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 $(success,echo "$(CC_VERSION_TEXT)" | grep -q gcc)
33 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
38 default $(shell,$(LD) --version | $(srctree)/scripts/ld-version.sh)
41 def_bool $(success,echo "$(CC_VERSION_TEXT)" | grep -q clang)
44 def_bool $(success,$(LD) -v | head -n 1 | grep -q LLD)
48 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
51 def_bool $(success,test "$(as-name)" = GNU)
54 def_bool $(success,test "$(as-name)" = LLVM)
58 # Use clang version if this is the integrated assembler
59 default CLANG_VERSION if AS_IS_LLVM
64 default $(shell,$(srctree)/scripts/lld-version.sh $(LD))
68 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
69 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
71 config CC_CAN_LINK_STATIC
73 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
74 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
76 config CC_HAS_ASM_GOTO
77 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
79 config CC_HAS_ASM_GOTO_OUTPUT
80 depends on CC_HAS_ASM_GOTO
81 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)
83 config CC_HAS_ASM_GOTO_TIED_OUTPUT
84 depends on CC_HAS_ASM_GOTO_OUTPUT
85 # Detect buggy gcc and clang, fixed in gcc-11 clang-14.
86 def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
88 config TOOLS_SUPPORT_RELR
89 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
91 config CC_HAS_ASM_INLINE
92 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
101 config BUILDTIME_TABLE_SORT
104 config THREAD_INFO_IN_TASK
107 Select this to move thread_info off the stack into task_struct. To
108 make this work, an arch will need to remove all thread_info fields
109 except flags and fix any runtime bugs.
111 One subtle change that will be needed is to use try_get_task_stack()
112 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
121 depends on BROKEN || !SMP
124 config INIT_ENV_ARG_LIMIT
129 Maximum of each of the number of arguments and environment
130 variables passed to init from the kernel command line.
133 bool "Compile also drivers which will not load"
136 Some drivers can be compiled on a different platform than they are
137 intended to be run on. Despite they cannot be loaded there (or even
138 when they load they cannot be used due to missing HW support),
139 developers still, opposing to distributors, might want to build such
140 drivers to compile-test them.
142 If you are a developer and want to build everything available, say Y
143 here. If you are a user/distributor, say N here to exclude useless
144 drivers to be distributed.
146 config UAPI_HEADER_TEST
147 bool "Compile test UAPI headers"
148 depends on HEADERS_INSTALL && CC_CAN_LINK
150 Compile test headers exported to user-space to ensure they are
151 self-contained, i.e. compilable as standalone units.
153 If you are a developer or tester and want to ensure the exported
154 headers are self-contained, say Y here. Otherwise, choose N.
157 string "Local version - append to kernel release"
159 Append an extra string to the end of your kernel version.
160 This will show up when you type uname, for example.
161 The string you set here will be appended after the contents of
162 any files with a filename matching localversion* in your
163 object and source tree, in that order. Your total string can
164 be a maximum of 64 characters.
166 config LOCALVERSION_AUTO
167 bool "Automatically append version information to the version string"
169 depends on !COMPILE_TEST
171 This will try to automatically determine if the current tree is a
172 release tree by looking for git tags that belong to the current
173 top of tree revision.
175 A string of the format -gxxxxxxxx will be added to the localversion
176 if a git-based tree is found. The string generated by this will be
177 appended after any matching localversion* files, and after the value
178 set in CONFIG_LOCALVERSION.
180 (The actual string used here is the first eight characters produced
181 by running the command:
183 $ git rev-parse --verify HEAD
185 which is done within the script "scripts/setlocalversion".)
188 string "Build ID Salt"
191 The build ID is used to link binaries and their debug info. Setting
192 this option will use the value in the calculation of the build id.
193 This is mostly useful for distributions which want to ensure the
194 build is unique between builds. It's safe to leave the default.
196 config HAVE_KERNEL_GZIP
199 config HAVE_KERNEL_BZIP2
202 config HAVE_KERNEL_LZMA
205 config HAVE_KERNEL_XZ
208 config HAVE_KERNEL_LZO
211 config HAVE_KERNEL_LZ4
214 config HAVE_KERNEL_ZSTD
217 config HAVE_KERNEL_UNCOMPRESSED
221 prompt "Kernel compression mode"
223 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
225 The linux kernel is a kind of self-extracting executable.
226 Several compression algorithms are available, which differ
227 in efficiency, compression and decompression speed.
228 Compression speed is only relevant when building a kernel.
229 Decompression speed is relevant at each boot.
231 If you have any problems with bzip2 or lzma compressed
232 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
233 version of this functionality (bzip2 only), for 2.4, was
234 supplied by Christian Ludwig)
236 High compression options are mostly useful for users, who
237 are low on disk space (embedded systems), but for whom ram
240 If in doubt, select 'gzip'
244 depends on HAVE_KERNEL_GZIP
246 The old and tried gzip compression. It provides a good balance
247 between compression ratio and decompression speed.
251 depends on HAVE_KERNEL_BZIP2
253 Its compression ratio and speed is intermediate.
254 Decompression speed is slowest among the choices. The kernel
255 size is about 10% smaller with bzip2, in comparison to gzip.
256 Bzip2 uses a large amount of memory. For modern kernels you
257 will need at least 8MB RAM or more for booting.
261 depends on HAVE_KERNEL_LZMA
263 This compression algorithm's ratio is best. Decompression speed
264 is between gzip and bzip2. Compression is slowest.
265 The kernel size is about 33% smaller with LZMA in comparison to gzip.
269 depends on HAVE_KERNEL_XZ
271 XZ uses the LZMA2 algorithm and instruction set specific
272 BCJ filters which can improve compression ratio of executable
273 code. The size of the kernel is about 30% smaller with XZ in
274 comparison to gzip. On architectures for which there is a BCJ
275 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
276 will create a few percent smaller kernel than plain LZMA.
278 The speed is about the same as with LZMA: The decompression
279 speed of XZ is better than that of bzip2 but worse than gzip
280 and LZO. Compression is slow.
284 depends on HAVE_KERNEL_LZO
286 Its compression ratio is the poorest among the choices. The kernel
287 size is about 10% bigger than gzip; however its speed
288 (both compression and decompression) is the fastest.
292 depends on HAVE_KERNEL_LZ4
294 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
295 A preliminary version of LZ4 de/compression tool is available at
296 <https://code.google.com/p/lz4/>.
298 Its compression ratio is worse than LZO. The size of the kernel
299 is about 8% bigger than LZO. But the decompression speed is
304 depends on HAVE_KERNEL_ZSTD
306 ZSTD is a compression algorithm targeting intermediate compression
307 with fast decompression speed. It will compress better than GZIP and
308 decompress around the same speed as LZO, but slower than LZ4. You
309 will need at least 192 KB RAM or more for booting. The zstd command
310 line tool is required for compression.
312 config KERNEL_UNCOMPRESSED
314 depends on HAVE_KERNEL_UNCOMPRESSED
316 Produce uncompressed kernel image. This option is usually not what
317 you want. It is useful for debugging the kernel in slow simulation
318 environments, where decompressing and moving the kernel is awfully
319 slow. This option allows early boot code to skip the decompressor
320 and jump right at uncompressed kernel image.
325 string "Default init path"
328 This option determines the default init for the system if no init=
329 option is passed on the kernel command line. If the requested path is
330 not present, we will still then move on to attempting further
331 locations (e.g. /sbin/init, etc). If this is empty, we will just use
332 the fallback list when init= is not passed.
334 config DEFAULT_HOSTNAME
335 string "Default hostname"
338 This option determines the default system hostname before userspace
339 calls sethostname(2). The kernel traditionally uses "(none)" here,
340 but you may wish to use a different default here to make a minimal
341 system more usable with less configuration.
344 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
345 # add proper SWAP support to them, in which case this can be remove.
351 bool "Support for paging of anonymous memory (swap)"
352 depends on MMU && BLOCK && !ARCH_NO_SWAP
355 This option allows you to choose whether you want to have support
356 for so called swap devices or swap files in your kernel that are
357 used to provide more virtual memory than the actual RAM present
358 in your computer. If unsure say Y.
