1 # SPDX-License-Identifier: GPL-2.0-only
4 default "$(CC_VERSION_TEXT)"
6 This is used in unclear ways:
8 - Re-run Kconfig when the compiler is updated
9 The 'default' property references the environment variable,
10 CC_VERSION_TEXT so it is recorded in include/config/auto.conf.cmd.
11 When the compiler is updated, Kconfig will be invoked.
13 - Ensure full rebuild when the compiler is updated
14 include/linux/compiler-version.h contains this option in the comment
15 line so fixdep adds include/config/CC_VERSION_TEXT into the
16 auto-generated dependency. When the compiler is updated, syncconfig
17 will touch it and then every file will be rebuilt.
20 def_bool $(success,test "$(cc-name)" = GCC)
24 default $(cc-version) if CC_IS_GCC
28 def_bool $(success,test "$(cc-name)" = Clang)
32 default $(cc-version) if CC_IS_CLANG
36 def_bool $(success,test "$(as-name)" = GNU)
39 def_bool $(success,test "$(as-name)" = LLVM)
43 # Use clang version if this is the integrated assembler
44 default CLANG_VERSION if AS_IS_LLVM
48 def_bool $(success,test "$(ld-name)" = BFD)
52 default $(ld-version) if LD_IS_BFD
56 def_bool $(success,test "$(ld-name)" = LLD)
60 default $(ld-version) if LD_IS_LLD
65 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
66 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
68 config CC_CAN_LINK_STATIC
70 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
71 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
73 config CC_HAS_ASM_GOTO
74 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
76 config CC_HAS_ASM_GOTO_OUTPUT
77 depends on CC_HAS_ASM_GOTO
78 def_bool $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
80 config TOOLS_SUPPORT_RELR
81 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
83 config CC_HAS_ASM_INLINE
84 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
86 config CC_HAS_NO_PROFILE_FN_ATTR
87 def_bool $(success,echo '__attribute__((no_profile_instrument_function)) int x();' | $(CC) -x c - -c -o /dev/null -Werror)
95 config BUILDTIME_TABLE_SORT
98 config THREAD_INFO_IN_TASK
101 Select this to move thread_info off the stack into task_struct. To
102 make this work, an arch will need to remove all thread_info fields
103 except flags and fix any runtime bugs.
105 One subtle change that will be needed is to use try_get_task_stack()
106 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
115 depends on BROKEN || !SMP
118 config INIT_ENV_ARG_LIMIT
123 Maximum of each of the number of arguments and environment
124 variables passed to init from the kernel command line.
127 bool "Compile also drivers which will not load"
130 Some drivers can be compiled on a different platform than they are
131 intended to be run on. Despite they cannot be loaded there (or even
132 when they load they cannot be used due to missing HW support),
133 developers still, opposing to distributors, might want to build such
134 drivers to compile-test them.
136 If you are a developer and want to build everything available, say Y
137 here. If you are a user/distributor, say N here to exclude useless
138 drivers to be distributed.
141 bool "Compile the kernel with warnings as errors"
144 A kernel build should not cause any compiler warnings, and this
145 enables the '-Werror' flag to enforce that rule by default.
147 However, if you have a new (or very old) compiler with odd and
148 unusual warnings, or you have some architecture with problems,
149 you may need to disable this config option in order to
150 successfully build the kernel.
154 config UAPI_HEADER_TEST
155 bool "Compile test UAPI headers"
156 depends on HEADERS_INSTALL && CC_CAN_LINK
158 Compile test headers exported to user-space to ensure they are
159 self-contained, i.e. compilable as standalone units.
161 If you are a developer or tester and want to ensure the exported
162 headers are self-contained, say Y here. Otherwise, choose N.
165 string "Local version - append to kernel release"
167 Append an extra string to the end of your kernel version.
168 This will show up when you type uname, for example.
169 The string you set here will be appended after the contents of
170 any files with a filename matching localversion* in your
171 object and source tree, in that order. Your total string can
172 be a maximum of 64 characters.
174 config LOCALVERSION_AUTO
175 bool "Automatically append version information to the version string"
177 depends on !COMPILE_TEST
179 This will try to automatically determine if the current tree is a
180 release tree by looking for git tags that belong to the current
181 top of tree revision.
183 A string of the format -gxxxxxxxx will be added to the localversion
184 if a git-based tree is found. The string generated by this will be
185 appended after any matching localversion* files, and after the value
186 set in CONFIG_LOCALVERSION.
188 (The actual string used here is the first eight characters produced
189 by running the command:
191 $ git rev-parse --verify HEAD
193 which is done within the script "scripts/setlocalversion".)
196 string "Build ID Salt"
199 The build ID is used to link binaries and their debug info. Setting
200 this option will use the value in the calculation of the build id.
201 This is mostly useful for distributions which want to ensure the
202 build is unique between builds. It's safe to leave the default.
204 config HAVE_KERNEL_GZIP
207 config HAVE_KERNEL_BZIP2
210 config HAVE_KERNEL_LZMA
213 config HAVE_KERNEL_XZ
216 config HAVE_KERNEL_LZO
219 config HAVE_KERNEL_LZ4
222 config HAVE_KERNEL_ZSTD
225 config HAVE_KERNEL_UNCOMPRESSED
229 prompt "Kernel compression mode"
231 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
233 The linux kernel is a kind of self-extracting executable.
234 Several compression algorithms are available, which differ
235 in efficiency, compression and decompression speed.
236 Compression speed is only relevant when building a kernel.
237 Decompression speed is relevant at each boot.
239 If you have any problems with bzip2 or lzma compressed
240 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
241 version of this functionality (bzip2 only), for 2.4, was
242 supplied by Christian Ludwig)
244 High compression options are mostly useful for users, who
245 are low on disk space (embedded systems), but for whom ram
248 If in doubt, select 'gzip'
252 depends on HAVE_KERNEL_GZIP
254 The old and tried gzip compression. It provides a good balance
255 between compression ratio and decompression speed.
259 depends on HAVE_KERNEL_BZIP2
261 Its compression ratio and speed is intermediate.
262 Decompression speed is slowest among the choices. The kernel
263 size is about 10% smaller with bzip2, in comparison to gzip.
264 Bzip2 uses a large amount of memory. For modern kernels you
265 will need at least 8MB RAM or more for booting.
269 depends on HAVE_KERNEL_LZMA
271 This compression algorithm's ratio is best. Decompression speed
272 is between gzip and bzip2. Compression is slowest.
273 The kernel size is about 33% smaller with LZMA in comparison to gzip.
277 depends on HAVE_KERNEL_XZ
279 XZ uses the LZMA2 algorithm and instruction set specific
280 BCJ filters which can improve compression ratio of executable
281 code. The size of the kernel is about 30% smaller with XZ in
282 comparison to gzip. On architectures for which there is a BCJ
283 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
284 will create a few percent smaller kernel than plain LZMA.
286 The speed is about the same as with LZMA: The decompression
287 speed of XZ is better than that of bzip2 but worse than gzip
288 and LZO. Compression is slow.
292 depends on HAVE_KERNEL_LZO
294 Its compression ratio is the poorest among the choices. The kernel
295 size is about 10% bigger than gzip; however its speed
296 (both compression and decompression) is the fastest.
300 depends on HAVE_KERNEL_LZ4
302 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
303 A preliminary version of LZ4 de/compression tool is available at
304 <https://code.google.com/p/lz4/>.
306 Its compression ratio is worse than LZO. The size of the kernel
307 is about 8% bigger than LZO. But the decompression speed is
312 depends on HAVE_KERNEL_ZSTD
314 ZSTD is a compression algorithm targeting intermediate compression
315 with fast decompression speed. It will compress better than GZIP and
316 decompress around the same speed as LZO, but slower than LZ4. You
317 will need at least 192 KB RAM or more for booting. The zstd command
318 line tool is required for compression.
320 config KERNEL_UNCOMPRESSED
322 depends on HAVE_KERNEL_UNCOMPRESSED
324 Produce uncompressed kernel image. This option is usually not what
325 you want. It is useful for debugging the kernel in slow simulation
326 environments, where decompressing and moving the kernel is awfully
327 slow. This option allows early boot code to skip the decompressor
328 and jump right at uncompressed kernel image.
