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)
25 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
28 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
30 config CC_HAS_ASM_GOTO
31 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
33 config TOOLS_SUPPORT_RELR
34 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
36 config CC_HAS_ASM_INLINE
37 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
39 config CC_HAS_WARN_MAYBE_UNINITIALIZED
40 def_bool $(cc-option,-Wmaybe-uninitialized)
42 GCC >= 4.7 supports this option.
44 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
46 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
47 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
49 GCC's -Wmaybe-uninitialized is not reliable by definition.
50 Lots of false positive warnings are produced in some cases.
52 If this option is enabled, -Wno-maybe-uninitialzed is passed
53 to the compiler to suppress maybe-uninitialized warnings.
62 config BUILDTIME_EXTABLE_SORT
65 config THREAD_INFO_IN_TASK
68 Select this to move thread_info off the stack into task_struct. To
69 make this work, an arch will need to remove all thread_info fields
70 except flags and fix any runtime bugs.
72 One subtle change that will be needed is to use try_get_task_stack()
73 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
82 depends on BROKEN || !SMP
85 config INIT_ENV_ARG_LIMIT
90 Maximum of each of the number of arguments and environment
91 variables passed to init from the kernel command line.
94 bool "Compile also drivers which will not load"
98 Some drivers can be compiled on a different platform than they are
99 intended to be run on. Despite they cannot be loaded there (or even
100 when they load they cannot be used due to missing HW support),
101 developers still, opposing to distributors, might want to build such
102 drivers to compile-test them.
104 If you are a developer and want to build everything available, say Y
105 here. If you are a user/distributor, say N here to exclude useless
106 drivers to be distributed.
109 bool "Compile test headers that should be standalone compilable"
111 Compile test headers listed in header-test-y target to ensure they are
112 self-contained, i.e. compilable as standalone units.
114 If you are a developer or tester and want to ensure the requested
115 headers are self-contained, say Y here. Otherwise, choose N.
117 config KERNEL_HEADER_TEST
118 bool "Compile test kernel headers"
119 depends on HEADER_TEST
121 Headers in include/ are used to build external moduls.
122 Compile test them to ensure they are self-contained, i.e.
123 compilable as standalone units.
125 If you are a developer or tester and want to ensure the headers
126 in include/ are self-contained, say Y here. Otherwise, choose N.
128 config UAPI_HEADER_TEST
129 bool "Compile test UAPI headers"
130 depends on HEADER_TEST && HEADERS_INSTALL && CC_CAN_LINK
132 Compile test headers exported to user-space to ensure they are
133 self-contained, i.e. compilable as standalone units.
135 If you are a developer or tester and want to ensure the exported
136 headers are self-contained, say Y here. Otherwise, choose N.
139 string "Local version - append to kernel release"
141 Append an extra string to the end of your kernel version.
142 This will show up when you type uname, for example.
143 The string you set here will be appended after the contents of
144 any files with a filename matching localversion* in your
145 object and source tree, in that order. Your total string can
146 be a maximum of 64 characters.
148 config LOCALVERSION_AUTO
149 bool "Automatically append version information to the version string"
151 depends on !COMPILE_TEST
153 This will try to automatically determine if the current tree is a
154 release tree by looking for git tags that belong to the current
155 top of tree revision.
157 A string of the format -gxxxxxxxx will be added to the localversion
158 if a git-based tree is found. The string generated by this will be
159 appended after any matching localversion* files, and after the value
160 set in CONFIG_LOCALVERSION.
162 (The actual string used here is the first eight characters produced
163 by running the command:
165 $ git rev-parse --verify HEAD
167 which is done within the script "scripts/setlocalversion".)
170 string "Build ID Salt"
173 The build ID is used to link binaries and their debug info. Setting
174 this option will use the value in the calculation of the build id.
175 This is mostly useful for distributions which want to ensure the
176 build is unique between builds. It's safe to leave the default.
178 config HAVE_KERNEL_GZIP
181 config HAVE_KERNEL_BZIP2
184 config HAVE_KERNEL_LZMA
187 config HAVE_KERNEL_XZ
190 config HAVE_KERNEL_LZO
193 config HAVE_KERNEL_LZ4
196 config HAVE_KERNEL_UNCOMPRESSED
200 prompt "Kernel compression mode"
202 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
204 The linux kernel is a kind of self-extracting executable.
205 Several compression algorithms are available, which differ
206 in efficiency, compression and decompression speed.
207 Compression speed is only relevant when building a kernel.
208 Decompression speed is relevant at each boot.
210 If you have any problems with bzip2 or lzma compressed
211 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
212 version of this functionality (bzip2 only), for 2.4, was
213 supplied by Christian Ludwig)
215 High compression options are mostly useful for users, who
216 are low on disk space (embedded systems), but for whom ram
219 If in doubt, select 'gzip'
223 depends on HAVE_KERNEL_GZIP
225 The old and tried gzip compression. It provides a good balance
226 between compression ratio and decompression speed.
230 depends on HAVE_KERNEL_BZIP2
232 Its compression ratio and speed is intermediate.
233 Decompression speed is slowest among the choices. The kernel
234 size is about 10% smaller with bzip2, in comparison to gzip.
235 Bzip2 uses a large amount of memory. For modern kernels you
236 will need at least 8MB RAM or more for booting.
240 depends on HAVE_KERNEL_LZMA
242 This compression algorithm's ratio is best. Decompression speed
243 is between gzip and bzip2. Compression is slowest.
244 The kernel size is about 33% smaller with LZMA in comparison to gzip.
248 depends on HAVE_KERNEL_XZ
250 XZ uses the LZMA2 algorithm and instruction set specific
251 BCJ filters which can improve compression ratio of executable
252 code. The size of the kernel is about 30% smaller with XZ in
253 comparison to gzip. On architectures for which there is a BCJ
254 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
255 will create a few percent smaller kernel than plain LZMA.
257 The speed is about the same as with LZMA: The decompression
258 speed of XZ is better than that of bzip2 but worse than gzip
259 and LZO. Compression is slow.
263 depends on HAVE_KERNEL_LZO
265 Its compression ratio is the poorest among the choices. The kernel
266 size is about 10% bigger than gzip; however its speed
267 (both compression and decompression) is the fastest.
271 depends on HAVE_KERNEL_LZ4
273 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
274 A preliminary version of LZ4 de/compression tool is available at
275 <https://code.google.com/p/lz4/>.
277 Its compression ratio is worse than LZO. The size of the kernel
278 is about 8% bigger than LZO. But the decompression speed is
281 config KERNEL_UNCOMPRESSED
283 depends on HAVE_KERNEL_UNCOMPRESSED
285 Produce uncompressed kernel image. This option is usually not what
286 you want. It is useful for debugging the kernel in slow simulation
287 environments, where decompressing and moving the kernel is awfully
288 slow. This option allows early boot code to skip the decompressor
289 and jump right at uncompressed kernel image.
293 config DEFAULT_HOSTNAME
294 string "Default hostname"
297 This option determines the default system hostname before userspace
298 calls sethostname(2). The kernel traditionally uses "(none)" here,
299 but you may wish to use a different default here to make a minimal
300 system more usable with less configuration.
303 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
304 # add proper SWAP support to them, in which case this can be remove.
310 bool "Support for paging of anonymous memory (swap)"
311 depends on MMU && BLOCK && !ARCH_NO_SWAP
314 This option allows you to choose whether you want to have support
315 for so called swap devices or swap files in your kernel that are
316 used to provide more virtual memory than the actual RAM present
317 in your computer. If unsure say Y.