363 Inter Process Communication is a suite of library functions and
364 system calls which let processes (running programs) synchronize and
365 exchange information. It is generally considered to be a good thing,
366 and some programs won't run unless you say Y here. In particular, if
367 you want to run the DOS emulator dosemu under Linux (read the
368 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
369 you'll need to say Y here.
371 You can find documentation about IPC with "info ipc" and also in
372 section 6.4 of the Linux Programmer's Guide, available from
373 <http://www.tldp.org/guides.html>.
375 config SYSVIPC_SYSCTL
382 bool "POSIX Message Queues"
385 POSIX variant of message queues is a part of IPC. In POSIX message
386 queues every message has a priority which decides about succession
387 of receiving it by a process. If you want to compile and run
388 programs written e.g. for Solaris with use of its POSIX message
389 queues (functions mq_*) say Y here.
391 POSIX message queues are visible as a filesystem called 'mqueue'
392 and can be mounted somewhere if you want to do filesystem
393 operations on message queues.
397 config POSIX_MQUEUE_SYSCTL
399 depends on POSIX_MQUEUE
404 bool "General notification queue"
408 This is a general notification queue for the kernel to pass events to
409 userspace by splicing them into pipes. It can be used in conjunction
410 with watches for key/keyring change notifications and device
413 See Documentation/watch_queue.rst
415 config CROSS_MEMORY_ATTACH
416 bool "Enable process_vm_readv/writev syscalls"
420 Enabling this option adds the system calls process_vm_readv and
421 process_vm_writev which allow a process with the correct privileges
422 to directly read from or write to another process' address space.
423 See the man page for more details.
426 bool "uselib syscall"
427 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
429 This option enables the uselib syscall, a system call used in the
430 dynamic linker from libc5 and earlier. glibc does not use this
431 system call. If you intend to run programs built on libc5 or
432 earlier, you may need to enable this syscall. Current systems
433 running glibc can safely disable this.
436 bool "Auditing support"
439 Enable auditing infrastructure that can be used with another
440 kernel subsystem, such as SELinux (which requires this for
441 logging of avc messages output). System call auditing is included
442 on architectures which support it.
444 config HAVE_ARCH_AUDITSYSCALL
449 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
452 source "kernel/irq/Kconfig"
453 source "kernel/time/Kconfig"
454 source "kernel/Kconfig.preempt"
456 menu "CPU/Task time and stats accounting"
458 config VIRT_CPU_ACCOUNTING
462 prompt "Cputime accounting"
463 default TICK_CPU_ACCOUNTING if !PPC64
464 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
466 # Kind of a stub config for the pure tick based cputime accounting
467 config TICK_CPU_ACCOUNTING
468 bool "Simple tick based cputime accounting"
469 depends on !S390 && !NO_HZ_FULL
471 This is the basic tick based cputime accounting that maintains
472 statistics about user, system and idle time spent on per jiffies
477 config VIRT_CPU_ACCOUNTING_NATIVE
478 bool "Deterministic task and CPU time accounting"
479 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
480 select VIRT_CPU_ACCOUNTING
482 Select this option to enable more accurate task and CPU time
483 accounting. This is done by reading a CPU counter on each
484 kernel entry and exit and on transitions within the kernel
485 between system, softirq and hardirq state, so there is a
486 small performance impact. In the case of s390 or IBM POWER > 5,
487 this also enables accounting of stolen time on logically-partitioned
490 config VIRT_CPU_ACCOUNTING_GEN
491 bool "Full dynticks CPU time accounting"
492 depends on HAVE_CONTEXT_TRACKING
493 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
494 depends on GENERIC_CLOCKEVENTS
495 select VIRT_CPU_ACCOUNTING
496 select CONTEXT_TRACKING
498 Select this option to enable task and CPU time accounting on full
499 dynticks systems. This accounting is implemented by watching every
500 kernel-user boundaries using the context tracking subsystem.
501 The accounting is thus performed at the expense of some significant
504 For now this is only useful if you are working on the full
505 dynticks subsystem development.
511 config IRQ_TIME_ACCOUNTING
512 bool "Fine granularity task level IRQ time accounting"
513 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
515 Select this option to enable fine granularity task irq time
516 accounting. This is done by reading a timestamp on each
517 transitions between softirq and hardirq state, so there can be a
518 small performance impact.
520 If in doubt, say N here.
522 config HAVE_SCHED_AVG_IRQ
524 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
527 config SCHED_THERMAL_PRESSURE
529 default y if ARM && ARM_CPU_TOPOLOGY
532 depends on CPU_FREQ_THERMAL
534 Select this option to enable thermal pressure accounting in the
535 scheduler. Thermal pressure is the value conveyed to the scheduler
536 that reflects the reduction in CPU compute capacity resulted from
537 thermal throttling. Thermal throttling occurs when the performance of
538 a CPU is capped due to high operating temperatures.
540 If selected, the scheduler will be able to balance tasks accordingly,
541 i.e. put less load on throttled CPUs than on non/less throttled ones.
543 This requires the architecture to implement
544 arch_set_thermal_pressure() and arch_get_thermal_pressure().
546 config BSD_PROCESS_ACCT
547 bool "BSD Process Accounting"
550 If you say Y here, a user level program will be able to instruct the
551 kernel (via a special system call) to write process accounting
552 information to a file: whenever a process exits, information about
553 that process will be appended to the file by the kernel. The
554 information includes things such as creation time, owning user,
555 command name, memory usage, controlling terminal etc. (the complete
556 list is in the struct acct in <file:include/linux/acct.h>). It is
557 up to the user level program to do useful things with this
558 information. This is generally a good idea, so say Y.
560 config BSD_PROCESS_ACCT_V3
561 bool "BSD Process Accounting version 3 file format"
562 depends on BSD_PROCESS_ACCT
565 If you say Y here, the process accounting information is written
566 in a new file format that also logs the process IDs of each
567 process and its parent. Note that this file format is incompatible
568 with previous v0/v1/v2 file formats, so you will need updated tools
569 for processing it. A preliminary version of these tools is available
570 at <http://www.gnu.org/software/acct/>.
573 bool "Export task/process statistics through netlink"
578 Export selected statistics for tasks/processes through the
579 generic netlink interface. Unlike BSD process accounting, the
580 statistics are available during the lifetime of tasks/processes as
581 responses to commands. Like BSD accounting, they are sent to user
586 config TASK_DELAY_ACCT
587 bool "Enable per-task delay accounting"
591 Collect information on time spent by a task waiting for system
592 resources like cpu, synchronous block I/O completion and swapping
593 in pages. Such statistics can help in setting a task's priorities
594 relative to other tasks for cpu, io, rss limits etc.
599 bool "Enable extended accounting over taskstats"
602 Collect extended task accounting data and send the data
603 to userland for processing over the taskstats interface.
607 config TASK_IO_ACCOUNTING
608 bool "Enable per-task storage I/O accounting"
609 depends on TASK_XACCT
611 Collect information on the number of bytes of storage I/O which this
617 bool "Pressure stall information tracking"
619 Collect metrics that indicate how overcommitted the CPU, memory,
620 and IO capacity are in the system.
622 If you say Y here, the kernel will create /proc/pressure/ with the
623 pressure statistics files cpu, memory, and io. These will indicate
624 the share of walltime in which some or all tasks in the system are
625 delayed due to contention of the respective resource.
627 In kernels with cgroup support, cgroups (cgroup2 only) will
628 have cpu.pressure, memory.pressure, and io.pressure files,
629 which aggregate pressure stalls for the grouped tasks only.
631 For more details see Documentation/accounting/psi.rst.
635 config PSI_DEFAULT_DISABLED
636 bool "Require boot parameter to enable pressure stall information tracking"
640 If set, pressure stall information tracking will be disabled
641 per default but can be enabled through passing psi=1 on the
642 kernel commandline during boot.
644 This feature adds some code to the task wakeup and sleep
645 paths of the scheduler. The overhead is too low to affect
646 common scheduling-intense workloads in practice (such as
647 webservers, memcache), but it does show up in artificial
648 scheduler stress tests, such as hackbench.
