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)
42 config CC_HAS_WARN_MAYBE_UNINITIALIZED
43 def_bool $(cc-option,-Wmaybe-uninitialized)
45 GCC >= 4.7 supports this option.
47 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
49 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
50 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
52 GCC's -Wmaybe-uninitialized is not reliable by definition.
53 Lots of false positive warnings are produced in some cases.
55 If this option is enabled, -Wno-maybe-uninitialzed is passed
56 to the compiler to suppress maybe-uninitialized warnings.
65 config BUILDTIME_TABLE_SORT
68 config THREAD_INFO_IN_TASK
71 Select this to move thread_info off the stack into task_struct. To
72 make this work, an arch will need to remove all thread_info fields
73 except flags and fix any runtime bugs.
75 One subtle change that will be needed is to use try_get_task_stack()
76 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
85 depends on BROKEN || !SMP
88 config INIT_ENV_ARG_LIMIT
93 Maximum of each of the number of arguments and environment
94 variables passed to init from the kernel command line.
97 bool "Compile also drivers which will not load"
101 Some drivers can be compiled on a different platform than they are
102 intended to be run on. Despite they cannot be loaded there (or even
103 when they load they cannot be used due to missing HW support),
104 developers still, opposing to distributors, might want to build such
105 drivers to compile-test them.
107 If you are a developer and want to build everything available, say Y
108 here. If you are a user/distributor, say N here to exclude useless
109 drivers to be distributed.
111 config UAPI_HEADER_TEST
112 bool "Compile test UAPI headers"
113 depends on HEADERS_INSTALL && CC_CAN_LINK
115 Compile test headers exported to user-space to ensure they are
116 self-contained, i.e. compilable as standalone units.
118 If you are a developer or tester and want to ensure the exported
119 headers are self-contained, say Y here. Otherwise, choose N.
122 string "Local version - append to kernel release"
124 Append an extra string to the end of your kernel version.
125 This will show up when you type uname, for example.
126 The string you set here will be appended after the contents of
127 any files with a filename matching localversion* in your
128 object and source tree, in that order. Your total string can
129 be a maximum of 64 characters.
131 config LOCALVERSION_AUTO
132 bool "Automatically append version information to the version string"
134 depends on !COMPILE_TEST
136 This will try to automatically determine if the current tree is a
137 release tree by looking for git tags that belong to the current
138 top of tree revision.
140 A string of the format -gxxxxxxxx will be added to the localversion
141 if a git-based tree is found. The string generated by this will be
142 appended after any matching localversion* files, and after the value
143 set in CONFIG_LOCALVERSION.
145 (The actual string used here is the first eight characters produced
146 by running the command:
148 $ git rev-parse --verify HEAD
150 which is done within the script "scripts/setlocalversion".)
153 string "Build ID Salt"
156 The build ID is used to link binaries and their debug info. Setting
157 this option will use the value in the calculation of the build id.
158 This is mostly useful for distributions which want to ensure the
159 build is unique between builds. It's safe to leave the default.
161 config HAVE_KERNEL_GZIP
164 config HAVE_KERNEL_BZIP2
167 config HAVE_KERNEL_LZMA
170 config HAVE_KERNEL_XZ
173 config HAVE_KERNEL_LZO
176 config HAVE_KERNEL_LZ4
179 config HAVE_KERNEL_UNCOMPRESSED
183 prompt "Kernel compression mode"
185 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
187 The linux kernel is a kind of self-extracting executable.
188 Several compression algorithms are available, which differ
189 in efficiency, compression and decompression speed.
190 Compression speed is only relevant when building a kernel.
191 Decompression speed is relevant at each boot.
193 If you have any problems with bzip2 or lzma compressed
194 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
195 version of this functionality (bzip2 only), for 2.4, was
196 supplied by Christian Ludwig)
198 High compression options are mostly useful for users, who
199 are low on disk space (embedded systems), but for whom ram
202 If in doubt, select 'gzip'
206 depends on HAVE_KERNEL_GZIP
208 The old and tried gzip compression. It provides a good balance
209 between compression ratio and decompression speed.
213 depends on HAVE_KERNEL_BZIP2
215 Its compression ratio and speed is intermediate.
216 Decompression speed is slowest among the choices. The kernel
217 size is about 10% smaller with bzip2, in comparison to gzip.
218 Bzip2 uses a large amount of memory. For modern kernels you
219 will need at least 8MB RAM or more for booting.
223 depends on HAVE_KERNEL_LZMA
225 This compression algorithm's ratio is best. Decompression speed
226 is between gzip and bzip2. Compression is slowest.
227 The kernel size is about 33% smaller with LZMA in comparison to gzip.
231 depends on HAVE_KERNEL_XZ
233 XZ uses the LZMA2 algorithm and instruction set specific
234 BCJ filters which can improve compression ratio of executable
235 code. The size of the kernel is about 30% smaller with XZ in
236 comparison to gzip. On architectures for which there is a BCJ
237 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
238 will create a few percent smaller kernel than plain LZMA.
240 The speed is about the same as with LZMA: The decompression
241 speed of XZ is better than that of bzip2 but worse than gzip
242 and LZO. Compression is slow.
246 depends on HAVE_KERNEL_LZO
248 Its compression ratio is the poorest among the choices. The kernel
249 size is about 10% bigger than gzip; however its speed
250 (both compression and decompression) is the fastest.
254 depends on HAVE_KERNEL_LZ4
256 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
257 A preliminary version of LZ4 de/compression tool is available at
258 <https://code.google.com/p/lz4/>.