322 Inter Process Communication is a suite of library functions and
323 system calls which let processes (running programs) synchronize and
324 exchange information. It is generally considered to be a good thing,
325 and some programs won't run unless you say Y here. In particular, if
326 you want to run the DOS emulator dosemu under Linux (read the
327 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
328 you'll need to say Y here.
330 You can find documentation about IPC with "info ipc" and also in
331 section 6.4 of the Linux Programmer's Guide, available from
332 <http://www.tldp.org/guides.html>.
334 config SYSVIPC_SYSCTL
341 bool "POSIX Message Queues"
344 POSIX variant of message queues is a part of IPC. In POSIX message
345 queues every message has a priority which decides about succession
346 of receiving it by a process. If you want to compile and run
347 programs written e.g. for Solaris with use of its POSIX message
348 queues (functions mq_*) say Y here.
350 POSIX message queues are visible as a filesystem called 'mqueue'
351 and can be mounted somewhere if you want to do filesystem
352 operations on message queues.
356 config POSIX_MQUEUE_SYSCTL
358 depends on POSIX_MQUEUE
362 config CROSS_MEMORY_ATTACH
363 bool "Enable process_vm_readv/writev syscalls"
367 Enabling this option adds the system calls process_vm_readv and
368 process_vm_writev which allow a process with the correct privileges
369 to directly read from or write to another process' address space.
370 See the man page for more details.
373 bool "uselib syscall"
374 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
376 This option enables the uselib syscall, a system call used in the
377 dynamic linker from libc5 and earlier. glibc does not use this
378 system call. If you intend to run programs built on libc5 or
379 earlier, you may need to enable this syscall. Current systems
380 running glibc can safely disable this.
383 bool "Auditing support"
386 Enable auditing infrastructure that can be used with another
387 kernel subsystem, such as SELinux (which requires this for
388 logging of avc messages output). System call auditing is included
389 on architectures which support it.
391 config HAVE_ARCH_AUDITSYSCALL
396 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
399 source "kernel/irq/Kconfig"
400 source "kernel/time/Kconfig"
401 source "kernel/Kconfig.preempt"
403 menu "CPU/Task time and stats accounting"
405 config VIRT_CPU_ACCOUNTING
409 prompt "Cputime accounting"
410 default TICK_CPU_ACCOUNTING if !PPC64
411 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
413 # Kind of a stub config for the pure tick based cputime accounting
414 config TICK_CPU_ACCOUNTING
415 bool "Simple tick based cputime accounting"
416 depends on !S390 && !NO_HZ_FULL
418 This is the basic tick based cputime accounting that maintains
419 statistics about user, system and idle time spent on per jiffies
424 config VIRT_CPU_ACCOUNTING_NATIVE
425 bool "Deterministic task and CPU time accounting"
426 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
427 select VIRT_CPU_ACCOUNTING
429 Select this option to enable more accurate task and CPU time
430 accounting. This is done by reading a CPU counter on each
431 kernel entry and exit and on transitions within the kernel
432 between system, softirq and hardirq state, so there is a
433 small performance impact. In the case of s390 or IBM POWER > 5,
434 this also enables accounting of stolen time on logically-partitioned
437 config VIRT_CPU_ACCOUNTING_GEN
438 bool "Full dynticks CPU time accounting"
439 depends on HAVE_CONTEXT_TRACKING
440 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
441 depends on GENERIC_CLOCKEVENTS
442 select VIRT_CPU_ACCOUNTING
443 select CONTEXT_TRACKING
445 Select this option to enable task and CPU time accounting on full
446 dynticks systems. This accounting is implemented by watching every
447 kernel-user boundaries using the context tracking subsystem.
448 The accounting is thus performed at the expense of some significant
451 For now this is only useful if you are working on the full
452 dynticks subsystem development.
458 config IRQ_TIME_ACCOUNTING
459 bool "Fine granularity task level IRQ time accounting"
460 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
462 Select this option to enable fine granularity task irq time
463 accounting. This is done by reading a timestamp on each
464 transitions between softirq and hardirq state, so there can be a
465 small performance impact.
467 If in doubt, say N here.
469 config HAVE_SCHED_AVG_IRQ
471 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
474 config BSD_PROCESS_ACCT
475 bool "BSD Process Accounting"
478 If you say Y here, a user level program will be able to instruct the
479 kernel (via a special system call) to write process accounting
480 information to a file: whenever a process exits, information about
481 that process will be appended to the file by the kernel. The
482 information includes things such as creation time, owning user,
483 command name, memory usage, controlling terminal etc. (the complete
484 list is in the struct acct in <file:include/linux/acct.h>). It is
485 up to the user level program to do useful things with this
486 information. This is generally a good idea, so say Y.
488 config BSD_PROCESS_ACCT_V3
489 bool "BSD Process Accounting version 3 file format"
490 depends on BSD_PROCESS_ACCT
493 If you say Y here, the process accounting information is written
494 in a new file format that also logs the process IDs of each
495 process and its parent. Note that this file format is incompatible
496 with previous v0/v1/v2 file formats, so you will need updated tools
497 for processing it. A preliminary version of these tools is available
498 at <http://www.gnu.org/software/acct/>.
501 bool "Export task/process statistics through netlink"
506 Export selected statistics for tasks/processes through the
507 generic netlink interface. Unlike BSD process accounting, the
508 statistics are available during the lifetime of tasks/processes as
509 responses to commands. Like BSD accounting, they are sent to user
514 config TASK_DELAY_ACCT
515 bool "Enable per-task delay accounting"
519 Collect information on time spent by a task waiting for system
520 resources like cpu, synchronous block I/O completion and swapping
521 in pages. Such statistics can help in setting a task's priorities
522 relative to other tasks for cpu, io, rss limits etc.
527 bool "Enable extended accounting over taskstats"
530 Collect extended task accounting data and send the data
531 to userland for processing over the taskstats interface.
535 config TASK_IO_ACCOUNTING
536 bool "Enable per-task storage I/O accounting"
537 depends on TASK_XACCT
539 Collect information on the number of bytes of storage I/O which this
545 bool "Pressure stall information tracking"
547 Collect metrics that indicate how overcommitted the CPU, memory,
548 and IO capacity are in the system.
550 If you say Y here, the kernel will create /proc/pressure/ with the
551 pressure statistics files cpu, memory, and io. These will indicate
552 the share of walltime in which some or all tasks in the system are
553 delayed due to contention of the respective resource.
555 In kernels with cgroup support, cgroups (cgroup2 only) will
556 have cpu.pressure, memory.pressure, and io.pressure files,
557 which aggregate pressure stalls for the grouped tasks only.
559 For more details see Documentation/accounting/psi.rst.
563 config PSI_DEFAULT_DISABLED
564 bool "Require boot parameter to enable pressure stall information tracking"
568 If set, pressure stall information tracking will be disabled
569 per default but can be enabled through passing psi=1 on the
570 kernel commandline during boot.
572 This feature adds some code to the task wakeup and sleep
573 paths of the scheduler. The overhead is too low to affect
574 common scheduling-intense workloads in practice (such as
575 webservers, memcache), but it does show up in artificial
576 scheduler stress tests, such as hackbench.
578 If you are paranoid and not sure what the kernel will be
583 endmenu # "CPU/Task time and stats accounting"
587 depends on SMP || COMPILE_TEST
590 Make sure that CPUs running critical tasks are not disturbed by
591 any source of "noise" such as unbound workqueues, timers, kthreads...