650 If you are paranoid and not sure what the kernel will be
655 endmenu # "CPU/Task time and stats accounting"
659 depends on SMP || COMPILE_TEST
662 Make sure that CPUs running critical tasks are not disturbed by
663 any source of "noise" such as unbound workqueues, timers, kthreads...
664 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
665 the "isolcpus=" boot parameter.
669 source "kernel/rcu/Kconfig"
676 tristate "Kernel .config support"
678 This option enables the complete Linux kernel ".config" file
679 contents to be saved in the kernel. It provides documentation
680 of which kernel options are used in a running kernel or in an
681 on-disk kernel. This information can be extracted from the kernel
682 image file with the script scripts/extract-ikconfig and used as
683 input to rebuild the current kernel or to build another kernel.
684 It can also be extracted from a running kernel by reading
685 /proc/config.gz if enabled (below).
688 bool "Enable access to .config through /proc/config.gz"
689 depends on IKCONFIG && PROC_FS
691 This option enables access to the kernel configuration file
692 through /proc/config.gz.
695 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
698 This option enables access to the in-kernel headers that are generated during
699 the build process. These can be used to build eBPF tracing programs,
700 or similar programs. If you build the headers as a module, a module called
701 kheaders.ko is built which can be loaded on-demand to get access to headers.
704 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
705 range 12 25 if !H8300
710 Select the minimal kernel log buffer size as a power of 2.
711 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
712 parameter, see below. Any higher size also might be forced
713 by "log_buf_len" boot parameter.
723 config LOG_CPU_MAX_BUF_SHIFT
724 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
727 default 12 if !BASE_SMALL
728 default 0 if BASE_SMALL
731 This option allows to increase the default ring buffer size
732 according to the number of CPUs. The value defines the contribution
733 of each CPU as a power of 2. The used space is typically only few
734 lines however it might be much more when problems are reported,
737 The increased size means that a new buffer has to be allocated and
738 the original static one is unused. It makes sense only on systems
739 with more CPUs. Therefore this value is used only when the sum of
740 contributions is greater than the half of the default kernel ring
741 buffer as defined by LOG_BUF_SHIFT. The default values are set
742 so that more than 16 CPUs are needed to trigger the allocation.
744 Also this option is ignored when "log_buf_len" kernel parameter is
745 used as it forces an exact (power of two) size of the ring buffer.
747 The number of possible CPUs is used for this computation ignoring
748 hotplugging making the computation optimal for the worst case
749 scenario while allowing a simple algorithm to be used from bootup.
751 Examples shift values and their meaning:
752 17 => 128 KB for each CPU
753 16 => 64 KB for each CPU
754 15 => 32 KB for each CPU
755 14 => 16 KB for each CPU
756 13 => 8 KB for each CPU
757 12 => 4 KB for each CPU
759 config PRINTK_SAFE_LOG_BUF_SHIFT
760 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
765 Select the size of an alternate printk per-CPU buffer where messages
766 printed from usafe contexts are temporary stored. One example would
767 be NMI messages, another one - printk recursion. The messages are
768 copied to the main log buffer in a safe context to avoid a deadlock.
769 The value defines the size as a power of 2.
771 Those messages are rare and limited. The largest one is when
772 a backtrace is printed. It usually fits into 4KB. Select
773 8KB if you want to be on the safe side.
776 17 => 128 KB for each CPU
777 16 => 64 KB for each CPU
778 15 => 32 KB for each CPU
779 14 => 16 KB for each CPU
780 13 => 8 KB for each CPU
781 12 => 4 KB for each CPU
784 # Architectures with an unreliable sched_clock() should select this:
786 config HAVE_UNSTABLE_SCHED_CLOCK
789 config GENERIC_SCHED_CLOCK
792 menu "Scheduler features"
795 bool "Enable utilization clamping for RT/FAIR tasks"
796 depends on CPU_FREQ_GOV_SCHEDUTIL
798 This feature enables the scheduler to track the clamped utilization
799 of each CPU based on RUNNABLE tasks scheduled on that CPU.
801 With this option, the user can specify the min and max CPU
802 utilization allowed for RUNNABLE tasks. The max utilization defines
803 the maximum frequency a task should use while the min utilization
804 defines the minimum frequency it should use.
806 Both min and max utilization clamp values are hints to the scheduler,
807 aiming at improving its frequency selection policy, but they do not
808 enforce or grant any specific bandwidth for tasks.
812 config UCLAMP_BUCKETS_COUNT
813 int "Number of supported utilization clamp buckets"
816 depends on UCLAMP_TASK
818 Defines the number of clamp buckets to use. The range of each bucket
819 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
820 number of clamp buckets the finer their granularity and the higher
821 the precision of clamping aggregation and tracking at run-time.
823 For example, with the minimum configuration value we will have 5
824 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
825 be refcounted in the [20..39]% bucket and will set the bucket clamp
826 effective value to 25%.
827 If a second 30% boosted task should be co-scheduled on the same CPU,
828 that task will be refcounted in the same bucket of the first task and
829 it will boost the bucket clamp effective value to 30%.
830 The clamp effective value of a bucket is reset to its nominal value
831 (20% in the example above) when there are no more tasks refcounted in
834 An additional boost/capping margin can be added to some tasks. In the
835 example above the 25% task will be boosted to 30% until it exits the
836 CPU. If that should be considered not acceptable on certain systems,
837 it's always possible to reduce the margin by increasing the number of
838 clamp buckets to trade off used memory for run-time tracking
841 If in doubt, use the default value.
846 # For architectures that want to enable the support for NUMA-affine scheduler
849 config ARCH_SUPPORTS_NUMA_BALANCING
853 # For architectures that prefer to flush all TLBs after a number of pages
854 # are unmapped instead of sending one IPI per page to flush. The architecture
855 # must provide guarantees on what happens if a clean TLB cache entry is
856 # written after the unmap. Details are in mm/rmap.c near the check for
857 # should_defer_flush. The architecture should also consider if the full flush
858 # and the refill costs are offset by the savings of sending fewer IPIs.
859 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
863 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
866 # For architectures that know their GCC __int128 support is sound
868 config ARCH_SUPPORTS_INT128
871 # For architectures that (ab)use NUMA to represent different memory regions
872 # all cpu-local but of different latencies, such as SuperH.
874 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
877 config NUMA_BALANCING
878 bool "Memory placement aware NUMA scheduler"
879 depends on ARCH_SUPPORTS_NUMA_BALANCING
880 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
881 depends on SMP && NUMA && MIGRATION
883 This option adds support for automatic NUMA aware memory/task placement.
884 The mechanism is quite primitive and is based on migrating memory when
885 it has references to the node the task is running on.
887 This system will be inactive on UMA systems.
889 config NUMA_BALANCING_DEFAULT_ENABLED
890 bool "Automatically enable NUMA aware memory/task placement"
892 depends on NUMA_BALANCING
894 If set, automatic NUMA balancing will be enabled if running on a NUMA
898 bool "Control Group support"
901 This option adds support for grouping sets of processes together, for
902 use with process control subsystems such as Cpusets, CFS, memory
903 controls or device isolation.
905 - Documentation/scheduler/sched-design-CFS.rst (CFS)
906 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
907 and resource control)
917 bool "Memory controller"
921 Provides control over the memory footprint of tasks in a cgroup.
925 depends on MEMCG && SWAP
930 depends on MEMCG && !SLOB
938 Generic block IO controller cgroup interface. This is the common
939 cgroup interface which should be used by various IO controlling
942 Currently, CFQ IO scheduler uses it to recognize task groups and
943 control disk bandwidth allocation (proportional time slice allocation)
944 to such task groups. It is also used by bio throttling logic in
945 block layer to implement upper limit in IO rates on a device.