333 string "Default init path"
336 This option determines the default init for the system if no init=
337 option is passed on the kernel command line. If the requested path is
338 not present, we will still then move on to attempting further
339 locations (e.g. /sbin/init, etc). If this is empty, we will just use
340 the fallback list when init= is not passed.
342 config DEFAULT_HOSTNAME
343 string "Default hostname"
346 This option determines the default system hostname before userspace
347 calls sethostname(2). The kernel traditionally uses "(none)" here,
348 but you may wish to use a different default here to make a minimal
349 system more usable with less configuration.
352 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
353 # add proper SWAP support to them, in which case this can be remove.
359 bool "Support for paging of anonymous memory (swap)"
360 depends on MMU && BLOCK && !ARCH_NO_SWAP
363 This option allows you to choose whether you want to have support
364 for so called swap devices or swap files in your kernel that are
365 used to provide more virtual memory than the actual RAM present
366 in your computer. If unsure say Y.
371 Inter Process Communication is a suite of library functions and
372 system calls which let processes (running programs) synchronize and
373 exchange information. It is generally considered to be a good thing,
374 and some programs won't run unless you say Y here. In particular, if
375 you want to run the DOS emulator dosemu under Linux (read the
376 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
377 you'll need to say Y here.
379 You can find documentation about IPC with "info ipc" and also in
380 section 6.4 of the Linux Programmer's Guide, available from
381 <http://www.tldp.org/guides.html>.
383 config SYSVIPC_SYSCTL
390 bool "POSIX Message Queues"
393 POSIX variant of message queues is a part of IPC. In POSIX message
394 queues every message has a priority which decides about succession
395 of receiving it by a process. If you want to compile and run
396 programs written e.g. for Solaris with use of its POSIX message
397 queues (functions mq_*) say Y here.
399 POSIX message queues are visible as a filesystem called 'mqueue'
400 and can be mounted somewhere if you want to do filesystem
401 operations on message queues.
405 config POSIX_MQUEUE_SYSCTL
407 depends on POSIX_MQUEUE
412 bool "General notification queue"
416 This is a general notification queue for the kernel to pass events to
417 userspace by splicing them into pipes. It can be used in conjunction
418 with watches for key/keyring change notifications and device
421 See Documentation/watch_queue.rst
423 config CROSS_MEMORY_ATTACH
424 bool "Enable process_vm_readv/writev syscalls"
428 Enabling this option adds the system calls process_vm_readv and
429 process_vm_writev which allow a process with the correct privileges
430 to directly read from or write to another process' address space.
431 See the man page for more details.
434 bool "uselib syscall"
435 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
437 This option enables the uselib syscall, a system call used in the
438 dynamic linker from libc5 and earlier. glibc does not use this
439 system call. If you intend to run programs built on libc5 or
440 earlier, you may need to enable this syscall. Current systems
441 running glibc can safely disable this.
444 bool "Auditing support"
447 Enable auditing infrastructure that can be used with another
448 kernel subsystem, such as SELinux (which requires this for
449 logging of avc messages output). System call auditing is included
450 on architectures which support it.
452 config HAVE_ARCH_AUDITSYSCALL
457 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
460 source "kernel/irq/Kconfig"
461 source "kernel/time/Kconfig"
462 source "kernel/bpf/Kconfig"
463 source "kernel/Kconfig.preempt"
465 menu "CPU/Task time and stats accounting"
467 config VIRT_CPU_ACCOUNTING
471 prompt "Cputime accounting"
472 default TICK_CPU_ACCOUNTING if !PPC64
473 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
475 # Kind of a stub config for the pure tick based cputime accounting
476 config TICK_CPU_ACCOUNTING
477 bool "Simple tick based cputime accounting"
478 depends on !S390 && !NO_HZ_FULL
480 This is the basic tick based cputime accounting that maintains
481 statistics about user, system and idle time spent on per jiffies
486 config VIRT_CPU_ACCOUNTING_NATIVE
487 bool "Deterministic task and CPU time accounting"
488 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
489 select VIRT_CPU_ACCOUNTING
491 Select this option to enable more accurate task and CPU time
492 accounting. This is done by reading a CPU counter on each
493 kernel entry and exit and on transitions within the kernel
494 between system, softirq and hardirq state, so there is a
495 small performance impact. In the case of s390 or IBM POWER > 5,
496 this also enables accounting of stolen time on logically-partitioned
499 config VIRT_CPU_ACCOUNTING_GEN
500 bool "Full dynticks CPU time accounting"
501 depends on HAVE_CONTEXT_TRACKING
502 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
503 depends on GENERIC_CLOCKEVENTS
504 select VIRT_CPU_ACCOUNTING
505 select CONTEXT_TRACKING
507 Select this option to enable task and CPU time accounting on full
508 dynticks systems. This accounting is implemented by watching every
509 kernel-user boundaries using the context tracking subsystem.
510 The accounting is thus performed at the expense of some significant
513 For now this is only useful if you are working on the full
514 dynticks subsystem development.
520 config IRQ_TIME_ACCOUNTING
521 bool "Fine granularity task level IRQ time accounting"
522 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
524 Select this option to enable fine granularity task irq time
525 accounting. This is done by reading a timestamp on each
526 transitions between softirq and hardirq state, so there can be a
527 small performance impact.
529 If in doubt, say N here.
531 config HAVE_SCHED_AVG_IRQ
533 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
536 config SCHED_THERMAL_PRESSURE
538 default y if ARM && ARM_CPU_TOPOLOGY
541 depends on CPU_FREQ_THERMAL
543 Select this option to enable thermal pressure accounting in the
544 scheduler. Thermal pressure is the value conveyed to the scheduler
545 that reflects the reduction in CPU compute capacity resulted from
546 thermal throttling. Thermal throttling occurs when the performance of
547 a CPU is capped due to high operating temperatures.
549 If selected, the scheduler will be able to balance tasks accordingly,
550 i.e. put less load on throttled CPUs than on non/less throttled ones.
552 This requires the architecture to implement
553 arch_set_thermal_pressure() and arch_scale_thermal_pressure().
555 config BSD_PROCESS_ACCT
556 bool "BSD Process Accounting"
559 If you say Y here, a user level program will be able to instruct the
560 kernel (via a special system call) to write process accounting
561 information to a file: whenever a process exits, information about
562 that process will be appended to the file by the kernel. The
563 information includes things such as creation time, owning user,
564 command name, memory usage, controlling terminal etc. (the complete
565 list is in the struct acct in <file:include/linux/acct.h>). It is
566 up to the user level program to do useful things with this
567 information. This is generally a good idea, so say Y.
569 config BSD_PROCESS_ACCT_V3
570 bool "BSD Process Accounting version 3 file format"
571 depends on BSD_PROCESS_ACCT
574 If you say Y here, the process accounting information is written
575 in a new file format that also logs the process IDs of each
576 process and its parent. Note that this file format is incompatible
577 with previous v0/v1/v2 file formats, so you will need updated tools
578 for processing it. A preliminary version of these tools is available
579 at <http://www.gnu.org/software/acct/>.
582 bool "Export task/process statistics through netlink"
587 Export selected statistics for tasks/processes through the
588 generic netlink interface. Unlike BSD process accounting, the
589 statistics are available during the lifetime of tasks/processes as
590 responses to commands. Like BSD accounting, they are sent to user
595 config TASK_DELAY_ACCT
596 bool "Enable per-task delay accounting"
600 Collect information on time spent by a task waiting for system
601 resources like cpu, synchronous block I/O completion and swapping
602 in pages. Such statistics can help in setting a task's priorities
603 relative to other tasks for cpu, io, rss limits etc.
608 bool "Enable extended accounting over taskstats"
611 Collect extended task accounting data and send the data
612 to userland for processing over the taskstats interface.
616 config TASK_IO_ACCOUNTING
617 bool "Enable per-task storage I/O accounting"
618 depends on TASK_XACCT
620 Collect information on the number of bytes of storage I/O which this
626 bool "Pressure stall information tracking"
628 Collect metrics that indicate how overcommitted the CPU, memory,
629 and IO capacity are in the system.
