1 # SPDX-License-Identifier: GPL-2.0-only
6 default "/lib/modules/$(shell,uname -r)/.config"
7 default "/etc/kernel-config"
8 default "/boot/config-$(shell,uname -r)"
10 default "arch/$(ARCH)/defconfig"
13 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
17 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
21 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
24 def_bool $(success,$(LD) -v | head -n 1 | grep -q LLD)
28 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
31 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
33 config CC_HAS_ASM_GOTO
34 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
36 config CC_HAS_ASM_GOTO_TIED_OUTPUT
37 depends on CC_HAS_ASM_GOTO_OUTPUT
38 # Detect buggy gcc and clang, fixed in gcc-11 clang-14.
39 def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .\n": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
41 config CC_HAS_ASM_GOTO_OUTPUT
42 depends on CC_HAS_ASM_GOTO
43 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)
45 config TOOLS_SUPPORT_RELR
46 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
48 config CC_HAS_ASM_INLINE
49 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
58 config BUILDTIME_EXTABLE_SORT
61 config THREAD_INFO_IN_TASK
64 Select this to move thread_info off the stack into task_struct. To
65 make this work, an arch will need to remove all thread_info fields
66 except flags and fix any runtime bugs.
68 One subtle change that will be needed is to use try_get_task_stack()
69 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
78 depends on BROKEN || !SMP
81 config INIT_ENV_ARG_LIMIT
86 Maximum of each of the number of arguments and environment
87 variables passed to init from the kernel command line.
90 bool "Compile also drivers which will not load"
93 Some drivers can be compiled on a different platform than they are
94 intended to be run on. Despite they cannot be loaded there (or even
95 when they load they cannot be used due to missing HW support),
96 developers still, opposing to distributors, might want to build such
97 drivers to compile-test them.
99 If you are a developer and want to build everything available, say Y
100 here. If you are a user/distributor, say N here to exclude useless
101 drivers to be distributed.
103 config UAPI_HEADER_TEST
104 bool "Compile test UAPI headers"
105 depends on HEADERS_INSTALL && CC_CAN_LINK
107 Compile test headers exported to user-space to ensure they are
108 self-contained, i.e. compilable as standalone units.
110 If you are a developer or tester and want to ensure the exported
111 headers are self-contained, say Y here. Otherwise, choose N.
114 string "Local version - append to kernel release"
116 Append an extra string to the end of your kernel version.
117 This will show up when you type uname, for example.
118 The string you set here will be appended after the contents of
119 any files with a filename matching localversion* in your
120 object and source tree, in that order. Your total string can
121 be a maximum of 64 characters.
123 config LOCALVERSION_AUTO
124 bool "Automatically append version information to the version string"
126 depends on !COMPILE_TEST
128 This will try to automatically determine if the current tree is a
129 release tree by looking for git tags that belong to the current
130 top of tree revision.
132 A string of the format -gxxxxxxxx will be added to the localversion
133 if a git-based tree is found. The string generated by this will be
134 appended after any matching localversion* files, and after the value
135 set in CONFIG_LOCALVERSION.
137 (The actual string used here is the first eight characters produced
138 by running the command:
140 $ git rev-parse --verify HEAD
142 which is done within the script "scripts/setlocalversion".)
145 string "Build ID Salt"
148 The build ID is used to link binaries and their debug info. Setting
149 this option will use the value in the calculation of the build id.
150 This is mostly useful for distributions which want to ensure the
151 build is unique between builds. It's safe to leave the default.
153 config HAVE_KERNEL_GZIP
156 config HAVE_KERNEL_BZIP2
159 config HAVE_KERNEL_LZMA
162 config HAVE_KERNEL_XZ
165 config HAVE_KERNEL_LZO
168 config HAVE_KERNEL_LZ4
171 config HAVE_KERNEL_UNCOMPRESSED
175 prompt "Kernel compression mode"
177 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
179 The linux kernel is a kind of self-extracting executable.
180 Several compression algorithms are available, which differ
181 in efficiency, compression and decompression speed.
182 Compression speed is only relevant when building a kernel.
183 Decompression speed is relevant at each boot.
185 If you have any problems with bzip2 or lzma compressed
186 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
187 version of this functionality (bzip2 only), for 2.4, was
188 supplied by Christian Ludwig)
190 High compression options are mostly useful for users, who
191 are low on disk space (embedded systems), but for whom ram
194 If in doubt, select 'gzip'
198 depends on HAVE_KERNEL_GZIP
200 The old and tried gzip compression. It provides a good balance
201 between compression ratio and decompression speed.
205 depends on HAVE_KERNEL_BZIP2
207 Its compression ratio and speed is intermediate.
208 Decompression speed is slowest among the choices. The kernel
209 size is about 10% smaller with bzip2, in comparison to gzip.
210 Bzip2 uses a large amount of memory. For modern kernels you
211 will need at least 8MB RAM or more for booting.
215 depends on HAVE_KERNEL_LZMA
217 This compression algorithm's ratio is best. Decompression speed
218 is between gzip and bzip2. Compression is slowest.
219 The kernel size is about 33% smaller with LZMA in comparison to gzip.
223 depends on HAVE_KERNEL_XZ
225 XZ uses the LZMA2 algorithm and instruction set specific
226 BCJ filters which can improve compression ratio of executable
227 code. The size of the kernel is about 30% smaller with XZ in
228 comparison to gzip. On architectures for which there is a BCJ
229 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
230 will create a few percent smaller kernel than plain LZMA.
232 The speed is about the same as with LZMA: The decompression
233 speed of XZ is better than that of bzip2 but worse than gzip
234 and LZO. Compression is slow.
238 depends on HAVE_KERNEL_LZO
240 Its compression ratio is the poorest among the choices. The kernel
241 size is about 10% bigger than gzip; however its speed
242 (both compression and decompression) is the fastest.
246 depends on HAVE_KERNEL_LZ4
248 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
249 A preliminary version of LZ4 de/compression tool is available at
250 <https://code.google.com/p/lz4/>.
252 Its compression ratio is worse than LZO. The size of the kernel
253 is about 8% bigger than LZO. But the decompression speed is
256 config KERNEL_UNCOMPRESSED
258 depends on HAVE_KERNEL_UNCOMPRESSED
260 Produce uncompressed kernel image. This option is usually not what
261 you want. It is useful for debugging the kernel in slow simulation
262 environments, where decompressing and moving the kernel is awfully
263 slow. This option allows early boot code to skip the decompressor
264 and jump right at uncompressed kernel image.
268 config DEFAULT_HOSTNAME
269 string "Default hostname"
272 This option determines the default system hostname before userspace
273 calls sethostname(2). The kernel traditionally uses "(none)" here,
274 but you may wish to use a different default here to make a minimal
275 system more usable with less configuration.
278 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
279 # add proper SWAP support to them, in which case this can be remove.
285 bool "Support for paging of anonymous memory (swap)"
286 depends on MMU && BLOCK && !ARCH_NO_SWAP
289 This option allows you to choose whether you want to have support
290 for so called swap devices or swap files in your kernel that are
291 used to provide more virtual memory than the actual RAM present
292 in your computer. If unsure say Y.
297 Inter Process Communication is a suite of library functions and
298 system calls which let processes (running programs) synchronize and
299 exchange information. It is generally considered to be a good thing,
300 and some programs won't run unless you say Y here. In particular, if
301 you want to run the DOS emulator dosemu under Linux (read the
302 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
303 you'll need to say Y here.
305 You can find documentation about IPC with "info ipc" and also in
306 section 6.4 of the Linux Programmer's Guide, available from
307 <http://www.tldp.org/guides.html>.