947 This option only enables generic Block IO controller infrastructure.
948 One needs to also enable actual IO controlling logic/policy. For
949 enabling proportional weight division of disk bandwidth in CFQ, set
950 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
951 CONFIG_BLK_DEV_THROTTLING=y.
953 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
955 config CGROUP_WRITEBACK
957 depends on MEMCG && BLK_CGROUP
960 menuconfig CGROUP_SCHED
961 bool "CPU controller"
964 This feature lets CPU scheduler recognize task groups and control CPU
965 bandwidth allocation to such task groups. It uses cgroups to group
969 config FAIR_GROUP_SCHED
970 bool "Group scheduling for SCHED_OTHER"
971 depends on CGROUP_SCHED
975 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
976 depends on FAIR_GROUP_SCHED
979 This option allows users to define CPU bandwidth rates (limits) for
980 tasks running within the fair group scheduler. Groups with no limit
981 set are considered to be unconstrained and will run with no
983 See Documentation/scheduler/sched-bwc.rst for more information.
985 config RT_GROUP_SCHED
986 bool "Group scheduling for SCHED_RR/FIFO"
987 depends on CGROUP_SCHED
990 This feature lets you explicitly allocate real CPU bandwidth
991 to task groups. If enabled, it will also make it impossible to
992 schedule realtime tasks for non-root users until you allocate
993 realtime bandwidth for them.
994 See Documentation/scheduler/sched-rt-group.rst for more information.
998 config UCLAMP_TASK_GROUP
999 bool "Utilization clamping per group of tasks"
1000 depends on CGROUP_SCHED
1001 depends on UCLAMP_TASK
1004 This feature enables the scheduler to track the clamped utilization
1005 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
1007 When this option is enabled, the user can specify a min and max
1008 CPU bandwidth which is allowed for each single task in a group.
1009 The max bandwidth allows to clamp the maximum frequency a task
1010 can use, while the min bandwidth allows to define a minimum
1011 frequency a task will always use.
1013 When task group based utilization clamping is enabled, an eventually
1014 specified task-specific clamp value is constrained by the cgroup
1015 specified clamp value. Both minimum and maximum task clamping cannot
1016 be bigger than the corresponding clamping defined at task group level.
1021 bool "PIDs controller"
1023 Provides enforcement of process number limits in the scope of a
1024 cgroup. Any attempt to fork more processes than is allowed in the
1025 cgroup will fail. PIDs are fundamentally a global resource because it
1026 is fairly trivial to reach PID exhaustion before you reach even a
1027 conservative kmemcg limit. As a result, it is possible to grind a
1028 system to halt without being limited by other cgroup policies. The
1029 PIDs controller is designed to stop this from happening.
1031 It should be noted that organisational operations (such as attaching
1032 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1033 since the PIDs limit only affects a process's ability to fork, not to
1037 bool "RDMA controller"
1039 Provides enforcement of RDMA resources defined by IB stack.
1040 It is fairly easy for consumers to exhaust RDMA resources, which
1041 can result into resource unavailability to other consumers.
1042 RDMA controller is designed to stop this from happening.
1043 Attaching processes with active RDMA resources to the cgroup
1044 hierarchy is allowed even if can cross the hierarchy's limit.
1046 config CGROUP_FREEZER
1047 bool "Freezer controller"
1049 Provides a way to freeze and unfreeze all tasks in a
1052 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1053 controller includes important in-kernel memory consumers per default.
1055 If you're using cgroup2, say N.
1057 config CGROUP_HUGETLB
1058 bool "HugeTLB controller"
1059 depends on HUGETLB_PAGE
1063 Provides a cgroup controller for HugeTLB pages.
1064 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1065 The limit is enforced during page fault. Since HugeTLB doesn't
1066 support page reclaim, enforcing the limit at page fault time implies
1067 that, the application will get SIGBUS signal if it tries to access
1068 HugeTLB pages beyond its limit. This requires the application to know
1069 beforehand how much HugeTLB pages it would require for its use. The
1070 control group is tracked in the third page lru pointer. This means
1071 that we cannot use the controller with huge page less than 3 pages.
1074 bool "Cpuset controller"
1077 This option will let you create and manage CPUSETs which
1078 allow dynamically partitioning a system into sets of CPUs and
1079 Memory Nodes and assigning tasks to run only within those sets.
1080 This is primarily useful on large SMP or NUMA systems.
1084 config PROC_PID_CPUSET
1085 bool "Include legacy /proc/<pid>/cpuset file"
1089 config CGROUP_DEVICE
1090 bool "Device controller"
1092 Provides a cgroup controller implementing whitelists for
1093 devices which a process in the cgroup can mknod or open.
1095 config CGROUP_CPUACCT
1096 bool "Simple CPU accounting controller"
1098 Provides a simple controller for monitoring the
1099 total CPU consumed by the tasks in a cgroup.
1102 bool "Perf controller"
1103 depends on PERF_EVENTS
1105 This option extends the perf per-cpu mode to restrict monitoring
1106 to threads which belong to the cgroup specified and run on the
1107 designated cpu. Or this can be used to have cgroup ID in samples
1108 so that it can monitor performance events among cgroups.
1113 bool "Support for eBPF programs attached to cgroups"
1114 depends on BPF_SYSCALL
1115 select SOCK_CGROUP_DATA
1117 Allow attaching eBPF programs to a cgroup using the bpf(2)
1118 syscall command BPF_PROG_ATTACH.
1120 In which context these programs are accessed depends on the type
1121 of attachment. For instance, programs that are attached using
1122 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1126 bool "Debug controller"
1128 depends on DEBUG_KERNEL
1130 This option enables a simple controller that exports
1131 debugging information about the cgroups framework. This
1132 controller is for control cgroup debugging only. Its
1133 interfaces are not stable.
1137 config SOCK_CGROUP_DATA
1143 menuconfig NAMESPACES
1144 bool "Namespaces support" if EXPERT
1145 depends on MULTIUSER
1148 Provides the way to make tasks work with different objects using
1149 the same id. For example same IPC id may refer to different objects
1150 or same user id or pid may refer to different tasks when used in
1151 different namespaces.
1156 bool "UTS namespace"
1159 In this namespace tasks see different info provided with the
1163 bool "TIME namespace"
1164 depends on GENERIC_VDSO_TIME_NS
1167 In this namespace boottime and monotonic clocks can be set.
1168 The time will keep going with the same pace.
1171 bool "IPC namespace"
1172 depends on (SYSVIPC || POSIX_MQUEUE)
1175 In this namespace tasks work with IPC ids which correspond to
1176 different IPC objects in different namespaces.
1179 bool "User namespace"
1182 This allows containers, i.e. vservers, to use user namespaces
1183 to provide different user info for different servers.
1185 When user namespaces are enabled in the kernel it is
1186 recommended that the MEMCG option also be enabled and that
1187 user-space use the memory control groups to limit the amount
1188 of memory a memory unprivileged users can use.
1193 bool "PID Namespaces"
1196 Support process id namespaces. This allows having multiple
1197 processes with the same pid as long as they are in different
1198 pid namespaces. This is a building block of containers.
1201 bool "Network namespace"
1205 Allow user space to create what appear to be multiple instances
1206 of the network stack.
1210 config CHECKPOINT_RESTORE
1211 bool "Checkpoint/restore support"
1212 select PROC_CHILDREN
1216 Enables additional kernel features in a sake of checkpoint/restore.
1217 In particular it adds auxiliary prctl codes to setup process text,
1218 data and heap segment sizes, and a few additional /proc filesystem
1221 If unsure, say N here.
1223 config SCHED_AUTOGROUP
1224 bool "Automatic process group scheduling"
1227 select FAIR_GROUP_SCHED
1229 This option optimizes the scheduler for common desktop workloads by
1230 automatically creating and populating task groups. This separation
1231 of workloads isolates aggressive CPU burners (like build jobs) from
1232 desktop applications. Task group autogeneration is currently based
1235 config SYSFS_DEPRECATED
1236 bool "Enable deprecated sysfs features to support old userspace tools"
1240 This option adds code that switches the layout of the "block" class
1241 devices, to not show up in /sys/class/block/, but only in
1244 This switch is only active when the sysfs.deprecated=1 boot option is
1245 passed or the SYSFS_DEPRECATED_V2 option is set.