260 Its compression ratio is worse than LZO. The size of the kernel
261 is about 8% bigger than LZO. But the decompression speed is
264 config KERNEL_UNCOMPRESSED
266 depends on HAVE_KERNEL_UNCOMPRESSED
268 Produce uncompressed kernel image. This option is usually not what
269 you want. It is useful for debugging the kernel in slow simulation
270 environments, where decompressing and moving the kernel is awfully
271 slow. This option allows early boot code to skip the decompressor
272 and jump right at uncompressed kernel image.
276 config DEFAULT_HOSTNAME
277 string "Default hostname"
280 This option determines the default system hostname before userspace
281 calls sethostname(2). The kernel traditionally uses "(none)" here,
282 but you may wish to use a different default here to make a minimal
283 system more usable with less configuration.
286 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
287 # add proper SWAP support to them, in which case this can be remove.
293 bool "Support for paging of anonymous memory (swap)"
294 depends on MMU && BLOCK && !ARCH_NO_SWAP
297 This option allows you to choose whether you want to have support
298 for so called swap devices or swap files in your kernel that are
299 used to provide more virtual memory than the actual RAM present
300 in your computer. If unsure say Y.
305 Inter Process Communication is a suite of library functions and
306 system calls which let processes (running programs) synchronize and
307 exchange information. It is generally considered to be a good thing,
308 and some programs won't run unless you say Y here. In particular, if
309 you want to run the DOS emulator dosemu under Linux (read the
310 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
311 you'll need to say Y here.
313 You can find documentation about IPC with "info ipc" and also in
314 section 6.4 of the Linux Programmer's Guide, available from
315 <http://www.tldp.org/guides.html>.
317 config SYSVIPC_SYSCTL
324 bool "POSIX Message Queues"
327 POSIX variant of message queues is a part of IPC. In POSIX message
328 queues every message has a priority which decides about succession
329 of receiving it by a process. If you want to compile and run
330 programs written e.g. for Solaris with use of its POSIX message
331 queues (functions mq_*) say Y here.
333 POSIX message queues are visible as a filesystem called 'mqueue'
334 and can be mounted somewhere if you want to do filesystem
335 operations on message queues.
339 config POSIX_MQUEUE_SYSCTL
341 depends on POSIX_MQUEUE
345 config CROSS_MEMORY_ATTACH
346 bool "Enable process_vm_readv/writev syscalls"
350 Enabling this option adds the system calls process_vm_readv and
351 process_vm_writev which allow a process with the correct privileges
352 to directly read from or write to another process' address space.
353 See the man page for more details.
356 bool "uselib syscall"
357 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
359 This option enables the uselib syscall, a system call used in the
360 dynamic linker from libc5 and earlier. glibc does not use this
361 system call. If you intend to run programs built on libc5 or
362 earlier, you may need to enable this syscall. Current systems
363 running glibc can safely disable this.
366 bool "Auditing support"
369 Enable auditing infrastructure that can be used with another
370 kernel subsystem, such as SELinux (which requires this for
371 logging of avc messages output). System call auditing is included
372 on architectures which support it.
374 config HAVE_ARCH_AUDITSYSCALL
379 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
382 source "kernel/irq/Kconfig"
383 source "kernel/time/Kconfig"
384 source "kernel/Kconfig.preempt"
386 menu "CPU/Task time and stats accounting"
388 config VIRT_CPU_ACCOUNTING
392 prompt "Cputime accounting"
393 default TICK_CPU_ACCOUNTING if !PPC64
394 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
396 # Kind of a stub config for the pure tick based cputime accounting
397 config TICK_CPU_ACCOUNTING
398 bool "Simple tick based cputime accounting"
399 depends on !S390 && !NO_HZ_FULL
401 This is the basic tick based cputime accounting that maintains
402 statistics about user, system and idle time spent on per jiffies
407 config VIRT_CPU_ACCOUNTING_NATIVE
408 bool "Deterministic task and CPU time accounting"
409 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
410 select VIRT_CPU_ACCOUNTING
412 Select this option to enable more accurate task and CPU time
413 accounting. This is done by reading a CPU counter on each
414 kernel entry and exit and on transitions within the kernel
415 between system, softirq and hardirq state, so there is a
416 small performance impact. In the case of s390 or IBM POWER > 5,
417 this also enables accounting of stolen time on logically-partitioned
420 config VIRT_CPU_ACCOUNTING_GEN
421 bool "Full dynticks CPU time accounting"
422 depends on HAVE_CONTEXT_TRACKING
423 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
424 depends on GENERIC_CLOCKEVENTS
425 select VIRT_CPU_ACCOUNTING
426 select CONTEXT_TRACKING
428 Select this option to enable task and CPU time accounting on full
429 dynticks systems. This accounting is implemented by watching every
430 kernel-user boundaries using the context tracking subsystem.
431 The accounting is thus performed at the expense of some significant
434 For now this is only useful if you are working on the full
435 dynticks subsystem development.
441 config IRQ_TIME_ACCOUNTING
442 bool "Fine granularity task level IRQ time accounting"
443 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
445 Select this option to enable fine granularity task irq time
446 accounting. This is done by reading a timestamp on each
447 transitions between softirq and hardirq state, so there can be a
448 small performance impact.
450 If in doubt, say N here.
452 config HAVE_SCHED_AVG_IRQ
454 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
457 config SCHED_THERMAL_PRESSURE
458 bool "Enable periodic averaging of thermal pressure"
461 config BSD_PROCESS_ACCT
462 bool "BSD Process Accounting"
465 If you say Y here, a user level program will be able to instruct the
466 kernel (via a special system call) to write process accounting
467 information to a file: whenever a process exits, information about
468 that process will be appended to the file by the kernel. The
469 information includes things such as creation time, owning user,
470 command name, memory usage, controlling terminal etc. (the complete
471 list is in the struct acct in <file:include/linux/acct.h>). It is
472 up to the user level program to do useful things with this
473 information. This is generally a good idea, so say Y.