592 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
593 the "isolcpus=" boot parameter.
597 source "kernel/rcu/Kconfig"
604 tristate "Kernel .config support"
606 This option enables the complete Linux kernel ".config" file
607 contents to be saved in the kernel. It provides documentation
608 of which kernel options are used in a running kernel or in an
609 on-disk kernel. This information can be extracted from the kernel
610 image file with the script scripts/extract-ikconfig and used as
611 input to rebuild the current kernel or to build another kernel.
612 It can also be extracted from a running kernel by reading
613 /proc/config.gz if enabled (below).
616 bool "Enable access to .config through /proc/config.gz"
617 depends on IKCONFIG && PROC_FS
619 This option enables access to the kernel configuration file
620 through /proc/config.gz.
623 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
626 This option enables access to the in-kernel headers that are generated during
627 the build process. These can be used to build eBPF tracing programs,
628 or similar programs. If you build the headers as a module, a module called
629 kheaders.ko is built which can be loaded on-demand to get access to headers.
632 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
637 Select the minimal kernel log buffer size as a power of 2.
638 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
639 parameter, see below. Any higher size also might be forced
640 by "log_buf_len" boot parameter.
650 config LOG_CPU_MAX_BUF_SHIFT
651 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
654 default 12 if !BASE_SMALL
655 default 0 if BASE_SMALL
658 This option allows to increase the default ring buffer size
659 according to the number of CPUs. The value defines the contribution
660 of each CPU as a power of 2. The used space is typically only few
661 lines however it might be much more when problems are reported,
664 The increased size means that a new buffer has to be allocated and
665 the original static one is unused. It makes sense only on systems
666 with more CPUs. Therefore this value is used only when the sum of
667 contributions is greater than the half of the default kernel ring
668 buffer as defined by LOG_BUF_SHIFT. The default values are set
669 so that more than 64 CPUs are needed to trigger the allocation.
671 Also this option is ignored when "log_buf_len" kernel parameter is
672 used as it forces an exact (power of two) size of the ring buffer.
674 The number of possible CPUs is used for this computation ignoring
675 hotplugging making the computation optimal for the worst case
676 scenario while allowing a simple algorithm to be used from bootup.
678 Examples shift values and their meaning:
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
686 config PRINTK_SAFE_LOG_BUF_SHIFT
687 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
692 Select the size of an alternate printk per-CPU buffer where messages
693 printed from usafe contexts are temporary stored. One example would
694 be NMI messages, another one - printk recursion. The messages are
695 copied to the main log buffer in a safe context to avoid a deadlock.
696 The value defines the size as a power of 2.
698 Those messages are rare and limited. The largest one is when
699 a backtrace is printed. It usually fits into 4KB. Select
700 8KB if you want to be on the safe side.
703 17 => 128 KB for each CPU
704 16 => 64 KB for each CPU
705 15 => 32 KB for each CPU
706 14 => 16 KB for each CPU
707 13 => 8 KB for each CPU
708 12 => 4 KB for each CPU
711 # Architectures with an unreliable sched_clock() should select this:
713 config HAVE_UNSTABLE_SCHED_CLOCK
716 config GENERIC_SCHED_CLOCK
719 menu "Scheduler features"
722 bool "Enable utilization clamping for RT/FAIR tasks"
723 depends on CPU_FREQ_GOV_SCHEDUTIL
725 This feature enables the scheduler to track the clamped utilization
726 of each CPU based on RUNNABLE tasks scheduled on that CPU.
728 With this option, the user can specify the min and max CPU
729 utilization allowed for RUNNABLE tasks. The max utilization defines
730 the maximum frequency a task should use while the min utilization
731 defines the minimum frequency it should use.
733 Both min and max utilization clamp values are hints to the scheduler,
734 aiming at improving its frequency selection policy, but they do not
735 enforce or grant any specific bandwidth for tasks.
739 config UCLAMP_BUCKETS_COUNT
740 int "Number of supported utilization clamp buckets"
743 depends on UCLAMP_TASK
745 Defines the number of clamp buckets to use. The range of each bucket
746 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
747 number of clamp buckets the finer their granularity and the higher
748 the precision of clamping aggregation and tracking at run-time.
750 For example, with the minimum configuration value we will have 5
751 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
752 be refcounted in the [20..39]% bucket and will set the bucket clamp
753 effective value to 25%.
754 If a second 30% boosted task should be co-scheduled on the same CPU,
755 that task will be refcounted in the same bucket of the first task and
756 it will boost the bucket clamp effective value to 30%.
757 The clamp effective value of a bucket is reset to its nominal value
758 (20% in the example above) when there are no more tasks refcounted in
761 An additional boost/capping margin can be added to some tasks. In the
762 example above the 25% task will be boosted to 30% until it exits the
763 CPU. If that should be considered not acceptable on certain systems,
764 it's always possible to reduce the margin by increasing the number of
765 clamp buckets to trade off used memory for run-time tracking
768 If in doubt, use the default value.
773 # For architectures that want to enable the support for NUMA-affine scheduler
776 config ARCH_SUPPORTS_NUMA_BALANCING
780 # For architectures that prefer to flush all TLBs after a number of pages
781 # are unmapped instead of sending one IPI per page to flush. The architecture
782 # must provide guarantees on what happens if a clean TLB cache entry is
783 # written after the unmap. Details are in mm/rmap.c near the check for
784 # should_defer_flush. The architecture should also consider if the full flush
785 # and the refill costs are offset by the savings of sending fewer IPIs.
786 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
790 # For architectures that know their GCC __int128 support is sound
792 config ARCH_SUPPORTS_INT128
795 # For architectures that (ab)use NUMA to represent different memory regions
796 # all cpu-local but of different latencies, such as SuperH.
798 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
801 config NUMA_BALANCING
802 bool "Memory placement aware NUMA scheduler"
803 depends on ARCH_SUPPORTS_NUMA_BALANCING
804 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
805 depends on SMP && NUMA && MIGRATION
807 This option adds support for automatic NUMA aware memory/task placement.
808 The mechanism is quite primitive and is based on migrating memory when
809 it has references to the node the task is running on.
811 This system will be inactive on UMA systems.
813 config NUMA_BALANCING_DEFAULT_ENABLED
814 bool "Automatically enable NUMA aware memory/task placement"
816 depends on NUMA_BALANCING
818 If set, automatic NUMA balancing will be enabled if running on a NUMA
822 bool "Control Group support"
825 This option adds support for grouping sets of processes together, for
826 use with process control subsystems such as Cpusets, CFS, memory
827 controls or device isolation.
829 - Documentation/scheduler/sched-design-CFS.rst (CFS)
830 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
831 and resource control)
841 bool "Memory controller"
845 Provides control over the memory footprint of tasks in a cgroup.
848 bool "Swap controller"
849 depends on MEMCG && SWAP
851 Provides control over the swap space consumed by tasks in a cgroup.
853 config MEMCG_SWAP_ENABLED
854 bool "Swap controller enabled by default"
855 depends on MEMCG_SWAP
858 Memory Resource Controller Swap Extension comes with its price in
859 a bigger memory consumption. General purpose distribution kernels
860 which want to enable the feature but keep it disabled by default
861 and let the user enable it by swapaccount=1 boot command line
862 parameter should have this option unselected.
863 For those who want to have the feature enabled by default should
864 select this option (if, for some reason, they need to disable it
865 then swapaccount=0 does the trick).