1247 This option allows new kernels to run on old distributions and tools,
1248 which might get confused by /sys/class/block/. Since 2007/2008 all
1249 major distributions and tools handle this just fine.
1251 Recent distributions and userspace tools after 2009/2010 depend on
1252 the existence of /sys/class/block/, and will not work with this
1255 Only if you are using a new kernel on an old distribution, you might
1258 config SYSFS_DEPRECATED_V2
1259 bool "Enable deprecated sysfs features by default"
1262 depends on SYSFS_DEPRECATED
1264 Enable deprecated sysfs by default.
1266 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1269 Only if you are using a new kernel on an old distribution, you might
1270 need to say Y here. Even then, odds are you would not need it
1271 enabled, you can always pass the boot option if absolutely necessary.
1274 bool "Kernel->user space relay support (formerly relayfs)"
1277 This option enables support for relay interface support in
1278 certain file systems (such as debugfs).
1279 It is designed to provide an efficient mechanism for tools and
1280 facilities to relay large amounts of data from kernel space to
1285 config BLK_DEV_INITRD
1286 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1288 The initial RAM filesystem is a ramfs which is loaded by the
1289 boot loader (loadlin or lilo) and that is mounted as root
1290 before the normal boot procedure. It is typically used to
1291 load modules needed to mount the "real" root file system,
1292 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1294 If RAM disk support (BLK_DEV_RAM) is also included, this
1295 also enables initial RAM disk (initrd) support and adds
1296 15 Kbytes (more on some other architectures) to the kernel size.
1302 source "usr/Kconfig"
1307 bool "Boot config support"
1308 select BLK_DEV_INITRD
1310 Extra boot config allows system admin to pass a config file as
1311 complemental extension of kernel cmdline when booting.
1312 The boot config file must be attached at the end of initramfs
1313 with checksum, size and magic word.
1314 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1319 prompt "Compiler optimization level"
1320 default CC_OPTIMIZE_FOR_PERFORMANCE
1322 config CC_OPTIMIZE_FOR_PERFORMANCE
1323 bool "Optimize for performance (-O2)"
1325 This is the default optimization level for the kernel, building
1326 with the "-O2" compiler flag for best performance and most
1327 helpful compile-time warnings.
1329 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1330 bool "Optimize more for performance (-O3)"
1333 Choosing this option will pass "-O3" to your compiler to optimize
1334 the kernel yet more for performance.
1336 config CC_OPTIMIZE_FOR_SIZE
1337 bool "Optimize for size (-Os)"
1339 Choosing this option will pass "-Os" to your compiler resulting
1340 in a smaller kernel.
1344 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1347 This requires that the arch annotates or otherwise protects
1348 its external entry points from being discarded. Linker scripts
1349 must also merge .text.*, .data.*, and .bss.* correctly into
1350 output sections. Care must be taken not to pull in unrelated
1351 sections (e.g., '.text.init'). Typically '.' in section names
1352 is used to distinguish them from label names / C identifiers.
1354 config LD_DEAD_CODE_DATA_ELIMINATION
1355 bool "Dead code and data elimination (EXPERIMENTAL)"
1356 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1358 depends on $(cc-option,-ffunction-sections -fdata-sections)
1359 depends on $(ld-option,--gc-sections)
1361 Enable this if you want to do dead code and data elimination with
1362 the linker by compiling with -ffunction-sections -fdata-sections,
1363 and linking with --gc-sections.
1365 This can reduce on disk and in-memory size of the kernel
1366 code and static data, particularly for small configs and
1367 on small systems. This has the possibility of introducing
1368 silently broken kernel if the required annotations are not
1369 present. This option is not well tested yet, so use at your
1372 config LD_ORPHAN_WARN
1374 depends on ARCH_WANT_LD_ORPHAN_WARN
1375 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1376 depends on $(ld-option,--orphan-handling=warn)
1384 config SYSCTL_EXCEPTION_TRACE
1387 Enable support for /proc/sys/debug/exception-trace.
1389 config SYSCTL_ARCH_UNALIGN_NO_WARN
1392 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1393 Allows arch to define/use @no_unaligned_warning to possibly warn
1394 about unaligned access emulation going on under the hood.
1396 config SYSCTL_ARCH_UNALIGN_ALLOW
1399 Enable support for /proc/sys/kernel/unaligned-trap
1400 Allows arches to define/use @unaligned_enabled to runtime toggle
1401 the unaligned access emulation.
1402 see arch/parisc/kernel/unaligned.c for reference
1404 config HAVE_PCSPKR_PLATFORM
1407 # interpreter that classic socket filters depend on
1412 bool "Configure standard kernel features (expert users)"
1413 # Unhide debug options, to make the on-by-default options visible
1416 This option allows certain base kernel options and settings
1417 to be disabled or tweaked. This is for specialized
1418 environments which can tolerate a "non-standard" kernel.
1419 Only use this if you really know what you are doing.
1422 bool "Enable 16-bit UID system calls" if EXPERT
1423 depends on HAVE_UID16 && MULTIUSER
1426 This enables the legacy 16-bit UID syscall wrappers.
1429 bool "Multiple users, groups and capabilities support" if EXPERT
1432 This option enables support for non-root users, groups and
1435 If you say N here, all processes will run with UID 0, GID 0, and all
1436 possible capabilities. Saying N here also compiles out support for
1437 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1440 If unsure, say Y here.
1442 config SGETMASK_SYSCALL
1443 bool "sgetmask/ssetmask syscalls support" if EXPERT
1444 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1446 sys_sgetmask and sys_ssetmask are obsolete system calls
1447 no longer supported in libc but still enabled by default in some
1450 If unsure, leave the default option here.
1452 config SYSFS_SYSCALL
1453 bool "Sysfs syscall support" if EXPERT
1456 sys_sysfs is an obsolete system call no longer supported in libc.
1457 Note that disabling this option is more secure but might break
1458 compatibility with some systems.
1460 If unsure say Y here.
1463 bool "open by fhandle syscalls" if EXPERT
1467 If you say Y here, a user level program will be able to map
1468 file names to handle and then later use the handle for
1469 different file system operations. This is useful in implementing
1470 userspace file servers, which now track files using handles instead
1471 of names. The handle would remain the same even if file names
1472 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1476 bool "Posix Clocks & timers" if EXPERT
1479 This includes native support for POSIX timers to the kernel.
1480 Some embedded systems have no use for them and therefore they
1481 can be configured out to reduce the size of the kernel image.
1483 When this option is disabled, the following syscalls won't be
1484 available: timer_create, timer_gettime: timer_getoverrun,
1485 timer_settime, timer_delete, clock_adjtime, getitimer,
1486 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1487 clock_getres and clock_nanosleep syscalls will be limited to
1488 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1494 bool "Enable support for printk" if EXPERT
1497 This option enables normal printk support. Removing it
1498 eliminates most of the message strings from the kernel image
1499 and makes the kernel more or less silent. As this makes it
1500 very difficult to diagnose system problems, saying N here is
1501 strongly discouraged.
1509 bool "BUG() support" if EXPERT
1512 Disabling this option eliminates support for BUG and WARN, reducing
1513 the size of your kernel image and potentially quietly ignoring
1514 numerous fatal conditions. You should only consider disabling this
1515 option for embedded systems with no facilities for reporting errors.
1521 bool "Enable ELF core dumps" if EXPERT
1523 Enable support for generating core dumps. Disabling saves about 4k.
1526 config PCSPKR_PLATFORM
1527 bool "Enable PC-Speaker support" if EXPERT
1528 depends on HAVE_PCSPKR_PLATFORM
1532 This option allows to disable the internal PC-Speaker
1533 support, saving some memory.
1537 bool "Enable full-sized data structures for core" if EXPERT
1539 Disabling this option reduces the size of miscellaneous core
1540 kernel data structures. This saves memory on small machines,
1541 but may reduce performance.
1544 bool "Enable futex support" if EXPERT
1548 Disabling this option will cause the kernel to be built without
1549 support for "fast userspace mutexes". The resulting kernel may not
1550 run glibc-based applications correctly.
