1 # SPDX-License-Identifier: GPL-2.0-only
6 default "/lib/modules/$(shell,uname -r)/.config"
7 default "/etc/kernel-config"
8 default "/boot/config-$(shell,uname -r)"
9 default "arch/$(SRCARCH)/configs/$(KBUILD_DEFCONFIG)"
12 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
16 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
21 default $(shell,$(LD) --version | $(srctree)/scripts/ld-version.sh)
24 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
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 TOOLS_SUPPORT_RELR
37 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
39 config CC_HAS_ASM_INLINE
40 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
49 config BUILDTIME_TABLE_SORT
52 config THREAD_INFO_IN_TASK
55 Select this to move thread_info off the stack into task_struct. To
56 make this work, an arch will need to remove all thread_info fields
57 except flags and fix any runtime bugs.
59 One subtle change that will be needed is to use try_get_task_stack()
60 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
69 depends on BROKEN || !SMP
72 config INIT_ENV_ARG_LIMIT
77 Maximum of each of the number of arguments and environment
78 variables passed to init from the kernel command line.
81 bool "Compile also drivers which will not load"
85 Some drivers can be compiled on a different platform than they are
86 intended to be run on. Despite they cannot be loaded there (or even
87 when they load they cannot be used due to missing HW support),
88 developers still, opposing to distributors, might want to build such
89 drivers to compile-test them.
91 If you are a developer and want to build everything available, say Y
92 here. If you are a user/distributor, say N here to exclude useless
93 drivers to be distributed.
95 config UAPI_HEADER_TEST
96 bool "Compile test UAPI headers"
97 depends on HEADERS_INSTALL && CC_CAN_LINK
99 Compile test headers exported to user-space to ensure they are
100 self-contained, i.e. compilable as standalone units.
102 If you are a developer or tester and want to ensure the exported
103 headers are self-contained, say Y here. Otherwise, choose N.
106 string "Local version - append to kernel release"
108 Append an extra string to the end of your kernel version.
109 This will show up when you type uname, for example.
110 The string you set here will be appended after the contents of
111 any files with a filename matching localversion* in your
112 object and source tree, in that order. Your total string can
113 be a maximum of 64 characters.
115 config LOCALVERSION_AUTO
116 bool "Automatically append version information to the version string"
118 depends on !COMPILE_TEST
120 This will try to automatically determine if the current tree is a
121 release tree by looking for git tags that belong to the current
122 top of tree revision.
124 A string of the format -gxxxxxxxx will be added to the localversion
125 if a git-based tree is found. The string generated by this will be
126 appended after any matching localversion* files, and after the value
127 set in CONFIG_LOCALVERSION.
129 (The actual string used here is the first eight characters produced
130 by running the command:
132 $ git rev-parse --verify HEAD
134 which is done within the script "scripts/setlocalversion".)
137 string "Build ID Salt"
140 The build ID is used to link binaries and their debug info. Setting
141 this option will use the value in the calculation of the build id.
142 This is mostly useful for distributions which want to ensure the
143 build is unique between builds. It's safe to leave the default.
145 config HAVE_KERNEL_GZIP
148 config HAVE_KERNEL_BZIP2
151 config HAVE_KERNEL_LZMA
154 config HAVE_KERNEL_XZ
157 config HAVE_KERNEL_LZO
160 config HAVE_KERNEL_LZ4
163 config HAVE_KERNEL_UNCOMPRESSED
167 prompt "Kernel compression mode"
169 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
171 The linux kernel is a kind of self-extracting executable.
172 Several compression algorithms are available, which differ
173 in efficiency, compression and decompression speed.
174 Compression speed is only relevant when building a kernel.
175 Decompression speed is relevant at each boot.
177 If you have any problems with bzip2 or lzma compressed
178 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
179 version of this functionality (bzip2 only), for 2.4, was
180 supplied by Christian Ludwig)
182 High compression options are mostly useful for users, who
183 are low on disk space (embedded systems), but for whom ram
186 If in doubt, select 'gzip'
190 depends on HAVE_KERNEL_GZIP
192 The old and tried gzip compression. It provides a good balance
193 between compression ratio and decompression speed.
197 depends on HAVE_KERNEL_BZIP2
199 Its compression ratio and speed is intermediate.
200 Decompression speed is slowest among the choices. The kernel
201 size is about 10% smaller with bzip2, in comparison to gzip.
202 Bzip2 uses a large amount of memory. For modern kernels you
203 will need at least 8MB RAM or more for booting.
207 depends on HAVE_KERNEL_LZMA
209 This compression algorithm's ratio is best. Decompression speed
210 is between gzip and bzip2. Compression is slowest.
211 The kernel size is about 33% smaller with LZMA in comparison to gzip.
215 depends on HAVE_KERNEL_XZ
217 XZ uses the LZMA2 algorithm and instruction set specific
218 BCJ filters which can improve compression ratio of executable
219 code. The size of the kernel is about 30% smaller with XZ in
220 comparison to gzip. On architectures for which there is a BCJ
221 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
222 will create a few percent smaller kernel than plain LZMA.
224 The speed is about the same as with LZMA: The decompression
225 speed of XZ is better than that of bzip2 but worse than gzip
226 and LZO. Compression is slow.
230 depends on HAVE_KERNEL_LZO
232 Its compression ratio is the poorest among the choices. The kernel
233 size is about 10% bigger than gzip; however its speed
234 (both compression and decompression) is the fastest.
238 depends on HAVE_KERNEL_LZ4
240 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
241 A preliminary version of LZ4 de/compression tool is available at
242 <https://code.google.com/p/lz4/>.
244 Its compression ratio is worse than LZO. The size of the kernel
245 is about 8% bigger than LZO. But the decompression speed is
248 config KERNEL_UNCOMPRESSED
250 depends on HAVE_KERNEL_UNCOMPRESSED
252 Produce uncompressed kernel image. This option is usually not what
253 you want. It is useful for debugging the kernel in slow simulation
254 environments, where decompressing and moving the kernel is awfully
255 slow. This option allows early boot code to skip the decompressor
256 and jump right at uncompressed kernel image.
260 config DEFAULT_HOSTNAME
261 string "Default hostname"
264 This option determines the default system hostname before userspace
265 calls sethostname(2). The kernel traditionally uses "(none)" here,
266 but you may wish to use a different default here to make a minimal
267 system more usable with less configuration.
270 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
271 # add proper SWAP support to them, in which case this can be remove.
277 bool "Support for paging of anonymous memory (swap)"
278 depends on MMU && BLOCK && !ARCH_NO_SWAP
281 This option allows you to choose whether you want to have support
282 for so called swap devices or swap files in your kernel that are
283 used to provide more virtual memory than the actual RAM present
284 in your computer. If unsure say Y.
289 Inter Process Communication is a suite of library functions and
290 system calls which let processes (running programs) synchronize and
291 exchange information. It is generally considered to be a good thing,
292 and some programs won't run unless you say Y here. In particular, if
293 you want to run the DOS emulator dosemu under Linux (read the
294 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
295 you'll need to say Y here.
297 You can find documentation about IPC with "info ipc" and also in
298 section 6.4 of the Linux Programmer's Guide, available from
299 <http://www.tldp.org/guides.html>.
301 config SYSVIPC_SYSCTL
308 bool "POSIX Message Queues"
311 POSIX variant of message queues is a part of IPC. In POSIX message
312 queues every message has a priority which decides about succession
313 of receiving it by a process. If you want to compile and run
314 programs written e.g. for Solaris with use of its POSIX message
315 queues (functions mq_*) say Y here.
317 POSIX message queues are visible as a filesystem called 'mqueue'
318 and can be mounted somewhere if you want to do filesystem
319 operations on message queues.
