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)"
11 config CC_VERSION_TEXT
13 default "$(CC_VERSION_TEXT)"
15 This is used in unclear ways:
17 - Re-run Kconfig when the compiler is updated
18 The 'default' property references the environment variable,
19 CC_VERSION_TEXT so it is recorded in include/config/auto.conf.cmd.
20 When the compiler is updated, Kconfig will be invoked.
22 - Ensure full rebuild when the compier is updated
23 include/linux/kconfig.h contains this option in the comment line so
24 fixdep adds include/config/cc/version/text.h into the auto-generated
25 dependency. When the compiler is updated, syncconfig will touch it
26 and then every file will be rebuilt.
29 def_bool $(success,test "$(cc-name)" = GCC)
33 default $(cc-version) if CC_IS_GCC
37 def_bool $(success,test "$(cc-name)" = Clang)
41 default $(cc-version) if CC_IS_CLANG
45 def_bool $(success,test "$(ld-name)" = BFD)
49 default $(ld-version) if LD_IS_BFD
53 def_bool $(success,test "$(ld-name)" = LLD)
57 default $(ld-version) if LD_IS_LLD
62 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
63 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
65 config CC_CAN_LINK_STATIC
67 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
68 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
70 config CC_HAS_ASM_GOTO
71 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
73 config CC_HAS_ASM_GOTO_OUTPUT
74 depends on CC_HAS_ASM_GOTO
75 def_bool $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
77 config TOOLS_SUPPORT_RELR
78 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
80 config CC_HAS_ASM_INLINE
81 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
89 config BUILDTIME_TABLE_SORT
92 config THREAD_INFO_IN_TASK
95 Select this to move thread_info off the stack into task_struct. To
96 make this work, an arch will need to remove all thread_info fields
97 except flags and fix any runtime bugs.
99 One subtle change that will be needed is to use try_get_task_stack()
100 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
109 depends on BROKEN || !SMP
112 config INIT_ENV_ARG_LIMIT
117 Maximum of each of the number of arguments and environment
118 variables passed to init from the kernel command line.
121 bool "Compile also drivers which will not load"
122 depends on !UML && !S390
125 Some drivers can be compiled on a different platform than they are
126 intended to be run on. Despite they cannot be loaded there (or even
127 when they load they cannot be used due to missing HW support),
128 developers still, opposing to distributors, might want to build such
129 drivers to compile-test them.
131 If you are a developer and want to build everything available, say Y
132 here. If you are a user/distributor, say N here to exclude useless
133 drivers to be distributed.
135 config UAPI_HEADER_TEST
136 bool "Compile test UAPI headers"
137 depends on HEADERS_INSTALL && CC_CAN_LINK
139 Compile test headers exported to user-space to ensure they are
140 self-contained, i.e. compilable as standalone units.
142 If you are a developer or tester and want to ensure the exported
143 headers are self-contained, say Y here. Otherwise, choose N.
146 string "Local version - append to kernel release"
148 Append an extra string to the end of your kernel version.
149 This will show up when you type uname, for example.
150 The string you set here will be appended after the contents of
151 any files with a filename matching localversion* in your
152 object and source tree, in that order. Your total string can
153 be a maximum of 64 characters.
155 config LOCALVERSION_AUTO
156 bool "Automatically append version information to the version string"
158 depends on !COMPILE_TEST
160 This will try to automatically determine if the current tree is a
161 release tree by looking for git tags that belong to the current
162 top of tree revision.
164 A string of the format -gxxxxxxxx will be added to the localversion
165 if a git-based tree is found. The string generated by this will be
166 appended after any matching localversion* files, and after the value
167 set in CONFIG_LOCALVERSION.
169 (The actual string used here is the first eight characters produced
170 by running the command:
172 $ git rev-parse --verify HEAD
174 which is done within the script "scripts/setlocalversion".)
177 string "Build ID Salt"
180 The build ID is used to link binaries and their debug info. Setting
181 this option will use the value in the calculation of the build id.
182 This is mostly useful for distributions which want to ensure the
183 build is unique between builds. It's safe to leave the default.
185 config HAVE_KERNEL_GZIP
188 config HAVE_KERNEL_BZIP2
191 config HAVE_KERNEL_LZMA
194 config HAVE_KERNEL_XZ
197 config HAVE_KERNEL_LZO
200 config HAVE_KERNEL_LZ4
203 config HAVE_KERNEL_ZSTD
206 config HAVE_KERNEL_UNCOMPRESSED
210 prompt "Kernel compression mode"
212 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_ZSTD || HAVE_KERNEL_UNCOMPRESSED
214 The linux kernel is a kind of self-extracting executable.
215 Several compression algorithms are available, which differ
216 in efficiency, compression and decompression speed.
217 Compression speed is only relevant when building a kernel.
218 Decompression speed is relevant at each boot.
220 If you have any problems with bzip2 or lzma compressed
221 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
222 version of this functionality (bzip2 only), for 2.4, was
223 supplied by Christian Ludwig)
225 High compression options are mostly useful for users, who
226 are low on disk space (embedded systems), but for whom ram
229 If in doubt, select 'gzip'
233 depends on HAVE_KERNEL_GZIP
235 The old and tried gzip compression. It provides a good balance
236 between compression ratio and decompression speed.
240 depends on HAVE_KERNEL_BZIP2
242 Its compression ratio and speed is intermediate.
243 Decompression speed is slowest among the choices. The kernel
244 size is about 10% smaller with bzip2, in comparison to gzip.
245 Bzip2 uses a large amount of memory. For modern kernels you
246 will need at least 8MB RAM or more for booting.
250 depends on HAVE_KERNEL_LZMA
252 This compression algorithm's ratio is best. Decompression speed
253 is between gzip and bzip2. Compression is slowest.
254 The kernel size is about 33% smaller with LZMA in comparison to gzip.
258 depends on HAVE_KERNEL_XZ
260 XZ uses the LZMA2 algorithm and instruction set specific
261 BCJ filters which can improve compression ratio of executable
262 code. The size of the kernel is about 30% smaller with XZ in
263 comparison to gzip. On architectures for which there is a BCJ
264 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
265 will create a few percent smaller kernel than plain LZMA.
267 The speed is about the same as with LZMA: The decompression
268 speed of XZ is better than that of bzip2 but worse than gzip
269 and LZO. Compression is slow.
273 depends on HAVE_KERNEL_LZO
275 Its compression ratio is the poorest among the choices. The kernel
276 size is about 10% bigger than gzip; however its speed
277 (both compression and decompression) is the fastest.
281 depends on HAVE_KERNEL_LZ4
283 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
284 A preliminary version of LZ4 de/compression tool is available at
285 <https://code.google.com/p/lz4/>.
287 Its compression ratio is worse than LZO. The size of the kernel
288 is about 8% bigger than LZO. But the decompression speed is
293 depends on HAVE_KERNEL_ZSTD
295 ZSTD is a compression algorithm targeting intermediate compression
296 with fast decompression speed. It will compress better than GZIP and
297 decompress around the same speed as LZO, but slower than LZ4. You
298 will need at least 192 KB RAM or more for booting. The zstd command
299 line tool is required for compression.
301 config KERNEL_UNCOMPRESSED
303 depends on HAVE_KERNEL_UNCOMPRESSED
305 Produce uncompressed kernel image. This option is usually not what
306 you want. It is useful for debugging the kernel in slow simulation
307 environments, where decompressing and moving the kernel is awfully
308 slow. This option allows early boot code to skip the decompressor
309 and jump right at uncompressed kernel image.
314 string "Default init path"
317 This option determines the default init for the system if no init=
318 option is passed on the kernel command line. If the requested path is
319 not present, we will still then move on to attempting further
320 locations (e.g. /sbin/init, etc). If this is empty, we will just use
321 the fallback list when init= is not passed.
323 config DEFAULT_HOSTNAME
324 string "Default hostname"
327 This option determines the default system hostname before userspace
328 calls sethostname(2). The kernel traditionally uses "(none)" here,
329 but you may wish to use a different default here to make a minimal
330 system more usable with less configuration.