869 depends on MEMCG && !SLOB
877 Generic block IO controller cgroup interface. This is the common
878 cgroup interface which should be used by various IO controlling
881 Currently, CFQ IO scheduler uses it to recognize task groups and
882 control disk bandwidth allocation (proportional time slice allocation)
883 to such task groups. It is also used by bio throttling logic in
884 block layer to implement upper limit in IO rates on a device.
886 This option only enables generic Block IO controller infrastructure.
887 One needs to also enable actual IO controlling logic/policy. For
888 enabling proportional weight division of disk bandwidth in CFQ, set
889 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
890 CONFIG_BLK_DEV_THROTTLING=y.
892 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
894 config CGROUP_WRITEBACK
896 depends on MEMCG && BLK_CGROUP
899 menuconfig CGROUP_SCHED
900 bool "CPU controller"
903 This feature lets CPU scheduler recognize task groups and control CPU
904 bandwidth allocation to such task groups. It uses cgroups to group
908 config FAIR_GROUP_SCHED
909 bool "Group scheduling for SCHED_OTHER"
910 depends on CGROUP_SCHED
914 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
915 depends on FAIR_GROUP_SCHED
918 This option allows users to define CPU bandwidth rates (limits) for
919 tasks running within the fair group scheduler. Groups with no limit
920 set are considered to be unconstrained and will run with no
922 See Documentation/scheduler/sched-bwc.rst for more information.
924 config RT_GROUP_SCHED
925 bool "Group scheduling for SCHED_RR/FIFO"
926 depends on CGROUP_SCHED
929 This feature lets you explicitly allocate real CPU bandwidth
930 to task groups. If enabled, it will also make it impossible to
931 schedule realtime tasks for non-root users until you allocate
932 realtime bandwidth for them.
933 See Documentation/scheduler/sched-rt-group.rst for more information.
937 config UCLAMP_TASK_GROUP
938 bool "Utilization clamping per group of tasks"
939 depends on CGROUP_SCHED
940 depends on UCLAMP_TASK
943 This feature enables the scheduler to track the clamped utilization
944 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
946 When this option is enabled, the user can specify a min and max
947 CPU bandwidth which is allowed for each single task in a group.
948 The max bandwidth allows to clamp the maximum frequency a task
949 can use, while the min bandwidth allows to define a minimum
950 frequency a task will always use.
952 When task group based utilization clamping is enabled, an eventually
953 specified task-specific clamp value is constrained by the cgroup
954 specified clamp value. Both minimum and maximum task clamping cannot
955 be bigger than the corresponding clamping defined at task group level.
960 bool "PIDs controller"
962 Provides enforcement of process number limits in the scope of a
963 cgroup. Any attempt to fork more processes than is allowed in the
964 cgroup will fail. PIDs are fundamentally a global resource because it
965 is fairly trivial to reach PID exhaustion before you reach even a
966 conservative kmemcg limit. As a result, it is possible to grind a
967 system to halt without being limited by other cgroup policies. The
968 PIDs controller is designed to stop this from happening.
970 It should be noted that organisational operations (such as attaching
971 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
972 since the PIDs limit only affects a process's ability to fork, not to
976 bool "RDMA controller"
978 Provides enforcement of RDMA resources defined by IB stack.
979 It is fairly easy for consumers to exhaust RDMA resources, which
980 can result into resource unavailability to other consumers.
981 RDMA controller is designed to stop this from happening.
982 Attaching processes with active RDMA resources to the cgroup
983 hierarchy is allowed even if can cross the hierarchy's limit.
985 config CGROUP_FREEZER
986 bool "Freezer controller"
988 Provides a way to freeze and unfreeze all tasks in a
991 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
992 controller includes important in-kernel memory consumers per default.
994 If you're using cgroup2, say N.
996 config CGROUP_HUGETLB
997 bool "HugeTLB controller"
998 depends on HUGETLB_PAGE
1002 Provides a cgroup controller for HugeTLB pages.
1003 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1004 The limit is enforced during page fault. Since HugeTLB doesn't
1005 support page reclaim, enforcing the limit at page fault time implies
1006 that, the application will get SIGBUS signal if it tries to access
1007 HugeTLB pages beyond its limit. This requires the application to know
1008 beforehand how much HugeTLB pages it would require for its use. The
1009 control group is tracked in the third page lru pointer. This means
1010 that we cannot use the controller with huge page less than 3 pages.
1013 bool "Cpuset controller"
1016 This option will let you create and manage CPUSETs which
1017 allow dynamically partitioning a system into sets of CPUs and
1018 Memory Nodes and assigning tasks to run only within those sets.
1019 This is primarily useful on large SMP or NUMA systems.
1023 config PROC_PID_CPUSET
1024 bool "Include legacy /proc/<pid>/cpuset file"
1028 config CGROUP_DEVICE
1029 bool "Device controller"
1031 Provides a cgroup controller implementing whitelists for
1032 devices which a process in the cgroup can mknod or open.
1034 config CGROUP_CPUACCT
1035 bool "Simple CPU accounting controller"
1037 Provides a simple controller for monitoring the
1038 total CPU consumed by the tasks in a cgroup.
1041 bool "Perf controller"
1042 depends on PERF_EVENTS
1044 This option extends the perf per-cpu mode to restrict monitoring
1045 to threads which belong to the cgroup specified and run on the
1051 bool "Support for eBPF programs attached to cgroups"
1052 depends on BPF_SYSCALL
1053 select SOCK_CGROUP_DATA
1055 Allow attaching eBPF programs to a cgroup using the bpf(2)
1056 syscall command BPF_PROG_ATTACH.
1058 In which context these programs are accessed depends on the type
1059 of attachment. For instance, programs that are attached using
1060 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1064 bool "Debug controller"
1066 depends on DEBUG_KERNEL
1068 This option enables a simple controller that exports
1069 debugging information about the cgroups framework. This
1070 controller is for control cgroup debugging only. Its
1071 interfaces are not stable.
1075 config SOCK_CGROUP_DATA
1081 menuconfig NAMESPACES
1082 bool "Namespaces support" if EXPERT
1083 depends on MULTIUSER
1086 Provides the way to make tasks work with different objects using
1087 the same id. For example same IPC id may refer to different objects
1088 or same user id or pid may refer to different tasks when used in
1089 different namespaces.
1094 bool "UTS namespace"
1097 In this namespace tasks see different info provided with the
1101 bool "IPC namespace"
1102 depends on (SYSVIPC || POSIX_MQUEUE)
1105 In this namespace tasks work with IPC ids which correspond to
1106 different IPC objects in different namespaces.
1109 bool "User namespace"
1112 This allows containers, i.e. vservers, to use user namespaces
1113 to provide different user info for different servers.
1115 When user namespaces are enabled in the kernel it is
1116 recommended that the MEMCG option also be enabled and that
1117 user-space use the memory control groups to limit the amount
1118 of memory a memory unprivileged users can use.
1123 bool "PID Namespaces"
1126 Support process id namespaces. This allows having multiple
1127 processes with the same pid as long as they are in different
1128 pid namespaces. This is a building block of containers.
1131 bool "Network namespace"
1135 Allow user space to create what appear to be multiple instances
1136 of the network stack.