631 If you say Y here, the kernel will create /proc/pressure/ with the
632 pressure statistics files cpu, memory, and io. These will indicate
633 the share of walltime in which some or all tasks in the system are
634 delayed due to contention of the respective resource.
636 In kernels with cgroup support, cgroups (cgroup2 only) will
637 have cpu.pressure, memory.pressure, and io.pressure files,
638 which aggregate pressure stalls for the grouped tasks only.
640 For more details see Documentation/accounting/psi.rst.
644 config PSI_DEFAULT_DISABLED
645 bool "Require boot parameter to enable pressure stall information tracking"
649 If set, pressure stall information tracking will be disabled
650 per default but can be enabled through passing psi=1 on the
651 kernel commandline during boot.
653 This feature adds some code to the task wakeup and sleep
654 paths of the scheduler. The overhead is too low to affect
655 common scheduling-intense workloads in practice (such as
656 webservers, memcache), but it does show up in artificial
657 scheduler stress tests, such as hackbench.
659 If you are paranoid and not sure what the kernel will be
664 endmenu # "CPU/Task time and stats accounting"
668 depends on SMP || COMPILE_TEST
671 Make sure that CPUs running critical tasks are not disturbed by
672 any source of "noise" such as unbound workqueues, timers, kthreads...
673 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
674 the "isolcpus=" boot parameter.
678 source "kernel/rcu/Kconfig"
685 tristate "Kernel .config support"
687 This option enables the complete Linux kernel ".config" file
688 contents to be saved in the kernel. It provides documentation
689 of which kernel options are used in a running kernel or in an
690 on-disk kernel. This information can be extracted from the kernel
691 image file with the script scripts/extract-ikconfig and used as
692 input to rebuild the current kernel or to build another kernel.
693 It can also be extracted from a running kernel by reading
694 /proc/config.gz if enabled (below).
697 bool "Enable access to .config through /proc/config.gz"
698 depends on IKCONFIG && PROC_FS
700 This option enables access to the kernel configuration file
701 through /proc/config.gz.
704 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
707 This option enables access to the in-kernel headers that are generated during
708 the build process. These can be used to build eBPF tracing programs,
709 or similar programs. If you build the headers as a module, a module called
710 kheaders.ko is built which can be loaded on-demand to get access to headers.
713 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
714 range 12 25 if !H8300
719 Select the minimal kernel log buffer size as a power of 2.
720 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
721 parameter, see below. Any higher size also might be forced
722 by "log_buf_len" boot parameter.
732 config LOG_CPU_MAX_BUF_SHIFT
733 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
736 default 12 if !BASE_SMALL
737 default 0 if BASE_SMALL
740 This option allows to increase the default ring buffer size
741 according to the number of CPUs. The value defines the contribution
742 of each CPU as a power of 2. The used space is typically only few
743 lines however it might be much more when problems are reported,
746 The increased size means that a new buffer has to be allocated and
747 the original static one is unused. It makes sense only on systems
748 with more CPUs. Therefore this value is used only when the sum of
749 contributions is greater than the half of the default kernel ring
750 buffer as defined by LOG_BUF_SHIFT. The default values are set
751 so that more than 16 CPUs are needed to trigger the allocation.
753 Also this option is ignored when "log_buf_len" kernel parameter is
754 used as it forces an exact (power of two) size of the ring buffer.
756 The number of possible CPUs is used for this computation ignoring
757 hotplugging making the computation optimal for the worst case
758 scenario while allowing a simple algorithm to be used from bootup.
760 Examples shift values and their meaning:
761 17 => 128 KB for each CPU
762 16 => 64 KB for each CPU
763 15 => 32 KB for each CPU
764 14 => 16 KB for each CPU
765 13 => 8 KB for each CPU
766 12 => 4 KB for each CPU
768 config PRINTK_SAFE_LOG_BUF_SHIFT
769 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
774 Select the size of an alternate printk per-CPU buffer where messages
775 printed from usafe contexts are temporary stored. One example would
776 be NMI messages, another one - printk recursion. The messages are
777 copied to the main log buffer in a safe context to avoid a deadlock.
778 The value defines the size as a power of 2.
780 Those messages are rare and limited. The largest one is when
781 a backtrace is printed. It usually fits into 4KB. Select
782 8KB if you want to be on the safe side.
785 17 => 128 KB for each CPU
786 16 => 64 KB for each CPU
787 15 => 32 KB for each CPU
788 14 => 16 KB for each CPU
789 13 => 8 KB for each CPU
790 12 => 4 KB for each CPU
793 bool "Printk indexing debugfs interface"
794 depends on PRINTK && DEBUG_FS
796 Add support for indexing of all printk formats known at compile time
797 at <debugfs>/printk/index/<module>.
799 This can be used as part of maintaining daemons which monitor
800 /dev/kmsg, as it permits auditing the printk formats present in a
801 kernel, allowing detection of cases where monitored printks are
802 changed or no longer present.
804 There is no additional runtime cost to printk with this enabled.
807 # Architectures with an unreliable sched_clock() should select this:
809 config HAVE_UNSTABLE_SCHED_CLOCK
812 config GENERIC_SCHED_CLOCK
815 menu "Scheduler features"
818 bool "Enable utilization clamping for RT/FAIR tasks"
819 depends on CPU_FREQ_GOV_SCHEDUTIL
821 This feature enables the scheduler to track the clamped utilization
822 of each CPU based on RUNNABLE tasks scheduled on that CPU.
824 With this option, the user can specify the min and max CPU
825 utilization allowed for RUNNABLE tasks. The max utilization defines
826 the maximum frequency a task should use while the min utilization
827 defines the minimum frequency it should use.
829 Both min and max utilization clamp values are hints to the scheduler,
830 aiming at improving its frequency selection policy, but they do not
831 enforce or grant any specific bandwidth for tasks.
835 config UCLAMP_BUCKETS_COUNT
836 int "Number of supported utilization clamp buckets"
839 depends on UCLAMP_TASK
841 Defines the number of clamp buckets to use. The range of each bucket
842 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
843 number of clamp buckets the finer their granularity and the higher
844 the precision of clamping aggregation and tracking at run-time.
846 For example, with the minimum configuration value we will have 5
847 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
848 be refcounted in the [20..39]% bucket and will set the bucket clamp
849 effective value to 25%.
850 If a second 30% boosted task should be co-scheduled on the same CPU,
851 that task will be refcounted in the same bucket of the first task and
852 it will boost the bucket clamp effective value to 30%.
853 The clamp effective value of a bucket is reset to its nominal value
854 (20% in the example above) when there are no more tasks refcounted in
857 An additional boost/capping margin can be added to some tasks. In the
858 example above the 25% task will be boosted to 30% until it exits the
859 CPU. If that should be considered not acceptable on certain systems,
860 it's always possible to reduce the margin by increasing the number of
861 clamp buckets to trade off used memory for run-time tracking
864 If in doubt, use the default value.
869 # For architectures that want to enable the support for NUMA-affine scheduler
872 config ARCH_SUPPORTS_NUMA_BALANCING
876 # For architectures that prefer to flush all TLBs after a number of pages
877 # are unmapped instead of sending one IPI per page to flush. The architecture
878 # must provide guarantees on what happens if a clean TLB cache entry is
879 # written after the unmap. Details are in mm/rmap.c near the check for
880 # should_defer_flush. The architecture should also consider if the full flush
881 # and the refill costs are offset by the savings of sending fewer IPIs.
882 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
886 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
889 # For architectures that know their GCC __int128 support is sound
891 config ARCH_SUPPORTS_INT128
894 # For architectures that (ab)use NUMA to represent different memory regions
895 # all cpu-local but of different latencies, such as SuperH.
897 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
900 config NUMA_BALANCING
901 bool "Memory placement aware NUMA scheduler"
902 depends on ARCH_SUPPORTS_NUMA_BALANCING
903 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
904 depends on SMP && NUMA && MIGRATION && !PREEMPT_RT
906 This option adds support for automatic NUMA aware memory/task placement.
907 The mechanism is quite primitive and is based on migrating memory when
908 it has references to the node the task is running on.