475 config BSD_PROCESS_ACCT_V3
476 bool "BSD Process Accounting version 3 file format"
477 depends on BSD_PROCESS_ACCT
480 If you say Y here, the process accounting information is written
481 in a new file format that also logs the process IDs of each
482 process and its parent. Note that this file format is incompatible
483 with previous v0/v1/v2 file formats, so you will need updated tools
484 for processing it. A preliminary version of these tools is available
485 at <http://www.gnu.org/software/acct/>.
488 bool "Export task/process statistics through netlink"
493 Export selected statistics for tasks/processes through the
494 generic netlink interface. Unlike BSD process accounting, the
495 statistics are available during the lifetime of tasks/processes as
496 responses to commands. Like BSD accounting, they are sent to user
501 config TASK_DELAY_ACCT
502 bool "Enable per-task delay accounting"
506 Collect information on time spent by a task waiting for system
507 resources like cpu, synchronous block I/O completion and swapping
508 in pages. Such statistics can help in setting a task's priorities
509 relative to other tasks for cpu, io, rss limits etc.
514 bool "Enable extended accounting over taskstats"
517 Collect extended task accounting data and send the data
518 to userland for processing over the taskstats interface.
522 config TASK_IO_ACCOUNTING
523 bool "Enable per-task storage I/O accounting"
524 depends on TASK_XACCT
526 Collect information on the number of bytes of storage I/O which this
532 bool "Pressure stall information tracking"
534 Collect metrics that indicate how overcommitted the CPU, memory,
535 and IO capacity are in the system.
537 If you say Y here, the kernel will create /proc/pressure/ with the
538 pressure statistics files cpu, memory, and io. These will indicate
539 the share of walltime in which some or all tasks in the system are
540 delayed due to contention of the respective resource.
542 In kernels with cgroup support, cgroups (cgroup2 only) will
543 have cpu.pressure, memory.pressure, and io.pressure files,
544 which aggregate pressure stalls for the grouped tasks only.
546 For more details see Documentation/accounting/psi.rst.
550 config PSI_DEFAULT_DISABLED
551 bool "Require boot parameter to enable pressure stall information tracking"
555 If set, pressure stall information tracking will be disabled
556 per default but can be enabled through passing psi=1 on the
557 kernel commandline during boot.
559 This feature adds some code to the task wakeup and sleep
560 paths of the scheduler. The overhead is too low to affect
561 common scheduling-intense workloads in practice (such as
562 webservers, memcache), but it does show up in artificial
563 scheduler stress tests, such as hackbench.
565 If you are paranoid and not sure what the kernel will be
570 endmenu # "CPU/Task time and stats accounting"
574 depends on SMP || COMPILE_TEST
577 Make sure that CPUs running critical tasks are not disturbed by
578 any source of "noise" such as unbound workqueues, timers, kthreads...
579 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
580 the "isolcpus=" boot parameter.
584 source "kernel/rcu/Kconfig"
591 tristate "Kernel .config support"
593 This option enables the complete Linux kernel ".config" file
594 contents to be saved in the kernel. It provides documentation
595 of which kernel options are used in a running kernel or in an
596 on-disk kernel. This information can be extracted from the kernel
597 image file with the script scripts/extract-ikconfig and used as
598 input to rebuild the current kernel or to build another kernel.
599 It can also be extracted from a running kernel by reading
600 /proc/config.gz if enabled (below).
603 bool "Enable access to .config through /proc/config.gz"
604 depends on IKCONFIG && PROC_FS
606 This option enables access to the kernel configuration file
607 through /proc/config.gz.
610 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
613 This option enables access to the in-kernel headers that are generated during
614 the build process. These can be used to build eBPF tracing programs,
615 or similar programs. If you build the headers as a module, a module called
616 kheaders.ko is built which can be loaded on-demand to get access to headers.
619 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
624 Select the minimal kernel log buffer size as a power of 2.
625 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
626 parameter, see below. Any higher size also might be forced
627 by "log_buf_len" boot parameter.
637 config LOG_CPU_MAX_BUF_SHIFT
638 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
641 default 12 if !BASE_SMALL
642 default 0 if BASE_SMALL
645 This option allows to increase the default ring buffer size
646 according to the number of CPUs. The value defines the contribution
647 of each CPU as a power of 2. The used space is typically only few
648 lines however it might be much more when problems are reported,
651 The increased size means that a new buffer has to be allocated and
652 the original static one is unused. It makes sense only on systems
653 with more CPUs. Therefore this value is used only when the sum of
654 contributions is greater than the half of the default kernel ring
655 buffer as defined by LOG_BUF_SHIFT. The default values are set
656 so that more than 64 CPUs are needed to trigger the allocation.
658 Also this option is ignored when "log_buf_len" kernel parameter is
659 used as it forces an exact (power of two) size of the ring buffer.
661 The number of possible CPUs is used for this computation ignoring
662 hotplugging making the computation optimal for the worst case
663 scenario while allowing a simple algorithm to be used from bootup.
665 Examples shift values and their meaning:
666 17 => 128 KB for each CPU
667 16 => 64 KB for each CPU
668 15 => 32 KB for each CPU
669 14 => 16 KB for each CPU
670 13 => 8 KB for each CPU
671 12 => 4 KB for each CPU
673 config PRINTK_SAFE_LOG_BUF_SHIFT
674 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
679 Select the size of an alternate printk per-CPU buffer where messages
680 printed from usafe contexts are temporary stored. One example would
681 be NMI messages, another one - printk recursion. The messages are
682 copied to the main log buffer in a safe context to avoid a deadlock.
683 The value defines the size as a power of 2.
685 Those messages are rare and limited. The largest one is when
686 a backtrace is printed. It usually fits into 4KB. Select
687 8KB if you want to be on the safe side.