323 config POSIX_MQUEUE_SYSCTL
325 depends on POSIX_MQUEUE
329 config CROSS_MEMORY_ATTACH
330 bool "Enable process_vm_readv/writev syscalls"
334 Enabling this option adds the system calls process_vm_readv and
335 process_vm_writev which allow a process with the correct privileges
336 to directly read from or write to another process' address space.
337 See the man page for more details.
340 bool "uselib syscall"
341 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
343 This option enables the uselib syscall, a system call used in the
344 dynamic linker from libc5 and earlier. glibc does not use this
345 system call. If you intend to run programs built on libc5 or
346 earlier, you may need to enable this syscall. Current systems
347 running glibc can safely disable this.
350 bool "Auditing support"
353 Enable auditing infrastructure that can be used with another
354 kernel subsystem, such as SELinux (which requires this for
355 logging of avc messages output). System call auditing is included
356 on architectures which support it.
358 config HAVE_ARCH_AUDITSYSCALL
363 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
366 source "kernel/irq/Kconfig"
367 source "kernel/time/Kconfig"
368 source "kernel/Kconfig.preempt"
370 menu "CPU/Task time and stats accounting"
372 config VIRT_CPU_ACCOUNTING
376 prompt "Cputime accounting"
377 default TICK_CPU_ACCOUNTING if !PPC64
378 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
380 # Kind of a stub config for the pure tick based cputime accounting
381 config TICK_CPU_ACCOUNTING
382 bool "Simple tick based cputime accounting"
383 depends on !S390 && !NO_HZ_FULL
385 This is the basic tick based cputime accounting that maintains
386 statistics about user, system and idle time spent on per jiffies
391 config VIRT_CPU_ACCOUNTING_NATIVE
392 bool "Deterministic task and CPU time accounting"
393 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
394 select VIRT_CPU_ACCOUNTING
396 Select this option to enable more accurate task and CPU time
397 accounting. This is done by reading a CPU counter on each
398 kernel entry and exit and on transitions within the kernel
399 between system, softirq and hardirq state, so there is a
400 small performance impact. In the case of s390 or IBM POWER > 5,
401 this also enables accounting of stolen time on logically-partitioned
404 config VIRT_CPU_ACCOUNTING_GEN
405 bool "Full dynticks CPU time accounting"
406 depends on HAVE_CONTEXT_TRACKING
407 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
408 depends on GENERIC_CLOCKEVENTS
409 select VIRT_CPU_ACCOUNTING
410 select CONTEXT_TRACKING
412 Select this option to enable task and CPU time accounting on full
413 dynticks systems. This accounting is implemented by watching every
414 kernel-user boundaries using the context tracking subsystem.
415 The accounting is thus performed at the expense of some significant
418 For now this is only useful if you are working on the full
419 dynticks subsystem development.
425 config IRQ_TIME_ACCOUNTING
426 bool "Fine granularity task level IRQ time accounting"
427 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
429 Select this option to enable fine granularity task irq time
430 accounting. This is done by reading a timestamp on each
431 transitions between softirq and hardirq state, so there can be a
432 small performance impact.
434 If in doubt, say N here.
436 config HAVE_SCHED_AVG_IRQ
438 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
441 config SCHED_THERMAL_PRESSURE
442 bool "Enable periodic averaging of thermal pressure"
445 config BSD_PROCESS_ACCT
446 bool "BSD Process Accounting"
449 If you say Y here, a user level program will be able to instruct the
450 kernel (via a special system call) to write process accounting
451 information to a file: whenever a process exits, information about
452 that process will be appended to the file by the kernel. The
453 information includes things such as creation time, owning user,
454 command name, memory usage, controlling terminal etc. (the complete
455 list is in the struct acct in <file:include/linux/acct.h>). It is
456 up to the user level program to do useful things with this
457 information. This is generally a good idea, so say Y.
459 config BSD_PROCESS_ACCT_V3
460 bool "BSD Process Accounting version 3 file format"
461 depends on BSD_PROCESS_ACCT
464 If you say Y here, the process accounting information is written
465 in a new file format that also logs the process IDs of each
466 process and its parent. Note that this file format is incompatible
467 with previous v0/v1/v2 file formats, so you will need updated tools
468 for processing it. A preliminary version of these tools is available
469 at <http://www.gnu.org/software/acct/>.
472 bool "Export task/process statistics through netlink"
477 Export selected statistics for tasks/processes through the
478 generic netlink interface. Unlike BSD process accounting, the
479 statistics are available during the lifetime of tasks/processes as
480 responses to commands. Like BSD accounting, they are sent to user
485 config TASK_DELAY_ACCT
486 bool "Enable per-task delay accounting"
490 Collect information on time spent by a task waiting for system
491 resources like cpu, synchronous block I/O completion and swapping
492 in pages. Such statistics can help in setting a task's priorities
493 relative to other tasks for cpu, io, rss limits etc.
498 bool "Enable extended accounting over taskstats"
501 Collect extended task accounting data and send the data
502 to userland for processing over the taskstats interface.
506 config TASK_IO_ACCOUNTING
507 bool "Enable per-task storage I/O accounting"
508 depends on TASK_XACCT
510 Collect information on the number of bytes of storage I/O which this
516 bool "Pressure stall information tracking"
518 Collect metrics that indicate how overcommitted the CPU, memory,
519 and IO capacity are in the system.
521 If you say Y here, the kernel will create /proc/pressure/ with the
522 pressure statistics files cpu, memory, and io. These will indicate
523 the share of walltime in which some or all tasks in the system are
524 delayed due to contention of the respective resource.
526 In kernels with cgroup support, cgroups (cgroup2 only) will
527 have cpu.pressure, memory.pressure, and io.pressure files,
528 which aggregate pressure stalls for the grouped tasks only.
530 For more details see Documentation/accounting/psi.rst.
534 config PSI_DEFAULT_DISABLED
535 bool "Require boot parameter to enable pressure stall information tracking"
539 If set, pressure stall information tracking will be disabled
540 per default but can be enabled through passing psi=1 on the
541 kernel commandline during boot.
543 This feature adds some code to the task wakeup and sleep
544 paths of the scheduler. The overhead is too low to affect
545 common scheduling-intense workloads in practice (such as
546 webservers, memcache), but it does show up in artificial
547 scheduler stress tests, such as hackbench.
549 If you are paranoid and not sure what the kernel will be
554 endmenu # "CPU/Task time and stats accounting"
558 depends on SMP || COMPILE_TEST
561 Make sure that CPUs running critical tasks are not disturbed by
562 any source of "noise" such as unbound workqueues, timers, kthreads...
563 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
564 the "isolcpus=" boot parameter.
568 source "kernel/rcu/Kconfig"
575 tristate "Kernel .config support"
577 This option enables the complete Linux kernel ".config" file
578 contents to be saved in the kernel. It provides documentation
579 of which kernel options are used in a running kernel or in an
580 on-disk kernel. This information can be extracted from the kernel
581 image file with the script scripts/extract-ikconfig and used as
582 input to rebuild the current kernel or to build another kernel.
583 It can also be extracted from a running kernel by reading
584 /proc/config.gz if enabled (below).
587 bool "Enable access to .config through /proc/config.gz"
588 depends on IKCONFIG && PROC_FS
590 This option enables access to the kernel configuration file
591 through /proc/config.gz.
594 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
597 This option enables access to the in-kernel headers that are generated during
598 the build process. These can be used to build eBPF tracing programs,
599 or similar programs. If you build the headers as a module, a module called
600 kheaders.ko is built which can be loaded on-demand to get access to headers.
603 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
608 Select the minimal kernel log buffer size as a power of 2.
609 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
610 parameter, see below. Any higher size also might be forced
611 by "log_buf_len" boot parameter.