910 This system will be inactive on UMA systems.
912 config NUMA_BALANCING_DEFAULT_ENABLED
913 bool "Automatically enable NUMA aware memory/task placement"
915 depends on NUMA_BALANCING
917 If set, automatic NUMA balancing will be enabled if running on a NUMA
921 bool "Control Group support"
924 This option adds support for grouping sets of processes together, for
925 use with process control subsystems such as Cpusets, CFS, memory
926 controls or device isolation.
928 - Documentation/scheduler/sched-design-CFS.rst (CFS)
929 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
930 and resource control)
940 bool "Memory controller"
944 Provides control over the memory footprint of tasks in a cgroup.
948 depends on MEMCG && SWAP
953 depends on MEMCG && !SLOB
961 Generic block IO controller cgroup interface. This is the common
962 cgroup interface which should be used by various IO controlling
965 Currently, CFQ IO scheduler uses it to recognize task groups and
966 control disk bandwidth allocation (proportional time slice allocation)
967 to such task groups. It is also used by bio throttling logic in
968 block layer to implement upper limit in IO rates on a device.
970 This option only enables generic Block IO controller infrastructure.
971 One needs to also enable actual IO controlling logic/policy. For
972 enabling proportional weight division of disk bandwidth in CFQ, set
973 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
974 CONFIG_BLK_DEV_THROTTLING=y.
976 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
978 config CGROUP_WRITEBACK
980 depends on MEMCG && BLK_CGROUP
983 menuconfig CGROUP_SCHED
984 bool "CPU controller"
987 This feature lets CPU scheduler recognize task groups and control CPU
988 bandwidth allocation to such task groups. It uses cgroups to group
992 config FAIR_GROUP_SCHED
993 bool "Group scheduling for SCHED_OTHER"
994 depends on CGROUP_SCHED
998 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
999 depends on FAIR_GROUP_SCHED
1002 This option allows users to define CPU bandwidth rates (limits) for
1003 tasks running within the fair group scheduler. Groups with no limit
1004 set are considered to be unconstrained and will run with no
1006 See Documentation/scheduler/sched-bwc.rst for more information.
1008 config RT_GROUP_SCHED
1009 bool "Group scheduling for SCHED_RR/FIFO"
1010 depends on CGROUP_SCHED
1013 This feature lets you explicitly allocate real CPU bandwidth
1014 to task groups. If enabled, it will also make it impossible to
1015 schedule realtime tasks for non-root users until you allocate
1016 realtime bandwidth for them.
1017 See Documentation/scheduler/sched-rt-group.rst for more information.
1021 config UCLAMP_TASK_GROUP
1022 bool "Utilization clamping per group of tasks"
1023 depends on CGROUP_SCHED
1024 depends on UCLAMP_TASK
1027 This feature enables the scheduler to track the clamped utilization
1028 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
1030 When this option is enabled, the user can specify a min and max
1031 CPU bandwidth which is allowed for each single task in a group.
1032 The max bandwidth allows to clamp the maximum frequency a task
1033 can use, while the min bandwidth allows to define a minimum
1034 frequency a task will always use.
1036 When task group based utilization clamping is enabled, an eventually
1037 specified task-specific clamp value is constrained by the cgroup
1038 specified clamp value. Both minimum and maximum task clamping cannot
1039 be bigger than the corresponding clamping defined at task group level.
1044 bool "PIDs controller"
1046 Provides enforcement of process number limits in the scope of a
1047 cgroup. Any attempt to fork more processes than is allowed in the
1048 cgroup will fail. PIDs are fundamentally a global resource because it
1049 is fairly trivial to reach PID exhaustion before you reach even a
1050 conservative kmemcg limit. As a result, it is possible to grind a
1051 system to halt without being limited by other cgroup policies. The
1052 PIDs controller is designed to stop this from happening.
1054 It should be noted that organisational operations (such as attaching
1055 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1056 since the PIDs limit only affects a process's ability to fork, not to
1060 bool "RDMA controller"
1062 Provides enforcement of RDMA resources defined by IB stack.
1063 It is fairly easy for consumers to exhaust RDMA resources, which
1064 can result into resource unavailability to other consumers.
1065 RDMA controller is designed to stop this from happening.
1066 Attaching processes with active RDMA resources to the cgroup
1067 hierarchy is allowed even if can cross the hierarchy's limit.
1069 config CGROUP_FREEZER
1070 bool "Freezer controller"
1072 Provides a way to freeze and unfreeze all tasks in a
1075 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1076 controller includes important in-kernel memory consumers per default.
1078 If you're using cgroup2, say N.
1080 config CGROUP_HUGETLB
1081 bool "HugeTLB controller"
1082 depends on HUGETLB_PAGE
1086 Provides a cgroup controller for HugeTLB pages.
1087 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1088 The limit is enforced during page fault. Since HugeTLB doesn't
1089 support page reclaim, enforcing the limit at page fault time implies
1090 that, the application will get SIGBUS signal if it tries to access
1091 HugeTLB pages beyond its limit. This requires the application to know
1092 beforehand how much HugeTLB pages it would require for its use. The
1093 control group is tracked in the third page lru pointer. This means
1094 that we cannot use the controller with huge page less than 3 pages.
1097 bool "Cpuset controller"
1100 This option will let you create and manage CPUSETs which
1101 allow dynamically partitioning a system into sets of CPUs and
1102 Memory Nodes and assigning tasks to run only within those sets.
1103 This is primarily useful on large SMP or NUMA systems.
1107 config PROC_PID_CPUSET
1108 bool "Include legacy /proc/<pid>/cpuset file"
1112 config CGROUP_DEVICE
1113 bool "Device controller"
1115 Provides a cgroup controller implementing whitelists for
1116 devices which a process in the cgroup can mknod or open.
1118 config CGROUP_CPUACCT
1119 bool "Simple CPU accounting controller"
1121 Provides a simple controller for monitoring the
1122 total CPU consumed by the tasks in a cgroup.
1125 bool "Perf controller"
1126 depends on PERF_EVENTS
1128 This option extends the perf per-cpu mode to restrict monitoring
1129 to threads which belong to the cgroup specified and run on the
1130 designated cpu. Or this can be used to have cgroup ID in samples
1131 so that it can monitor performance events among cgroups.
1136 bool "Support for eBPF programs attached to cgroups"
1137 depends on BPF_SYSCALL
1138 select SOCK_CGROUP_DATA
1140 Allow attaching eBPF programs to a cgroup using the bpf(2)
1141 syscall command BPF_PROG_ATTACH.
1143 In which context these programs are accessed depends on the type
1144 of attachment. For instance, programs that are attached using
1145 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1149 bool "Misc resource controller"
1152 Provides a controller for miscellaneous resources on a host.
1154 Miscellaneous scalar resources are the resources on the host system
1155 which cannot be abstracted like the other cgroups. This controller
1156 tracks and limits the miscellaneous resources used by a process
1157 attached to a cgroup hierarchy.
1159 For more information, please check misc cgroup section in
1160 /Documentation/admin-guide/cgroup-v2.rst.
1163 bool "Debug controller"
1165 depends on DEBUG_KERNEL
1167 This option enables a simple controller that exports
1168 debugging information about the cgroups framework. This
1169 controller is for control cgroup debugging only. Its
1170 interfaces are not stable.
1174 config SOCK_CGROUP_DATA
1180 menuconfig NAMESPACES
1181 bool "Namespaces support" if EXPERT
1182 depends on MULTIUSER
1185 Provides the way to make tasks work with different objects using
1186 the same id. For example same IPC id may refer to different objects
1187 or same user id or pid may refer to different tasks when used in
1188 different namespaces.
1193 bool "UTS namespace"
1196 In this namespace tasks see different info provided with the
1200 bool "TIME namespace"
1201 depends on GENERIC_VDSO_TIME_NS
1204 In this namespace boottime and monotonic clocks can be set.
1205 The time will keep going with the same pace.
1208 bool "IPC namespace"
1209 depends on (SYSVIPC || POSIX_MQUEUE)
1212 In this namespace tasks work with IPC ids which correspond to
1213 different IPC objects in different namespaces.