333 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
334 # add proper SWAP support to them, in which case this can be remove.
340 bool "Support for paging of anonymous memory (swap)"
341 depends on MMU && BLOCK && !ARCH_NO_SWAP
344 This option allows you to choose whether you want to have support
345 for so called swap devices or swap files in your kernel that are
346 used to provide more virtual memory than the actual RAM present
347 in your computer. If unsure say Y.
352 Inter Process Communication is a suite of library functions and
353 system calls which let processes (running programs) synchronize and
354 exchange information. It is generally considered to be a good thing,
355 and some programs won't run unless you say Y here. In particular, if
356 you want to run the DOS emulator dosemu under Linux (read the
357 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
358 you'll need to say Y here.
360 You can find documentation about IPC with "info ipc" and also in
361 section 6.4 of the Linux Programmer's Guide, available from
362 <http://www.tldp.org/guides.html>.
364 config SYSVIPC_SYSCTL
371 bool "POSIX Message Queues"
374 POSIX variant of message queues is a part of IPC. In POSIX message
375 queues every message has a priority which decides about succession
376 of receiving it by a process. If you want to compile and run
377 programs written e.g. for Solaris with use of its POSIX message
378 queues (functions mq_*) say Y here.
380 POSIX message queues are visible as a filesystem called 'mqueue'
381 and can be mounted somewhere if you want to do filesystem
382 operations on message queues.
386 config POSIX_MQUEUE_SYSCTL
388 depends on POSIX_MQUEUE
393 bool "General notification queue"
397 This is a general notification queue for the kernel to pass events to
398 userspace by splicing them into pipes. It can be used in conjunction
399 with watches for key/keyring change notifications and device
402 See Documentation/watch_queue.rst
404 config CROSS_MEMORY_ATTACH
405 bool "Enable process_vm_readv/writev syscalls"
409 Enabling this option adds the system calls process_vm_readv and
410 process_vm_writev which allow a process with the correct privileges
411 to directly read from or write to another process' address space.
412 See the man page for more details.
415 bool "uselib syscall"
416 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
418 This option enables the uselib syscall, a system call used in the
419 dynamic linker from libc5 and earlier. glibc does not use this
420 system call. If you intend to run programs built on libc5 or
421 earlier, you may need to enable this syscall. Current systems
422 running glibc can safely disable this.
425 bool "Auditing support"
428 Enable auditing infrastructure that can be used with another
429 kernel subsystem, such as SELinux (which requires this for
430 logging of avc messages output). System call auditing is included
431 on architectures which support it.
433 config HAVE_ARCH_AUDITSYSCALL
438 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
441 source "kernel/irq/Kconfig"
442 source "kernel/time/Kconfig"
443 source "kernel/Kconfig.preempt"
445 menu "CPU/Task time and stats accounting"
447 config VIRT_CPU_ACCOUNTING
451 prompt "Cputime accounting"
452 default TICK_CPU_ACCOUNTING if !PPC64
453 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
455 # Kind of a stub config for the pure tick based cputime accounting
456 config TICK_CPU_ACCOUNTING
457 bool "Simple tick based cputime accounting"
458 depends on !S390 && !NO_HZ_FULL
460 This is the basic tick based cputime accounting that maintains
461 statistics about user, system and idle time spent on per jiffies
466 config VIRT_CPU_ACCOUNTING_NATIVE
467 bool "Deterministic task and CPU time accounting"
468 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
469 select VIRT_CPU_ACCOUNTING
471 Select this option to enable more accurate task and CPU time
472 accounting. This is done by reading a CPU counter on each
473 kernel entry and exit and on transitions within the kernel
474 between system, softirq and hardirq state, so there is a
475 small performance impact. In the case of s390 or IBM POWER > 5,
476 this also enables accounting of stolen time on logically-partitioned
479 config VIRT_CPU_ACCOUNTING_GEN
480 bool "Full dynticks CPU time accounting"
481 depends on HAVE_CONTEXT_TRACKING
482 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
483 depends on GENERIC_CLOCKEVENTS
484 select VIRT_CPU_ACCOUNTING
485 select CONTEXT_TRACKING
487 Select this option to enable task and CPU time accounting on full
488 dynticks systems. This accounting is implemented by watching every
489 kernel-user boundaries using the context tracking subsystem.
490 The accounting is thus performed at the expense of some significant
493 For now this is only useful if you are working on the full
494 dynticks subsystem development.
500 config IRQ_TIME_ACCOUNTING
501 bool "Fine granularity task level IRQ time accounting"
502 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
504 Select this option to enable fine granularity task irq time
505 accounting. This is done by reading a timestamp on each
506 transitions between softirq and hardirq state, so there can be a
507 small performance impact.
509 If in doubt, say N here.
511 config HAVE_SCHED_AVG_IRQ
513 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
516 config SCHED_THERMAL_PRESSURE
518 default y if ARM && ARM_CPU_TOPOLOGY
521 depends on CPU_FREQ_THERMAL
523 Select this option to enable thermal pressure accounting in the
524 scheduler. Thermal pressure is the value conveyed to the scheduler
525 that reflects the reduction in CPU compute capacity resulted from
526 thermal throttling. Thermal throttling occurs when the performance of
527 a CPU is capped due to high operating temperatures.
529 If selected, the scheduler will be able to balance tasks accordingly,
530 i.e. put less load on throttled CPUs than on non/less throttled ones.
532 This requires the architecture to implement
533 arch_set_thermal_pressure() and arch_get_thermal_pressure().
535 config BSD_PROCESS_ACCT
536 bool "BSD Process Accounting"
539 If you say Y here, a user level program will be able to instruct the
540 kernel (via a special system call) to write process accounting
541 information to a file: whenever a process exits, information about
542 that process will be appended to the file by the kernel. The
543 information includes things such as creation time, owning user,
544 command name, memory usage, controlling terminal etc. (the complete
545 list is in the struct acct in <file:include/linux/acct.h>). It is
546 up to the user level program to do useful things with this
547 information. This is generally a good idea, so say Y.
549 config BSD_PROCESS_ACCT_V3
550 bool "BSD Process Accounting version 3 file format"
551 depends on BSD_PROCESS_ACCT
554 If you say Y here, the process accounting information is written
555 in a new file format that also logs the process IDs of each
556 process and its parent. Note that this file format is incompatible
557 with previous v0/v1/v2 file formats, so you will need updated tools
558 for processing it. A preliminary version of these tools is available
559 at <http://www.gnu.org/software/acct/>.
562 bool "Export task/process statistics through netlink"
567 Export selected statistics for tasks/processes through the
568 generic netlink interface. Unlike BSD process accounting, the
569 statistics are available during the lifetime of tasks/processes as
570 responses to commands. Like BSD accounting, they are sent to user
575 config TASK_DELAY_ACCT
576 bool "Enable per-task delay accounting"
580 Collect information on time spent by a task waiting for system
581 resources like cpu, synchronous block I/O completion and swapping
582 in pages. Such statistics can help in setting a task's priorities
583 relative to other tasks for cpu, io, rss limits etc.
588 bool "Enable extended accounting over taskstats"
591 Collect extended task accounting data and send the data
592 to userland for processing over the taskstats interface.
596 config TASK_IO_ACCOUNTING
597 bool "Enable per-task storage I/O accounting"
598 depends on TASK_XACCT
600 Collect information on the number of bytes of storage I/O which this
606 bool "Pressure stall information tracking"
608 Collect metrics that indicate how overcommitted the CPU, memory,
609 and IO capacity are in the system.
611 If you say Y here, the kernel will create /proc/pressure/ with the
612 pressure statistics files cpu, memory, and io. These will indicate
613 the share of walltime in which some or all tasks in the system are
614 delayed due to contention of the respective resource.
616 In kernels with cgroup support, cgroups (cgroup2 only) will
617 have cpu.pressure, memory.pressure, and io.pressure files,
618 which aggregate pressure stalls for the grouped tasks only.