309 config SYSVIPC_SYSCTL
316 bool "POSIX Message Queues"
319 POSIX variant of message queues is a part of IPC. In POSIX message
320 queues every message has a priority which decides about succession
321 of receiving it by a process. If you want to compile and run
322 programs written e.g. for Solaris with use of its POSIX message
323 queues (functions mq_*) say Y here.
325 POSIX message queues are visible as a filesystem called 'mqueue'
326 and can be mounted somewhere if you want to do filesystem
327 operations on message queues.
331 config POSIX_MQUEUE_SYSCTL
333 depends on POSIX_MQUEUE
337 config CROSS_MEMORY_ATTACH
338 bool "Enable process_vm_readv/writev syscalls"
342 Enabling this option adds the system calls process_vm_readv and
343 process_vm_writev which allow a process with the correct privileges
344 to directly read from or write to another process' address space.
345 See the man page for more details.
348 bool "uselib syscall"
349 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
351 This option enables the uselib syscall, a system call used in the
352 dynamic linker from libc5 and earlier. glibc does not use this
353 system call. If you intend to run programs built on libc5 or
354 earlier, you may need to enable this syscall. Current systems
355 running glibc can safely disable this.
358 bool "Auditing support"
361 Enable auditing infrastructure that can be used with another
362 kernel subsystem, such as SELinux (which requires this for
363 logging of avc messages output). System call auditing is included
364 on architectures which support it.
366 config HAVE_ARCH_AUDITSYSCALL
371 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
374 source "kernel/irq/Kconfig"
375 source "kernel/time/Kconfig"
376 source "kernel/Kconfig.preempt"
378 menu "CPU/Task time and stats accounting"
380 config VIRT_CPU_ACCOUNTING
384 prompt "Cputime accounting"
385 default TICK_CPU_ACCOUNTING if !PPC64
386 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
388 # Kind of a stub config for the pure tick based cputime accounting
389 config TICK_CPU_ACCOUNTING
390 bool "Simple tick based cputime accounting"
391 depends on !S390 && !NO_HZ_FULL
393 This is the basic tick based cputime accounting that maintains
394 statistics about user, system and idle time spent on per jiffies
399 config VIRT_CPU_ACCOUNTING_NATIVE
400 bool "Deterministic task and CPU time accounting"
401 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
402 select VIRT_CPU_ACCOUNTING
404 Select this option to enable more accurate task and CPU time
405 accounting. This is done by reading a CPU counter on each
406 kernel entry and exit and on transitions within the kernel
407 between system, softirq and hardirq state, so there is a
408 small performance impact. In the case of s390 or IBM POWER > 5,
409 this also enables accounting of stolen time on logically-partitioned
412 config VIRT_CPU_ACCOUNTING_GEN
413 bool "Full dynticks CPU time accounting"
414 depends on HAVE_CONTEXT_TRACKING
415 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
416 depends on GENERIC_CLOCKEVENTS
417 select VIRT_CPU_ACCOUNTING
418 select CONTEXT_TRACKING
420 Select this option to enable task and CPU time accounting on full
421 dynticks systems. This accounting is implemented by watching every
422 kernel-user boundaries using the context tracking subsystem.
423 The accounting is thus performed at the expense of some significant
426 For now this is only useful if you are working on the full
427 dynticks subsystem development.
433 config IRQ_TIME_ACCOUNTING
434 bool "Fine granularity task level IRQ time accounting"
435 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
437 Select this option to enable fine granularity task irq time
438 accounting. This is done by reading a timestamp on each
439 transitions between softirq and hardirq state, so there can be a
440 small performance impact.
442 If in doubt, say N here.
444 config HAVE_SCHED_AVG_IRQ
446 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
449 config BSD_PROCESS_ACCT
450 bool "BSD Process Accounting"
453 If you say Y here, a user level program will be able to instruct the
454 kernel (via a special system call) to write process accounting
455 information to a file: whenever a process exits, information about
456 that process will be appended to the file by the kernel. The
457 information includes things such as creation time, owning user,
458 command name, memory usage, controlling terminal etc. (the complete
459 list is in the struct acct in <file:include/linux/acct.h>). It is
460 up to the user level program to do useful things with this
461 information. This is generally a good idea, so say Y.
463 config BSD_PROCESS_ACCT_V3
464 bool "BSD Process Accounting version 3 file format"
465 depends on BSD_PROCESS_ACCT
468 If you say Y here, the process accounting information is written
469 in a new file format that also logs the process IDs of each
470 process and its parent. Note that this file format is incompatible
471 with previous v0/v1/v2 file formats, so you will need updated tools
472 for processing it. A preliminary version of these tools is available
473 at <http://www.gnu.org/software/acct/>.
476 bool "Export task/process statistics through netlink"
481 Export selected statistics for tasks/processes through the
482 generic netlink interface. Unlike BSD process accounting, the
483 statistics are available during the lifetime of tasks/processes as
484 responses to commands. Like BSD accounting, they are sent to user
489 config TASK_DELAY_ACCT
490 bool "Enable per-task delay accounting"
494 Collect information on time spent by a task waiting for system
495 resources like cpu, synchronous block I/O completion and swapping
496 in pages. Such statistics can help in setting a task's priorities
497 relative to other tasks for cpu, io, rss limits etc.
502 bool "Enable extended accounting over taskstats"
505 Collect extended task accounting data and send the data
506 to userland for processing over the taskstats interface.
510 config TASK_IO_ACCOUNTING
511 bool "Enable per-task storage I/O accounting"
512 depends on TASK_XACCT
514 Collect information on the number of bytes of storage I/O which this
520 bool "Pressure stall information tracking"
522 Collect metrics that indicate how overcommitted the CPU, memory,
523 and IO capacity are in the system.
525 If you say Y here, the kernel will create /proc/pressure/ with the
526 pressure statistics files cpu, memory, and io. These will indicate
527 the share of walltime in which some or all tasks in the system are
528 delayed due to contention of the respective resource.
530 In kernels with cgroup support, cgroups (cgroup2 only) will
531 have cpu.pressure, memory.pressure, and io.pressure files,
532 which aggregate pressure stalls for the grouped tasks only.
534 For more details see Documentation/accounting/psi.rst.
538 config PSI_DEFAULT_DISABLED
539 bool "Require boot parameter to enable pressure stall information tracking"
543 If set, pressure stall information tracking will be disabled
544 per default but can be enabled through passing psi=1 on the
545 kernel commandline during boot.
547 This feature adds some code to the task wakeup and sleep
548 paths of the scheduler. The overhead is too low to affect
549 common scheduling-intense workloads in practice (such as
550 webservers, memcache), but it does show up in artificial
551 scheduler stress tests, such as hackbench.
553 If you are paranoid and not sure what the kernel will be
558 endmenu # "CPU/Task time and stats accounting"
562 depends on SMP || COMPILE_TEST
565 Make sure that CPUs running critical tasks are not disturbed by
566 any source of "noise" such as unbound workqueues, timers, kthreads...
567 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
568 the "isolcpus=" boot parameter.
572 source "kernel/rcu/Kconfig"
579 tristate "Kernel .config support"
581 This option enables the complete Linux kernel ".config" file
582 contents to be saved in the kernel. It provides documentation
583 of which kernel options are used in a running kernel or in an
584 on-disk kernel. This information can be extracted from the kernel
585 image file with the script scripts/extract-ikconfig and used as
586 input to rebuild the current kernel or to build another kernel.
587 It can also be extracted from a running kernel by reading
588 /proc/config.gz if enabled (below).
591 bool "Enable access to .config through /proc/config.gz"
592 depends on IKCONFIG && PROC_FS
594 This option enables access to the kernel configuration file
595 through /proc/config.gz.