1140 config CHECKPOINT_RESTORE
1141 bool "Checkpoint/restore support"
1142 select PROC_CHILDREN
1145 Enables additional kernel features in a sake of checkpoint/restore.
1146 In particular it adds auxiliary prctl codes to setup process text,
1147 data and heap segment sizes, and a few additional /proc filesystem
1150 If unsure, say N here.
1152 config SCHED_AUTOGROUP
1153 bool "Automatic process group scheduling"
1156 select FAIR_GROUP_SCHED
1158 This option optimizes the scheduler for common desktop workloads by
1159 automatically creating and populating task groups. This separation
1160 of workloads isolates aggressive CPU burners (like build jobs) from
1161 desktop applications. Task group autogeneration is currently based
1164 config SYSFS_DEPRECATED
1165 bool "Enable deprecated sysfs features to support old userspace tools"
1169 This option adds code that switches the layout of the "block" class
1170 devices, to not show up in /sys/class/block/, but only in
1173 This switch is only active when the sysfs.deprecated=1 boot option is
1174 passed or the SYSFS_DEPRECATED_V2 option is set.
1176 This option allows new kernels to run on old distributions and tools,
1177 which might get confused by /sys/class/block/. Since 2007/2008 all
1178 major distributions and tools handle this just fine.
1180 Recent distributions and userspace tools after 2009/2010 depend on
1181 the existence of /sys/class/block/, and will not work with this
1184 Only if you are using a new kernel on an old distribution, you might
1187 config SYSFS_DEPRECATED_V2
1188 bool "Enable deprecated sysfs features by default"
1191 depends on SYSFS_DEPRECATED
1193 Enable deprecated sysfs by default.
1195 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1198 Only if you are using a new kernel on an old distribution, you might
1199 need to say Y here. Even then, odds are you would not need it
1200 enabled, you can always pass the boot option if absolutely necessary.
1203 bool "Kernel->user space relay support (formerly relayfs)"
1206 This option enables support for relay interface support in
1207 certain file systems (such as debugfs).
1208 It is designed to provide an efficient mechanism for tools and
1209 facilities to relay large amounts of data from kernel space to
1214 config BLK_DEV_INITRD
1215 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1217 The initial RAM filesystem is a ramfs which is loaded by the
1218 boot loader (loadlin or lilo) and that is mounted as root
1219 before the normal boot procedure. It is typically used to
1220 load modules needed to mount the "real" root file system,
1221 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1223 If RAM disk support (BLK_DEV_RAM) is also included, this
1224 also enables initial RAM disk (initrd) support and adds
1225 15 Kbytes (more on some other architectures) to the kernel size.
1231 source "usr/Kconfig"
1236 prompt "Compiler optimization level"
1237 default CC_OPTIMIZE_FOR_PERFORMANCE
1239 config CC_OPTIMIZE_FOR_PERFORMANCE
1240 bool "Optimize for performance (-O2)"
1242 This is the default optimization level for the kernel, building
1243 with the "-O2" compiler flag for best performance and most
1244 helpful compile-time warnings.
1246 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1247 bool "Optimize more for performance (-O3)"
1249 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1251 Choosing this option will pass "-O3" to your compiler to optimize
1252 the kernel yet more for performance.
1254 config CC_OPTIMIZE_FOR_SIZE
1255 bool "Optimize for size (-Os)"
1256 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1258 Choosing this option will pass "-Os" to your compiler resulting
1259 in a smaller kernel.
1263 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1266 This requires that the arch annotates or otherwise protects
1267 its external entry points from being discarded. Linker scripts
1268 must also merge .text.*, .data.*, and .bss.* correctly into
1269 output sections. Care must be taken not to pull in unrelated
1270 sections (e.g., '.text.init'). Typically '.' in section names
1271 is used to distinguish them from label names / C identifiers.
1273 config LD_DEAD_CODE_DATA_ELIMINATION
1274 bool "Dead code and data elimination (EXPERIMENTAL)"
1275 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1277 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1278 depends on $(cc-option,-ffunction-sections -fdata-sections)
1279 depends on $(ld-option,--gc-sections)
1281 Enable this if you want to do dead code and data elimination with
1282 the linker by compiling with -ffunction-sections -fdata-sections,
1283 and linking with --gc-sections.
1285 This can reduce on disk and in-memory size of the kernel
1286 code and static data, particularly for small configs and
1287 on small systems. This has the possibility of introducing
1288 silently broken kernel if the required annotations are not
1289 present. This option is not well tested yet, so use at your
1298 config SYSCTL_EXCEPTION_TRACE
1301 Enable support for /proc/sys/debug/exception-trace.
1303 config SYSCTL_ARCH_UNALIGN_NO_WARN
1306 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1307 Allows arch to define/use @no_unaligned_warning to possibly warn
1308 about unaligned access emulation going on under the hood.
1310 config SYSCTL_ARCH_UNALIGN_ALLOW
1313 Enable support for /proc/sys/kernel/unaligned-trap
1314 Allows arches to define/use @unaligned_enabled to runtime toggle
1315 the unaligned access emulation.
1316 see arch/parisc/kernel/unaligned.c for reference
1318 config HAVE_PCSPKR_PLATFORM
1321 # interpreter that classic socket filters depend on
1326 bool "Configure standard kernel features (expert users)"
1327 # Unhide debug options, to make the on-by-default options visible
1330 This option allows certain base kernel options and settings
1331 to be disabled or tweaked. This is for specialized
1332 environments which can tolerate a "non-standard" kernel.
1333 Only use this if you really know what you are doing.
1336 bool "Enable 16-bit UID system calls" if EXPERT
1337 depends on HAVE_UID16 && MULTIUSER
1340 This enables the legacy 16-bit UID syscall wrappers.
1343 bool "Multiple users, groups and capabilities support" if EXPERT
1346 This option enables support for non-root users, groups and
1349 If you say N here, all processes will run with UID 0, GID 0, and all
1350 possible capabilities. Saying N here also compiles out support for
1351 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1354 If unsure, say Y here.
1356 config SGETMASK_SYSCALL
1357 bool "sgetmask/ssetmask syscalls support" if EXPERT
1358 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1360 sys_sgetmask and sys_ssetmask are obsolete system calls
1361 no longer supported in libc but still enabled by default in some
1364 If unsure, leave the default option here.
1366 config SYSFS_SYSCALL
1367 bool "Sysfs syscall support" if EXPERT
1370 sys_sysfs is an obsolete system call no longer supported in libc.
1371 Note that disabling this option is more secure but might break
1372 compatibility with some systems.
1374 If unsure say Y here.
1376 config SYSCTL_SYSCALL
1377 bool "Sysctl syscall support" if EXPERT
1378 depends on PROC_SYSCTL
1382 sys_sysctl uses binary paths that have been found challenging
1383 to properly maintain and use. The interface in /proc/sys
1384 using paths with ascii names is now the primary path to this
1387 Almost nothing using the binary sysctl interface so if you are
1388 trying to save some space it is probably safe to disable this,
1389 making your kernel marginally smaller.
1391 If unsure say N here.
1394 bool "open by fhandle syscalls" if EXPERT
1398 If you say Y here, a user level program will be able to map
1399 file names to handle and then later use the handle for
1400 different file system operations. This is useful in implementing
1401 userspace file servers, which now track files using handles instead
1402 of names. The handle would remain the same even if file names
1403 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1407 bool "Posix Clocks & timers" if EXPERT
1410 This includes native support for POSIX timers to the kernel.
1411 Some embedded systems have no use for them and therefore they
1412 can be configured out to reduce the size of the kernel image.