620 For more details see Documentation/accounting/psi.rst.
624 config PSI_DEFAULT_DISABLED
625 bool "Require boot parameter to enable pressure stall information tracking"
629 If set, pressure stall information tracking will be disabled
630 per default but can be enabled through passing psi=1 on the
631 kernel commandline during boot.
633 This feature adds some code to the task wakeup and sleep
634 paths of the scheduler. The overhead is too low to affect
635 common scheduling-intense workloads in practice (such as
636 webservers, memcache), but it does show up in artificial
637 scheduler stress tests, such as hackbench.
639 If you are paranoid and not sure what the kernel will be
644 endmenu # "CPU/Task time and stats accounting"
648 depends on SMP || COMPILE_TEST
651 Make sure that CPUs running critical tasks are not disturbed by
652 any source of "noise" such as unbound workqueues, timers, kthreads...
653 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
654 the "isolcpus=" boot parameter.
658 source "kernel/rcu/Kconfig"
665 tristate "Kernel .config support"
667 This option enables the complete Linux kernel ".config" file
668 contents to be saved in the kernel. It provides documentation
669 of which kernel options are used in a running kernel or in an
670 on-disk kernel. This information can be extracted from the kernel
671 image file with the script scripts/extract-ikconfig and used as
672 input to rebuild the current kernel or to build another kernel.
673 It can also be extracted from a running kernel by reading
674 /proc/config.gz if enabled (below).
677 bool "Enable access to .config through /proc/config.gz"
678 depends on IKCONFIG && PROC_FS
680 This option enables access to the kernel configuration file
681 through /proc/config.gz.
684 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
687 This option enables access to the in-kernel headers that are generated during
688 the build process. These can be used to build eBPF tracing programs,
689 or similar programs. If you build the headers as a module, a module called
690 kheaders.ko is built which can be loaded on-demand to get access to headers.
693 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
694 range 12 25 if !H8300
699 Select the minimal kernel log buffer size as a power of 2.
700 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
701 parameter, see below. Any higher size also might be forced
702 by "log_buf_len" boot parameter.
712 config LOG_CPU_MAX_BUF_SHIFT
713 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
716 default 12 if !BASE_SMALL
717 default 0 if BASE_SMALL
720 This option allows to increase the default ring buffer size
721 according to the number of CPUs. The value defines the contribution
722 of each CPU as a power of 2. The used space is typically only few
723 lines however it might be much more when problems are reported,
726 The increased size means that a new buffer has to be allocated and
727 the original static one is unused. It makes sense only on systems
728 with more CPUs. Therefore this value is used only when the sum of
729 contributions is greater than the half of the default kernel ring
730 buffer as defined by LOG_BUF_SHIFT. The default values are set
731 so that more than 16 CPUs are needed to trigger the allocation.
733 Also this option is ignored when "log_buf_len" kernel parameter is
734 used as it forces an exact (power of two) size of the ring buffer.
736 The number of possible CPUs is used for this computation ignoring
737 hotplugging making the computation optimal for the worst case
738 scenario while allowing a simple algorithm to be used from bootup.
740 Examples shift values and their meaning:
741 17 => 128 KB for each CPU
742 16 => 64 KB for each CPU
743 15 => 32 KB for each CPU
744 14 => 16 KB for each CPU
745 13 => 8 KB for each CPU
746 12 => 4 KB for each CPU
748 config PRINTK_SAFE_LOG_BUF_SHIFT
749 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
754 Select the size of an alternate printk per-CPU buffer where messages
755 printed from usafe contexts are temporary stored. One example would
756 be NMI messages, another one - printk recursion. The messages are
757 copied to the main log buffer in a safe context to avoid a deadlock.
758 The value defines the size as a power of 2.
760 Those messages are rare and limited. The largest one is when
761 a backtrace is printed. It usually fits into 4KB. Select
762 8KB if you want to be on the safe side.
765 17 => 128 KB for each CPU
766 16 => 64 KB for each CPU
767 15 => 32 KB for each CPU
768 14 => 16 KB for each CPU
769 13 => 8 KB for each CPU
770 12 => 4 KB for each CPU
773 # Architectures with an unreliable sched_clock() should select this:
775 config HAVE_UNSTABLE_SCHED_CLOCK
778 config GENERIC_SCHED_CLOCK
781 menu "Scheduler features"
784 bool "Enable utilization clamping for RT/FAIR tasks"
785 depends on CPU_FREQ_GOV_SCHEDUTIL
787 This feature enables the scheduler to track the clamped utilization
788 of each CPU based on RUNNABLE tasks scheduled on that CPU.
790 With this option, the user can specify the min and max CPU
791 utilization allowed for RUNNABLE tasks. The max utilization defines
792 the maximum frequency a task should use while the min utilization
793 defines the minimum frequency it should use.
795 Both min and max utilization clamp values are hints to the scheduler,
796 aiming at improving its frequency selection policy, but they do not
797 enforce or grant any specific bandwidth for tasks.
801 config UCLAMP_BUCKETS_COUNT
802 int "Number of supported utilization clamp buckets"
805 depends on UCLAMP_TASK
807 Defines the number of clamp buckets to use. The range of each bucket
808 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
809 number of clamp buckets the finer their granularity and the higher
810 the precision of clamping aggregation and tracking at run-time.
812 For example, with the minimum configuration value we will have 5
813 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
814 be refcounted in the [20..39]% bucket and will set the bucket clamp
815 effective value to 25%.
816 If a second 30% boosted task should be co-scheduled on the same CPU,
817 that task will be refcounted in the same bucket of the first task and
818 it will boost the bucket clamp effective value to 30%.
819 The clamp effective value of a bucket is reset to its nominal value
820 (20% in the example above) when there are no more tasks refcounted in
823 An additional boost/capping margin can be added to some tasks. In the
824 example above the 25% task will be boosted to 30% until it exits the
825 CPU. If that should be considered not acceptable on certain systems,
826 it's always possible to reduce the margin by increasing the number of
827 clamp buckets to trade off used memory for run-time tracking
830 If in doubt, use the default value.
835 # For architectures that want to enable the support for NUMA-affine scheduler
838 config ARCH_SUPPORTS_NUMA_BALANCING
842 # For architectures that prefer to flush all TLBs after a number of pages
843 # are unmapped instead of sending one IPI per page to flush. The architecture
844 # must provide guarantees on what happens if a clean TLB cache entry is
845 # written after the unmap. Details are in mm/rmap.c near the check for
846 # should_defer_flush. The architecture should also consider if the full flush
847 # and the refill costs are offset by the savings of sending fewer IPIs.
848 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
852 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
855 # For architectures that know their GCC __int128 support is sound
857 config ARCH_SUPPORTS_INT128
860 # For architectures that (ab)use NUMA to represent different memory regions
861 # all cpu-local but of different latencies, such as SuperH.
863 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
866 config NUMA_BALANCING
867 bool "Memory placement aware NUMA scheduler"
868 depends on ARCH_SUPPORTS_NUMA_BALANCING
869 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
870 depends on SMP && NUMA && MIGRATION
872 This option adds support for automatic NUMA aware memory/task placement.
873 The mechanism is quite primitive and is based on migrating memory when
874 it has references to the node the task is running on.
876 This system will be inactive on UMA systems.
878 config NUMA_BALANCING_DEFAULT_ENABLED
879 bool "Automatically enable NUMA aware memory/task placement"
881 depends on NUMA_BALANCING
883 If set, automatic NUMA balancing will be enabled if running on a NUMA
887 bool "Control Group support"
890 This option adds support for grouping sets of processes together, for
891 use with process control subsystems such as Cpusets, CFS, memory
892 controls or device isolation.
894 - Documentation/scheduler/sched-design-CFS.rst (CFS)
895 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
896 and resource control)
906 bool "Memory controller"
910 Provides control over the memory footprint of tasks in a cgroup.
914 depends on MEMCG && SWAP
919 depends on MEMCG && !SLOB
927 Generic block IO controller cgroup interface. This is the common
928 cgroup interface which should be used by various IO controlling
931 Currently, CFQ IO scheduler uses it to recognize task groups and
932 control disk bandwidth allocation (proportional time slice allocation)
933 to such task groups. It is also used by bio throttling logic in
934 block layer to implement upper limit in IO rates on a device.