598 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
601 This option enables access to the in-kernel headers that are generated during
602 the build process. These can be used to build eBPF tracing programs,
603 or similar programs. If you build the headers as a module, a module called
604 kheaders.ko is built which can be loaded on-demand to get access to headers.
607 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
608 range 12 25 if !H8300
613 Select the minimal kernel log buffer size as a power of 2.
614 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
615 parameter, see below. Any higher size also might be forced
616 by "log_buf_len" boot parameter.
626 config LOG_CPU_MAX_BUF_SHIFT
627 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
630 default 12 if !BASE_SMALL
631 default 0 if BASE_SMALL
634 This option allows to increase the default ring buffer size
635 according to the number of CPUs. The value defines the contribution
636 of each CPU as a power of 2. The used space is typically only few
637 lines however it might be much more when problems are reported,
640 The increased size means that a new buffer has to be allocated and
641 the original static one is unused. It makes sense only on systems
642 with more CPUs. Therefore this value is used only when the sum of
643 contributions is greater than the half of the default kernel ring
644 buffer as defined by LOG_BUF_SHIFT. The default values are set
645 so that more than 64 CPUs are needed to trigger the allocation.
647 Also this option is ignored when "log_buf_len" kernel parameter is
648 used as it forces an exact (power of two) size of the ring buffer.
650 The number of possible CPUs is used for this computation ignoring
651 hotplugging making the computation optimal for the worst case
652 scenario while allowing a simple algorithm to be used from bootup.
654 Examples shift values and their meaning:
655 17 => 128 KB for each CPU
656 16 => 64 KB for each CPU
657 15 => 32 KB for each CPU
658 14 => 16 KB for each CPU
659 13 => 8 KB for each CPU
660 12 => 4 KB for each CPU
662 config PRINTK_SAFE_LOG_BUF_SHIFT
663 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
668 Select the size of an alternate printk per-CPU buffer where messages
669 printed from usafe contexts are temporary stored. One example would
670 be NMI messages, another one - printk recursion. The messages are
671 copied to the main log buffer in a safe context to avoid a deadlock.
672 The value defines the size as a power of 2.
674 Those messages are rare and limited. The largest one is when
675 a backtrace is printed. It usually fits into 4KB. Select
676 8KB if you want to be on the safe side.
679 17 => 128 KB for each CPU
680 16 => 64 KB for each CPU
681 15 => 32 KB for each CPU
682 14 => 16 KB for each CPU
683 13 => 8 KB for each CPU
684 12 => 4 KB for each CPU
687 # Architectures with an unreliable sched_clock() should select this:
689 config HAVE_UNSTABLE_SCHED_CLOCK
692 config GENERIC_SCHED_CLOCK
695 menu "Scheduler features"
698 bool "Enable utilization clamping for RT/FAIR tasks"
699 depends on CPU_FREQ_GOV_SCHEDUTIL
701 This feature enables the scheduler to track the clamped utilization
702 of each CPU based on RUNNABLE tasks scheduled on that CPU.
704 With this option, the user can specify the min and max CPU
705 utilization allowed for RUNNABLE tasks. The max utilization defines
706 the maximum frequency a task should use while the min utilization
707 defines the minimum frequency it should use.
709 Both min and max utilization clamp values are hints to the scheduler,
710 aiming at improving its frequency selection policy, but they do not
711 enforce or grant any specific bandwidth for tasks.
715 config UCLAMP_BUCKETS_COUNT
716 int "Number of supported utilization clamp buckets"
719 depends on UCLAMP_TASK
721 Defines the number of clamp buckets to use. The range of each bucket
722 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
723 number of clamp buckets the finer their granularity and the higher
724 the precision of clamping aggregation and tracking at run-time.
726 For example, with the minimum configuration value we will have 5
727 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
728 be refcounted in the [20..39]% bucket and will set the bucket clamp
729 effective value to 25%.
730 If a second 30% boosted task should be co-scheduled on the same CPU,
731 that task will be refcounted in the same bucket of the first task and
732 it will boost the bucket clamp effective value to 30%.
733 The clamp effective value of a bucket is reset to its nominal value
734 (20% in the example above) when there are no more tasks refcounted in
737 An additional boost/capping margin can be added to some tasks. In the
738 example above the 25% task will be boosted to 30% until it exits the
739 CPU. If that should be considered not acceptable on certain systems,
740 it's always possible to reduce the margin by increasing the number of
741 clamp buckets to trade off used memory for run-time tracking
744 If in doubt, use the default value.
749 # For architectures that want to enable the support for NUMA-affine scheduler
752 config ARCH_SUPPORTS_NUMA_BALANCING
756 # For architectures that prefer to flush all TLBs after a number of pages
757 # are unmapped instead of sending one IPI per page to flush. The architecture
758 # must provide guarantees on what happens if a clean TLB cache entry is
759 # written after the unmap. Details are in mm/rmap.c near the check for
760 # should_defer_flush. The architecture should also consider if the full flush
761 # and the refill costs are offset by the savings of sending fewer IPIs.
762 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
766 # For architectures that know their GCC __int128 support is sound
768 config ARCH_SUPPORTS_INT128
771 # For architectures that (ab)use NUMA to represent different memory regions
772 # all cpu-local but of different latencies, such as SuperH.
774 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
777 config NUMA_BALANCING
778 bool "Memory placement aware NUMA scheduler"
779 depends on ARCH_SUPPORTS_NUMA_BALANCING
780 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
781 depends on SMP && NUMA && MIGRATION
783 This option adds support for automatic NUMA aware memory/task placement.
784 The mechanism is quite primitive and is based on migrating memory when
785 it has references to the node the task is running on.
787 This system will be inactive on UMA systems.
789 config NUMA_BALANCING_DEFAULT_ENABLED
790 bool "Automatically enable NUMA aware memory/task placement"
792 depends on NUMA_BALANCING
794 If set, automatic NUMA balancing will be enabled if running on a NUMA
798 bool "Control Group support"
801 This option adds support for grouping sets of processes together, for
802 use with process control subsystems such as Cpusets, CFS, memory
803 controls or device isolation.
805 - Documentation/scheduler/sched-design-CFS.rst (CFS)
806 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
807 and resource control)
817 bool "Memory controller"
821 Provides control over the memory footprint of tasks in a cgroup.
824 bool "Swap controller"
825 depends on MEMCG && SWAP
827 Provides control over the swap space consumed by tasks in a cgroup.
829 config MEMCG_SWAP_ENABLED
830 bool "Swap controller enabled by default"
831 depends on MEMCG_SWAP
834 Memory Resource Controller Swap Extension comes with its price in
835 a bigger memory consumption. General purpose distribution kernels
836 which want to enable the feature but keep it disabled by default
837 and let the user enable it by swapaccount=1 boot command line
838 parameter should have this option unselected.
839 For those who want to have the feature enabled by default should
840 select this option (if, for some reason, they need to disable it
841 then swapaccount=0 does the trick).
845 depends on MEMCG && !SLOB
853 Generic block IO controller cgroup interface. This is the common
854 cgroup interface which should be used by various IO controlling
857 Currently, CFQ IO scheduler uses it to recognize task groups and
858 control disk bandwidth allocation (proportional time slice allocation)
859 to such task groups. It is also used by bio throttling logic in
860 block layer to implement upper limit in IO rates on a device.