1414 When this option is disabled, the following syscalls won't be
1415 available: timer_create, timer_gettime: timer_getoverrun,
1416 timer_settime, timer_delete, clock_adjtime, getitimer,
1417 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1418 clock_getres and clock_nanosleep syscalls will be limited to
1419 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1425 bool "Enable support for printk" if EXPERT
1428 This option enables normal printk support. Removing it
1429 eliminates most of the message strings from the kernel image
1430 and makes the kernel more or less silent. As this makes it
1431 very difficult to diagnose system problems, saying N here is
1432 strongly discouraged.
1440 bool "BUG() support" if EXPERT
1443 Disabling this option eliminates support for BUG and WARN, reducing
1444 the size of your kernel image and potentially quietly ignoring
1445 numerous fatal conditions. You should only consider disabling this
1446 option for embedded systems with no facilities for reporting errors.
1452 bool "Enable ELF core dumps" if EXPERT
1454 Enable support for generating core dumps. Disabling saves about 4k.
1457 config PCSPKR_PLATFORM
1458 bool "Enable PC-Speaker support" if EXPERT
1459 depends on HAVE_PCSPKR_PLATFORM
1463 This option allows to disable the internal PC-Speaker
1464 support, saving some memory.
1468 bool "Enable full-sized data structures for core" if EXPERT
1470 Disabling this option reduces the size of miscellaneous core
1471 kernel data structures. This saves memory on small machines,
1472 but may reduce performance.
1475 bool "Enable futex support" if EXPERT
1479 Disabling this option will cause the kernel to be built without
1480 support for "fast userspace mutexes". The resulting kernel may not
1481 run glibc-based applications correctly.
1485 depends on FUTEX && RT_MUTEXES
1488 config HAVE_FUTEX_CMPXCHG
1492 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1493 is implemented and always working. This removes a couple of runtime
1497 bool "Enable eventpoll support" if EXPERT
1500 Disabling this option will cause the kernel to be built without
1501 support for epoll family of system calls.
1504 bool "Enable signalfd() system call" if EXPERT
1507 Enable the signalfd() system call that allows to receive signals
1508 on a file descriptor.
1513 bool "Enable timerfd() system call" if EXPERT
1516 Enable the timerfd() system call that allows to receive timer
1517 events on a file descriptor.
1522 bool "Enable eventfd() system call" if EXPERT
1525 Enable the eventfd() system call that allows to receive both
1526 kernel notification (ie. KAIO) or userspace notifications.
1531 bool "Use full shmem filesystem" if EXPERT
1535 The shmem is an internal filesystem used to manage shared memory.
1536 It is backed by swap and manages resource limits. It is also exported
1537 to userspace as tmpfs if TMPFS is enabled. Disabling this
1538 option replaces shmem and tmpfs with the much simpler ramfs code,
1539 which may be appropriate on small systems without swap.
1542 bool "Enable AIO support" if EXPERT
1545 This option enables POSIX asynchronous I/O which may by used
1546 by some high performance threaded applications. Disabling
1547 this option saves about 7k.
1550 bool "Enable IO uring support" if EXPERT
1554 This option enables support for the io_uring interface, enabling
1555 applications to submit and complete IO through submission and
1556 completion rings that are shared between the kernel and application.
1558 config ADVISE_SYSCALLS
1559 bool "Enable madvise/fadvise syscalls" if EXPERT
1562 This option enables the madvise and fadvise syscalls, used by
1563 applications to advise the kernel about their future memory or file
1564 usage, improving performance. If building an embedded system where no
1565 applications use these syscalls, you can disable this option to save
1569 bool "Enable membarrier() system call" if EXPERT
1572 Enable the membarrier() system call that allows issuing memory
1573 barriers across all running threads, which can be used to distribute
1574 the cost of user-space memory barriers asymmetrically by transforming
1575 pairs of memory barriers into pairs consisting of membarrier() and a
1581 bool "Load all symbols for debugging/ksymoops" if EXPERT
1584 Say Y here to let the kernel print out symbolic crash information and
1585 symbolic stack backtraces. This increases the size of the kernel
1586 somewhat, as all symbols have to be loaded into the kernel image.
1589 bool "Include all symbols in kallsyms"
1590 depends on DEBUG_KERNEL && KALLSYMS
1592 Normally kallsyms only contains the symbols of functions for nicer
1593 OOPS messages and backtraces (i.e., symbols from the text and inittext
1594 sections). This is sufficient for most cases. And only in very rare
1595 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1596 names of variables from the data sections, etc).
1598 This option makes sure that all symbols are loaded into the kernel
1599 image (i.e., symbols from all sections) in cost of increased kernel
1600 size (depending on the kernel configuration, it may be 300KiB or
1601 something like this).
1603 Say N unless you really need all symbols.
1605 config KALLSYMS_ABSOLUTE_PERCPU
1608 default X86_64 && SMP
1610 config KALLSYMS_BASE_RELATIVE
1615 Instead of emitting them as absolute values in the native word size,
1616 emit the symbol references in the kallsyms table as 32-bit entries,
1617 each containing a relative value in the range [base, base + U32_MAX]
1618 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1619 an absolute value in the range [0, S32_MAX] or a relative value in the
1620 range [base, base + S32_MAX], where base is the lowest relative symbol
1621 address encountered in the image.
1623 On 64-bit builds, this reduces the size of the address table by 50%,
1624 but more importantly, it results in entries whose values are build
1625 time constants, and no relocation pass is required at runtime to fix
1626 up the entries based on the runtime load address of the kernel.
1628 # end of the "standard kernel features (expert users)" menu
1630 # syscall, maps, verifier
1632 bool "Enable bpf() system call"
1637 Enable the bpf() system call that allows to manipulate eBPF
1638 programs and maps via file descriptors.
1640 config BPF_JIT_ALWAYS_ON
1641 bool "Permanently enable BPF JIT and remove BPF interpreter"
1642 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1644 Enables BPF JIT and removes BPF interpreter to avoid
1645 speculative execution of BPF instructions by the interpreter
1648 bool "Enable userfaultfd() system call"
1651 Enable the userfaultfd() system call that allows to intercept and
1652 handle page faults in userland.
1654 config ARCH_HAS_MEMBARRIER_CALLBACKS
1657 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1661 bool "Enable rseq() system call" if EXPERT
1663 depends on HAVE_RSEQ
1666 Enable the restartable sequences system call. It provides a
1667 user-space cache for the current CPU number value, which
1668 speeds up getting the current CPU number from user-space,
1669 as well as an ABI to speed up user-space operations on
1676 bool "Enabled debugging of rseq() system call" if EXPERT
1677 depends on RSEQ && DEBUG_KERNEL
1679 Enable extra debugging checks for the rseq system call.
1684 bool "Embedded system"
1685 option allnoconfig_y
1688 This option should be enabled if compiling the kernel for
1689 an embedded system so certain expert options are available
1692 config HAVE_PERF_EVENTS
1695 See tools/perf/design.txt for details.
1697 config PERF_USE_VMALLOC
1700 See tools/perf/design.txt for details
1703 bool "PC/104 support" if EXPERT
1705 Expose PC/104 form factor device drivers and options available for
1706 selection and configuration. Enable this option if your target
1707 machine has a PC/104 bus.
1709 menu "Kernel Performance Events And Counters"
1712 bool "Kernel performance events and counters"
1713 default y if PROFILING
1714 depends on HAVE_PERF_EVENTS
1718 Enable kernel support for various performance events provided
1719 by software and hardware.
1721 Software events are supported either built-in or via the
1722 use of generic tracepoints.