690 17 => 128 KB for each CPU
691 16 => 64 KB for each CPU
692 15 => 32 KB for each CPU
693 14 => 16 KB for each CPU
694 13 => 8 KB for each CPU
695 12 => 4 KB for each CPU
698 # Architectures with an unreliable sched_clock() should select this:
700 config HAVE_UNSTABLE_SCHED_CLOCK
703 config GENERIC_SCHED_CLOCK
706 menu "Scheduler features"
709 bool "Enable utilization clamping for RT/FAIR tasks"
710 depends on CPU_FREQ_GOV_SCHEDUTIL
712 This feature enables the scheduler to track the clamped utilization
713 of each CPU based on RUNNABLE tasks scheduled on that CPU.
715 With this option, the user can specify the min and max CPU
716 utilization allowed for RUNNABLE tasks. The max utilization defines
717 the maximum frequency a task should use while the min utilization
718 defines the minimum frequency it should use.
720 Both min and max utilization clamp values are hints to the scheduler,
721 aiming at improving its frequency selection policy, but they do not
722 enforce or grant any specific bandwidth for tasks.
726 config UCLAMP_BUCKETS_COUNT
727 int "Number of supported utilization clamp buckets"
730 depends on UCLAMP_TASK
732 Defines the number of clamp buckets to use. The range of each bucket
733 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
734 number of clamp buckets the finer their granularity and the higher
735 the precision of clamping aggregation and tracking at run-time.
737 For example, with the minimum configuration value we will have 5
738 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
739 be refcounted in the [20..39]% bucket and will set the bucket clamp
740 effective value to 25%.
741 If a second 30% boosted task should be co-scheduled on the same CPU,
742 that task will be refcounted in the same bucket of the first task and
743 it will boost the bucket clamp effective value to 30%.
744 The clamp effective value of a bucket is reset to its nominal value
745 (20% in the example above) when there are no more tasks refcounted in
748 An additional boost/capping margin can be added to some tasks. In the
749 example above the 25% task will be boosted to 30% until it exits the
750 CPU. If that should be considered not acceptable on certain systems,
751 it's always possible to reduce the margin by increasing the number of
752 clamp buckets to trade off used memory for run-time tracking
755 If in doubt, use the default value.
760 # For architectures that want to enable the support for NUMA-affine scheduler
763 config ARCH_SUPPORTS_NUMA_BALANCING
767 # For architectures that prefer to flush all TLBs after a number of pages
768 # are unmapped instead of sending one IPI per page to flush. The architecture
769 # must provide guarantees on what happens if a clean TLB cache entry is
770 # written after the unmap. Details are in mm/rmap.c near the check for
771 # should_defer_flush. The architecture should also consider if the full flush
772 # and the refill costs are offset by the savings of sending fewer IPIs.
773 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
777 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
780 # For architectures that know their GCC __int128 support is sound
782 config ARCH_SUPPORTS_INT128
785 # For architectures that (ab)use NUMA to represent different memory regions
786 # all cpu-local but of different latencies, such as SuperH.
788 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
791 config NUMA_BALANCING
792 bool "Memory placement aware NUMA scheduler"
793 depends on ARCH_SUPPORTS_NUMA_BALANCING
794 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
795 depends on SMP && NUMA && MIGRATION
797 This option adds support for automatic NUMA aware memory/task placement.
798 The mechanism is quite primitive and is based on migrating memory when
799 it has references to the node the task is running on.
801 This system will be inactive on UMA systems.
803 config NUMA_BALANCING_DEFAULT_ENABLED
804 bool "Automatically enable NUMA aware memory/task placement"
806 depends on NUMA_BALANCING
808 If set, automatic NUMA balancing will be enabled if running on a NUMA
812 bool "Control Group support"
815 This option adds support for grouping sets of processes together, for
816 use with process control subsystems such as Cpusets, CFS, memory
817 controls or device isolation.
819 - Documentation/scheduler/sched-design-CFS.rst (CFS)
820 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
821 and resource control)
831 bool "Memory controller"
835 Provides control over the memory footprint of tasks in a cgroup.
838 bool "Swap controller"
839 depends on MEMCG && SWAP
841 Provides control over the swap space consumed by tasks in a cgroup.
843 config MEMCG_SWAP_ENABLED
844 bool "Swap controller enabled by default"
845 depends on MEMCG_SWAP
848 Memory Resource Controller Swap Extension comes with its price in
849 a bigger memory consumption. General purpose distribution kernels
850 which want to enable the feature but keep it disabled by default
851 and let the user enable it by swapaccount=1 boot command line
852 parameter should have this option unselected.
853 For those who want to have the feature enabled by default should
854 select this option (if, for some reason, they need to disable it
855 then swapaccount=0 does the trick).
859 depends on MEMCG && !SLOB
867 Generic block IO controller cgroup interface. This is the common
868 cgroup interface which should be used by various IO controlling
871 Currently, CFQ IO scheduler uses it to recognize task groups and
872 control disk bandwidth allocation (proportional time slice allocation)
873 to such task groups. It is also used by bio throttling logic in
874 block layer to implement upper limit in IO rates on a device.