862 This option only enables generic Block IO controller infrastructure.
863 One needs to also enable actual IO controlling logic/policy. For
864 enabling proportional weight division of disk bandwidth in CFQ, set
865 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
866 CONFIG_BLK_DEV_THROTTLING=y.
868 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
870 config CGROUP_WRITEBACK
872 depends on MEMCG && BLK_CGROUP
875 menuconfig CGROUP_SCHED
876 bool "CPU controller"
879 This feature lets CPU scheduler recognize task groups and control CPU
880 bandwidth allocation to such task groups. It uses cgroups to group
884 config FAIR_GROUP_SCHED
885 bool "Group scheduling for SCHED_OTHER"
886 depends on CGROUP_SCHED
890 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
891 depends on FAIR_GROUP_SCHED
894 This option allows users to define CPU bandwidth rates (limits) for
895 tasks running within the fair group scheduler. Groups with no limit
896 set are considered to be unconstrained and will run with no
898 See Documentation/scheduler/sched-bwc.rst for more information.
900 config RT_GROUP_SCHED
901 bool "Group scheduling for SCHED_RR/FIFO"
902 depends on CGROUP_SCHED
905 This feature lets you explicitly allocate real CPU bandwidth
906 to task groups. If enabled, it will also make it impossible to
907 schedule realtime tasks for non-root users until you allocate
908 realtime bandwidth for them.
909 See Documentation/scheduler/sched-rt-group.rst for more information.
913 config UCLAMP_TASK_GROUP
914 bool "Utilization clamping per group of tasks"
915 depends on CGROUP_SCHED
916 depends on UCLAMP_TASK
919 This feature enables the scheduler to track the clamped utilization
920 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
922 When this option is enabled, the user can specify a min and max
923 CPU bandwidth which is allowed for each single task in a group.
924 The max bandwidth allows to clamp the maximum frequency a task
925 can use, while the min bandwidth allows to define a minimum
926 frequency a task will always use.
928 When task group based utilization clamping is enabled, an eventually
929 specified task-specific clamp value is constrained by the cgroup
930 specified clamp value. Both minimum and maximum task clamping cannot
931 be bigger than the corresponding clamping defined at task group level.
936 bool "PIDs controller"
938 Provides enforcement of process number limits in the scope of a
939 cgroup. Any attempt to fork more processes than is allowed in the
940 cgroup will fail. PIDs are fundamentally a global resource because it
941 is fairly trivial to reach PID exhaustion before you reach even a
942 conservative kmemcg limit. As a result, it is possible to grind a
943 system to halt without being limited by other cgroup policies. The
944 PIDs controller is designed to stop this from happening.
946 It should be noted that organisational operations (such as attaching
947 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
948 since the PIDs limit only affects a process's ability to fork, not to
952 bool "RDMA controller"
954 Provides enforcement of RDMA resources defined by IB stack.
955 It is fairly easy for consumers to exhaust RDMA resources, which
956 can result into resource unavailability to other consumers.
957 RDMA controller is designed to stop this from happening.
958 Attaching processes with active RDMA resources to the cgroup
959 hierarchy is allowed even if can cross the hierarchy's limit.
961 config CGROUP_FREEZER
962 bool "Freezer controller"
964 Provides a way to freeze and unfreeze all tasks in a
967 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
968 controller includes important in-kernel memory consumers per default.
970 If you're using cgroup2, say N.
972 config CGROUP_HUGETLB
973 bool "HugeTLB controller"
974 depends on HUGETLB_PAGE
978 Provides a cgroup controller for HugeTLB pages.
979 When you enable this, you can put a per cgroup limit on HugeTLB usage.
980 The limit is enforced during page fault. Since HugeTLB doesn't
981 support page reclaim, enforcing the limit at page fault time implies
982 that, the application will get SIGBUS signal if it tries to access
983 HugeTLB pages beyond its limit. This requires the application to know
984 beforehand how much HugeTLB pages it would require for its use. The
985 control group is tracked in the third page lru pointer. This means
986 that we cannot use the controller with huge page less than 3 pages.
989 bool "Cpuset controller"
992 This option will let you create and manage CPUSETs which
993 allow dynamically partitioning a system into sets of CPUs and
994 Memory Nodes and assigning tasks to run only within those sets.
995 This is primarily useful on large SMP or NUMA systems.
999 config PROC_PID_CPUSET
1000 bool "Include legacy /proc/<pid>/cpuset file"
1004 config CGROUP_DEVICE
1005 bool "Device controller"
1007 Provides a cgroup controller implementing whitelists for
1008 devices which a process in the cgroup can mknod or open.
1010 config CGROUP_CPUACCT
1011 bool "Simple CPU accounting controller"
1013 Provides a simple controller for monitoring the
1014 total CPU consumed by the tasks in a cgroup.
1017 bool "Perf controller"
1018 depends on PERF_EVENTS
1020 This option extends the perf per-cpu mode to restrict monitoring
1021 to threads which belong to the cgroup specified and run on the
1027 bool "Support for eBPF programs attached to cgroups"
1028 depends on BPF_SYSCALL
1029 select SOCK_CGROUP_DATA
1031 Allow attaching eBPF programs to a cgroup using the bpf(2)
1032 syscall command BPF_PROG_ATTACH.
1034 In which context these programs are accessed depends on the type
1035 of attachment. For instance, programs that are attached using
1036 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1040 bool "Debug controller"
1042 depends on DEBUG_KERNEL
1044 This option enables a simple controller that exports
1045 debugging information about the cgroups framework. This
1046 controller is for control cgroup debugging only. Its
1047 interfaces are not stable.
1051 config SOCK_CGROUP_DATA
1057 menuconfig NAMESPACES
1058 bool "Namespaces support" if EXPERT
1059 depends on MULTIUSER
1062 Provides the way to make tasks work with different objects using
1063 the same id. For example same IPC id may refer to different objects
1064 or same user id or pid may refer to different tasks when used in
1065 different namespaces.
1070 bool "UTS namespace"
1073 In this namespace tasks see different info provided with the
1077 bool "IPC namespace"
1078 depends on (SYSVIPC || POSIX_MQUEUE)
1081 In this namespace tasks work with IPC ids which correspond to
1082 different IPC objects in different namespaces.
1085 bool "User namespace"
1088 This allows containers, i.e. vservers, to use user namespaces
1089 to provide different user info for different servers.
1091 When user namespaces are enabled in the kernel it is
1092 recommended that the MEMCG option also be enabled and that
1093 user-space use the memory control groups to limit the amount
1094 of memory a memory unprivileged users can use.
1099 bool "PID Namespaces"
1102 Support process id namespaces. This allows having multiple
1103 processes with the same pid as long as they are in different
1104 pid namespaces. This is a building block of containers.
1107 bool "Network namespace"
1111 Allow user space to create what appear to be multiple instances
1112 of the network stack.
1116 config CHECKPOINT_RESTORE
1117 bool "Checkpoint/restore support"
1118 select PROC_CHILDREN
1121 Enables additional kernel features in a sake of checkpoint/restore.
1122 In particular it adds auxiliary prctl codes to setup process text,
1123 data and heap segment sizes, and a few additional /proc filesystem
1126 If unsure, say N here.
1128 config SCHED_AUTOGROUP
1129 bool "Automatic process group scheduling"
1132 select FAIR_GROUP_SCHED
1134 This option optimizes the scheduler for common desktop workloads by
1135 automatically creating and populating task groups. This separation
1136 of workloads isolates aggressive CPU burners (like build jobs) from
1137 desktop applications. Task group autogeneration is currently based
1140 config SYSFS_DEPRECATED
1141 bool "Enable deprecated sysfs features to support old userspace tools"
1145 This option adds code that switches the layout of the "block" class
1146 devices, to not show up in /sys/class/block/, but only in
1149 This switch is only active when the sysfs.deprecated=1 boot option is
1150 passed or the SYSFS_DEPRECATED_V2 option is set.
1152 This option allows new kernels to run on old distributions and tools,
1153 which might get confused by /sys/class/block/. Since 2007/2008 all
1154 major distributions and tools handle this just fine.