936 This option only enables generic Block IO controller infrastructure.
937 One needs to also enable actual IO controlling logic/policy. For
938 enabling proportional weight division of disk bandwidth in CFQ, set
939 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
940 CONFIG_BLK_DEV_THROTTLING=y.
942 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
944 config CGROUP_WRITEBACK
946 depends on MEMCG && BLK_CGROUP
949 menuconfig CGROUP_SCHED
950 bool "CPU controller"
953 This feature lets CPU scheduler recognize task groups and control CPU
954 bandwidth allocation to such task groups. It uses cgroups to group
958 config FAIR_GROUP_SCHED
959 bool "Group scheduling for SCHED_OTHER"
960 depends on CGROUP_SCHED
964 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
965 depends on FAIR_GROUP_SCHED
968 This option allows users to define CPU bandwidth rates (limits) for
969 tasks running within the fair group scheduler. Groups with no limit
970 set are considered to be unconstrained and will run with no
972 See Documentation/scheduler/sched-bwc.rst for more information.
974 config RT_GROUP_SCHED
975 bool "Group scheduling for SCHED_RR/FIFO"
976 depends on CGROUP_SCHED
979 This feature lets you explicitly allocate real CPU bandwidth
980 to task groups. If enabled, it will also make it impossible to
981 schedule realtime tasks for non-root users until you allocate
982 realtime bandwidth for them.
983 See Documentation/scheduler/sched-rt-group.rst for more information.
987 config UCLAMP_TASK_GROUP
988 bool "Utilization clamping per group of tasks"
989 depends on CGROUP_SCHED
990 depends on UCLAMP_TASK
993 This feature enables the scheduler to track the clamped utilization
994 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
996 When this option is enabled, the user can specify a min and max
997 CPU bandwidth which is allowed for each single task in a group.
998 The max bandwidth allows to clamp the maximum frequency a task
999 can use, while the min bandwidth allows to define a minimum
1000 frequency a task will always use.
1002 When task group based utilization clamping is enabled, an eventually
1003 specified task-specific clamp value is constrained by the cgroup
1004 specified clamp value. Both minimum and maximum task clamping cannot
1005 be bigger than the corresponding clamping defined at task group level.
1010 bool "PIDs controller"
1012 Provides enforcement of process number limits in the scope of a
1013 cgroup. Any attempt to fork more processes than is allowed in the
1014 cgroup will fail. PIDs are fundamentally a global resource because it
1015 is fairly trivial to reach PID exhaustion before you reach even a
1016 conservative kmemcg limit. As a result, it is possible to grind a
1017 system to halt without being limited by other cgroup policies. The
1018 PIDs controller is designed to stop this from happening.
1020 It should be noted that organisational operations (such as attaching
1021 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1022 since the PIDs limit only affects a process's ability to fork, not to
1026 bool "RDMA controller"
1028 Provides enforcement of RDMA resources defined by IB stack.
1029 It is fairly easy for consumers to exhaust RDMA resources, which
1030 can result into resource unavailability to other consumers.
1031 RDMA controller is designed to stop this from happening.
1032 Attaching processes with active RDMA resources to the cgroup
1033 hierarchy is allowed even if can cross the hierarchy's limit.
1035 config CGROUP_FREEZER
1036 bool "Freezer controller"
1038 Provides a way to freeze and unfreeze all tasks in a
1041 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1042 controller includes important in-kernel memory consumers per default.
1044 If you're using cgroup2, say N.
1046 config CGROUP_HUGETLB
1047 bool "HugeTLB controller"
1048 depends on HUGETLB_PAGE
1052 Provides a cgroup controller for HugeTLB pages.
1053 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1054 The limit is enforced during page fault. Since HugeTLB doesn't
1055 support page reclaim, enforcing the limit at page fault time implies
1056 that, the application will get SIGBUS signal if it tries to access
1057 HugeTLB pages beyond its limit. This requires the application to know
1058 beforehand how much HugeTLB pages it would require for its use. The
1059 control group is tracked in the third page lru pointer. This means
1060 that we cannot use the controller with huge page less than 3 pages.
1063 bool "Cpuset controller"
1066 This option will let you create and manage CPUSETs which
1067 allow dynamically partitioning a system into sets of CPUs and
1068 Memory Nodes and assigning tasks to run only within those sets.
1069 This is primarily useful on large SMP or NUMA systems.
1073 config PROC_PID_CPUSET
1074 bool "Include legacy /proc/<pid>/cpuset file"
1078 config CGROUP_DEVICE
1079 bool "Device controller"
1081 Provides a cgroup controller implementing whitelists for
1082 devices which a process in the cgroup can mknod or open.
1084 config CGROUP_CPUACCT
1085 bool "Simple CPU accounting controller"
1087 Provides a simple controller for monitoring the
1088 total CPU consumed by the tasks in a cgroup.
1091 bool "Perf controller"
1092 depends on PERF_EVENTS
1094 This option extends the perf per-cpu mode to restrict monitoring
1095 to threads which belong to the cgroup specified and run on the
1096 designated cpu. Or this can be used to have cgroup ID in samples
1097 so that it can monitor performance events among cgroups.
1102 bool "Support for eBPF programs attached to cgroups"
1103 depends on BPF_SYSCALL
1104 select SOCK_CGROUP_DATA
1106 Allow attaching eBPF programs to a cgroup using the bpf(2)
1107 syscall command BPF_PROG_ATTACH.
1109 In which context these programs are accessed depends on the type
1110 of attachment. For instance, programs that are attached using
1111 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1115 bool "Debug controller"
1117 depends on DEBUG_KERNEL
1119 This option enables a simple controller that exports
1120 debugging information about the cgroups framework. This
1121 controller is for control cgroup debugging only. Its
1122 interfaces are not stable.
1126 config SOCK_CGROUP_DATA
1132 menuconfig NAMESPACES
1133 bool "Namespaces support" if EXPERT
1134 depends on MULTIUSER
1137 Provides the way to make tasks work with different objects using
1138 the same id. For example same IPC id may refer to different objects
1139 or same user id or pid may refer to different tasks when used in
1140 different namespaces.
1145 bool "UTS namespace"
1148 In this namespace tasks see different info provided with the
1152 bool "TIME namespace"
1153 depends on GENERIC_VDSO_TIME_NS
1156 In this namespace boottime and monotonic clocks can be set.
1157 The time will keep going with the same pace.
1160 bool "IPC namespace"
1161 depends on (SYSVIPC || POSIX_MQUEUE)
1164 In this namespace tasks work with IPC ids which correspond to
1165 different IPC objects in different namespaces.
1168 bool "User namespace"
1171 This allows containers, i.e. vservers, to use user namespaces
1172 to provide different user info for different servers.
1174 When user namespaces are enabled in the kernel it is
1175 recommended that the MEMCG option also be enabled and that
1176 user-space use the memory control groups to limit the amount
1177 of memory a memory unprivileged users can use.
1182 bool "PID Namespaces"
1185 Support process id namespaces. This allows having multiple
1186 processes with the same pid as long as they are in different
1187 pid namespaces. This is a building block of containers.
1190 bool "Network namespace"
1194 Allow user space to create what appear to be multiple instances
1195 of the network stack.
1199 config CHECKPOINT_RESTORE
1200 bool "Checkpoint/restore support"
1201 select PROC_CHILDREN
1204 Enables additional kernel features in a sake of checkpoint/restore.
1205 In particular it adds auxiliary prctl codes to setup process text,
1206 data and heap segment sizes, and a few additional /proc filesystem
1209 If unsure, say N here.
1211 config SCHED_AUTOGROUP
1212 bool "Automatic process group scheduling"
1215 select FAIR_GROUP_SCHED
1217 This option optimizes the scheduler for common desktop workloads by
1218 automatically creating and populating task groups. This separation
1219 of workloads isolates aggressive CPU burners (like build jobs) from
1220 desktop applications. Task group autogeneration is currently based
1223 config SYSFS_DEPRECATED
1224 bool "Enable deprecated sysfs features to support old userspace tools"
1228 This option adds code that switches the layout of the "block" class
1229 devices, to not show up in /sys/class/block/, but only in
1232 This switch is only active when the sysfs.deprecated=1 boot option is
1233 passed or the SYSFS_DEPRECATED_V2 option is set.