876 This option only enables generic Block IO controller infrastructure.
877 One needs to also enable actual IO controlling logic/policy. For
878 enabling proportional weight division of disk bandwidth in CFQ, set
879 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
880 CONFIG_BLK_DEV_THROTTLING=y.
882 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
884 config CGROUP_WRITEBACK
886 depends on MEMCG && BLK_CGROUP
889 menuconfig CGROUP_SCHED
890 bool "CPU controller"
893 This feature lets CPU scheduler recognize task groups and control CPU
894 bandwidth allocation to such task groups. It uses cgroups to group
898 config FAIR_GROUP_SCHED
899 bool "Group scheduling for SCHED_OTHER"
900 depends on CGROUP_SCHED
904 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
905 depends on FAIR_GROUP_SCHED
908 This option allows users to define CPU bandwidth rates (limits) for
909 tasks running within the fair group scheduler. Groups with no limit
910 set are considered to be unconstrained and will run with no
912 See Documentation/scheduler/sched-bwc.rst for more information.
914 config RT_GROUP_SCHED
915 bool "Group scheduling for SCHED_RR/FIFO"
916 depends on CGROUP_SCHED
919 This feature lets you explicitly allocate real CPU bandwidth
920 to task groups. If enabled, it will also make it impossible to
921 schedule realtime tasks for non-root users until you allocate
922 realtime bandwidth for them.
923 See Documentation/scheduler/sched-rt-group.rst for more information.
927 config UCLAMP_TASK_GROUP
928 bool "Utilization clamping per group of tasks"
929 depends on CGROUP_SCHED
930 depends on UCLAMP_TASK
933 This feature enables the scheduler to track the clamped utilization
934 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
936 When this option is enabled, the user can specify a min and max
937 CPU bandwidth which is allowed for each single task in a group.
938 The max bandwidth allows to clamp the maximum frequency a task
939 can use, while the min bandwidth allows to define a minimum
940 frequency a task will always use.
942 When task group based utilization clamping is enabled, an eventually
943 specified task-specific clamp value is constrained by the cgroup
944 specified clamp value. Both minimum and maximum task clamping cannot
945 be bigger than the corresponding clamping defined at task group level.
950 bool "PIDs controller"
952 Provides enforcement of process number limits in the scope of a
953 cgroup. Any attempt to fork more processes than is allowed in the
954 cgroup will fail. PIDs are fundamentally a global resource because it
955 is fairly trivial to reach PID exhaustion before you reach even a
956 conservative kmemcg limit. As a result, it is possible to grind a
957 system to halt without being limited by other cgroup policies. The
958 PIDs controller is designed to stop this from happening.
960 It should be noted that organisational operations (such as attaching
961 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
962 since the PIDs limit only affects a process's ability to fork, not to
966 bool "RDMA controller"
968 Provides enforcement of RDMA resources defined by IB stack.
969 It is fairly easy for consumers to exhaust RDMA resources, which
970 can result into resource unavailability to other consumers.
971 RDMA controller is designed to stop this from happening.
972 Attaching processes with active RDMA resources to the cgroup
973 hierarchy is allowed even if can cross the hierarchy's limit.
975 config CGROUP_FREEZER
976 bool "Freezer controller"
978 Provides a way to freeze and unfreeze all tasks in a
981 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
982 controller includes important in-kernel memory consumers per default.
984 If you're using cgroup2, say N.
986 config CGROUP_HUGETLB
987 bool "HugeTLB controller"
988 depends on HUGETLB_PAGE
992 Provides a cgroup controller for HugeTLB pages.
993 When you enable this, you can put a per cgroup limit on HugeTLB usage.
994 The limit is enforced during page fault. Since HugeTLB doesn't
995 support page reclaim, enforcing the limit at page fault time implies
996 that, the application will get SIGBUS signal if it tries to access
997 HugeTLB pages beyond its limit. This requires the application to know
998 beforehand how much HugeTLB pages it would require for its use. The
999 control group is tracked in the third page lru pointer. This means
1000 that we cannot use the controller with huge page less than 3 pages.
1003 bool "Cpuset controller"
1006 This option will let you create and manage CPUSETs which
1007 allow dynamically partitioning a system into sets of CPUs and
1008 Memory Nodes and assigning tasks to run only within those sets.
1009 This is primarily useful on large SMP or NUMA systems.
1013 config PROC_PID_CPUSET
1014 bool "Include legacy /proc/<pid>/cpuset file"
1018 config CGROUP_DEVICE
1019 bool "Device controller"
1021 Provides a cgroup controller implementing whitelists for
1022 devices which a process in the cgroup can mknod or open.
1024 config CGROUP_CPUACCT
1025 bool "Simple CPU accounting controller"
1027 Provides a simple controller for monitoring the
1028 total CPU consumed by the tasks in a cgroup.
1031 bool "Perf controller"
1032 depends on PERF_EVENTS
1034 This option extends the perf per-cpu mode to restrict monitoring
1035 to threads which belong to the cgroup specified and run on the
1036 designated cpu. Or this can be used to have cgroup ID in samples
1037 so that it can monitor performance events among cgroups.
1042 bool "Support for eBPF programs attached to cgroups"
1043 depends on BPF_SYSCALL
1044 select SOCK_CGROUP_DATA
1046 Allow attaching eBPF programs to a cgroup using the bpf(2)
1047 syscall command BPF_PROG_ATTACH.
1049 In which context these programs are accessed depends on the type
1050 of attachment. For instance, programs that are attached using
1051 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1055 bool "Debug controller"
1057 depends on DEBUG_KERNEL
1059 This option enables a simple controller that exports
1060 debugging information about the cgroups framework. This
1061 controller is for control cgroup debugging only. Its
1062 interfaces are not stable.
1066 config SOCK_CGROUP_DATA
1072 menuconfig NAMESPACES
1073 bool "Namespaces support" if EXPERT
1074 depends on MULTIUSER
1077 Provides the way to make tasks work with different objects using
1078 the same id. For example same IPC id may refer to different objects
1079 or same user id or pid may refer to different tasks when used in
1080 different namespaces.
1085 bool "UTS namespace"
1088 In this namespace tasks see different info provided with the
1092 bool "TIME namespace"
1093 depends on GENERIC_VDSO_TIME_NS
1096 In this namespace boottime and monotonic clocks can be set.
1097 The time will keep going with the same pace.
1100 bool "IPC namespace"
1101 depends on (SYSVIPC || POSIX_MQUEUE)
1104 In this namespace tasks work with IPC ids which correspond to
1105 different IPC objects in different namespaces.