1156 Recent distributions and userspace tools after 2009/2010 depend on
1157 the existence of /sys/class/block/, and will not work with this
1160 Only if you are using a new kernel on an old distribution, you might
1163 config SYSFS_DEPRECATED_V2
1164 bool "Enable deprecated sysfs features by default"
1167 depends on SYSFS_DEPRECATED
1169 Enable deprecated sysfs by default.
1171 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1174 Only if you are using a new kernel on an old distribution, you might
1175 need to say Y here. Even then, odds are you would not need it
1176 enabled, you can always pass the boot option if absolutely necessary.
1179 bool "Kernel->user space relay support (formerly relayfs)"
1182 This option enables support for relay interface support in
1183 certain file systems (such as debugfs).
1184 It is designed to provide an efficient mechanism for tools and
1185 facilities to relay large amounts of data from kernel space to
1190 config BLK_DEV_INITRD
1191 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1193 The initial RAM filesystem is a ramfs which is loaded by the
1194 boot loader (loadlin or lilo) and that is mounted as root
1195 before the normal boot procedure. It is typically used to
1196 load modules needed to mount the "real" root file system,
1197 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1199 If RAM disk support (BLK_DEV_RAM) is also included, this
1200 also enables initial RAM disk (initrd) support and adds
1201 15 Kbytes (more on some other architectures) to the kernel size.
1207 source "usr/Kconfig"
1212 prompt "Compiler optimization level"
1213 default CC_OPTIMIZE_FOR_PERFORMANCE
1215 config CC_OPTIMIZE_FOR_PERFORMANCE
1216 bool "Optimize for performance (-O2)"
1218 This is the default optimization level for the kernel, building
1219 with the "-O2" compiler flag for best performance and most
1220 helpful compile-time warnings.
1222 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1223 bool "Optimize more for performance (-O3)"
1226 Choosing this option will pass "-O3" to your compiler to optimize
1227 the kernel yet more for performance.
1229 config CC_OPTIMIZE_FOR_SIZE
1230 bool "Optimize for size (-Os)"
1232 Choosing this option will pass "-Os" to your compiler resulting
1233 in a smaller kernel.
1237 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1240 This requires that the arch annotates or otherwise protects
1241 its external entry points from being discarded. Linker scripts
1242 must also merge .text.*, .data.*, and .bss.* correctly into
1243 output sections. Care must be taken not to pull in unrelated
1244 sections (e.g., '.text.init'). Typically '.' in section names
1245 is used to distinguish them from label names / C identifiers.
1247 config LD_DEAD_CODE_DATA_ELIMINATION
1248 bool "Dead code and data elimination (EXPERIMENTAL)"
1249 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1251 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1252 depends on $(cc-option,-ffunction-sections -fdata-sections)
1253 depends on $(ld-option,--gc-sections)
1255 Enable this if you want to do dead code and data elimination with
1256 the linker by compiling with -ffunction-sections -fdata-sections,
1257 and linking with --gc-sections.
1259 This can reduce on disk and in-memory size of the kernel
1260 code and static data, particularly for small configs and
1261 on small systems. This has the possibility of introducing
1262 silently broken kernel if the required annotations are not
1263 present. This option is not well tested yet, so use at your
1272 config SYSCTL_EXCEPTION_TRACE
1275 Enable support for /proc/sys/debug/exception-trace.
1277 config SYSCTL_ARCH_UNALIGN_NO_WARN
1280 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1281 Allows arch to define/use @no_unaligned_warning to possibly warn
1282 about unaligned access emulation going on under the hood.
1284 config SYSCTL_ARCH_UNALIGN_ALLOW
1287 Enable support for /proc/sys/kernel/unaligned-trap
1288 Allows arches to define/use @unaligned_enabled to runtime toggle
1289 the unaligned access emulation.
1290 see arch/parisc/kernel/unaligned.c for reference
1292 config HAVE_PCSPKR_PLATFORM
1295 # interpreter that classic socket filters depend on
1300 bool "Configure standard kernel features (expert users)"
1301 # Unhide debug options, to make the on-by-default options visible
1304 This option allows certain base kernel options and settings
1305 to be disabled or tweaked. This is for specialized
1306 environments which can tolerate a "non-standard" kernel.
1307 Only use this if you really know what you are doing.
1310 bool "Enable 16-bit UID system calls" if EXPERT
1311 depends on HAVE_UID16 && MULTIUSER
1314 This enables the legacy 16-bit UID syscall wrappers.
1317 bool "Multiple users, groups and capabilities support" if EXPERT
1320 This option enables support for non-root users, groups and
1323 If you say N here, all processes will run with UID 0, GID 0, and all
1324 possible capabilities. Saying N here also compiles out support for
1325 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1328 If unsure, say Y here.
1330 config SGETMASK_SYSCALL
1331 bool "sgetmask/ssetmask syscalls support" if EXPERT
1332 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1334 sys_sgetmask and sys_ssetmask are obsolete system calls
1335 no longer supported in libc but still enabled by default in some
1338 If unsure, leave the default option here.
1340 config SYSFS_SYSCALL
1341 bool "Sysfs syscall support" if EXPERT
1344 sys_sysfs is an obsolete system call no longer supported in libc.
1345 Note that disabling this option is more secure but might break
1346 compatibility with some systems.
1348 If unsure say Y here.
1350 config SYSCTL_SYSCALL
1351 bool "Sysctl syscall support" if EXPERT
1352 depends on PROC_SYSCTL
1356 sys_sysctl uses binary paths that have been found challenging
1357 to properly maintain and use. The interface in /proc/sys
1358 using paths with ascii names is now the primary path to this
1361 Almost nothing using the binary sysctl interface so if you are
1362 trying to save some space it is probably safe to disable this,
1363 making your kernel marginally smaller.
1365 If unsure say N here.
1368 bool "open by fhandle syscalls" if EXPERT
1372 If you say Y here, a user level program will be able to map
1373 file names to handle and then later use the handle for
1374 different file system operations. This is useful in implementing
1375 userspace file servers, which now track files using handles instead
1376 of names. The handle would remain the same even if file names
1377 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1381 bool "Posix Clocks & timers" if EXPERT
1384 This includes native support for POSIX timers to the kernel.
1385 Some embedded systems have no use for them and therefore they
1386 can be configured out to reduce the size of the kernel image.
1388 When this option is disabled, the following syscalls won't be
1389 available: timer_create, timer_gettime: timer_getoverrun,
1390 timer_settime, timer_delete, clock_adjtime, getitimer,
1391 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1392 clock_getres and clock_nanosleep syscalls will be limited to
1393 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1399 bool "Enable support for printk" if EXPERT
1402 This option enables normal printk support. Removing it
1403 eliminates most of the message strings from the kernel image
1404 and makes the kernel more or less silent. As this makes it
1405 very difficult to diagnose system problems, saying N here is
1406 strongly discouraged.
1414 bool "BUG() support" if EXPERT
1417 Disabling this option eliminates support for BUG and WARN, reducing
1418 the size of your kernel image and potentially quietly ignoring
1419 numerous fatal conditions. You should only consider disabling this
1420 option for embedded systems with no facilities for reporting errors.
1426 bool "Enable ELF core dumps" if EXPERT
1428 Enable support for generating core dumps. Disabling saves about 4k.
1431 config PCSPKR_PLATFORM
1432 bool "Enable PC-Speaker support" if EXPERT
1433 depends on HAVE_PCSPKR_PLATFORM
1437 This option allows to disable the internal PC-Speaker
1438 support, saving some memory.
1442 bool "Enable full-sized data structures for core" if EXPERT
1444 Disabling this option reduces the size of miscellaneous core
1445 kernel data structures. This saves memory on small machines,
1446 but may reduce performance.