1554 depends on FUTEX && RT_MUTEXES
1557 config HAVE_FUTEX_CMPXCHG
1561 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1562 is implemented and always working. This removes a couple of runtime
1566 bool "Enable eventpoll support" if EXPERT
1569 Disabling this option will cause the kernel to be built without
1570 support for epoll family of system calls.
1573 bool "Enable signalfd() system call" if EXPERT
1576 Enable the signalfd() system call that allows to receive signals
1577 on a file descriptor.
1582 bool "Enable timerfd() system call" if EXPERT
1585 Enable the timerfd() system call that allows to receive timer
1586 events on a file descriptor.
1591 bool "Enable eventfd() system call" if EXPERT
1594 Enable the eventfd() system call that allows to receive both
1595 kernel notification (ie. KAIO) or userspace notifications.
1600 bool "Use full shmem filesystem" if EXPERT
1604 The shmem is an internal filesystem used to manage shared memory.
1605 It is backed by swap and manages resource limits. It is also exported
1606 to userspace as tmpfs if TMPFS is enabled. Disabling this
1607 option replaces shmem and tmpfs with the much simpler ramfs code,
1608 which may be appropriate on small systems without swap.
1611 bool "Enable AIO support" if EXPERT
1614 This option enables POSIX asynchronous I/O which may by used
1615 by some high performance threaded applications. Disabling
1616 this option saves about 7k.
1619 bool "Enable IO uring support" if EXPERT
1623 This option enables support for the io_uring interface, enabling
1624 applications to submit and complete IO through submission and
1625 completion rings that are shared between the kernel and application.
1627 config ADVISE_SYSCALLS
1628 bool "Enable madvise/fadvise syscalls" if EXPERT
1631 This option enables the madvise and fadvise syscalls, used by
1632 applications to advise the kernel about their future memory or file
1633 usage, improving performance. If building an embedded system where no
1634 applications use these syscalls, you can disable this option to save
1637 config HAVE_ARCH_USERFAULTFD_WP
1640 Arch has userfaultfd write protection support
1643 bool "Enable membarrier() system call" if EXPERT
1646 Enable the membarrier() system call that allows issuing memory
1647 barriers across all running threads, which can be used to distribute
1648 the cost of user-space memory barriers asymmetrically by transforming
1649 pairs of memory barriers into pairs consisting of membarrier() and a
1655 bool "Load all symbols for debugging/ksymoops" if EXPERT
1658 Say Y here to let the kernel print out symbolic crash information and
1659 symbolic stack backtraces. This increases the size of the kernel
1660 somewhat, as all symbols have to be loaded into the kernel image.
1663 bool "Include all symbols in kallsyms"
1664 depends on DEBUG_KERNEL && KALLSYMS
1666 Normally kallsyms only contains the symbols of functions for nicer
1667 OOPS messages and backtraces (i.e., symbols from the text and inittext
1668 sections). This is sufficient for most cases. And only in very rare
1669 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1670 names of variables from the data sections, etc).
1672 This option makes sure that all symbols are loaded into the kernel
1673 image (i.e., symbols from all sections) in cost of increased kernel
1674 size (depending on the kernel configuration, it may be 300KiB or
1675 something like this).
1677 Say N unless you really need all symbols.
1679 config KALLSYMS_ABSOLUTE_PERCPU
1682 default X86_64 && SMP
1684 config KALLSYMS_BASE_RELATIVE
1689 Instead of emitting them as absolute values in the native word size,
1690 emit the symbol references in the kallsyms table as 32-bit entries,
1691 each containing a relative value in the range [base, base + U32_MAX]
1692 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1693 an absolute value in the range [0, S32_MAX] or a relative value in the
1694 range [base, base + S32_MAX], where base is the lowest relative symbol
1695 address encountered in the image.
1697 On 64-bit builds, this reduces the size of the address table by 50%,
1698 but more importantly, it results in entries whose values are build
1699 time constants, and no relocation pass is required at runtime to fix
1700 up the entries based on the runtime load address of the kernel.
1702 # end of the "standard kernel features (expert users)" menu
1704 # syscall, maps, verifier
1707 bool "LSM Instrumentation with BPF"
1708 depends on BPF_EVENTS
1709 depends on BPF_SYSCALL
1713 Enables instrumentation of the security hooks with eBPF programs for
1714 implementing dynamic MAC and Audit Policies.
1716 If you are unsure how to answer this question, answer N.
1719 bool "Enable bpf() system call"
1722 select TASKS_TRACE_RCU
1725 Enable the bpf() system call that allows to manipulate eBPF
1726 programs and maps via file descriptors.
1728 config ARCH_WANT_DEFAULT_BPF_JIT
1731 config BPF_JIT_ALWAYS_ON
1732 bool "Permanently enable BPF JIT and remove BPF interpreter"
1733 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1735 Enables BPF JIT and removes BPF interpreter to avoid
1736 speculative execution of BPF instructions by the interpreter
1738 config BPF_JIT_DEFAULT_ON
1739 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1740 depends on HAVE_EBPF_JIT && BPF_JIT
1742 config BPF_UNPRIV_DEFAULT_OFF
1743 bool "Disable unprivileged BPF by default"
1744 depends on BPF_SYSCALL
1746 Disables unprivileged BPF by default by setting the corresponding
1747 /proc/sys/kernel/unprivileged_bpf_disabled knob to 2. An admin can
1748 still reenable it by setting it to 0 later on, or permanently
1749 disable it by setting it to 1 (from which no other transition to
1750 0 is possible anymore).
1752 source "kernel/bpf/preload/Kconfig"
1755 bool "Enable userfaultfd() system call"
1758 Enable the userfaultfd() system call that allows to intercept and
1759 handle page faults in userland.
1761 config ARCH_HAS_MEMBARRIER_CALLBACKS
1764 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1768 bool "Enable kcmp() system call" if EXPERT
1770 Enable the kernel resource comparison system call. It provides
1771 user-space with the ability to compare two processes to see if they
1772 share a common resource, such as a file descriptor or even virtual
1778 bool "Enable rseq() system call" if EXPERT
1780 depends on HAVE_RSEQ
1783 Enable the restartable sequences system call. It provides a
1784 user-space cache for the current CPU number value, which
1785 speeds up getting the current CPU number from user-space,
1786 as well as an ABI to speed up user-space operations on
1793 bool "Enabled debugging of rseq() system call" if EXPERT
1794 depends on RSEQ && DEBUG_KERNEL
1796 Enable extra debugging checks for the rseq system call.
1801 bool "Embedded system"
1802 option allnoconfig_y
1805 This option should be enabled if compiling the kernel for
1806 an embedded system so certain expert options are available
1809 config HAVE_PERF_EVENTS
1812 See tools/perf/design.txt for details.
1814 config PERF_USE_VMALLOC
1817 See tools/perf/design.txt for details
1820 bool "PC/104 support" if EXPERT
1822 Expose PC/104 form factor device drivers and options available for
1823 selection and configuration. Enable this option if your target
1824 machine has a PC/104 bus.
1826 menu "Kernel Performance Events And Counters"
1829 bool "Kernel performance events and counters"
1830 default y if PROFILING
1831 depends on HAVE_PERF_EVENTS
1835 Enable kernel support for various performance events provided
1836 by software and hardware.
1838 Software events are supported either built-in or via the
1839 use of generic tracepoints.
1841 Most modern CPUs support performance events via performance
1842 counter registers. These registers count the number of certain
1843 types of hw events: such as instructions executed, cachemisses
1844 suffered, or branches mis-predicted - without slowing down the
1845 kernel or applications. These registers can also trigger interrupts
1846 when a threshold number of events have passed - and can thus be
1847 used to profile the code that runs on that CPU.
1849 The Linux Performance Event subsystem provides an abstraction of
1850 these software and hardware event capabilities, available via a
1851 system call and used by the "perf" utility in tools/perf/. It
1852 provides per task and per CPU counters, and it provides event
1853 capabilities on top of those.
1857 config DEBUG_PERF_USE_VMALLOC
1859 bool "Debug: use vmalloc to back perf mmap() buffers"
1860 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1861 select PERF_USE_VMALLOC
1863 Use vmalloc memory to back perf mmap() buffers.
1865 Mostly useful for debugging the vmalloc code on platforms
1866 that don't require it.