621 config LOG_CPU_MAX_BUF_SHIFT
622 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
625 default 12 if !BASE_SMALL
626 default 0 if BASE_SMALL
629 This option allows to increase the default ring buffer size
630 according to the number of CPUs. The value defines the contribution
631 of each CPU as a power of 2. The used space is typically only few
632 lines however it might be much more when problems are reported,
635 The increased size means that a new buffer has to be allocated and
636 the original static one is unused. It makes sense only on systems
637 with more CPUs. Therefore this value is used only when the sum of
638 contributions is greater than the half of the default kernel ring
639 buffer as defined by LOG_BUF_SHIFT. The default values are set
640 so that more than 64 CPUs are needed to trigger the allocation.
642 Also this option is ignored when "log_buf_len" kernel parameter is
643 used as it forces an exact (power of two) size of the ring buffer.
645 The number of possible CPUs is used for this computation ignoring
646 hotplugging making the computation optimal for the worst case
647 scenario while allowing a simple algorithm to be used from bootup.
649 Examples shift values and their meaning:
650 17 => 128 KB for each CPU
651 16 => 64 KB for each CPU
652 15 => 32 KB for each CPU
653 14 => 16 KB for each CPU
654 13 => 8 KB for each CPU
655 12 => 4 KB for each CPU
657 config PRINTK_SAFE_LOG_BUF_SHIFT
658 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
663 Select the size of an alternate printk per-CPU buffer where messages
664 printed from usafe contexts are temporary stored. One example would
665 be NMI messages, another one - printk recursion. The messages are
666 copied to the main log buffer in a safe context to avoid a deadlock.
667 The value defines the size as a power of 2.
669 Those messages are rare and limited. The largest one is when
670 a backtrace is printed. It usually fits into 4KB. Select
671 8KB if you want to be on the safe side.
674 17 => 128 KB for each CPU
675 16 => 64 KB for each CPU
676 15 => 32 KB for each CPU
677 14 => 16 KB for each CPU
678 13 => 8 KB for each CPU
679 12 => 4 KB for each CPU
682 # Architectures with an unreliable sched_clock() should select this:
684 config HAVE_UNSTABLE_SCHED_CLOCK
687 config GENERIC_SCHED_CLOCK
690 menu "Scheduler features"
693 bool "Enable utilization clamping for RT/FAIR tasks"
694 depends on CPU_FREQ_GOV_SCHEDUTIL
696 This feature enables the scheduler to track the clamped utilization
697 of each CPU based on RUNNABLE tasks scheduled on that CPU.
699 With this option, the user can specify the min and max CPU
700 utilization allowed for RUNNABLE tasks. The max utilization defines
701 the maximum frequency a task should use while the min utilization
702 defines the minimum frequency it should use.
704 Both min and max utilization clamp values are hints to the scheduler,
705 aiming at improving its frequency selection policy, but they do not
706 enforce or grant any specific bandwidth for tasks.
710 config UCLAMP_BUCKETS_COUNT
711 int "Number of supported utilization clamp buckets"
714 depends on UCLAMP_TASK
716 Defines the number of clamp buckets to use. The range of each bucket
717 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
718 number of clamp buckets the finer their granularity and the higher
719 the precision of clamping aggregation and tracking at run-time.
721 For example, with the minimum configuration value we will have 5
722 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
723 be refcounted in the [20..39]% bucket and will set the bucket clamp
724 effective value to 25%.
725 If a second 30% boosted task should be co-scheduled on the same CPU,
726 that task will be refcounted in the same bucket of the first task and
727 it will boost the bucket clamp effective value to 30%.
728 The clamp effective value of a bucket is reset to its nominal value
729 (20% in the example above) when there are no more tasks refcounted in
732 An additional boost/capping margin can be added to some tasks. In the
733 example above the 25% task will be boosted to 30% until it exits the
734 CPU. If that should be considered not acceptable on certain systems,
735 it's always possible to reduce the margin by increasing the number of
736 clamp buckets to trade off used memory for run-time tracking
739 If in doubt, use the default value.
744 # For architectures that want to enable the support for NUMA-affine scheduler
747 config ARCH_SUPPORTS_NUMA_BALANCING
751 # For architectures that prefer to flush all TLBs after a number of pages
752 # are unmapped instead of sending one IPI per page to flush. The architecture
753 # must provide guarantees on what happens if a clean TLB cache entry is
754 # written after the unmap. Details are in mm/rmap.c near the check for
755 # should_defer_flush. The architecture should also consider if the full flush
756 # and the refill costs are offset by the savings of sending fewer IPIs.
757 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
761 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
764 # For architectures that know their GCC __int128 support is sound
766 config ARCH_SUPPORTS_INT128
769 # For architectures that (ab)use NUMA to represent different memory regions
770 # all cpu-local but of different latencies, such as SuperH.
772 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
775 config NUMA_BALANCING
776 bool "Memory placement aware NUMA scheduler"
777 depends on ARCH_SUPPORTS_NUMA_BALANCING
778 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
779 depends on SMP && NUMA && MIGRATION
781 This option adds support for automatic NUMA aware memory/task placement.
782 The mechanism is quite primitive and is based on migrating memory when
783 it has references to the node the task is running on.
785 This system will be inactive on UMA systems.
787 config NUMA_BALANCING_DEFAULT_ENABLED
788 bool "Automatically enable NUMA aware memory/task placement"
790 depends on NUMA_BALANCING
792 If set, automatic NUMA balancing will be enabled if running on a NUMA
796 bool "Control Group support"
799 This option adds support for grouping sets of processes together, for
800 use with process control subsystems such as Cpusets, CFS, memory
801 controls or device isolation.
803 - Documentation/scheduler/sched-design-CFS.rst (CFS)
804 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
805 and resource control)
815 bool "Memory controller"
819 Provides control over the memory footprint of tasks in a cgroup.
822 bool "Swap controller"
823 depends on MEMCG && SWAP
825 Provides control over the swap space consumed by tasks in a cgroup.
827 config MEMCG_SWAP_ENABLED
828 bool "Swap controller enabled by default"
829 depends on MEMCG_SWAP
832 Memory Resource Controller Swap Extension comes with its price in
833 a bigger memory consumption. General purpose distribution kernels
834 which want to enable the feature but keep it disabled by default
835 and let the user enable it by swapaccount=1 boot command line
836 parameter should have this option unselected.
837 For those who want to have the feature enabled by default should
838 select this option (if, for some reason, they need to disable it
839 then swapaccount=0 does the trick).
843 depends on MEMCG && !SLOB
851 Generic block IO controller cgroup interface. This is the common
852 cgroup interface which should be used by various IO controlling
855 Currently, CFQ IO scheduler uses it to recognize task groups and
856 control disk bandwidth allocation (proportional time slice allocation)
857 to such task groups. It is also used by bio throttling logic in
858 block layer to implement upper limit in IO rates on a device.
860 This option only enables generic Block IO controller infrastructure.
861 One needs to also enable actual IO controlling logic/policy. For
862 enabling proportional weight division of disk bandwidth in CFQ, set
863 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
864 CONFIG_BLK_DEV_THROTTLING=y.
866 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
868 config CGROUP_WRITEBACK
870 depends on MEMCG && BLK_CGROUP
873 menuconfig CGROUP_SCHED
874 bool "CPU controller"
877 This feature lets CPU scheduler recognize task groups and control CPU
878 bandwidth allocation to such task groups. It uses cgroups to group
882 config FAIR_GROUP_SCHED
883 bool "Group scheduling for SCHED_OTHER"
884 depends on CGROUP_SCHED
888 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
889 depends on FAIR_GROUP_SCHED
892 This option allows users to define CPU bandwidth rates (limits) for
893 tasks running within the fair group scheduler. Groups with no limit
894 set are considered to be unconstrained and will run with no
896 See Documentation/scheduler/sched-bwc.rst for more information.
898 config RT_GROUP_SCHED
899 bool "Group scheduling for SCHED_RR/FIFO"
900 depends on CGROUP_SCHED
903 This feature lets you explicitly allocate real CPU bandwidth
904 to task groups. If enabled, it will also make it impossible to
905 schedule realtime tasks for non-root users until you allocate
906 realtime bandwidth for them.