1449 bool "Enable futex support" if EXPERT
1453 Disabling this option will cause the kernel to be built without
1454 support for "fast userspace mutexes". The resulting kernel may not
1455 run glibc-based applications correctly.
1459 depends on FUTEX && RT_MUTEXES
1462 config HAVE_FUTEX_CMPXCHG
1466 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1467 is implemented and always working. This removes a couple of runtime
1471 bool "Enable eventpoll support" if EXPERT
1474 Disabling this option will cause the kernel to be built without
1475 support for epoll family of system calls.
1478 bool "Enable signalfd() system call" if EXPERT
1481 Enable the signalfd() system call that allows to receive signals
1482 on a file descriptor.
1487 bool "Enable timerfd() system call" if EXPERT
1490 Enable the timerfd() system call that allows to receive timer
1491 events on a file descriptor.
1496 bool "Enable eventfd() system call" if EXPERT
1499 Enable the eventfd() system call that allows to receive both
1500 kernel notification (ie. KAIO) or userspace notifications.
1505 bool "Use full shmem filesystem" if EXPERT
1509 The shmem is an internal filesystem used to manage shared memory.
1510 It is backed by swap and manages resource limits. It is also exported
1511 to userspace as tmpfs if TMPFS is enabled. Disabling this
1512 option replaces shmem and tmpfs with the much simpler ramfs code,
1513 which may be appropriate on small systems without swap.
1516 bool "Enable AIO support" if EXPERT
1519 This option enables POSIX asynchronous I/O which may by used
1520 by some high performance threaded applications. Disabling
1521 this option saves about 7k.
1524 bool "Enable IO uring support" if EXPERT
1528 This option enables support for the io_uring interface, enabling
1529 applications to submit and complete IO through submission and
1530 completion rings that are shared between the kernel and application.
1532 config ADVISE_SYSCALLS
1533 bool "Enable madvise/fadvise syscalls" if EXPERT
1536 This option enables the madvise and fadvise syscalls, used by
1537 applications to advise the kernel about their future memory or file
1538 usage, improving performance. If building an embedded system where no
1539 applications use these syscalls, you can disable this option to save
1543 bool "Enable membarrier() system call" if EXPERT
1546 Enable the membarrier() system call that allows issuing memory
1547 barriers across all running threads, which can be used to distribute
1548 the cost of user-space memory barriers asymmetrically by transforming
1549 pairs of memory barriers into pairs consisting of membarrier() and a
1555 bool "Load all symbols for debugging/ksymoops" if EXPERT
1558 Say Y here to let the kernel print out symbolic crash information and
1559 symbolic stack backtraces. This increases the size of the kernel
1560 somewhat, as all symbols have to be loaded into the kernel image.
1563 bool "Include all symbols in kallsyms"
1564 depends on DEBUG_KERNEL && KALLSYMS
1566 Normally kallsyms only contains the symbols of functions for nicer
1567 OOPS messages and backtraces (i.e., symbols from the text and inittext
1568 sections). This is sufficient for most cases. And only in very rare
1569 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1570 names of variables from the data sections, etc).
1572 This option makes sure that all symbols are loaded into the kernel
1573 image (i.e., symbols from all sections) in cost of increased kernel
1574 size (depending on the kernel configuration, it may be 300KiB or
1575 something like this).
1577 Say N unless you really need all symbols.
1579 config KALLSYMS_ABSOLUTE_PERCPU
1582 default X86_64 && SMP
1584 config KALLSYMS_BASE_RELATIVE
1589 Instead of emitting them as absolute values in the native word size,
1590 emit the symbol references in the kallsyms table as 32-bit entries,
1591 each containing a relative value in the range [base, base + U32_MAX]
1592 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1593 an absolute value in the range [0, S32_MAX] or a relative value in the
1594 range [base, base + S32_MAX], where base is the lowest relative symbol
1595 address encountered in the image.
1597 On 64-bit builds, this reduces the size of the address table by 50%,
1598 but more importantly, it results in entries whose values are build
1599 time constants, and no relocation pass is required at runtime to fix
1600 up the entries based on the runtime load address of the kernel.
1602 # end of the "standard kernel features (expert users)" menu
1604 # syscall, maps, verifier
1606 bool "Enable bpf() system call"
1611 Enable the bpf() system call that allows to manipulate eBPF
1612 programs and maps via file descriptors.
1614 config BPF_JIT_ALWAYS_ON
1615 bool "Permanently enable BPF JIT and remove BPF interpreter"
1616 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1618 Enables BPF JIT and removes BPF interpreter to avoid
1619 speculative execution of BPF instructions by the interpreter
1621 config BPF_UNPRIV_DEFAULT_OFF
1622 bool "Disable unprivileged BPF by default"
1623 depends on BPF_SYSCALL
1625 Disables unprivileged BPF by default by setting the corresponding
1626 /proc/sys/kernel/unprivileged_bpf_disabled knob to 2. An admin can
1627 still reenable it by setting it to 0 later on, or permanently
1628 disable it by setting it to 1 (from which no other transition to
1629 0 is possible anymore).
1632 bool "Enable userfaultfd() system call"
1635 Enable the userfaultfd() system call that allows to intercept and
1636 handle page faults in userland.
1638 config ARCH_HAS_MEMBARRIER_CALLBACKS
1641 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1645 bool "Enable rseq() system call" if EXPERT
1647 depends on HAVE_RSEQ
1650 Enable the restartable sequences system call. It provides a
1651 user-space cache for the current CPU number value, which
1652 speeds up getting the current CPU number from user-space,
1653 as well as an ABI to speed up user-space operations on
1660 bool "Enabled debugging of rseq() system call" if EXPERT
1661 depends on RSEQ && DEBUG_KERNEL
1663 Enable extra debugging checks for the rseq system call.
1668 bool "Embedded system"
1669 option allnoconfig_y
1672 This option should be enabled if compiling the kernel for
1673 an embedded system so certain expert options are available
1676 config HAVE_PERF_EVENTS
1679 See tools/perf/design.txt for details.
1681 config PERF_USE_VMALLOC
1684 See tools/perf/design.txt for details
1687 bool "PC/104 support" if EXPERT
1689 Expose PC/104 form factor device drivers and options available for
1690 selection and configuration. Enable this option if your target
1691 machine has a PC/104 bus.
1693 menu "Kernel Performance Events And Counters"
1696 bool "Kernel performance events and counters"
1697 default y if PROFILING
1698 depends on HAVE_PERF_EVENTS
1702 Enable kernel support for various performance events provided
1703 by software and hardware.
1705 Software events are supported either built-in or via the
1706 use of generic tracepoints.
1708 Most modern CPUs support performance events via performance
1709 counter registers. These registers count the number of certain
1710 types of hw events: such as instructions executed, cachemisses
1711 suffered, or branches mis-predicted - without slowing down the
1712 kernel or applications. These registers can also trigger interrupts
1713 when a threshold number of events have passed - and can thus be
1714 used to profile the code that runs on that CPU.
1716 The Linux Performance Event subsystem provides an abstraction of
1717 these software and hardware event capabilities, available via a
1718 system call and used by the "perf" utility in tools/perf/. It
1719 provides per task and per CPU counters, and it provides event
1720 capabilities on top of those.
1724 config DEBUG_PERF_USE_VMALLOC
1726 bool "Debug: use vmalloc to back perf mmap() buffers"
1727 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1728 select PERF_USE_VMALLOC
1730 Use vmalloc memory to back perf mmap() buffers.
1732 Mostly useful for debugging the vmalloc code on platforms
1733 that don't require it.
1739 config VM_EVENT_COUNTERS
1741 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1743 VM event counters are needed for event counts to be shown.