1216 bool "User namespace"
1219 This allows containers, i.e. vservers, to use user namespaces
1220 to provide different user info for different servers.
1222 When user namespaces are enabled in the kernel it is
1223 recommended that the MEMCG option also be enabled and that
1224 user-space use the memory control groups to limit the amount
1225 of memory a memory unprivileged users can use.
1230 bool "PID Namespaces"
1233 Support process id namespaces. This allows having multiple
1234 processes with the same pid as long as they are in different
1235 pid namespaces. This is a building block of containers.
1238 bool "Network namespace"
1242 Allow user space to create what appear to be multiple instances
1243 of the network stack.
1247 config CHECKPOINT_RESTORE
1248 bool "Checkpoint/restore support"
1249 select PROC_CHILDREN
1253 Enables additional kernel features in a sake of checkpoint/restore.
1254 In particular it adds auxiliary prctl codes to setup process text,
1255 data and heap segment sizes, and a few additional /proc filesystem
1258 If unsure, say N here.
1260 config SCHED_AUTOGROUP
1261 bool "Automatic process group scheduling"
1264 select FAIR_GROUP_SCHED
1266 This option optimizes the scheduler for common desktop workloads by
1267 automatically creating and populating task groups. This separation
1268 of workloads isolates aggressive CPU burners (like build jobs) from
1269 desktop applications. Task group autogeneration is currently based
1272 config SYSFS_DEPRECATED
1273 bool "Enable deprecated sysfs features to support old userspace tools"
1277 This option adds code that switches the layout of the "block" class
1278 devices, to not show up in /sys/class/block/, but only in
1281 This switch is only active when the sysfs.deprecated=1 boot option is
1282 passed or the SYSFS_DEPRECATED_V2 option is set.
1284 This option allows new kernels to run on old distributions and tools,
1285 which might get confused by /sys/class/block/. Since 2007/2008 all
1286 major distributions and tools handle this just fine.
1288 Recent distributions and userspace tools after 2009/2010 depend on
1289 the existence of /sys/class/block/, and will not work with this
1292 Only if you are using a new kernel on an old distribution, you might
1295 config SYSFS_DEPRECATED_V2
1296 bool "Enable deprecated sysfs features by default"
1299 depends on SYSFS_DEPRECATED
1301 Enable deprecated sysfs by default.
1303 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1306 Only if you are using a new kernel on an old distribution, you might
1307 need to say Y here. Even then, odds are you would not need it
1308 enabled, you can always pass the boot option if absolutely necessary.
1311 bool "Kernel->user space relay support (formerly relayfs)"
1314 This option enables support for relay interface support in
1315 certain file systems (such as debugfs).
1316 It is designed to provide an efficient mechanism for tools and
1317 facilities to relay large amounts of data from kernel space to
1322 config BLK_DEV_INITRD
1323 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1325 The initial RAM filesystem is a ramfs which is loaded by the
1326 boot loader (loadlin or lilo) and that is mounted as root
1327 before the normal boot procedure. It is typically used to
1328 load modules needed to mount the "real" root file system,
1329 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1331 If RAM disk support (BLK_DEV_RAM) is also included, this
1332 also enables initial RAM disk (initrd) support and adds
1333 15 Kbytes (more on some other architectures) to the kernel size.
1339 source "usr/Kconfig"
1344 bool "Boot config support"
1345 select BLK_DEV_INITRD
1347 Extra boot config allows system admin to pass a config file as
1348 complemental extension of kernel cmdline when booting.
1349 The boot config file must be attached at the end of initramfs
1350 with checksum, size and magic word.
1351 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1356 prompt "Compiler optimization level"
1357 default CC_OPTIMIZE_FOR_PERFORMANCE
1359 config CC_OPTIMIZE_FOR_PERFORMANCE
1360 bool "Optimize for performance (-O2)"
1362 This is the default optimization level for the kernel, building
1363 with the "-O2" compiler flag for best performance and most
1364 helpful compile-time warnings.
1366 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1367 bool "Optimize more for performance (-O3)"
1370 Choosing this option will pass "-O3" to your compiler to optimize
1371 the kernel yet more for performance.
1373 config CC_OPTIMIZE_FOR_SIZE
1374 bool "Optimize for size (-Os)"
1376 Choosing this option will pass "-Os" to your compiler resulting
1377 in a smaller kernel.
1381 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1384 This requires that the arch annotates or otherwise protects
1385 its external entry points from being discarded. Linker scripts
1386 must also merge .text.*, .data.*, and .bss.* correctly into
1387 output sections. Care must be taken not to pull in unrelated
1388 sections (e.g., '.text.init'). Typically '.' in section names
1389 is used to distinguish them from label names / C identifiers.
1391 config LD_DEAD_CODE_DATA_ELIMINATION
1392 bool "Dead code and data elimination (EXPERIMENTAL)"
1393 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1395 depends on $(cc-option,-ffunction-sections -fdata-sections)
1396 depends on $(ld-option,--gc-sections)
1398 Enable this if you want to do dead code and data elimination with
1399 the linker by compiling with -ffunction-sections -fdata-sections,
1400 and linking with --gc-sections.
1402 This can reduce on disk and in-memory size of the kernel
1403 code and static data, particularly for small configs and
1404 on small systems. This has the possibility of introducing
1405 silently broken kernel if the required annotations are not
1406 present. This option is not well tested yet, so use at your
1409 config LD_ORPHAN_WARN
1411 depends on ARCH_WANT_LD_ORPHAN_WARN
1412 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1413 depends on $(ld-option,--orphan-handling=warn)
1421 config SYSCTL_EXCEPTION_TRACE
1424 Enable support for /proc/sys/debug/exception-trace.
1426 config SYSCTL_ARCH_UNALIGN_NO_WARN
1429 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1430 Allows arch to define/use @no_unaligned_warning to possibly warn
1431 about unaligned access emulation going on under the hood.
1433 config SYSCTL_ARCH_UNALIGN_ALLOW
1436 Enable support for /proc/sys/kernel/unaligned-trap
1437 Allows arches to define/use @unaligned_enabled to runtime toggle
1438 the unaligned access emulation.
1439 see arch/parisc/kernel/unaligned.c for reference
1441 config HAVE_PCSPKR_PLATFORM
1444 # interpreter that classic socket filters depend on
1449 bool "Configure standard kernel features (expert users)"
1450 # Unhide debug options, to make the on-by-default options visible
1453 This option allows certain base kernel options and settings
1454 to be disabled or tweaked. This is for specialized
1455 environments which can tolerate a "non-standard" kernel.
1456 Only use this if you really know what you are doing.
1459 bool "Enable 16-bit UID system calls" if EXPERT
1460 depends on HAVE_UID16 && MULTIUSER
1463 This enables the legacy 16-bit UID syscall wrappers.
1466 bool "Multiple users, groups and capabilities support" if EXPERT
1469 This option enables support for non-root users, groups and
1472 If you say N here, all processes will run with UID 0, GID 0, and all
1473 possible capabilities. Saying N here also compiles out support for
1474 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1477 If unsure, say Y here.
1479 config SGETMASK_SYSCALL
1480 bool "sgetmask/ssetmask syscalls support" if EXPERT
1481 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1483 sys_sgetmask and sys_ssetmask are obsolete system calls
1484 no longer supported in libc but still enabled by default in some
1487 If unsure, leave the default option here.
1489 config SYSFS_SYSCALL
1490 bool "Sysfs syscall support" if EXPERT
1493 sys_sysfs is an obsolete system call no longer supported in libc.
1494 Note that disabling this option is more secure but might break
1495 compatibility with some systems.
1497 If unsure say Y here.
1500 bool "open by fhandle syscalls" if EXPERT
1504 If you say Y here, a user level program will be able to map
1505 file names to handle and then later use the handle for
1506 different file system operations. This is useful in implementing
1507 userspace file servers, which now track files using handles instead
1508 of names. The handle would remain the same even if file names
1509 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1513 bool "Posix Clocks & timers" if EXPERT
1516 This includes native support for POSIX timers to the kernel.
1517 Some embedded systems have no use for them and therefore they
1518 can be configured out to reduce the size of the kernel image.