907 See Documentation/scheduler/sched-rt-group.rst for more information.
911 config UCLAMP_TASK_GROUP
912 bool "Utilization clamping per group of tasks"
913 depends on CGROUP_SCHED
914 depends on UCLAMP_TASK
917 This feature enables the scheduler to track the clamped utilization
918 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
920 When this option is enabled, the user can specify a min and max
921 CPU bandwidth which is allowed for each single task in a group.
922 The max bandwidth allows to clamp the maximum frequency a task
923 can use, while the min bandwidth allows to define a minimum
924 frequency a task will always use.
926 When task group based utilization clamping is enabled, an eventually
927 specified task-specific clamp value is constrained by the cgroup
928 specified clamp value. Both minimum and maximum task clamping cannot
929 be bigger than the corresponding clamping defined at task group level.
934 bool "PIDs controller"
936 Provides enforcement of process number limits in the scope of a
937 cgroup. Any attempt to fork more processes than is allowed in the
938 cgroup will fail. PIDs are fundamentally a global resource because it
939 is fairly trivial to reach PID exhaustion before you reach even a
940 conservative kmemcg limit. As a result, it is possible to grind a
941 system to halt without being limited by other cgroup policies. The
942 PIDs controller is designed to stop this from happening.
944 It should be noted that organisational operations (such as attaching
945 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
946 since the PIDs limit only affects a process's ability to fork, not to
950 bool "RDMA controller"
952 Provides enforcement of RDMA resources defined by IB stack.
953 It is fairly easy for consumers to exhaust RDMA resources, which
954 can result into resource unavailability to other consumers.
955 RDMA controller is designed to stop this from happening.
956 Attaching processes with active RDMA resources to the cgroup
957 hierarchy is allowed even if can cross the hierarchy's limit.
959 config CGROUP_FREEZER
960 bool "Freezer controller"
962 Provides a way to freeze and unfreeze all tasks in a
965 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
966 controller includes important in-kernel memory consumers per default.
968 If you're using cgroup2, say N.
970 config CGROUP_HUGETLB
971 bool "HugeTLB controller"
972 depends on HUGETLB_PAGE
976 Provides a cgroup controller for HugeTLB pages.
977 When you enable this, you can put a per cgroup limit on HugeTLB usage.
978 The limit is enforced during page fault. Since HugeTLB doesn't
979 support page reclaim, enforcing the limit at page fault time implies
980 that, the application will get SIGBUS signal if it tries to access
981 HugeTLB pages beyond its limit. This requires the application to know
982 beforehand how much HugeTLB pages it would require for its use. The
983 control group is tracked in the third page lru pointer. This means
984 that we cannot use the controller with huge page less than 3 pages.
987 bool "Cpuset controller"
990 This option will let you create and manage CPUSETs which
991 allow dynamically partitioning a system into sets of CPUs and
992 Memory Nodes and assigning tasks to run only within those sets.
993 This is primarily useful on large SMP or NUMA systems.
997 config PROC_PID_CPUSET
998 bool "Include legacy /proc/<pid>/cpuset file"
1002 config CGROUP_DEVICE
1003 bool "Device controller"
1005 Provides a cgroup controller implementing whitelists for
1006 devices which a process in the cgroup can mknod or open.
1008 config CGROUP_CPUACCT
1009 bool "Simple CPU accounting controller"
1011 Provides a simple controller for monitoring the
1012 total CPU consumed by the tasks in a cgroup.
1015 bool "Perf controller"
1016 depends on PERF_EVENTS
1018 This option extends the perf per-cpu mode to restrict monitoring
1019 to threads which belong to the cgroup specified and run on the
1020 designated cpu. Or this can be used to have cgroup ID in samples
1021 so that it can monitor performance events among cgroups.
1026 bool "Support for eBPF programs attached to cgroups"
1027 depends on BPF_SYSCALL
1028 select SOCK_CGROUP_DATA
1030 Allow attaching eBPF programs to a cgroup using the bpf(2)
1031 syscall command BPF_PROG_ATTACH.
1033 In which context these programs are accessed depends on the type
1034 of attachment. For instance, programs that are attached using
1035 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1039 bool "Debug controller"
1041 depends on DEBUG_KERNEL
1043 This option enables a simple controller that exports
1044 debugging information about the cgroups framework. This
1045 controller is for control cgroup debugging only. Its
1046 interfaces are not stable.
1050 config SOCK_CGROUP_DATA
1056 menuconfig NAMESPACES
1057 bool "Namespaces support" if EXPERT
1058 depends on MULTIUSER
1061 Provides the way to make tasks work with different objects using
1062 the same id. For example same IPC id may refer to different objects
1063 or same user id or pid may refer to different tasks when used in
1064 different namespaces.
1069 bool "UTS namespace"
1072 In this namespace tasks see different info provided with the
1076 bool "TIME namespace"
1077 depends on GENERIC_VDSO_TIME_NS
1080 In this namespace boottime and monotonic clocks can be set.
1081 The time will keep going with the same pace.
1084 bool "IPC namespace"
1085 depends on (SYSVIPC || POSIX_MQUEUE)
1088 In this namespace tasks work with IPC ids which correspond to
1089 different IPC objects in different namespaces.
1092 bool "User namespace"
1095 This allows containers, i.e. vservers, to use user namespaces
1096 to provide different user info for different servers.
1098 When user namespaces are enabled in the kernel it is
1099 recommended that the MEMCG option also be enabled and that
1100 user-space use the memory control groups to limit the amount
1101 of memory a memory unprivileged users can use.
1106 bool "PID Namespaces"
1109 Support process id namespaces. This allows having multiple
1110 processes with the same pid as long as they are in different
1111 pid namespaces. This is a building block of containers.
1114 bool "Network namespace"
1118 Allow user space to create what appear to be multiple instances
1119 of the network stack.
1123 config CHECKPOINT_RESTORE
1124 bool "Checkpoint/restore support"
1125 select PROC_CHILDREN
1128 Enables additional kernel features in a sake of checkpoint/restore.
1129 In particular it adds auxiliary prctl codes to setup process text,
1130 data and heap segment sizes, and a few additional /proc filesystem
1133 If unsure, say N here.
1135 config SCHED_AUTOGROUP
1136 bool "Automatic process group scheduling"
1139 select FAIR_GROUP_SCHED
1141 This option optimizes the scheduler for common desktop workloads by
1142 automatically creating and populating task groups. This separation
1143 of workloads isolates aggressive CPU burners (like build jobs) from
1144 desktop applications. Task group autogeneration is currently based
1147 config SYSFS_DEPRECATED
1148 bool "Enable deprecated sysfs features to support old userspace tools"
1152 This option adds code that switches the layout of the "block" class
1153 devices, to not show up in /sys/class/block/, but only in
1156 This switch is only active when the sysfs.deprecated=1 boot option is
1157 passed or the SYSFS_DEPRECATED_V2 option is set.
1159 This option allows new kernels to run on old distributions and tools,
1160 which might get confused by /sys/class/block/. Since 2007/2008 all
1161 major distributions and tools handle this just fine.