1108 bool "User namespace"
1111 This allows containers, i.e. vservers, to use user namespaces
1112 to provide different user info for different servers.
1114 When user namespaces are enabled in the kernel it is
1115 recommended that the MEMCG option also be enabled and that
1116 user-space use the memory control groups to limit the amount
1117 of memory a memory unprivileged users can use.
1122 bool "PID Namespaces"
1125 Support process id namespaces. This allows having multiple
1126 processes with the same pid as long as they are in different
1127 pid namespaces. This is a building block of containers.
1130 bool "Network namespace"
1134 Allow user space to create what appear to be multiple instances
1135 of the network stack.
1139 config CHECKPOINT_RESTORE
1140 bool "Checkpoint/restore support"
1141 select PROC_CHILDREN
1144 Enables additional kernel features in a sake of checkpoint/restore.
1145 In particular it adds auxiliary prctl codes to setup process text,
1146 data and heap segment sizes, and a few additional /proc filesystem
1149 If unsure, say N here.
1151 config SCHED_AUTOGROUP
1152 bool "Automatic process group scheduling"
1155 select FAIR_GROUP_SCHED
1157 This option optimizes the scheduler for common desktop workloads by
1158 automatically creating and populating task groups. This separation
1159 of workloads isolates aggressive CPU burners (like build jobs) from
1160 desktop applications. Task group autogeneration is currently based
1163 config SYSFS_DEPRECATED
1164 bool "Enable deprecated sysfs features to support old userspace tools"
1168 This option adds code that switches the layout of the "block" class
1169 devices, to not show up in /sys/class/block/, but only in
1172 This switch is only active when the sysfs.deprecated=1 boot option is
1173 passed or the SYSFS_DEPRECATED_V2 option is set.
1175 This option allows new kernels to run on old distributions and tools,
1176 which might get confused by /sys/class/block/. Since 2007/2008 all
1177 major distributions and tools handle this just fine.
1179 Recent distributions and userspace tools after 2009/2010 depend on
1180 the existence of /sys/class/block/, and will not work with this
1183 Only if you are using a new kernel on an old distribution, you might
1186 config SYSFS_DEPRECATED_V2
1187 bool "Enable deprecated sysfs features by default"
1190 depends on SYSFS_DEPRECATED
1192 Enable deprecated sysfs by default.
1194 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1197 Only if you are using a new kernel on an old distribution, you might
1198 need to say Y here. Even then, odds are you would not need it
1199 enabled, you can always pass the boot option if absolutely necessary.
1202 bool "Kernel->user space relay support (formerly relayfs)"
1205 This option enables support for relay interface support in
1206 certain file systems (such as debugfs).
1207 It is designed to provide an efficient mechanism for tools and
1208 facilities to relay large amounts of data from kernel space to
1213 config BLK_DEV_INITRD
1214 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1216 The initial RAM filesystem is a ramfs which is loaded by the
1217 boot loader (loadlin or lilo) and that is mounted as root
1218 before the normal boot procedure. It is typically used to
1219 load modules needed to mount the "real" root file system,
1220 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1222 If RAM disk support (BLK_DEV_RAM) is also included, this
1223 also enables initial RAM disk (initrd) support and adds
1224 15 Kbytes (more on some other architectures) to the kernel size.
1230 source "usr/Kconfig"
1235 bool "Boot config support"
1236 select BLK_DEV_INITRD
1238 Extra boot config allows system admin to pass a config file as
1239 complemental extension of kernel cmdline when booting.
1240 The boot config file must be attached at the end of initramfs
1241 with checksum, size and magic word.
1242 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1247 prompt "Compiler optimization level"
1248 default CC_OPTIMIZE_FOR_PERFORMANCE
1250 config CC_OPTIMIZE_FOR_PERFORMANCE
1251 bool "Optimize for performance (-O2)"
1253 This is the default optimization level for the kernel, building
1254 with the "-O2" compiler flag for best performance and most
1255 helpful compile-time warnings.
1257 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1258 bool "Optimize more for performance (-O3)"
1260 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1262 Choosing this option will pass "-O3" to your compiler to optimize
1263 the kernel yet more for performance.
1265 config CC_OPTIMIZE_FOR_SIZE
1266 bool "Optimize for size (-Os)"
1267 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1269 Choosing this option will pass "-Os" to your compiler resulting
1270 in a smaller kernel.
1274 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1277 This requires that the arch annotates or otherwise protects
1278 its external entry points from being discarded. Linker scripts
1279 must also merge .text.*, .data.*, and .bss.* correctly into
1280 output sections. Care must be taken not to pull in unrelated
1281 sections (e.g., '.text.init'). Typically '.' in section names
1282 is used to distinguish them from label names / C identifiers.
1284 config LD_DEAD_CODE_DATA_ELIMINATION
1285 bool "Dead code and data elimination (EXPERIMENTAL)"
1286 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1288 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1289 depends on $(cc-option,-ffunction-sections -fdata-sections)
1290 depends on $(ld-option,--gc-sections)
1292 Enable this if you want to do dead code and data elimination with
1293 the linker by compiling with -ffunction-sections -fdata-sections,
1294 and linking with --gc-sections.
1296 This can reduce on disk and in-memory size of the kernel
1297 code and static data, particularly for small configs and
1298 on small systems. This has the possibility of introducing
1299 silently broken kernel if the required annotations are not
1300 present. This option is not well tested yet, so use at your
1309 config SYSCTL_EXCEPTION_TRACE
1312 Enable support for /proc/sys/debug/exception-trace.
1314 config SYSCTL_ARCH_UNALIGN_NO_WARN
1317 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1318 Allows arch to define/use @no_unaligned_warning to possibly warn
1319 about unaligned access emulation going on under the hood.
