value.
1799 bool "Disable heap randomization"
1802 Randomizing heap placement makes heap exploits harder, but it
1803 also breaks ancient binaries (including anything libc5 based).
1804 This option changes the bootup default to heap randomization
1805 disabled, and can be overridden at runtime by setting
1806 /proc/sys/kernel/randomize_va_space to 2.
1808 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1811 prompt "Choose SLAB allocator"
1814 This option allows to select a slab allocator.
1818 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1820 The regular slab allocator that is established and known to work
1821 well in all environments. It organizes cache hot objects in
1822 per cpu and per node queues.
1825 bool "SLUB (Unqueued Allocator)"
1826 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1828 SLUB is a slab allocator that minimizes cache line usage
1829 instead of managing queues of cached objects (SLAB approach).
1830 Per cpu caching is realized using slabs of objects instead
1831 of queues of objects. SLUB can use memory efficiently
1832 and has enhanced diagnostics. SLUB is the default choice for
1837 bool "SLOB (Simple Allocator)"
1839 SLOB replaces the stock allocator with a drastically simpler
1840 allocator. SLOB is generally more space efficient but
1841 does not perform as well on large systems.
1845 config SLAB_MERGE_DEFAULT
1846 bool "Allow slab caches to be merged"
1849 For reduced kernel memory fragmentation, slab caches can be
1850 merged when they share the same size and other characteristics.
1851 This carries a risk of kernel heap overflows being able to
1852 overwrite objects from merged caches (and more easily control
1853 cache layout), which makes such heap attacks easier to exploit
1854 by attackers. By keeping caches unmerged, these kinds of exploits
1855 can usually only damage objects in the same cache. To disable
1856 merging at runtime, "slab_nomerge" can be passed on the kernel
1859 config SLAB_FREELIST_RANDOM
1861 depends on SLAB || SLUB
1862 bool "SLAB freelist randomization"
1864 Randomizes the freelist order used on creating new pages. This
1865 security feature reduces the predictability of the kernel slab
1866 allocator against heap overflows.
1868 config SLAB_FREELIST_HARDENED
1869 bool "Harden slab freelist metadata"
1872 Many kernel heap attacks try to target slab cache metadata and
1873 other infrastructure. This options makes minor performance
1874 sacrifices to harden the kernel slab allocator against common
1875 freelist exploit methods.
1877 config SHUFFLE_PAGE_ALLOCATOR
1878 bool "Page allocator randomization"
1879 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1881 Randomization of the page allocator improves the average
1882 utilization of a direct-mapped memory-side-cache. See section
1883 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1884 6.2a specification for an example of how a platform advertises
1885 the presence of a memory-side-cache. There are also incidental
1886 security benefits as it reduces the predictability of page
1887 allocations to compliment SLAB_FREELIST_RANDOM, but the
1888 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1889 10th order of pages is selected based on cache utilization
1892 While the randomization improves cache utilization it may
1893 negatively impact workloads on platforms without a cache. For
1894 this reason, by default, the randomization is enabled only
1895 after runtime detection of a direct-mapped memory-side-cache.
1896 Otherwise, the randomization may be force enabled with the
1897 'page_alloc.shuffle' kernel command line parameter.
1901 config SLUB_CPU_PARTIAL
1903 depends on SLUB && SMP
1904 bool "SLUB per cpu partial cache"
1906 Per cpu partial caches accelerate objects allocation and freeing
1907 that is local to a processor at the price of more indeterminism
1908 in the latency of the free. On overflow these caches will be cleared
1909 which requires the taking of locks that may cause latency spikes.
1910 Typically one would choose no for a realtime system.
1912 config MMAP_ALLOW_UNINITIALIZED
1913 bool "Allow mmapped anonymous memory to be uninitialized"
1914 depends on EXPERT && !MMU
1917 Normally, and according to the Linux spec, anonymous memory obtained
1918 from mmap() has its contents cleared before it is passed to
1919 userspace. Enabling this config option allows you to request that
1920 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1921 providing a huge performance boost. If this option is not enabled,
1922 then the flag will be ignored.
1924 This is taken advantage of by uClibc's malloc(), and also by
1925 ELF-FDPIC binfmt's brk and stack allocator.
1927 Because of the obvious security issues, this option should only be
1928 enabled on embedded devices where you control what is run in
1929 userspace. Since that isn't generally a problem on no-MMU systems,
1930 it is normally safe to say Y here.
1932 See Documentation/nommu-mmap.txt for more information.
1934 config SYSTEM_DATA_VERIFICATION
1936 select SYSTEM_TRUSTED_KEYRING
1940 select ASYMMETRIC_KEY_TYPE
1941 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1944 select X509_CERTIFICATE_PARSER
1945 select PKCS7_MESSAGE_PARSER
1947 Provide PKCS#7 message verification using the contents of the system
1948 trusted keyring to provide public keys. This then can be used for
1949 module verification, kexec image verification and firmware blob
1953 bool "Profiling support"
1955 Say Y here to enable the extended profiling support mechanisms used
1956 by profilers such as OProfile.