1163 Recent distributions and userspace tools after 2009/2010 depend on
1164 the existence of /sys/class/block/, and will not work with this
1167 Only if you are using a new kernel on an old distribution, you might
1170 config SYSFS_DEPRECATED_V2
1171 bool "Enable deprecated sysfs features by default"
1174 depends on SYSFS_DEPRECATED
1176 Enable deprecated sysfs by default.
1178 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1181 Only if you are using a new kernel on an old distribution, you might
1182 need to say Y here. Even then, odds are you would not need it
1183 enabled, you can always pass the boot option if absolutely necessary.
1186 bool "Kernel->user space relay support (formerly relayfs)"
1189 This option enables support for relay interface support in
1190 certain file systems (such as debugfs).
1191 It is designed to provide an efficient mechanism for tools and
1192 facilities to relay large amounts of data from kernel space to
1197 config BLK_DEV_INITRD
1198 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1200 The initial RAM filesystem is a ramfs which is loaded by the
1201 boot loader (loadlin or lilo) and that is mounted as root
1202 before the normal boot procedure. It is typically used to
1203 load modules needed to mount the "real" root file system,
1204 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1206 If RAM disk support (BLK_DEV_RAM) is also included, this
1207 also enables initial RAM disk (initrd) support and adds
1208 15 Kbytes (more on some other architectures) to the kernel size.
1214 source "usr/Kconfig"
1219 bool "Boot config support"
1220 select BLK_DEV_INITRD
1222 Extra boot config allows system admin to pass a config file as
1223 complemental extension of kernel cmdline when booting.
1224 The boot config file must be attached at the end of initramfs
1225 with checksum, size and magic word.
1226 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1231 prompt "Compiler optimization level"
1232 default CC_OPTIMIZE_FOR_PERFORMANCE
1234 config CC_OPTIMIZE_FOR_PERFORMANCE
1235 bool "Optimize for performance (-O2)"
1237 This is the default optimization level for the kernel, building
1238 with the "-O2" compiler flag for best performance and most
1239 helpful compile-time warnings.
1241 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1242 bool "Optimize more for performance (-O3)"
1245 Choosing this option will pass "-O3" to your compiler to optimize
1246 the kernel yet more for performance.
1248 config CC_OPTIMIZE_FOR_SIZE
1249 bool "Optimize for size (-Os)"
1251 Choosing this option will pass "-Os" to your compiler resulting
1252 in a smaller kernel.
1256 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1259 This requires that the arch annotates or otherwise protects
1260 its external entry points from being discarded. Linker scripts
1261 must also merge .text.*, .data.*, and .bss.* correctly into
1262 output sections. Care must be taken not to pull in unrelated
1263 sections (e.g., '.text.init'). Typically '.' in section names
1264 is used to distinguish them from label names / C identifiers.
1266 config LD_DEAD_CODE_DATA_ELIMINATION
1267 bool "Dead code and data elimination (EXPERIMENTAL)"
1268 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1270 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1271 depends on $(cc-option,-ffunction-sections -fdata-sections)
1272 depends on $(ld-option,--gc-sections)
1274 Enable this if you want to do dead code and data elimination with
1275 the linker by compiling with -ffunction-sections -fdata-sections,
1276 and linking with --gc-sections.
1278 This can reduce on disk and in-memory size of the kernel
1279 code and static data, particularly for small configs and
1280 on small systems. This has the possibility of introducing
1281 silently broken kernel if the required annotations are not
1282 present. This option is not well tested yet, so use at your
1291 config SYSCTL_EXCEPTION_TRACE
1294 Enable support for /proc/sys/debug/exception-trace.
1296 config SYSCTL_ARCH_UNALIGN_NO_WARN
1299 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1300 Allows arch to define/use @no_unaligned_warning to possibly warn
1301 about unaligned access emulation going on under the hood.
1303 config SYSCTL_ARCH_UNALIGN_ALLOW
1306 Enable support for /proc/sys/kernel/unaligned-trap
1307 Allows arches to define/use @unaligned_enabled to runtime toggle
1308 the unaligned access emulation.
1309 see arch/parisc/kernel/unaligned.c for reference
1311 config HAVE_PCSPKR_PLATFORM
1314 # interpreter that classic socket filters depend on
1319 bool "Configure standard kernel features (expert users)"
1320 # Unhide debug options, to make the on-by-default options visible
1323 This option allows certain base kernel options and settings
1324 to be disabled or tweaked. This is for specialized
1325 environments which can tolerate a "non-standard" kernel.
1326 Only use this if you really know what you are doing.
1329 bool "Enable 16-bit UID system calls" if EXPERT
1330 depends on HAVE_UID16 && MULTIUSER
1333 This enables the legacy 16-bit UID syscall wrappers.
1336 bool "Multiple users, groups and capabilities support" if EXPERT
1339 This option enables support for non-root users, groups and
1342 If you say N here, all processes will run with UID 0, GID 0, and all
1343 possible capabilities. Saying N here also compiles out support for
1344 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1347 If unsure, say Y here.
1349 config SGETMASK_SYSCALL
1350 bool "sgetmask/ssetmask syscalls support" if EXPERT
1351 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1353 sys_sgetmask and sys_ssetmask are obsolete system calls
1354 no longer supported in libc but still enabled by default in some
1357 If unsure, leave the default option here.
1359 config SYSFS_SYSCALL
1360 bool "Sysfs syscall support" if EXPERT
1363 sys_sysfs is an obsolete system call no longer supported in libc.
1364 Note that disabling this option is more secure but might break
1365 compatibility with some systems.
1367 If unsure say Y here.
1370 bool "open by fhandle syscalls" if EXPERT
1374 If you say Y here, a user level program will be able to map
1375 file names to handle and then later use the handle for
1376 different file system operations. This is useful in implementing
1377 userspace file servers, which now track files using handles instead
1378 of names. The handle would remain the same even if file names
1379 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1383 bool "Posix Clocks & timers" if EXPERT
1386 This includes native support for POSIX timers to the kernel.
1387 Some embedded systems have no use for them and therefore they
1388 can be configured out to reduce the size of the kernel image.
1390 When this option is disabled, the following syscalls won't be
1391 available: timer_create, timer_gettime: timer_getoverrun,
1392 timer_settime, timer_delete, clock_adjtime, getitimer,
1393 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1394 clock_getres and clock_nanosleep syscalls will be limited to
1395 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1401 bool "Enable support for printk" if EXPERT
1404 This option enables normal printk support. Removing it
1405 eliminates most of the message strings from the kernel image
1406 and makes the kernel more or less silent. As this makes it
1407 very difficult to diagnose system problems, saying N here is
1408 strongly discouraged.
1416 bool "BUG() support" if EXPERT
1419 Disabling this option eliminates support for BUG and WARN, reducing
1420 the size of your kernel image and potentially quietly ignoring
1421 numerous fatal conditions. You should only consider disabling this
1422 option for embedded systems with no facilities for reporting errors.
1428 bool "Enable ELF core dumps" if EXPERT
1430 Enable support for generating core dumps. Disabling saves about 4k.
1433 config PCSPKR_PLATFORM
1434 bool "Enable PC-Speaker support" if EXPERT
1435 depends on HAVE_PCSPKR_PLATFORM
1439 This option allows to disable the internal PC-Speaker
1440 support, saving some memory.
1444 bool "Enable full-sized data structures for core" if EXPERT
1446 Disabling this option reduces the size of miscellaneous core
1447 kernel data structures. This saves memory on small machines,
1448 but may reduce performance.
1451 bool "Enable futex support" if EXPERT
1455 Disabling this option will cause the kernel to be built without
1456 support for "fast userspace mutexes". The resulting kernel may not
1457 run glibc-based applications correctly.
1461 depends on FUTEX && RT_MUTEXES
1464 config HAVE_FUTEX_CMPXCHG
1468 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1469 is implemented and always working. This removes a couple of runtime
1473 bool "Enable eventpoll support" if EXPERT
1476 Disabling this option will cause the kernel to be built without
1477 support for epoll family of system calls.