1235 This option allows new kernels to run on old distributions and tools,
1236 which might get confused by /sys/class/block/. Since 2007/2008 all
1237 major distributions and tools handle this just fine.
1239 Recent distributions and userspace tools after 2009/2010 depend on
1240 the existence of /sys/class/block/, and will not work with this
1243 Only if you are using a new kernel on an old distribution, you might
1246 config SYSFS_DEPRECATED_V2
1247 bool "Enable deprecated sysfs features by default"
1250 depends on SYSFS_DEPRECATED
1252 Enable deprecated sysfs by default.
1254 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1257 Only if you are using a new kernel on an old distribution, you might
1258 need to say Y here. Even then, odds are you would not need it
1259 enabled, you can always pass the boot option if absolutely necessary.
1262 bool "Kernel->user space relay support (formerly relayfs)"
1265 This option enables support for relay interface support in
1266 certain file systems (such as debugfs).
1267 It is designed to provide an efficient mechanism for tools and
1268 facilities to relay large amounts of data from kernel space to
1273 config BLK_DEV_INITRD
1274 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1276 The initial RAM filesystem is a ramfs which is loaded by the
1277 boot loader (loadlin or lilo) and that is mounted as root
1278 before the normal boot procedure. It is typically used to
1279 load modules needed to mount the "real" root file system,
1280 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1282 If RAM disk support (BLK_DEV_RAM) is also included, this
1283 also enables initial RAM disk (initrd) support and adds
1284 15 Kbytes (more on some other architectures) to the kernel size.
1290 source "usr/Kconfig"
1295 bool "Boot config support"
1296 select BLK_DEV_INITRD
1298 Extra boot config allows system admin to pass a config file as
1299 complemental extension of kernel cmdline when booting.
1300 The boot config file must be attached at the end of initramfs
1301 with checksum, size and magic word.
1302 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1307 prompt "Compiler optimization level"
1308 default CC_OPTIMIZE_FOR_PERFORMANCE
1310 config CC_OPTIMIZE_FOR_PERFORMANCE
1311 bool "Optimize for performance (-O2)"
1313 This is the default optimization level for the kernel, building
1314 with the "-O2" compiler flag for best performance and most
1315 helpful compile-time warnings.
1317 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1318 bool "Optimize more for performance (-O3)"
1321 Choosing this option will pass "-O3" to your compiler to optimize
1322 the kernel yet more for performance.
1324 config CC_OPTIMIZE_FOR_SIZE
1325 bool "Optimize for size (-Os)"
1327 Choosing this option will pass "-Os" to your compiler resulting
1328 in a smaller kernel.
1332 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1335 This requires that the arch annotates or otherwise protects
1336 its external entry points from being discarded. Linker scripts
1337 must also merge .text.*, .data.*, and .bss.* correctly into
1338 output sections. Care must be taken not to pull in unrelated
1339 sections (e.g., '.text.init'). Typically '.' in section names
1340 is used to distinguish them from label names / C identifiers.
1342 config LD_DEAD_CODE_DATA_ELIMINATION
1343 bool "Dead code and data elimination (EXPERIMENTAL)"
1344 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1346 depends on $(cc-option,-ffunction-sections -fdata-sections)
1347 depends on $(ld-option,--gc-sections)
1349 Enable this if you want to do dead code and data elimination with
1350 the linker by compiling with -ffunction-sections -fdata-sections,
1351 and linking with --gc-sections.
1353 This can reduce on disk and in-memory size of the kernel
1354 code and static data, particularly for small configs and
1355 on small systems. This has the possibility of introducing
1356 silently broken kernel if the required annotations are not
1357 present. This option is not well tested yet, so use at your
1360 config LD_ORPHAN_WARN
1362 depends on ARCH_WANT_LD_ORPHAN_WARN
1363 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1364 depends on $(ld-option,--orphan-handling=warn)
1372 config SYSCTL_EXCEPTION_TRACE
1375 Enable support for /proc/sys/debug/exception-trace.
1377 config SYSCTL_ARCH_UNALIGN_NO_WARN
1380 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1381 Allows arch to define/use @no_unaligned_warning to possibly warn
1382 about unaligned access emulation going on under the hood.
1384 config SYSCTL_ARCH_UNALIGN_ALLOW
1387 Enable support for /proc/sys/kernel/unaligned-trap
1388 Allows arches to define/use @unaligned_enabled to runtime toggle
1389 the unaligned access emulation.
1390 see arch/parisc/kernel/unaligned.c for reference
1392 config HAVE_PCSPKR_PLATFORM
1395 # interpreter that classic socket filters depend on
1400 bool "Configure standard kernel features (expert users)"
1401 # Unhide debug options, to make the on-by-default options visible
1404 This option allows certain base kernel options and settings
1405 to be disabled or tweaked. This is for specialized
1406 environments which can tolerate a "non-standard" kernel.
1407 Only use this if you really know what you are doing.
1410 bool "Enable 16-bit UID system calls" if EXPERT
1411 depends on HAVE_UID16 && MULTIUSER
1414 This enables the legacy 16-bit UID syscall wrappers.
1417 bool "Multiple users, groups and capabilities support" if EXPERT
1420 This option enables support for non-root users, groups and
1423 If you say N here, all processes will run with UID 0, GID 0, and all
1424 possible capabilities. Saying N here also compiles out support for
1425 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1428 If unsure, say Y here.
1430 config SGETMASK_SYSCALL
1431 bool "sgetmask/ssetmask syscalls support" if EXPERT
1432 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1434 sys_sgetmask and sys_ssetmask are obsolete system calls
1435 no longer supported in libc but still enabled by default in some
1438 If unsure, leave the default option here.
1440 config SYSFS_SYSCALL
1441 bool "Sysfs syscall support" if EXPERT
1444 sys_sysfs is an obsolete system call no longer supported in libc.
1445 Note that disabling this option is more secure but might break
1446 compatibility with some systems.
1448 If unsure say Y here.
1451 bool "open by fhandle syscalls" if EXPERT
1455 If you say Y here, a user level program will be able to map
1456 file names to handle and then later use the handle for
1457 different file system operations. This is useful in implementing
1458 userspace file servers, which now track files using handles instead
1459 of names. The handle would remain the same even if file names
1460 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1464 bool "Posix Clocks & timers" if EXPERT
1467 This includes native support for POSIX timers to the kernel.
1468 Some embedded systems have no use for them and therefore they
1469 can be configured out to reduce the size of the kernel image.
1471 When this option is disabled, the following syscalls won't be
1472 available: timer_create, timer_gettime: timer_getoverrun,
1473 timer_settime, timer_delete, clock_adjtime, getitimer,
1474 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1475 clock_getres and clock_nanosleep syscalls will be limited to
1476 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1482 bool "Enable support for printk" if EXPERT
1485 This option enables normal printk support. Removing it
1486 eliminates most of the message strings from the kernel image
1487 and makes the kernel more or less silent. As this makes it
1488 very difficult to diagnose system problems, saying N here is
1489 strongly discouraged.
1497 bool "BUG() support" if EXPERT
1500 Disabling this option eliminates support for BUG and WARN, reducing
1501 the size of your kernel image and potentially quietly ignoring
1502 numerous fatal conditions. You should only consider disabling this
1503 option for embedded systems with no facilities for reporting errors.
1509 bool "Enable ELF core dumps" if EXPERT
1511 Enable support for generating core dumps. Disabling saves about 4k.
1514 config PCSPKR_PLATFORM
1515 bool "Enable PC-Speaker support" if EXPERT
1516 depends on HAVE_PCSPKR_PLATFORM
1520 This option allows to disable the internal PC-Speaker
1521 support, saving some memory.