1321 config SYSCTL_ARCH_UNALIGN_ALLOW
1324 Enable support for /proc/sys/kernel/unaligned-trap
1325 Allows arches to define/use @unaligned_enabled to runtime toggle
1326 the unaligned access emulation.
1327 see arch/parisc/kernel/unaligned.c for reference
1329 config HAVE_PCSPKR_PLATFORM
1332 # interpreter that classic socket filters depend on
1337 bool "Configure standard kernel features (expert users)"
1338 # Unhide debug options, to make the on-by-default options visible
1341 This option allows certain base kernel options and settings
1342 to be disabled or tweaked. This is for specialized
1343 environments which can tolerate a "non-standard" kernel.
1344 Only use this if you really know what you are doing.
1347 bool "Enable 16-bit UID system calls" if EXPERT
1348 depends on HAVE_UID16 && MULTIUSER
1351 This enables the legacy 16-bit UID syscall wrappers.
1354 bool "Multiple users, groups and capabilities support" if EXPERT
1357 This option enables support for non-root users, groups and
1360 If you say N here, all processes will run with UID 0, GID 0, and all
1361 possible capabilities. Saying N here also compiles out support for
1362 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1365 If unsure, say Y here.
1367 config SGETMASK_SYSCALL
1368 bool "sgetmask/ssetmask syscalls support" if EXPERT
1369 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1371 sys_sgetmask and sys_ssetmask are obsolete system calls
1372 no longer supported in libc but still enabled by default in some
1375 If unsure, leave the default option here.
1377 config SYSFS_SYSCALL
1378 bool "Sysfs syscall support" if EXPERT
1381 sys_sysfs is an obsolete system call no longer supported in libc.
1382 Note that disabling this option is more secure but might break
1383 compatibility with some systems.
1385 If unsure say Y here.
1388 bool "open by fhandle syscalls" if EXPERT
1392 If you say Y here, a user level program will be able to map
1393 file names to handle and then later use the handle for
1394 different file system operations. This is useful in implementing
1395 userspace file servers, which now track files using handles instead
1396 of names. The handle would remain the same even if file names
1397 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1401 bool "Posix Clocks & timers" if EXPERT
1404 This includes native support for POSIX timers to the kernel.
1405 Some embedded systems have no use for them and therefore they
1406 can be configured out to reduce the size of the kernel image.
1408 When this option is disabled, the following syscalls won't be
1409 available: timer_create, timer_gettime: timer_getoverrun,
1410 timer_settime, timer_delete, clock_adjtime, getitimer,
1411 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1412 clock_getres and clock_nanosleep syscalls will be limited to
1413 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1419 bool "Enable support for printk" if EXPERT
1422 This option enables normal printk support. Removing it
1423 eliminates most of the message strings from the kernel image
1424 and makes the kernel more or less silent. As this makes it
1425 very difficult to diagnose system problems, saying N here is
1426 strongly discouraged.
1434 bool "BUG() support" if EXPERT
1437 Disabling this option eliminates support for BUG and WARN, reducing
1438 the size of your kernel image and potentially quietly ignoring
1439 numerous fatal conditions. You should only consider disabling this
1440 option for embedded systems with no facilities for reporting errors.
1446 bool "Enable ELF core dumps" if EXPERT
1448 Enable support for generating core dumps. Disabling saves about 4k.
1451 config PCSPKR_PLATFORM
1452 bool "Enable PC-Speaker support" if EXPERT
1453 depends on HAVE_PCSPKR_PLATFORM
1457 This option allows to disable the internal PC-Speaker
1458 support, saving some memory.
1462 bool "Enable full-sized data structures for core" if EXPERT
1464 Disabling this option reduces the size of miscellaneous core
1465 kernel data structures. This saves memory on small machines,
1466 but may reduce performance.
1469 bool "Enable futex support" if EXPERT
1473 Disabling this option will cause the kernel to be built without
1474 support for "fast userspace mutexes". The resulting kernel may not
1475 run glibc-based applications correctly.
1479 depends on FUTEX && RT_MUTEXES
1482 config HAVE_FUTEX_CMPXCHG
1486 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1487 is implemented and always working. This removes a couple of runtime
1491 bool "Enable eventpoll support" if EXPERT
1494 Disabling this option will cause the kernel to be built without
1495 support for epoll family of system calls.
1498 bool "Enable signalfd() system call" if EXPERT
1501 Enable the signalfd() system call that allows to receive signals
1502 on a file descriptor.
1507 bool "Enable timerfd() system call" if EXPERT
1510 Enable the timerfd() system call that allows to receive timer
1511 events on a file descriptor.
1516 bool "Enable eventfd() system call" if EXPERT
1519 Enable the eventfd() system call that allows to receive both
1520 kernel notification (ie. KAIO) or userspace notifications.
1525 bool "Use full shmem filesystem" if EXPERT
1529 The shmem is an internal filesystem used to manage shared memory.
1530 It is backed by swap and manages resource limits. It is also exported
1531 to userspace as tmpfs if TMPFS is enabled. Disabling this
1532 option replaces shmem and tmpfs with the much simpler ramfs code,
1533 which may be appropriate on small systems without swap.
1536 bool "Enable AIO support" if EXPERT
1539 This option enables POSIX asynchronous I/O which may by used
1540 by some high performance threaded applications. Disabling
1541 this option saves about 7k.
1544 bool "Enable IO uring support" if EXPERT
1548 This option enables support for the io_uring interface, enabling
1549 applications to submit and complete IO through submission and
1550 completion rings that are shared between the kernel and application.