1480 bool "Enable signalfd() system call" if EXPERT
1483 Enable the signalfd() system call that allows to receive signals
1484 on a file descriptor.
1489 bool "Enable timerfd() system call" if EXPERT
1492 Enable the timerfd() system call that allows to receive timer
1493 events on a file descriptor.
1498 bool "Enable eventfd() system call" if EXPERT
1501 Enable the eventfd() system call that allows to receive both
1502 kernel notification (ie. KAIO) or userspace notifications.
1507 bool "Use full shmem filesystem" if EXPERT
1511 The shmem is an internal filesystem used to manage shared memory.
1512 It is backed by swap and manages resource limits. It is also exported
1513 to userspace as tmpfs if TMPFS is enabled. Disabling this
1514 option replaces shmem and tmpfs with the much simpler ramfs code,
1515 which may be appropriate on small systems without swap.
1518 bool "Enable AIO support" if EXPERT
1521 This option enables POSIX asynchronous I/O which may by used
1522 by some high performance threaded applications. Disabling
1523 this option saves about 7k.
1526 bool "Enable IO uring support" if EXPERT
1530 This option enables support for the io_uring interface, enabling
1531 applications to submit and complete IO through submission and
1532 completion rings that are shared between the kernel and application.
1534 config ADVISE_SYSCALLS
1535 bool "Enable madvise/fadvise syscalls" if EXPERT
1538 This option enables the madvise and fadvise syscalls, used by
1539 applications to advise the kernel about their future memory or file
1540 usage, improving performance. If building an embedded system where no
1541 applications use these syscalls, you can disable this option to save
1544 config HAVE_ARCH_USERFAULTFD_WP
1547 Arch has userfaultfd write protection support
1550 bool "Enable membarrier() system call" if EXPERT
1553 Enable the membarrier() system call that allows issuing memory
1554 barriers across all running threads, which can be used to distribute
1555 the cost of user-space memory barriers asymmetrically by transforming
1556 pairs of memory barriers into pairs consisting of membarrier() and a
1562 bool "Load all symbols for debugging/ksymoops" if EXPERT
1565 Say Y here to let the kernel print out symbolic crash information and
1566 symbolic stack backtraces. This increases the size of the kernel
1567 somewhat, as all symbols have to be loaded into the kernel image.
1570 bool "Include all symbols in kallsyms"
1571 depends on DEBUG_KERNEL && KALLSYMS
1573 Normally kallsyms only contains the symbols of functions for nicer
1574 OOPS messages and backtraces (i.e., symbols from the text and inittext
1575 sections). This is sufficient for most cases. And only in very rare
1576 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1577 names of variables from the data sections, etc).
1579 This option makes sure that all symbols are loaded into the kernel
1580 image (i.e., symbols from all sections) in cost of increased kernel
1581 size (depending on the kernel configuration, it may be 300KiB or
1582 something like this).
1584 Say N unless you really need all symbols.
1586 config KALLSYMS_ABSOLUTE_PERCPU
1589 default X86_64 && SMP
1591 config KALLSYMS_BASE_RELATIVE
1596 Instead of emitting them as absolute values in the native word size,
1597 emit the symbol references in the kallsyms table as 32-bit entries,
1598 each containing a relative value in the range [base, base + U32_MAX]
1599 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1600 an absolute value in the range [0, S32_MAX] or a relative value in the
1601 range [base, base + S32_MAX], where base is the lowest relative symbol
1602 address encountered in the image.
1604 On 64-bit builds, this reduces the size of the address table by 50%,
1605 but more importantly, it results in entries whose values are build
1606 time constants, and no relocation pass is required at runtime to fix
1607 up the entries based on the runtime load address of the kernel.
1609 # end of the "standard kernel features (expert users)" menu
1611 # syscall, maps, verifier
1614 bool "LSM Instrumentation with BPF"
1615 depends on BPF_EVENTS
1616 depends on BPF_SYSCALL
1620 Enables instrumentation of the security hooks with eBPF programs for
1621 implementing dynamic MAC and Audit Policies.
1623 If you are unsure how to answer this question, answer N.
1626 bool "Enable bpf() system call"
1631 Enable the bpf() system call that allows to manipulate eBPF
1632 programs and maps via file descriptors.
1634 config ARCH_WANT_DEFAULT_BPF_JIT
1637 config BPF_JIT_ALWAYS_ON
1638 bool "Permanently enable BPF JIT and remove BPF interpreter"
1639 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1641 Enables BPF JIT and removes BPF interpreter to avoid
1642 speculative execution of BPF instructions by the interpreter
1644 config BPF_JIT_DEFAULT_ON
1645 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1646 depends on HAVE_EBPF_JIT && BPF_JIT
1649 bool "Enable userfaultfd() system call"
1652 Enable the userfaultfd() system call that allows to intercept and
1653 handle page faults in userland.
1655 config ARCH_HAS_MEMBARRIER_CALLBACKS
1658 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1662 bool "Enable rseq() system call" if EXPERT
1664 depends on HAVE_RSEQ
1667 Enable the restartable sequences system call. It provides a
1668 user-space cache for the current CPU number value, which
1669 speeds up getting the current CPU number from user-space,
1670 as well as an ABI to speed up user-space operations on
1677 bool "Enabled debugging of rseq() system call" if EXPERT
1678 depends on RSEQ && DEBUG_KERNEL
1680 Enable extra debugging checks for the rseq system call.
1685 bool "Embedded system"
1686 option allnoconfig_y
1689 This option should be enabled if compiling the kernel for
1690 an embedded system so certain expert options are available
1693 config HAVE_PERF_EVENTS
1696 See tools/perf/design.txt for details.
1698 config PERF_USE_VMALLOC
1701 See tools/perf/design.txt for details
1704 bool "PC/104 support" if EXPERT
1706 Expose PC/104 form factor device drivers and options available for
1707 selection and configuration. Enable this option if your target
1708 machine has a PC/104 bus.
1710 menu "Kernel Performance Events And Counters"
1713 bool "Kernel performance events and counters"
1714 default y if PROFILING
1715 depends on HAVE_PERF_EVENTS
1719 Enable kernel support for various performance events provided
1720 by software and hardware.
1722 Software events are supported either built-in or via the
1723 use of generic tracepoints.
1725 Most modern CPUs support performance events via performance
1726 counter registers. These registers count the number of certain
1727 types of hw events: such as instructions executed, cachemisses
1728 suffered, or branches mis-predicted - without slowing down the
1729 kernel or applications. These registers can also trigger interrupts
1730 when a threshold number of events have passed - and can thus be
1731 used to profile the code that runs on that CPU.
1733 The Linux Performance Event subsystem provides an abstraction of
1734 these software and hardware event capabilities, available via a
1735 system call and used by the "perf" utility in tools/perf/. It
1736 provides per task and per CPU counters, and it provides event
1737 capabilities on top of those.
1741 config DEBUG_PERF_USE_VMALLOC
1743 bool "Debug: use vmalloc to back perf mmap() buffers"
1744 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1745 select PERF_USE_VMALLOC
1747 Use vmalloc memory to back perf mmap() buffers.
1749 Mostly useful for debugging the vmalloc code on platforms
1750 that don't require it.
1756 config VM_EVENT_COUNTERS
1758 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1760 VM event counters are needed for event counts to be shown.
1761 This option allows the disabling of the VM event counters
1762 on EXPERT systems. /proc/vmstat will only show page counts
1763 if VM event counters are disabled.
1767 bool "Enable SLUB debugging support" if EXPERT
1768 depends on SLUB && SYSFS
1770 SLUB has extensive debug support features. Disabling these can
1771 result in significant savings in code size. This also disables
1772 SLUB sysfs support. /sys/slab will not exist and there will be
1773 no support for cache validation etc.