1724 Most modern CPUs support performance events via performance
1725 counter registers. These registers count the number of certain
1726 types of hw events: such as instructions executed, cachemisses
1727 suffered, or branches mis-predicted - without slowing down the
1728 kernel or applications. These registers can also trigger interrupts
1729 when a threshold number of events have passed - and can thus be
1730 used to profile the code that runs on that CPU.
1732 The Linux Performance Event subsystem provides an abstraction of
1733 these software and hardware event capabilities, available via a
1734 system call and used by the "perf" utility in tools/perf/. It
1735 provides per task and per CPU counters, and it provides event
1736 capabilities on top of those.
1740 config DEBUG_PERF_USE_VMALLOC
1742 bool "Debug: use vmalloc to back perf mmap() buffers"
1743 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1744 select PERF_USE_VMALLOC
1746 Use vmalloc memory to back perf mmap() buffers.
1748 Mostly useful for debugging the vmalloc code on platforms
1749 that don't require it.
1755 config VM_EVENT_COUNTERS
1757 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1759 VM event counters are needed for event counts to be shown.
1760 This option allows the disabling of the VM event counters
1761 on EXPERT systems. /proc/vmstat will only show page counts
1762 if VM event counters are disabled.
1766 bool "Enable SLUB debugging support" if EXPERT
1767 depends on SLUB && SYSFS
1769 SLUB has extensive debug support features. Disabling these can
1770 result in significant savings in code size. This also disables
1771 SLUB sysfs support. /sys/slab will not exist and there will be
1772 no support for cache validation etc.
1774 config SLUB_MEMCG_SYSFS_ON
1776 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1777 depends on SLUB && SYSFS && MEMCG
1779 SLUB creates a directory under /sys/kernel/slab for each
1780 allocation cache to host info and debug files. If memory
1781 cgroup is enabled, each cache can have per memory cgroup
1782 caches. SLUB can create the same sysfs directories for these
1783 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1784 to a very high number of debug files being created. This is
1785 controlled by slub_memcg_sysfs boot parameter and this
1786 config option determines the parameter's default value.
1789 bool "Disable heap randomization"
1792 Randomizing heap placement makes heap exploits harder, but it
1793 also breaks ancient binaries (including anything libc5 based).
1794 This option changes the bootup default to heap randomization
1795 disabled, and can be overridden at runtime by setting
1796 /proc/sys/kernel/randomize_va_space to 2.
1798 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1801 prompt "Choose SLAB allocator"
1804 This option allows to select a slab allocator.
1808 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1810 The regular slab allocator that is established and known to work
1811 well in all environments. It organizes cache hot objects in
1812 per cpu and per node queues.
1815 bool "SLUB (Unqueued Allocator)"
1816 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1818 SLUB is a slab allocator that minimizes cache line usage
1819 instead of managing queues of cached objects (SLAB approach).
1820 Per cpu caching is realized using slabs of objects instead
1821 of queues of objects. SLUB can use memory efficiently
1822 and has enhanced diagnostics. SLUB is the default choice for
1827 bool "SLOB (Simple Allocator)"
1829 SLOB replaces the stock allocator with a drastically simpler
1830 allocator. SLOB is generally more space efficient but
1831 does not perform as well on large systems.
1835 config SLAB_MERGE_DEFAULT
1836 bool "Allow slab caches to be merged"
1839 For reduced kernel memory fragmentation, slab caches can be
1840 merged when they share the same size and other characteristics.
1841 This carries a risk of kernel heap overflows being able to
1842 overwrite objects from merged caches (and more easily control
1843 cache layout), which makes such heap attacks easier to exploit
1844 by attackers. By keeping caches unmerged, these kinds of exploits
1845 can usually only damage objects in the same cache. To disable
1846 merging at runtime, "slab_nomerge" can be passed on the kernel
1849 config SLAB_FREELIST_RANDOM
1851 depends on SLAB || SLUB
1852 bool "SLAB freelist randomization"
1854 Randomizes the freelist order used on creating new pages. This
1855 security feature reduces the predictability of the kernel slab
1856 allocator against heap overflows.
1858 config SLAB_FREELIST_HARDENED
1859 bool "Harden slab freelist metadata"
1862 Many kernel heap attacks try to target slab cache metadata and
1863 other infrastructure. This options makes minor performance
1864 sacrifices to harden the kernel slab allocator against common
1865 freelist exploit methods.
1867 config SHUFFLE_PAGE_ALLOCATOR
1868 bool "Page allocator randomization"
1869 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1871 Randomization of the page allocator improves the average
1872 utilization of a direct-mapped memory-side-cache. See section
1873 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1874 6.2a specification for an example of how a platform advertises
1875 the presence of a memory-side-cache. There are also incidental
1876 security benefits as it reduces the predictability of page
1877 allocations to compliment SLAB_FREELIST_RANDOM, but the
1878 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1879 10th order of pages is selected based on cache utilization
1882 While the randomization improves cache utilization it may
1883 negatively impact workloads on platforms without a cache. For
1884 this reason, by default, the randomization is enabled only
1885 after runtime detection of a direct-mapped memory-side-cache.
1886 Otherwise, the randomization may be force enabled with the
1887 'page_alloc.shuffle' kernel command line parameter.
1891 config SLUB_CPU_PARTIAL
1893 depends on SLUB && SMP
1894 bool "SLUB per cpu partial cache"
1896 Per cpu partial caches accelerate objects allocation and freeing
1897 that is local to a processor at the price of more indeterminism
1898 in the latency of the free. On overflow these caches will be cleared
1899 which requires the taking of locks that may cause latency spikes.
1900 Typically one would choose no for a realtime system.
1902 config MMAP_ALLOW_UNINITIALIZED
1903 bool "Allow mmapped anonymous memory to be uninitialized"
1904 depends on EXPERT && !MMU
1907 Normally, and according to the Linux spec, anonymous memory obtained
1908 from mmap() has its contents cleared before it is passed to
1909 userspace. Enabling this config option allows you to request that
1910 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1911 providing a huge performance boost. If this option is not enabled,
1912 then the flag will be ignored.
1914 This is taken advantage of by uClibc's malloc(), and also by
1915 ELF-FDPIC binfmt's brk and stack allocator.
1917 Because of the obvious security issues, this option should only be
1918 enabled on embedded devices where you control what is run in
1919 userspace. Since that isn't generally a problem on no-MMU systems,
1920 it is normally safe to say Y here.
1922 See Documentation/nommu-mmap.txt for more information.
1924 config SYSTEM_DATA_VERIFICATION
1926 select SYSTEM_TRUSTED_KEYRING
1930 select ASYMMETRIC_KEY_TYPE
1931 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1934 select X509_CERTIFICATE_PARSER
1935 select PKCS7_MESSAGE_PARSER
1937 Provide PKCS#7 message verification using the contents of the system
1938 trusted keyring to provide public keys. This then can be used for
1939 module verification, kexec image verification and firmware blob
1943 bool "Profiling support"
1945 Say Y here to enable the extended profiling support mechanisms used
1946 by profilers such as OProfile.
1949 # Place an empty function call at each tracepoint site. Can be
1950 # dynamically changed for a probe function.
1955 endmenu # General setup
1957 source "arch/Kconfig"
1964 default 0 if BASE_FULL
1965 default 1 if !BASE_FULL
1968 bool "Enable loadable module support"
1971 Kernel modules are small pieces of compiled code which can
1972 be inserted in the running kernel, rather than being
1973 permanently built into the kernel. You use the "modprobe"
1974 tool to add (and sometimes remove) them. If you say Y here,
1975 many parts of the kernel can be built as modules (by
1976 answering M instead of Y where indicated): this is most
1977 useful for infrequently used options which are not required
1978 for booting. For more information, see the man pages for
1979 modprobe, lsmod, modinfo, insmod and rmmod.