1959 # Place an empty function call at each tracepoint site. Can be
1960 # dynamically changed for a probe function.
1965 endmenu # General setup
1967 source "arch/Kconfig"
1974 default 0 if BASE_FULL
1975 default 1 if !BASE_FULL
1977 config MODULE_SIG_FORMAT
1979 select SYSTEM_DATA_VERIFICATION
1982 bool "Enable loadable module support"
1985 Kernel modules are small pieces of compiled code which can
1986 be inserted in the running kernel, rather than being
1987 permanently built into the kernel. You use the "modprobe"
1988 tool to add (and sometimes remove) them. If you say Y here,
1989 many parts of the kernel can be built as modules (by
1990 answering M instead of Y where indicated): this is most
1991 useful for infrequently used options which are not required
1992 for booting. For more information, see the man pages for
1993 modprobe, lsmod, modinfo, insmod and rmmod.
1995 If you say Y here, you will need to run "make
1996 modules_install" to put the modules under /lib/modules/
1997 where modprobe can find them (you may need to be root to do
2004 config MODULE_FORCE_LOAD
2005 bool "Forced module loading"
2008 Allow loading of modules without version information (ie. modprobe
2009 --force). Forced module loading sets the 'F' (forced) taint flag and
2010 is usually a really bad idea.
2012 config MODULE_UNLOAD
2013 bool "Module unloading"
2015 Without this option you will not be able to unload any
2016 modules (note that some modules may not be unloadable
2017 anyway), which makes your kernel smaller, faster
2018 and simpler. If unsure, say Y.
2020 config MODULE_FORCE_UNLOAD
2021 bool "Forced module unloading"
2022 depends on MODULE_UNLOAD
2024 This option allows you to force a module to unload, even if the
2025 kernel believes it is unsafe: the kernel will remove the module
2026 without waiting for anyone to stop using it (using the -f option to
2027 rmmod). This is mainly for kernel developers and desperate users.
2031 bool "Module versioning support"
2033 Usually, you have to use modules compiled with your kernel.
2034 Saying Y here makes it sometimes possible to use modules
2035 compiled for different kernels, by adding enough information
2036 to the modules to (hopefully) spot any changes which would
2037 make them incompatible with the kernel you are running. If
2040 config ASM_MODVERSIONS
2042 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2044 This enables module versioning for exported symbols also from
2045 assembly. This can be enabled only when the target architecture
2048 config MODULE_REL_CRCS
2050 depends on MODVERSIONS
2052 config MODULE_SRCVERSION_ALL
2053 bool "Source checksum for all modules"
2055 Modules which contain a MODULE_VERSION get an extra "srcversion"
2056 field inserted into their modinfo section, which contains a
2057 sum of the source files which made it. This helps maintainers
2058 see exactly which source was used to build a module (since
2059 others sometimes change the module source without updating
2060 the version). With this option, such a "srcversion" field
2061 will be created for all modules. If unsure, say N.
2064 bool "Module signature verification"
2065 select MODULE_SIG_FORMAT
2067 Check modules for valid signatures upon load: the signature
2068 is simply appended to the module. For more information see
2069 <file:Documentation/admin-guide/module-signing.rst>.
2071 Note that this option adds the OpenSSL development packages as a
2072 kernel build dependency so that the signing tool can use its crypto
2075 You should enable this option if you wish to use either
2076 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2077 another LSM - otherwise unsigned modules will be loadable regardless
2078 of the lockdown policy.
2080 !!!WARNING!!! If you enable this option, you MUST make sure that the
2081 module DOES NOT get stripped after being signed. This includes the
2082 debuginfo strip done by some packagers (such as rpmbuild) and
2083 inclusion into an initramfs that wants the module size reduced.
2085 config MODULE_SIG_FORCE
2086 bool "Require modules to be validly signed"
2087 depends on MODULE_SIG
2089 Reject unsigned modules or signed modules for which we don't have a
2090 key. Without this, such modules will simply taint the kernel.
2092 config MODULE_SIG_ALL
2093 bool "Automatically sign all modules"
2095 depends on MODULE_SIG
2097 Sign all modules during make modules_install. Without this option,
2098 modules must be signed manually, using the scripts/sign-file tool.
2100 comment "Do not forget to sign required modules with scripts/sign-file"
2101 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2104 prompt "Which hash algorithm should modules be signed with?"
2105 depends on MODULE_SIG
2107 This determines which sort of hashing algorithm will be used during
2108 signature generation. This algorithm _must_ be built into the kernel
2109 directly so that signature verification can take place. It is not
2110 possible to load a signed module containing the algorithm to check
2111 the signature on that module.