1775 config SLUB_MEMCG_SYSFS_ON
1777 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1778 depends on SLUB && SYSFS && MEMCG
1780 SLUB creates a directory under /sys/kernel/slab for each
1781 allocation cache to host info and debug files. If memory
1782 cgroup is enabled, each cache can have per memory cgroup
1783 caches. SLUB can create the same sysfs directories for these
1784 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1785 to a very high number of debug files being created. This is
1786 controlled by slub_memcg_sysfs boot parameter and this
1787 config option determines the parameter's default value.
1790 bool "Disable heap randomization"
1793 Randomizing heap placement makes heap exploits harder, but it
1794 also breaks ancient binaries (including anything libc5 based).
1795 This option changes the bootup default to heap randomization
1796 disabled, and can be overridden at runtime by setting
1797 /proc/sys/kernel/randomize_va_space to 2.
1799 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1802 prompt "Choose SLAB allocator"
1805 This option allows to select a slab allocator.
1809 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1811 The regular slab allocator that is established and known to work
1812 well in all environments. It organizes cache hot objects in
1813 per cpu and per node queues.
1816 bool "SLUB (Unqueued Allocator)"
1817 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1819 SLUB is a slab allocator that minimizes cache line usage
1820 instead of managing queues of cached objects (SLAB approach).
1821 Per cpu caching is realized using slabs of objects instead
1822 of queues of objects. SLUB can use memory efficiently
1823 and has enhanced diagnostics. SLUB is the default choice for
1828 bool "SLOB (Simple Allocator)"
1830 SLOB replaces the stock allocator with a drastically simpler
1831 allocator. SLOB is generally more space efficient but
1832 does not perform as well on large systems.
1836 config SLAB_MERGE_DEFAULT
1837 bool "Allow slab caches to be merged"
1840 For reduced kernel memory fragmentation, slab caches can be
1841 merged when they share the same size and other characteristics.
1842 This carries a risk of kernel heap overflows being able to
1843 overwrite objects from merged caches (and more easily control
1844 cache layout), which makes such heap attacks easier to exploit
1845 by attackers. By keeping caches unmerged, these kinds of exploits
1846 can usually only damage objects in the same cache. To disable
1847 merging at runtime, "slab_nomerge" can be passed on the kernel
1850 config SLAB_FREELIST_RANDOM
1852 depends on SLAB || SLUB
1853 bool "SLAB freelist randomization"
1855 Randomizes the freelist order used on creating new pages. This
1856 security feature reduces the predictability of the kernel slab
1857 allocator against heap overflows.
1859 config SLAB_FREELIST_HARDENED
1860 bool "Harden slab freelist metadata"
1863 Many kernel heap attacks try to target slab cache metadata and
1864 other infrastructure. This options makes minor performance
1865 sacrifices to harden the kernel slab allocator against common
1866 freelist exploit methods.
1868 config SHUFFLE_PAGE_ALLOCATOR
1869 bool "Page allocator randomization"
1870 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1872 Randomization of the page allocator improves the average
1873 utilization of a direct-mapped memory-side-cache. See section
1874 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1875 6.2a specification for an example of how a platform advertises
1876 the presence of a memory-side-cache. There are also incidental
1877 security benefits as it reduces the predictability of page
1878 allocations to compliment SLAB_FREELIST_RANDOM, but the
1879 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1880 10th order of pages is selected based on cache utilization
1883 While the randomization improves cache utilization it may
1884 negatively impact workloads on platforms without a cache. For
1885 this reason, by default, the randomization is enabled only
1886 after runtime detection of a direct-mapped memory-side-cache.
1887 Otherwise, the randomization may be force enabled with the
1888 'page_alloc.shuffle' kernel command line parameter.
1892 config SLUB_CPU_PARTIAL
1894 depends on SLUB && SMP
1895 bool "SLUB per cpu partial cache"
1897 Per cpu partial caches accelerate objects allocation and freeing
1898 that is local to a processor at the price of more indeterminism
1899 in the latency of the free. On overflow these caches will be cleared
1900 which requires the taking of locks that may cause latency spikes.
1901 Typically one would choose no for a realtime system.
1903 config MMAP_ALLOW_UNINITIALIZED
1904 bool "Allow mmapped anonymous memory to be uninitialized"
1905 depends on EXPERT && !MMU
1908 Normally, and according to the Linux spec, anonymous memory obtained
1909 from mmap() has its contents cleared before it is passed to
1910 userspace. Enabling this config option allows you to request that
1911 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1912 providing a huge performance boost. If this option is not enabled,
1913 then the flag will be ignored.
1915 This is taken advantage of by uClibc's malloc(), and also by
1916 ELF-FDPIC binfmt's brk and stack allocator.
1918 Because of the obvious security issues, this option should only be
1919 enabled on embedded devices where you control what is run in
1920 userspace. Since that isn't generally a problem on no-MMU systems,
1921 it is normally safe to say Y here.
1923 See Documentation/nommu-mmap.txt for more information.
1925 config SYSTEM_DATA_VERIFICATION
1927 select SYSTEM_TRUSTED_KEYRING
1931 select ASYMMETRIC_KEY_TYPE
1932 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1935 select X509_CERTIFICATE_PARSER
1936 select PKCS7_MESSAGE_PARSER
1938 Provide PKCS#7 message verification using the contents of the system
1939 trusted keyring to provide public keys. This then can be used for
1940 module verification, kexec image verification and firmware blob
1944 bool "Profiling support"
1946 Say Y here to enable the extended profiling support mechanisms used
1947 by profilers such as OProfile.
1950 # Place an empty function call at each tracepoint site. Can be
1951 # dynamically changed for a probe function.
1956 endmenu # General setup
1958 source "arch/Kconfig"
1965 default 0 if BASE_FULL
1966 default 1 if !BASE_FULL
1968 config MODULE_SIG_FORMAT
1970 select SYSTEM_DATA_VERIFICATION
1973 bool "Enable loadable module support"
1976 Kernel modules are small pieces of compiled code which can
1977 be inserted in the running kernel, rather than being
1978 permanently built into the kernel. You use the "modprobe"
1979 tool to add (and sometimes remove) them. If you say Y here,
1980 many parts of the kernel can be built as modules (by
1981 answering M instead of Y where indicated): this is most
1982 useful for infrequently used options which are not required
1983 for booting. For more information, see the man pages for
1984 modprobe, lsmod, modinfo, insmod and rmmod.
1986 If you say Y here, you will need to run "make
1987 modules_install" to put the modules under /lib/modules/
1988 where modprobe can find them (you may need to be root to do
1995 config MODULE_FORCE_LOAD
1996 bool "Forced module loading"
1999 Allow loading of modules without version information (ie. modprobe
2000 --force). Forced module loading sets the 'F' (forced) taint flag and
2001 is usually a really bad idea.
2003 config MODULE_UNLOAD
2004 bool "Module unloading"
2006 Without this option you will not be able to unload any
2007 modules (note that some modules may not be unloadable
2008 anyway), which makes your kernel smaller, faster
2009 and simpler. If unsure, say Y.
2011 config MODULE_FORCE_UNLOAD
2012 bool "Forced module unloading"
2013 depends on MODULE_UNLOAD
2015 This option allows you to force a module to unload, even if the
2016 kernel believes it is unsafe: the kernel will remove the module
2017 without waiting for anyone to stop using it (using the -f option to
2018 rmmod). This is mainly for kernel developers and desperate users.