1520 When this option is disabled, the following syscalls won't be
1521 available: timer_create, timer_gettime: timer_getoverrun,
1522 timer_settime, timer_delete, clock_adjtime, getitimer,
1523 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1524 clock_getres and clock_nanosleep syscalls will be limited to
1525 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1531 bool "Enable support for printk" if EXPERT
1534 This option enables normal printk support. Removing it
1535 eliminates most of the message strings from the kernel image
1536 and makes the kernel more or less silent. As this makes it
1537 very difficult to diagnose system problems, saying N here is
1538 strongly discouraged.
1541 bool "BUG() support" if EXPERT
1544 Disabling this option eliminates support for BUG and WARN, reducing
1545 the size of your kernel image and potentially quietly ignoring
1546 numerous fatal conditions. You should only consider disabling this
1547 option for embedded systems with no facilities for reporting errors.
1553 bool "Enable ELF core dumps" if EXPERT
1555 Enable support for generating core dumps. Disabling saves about 4k.
1558 config PCSPKR_PLATFORM
1559 bool "Enable PC-Speaker support" if EXPERT
1560 depends on HAVE_PCSPKR_PLATFORM
1564 This option allows to disable the internal PC-Speaker
1565 support, saving some memory.
1569 bool "Enable full-sized data structures for core" if EXPERT
1571 Disabling this option reduces the size of miscellaneous core
1572 kernel data structures. This saves memory on small machines,
1573 but may reduce performance.
1576 bool "Enable futex support" if EXPERT
1580 Disabling this option will cause the kernel to be built without
1581 support for "fast userspace mutexes". The resulting kernel may not
1582 run glibc-based applications correctly.
1586 depends on FUTEX && RT_MUTEXES
1589 config HAVE_FUTEX_CMPXCHG
1593 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1594 is implemented and always working. This removes a couple of runtime
1598 bool "Enable eventpoll support" if EXPERT
1601 Disabling this option will cause the kernel to be built without
1602 support for epoll family of system calls.
1605 bool "Enable signalfd() system call" if EXPERT
1608 Enable the signalfd() system call that allows to receive signals
1609 on a file descriptor.
1614 bool "Enable timerfd() system call" if EXPERT
1617 Enable the timerfd() system call that allows to receive timer
1618 events on a file descriptor.
1623 bool "Enable eventfd() system call" if EXPERT
1626 Enable the eventfd() system call that allows to receive both
1627 kernel notification (ie. KAIO) or userspace notifications.
1632 bool "Use full shmem filesystem" if EXPERT
1636 The shmem is an internal filesystem used to manage shared memory.
1637 It is backed by swap and manages resource limits. It is also exported
1638 to userspace as tmpfs if TMPFS is enabled. Disabling this
1639 option replaces shmem and tmpfs with the much simpler ramfs code,
1640 which may be appropriate on small systems without swap.
1643 bool "Enable AIO support" if EXPERT
1646 This option enables POSIX asynchronous I/O which may by used
1647 by some high performance threaded applications. Disabling
1648 this option saves about 7k.
1651 bool "Enable IO uring support" if EXPERT
1655 This option enables support for the io_uring interface, enabling
1656 applications to submit and complete IO through submission and
1657 completion rings that are shared between the kernel and application.
1659 config ADVISE_SYSCALLS
1660 bool "Enable madvise/fadvise syscalls" if EXPERT
1663 This option enables the madvise and fadvise syscalls, used by
1664 applications to advise the kernel about their future memory or file
1665 usage, improving performance. If building an embedded system where no
1666 applications use these syscalls, you can disable this option to save
1669 config HAVE_ARCH_USERFAULTFD_WP
1672 Arch has userfaultfd write protection support
1674 config HAVE_ARCH_USERFAULTFD_MINOR
1677 Arch has userfaultfd minor fault support
1680 bool "Enable membarrier() system call" if EXPERT
1683 Enable the membarrier() system call that allows issuing memory
1684 barriers across all running threads, which can be used to distribute
1685 the cost of user-space memory barriers asymmetrically by transforming
1686 pairs of memory barriers into pairs consisting of membarrier() and a
1692 bool "Load all symbols for debugging/ksymoops" if EXPERT
1695 Say Y here to let the kernel print out symbolic crash information and
1696 symbolic stack backtraces. This increases the size of the kernel
1697 somewhat, as all symbols have to be loaded into the kernel image.
1700 bool "Include all symbols in kallsyms"
1701 depends on DEBUG_KERNEL && KALLSYMS
1703 Normally kallsyms only contains the symbols of functions for nicer
1704 OOPS messages and backtraces (i.e., symbols from the text and inittext
1705 sections). This is sufficient for most cases. And only in very rare
1706 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1707 names of variables from the data sections, etc).
1709 This option makes sure that all symbols are loaded into the kernel
1710 image (i.e., symbols from all sections) in cost of increased kernel
1711 size (depending on the kernel configuration, it may be 300KiB or
1712 something like this).
1714 Say N unless you really need all symbols.
1716 config KALLSYMS_ABSOLUTE_PERCPU
1719 default X86_64 && SMP
1721 config KALLSYMS_BASE_RELATIVE
1726 Instead of emitting them as absolute values in the native word size,
1727 emit the symbol references in the kallsyms table as 32-bit entries,
1728 each containing a relative value in the range [base, base + U32_MAX]
1729 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1730 an absolute value in the range [0, S32_MAX] or a relative value in the
1731 range [base, base + S32_MAX], where base is the lowest relative symbol
1732 address encountered in the image.
1734 On 64-bit builds, this reduces the size of the address table by 50%,
1735 but more importantly, it results in entries whose values are build
1736 time constants, and no relocation pass is required at runtime to fix
1737 up the entries based on the runtime load address of the kernel.
1739 # end of the "standard kernel features (expert users)" menu
1741 # syscall, maps, verifier
1744 bool "Enable userfaultfd() system call"
1747 Enable the userfaultfd() system call that allows to intercept and
1748 handle page faults in userland.
1750 config ARCH_HAS_MEMBARRIER_CALLBACKS
1753 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1757 bool "Enable kcmp() system call" if EXPERT
1759 Enable the kernel resource comparison system call. It provides
1760 user-space with the ability to compare two processes to see if they
1761 share a common resource, such as a file descriptor or even virtual
1767 bool "Enable rseq() system call" if EXPERT
1769 depends on HAVE_RSEQ
1772 Enable the restartable sequences system call. It provides a
1773 user-space cache for the current CPU number value, which
1774 speeds up getting the current CPU number from user-space,
1775 as well as an ABI to speed up user-space operations on
1782 bool "Enabled debugging of rseq() system call" if EXPERT
1783 depends on RSEQ && DEBUG_KERNEL
1785 Enable extra debugging checks for the rseq system call.
1790 bool "Embedded system"
1793 This option should be enabled if compiling the kernel for
1794 an embedded system so certain expert options are available
1797 config HAVE_PERF_EVENTS
1800 See tools/perf/design.txt for details.
1802 config PERF_USE_VMALLOC
1805 See tools/perf/design.txt for details
1808 bool "PC/104 support" if EXPERT
1810 Expose PC/104 form factor device drivers and options available for
1811 selection and configuration. Enable this option if your target
1812 machine has a PC/104 bus.
1814 menu "Kernel Performance Events And Counters"
1817 bool "Kernel performance events and counters"
1818 default y if PROFILING
1819 depends on HAVE_PERF_EVENTS
1823 Enable kernel support for various performance events provided
1824 by software and hardware.
1826 Software events are supported either built-in or via the
1827 use of generic tracepoints.
1829 Most modern CPUs support performance events via performance
1830 counter registers. These registers count the number of certain
1831 types of hw events: such as instructions executed, cachemisses
1832 suffered, or branches mis-predicted - without slowing down the
1833 kernel or applications. These registers can also trigger interrupts
1834 when a threshold number of events have passed - and can thus be
1835 used to profile the code that runs on that CPU.