1744 This option allows the disabling of the VM event counters
1745 on EXPERT systems. /proc/vmstat will only show page counts
1746 if VM event counters are disabled.
1750 bool "Enable SLUB debugging support" if EXPERT
1751 depends on SLUB && SYSFS
1753 SLUB has extensive debug support features. Disabling these can
1754 result in significant savings in code size. This also disables
1755 SLUB sysfs support. /sys/slab will not exist and there will be
1756 no support for cache validation etc.
1758 config SLUB_MEMCG_SYSFS_ON
1760 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1761 depends on SLUB && SYSFS && MEMCG
1763 SLUB creates a directory under /sys/kernel/slab for each
1764 allocation cache to host info and debug files. If memory
1765 cgroup is enabled, each cache can have per memory cgroup
1766 caches. SLUB can create the same sysfs directories for these
1767 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1768 to a very high number of debug files being created. This is
1769 controlled by slub_memcg_sysfs boot parameter and this
1770 config option determines the parameter's default value.
1773 bool "Disable heap randomization"
1776 Randomizing heap placement makes heap exploits harder, but it
1777 also breaks ancient binaries (including anything libc5 based).
1778 This option changes the bootup default to heap randomization
1779 disabled, and can be overridden at runtime by setting
1780 /proc/sys/kernel/randomize_va_space to 2.
1782 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1785 prompt "Choose SLAB allocator"
1788 This option allows to select a slab allocator.
1792 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1794 The regular slab allocator that is established and known to work
1795 well in all environments. It organizes cache hot objects in
1796 per cpu and per node queues.
1799 bool "SLUB (Unqueued Allocator)"
1800 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1802 SLUB is a slab allocator that minimizes cache line usage
1803 instead of managing queues of cached objects (SLAB approach).
1804 Per cpu caching is realized using slabs of objects instead
1805 of queues of objects. SLUB can use memory efficiently
1806 and has enhanced diagnostics. SLUB is the default choice for
1811 bool "SLOB (Simple Allocator)"
1813 SLOB replaces the stock allocator with a drastically simpler
1814 allocator. SLOB is generally more space efficient but
1815 does not perform as well on large systems.
1819 config SLAB_MERGE_DEFAULT
1820 bool "Allow slab caches to be merged"
1823 For reduced kernel memory fragmentation, slab caches can be
1824 merged when they share the same size and other characteristics.
1825 This carries a risk of kernel heap overflows being able to
1826 overwrite objects from merged caches (and more easily control
1827 cache layout), which makes such heap attacks easier to exploit
1828 by attackers. By keeping caches unmerged, these kinds of exploits
1829 can usually only damage objects in the same cache. To disable
1830 merging at runtime, "slab_nomerge" can be passed on the kernel
1833 config SLAB_FREELIST_RANDOM
1835 depends on SLAB || SLUB
1836 bool "SLAB freelist randomization"
1838 Randomizes the freelist order used on creating new pages. This
1839 security feature reduces the predictability of the kernel slab
1840 allocator against heap overflows.
1842 config SLAB_FREELIST_HARDENED
1843 bool "Harden slab freelist metadata"
1846 Many kernel heap attacks try to target slab cache metadata and
1847 other infrastructure. This options makes minor performance
1848 sacrifices to harden the kernel slab allocator against common
1849 freelist exploit methods.
1851 config SHUFFLE_PAGE_ALLOCATOR
1852 bool "Page allocator randomization"
1853 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1855 Randomization of the page allocator improves the average
1856 utilization of a direct-mapped memory-side-cache. See section
1857 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1858 6.2a specification for an example of how a platform advertises
1859 the presence of a memory-side-cache. There are also incidental
1860 security benefits as it reduces the predictability of page
1861 allocations to compliment SLAB_FREELIST_RANDOM, but the
1862 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1863 10th order of pages is selected based on cache utilization
1866 While the randomization improves cache utilization it may
1867 negatively impact workloads on platforms without a cache. For
1868 this reason, by default, the randomization is enabled only
1869 after runtime detection of a direct-mapped memory-side-cache.
1870 Otherwise, the randomization may be force enabled with the
1871 'page_alloc.shuffle' kernel command line parameter.
1875 config SLUB_CPU_PARTIAL
1877 depends on SLUB && SMP
1878 bool "SLUB per cpu partial cache"
1880 Per cpu partial caches accelerate objects allocation and freeing
1881 that is local to a processor at the price of more indeterminism
1882 in the latency of the free. On overflow these caches will be cleared
1883 which requires the taking of locks that may cause latency spikes.
1884 Typically one would choose no for a realtime system.
1886 config MMAP_ALLOW_UNINITIALIZED
1887 bool "Allow mmapped anonymous memory to be uninitialized"
1888 depends on EXPERT && !MMU
1891 Normally, and according to the Linux spec, anonymous memory obtained
1892 from mmap() has its contents cleared before it is passed to
1893 userspace. Enabling this config option allows you to request that
1894 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1895 providing a huge performance boost. If this option is not enabled,
1896 then the flag will be ignored.
1898 This is taken advantage of by uClibc's malloc(), and also by
1899 ELF-FDPIC binfmt's brk and stack allocator.
1901 Because of the obvious security issues, this option should only be
1902 enabled on embedded devices where you control what is run in
1903 userspace. Since that isn't generally a problem on no-MMU systems,
1904 it is normally safe to say Y here.
1906 See Documentation/nommu-mmap.txt for more information.
1908 config SYSTEM_DATA_VERIFICATION
1910 select SYSTEM_TRUSTED_KEYRING
1914 select ASYMMETRIC_KEY_TYPE
1915 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1918 select X509_CERTIFICATE_PARSER
1919 select PKCS7_MESSAGE_PARSER
1921 Provide PKCS#7 message verification using the contents of the system
1922 trusted keyring to provide public keys. This then can be used for
1923 module verification, kexec image verification and firmware blob
1927 bool "Profiling support"
1929 Say Y here to enable the extended profiling support mechanisms used
1930 by profilers such as OProfile.
1933 # Place an empty function call at each tracepoint site. Can be
1934 # dynamically changed for a probe function.
1939 endmenu # General setup
1941 source "arch/Kconfig"
1948 default 0 if BASE_FULL
1949 default 1 if !BASE_FULL
1951 config MODULE_SIG_FORMAT
1953 select SYSTEM_DATA_VERIFICATION
1956 bool "Enable loadable module support"
1959 Kernel modules are small pieces of compiled code which can
1960 be inserted in the running kernel, rather than being
1961 permanently built into the kernel. You use the "modprobe"
1962 tool to add (and sometimes remove) them. If you say Y here,
1963 many parts of the kernel can be built as modules (by
1964 answering M instead of Y where indicated): this is most
1965 useful for infrequently used options which are not required
1966 for booting. For more information, see the man pages for
1967 modprobe, lsmod, modinfo, insmod and rmmod.
1969 If you say Y here, you will need to run "make
1970 modules_install" to put the modules under /lib/modules/
1971 where modprobe can find them (you may need to be root to do
1978 config MODULE_FORCE_LOAD
1979 bool "Forced module loading"
1982 Allow loading of modules without version information (ie. modprobe
1983 --force). Forced module loading sets the 'F' (forced) taint flag and
1984 is usually a really bad idea.
1986 config MODULE_UNLOAD
1987 bool "Module unloading"
1989 Without this option you will not be able to unload any
1990 modules (note that some modules may not be unloadable
1991 anyway), which makes your kernel smaller, faster
1992 and simpler. If unsure, say Y.