1872 config VM_EVENT_COUNTERS
1874 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1876 VM event counters are needed for event counts to be shown.
1877 This option allows the disabling of the VM event counters
1878 on EXPERT systems. /proc/vmstat will only show page counts
1879 if VM event counters are disabled.
1883 bool "Enable SLUB debugging support" if EXPERT
1884 depends on SLUB && SYSFS
1886 SLUB has extensive debug support features. Disabling these can
1887 result in significant savings in code size. This also disables
1888 SLUB sysfs support. /sys/slab will not exist and there will be
1889 no support for cache validation etc.
1891 config SLUB_MEMCG_SYSFS_ON
1893 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1894 depends on SLUB && SYSFS && MEMCG
1896 SLUB creates a directory under /sys/kernel/slab for each
1897 allocation cache to host info and debug files. If memory
1898 cgroup is enabled, each cache can have per memory cgroup
1899 caches. SLUB can create the same sysfs directories for these
1900 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1901 to a very high number of debug files being created. This is
1902 controlled by slub_memcg_sysfs boot parameter and this
1903 config option determines the parameter's default value.
1906 bool "Disable heap randomization"
1909 Randomizing heap placement makes heap exploits harder, but it
1910 also breaks ancient binaries (including anything libc5 based).
1911 This option changes the bootup default to heap randomization
1912 disabled, and can be overridden at runtime by setting
1913 /proc/sys/kernel/randomize_va_space to 2.
1915 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1918 prompt "Choose SLAB allocator"
1921 This option allows to select a slab allocator.
1925 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1927 The regular slab allocator that is established and known to work
1928 well in all environments. It organizes cache hot objects in
1929 per cpu and per node queues.
1932 bool "SLUB (Unqueued Allocator)"
1933 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1935 SLUB is a slab allocator that minimizes cache line usage
1936 instead of managing queues of cached objects (SLAB approach).
1937 Per cpu caching is realized using slabs of objects instead
1938 of queues of objects. SLUB can use memory efficiently
1939 and has enhanced diagnostics. SLUB is the default choice for
1944 bool "SLOB (Simple Allocator)"
1946 SLOB replaces the stock allocator with a drastically simpler
1947 allocator. SLOB is generally more space efficient but
1948 does not perform as well on large systems.
1952 config SLAB_MERGE_DEFAULT
1953 bool "Allow slab caches to be merged"
1956 For reduced kernel memory fragmentation, slab caches can be
1957 merged when they share the same size and other characteristics.
1958 This carries a risk of kernel heap overflows being able to
1959 overwrite objects from merged caches (and more easily control
1960 cache layout), which makes such heap attacks easier to exploit
1961 by attackers. By keeping caches unmerged, these kinds of exploits
1962 can usually only damage objects in the same cache. To disable
1963 merging at runtime, "slab_nomerge" can be passed on the kernel
1966 config SLAB_FREELIST_RANDOM
1967 bool "Randomize slab freelist"
1968 depends on SLAB || SLUB
1970 Randomizes the freelist order used on creating new pages. This
1971 security feature reduces the predictability of the kernel slab
1972 allocator against heap overflows.
1974 config SLAB_FREELIST_HARDENED
1975 bool "Harden slab freelist metadata"
1976 depends on SLAB || SLUB
1978 Many kernel heap attacks try to target slab cache metadata and
1979 other infrastructure. This options makes minor performance
1980 sacrifices to harden the kernel slab allocator against common
1981 freelist exploit methods. Some slab implementations have more
1982 sanity-checking than others. This option is most effective with
1985 config SHUFFLE_PAGE_ALLOCATOR
1986 bool "Page allocator randomization"
1987 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1989 Randomization of the page allocator improves the average
1990 utilization of a direct-mapped memory-side-cache. See section
1991 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1992 6.2a specification for an example of how a platform advertises
1993 the presence of a memory-side-cache. There are also incidental
1994 security benefits as it reduces the predictability of page
1995 allocations to compliment SLAB_FREELIST_RANDOM, but the
1996 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1997 10th order of pages is selected based on cache utilization
2000 While the randomization improves cache utilization it may
2001 negatively impact workloads on platforms without a cache. For
2002 this reason, by default, the randomization is enabled only
2003 after runtime detection of a direct-mapped memory-side-cache.
2004 Otherwise, the randomization may be force enabled with the
2005 'page_alloc.shuffle' kernel command line parameter.
2009 config SLUB_CPU_PARTIAL
2011 depends on SLUB && SMP
2012 bool "SLUB per cpu partial cache"
2014 Per cpu partial caches accelerate objects allocation and freeing
2015 that is local to a processor at the price of more indeterminism
2016 in the latency of the free. On overflow these caches will be cleared
2017 which requires the taking of locks that may cause latency spikes.
2018 Typically one would choose no for a realtime system.
2020 config MMAP_ALLOW_UNINITIALIZED
2021 bool "Allow mmapped anonymous memory to be uninitialized"
2022 depends on EXPERT && !MMU
2025 Normally, and according to the Linux spec, anonymous memory obtained
2026 from mmap() has its contents cleared before it is passed to
2027 userspace. Enabling this config option allows you to request that
2028 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
2029 providing a huge performance boost. If this option is not enabled,
2030 then the flag will be ignored.
2032 This is taken advantage of by uClibc's malloc(), and also by
2033 ELF-FDPIC binfmt's brk and stack allocator.
2035 Because of the obvious security issues, this option should only be
2036 enabled on embedded devices where you control what is run in
2037 userspace. Since that isn't generally a problem on no-MMU systems,
2038 it is normally safe to say Y here.
2040 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2042 config SYSTEM_DATA_VERIFICATION
2044 select SYSTEM_TRUSTED_KEYRING
2048 select ASYMMETRIC_KEY_TYPE
2049 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2052 select X509_CERTIFICATE_PARSER
2053 select PKCS7_MESSAGE_PARSER
2055 Provide PKCS#7 message verification using the contents of the system
2056 trusted keyring to provide public keys. This then can be used for
2057 module verification, kexec image verification and firmware blob
2061 bool "Profiling support"
2063 Say Y here to enable the extended profiling support mechanisms used
2064 by profilers such as OProfile.
2067 # Place an empty function call at each tracepoint site. Can be
2068 # dynamically changed for a probe function.
2073 endmenu # General setup
2075 source "arch/Kconfig"
2082 default 0 if BASE_FULL
2083 default 1 if !BASE_FULL
2085 config MODULE_SIG_FORMAT
2087 select SYSTEM_DATA_VERIFICATION
2090 bool "Enable loadable module support"
2093 Kernel modules are small pieces of compiled code which can
2094 be inserted in the running kernel, rather than being
2095 permanently built into the kernel. You use the "modprobe"
2096 tool to add (and sometimes remove) them. If you say Y here,
2097 many parts of the kernel can be built as modules (by
2098 answering M instead of Y where indicated): this is most
2099 useful for infrequently used options which are not required
2100 for booting. For more information, see the man pages for
2101 modprobe, lsmod, modinfo, insmod and rmmod.
2103 If you say Y here, you will need to run "make
2104 modules_install" to put the modules under /lib/modules/
2105 where modprobe can find them (you may need to be root to do
2112 config MODULE_FORCE_LOAD
2113 bool "Forced module loading"
2116 Allow loading of modules without version information (ie. modprobe
2117 --force). Forced module loading sets the 'F' (forced) taint flag and
2118 is usually a really bad idea.
2120 config MODULE_UNLOAD
2121 bool "Module unloading"
2123 Without this option you will not be able to unload any
2124 modules (note that some modules may not be unloadable
2125 anyway), which makes your kernel smaller, faster
2126 and simpler. If unsure, say Y.
2128 config MODULE_FORCE_UNLOAD
2129 bool "Forced module unloading"
2130 depends on MODULE_UNLOAD
2132 This option allows you to force a module to unload, even if the
2133 kernel believes it is unsafe: the kernel will remove the module
2134 without waiting for anyone to stop using it (using the -f option to
2135 rmmod). This is mainly for kernel developers and desperate users.