1525 bool "Enable full-sized data structures for core" if EXPERT
1527 Disabling this option reduces the size of miscellaneous core
1528 kernel data structures. This saves memory on small machines,
1529 but may reduce performance.
1532 bool "Enable futex support" if EXPERT
1536 Disabling this option will cause the kernel to be built without
1537 support for "fast userspace mutexes". The resulting kernel may not
1538 run glibc-based applications correctly.
1542 depends on FUTEX && RT_MUTEXES
1545 config HAVE_FUTEX_CMPXCHG
1549 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1550 is implemented and always working. This removes a couple of runtime
1554 bool "Enable eventpoll support" if EXPERT
1557 Disabling this option will cause the kernel to be built without
1558 support for epoll family of system calls.
1561 bool "Enable signalfd() system call" if EXPERT
1564 Enable the signalfd() system call that allows to receive signals
1565 on a file descriptor.
1570 bool "Enable timerfd() system call" if EXPERT
1573 Enable the timerfd() system call that allows to receive timer
1574 events on a file descriptor.
1579 bool "Enable eventfd() system call" if EXPERT
1582 Enable the eventfd() system call that allows to receive both
1583 kernel notification (ie. KAIO) or userspace notifications.
1588 bool "Use full shmem filesystem" if EXPERT
1592 The shmem is an internal filesystem used to manage shared memory.
1593 It is backed by swap and manages resource limits. It is also exported
1594 to userspace as tmpfs if TMPFS is enabled. Disabling this
1595 option replaces shmem and tmpfs with the much simpler ramfs code,
1596 which may be appropriate on small systems without swap.
1599 bool "Enable AIO support" if EXPERT
1602 This option enables POSIX asynchronous I/O which may by used
1603 by some high performance threaded applications. Disabling
1604 this option saves about 7k.
1607 bool "Enable IO uring support" if EXPERT
1611 This option enables support for the io_uring interface, enabling
1612 applications to submit and complete IO through submission and
1613 completion rings that are shared between the kernel and application.
1615 config ADVISE_SYSCALLS
1616 bool "Enable madvise/fadvise syscalls" if EXPERT
1619 This option enables the madvise and fadvise syscalls, used by
1620 applications to advise the kernel about their future memory or file
1621 usage, improving performance. If building an embedded system where no
1622 applications use these syscalls, you can disable this option to save
1625 config HAVE_ARCH_USERFAULTFD_WP
1628 Arch has userfaultfd write protection support
1631 bool "Enable membarrier() system call" if EXPERT
1634 Enable the membarrier() system call that allows issuing memory
1635 barriers across all running threads, which can be used to distribute
1636 the cost of user-space memory barriers asymmetrically by transforming
1637 pairs of memory barriers into pairs consisting of membarrier() and a
1643 bool "Load all symbols for debugging/ksymoops" if EXPERT
1646 Say Y here to let the kernel print out symbolic crash information and
1647 symbolic stack backtraces. This increases the size of the kernel
1648 somewhat, as all symbols have to be loaded into the kernel image.
1651 bool "Include all symbols in kallsyms"
1652 depends on DEBUG_KERNEL && KALLSYMS
1654 Normally kallsyms only contains the symbols of functions for nicer
1655 OOPS messages and backtraces (i.e., symbols from the text and inittext
1656 sections). This is sufficient for most cases. And only in very rare
1657 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1658 names of variables from the data sections, etc).
1660 This option makes sure that all symbols are loaded into the kernel
1661 image (i.e., symbols from all sections) in cost of increased kernel
1662 size (depending on the kernel configuration, it may be 300KiB or
1663 something like this).
1665 Say N unless you really need all symbols.
1667 config KALLSYMS_ABSOLUTE_PERCPU
1670 default X86_64 && SMP
1672 config KALLSYMS_BASE_RELATIVE
1677 Instead of emitting them as absolute values in the native word size,
1678 emit the symbol references in the kallsyms table as 32-bit entries,
1679 each containing a relative value in the range [base, base + U32_MAX]
1680 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1681 an absolute value in the range [0, S32_MAX] or a relative value in the
1682 range [base, base + S32_MAX], where base is the lowest relative symbol
1683 address encountered in the image.
1685 On 64-bit builds, this reduces the size of the address table by 50%,
1686 but more importantly, it results in entries whose values are build
1687 time constants, and no relocation pass is required at runtime to fix
1688 up the entries based on the runtime load address of the kernel.
1690 # end of the "standard kernel features (expert users)" menu
1692 # syscall, maps, verifier
1695 bool "LSM Instrumentation with BPF"
1696 depends on BPF_EVENTS
1697 depends on BPF_SYSCALL
1701 Enables instrumentation of the security hooks with eBPF programs for
1702 implementing dynamic MAC and Audit Policies.
1704 If you are unsure how to answer this question, answer N.
1707 bool "Enable bpf() system call"
1710 select TASKS_TRACE_RCU
1713 Enable the bpf() system call that allows to manipulate eBPF
1714 programs and maps via file descriptors.
1716 config ARCH_WANT_DEFAULT_BPF_JIT
1719 config BPF_JIT_ALWAYS_ON
1720 bool "Permanently enable BPF JIT and remove BPF interpreter"
1721 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1723 Enables BPF JIT and removes BPF interpreter to avoid
1724 speculative execution of BPF instructions by the interpreter
1726 config BPF_JIT_DEFAULT_ON
1727 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1728 depends on HAVE_EBPF_JIT && BPF_JIT
1730 source "kernel/bpf/preload/Kconfig"
1733 bool "Enable userfaultfd() system call"
1736 Enable the userfaultfd() system call that allows to intercept and
1737 handle page faults in userland.
1739 config ARCH_HAS_MEMBARRIER_CALLBACKS
1742 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1746 bool "Enable rseq() system call" if EXPERT
1748 depends on HAVE_RSEQ
1751 Enable the restartable sequences system call. It provides a
1752 user-space cache for the current CPU number value, which
1753 speeds up getting the current CPU number from user-space,
1754 as well as an ABI to speed up user-space operations on
1761 bool "Enabled debugging of rseq() system call" if EXPERT
1762 depends on RSEQ && DEBUG_KERNEL
1764 Enable extra debugging checks for the rseq system call.
1769 bool "Embedded system"
1770 option allnoconfig_y
1773 This option should be enabled if compiling the kernel for
1774 an embedded system so certain expert options are available
1777 config HAVE_PERF_EVENTS
1780 See tools/perf/design.txt for details.
1782 config PERF_USE_VMALLOC
1785 See tools/perf/design.txt for details
1788 bool "PC/104 support" if EXPERT
1790 Expose PC/104 form factor device drivers and options available for
1791 selection and configuration. Enable this option if your target
1792 machine has a PC/104 bus.
1794 menu "Kernel Performance Events And Counters"
1797 bool "Kernel performance events and counters"
1798 default y if PROFILING
1799 depends on HAVE_PERF_EVENTS
1803 Enable kernel support for various performance events provided
1804 by software and hardware.
1806 Software events are supported either built-in or via the
1807 use of generic tracepoints.
1809 Most modern CPUs support performance events via performance
1810 counter registers. These registers count the number of certain
1811 types of hw events: such as instructions executed, cachemisses
1812 suffered, or branches mis-predicted - without slowing down the
1813 kernel or applications. These registers can also trigger interrupts
1814 when a threshold number of events have passed - and can thus be
1815 used to profile the code that runs on that CPU.
1817 The Linux Performance Event subsystem provides an abstraction of
1818 these software and hardware event capabilities, available via a
1819 system call and used by the "perf" utility in tools/perf/. It
1820 provides per task and per CPU counters, and it provides event
1821 capabilities on top of those.
1825 config DEBUG_PERF_USE_VMALLOC
1827 bool "Debug: use vmalloc to back perf mmap() buffers"
1828 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1829 select PERF_USE_VMALLOC
1831 Use vmalloc memory to back perf mmap() buffers.
1833 Mostly useful for debugging the vmalloc code on platforms
1834 that don't require it.