1994 config MODULE_FORCE_UNLOAD
1995 bool "Forced module unloading"
1996 depends on MODULE_UNLOAD
1998 This option allows you to force a module to unload, even if the
1999 kernel believes it is unsafe: the kernel will remove the module
2000 without waiting for anyone to stop using it (using the -f option to
2001 rmmod). This is mainly for kernel developers and desperate users.
2005 bool "Module versioning support"
2007 Usually, you have to use modules compiled with your kernel.
2008 Saying Y here makes it sometimes possible to use modules
2009 compiled for different kernels, by adding enough information
2010 to the modules to (hopefully) spot any changes which would
2011 make them incompatible with the kernel you are running. If
2014 config ASM_MODVERSIONS
2016 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2018 This enables module versioning for exported symbols also from
2019 assembly. This can be enabled only when the target architecture
2022 config MODULE_REL_CRCS
2024 depends on MODVERSIONS
2026 config MODULE_SRCVERSION_ALL
2027 bool "Source checksum for all modules"
2029 Modules which contain a MODULE_VERSION get an extra "srcversion"
2030 field inserted into their modinfo section, which contains a
2031 sum of the source files which made it. This helps maintainers
2032 see exactly which source was used to build a module (since
2033 others sometimes change the module source without updating
2034 the version). With this option, such a "srcversion" field
2035 will be created for all modules. If unsure, say N.
2038 bool "Module signature verification"
2039 select MODULE_SIG_FORMAT
2041 Check modules for valid signatures upon load: the signature
2042 is simply appended to the module. For more information see
2043 <file:Documentation/admin-guide/module-signing.rst>.
2045 Note that this option adds the OpenSSL development packages as a
2046 kernel build dependency so that the signing tool can use its crypto
2049 You should enable this option if you wish to use either
2050 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2051 another LSM - otherwise unsigned modules will be loadable regardless
2052 of the lockdown policy.
2054 !!!WARNING!!! If you enable this option, you MUST make sure that the
2055 module DOES NOT get stripped after being signed. This includes the
2056 debuginfo strip done by some packagers (such as rpmbuild) and
2057 inclusion into an initramfs that wants the module size reduced.
2059 config MODULE_SIG_FORCE
2060 bool "Require modules to be validly signed"
2061 depends on MODULE_SIG
2063 Reject unsigned modules or signed modules for which we don't have a
2064 key. Without this, such modules will simply taint the kernel.
2066 config MODULE_SIG_ALL
2067 bool "Automatically sign all modules"
2069 depends on MODULE_SIG
2071 Sign all modules during make modules_install. Without this option,
2072 modules must be signed manually, using the scripts/sign-file tool.
2074 comment "Do not forget to sign required modules with scripts/sign-file"
2075 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2078 prompt "Which hash algorithm should modules be signed with?"
2079 depends on MODULE_SIG
2081 This determines which sort of hashing algorithm will be used during
2082 signature generation. This algorithm _must_ be built into the kernel
2083 directly so that signature verification can take place. It is not
2084 possible to load a signed module containing the algorithm to check
2085 the signature on that module.
2087 config MODULE_SIG_SHA1
2088 bool "Sign modules with SHA-1"
2091 config MODULE_SIG_SHA224
2092 bool "Sign modules with SHA-224"
2093 select CRYPTO_SHA256
2095 config MODULE_SIG_SHA256
2096 bool "Sign modules with SHA-256"
2097 select CRYPTO_SHA256
2099 config MODULE_SIG_SHA384
2100 bool "Sign modules with SHA-384"
2101 select CRYPTO_SHA512
2103 config MODULE_SIG_SHA512
2104 bool "Sign modules with SHA-512"
2105 select CRYPTO_SHA512
2109 config MODULE_SIG_HASH
2111 depends on MODULE_SIG
2112 default "sha1" if MODULE_SIG_SHA1
2113 default "sha224" if MODULE_SIG_SHA224
2114 default "sha256" if MODULE_SIG_SHA256
2115 default "sha384" if MODULE_SIG_SHA384
2116 default "sha512" if MODULE_SIG_SHA512
2118 config MODULE_COMPRESS
2119 bool "Compress modules on installation"
2122 Compresses kernel modules when 'make modules_install' is run; gzip or
2123 xz depending on "Compression algorithm" below.
2125 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2127 Out-of-tree kernel modules installed using Kbuild will also be
2128 compressed upon installation.
2130 Note: for modules inside an initrd or initramfs, it's more efficient
2131 to compress the whole initrd or initramfs instead.
2133 Note: This is fully compatible with signed modules.
2138 prompt "Compression algorithm"
2139 depends on MODULE_COMPRESS
2140 default MODULE_COMPRESS_GZIP
2142 This determines which sort of compression will be used during
2143 'make modules_install'.
2145 GZIP (default) and XZ are supported.
2147 config MODULE_COMPRESS_GZIP
2150 config MODULE_COMPRESS_XZ
2155 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2156 bool "Allow loading of modules with missing namespace imports"
2158 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2159 a namespace. A module that makes use of a symbol exported with such a
2160 namespace is required to import the namespace via MODULE_IMPORT_NS().
2161 There is no technical reason to enforce correct namespace imports,
2162 but it creates consistency between symbols defining namespaces and
2163 users importing namespaces they make use of. This option relaxes this
2164 requirement and lifts the enforcement when loading a module.
2168 config UNUSED_SYMBOLS
2169 bool "Enable unused/obsolete exported symbols"
2172 Unused but exported symbols make the kernel needlessly bigger. For
2173 that reason most of these unused exports will soon be removed. This
2174 option is provided temporarily to provide a transition period in case
2175 some external kernel module needs one of these symbols anyway. If you
2176 encounter such a case in your module, consider if you are actually
2177 using the right API. (rationale: since nobody in the kernel is using
2178 this in a module, there is a pretty good chance it's actually the
2179 wrong interface to use). If you really need the symbol, please send a
2180 mail to the linux kernel mailing list mentioning the symbol and why
2181 you really need it, and what the merge plan to the mainline kernel for
2184 config TRIM_UNUSED_KSYMS
2185 bool "Trim unused exported kernel symbols"
2186 depends on !UNUSED_SYMBOLS
2188 The kernel and some modules make many symbols available for
2189 other modules to use via EXPORT_SYMBOL() and variants. Depending
2190 on the set of modules being selected in your kernel configuration,
2191 many of those exported symbols might never be used.
2193 This option allows for unused exported symbols to be dropped from
2194 the build. In turn, this provides the compiler more opportunities
2195 (especially when using LTO) for optimizing the code and reducing
2196 binary size. This might have some security advantages as well.
2198 If unsure, or if you need to build out-of-tree modules, say N.
2202 config MODULES_TREE_LOOKUP
2204 depends on PERF_EVENTS || TRACING
2206 config INIT_ALL_POSSIBLE
2209 Back when each arch used to define their own cpu_online_mask and
2210 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2211 with all 1s, and others with all 0s. When they were centralised,
2212 it was better to provide this option than to break all the archs
2213 and have several arch maintainers pursuing me down dark alleys.
2215 source "block/Kconfig"
2217 config PREEMPT_NOTIFIERS
2227 Build a simple ASN.1 grammar compiler that produces a bytecode output
2228 that can be interpreted by the ASN.1 stream decoder and used to
2229 inform it as to what tags are to be expected in a stream and what
2230 functions to call on what tags.
2232 source "kernel/Kconfig.locks"
2234 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2237 # It may be useful for an architecture to override the definitions of the
2238 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2239 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2240 # different calling convention for syscalls. They can also override the
2241 # macros for not-implemented syscalls in kernel/sys_ni.c and
2242 # kernel/time/posix-stubs.c. All these overrides need to be available in
2243 # <asm/syscall_wrapper.h>.
2244 config ARCH_HAS_SYSCALL_WRAPPER