2139 bool "Module versioning support"
2141 Usually, you have to use modules compiled with your kernel.
2142 Saying Y here makes it sometimes possible to use modules
2143 compiled for different kernels, by adding enough information
2144 to the modules to (hopefully) spot any changes which would
2145 make them incompatible with the kernel you are running. If
2148 config ASM_MODVERSIONS
2150 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2152 This enables module versioning for exported symbols also from
2153 assembly. This can be enabled only when the target architecture
2156 config MODULE_REL_CRCS
2158 depends on MODVERSIONS
2160 config MODULE_SRCVERSION_ALL
2161 bool "Source checksum for all modules"
2163 Modules which contain a MODULE_VERSION get an extra "srcversion"
2164 field inserted into their modinfo section, which contains a
2165 sum of the source files which made it. This helps maintainers
2166 see exactly which source was used to build a module (since
2167 others sometimes change the module source without updating
2168 the version). With this option, such a "srcversion" field
2169 will be created for all modules. If unsure, say N.
2172 bool "Module signature verification"
2173 select MODULE_SIG_FORMAT
2175 Check modules for valid signatures upon load: the signature
2176 is simply appended to the module. For more information see
2177 <file:Documentation/admin-guide/module-signing.rst>.
2179 Note that this option adds the OpenSSL development packages as a
2180 kernel build dependency so that the signing tool can use its crypto
2183 You should enable this option if you wish to use either
2184 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2185 another LSM - otherwise unsigned modules will be loadable regardless
2186 of the lockdown policy.
2188 !!!WARNING!!! If you enable this option, you MUST make sure that the
2189 module DOES NOT get stripped after being signed. This includes the
2190 debuginfo strip done by some packagers (such as rpmbuild) and
2191 inclusion into an initramfs that wants the module size reduced.
2193 config MODULE_SIG_FORCE
2194 bool "Require modules to be validly signed"
2195 depends on MODULE_SIG
2197 Reject unsigned modules or signed modules for which we don't have a
2198 key. Without this, such modules will simply taint the kernel.
2200 config MODULE_SIG_ALL
2201 bool "Automatically sign all modules"
2203 depends on MODULE_SIG
2205 Sign all modules during make modules_install. Without this option,
2206 modules must be signed manually, using the scripts/sign-file tool.
2208 comment "Do not forget to sign required modules with scripts/sign-file"
2209 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2212 prompt "Which hash algorithm should modules be signed with?"
2213 depends on MODULE_SIG
2215 This determines which sort of hashing algorithm will be used during
2216 signature generation. This algorithm _must_ be built into the kernel
2217 directly so that signature verification can take place. It is not
2218 possible to load a signed module containing the algorithm to check
2219 the signature on that module.
2221 config MODULE_SIG_SHA1
2222 bool "Sign modules with SHA-1"
2225 config MODULE_SIG_SHA224
2226 bool "Sign modules with SHA-224"
2227 select CRYPTO_SHA256
2229 config MODULE_SIG_SHA256
2230 bool "Sign modules with SHA-256"
2231 select CRYPTO_SHA256
2233 config MODULE_SIG_SHA384
2234 bool "Sign modules with SHA-384"
2235 select CRYPTO_SHA512
2237 config MODULE_SIG_SHA512
2238 bool "Sign modules with SHA-512"
2239 select CRYPTO_SHA512
2243 config MODULE_SIG_HASH
2245 depends on MODULE_SIG
2246 default "sha1" if MODULE_SIG_SHA1
2247 default "sha224" if MODULE_SIG_SHA224
2248 default "sha256" if MODULE_SIG_SHA256
2249 default "sha384" if MODULE_SIG_SHA384
2250 default "sha512" if MODULE_SIG_SHA512
2252 config MODULE_COMPRESS
2253 bool "Compress modules on installation"
2256 Compresses kernel modules when 'make modules_install' is run; gzip or
2257 xz depending on "Compression algorithm" below.
2259 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2261 Out-of-tree kernel modules installed using Kbuild will also be
2262 compressed upon installation.
2264 Note: for modules inside an initrd or initramfs, it's more efficient
2265 to compress the whole initrd or initramfs instead.
2267 Note: This is fully compatible with signed modules.
2272 prompt "Compression algorithm"
2273 depends on MODULE_COMPRESS
2274 default MODULE_COMPRESS_GZIP
2276 This determines which sort of compression will be used during
2277 'make modules_install'.
2279 GZIP (default) and XZ are supported.
2281 config MODULE_COMPRESS_GZIP
2284 config MODULE_COMPRESS_XZ
2289 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2290 bool "Allow loading of modules with missing namespace imports"
2292 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2293 a namespace. A module that makes use of a symbol exported with such a
2294 namespace is required to import the namespace via MODULE_IMPORT_NS().
2295 There is no technical reason to enforce correct namespace imports,
2296 but it creates consistency between symbols defining namespaces and
2297 users importing namespaces they make use of. This option relaxes this
2298 requirement and lifts the enforcement when loading a module.
2302 config UNUSED_SYMBOLS
2303 bool "Enable unused/obsolete exported symbols"
2306 Unused but exported symbols make the kernel needlessly bigger. For
2307 that reason most of these unused exports will soon be removed. This
2308 option is provided temporarily to provide a transition period in case
2309 some external kernel module needs one of these symbols anyway. If you
2310 encounter such a case in your module, consider if you are actually
2311 using the right API. (rationale: since nobody in the kernel is using
2312 this in a module, there is a pretty good chance it's actually the
2313 wrong interface to use). If you really need the symbol, please send a
2314 mail to the linux kernel mailing list mentioning the symbol and why
2315 you really need it, and what the merge plan to the mainline kernel for
2318 config TRIM_UNUSED_KSYMS
2319 bool "Trim unused exported kernel symbols"
2320 depends on !UNUSED_SYMBOLS
2322 The kernel and some modules make many symbols available for
2323 other modules to use via EXPORT_SYMBOL() and variants. Depending
2324 on the set of modules being selected in your kernel configuration,
2325 many of those exported symbols might never be used.
2327 This option allows for unused exported symbols to be dropped from
2328 the build. In turn, this provides the compiler more opportunities
2329 (especially when using LTO) for optimizing the code and reducing
2330 binary size. This might have some security advantages as well.
2332 If unsure, or if you need to build out-of-tree modules, say N.
2334 config UNUSED_KSYMS_WHITELIST
2335 string "Whitelist of symbols to keep in ksymtab"
2336 depends on TRIM_UNUSED_KSYMS
2338 By default, all unused exported symbols will be un-exported from the
2339 build when TRIM_UNUSED_KSYMS is selected.
2341 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2342 exported at all times, even in absence of in-tree users. The value to
2343 set here is the path to a text file containing the list of symbols,
2344 one per line. The path can be absolute, or relative to the kernel
2349 config MODULES_TREE_LOOKUP
2351 depends on PERF_EVENTS || TRACING
2353 config INIT_ALL_POSSIBLE
2356 Back when each arch used to define their own cpu_online_mask and
2357 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2358 with all 1s, and others with all 0s. When they were centralised,
2359 it was better to provide this option than to break all the archs
2360 and have several arch maintainers pursuing me down dark alleys.
2362 source "block/Kconfig"
2364 config PREEMPT_NOTIFIERS
2374 Build a simple ASN.1 grammar compiler that produces a bytecode output
2375 that can be interpreted by the ASN.1 stream decoder and used to
2376 inform it as to what tags are to be expected in a stream and what
2377 functions to call on what tags.
2379 source "kernel/Kconfig.locks"
2381 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2384 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2387 # It may be useful for an architecture to override the definitions of the
2388 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2389 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2390 # different calling convention for syscalls. They can also override the
2391 # macros for not-implemented syscalls in kernel/sys_ni.c and
2392 # kernel/time/posix-stubs.c. All these overrides need to be available in
2393 # <asm/syscall_wrapper.h>.
2394 config ARCH_HAS_SYSCALL_WRAPPER