1981 If you say Y here, you will need to run "make
1982 modules_install" to put the modules under /lib/modules/
1983 where modprobe can find them (you may need to be root to do
1990 config MODULE_FORCE_LOAD
1991 bool "Forced module loading"
1994 Allow loading of modules without version information (ie. modprobe
1995 --force). Forced module loading sets the 'F' (forced) taint flag and
1996 is usually a really bad idea.
1998 config MODULE_UNLOAD
1999 bool "Module unloading"
2001 Without this option you will not be able to unload any
2002 modules (note that some modules may not be unloadable
2003 anyway), which makes your kernel smaller, faster
2004 and simpler. If unsure, say Y.
2006 config MODULE_FORCE_UNLOAD
2007 bool "Forced module unloading"
2008 depends on MODULE_UNLOAD
2010 This option allows you to force a module to unload, even if the
2011 kernel believes it is unsafe: the kernel will remove the module
2012 without waiting for anyone to stop using it (using the -f option to
2013 rmmod). This is mainly for kernel developers and desperate users.
2017 bool "Module versioning support"
2019 Usually, you have to use modules compiled with your kernel.
2020 Saying Y here makes it sometimes possible to use modules
2021 compiled for different kernels, by adding enough information
2022 to the modules to (hopefully) spot any changes which would
2023 make them incompatible with the kernel you are running. If
2026 config ASM_MODVERSIONS
2028 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2030 This enables module versioning for exported symbols also from
2031 assembly. This can be enabled only when the target architecture
2034 config MODULE_REL_CRCS
2036 depends on MODVERSIONS
2038 config MODULE_SRCVERSION_ALL
2039 bool "Source checksum for all modules"
2041 Modules which contain a MODULE_VERSION get an extra "srcversion"
2042 field inserted into their modinfo section, which contains a
2043 sum of the source files which made it. This helps maintainers
2044 see exactly which source was used to build a module (since
2045 others sometimes change the module source without updating
2046 the version). With this option, such a "srcversion" field
2047 will be created for all modules. If unsure, say N.
2050 bool "Module signature verification"
2052 select SYSTEM_DATA_VERIFICATION
2054 Check modules for valid signatures upon load: the signature
2055 is simply appended to the module. For more information see
2056 <file:Documentation/admin-guide/module-signing.rst>.
2058 Note that this option adds the OpenSSL development packages as a
2059 kernel build dependency so that the signing tool can use its crypto
2062 !!!WARNING!!! If you enable this option, you MUST make sure that the
2063 module DOES NOT get stripped after being signed. This includes the
2064 debuginfo strip done by some packagers (such as rpmbuild) and
2065 inclusion into an initramfs that wants the module size reduced.
2067 config MODULE_SIG_FORCE
2068 bool "Require modules to be validly signed"
2069 depends on MODULE_SIG
2071 Reject unsigned modules or signed modules for which we don't have a
2072 key. Without this, such modules will simply taint the kernel.
2074 config MODULE_SIG_ALL
2075 bool "Automatically sign all modules"
2077 depends on MODULE_SIG
2079 Sign all modules during make modules_install. Without this option,
2080 modules must be signed manually, using the scripts/sign-file tool.
2082 comment "Do not forget to sign required modules with scripts/sign-file"
2083 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2086 prompt "Which hash algorithm should modules be signed with?"
2087 depends on MODULE_SIG
2089 This determines which sort of hashing algorithm will be used during
2090 signature generation. This algorithm _must_ be built into the kernel
2091 directly so that signature verification can take place. It is not
2092 possible to load a signed module containing the algorithm to check
2093 the signature on that module.
2095 config MODULE_SIG_SHA1
2096 bool "Sign modules with SHA-1"
2099 config MODULE_SIG_SHA224
2100 bool "Sign modules with SHA-224"
2101 select CRYPTO_SHA256
2103 config MODULE_SIG_SHA256
2104 bool "Sign modules with SHA-256"
2105 select CRYPTO_SHA256
2107 config MODULE_SIG_SHA384
2108 bool "Sign modules with SHA-384"
2109 select CRYPTO_SHA512
2111 config MODULE_SIG_SHA512
2112 bool "Sign modules with SHA-512"
2113 select CRYPTO_SHA512
2117 config MODULE_SIG_HASH
2119 depends on MODULE_SIG
2120 default "sha1" if MODULE_SIG_SHA1
2121 default "sha224" if MODULE_SIG_SHA224
2122 default "sha256" if MODULE_SIG_SHA256
2123 default "sha384" if MODULE_SIG_SHA384
2124 default "sha512" if MODULE_SIG_SHA512
2126 config MODULE_COMPRESS
2127 bool "Compress modules on installation"
2131 Compresses kernel modules when 'make modules_install' is run; gzip or
2132 xz depending on "Compression algorithm" below.
2134 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2136 Out-of-tree kernel modules installed using Kbuild will also be
2137 compressed upon installation.
2139 Note: for modules inside an initrd or initramfs, it's more efficient
2140 to compress the whole initrd or initramfs instead.
2142 Note: This is fully compatible with signed modules.
2147 prompt "Compression algorithm"
2148 depends on MODULE_COMPRESS
2149 default MODULE_COMPRESS_GZIP
2151 This determines which sort of compression will be used during
2152 'make modules_install'.
2154 GZIP (default) and XZ are supported.
2156 config MODULE_COMPRESS_GZIP
2159 config MODULE_COMPRESS_XZ
2164 config TRIM_UNUSED_KSYMS
2165 bool "Trim unused exported kernel symbols"
2166 depends on MODULES && !UNUSED_SYMBOLS
2168 The kernel and some modules make many symbols available for
2169 other modules to use via EXPORT_SYMBOL() and variants. Depending
2170 on the set of modules being selected in your kernel configuration,
2171 many of those exported symbols might never be used.
2173 This option allows for unused exported symbols to be dropped from
2174 the build. In turn, this provides the compiler more opportunities
2175 (especially when using LTO) for optimizing the code and reducing
2176 binary size. This might have some security advantages as well.
2178 If unsure, or if you need to build out-of-tree modules, say N.
2182 config MODULES_TREE_LOOKUP
2184 depends on PERF_EVENTS || TRACING
2186 config INIT_ALL_POSSIBLE
2189 Back when each arch used to define their own cpu_online_mask and
2190 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2191 with all 1s, and others with all 0s. When they were centralised,
2192 it was better to provide this option than to break all the archs
2193 and have several arch maintainers pursuing me down dark alleys.
2195 source "block/Kconfig"
2197 config PREEMPT_NOTIFIERS
2207 Build a simple ASN.1 grammar compiler that produces a bytecode output
2208 that can be interpreted by the ASN.1 stream decoder and used to
2209 inform it as to what tags are to be expected in a stream and what
2210 functions to call on what tags.
2212 source "kernel/Kconfig.locks"
2214 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2217 # It may be useful for an architecture to override the definitions of the
2218 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2219 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2220 # different calling convention for syscalls. They can also override the
2221 # macros for not-implemented syscalls in kernel/sys_ni.c and
2222 # kernel/time/posix-stubs.c. All these overrides need to be available in
2223 # <asm/syscall_wrapper.h>.
2224 config ARCH_HAS_SYSCALL_WRAPPER