2022 bool "Module versioning support"
2024 Usually, you have to use modules compiled with your kernel.
2025 Saying Y here makes it sometimes possible to use modules
2026 compiled for different kernels, by adding enough information
2027 to the modules to (hopefully) spot any changes which would
2028 make them incompatible with the kernel you are running. If
2031 config ASM_MODVERSIONS
2033 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2035 This enables module versioning for exported symbols also from
2036 assembly. This can be enabled only when the target architecture
2039 config MODULE_REL_CRCS
2041 depends on MODVERSIONS
2043 config MODULE_SRCVERSION_ALL
2044 bool "Source checksum for all modules"
2046 Modules which contain a MODULE_VERSION get an extra "srcversion"
2047 field inserted into their modinfo section, which contains a
2048 sum of the source files which made it. This helps maintainers
2049 see exactly which source was used to build a module (since
2050 others sometimes change the module source without updating
2051 the version). With this option, such a "srcversion" field
2052 will be created for all modules. If unsure, say N.
2055 bool "Module signature verification"
2056 select MODULE_SIG_FORMAT
2058 Check modules for valid signatures upon load: the signature
2059 is simply appended to the module. For more information see
2060 <file:Documentation/admin-guide/module-signing.rst>.
2062 Note that this option adds the OpenSSL development packages as a
2063 kernel build dependency so that the signing tool can use its crypto
2066 You should enable this option if you wish to use either
2067 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2068 another LSM - otherwise unsigned modules will be loadable regardless
2069 of the lockdown policy.
2071 !!!WARNING!!! If you enable this option, you MUST make sure that the
2072 module DOES NOT get stripped after being signed. This includes the
2073 debuginfo strip done by some packagers (such as rpmbuild) and
2074 inclusion into an initramfs that wants the module size reduced.
2076 config MODULE_SIG_FORCE
2077 bool "Require modules to be validly signed"
2078 depends on MODULE_SIG
2080 Reject unsigned modules or signed modules for which we don't have a
2081 key. Without this, such modules will simply taint the kernel.
2083 config MODULE_SIG_ALL
2084 bool "Automatically sign all modules"
2086 depends on MODULE_SIG
2088 Sign all modules during make modules_install. Without this option,
2089 modules must be signed manually, using the scripts/sign-file tool.
2091 comment "Do not forget to sign required modules with scripts/sign-file"
2092 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2095 prompt "Which hash algorithm should modules be signed with?"
2096 depends on MODULE_SIG
2098 This determines which sort of hashing algorithm will be used during
2099 signature generation. This algorithm _must_ be built into the kernel
2100 directly so that signature verification can take place. It is not
2101 possible to load a signed module containing the algorithm to check
2102 the signature on that module.
2104 config MODULE_SIG_SHA1
2105 bool "Sign modules with SHA-1"
2108 config MODULE_SIG_SHA224
2109 bool "Sign modules with SHA-224"
2110 select CRYPTO_SHA256
2112 config MODULE_SIG_SHA256
2113 bool "Sign modules with SHA-256"
2114 select CRYPTO_SHA256
2116 config MODULE_SIG_SHA384
2117 bool "Sign modules with SHA-384"
2118 select CRYPTO_SHA512
2120 config MODULE_SIG_SHA512
2121 bool "Sign modules with SHA-512"
2122 select CRYPTO_SHA512
2126 config MODULE_SIG_HASH
2128 depends on MODULE_SIG
2129 default "sha1" if MODULE_SIG_SHA1
2130 default "sha224" if MODULE_SIG_SHA224
2131 default "sha256" if MODULE_SIG_SHA256
2132 default "sha384" if MODULE_SIG_SHA384
2133 default "sha512" if MODULE_SIG_SHA512
2135 config MODULE_COMPRESS
2136 bool "Compress modules on installation"
2139 Compresses kernel modules when 'make modules_install' is run; gzip or
2140 xz depending on "Compression algorithm" below.
2142 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2144 Out-of-tree kernel modules installed using Kbuild will also be
2145 compressed upon installation.
2147 Note: for modules inside an initrd or initramfs, it's more efficient
2148 to compress the whole initrd or initramfs instead.
2150 Note: This is fully compatible with signed modules.
2155 prompt "Compression algorithm"
2156 depends on MODULE_COMPRESS
2157 default MODULE_COMPRESS_GZIP
2159 This determines which sort of compression will be used during
2160 'make modules_install'.
2162 GZIP (default) and XZ are supported.
2164 config MODULE_COMPRESS_GZIP
2167 config MODULE_COMPRESS_XZ
2172 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2173 bool "Allow loading of modules with missing namespace imports"
2175 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2176 a namespace. A module that makes use of a symbol exported with such a
2177 namespace is required to import the namespace via MODULE_IMPORT_NS().
2178 There is no technical reason to enforce correct namespace imports,
2179 but it creates consistency between symbols defining namespaces and
2180 users importing namespaces they make use of. This option relaxes this
2181 requirement and lifts the enforcement when loading a module.
2185 config UNUSED_SYMBOLS
2186 bool "Enable unused/obsolete exported symbols"
2189 Unused but exported symbols make the kernel needlessly bigger. For
2190 that reason most of these unused exports will soon be removed. This
2191 option is provided temporarily to provide a transition period in case
2192 some external kernel module needs one of these symbols anyway. If you
2193 encounter such a case in your module, consider if you are actually
2194 using the right API. (rationale: since nobody in the kernel is using
2195 this in a module, there is a pretty good chance it's actually the
2196 wrong interface to use). If you really need the symbol, please send a
2197 mail to the linux kernel mailing list mentioning the symbol and why
2198 you really need it, and what the merge plan to the mainline kernel for
2201 config TRIM_UNUSED_KSYMS
2202 bool "Trim unused exported kernel symbols"
2203 depends on !UNUSED_SYMBOLS
2205 The kernel and some modules make many symbols available for
2206 other modules to use via EXPORT_SYMBOL() and variants. Depending
2207 on the set of modules being selected in your kernel configuration,
2208 many of those exported symbols might never be used.
2210 This option allows for unused exported symbols to be dropped from
2211 the build. In turn, this provides the compiler more opportunities
2212 (especially when using LTO) for optimizing the code and reducing
2213 binary size. This might have some security advantages as well.
2215 If unsure, or if you need to build out-of-tree modules, say N.
2217 config UNUSED_KSYMS_WHITELIST
2218 string "Whitelist of symbols to keep in ksymtab"
2219 depends on TRIM_UNUSED_KSYMS
2221 By default, all unused exported symbols will be un-exported from the
2222 build when TRIM_UNUSED_KSYMS is selected.
2224 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2225 exported at all times, even in absence of in-tree users. The value to
2226 set here is the path to a text file containing the list of symbols,
2227 one per line. The path can be absolute, or relative to the kernel
2232 config MODULES_TREE_LOOKUP
2234 depends on PERF_EVENTS || TRACING
2236 config INIT_ALL_POSSIBLE
2239 Back when each arch used to define their own cpu_online_mask and
2240 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2241 with all 1s, and others with all 0s. When they were centralised,
2242 it was better to provide this option than to break all the archs
2243 and have several arch maintainers pursuing me down dark alleys.
2245 source "block/Kconfig"
2247 config PREEMPT_NOTIFIERS
2257 Build a simple ASN.1 grammar compiler that produces a bytecode output
2258 that can be interpreted by the ASN.1 stream decoder and used to
2259 inform it as to what tags are to be expected in a stream and what
2260 functions to call on what tags.
2262 source "kernel/Kconfig.locks"
2264 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2267 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2270 # It may be useful for an architecture to override the definitions of the
2271 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2272 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2273 # different calling convention for syscalls. They can also override the
2274 # macros for not-implemented syscalls in kernel/sys_ni.c and
2275 # kernel/time/posix-stubs.c. All these overrides need to be available in
2276 # <asm/syscall_wrapper.h>.
2277 config ARCH_HAS_SYSCALL_WRAPPER
OSDN Git Service