1840 config VM_EVENT_COUNTERS
1842 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1844 VM event counters are needed for event counts to be shown.
1845 This option allows the disabling of the VM event counters
1846 on EXPERT systems. /proc/vmstat will only show page counts
1847 if VM event counters are disabled.
1851 bool "Enable SLUB debugging support" if EXPERT
1852 depends on SLUB && SYSFS
1854 SLUB has extensive debug support features. Disabling these can
1855 result in significant savings in code size. This also disables
1856 SLUB sysfs support. /sys/slab will not exist and there will be
1857 no support for cache validation etc.
1859 config SLUB_MEMCG_SYSFS_ON
1861 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1862 depends on SLUB && SYSFS && MEMCG
1864 SLUB creates a directory under /sys/kernel/slab for each
1865 allocation cache to host info and debug files. If memory
1866 cgroup is enabled, each cache can have per memory cgroup
1867 caches. SLUB can create the same sysfs directories for these
1868 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1869 to a very high number of debug files being created. This is
1870 controlled by slub_memcg_sysfs boot parameter and this
1871 config option determines the parameter's default value.
1874 bool "Disable heap randomization"
1877 Randomizing heap placement makes heap exploits harder, but it
1878 also breaks ancient binaries (including anything libc5 based).
1879 This option changes the bootup default to heap randomization
1880 disabled, and can be overridden at runtime by setting
1881 /proc/sys/kernel/randomize_va_space to 2.
1883 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1886 prompt "Choose SLAB allocator"
1889 This option allows to select a slab allocator.
1893 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1895 The regular slab allocator that is established and known to work
1896 well in all environments. It organizes cache hot objects in
1897 per cpu and per node queues.
1900 bool "SLUB (Unqueued Allocator)"
1901 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1903 SLUB is a slab allocator that minimizes cache line usage
1904 instead of managing queues of cached objects (SLAB approach).
1905 Per cpu caching is realized using slabs of objects instead
1906 of queues of objects. SLUB can use memory efficiently
1907 and has enhanced diagnostics. SLUB is the default choice for
1912 bool "SLOB (Simple Allocator)"
1914 SLOB replaces the stock allocator with a drastically simpler
1915 allocator. SLOB is generally more space efficient but
1916 does not perform as well on large systems.
1920 config SLAB_MERGE_DEFAULT
1921 bool "Allow slab caches to be merged"
1924 For reduced kernel memory fragmentation, slab caches can be
1925 merged when they share the same size and other characteristics.
1926 This carries a risk of kernel heap overflows being able to
1927 overwrite objects from merged caches (and more easily control
1928 cache layout), which makes such heap attacks easier to exploit
1929 by attackers. By keeping caches unmerged, these kinds of exploits
1930 can usually only damage objects in the same cache. To disable
1931 merging at runtime, "slab_nomerge" can be passed on the kernel
1934 config SLAB_FREELIST_RANDOM
1935 bool "Randomize slab freelist"
1936 depends on SLAB || SLUB
1938 Randomizes the freelist order used on creating new pages. This
1939 security feature reduces the predictability of the kernel slab
1940 allocator against heap overflows.
1942 config SLAB_FREELIST_HARDENED
1943 bool "Harden slab freelist metadata"
1944 depends on SLAB || SLUB
1946 Many kernel heap attacks try to target slab cache metadata and
1947 other infrastructure. This options makes minor performance
1948 sacrifices to harden the kernel slab allocator against common
1949 freelist exploit methods. Some slab implementations have more
1950 sanity-checking than others. This option is most effective with
1953 config SHUFFLE_PAGE_ALLOCATOR
1954 bool "Page allocator randomization"
1955 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1957 Randomization of the page allocator improves the average
1958 utilization of a direct-mapped memory-side-cache. See section
1959 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1960 6.2a specification for an example of how a platform advertises
1961 the presence of a memory-side-cache. There are also incidental
1962 security benefits as it reduces the predictability of page
1963 allocations to compliment SLAB_FREELIST_RANDOM, but the
1964 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1965 10th order of pages is selected based on cache utilization
1968 While the randomization improves cache utilization it may
1969 negatively impact workloads on platforms without a cache. For
1970 this reason, by default, the randomization is enabled only
1971 after runtime detection of a direct-mapped memory-side-cache.
1972 Otherwise, the randomization may be force enabled with the
1973 'page_alloc.shuffle' kernel command line parameter.
1977 config SLUB_CPU_PARTIAL
1979 depends on SLUB && SMP
1980 bool "SLUB per cpu partial cache"
1982 Per cpu partial caches accelerate objects allocation and freeing
1983 that is local to a processor at the price of more indeterminism
1984 in the latency of the free. On overflow these caches will be cleared
1985 which requires the taking of locks that may cause latency spikes.
1986 Typically one would choose no for a realtime system.
1988 config MMAP_ALLOW_UNINITIALIZED
1989 bool "Allow mmapped anonymous memory to be uninitialized"
1990 depends on EXPERT && !MMU
1993 Normally, and according to the Linux spec, anonymous memory obtained
1994 from mmap() has its contents cleared before it is passed to
1995 userspace. Enabling this config option allows you to request that
1996 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1997 providing a huge performance boost. If this option is not enabled,
1998 then the flag will be ignored.
2000 This is taken advantage of by uClibc's malloc(), and also by
2001 ELF-FDPIC binfmt's brk and stack allocator.
2003 Because of the obvious security issues, this option should only be
2004 enabled on embedded devices where you control what is run in
2005 userspace. Since that isn't generally a problem on no-MMU systems,
2006 it is normally safe to say Y here.
2008 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2010 config SYSTEM_DATA_VERIFICATION
2012 select SYSTEM_TRUSTED_KEYRING
2016 select ASYMMETRIC_KEY_TYPE
2017 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2020 select X509_CERTIFICATE_PARSER
2021 select PKCS7_MESSAGE_PARSER
2023 Provide PKCS#7 message verification using the contents of the system
2024 trusted keyring to provide public keys. This then can be used for
2025 module verification, kexec image verification and firmware blob
2029 bool "Profiling support"
2031 Say Y here to enable the extended profiling support mechanisms used
2032 by profilers such as OProfile.
2035 # Place an empty function call at each tracepoint site. Can be
2036 # dynamically changed for a probe function.
2041 endmenu # General setup
2043 source "arch/Kconfig"
2050 default 0 if BASE_FULL
2051 default 1 if !BASE_FULL
2053 config MODULE_SIG_FORMAT
2055 select SYSTEM_DATA_VERIFICATION
2058 bool "Enable loadable module support"
2061 Kernel modules are small pieces of compiled code which can
2062 be inserted in the running kernel, rather than being
2063 permanently built into the kernel. You use the "modprobe"
2064 tool to add (and sometimes remove) them. If you say Y here,
2065 many parts of the kernel can be built as modules (by
2066 answering M instead of Y where indicated): this is most
2067 useful for infrequently used options which are not required
2068 for booting. For more information, see the man pages for
2069 modprobe, lsmod, modinfo, insmod and rmmod.
2071 If you say Y here, you will need to run "make
2072 modules_install" to put the modules under /lib/modules/
2073 where modprobe can find them (you may need to be root to do
2080 config MODULE_FORCE_LOAD
2081 bool "Forced module loading"
2084 Allow loading of modules without version information (ie. modprobe
2085 --force). Forced module loading sets the 'F' (forced) taint flag and
2086 is usually a really bad idea.
2088 config MODULE_UNLOAD
2089 bool "Module unloading"
2091 Without this option you will not be able to unload any
2092 modules (note that some modules may not be unloadable
2093 anyway), which makes your kernel smaller, faster
2094 and simpler. If unsure, say Y.
2096 config MODULE_FORCE_UNLOAD
2097 bool "Forced module unloading"
2098 depends on MODULE_UNLOAD
2100 This option allows you to force a module to unload, even if the
2101 kernel believes it is unsafe: the kernel will remove the module
2102 without waiting for anyone to stop using it (using the -f option to
2103 rmmod). This is mainly for kernel developers and desperate users.