2113 config MODULE_SIG_SHA1
2114 bool "Sign modules with SHA-1"
2117 config MODULE_SIG_SHA224
2118 bool "Sign modules with SHA-224"
2119 select CRYPTO_SHA256
2121 config MODULE_SIG_SHA256
2122 bool "Sign modules with SHA-256"
2123 select CRYPTO_SHA256
2125 config MODULE_SIG_SHA384
2126 bool "Sign modules with SHA-384"
2127 select CRYPTO_SHA512
2129 config MODULE_SIG_SHA512
2130 bool "Sign modules with SHA-512"
2131 select CRYPTO_SHA512
2135 config MODULE_SIG_HASH
2137 depends on MODULE_SIG
2138 default "sha1" if MODULE_SIG_SHA1
2139 default "sha224" if MODULE_SIG_SHA224
2140 default "sha256" if MODULE_SIG_SHA256
2141 default "sha384" if MODULE_SIG_SHA384
2142 default "sha512" if MODULE_SIG_SHA512
2144 config MODULE_COMPRESS
2145 bool "Compress modules on installation"
2148 Compresses kernel modules when 'make modules_install' is run; gzip or
2149 xz depending on "Compression algorithm" below.
2151 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2153 Out-of-tree kernel modules installed using Kbuild will also be
2154 compressed upon installation.
2156 Note: for modules inside an initrd or initramfs, it's more efficient
2157 to compress the whole initrd or initramfs instead.
2159 Note: This is fully compatible with signed modules.
2164 prompt "Compression algorithm"
2165 depends on MODULE_COMPRESS
2166 default MODULE_COMPRESS_GZIP
2168 This determines which sort of compression will be used during
2169 'make modules_install'.
2171 GZIP (default) and XZ are supported.
2173 config MODULE_COMPRESS_GZIP
2176 config MODULE_COMPRESS_XZ
2181 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2182 bool "Allow loading of modules with missing namespace imports"
2184 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2185 a namespace. A module that makes use of a symbol exported with such a
2186 namespace is required to import the namespace via MODULE_IMPORT_NS().
2187 There is no technical reason to enforce correct namespace imports,
2188 but it creates consistency between symbols defining namespaces and
2189 users importing namespaces they make use of. This option relaxes this
2190 requirement and lifts the enforcement when loading a module.
2194 config UNUSED_SYMBOLS
2195 bool "Enable unused/obsolete exported symbols"
2198 Unused but exported symbols make the kernel needlessly bigger. For
2199 that reason most of these unused exports will soon be removed. This
2200 option is provided temporarily to provide a transition period in case
2201 some external kernel module needs one of these symbols anyway. If you
2202 encounter such a case in your module, consider if you are actually
2203 using the right API. (rationale: since nobody in the kernel is using
2204 this in a module, there is a pretty good chance it's actually the
2205 wrong interface to use). If you really need the symbol, please send a
2206 mail to the linux kernel mailing list mentioning the symbol and why
2207 you really need it, and what the merge plan to the mainline kernel for
2210 config TRIM_UNUSED_KSYMS
2211 bool "Trim unused exported kernel symbols"
2212 depends on !UNUSED_SYMBOLS
2214 The kernel and some modules make many symbols available for
2215 other modules to use via EXPORT_SYMBOL() and variants. Depending
2216 on the set of modules being selected in your kernel configuration,
2217 many of those exported symbols might never be used.
2219 This option allows for unused exported symbols to be dropped from
2220 the build. In turn, this provides the compiler more opportunities
2221 (especially when using LTO) for optimizing the code and reducing
2222 binary size. This might have some security advantages as well.
2224 If unsure, or if you need to build out-of-tree modules, say N.
2226 config UNUSED_KSYMS_WHITELIST
2227 string "Whitelist of symbols to keep in ksymtab"
2228 depends on TRIM_UNUSED_KSYMS
2230 By default, all unused exported symbols will be un-exported from the
2231 build when TRIM_UNUSED_KSYMS is selected.
2233 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2234 exported at all times, even in absence of in-tree users. The value to
2235 set here is the path to a text file containing the list of symbols,
2236 one per line. The path can be absolute, or relative to the kernel
2241 config MODULES_TREE_LOOKUP
2243 depends on PERF_EVENTS || TRACING
2245 config INIT_ALL_POSSIBLE
2248 Back when each arch used to define their own cpu_online_mask and
2249 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2250 with all 1s, and others with all 0s. When they were centralised,
2251 it was better to provide this option than to break all the archs
2252 and have several arch maintainers pursuing me down dark alleys.
2254 source "block/Kconfig"
2256 config PREEMPT_NOTIFIERS
2266 Build a simple ASN.1 grammar compiler that produces a bytecode output
2267 that can be interpreted by the ASN.1 stream decoder and used to
2268 inform it as to what tags are to be expected in a stream and what
2269 functions to call on what tags.
2271 source "kernel/Kconfig.locks"
2273 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2276 # It may be useful for an architecture to override the definitions of the
2277 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2278 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2279 # different calling convention for syscalls. They can also override the
2280 # macros for not-implemented syscalls in kernel/sys_ni.c and
2281 # kernel/time/posix-stubs.c. All these overrides need to be available in
2282 # <asm/syscall_wrapper.h>.
2283 config ARCH_HAS_SYSCALL_WRAPPER