1837 The Linux Performance Event subsystem provides an abstraction of
1838 these software and hardware event capabilities, available via a
1839 system call and used by the "perf" utility in tools/perf/. It
1840 provides per task and per CPU counters, and it provides event
1841 capabilities on top of those.
1845 config DEBUG_PERF_USE_VMALLOC
1847 bool "Debug: use vmalloc to back perf mmap() buffers"
1848 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1849 select PERF_USE_VMALLOC
1851 Use vmalloc memory to back perf mmap() buffers.
1853 Mostly useful for debugging the vmalloc code on platforms
1854 that don't require it.
1860 config VM_EVENT_COUNTERS
1862 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1864 VM event counters are needed for event counts to be shown.
1865 This option allows the disabling of the VM event counters
1866 on EXPERT systems. /proc/vmstat will only show page counts
1867 if VM event counters are disabled.
1871 bool "Enable SLUB debugging support" if EXPERT
1872 depends on SLUB && SYSFS
1874 SLUB has extensive debug support features. Disabling these can
1875 result in significant savings in code size. This also disables
1876 SLUB sysfs support. /sys/slab will not exist and there will be
1877 no support for cache validation etc.
1880 bool "Disable heap randomization"
1883 Randomizing heap placement makes heap exploits harder, but it
1884 also breaks ancient binaries (including anything libc5 based).
1885 This option changes the bootup default to heap randomization
1886 disabled, and can be overridden at runtime by setting
1887 /proc/sys/kernel/randomize_va_space to 2.
1889 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1892 prompt "Choose SLAB allocator"
1895 This option allows to select a slab allocator.
1899 depends on !PREEMPT_RT
1900 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1902 The regular slab allocator that is established and known to work
1903 well in all environments. It organizes cache hot objects in
1904 per cpu and per node queues.
1907 bool "SLUB (Unqueued Allocator)"
1908 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1910 SLUB is a slab allocator that minimizes cache line usage
1911 instead of managing queues of cached objects (SLAB approach).
1912 Per cpu caching is realized using slabs of objects instead
1913 of queues of objects. SLUB can use memory efficiently
1914 and has enhanced diagnostics. SLUB is the default choice for
1919 bool "SLOB (Simple Allocator)"
1920 depends on !PREEMPT_RT
1922 SLOB replaces the stock allocator with a drastically simpler
1923 allocator. SLOB is generally more space efficient but
1924 does not perform as well on large systems.
1928 config SLAB_MERGE_DEFAULT
1929 bool "Allow slab caches to be merged"
1932 For reduced kernel memory fragmentation, slab caches can be
1933 merged when they share the same size and other characteristics.
1934 This carries a risk of kernel heap overflows being able to
1935 overwrite objects from merged caches (and more easily control
1936 cache layout), which makes such heap attacks easier to exploit
1937 by attackers. By keeping caches unmerged, these kinds of exploits
1938 can usually only damage objects in the same cache. To disable
1939 merging at runtime, "slab_nomerge" can be passed on the kernel
1942 config SLAB_FREELIST_RANDOM
1943 bool "Randomize slab freelist"
1944 depends on SLAB || SLUB
1946 Randomizes the freelist order used on creating new pages. This
1947 security feature reduces the predictability of the kernel slab
1948 allocator against heap overflows.
1950 config SLAB_FREELIST_HARDENED
1951 bool "Harden slab freelist metadata"
1952 depends on SLAB || SLUB
1954 Many kernel heap attacks try to target slab cache metadata and
1955 other infrastructure. This options makes minor performance
1956 sacrifices to harden the kernel slab allocator against common
1957 freelist exploit methods. Some slab implementations have more
1958 sanity-checking than others. This option is most effective with
1961 config SHUFFLE_PAGE_ALLOCATOR
1962 bool "Page allocator randomization"
1963 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1965 Randomization of the page allocator improves the average
1966 utilization of a direct-mapped memory-side-cache. See section
1967 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1968 6.2a specification for an example of how a platform advertises
1969 the presence of a memory-side-cache. There are also incidental
1970 security benefits as it reduces the predictability of page
1971 allocations to compliment SLAB_FREELIST_RANDOM, but the
1972 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1973 10th order of pages is selected based on cache utilization
1976 While the randomization improves cache utilization it may
1977 negatively impact workloads on platforms without a cache. For
1978 this reason, by default, the randomization is enabled only
1979 after runtime detection of a direct-mapped memory-side-cache.
1980 Otherwise, the randomization may be force enabled with the
1981 'page_alloc.shuffle' kernel command line parameter.
1985 config SLUB_CPU_PARTIAL
1987 depends on SLUB && SMP
1988 bool "SLUB per cpu partial cache"
1990 Per cpu partial caches accelerate objects allocation and freeing
1991 that is local to a processor at the price of more indeterminism
1992 in the latency of the free. On overflow these caches will be cleared
1993 which requires the taking of locks that may cause latency spikes.
1994 Typically one would choose no for a realtime system.
1996 config MMAP_ALLOW_UNINITIALIZED
1997 bool "Allow mmapped anonymous memory to be uninitialized"
1998 depends on EXPERT && !MMU
2001 Normally, and according to the Linux spec, anonymous memory obtained
2002 from mmap() has its contents cleared before it is passed to
2003 userspace. Enabling this config option allows you to request that
2004 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
2005 providing a huge performance boost. If this option is not enabled,
2006 then the flag will be ignored.
2008 This is taken advantage of by uClibc's malloc(), and also by
2009 ELF-FDPIC binfmt's brk and stack allocator.
2011 Because of the obvious security issues, this option should only be
2012 enabled on embedded devices where you control what is run in
2013 userspace. Since that isn't generally a problem on no-MMU systems,
2014 it is normally safe to say Y here.
2016 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2018 config SYSTEM_DATA_VERIFICATION
2020 select SYSTEM_TRUSTED_KEYRING
2024 select ASYMMETRIC_KEY_TYPE
2025 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2028 select X509_CERTIFICATE_PARSER
2029 select PKCS7_MESSAGE_PARSER
2031 Provide PKCS#7 message verification using the contents of the system
2032 trusted keyring to provide public keys. This then can be used for
2033 module verification, kexec image verification and firmware blob
2037 bool "Profiling support"
2039 Say Y here to enable the extended profiling support mechanisms used
2043 # Place an empty function call at each tracepoint site. Can be
2044 # dynamically changed for a probe function.
2049 endmenu # General setup
2051 source "arch/Kconfig"
2058 default 0 if BASE_FULL
2059 default 1 if !BASE_FULL
2061 config MODULE_SIG_FORMAT
2063 select SYSTEM_DATA_VERIFICATION
2066 bool "Enable loadable module support"
2069 Kernel modules are small pieces of compiled code which can
2070 be inserted in the running kernel, rather than being
2071 permanently built into the kernel. You use the "modprobe"
2072 tool to add (and sometimes remove) them. If you say Y here,
2073 many parts of the kernel can be built as modules (by
2074 answering M instead of Y where indicated): this is most
2075 useful for infrequently used options which are not required
2076 for booting. For more information, see the man pages for
2077 modprobe, lsmod, modinfo, insmod and rmmod.
2079 If you say Y here, you will need to run "make
2080 modules_install" to put the modules under /lib/modules/
2081 where modprobe can find them (you may need to be root to do
2088 config MODULE_FORCE_LOAD
2089 bool "Forced module loading"
2092 Allow loading of modules without version information (ie. modprobe
2093 --force). Forced module loading sets the 'F' (forced) taint flag and
2094 is usually a really bad idea.
2096 config MODULE_UNLOAD
2097 bool "Module unloading"
2099 Without this option you will not be able to unload any
2100 modules (note that some modules may not be unloadable
2101 anyway), which makes your kernel smaller, faster
2102 and simpler. If unsure, say Y.
2104 config MODULE_FORCE_UNLOAD
2105 bool "Forced module unloading"
2106 depends on MODULE_UNLOAD
2108 This option allows you to force a module to unload, even if the
2109 kernel believes it is unsafe: the kernel will remove the module
2110 without waiting for anyone to stop using it (using the -f option to
2111 rmmod). This is mainly for kernel developers and desperate users.