2107 bool "Module versioning support"
2109 Usually, you have to use modules compiled with your kernel.
2110 Saying Y here makes it sometimes possible to use modules
2111 compiled for different kernels, by adding enough information
2112 to the modules to (hopefully) spot any changes which would
2113 make them incompatible with the kernel you are running. If
2116 config ASM_MODVERSIONS
2118 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2120 This enables module versioning for exported symbols also from
2121 assembly. This can be enabled only when the target architecture
2124 config MODULE_REL_CRCS
2126 depends on MODVERSIONS
2128 config MODULE_SRCVERSION_ALL
2129 bool "Source checksum for all modules"
2131 Modules which contain a MODULE_VERSION get an extra "srcversion"
2132 field inserted into their modinfo section, which contains a
2133 sum of the source files which made it. This helps maintainers
2134 see exactly which source was used to build a module (since
2135 others sometimes change the module source without updating
2136 the version). With this option, such a "srcversion" field
2137 will be created for all modules. If unsure, say N.
2140 bool "Module signature verification"
2141 select MODULE_SIG_FORMAT
2143 Check modules for valid signatures upon load: the signature
2144 is simply appended to the module. For more information see
2145 <file:Documentation/admin-guide/module-signing.rst>.
2147 Note that this option adds the OpenSSL development packages as a
2148 kernel build dependency so that the signing tool can use its crypto
2151 You should enable this option if you wish to use either
2152 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2153 another LSM - otherwise unsigned modules will be loadable regardless
2154 of the lockdown policy.
2156 !!!WARNING!!! If you enable this option, you MUST make sure that the
2157 module DOES NOT get stripped after being signed. This includes the
2158 debuginfo strip done by some packagers (such as rpmbuild) and
2159 inclusion into an initramfs that wants the module size reduced.
2161 config MODULE_SIG_FORCE
2162 bool "Require modules to be validly signed"
2163 depends on MODULE_SIG
2165 Reject unsigned modules or signed modules for which we don't have a
2166 key. Without this, such modules will simply taint the kernel.
2168 config MODULE_SIG_ALL
2169 bool "Automatically sign all modules"
2171 depends on MODULE_SIG
2173 Sign all modules during make modules_install. Without this option,
2174 modules must be signed manually, using the scripts/sign-file tool.
2176 comment "Do not forget to sign required modules with scripts/sign-file"
2177 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2180 prompt "Which hash algorithm should modules be signed with?"
2181 depends on MODULE_SIG
2183 This determines which sort of hashing algorithm will be used during
2184 signature generation. This algorithm _must_ be built into the kernel
2185 directly so that signature verification can take place. It is not
2186 possible to load a signed module containing the algorithm to check
2187 the signature on that module.
2189 config MODULE_SIG_SHA1
2190 bool "Sign modules with SHA-1"
2193 config MODULE_SIG_SHA224
2194 bool "Sign modules with SHA-224"
2195 select CRYPTO_SHA256
2197 config MODULE_SIG_SHA256
2198 bool "Sign modules with SHA-256"
2199 select CRYPTO_SHA256
2201 config MODULE_SIG_SHA384
2202 bool "Sign modules with SHA-384"
2203 select CRYPTO_SHA512
2205 config MODULE_SIG_SHA512
2206 bool "Sign modules with SHA-512"
2207 select CRYPTO_SHA512
2211 config MODULE_SIG_HASH
2213 depends on MODULE_SIG
2214 default "sha1" if MODULE_SIG_SHA1
2215 default "sha224" if MODULE_SIG_SHA224
2216 default "sha256" if MODULE_SIG_SHA256
2217 default "sha384" if MODULE_SIG_SHA384
2218 default "sha512" if MODULE_SIG_SHA512
2220 config MODULE_COMPRESS
2221 bool "Compress modules on installation"
2224 Compresses kernel modules when 'make modules_install' is run; gzip or
2225 xz depending on "Compression algorithm" below.
2227 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2229 Out-of-tree kernel modules installed using Kbuild will also be
2230 compressed upon installation.
2232 Note: for modules inside an initrd or initramfs, it's more efficient
2233 to compress the whole initrd or initramfs instead.
2235 Note: This is fully compatible with signed modules.
2240 prompt "Compression algorithm"
2241 depends on MODULE_COMPRESS
2242 default MODULE_COMPRESS_GZIP
2244 This determines which sort of compression will be used during
2245 'make modules_install'.
2247 GZIP (default) and XZ are supported.
2249 config MODULE_COMPRESS_GZIP
2252 config MODULE_COMPRESS_XZ
2257 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2258 bool "Allow loading of modules with missing namespace imports"
2260 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2261 a namespace. A module that makes use of a symbol exported with such a
2262 namespace is required to import the namespace via MODULE_IMPORT_NS().
2263 There is no technical reason to enforce correct namespace imports,
2264 but it creates consistency between symbols defining namespaces and
2265 users importing namespaces they make use of. This option relaxes this
2266 requirement and lifts the enforcement when loading a module.
2270 config UNUSED_SYMBOLS
2271 bool "Enable unused/obsolete exported symbols"
2274 Unused but exported symbols make the kernel needlessly bigger. For
2275 that reason most of these unused exports will soon be removed. This
2276 option is provided temporarily to provide a transition period in case
2277 some external kernel module needs one of these symbols anyway. If you
2278 encounter such a case in your module, consider if you are actually
2279 using the right API. (rationale: since nobody in the kernel is using
2280 this in a module, there is a pretty good chance it's actually the
2281 wrong interface to use). If you really need the symbol, please send a
2282 mail to the linux kernel mailing list mentioning the symbol and why
2283 you really need it, and what the merge plan to the mainline kernel for
2286 config TRIM_UNUSED_KSYMS
2287 bool "Trim unused exported kernel symbols"
2288 depends on !UNUSED_SYMBOLS
2290 The kernel and some modules make many symbols available for
2291 other modules to use via EXPORT_SYMBOL() and variants. Depending
2292 on the set of modules being selected in your kernel configuration,
2293 many of those exported symbols might never be used.
2295 This option allows for unused exported symbols to be dropped from
2296 the build. In turn, this provides the compiler more opportunities
2297 (especially when using LTO) for optimizing the code and reducing
2298 binary size. This might have some security advantages as well.
2300 If unsure, or if you need to build out-of-tree modules, say N.
2302 config UNUSED_KSYMS_WHITELIST
2303 string "Whitelist of symbols to keep in ksymtab"
2304 depends on TRIM_UNUSED_KSYMS
2306 By default, all unused exported symbols will be un-exported from the
2307 build when TRIM_UNUSED_KSYMS is selected.
2309 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2310 exported at all times, even in absence of in-tree users. The value to
2311 set here is the path to a text file containing the list of symbols,
2312 one per line. The path can be absolute, or relative to the kernel
2317 config MODULES_TREE_LOOKUP
2319 depends on PERF_EVENTS || TRACING
2321 config INIT_ALL_POSSIBLE
2324 Back when each arch used to define their own cpu_online_mask and
2325 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2326 with all 1s, and others with all 0s. When they were centralised,
2327 it was better to provide this option than to break all the archs
2328 and have several arch maintainers pursuing me down dark alleys.
2330 source "block/Kconfig"
2332 config PREEMPT_NOTIFIERS
2342 Build a simple ASN.1 grammar compiler that produces a bytecode output
2343 that can be interpreted by the ASN.1 stream decoder and used to
2344 inform it as to what tags are to be expected in a stream and what
2345 functions to call on what tags.
2347 source "kernel/Kconfig.locks"
2349 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2352 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2355 # It may be useful for an architecture to override the definitions of the
2356 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2357 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2358 # different calling convention for syscalls. They can also override the
2359 # macros for not-implemented syscalls in kernel/sys_ni.c and
2360 # kernel/time/posix-stubs.c. All these overrides need to be available in
2361 # <asm/syscall_wrapper.h>.
2362 config ARCH_HAS_SYSCALL_WRAPPER