2115 bool "Module versioning support"
2117 Usually, you have to use modules compiled with your kernel.
2118 Saying Y here makes it sometimes possible to use modules
2119 compiled for different kernels, by adding enough information
2120 to the modules to (hopefully) spot any changes which would
2121 make them incompatible with the kernel you are running. If
2124 config ASM_MODVERSIONS
2126 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2128 This enables module versioning for exported symbols also from
2129 assembly. This can be enabled only when the target architecture
2132 config MODULE_REL_CRCS
2134 depends on MODVERSIONS
2136 config MODULE_SRCVERSION_ALL
2137 bool "Source checksum for all modules"
2139 Modules which contain a MODULE_VERSION get an extra "srcversion"
2140 field inserted into their modinfo section, which contains a
2141 sum of the source files which made it. This helps maintainers
2142 see exactly which source was used to build a module (since
2143 others sometimes change the module source without updating
2144 the version). With this option, such a "srcversion" field
2145 will be created for all modules. If unsure, say N.
2148 bool "Module signature verification"
2149 select MODULE_SIG_FORMAT
2151 Check modules for valid signatures upon load: the signature
2152 is simply appended to the module. For more information see
2153 <file:Documentation/admin-guide/module-signing.rst>.
2155 Note that this option adds the OpenSSL development packages as a
2156 kernel build dependency so that the signing tool can use its crypto
2159 You should enable this option if you wish to use either
2160 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2161 another LSM - otherwise unsigned modules will be loadable regardless
2162 of the lockdown policy.
2164 !!!WARNING!!! If you enable this option, you MUST make sure that the
2165 module DOES NOT get stripped after being signed. This includes the
2166 debuginfo strip done by some packagers (such as rpmbuild) and
2167 inclusion into an initramfs that wants the module size reduced.
2169 config MODULE_SIG_FORCE
2170 bool "Require modules to be validly signed"
2171 depends on MODULE_SIG
2173 Reject unsigned modules or signed modules for which we don't have a
2174 key. Without this, such modules will simply taint the kernel.
2176 config MODULE_SIG_ALL
2177 bool "Automatically sign all modules"
2179 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2181 Sign all modules during make modules_install. Without this option,
2182 modules must be signed manually, using the scripts/sign-file tool.
2184 comment "Do not forget to sign required modules with scripts/sign-file"
2185 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2188 prompt "Which hash algorithm should modules be signed with?"
2189 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2191 This determines which sort of hashing algorithm will be used during
2192 signature generation. This algorithm _must_ be built into the kernel
2193 directly so that signature verification can take place. It is not
2194 possible to load a signed module containing the algorithm to check
2195 the signature on that module.
2197 config MODULE_SIG_SHA1
2198 bool "Sign modules with SHA-1"
2201 config MODULE_SIG_SHA224
2202 bool "Sign modules with SHA-224"
2203 select CRYPTO_SHA256
2205 config MODULE_SIG_SHA256
2206 bool "Sign modules with SHA-256"
2207 select CRYPTO_SHA256
2209 config MODULE_SIG_SHA384
2210 bool "Sign modules with SHA-384"
2211 select CRYPTO_SHA512
2213 config MODULE_SIG_SHA512
2214 bool "Sign modules with SHA-512"
2215 select CRYPTO_SHA512
2219 config MODULE_SIG_HASH
2221 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2222 default "sha1" if MODULE_SIG_SHA1
2223 default "sha224" if MODULE_SIG_SHA224
2224 default "sha256" if MODULE_SIG_SHA256
2225 default "sha384" if MODULE_SIG_SHA384
2226 default "sha512" if MODULE_SIG_SHA512
2229 prompt "Module compression mode"
2231 This option allows you to choose the algorithm which will be used to
2232 compress modules when 'make modules_install' is run. (or, you can
2233 choose to not compress modules at all.)
2235 External modules will also be compressed in the same way during the
2238 For modules inside an initrd or initramfs, it's more efficient to
2239 compress the whole initrd or initramfs instead.
2241 This is fully compatible with signed modules.
2243 Please note that the tool used to load modules needs to support the
2244 corresponding algorithm. module-init-tools MAY support gzip, and kmod
2245 MAY support gzip, xz and zstd.
2247 Your build system needs to provide the appropriate compression tool
2248 to compress the modules.
2250 If in doubt, select 'None'.
2252 config MODULE_COMPRESS_NONE
2255 Do not compress modules. The installed modules are suffixed
2258 config MODULE_COMPRESS_GZIP
2261 Compress modules with GZIP. The installed modules are suffixed
2264 config MODULE_COMPRESS_XZ
2267 Compress modules with XZ. The installed modules are suffixed
2270 config MODULE_COMPRESS_ZSTD
2273 Compress modules with ZSTD. The installed modules are suffixed
2278 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2279 bool "Allow loading of modules with missing namespace imports"
2281 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2282 a namespace. A module that makes use of a symbol exported with such a
2283 namespace is required to import the namespace via MODULE_IMPORT_NS().
2284 There is no technical reason to enforce correct namespace imports,
2285 but it creates consistency between symbols defining namespaces and
2286 users importing namespaces they make use of. This option relaxes this
2287 requirement and lifts the enforcement when loading a module.
2291 config MODPROBE_PATH
2292 string "Path to modprobe binary"
2293 default "/sbin/modprobe"
2295 When kernel code requests a module, it does so by calling
2296 the "modprobe" userspace utility. This option allows you to
2297 set the path where that binary is found. This can be changed
2298 at runtime via the sysctl file
2299 /proc/sys/kernel/modprobe. Setting this to the empty string
2300 removes the kernel's ability to request modules (but
2301 userspace can still load modules explicitly).
2303 config TRIM_UNUSED_KSYMS
2304 bool "Trim unused exported kernel symbols" if EXPERT
2305 depends on !COMPILE_TEST
2307 The kernel and some modules make many symbols available for
2308 other modules to use via EXPORT_SYMBOL() and variants. Depending
2309 on the set of modules being selected in your kernel configuration,
2310 many of those exported symbols might never be used.
2312 This option allows for unused exported symbols to be dropped from
2313 the build. In turn, this provides the compiler more opportunities
2314 (especially when using LTO) for optimizing the code and reducing
2315 binary size. This might have some security advantages as well.
2317 If unsure, or if you need to build out-of-tree modules, say N.
2319 config UNUSED_KSYMS_WHITELIST
2320 string "Whitelist of symbols to keep in ksymtab"
2321 depends on TRIM_UNUSED_KSYMS
2323 By default, all unused exported symbols will be un-exported from the
2324 build when TRIM_UNUSED_KSYMS is selected.
2326 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2327 exported at all times, even in absence of in-tree users. The value to
2328 set here is the path to a text file containing the list of symbols,
2329 one per line. The path can be absolute, or relative to the kernel
2334 config MODULES_TREE_LOOKUP
2336 depends on PERF_EVENTS || TRACING || CFI_CLANG
2338 config INIT_ALL_POSSIBLE
2341 Back when each arch used to define their own cpu_online_mask and
2342 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2343 with all 1s, and others with all 0s. When they were centralised,
2344 it was better to provide this option than to break all the archs
2345 and have several arch maintainers pursuing me down dark alleys.
2347 source "block/Kconfig"
2349 config PREEMPT_NOTIFIERS
2359 Build a simple ASN.1 grammar compiler that produces a bytecode output
2360 that can be interpreted by the ASN.1 stream decoder and used to
2361 inform it as to what tags are to be expected in a stream and what
2362 functions to call on what tags.
2364 source "kernel/Kconfig.locks"
2366 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2369 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2372 # It may be useful for an architecture to override the definitions of the
2373 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2374 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2375 # different calling convention for syscalls. They can also override the
2376 # macros for not-implemented syscalls in kernel/sys_ni.c and
2377 # kernel/time/posix-stubs.c. All these overrides need to be available in
2378 # <asm/syscall_wrapper.h>.
2379 config ARCH_HAS_SYSCALL_WRAPPER