1552 config ADVISE_SYSCALLS
1553 bool "Enable madvise/fadvise syscalls" if EXPERT
1556 This option enables the madvise and fadvise syscalls, used by
1557 applications to advise the kernel about their future memory or file
1558 usage, improving performance. If building an embedded system where no
1559 applications use these syscalls, you can disable this option to save
1562 config HAVE_ARCH_USERFAULTFD_WP
1565 Arch has userfaultfd write protection support
1568 bool "Enable membarrier() system call" if EXPERT
1571 Enable the membarrier() system call that allows issuing memory
1572 barriers across all running threads, which can be used to distribute
1573 the cost of user-space memory barriers asymmetrically by transforming
1574 pairs of memory barriers into pairs consisting of membarrier() and a
1580 bool "Load all symbols for debugging/ksymoops" if EXPERT
1583 Say Y here to let the kernel print out symbolic crash information and
1584 symbolic stack backtraces. This increases the size of the kernel
1585 somewhat, as all symbols have to be loaded into the kernel image.
1588 bool "Include all symbols in kallsyms"
1589 depends on DEBUG_KERNEL && KALLSYMS
1591 Normally kallsyms only contains the symbols of functions for nicer
1592 OOPS messages and backtraces (i.e., symbols from the text and inittext
1593 sections). This is sufficient for most cases. And only in very rare
1594 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1595 names of variables from the data sections, etc).
1597 This option makes sure that all symbols are loaded into the kernel
1598 image (i.e., symbols from all sections) in cost of increased kernel
1599 size (depending on the kernel configuration, it may be 300KiB or
1600 something like this).
1602 Say N unless you really need all symbols.
1604 config KALLSYMS_ABSOLUTE_PERCPU
1607 default X86_64 && SMP
1609 config KALLSYMS_BASE_RELATIVE
1614 Instead of emitting them as absolute values in the native word size,
1615 emit the symbol references in the kallsyms table as 32-bit entries,
1616 each containing a relative value in the range [base, base + U32_MAX]
1617 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1618 an absolute value in the range [0, S32_MAX] or a relative value in the
1619 range [base, base + S32_MAX], where base is the lowest relative symbol
1620 address encountered in the image.
1622 On 64-bit builds, this reduces the size of the address table by 50%,
1623 but more importantly, it results in entries whose values are build
1624 time constants, and no relocation pass is required at runtime to fix
1625 up the entries based on the runtime load address of the kernel.
1627 # end of the "standard kernel features (expert users)" menu
1629 # syscall, maps, verifier
1632 bool "LSM Instrumentation with BPF"
1633 depends on BPF_EVENTS
1634 depends on BPF_SYSCALL
1638 Enables instrumentation of the security hooks with eBPF programs for
1639 implementing dynamic MAC and Audit Policies.
1641 If you are unsure how to answer this question, answer N.
1644 bool "Enable bpf() system call"
1649 Enable the bpf() system call that allows to manipulate eBPF
1650 programs and maps via file descriptors.
1652 config ARCH_WANT_DEFAULT_BPF_JIT
1655 config BPF_JIT_ALWAYS_ON
1656 bool "Permanently enable BPF JIT and remove BPF interpreter"
1657 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1659 Enables BPF JIT and removes BPF interpreter to avoid
1660 speculative execution of BPF instructions by the interpreter
1662 config BPF_JIT_DEFAULT_ON
1663 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1664 depends on HAVE_EBPF_JIT && BPF_JIT
1667 bool "Enable userfaultfd() system call"
1670 Enable the userfaultfd() system call that allows to intercept and
1671 handle page faults in userland.
1673 config ARCH_HAS_MEMBARRIER_CALLBACKS
1676 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1680 bool "Enable rseq() system call" if EXPERT
1682 depends on HAVE_RSEQ
1685 Enable the restartable sequences system call. It provides a
1686 user-space cache for the current CPU number value, which
1687 speeds up getting the current CPU number from user-space,
1688 as well as an ABI to speed up user-space operations on
1695 bool "Enabled debugging of rseq() system call" if EXPERT
1696 depends on RSEQ && DEBUG_KERNEL
1698 Enable extra debugging checks for the rseq system call.
1703 bool "Embedded system"
1704 option allnoconfig_y
1707 This option should be enabled if compiling the kernel for
1708 an embedded system so certain expert options are available
1711 config HAVE_PERF_EVENTS
1714 See tools/perf/design.txt for details.
1716 config PERF_USE_VMALLOC
1719 See tools/perf/design.txt for details
1722 bool "PC/104 support" if EXPERT
1724 Expose PC/104 form factor device drivers and options available for
1725 selection and configuration. Enable this option if your target
1726 machine has a PC/104 bus.
1728 menu "Kernel Performance Events And Counters"
1731 bool "Kernel performance events and counters"
1732 default y if PROFILING
1733 depends on HAVE_PERF_EVENTS
1737 Enable kernel support for various performance events provided
1738 by software and hardware.
1740 Software events are supported either built-in or via the
1741 use of generic tracepoints.
1743 Most modern CPUs support performance events via performance
1744 counter registers. These registers count the number of certain
1745 types of hw events: such as instructions executed, cachemisses
1746 suffered, or branches mis-predicted - without slowing down the
1747 kernel or applications. These registers can also trigger interrupts
1748 when a threshold number of events have passed - and can thus be
1749 used to profile the code that runs on that CPU.
1751 The Linux Performance Event subsystem provides an abstraction of
1752 these software and hardware event capabilities, available via a
1753 system call and used by the "perf" utility in tools/perf/. It
1754 provides per task and per CPU counters, and it provides event
1755 capabilities on top of those.
1759 config DEBUG_PERF_USE_VMALLOC
1761 bool "Debug: use vmalloc to back perf mmap() buffers"
1762 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1763 select PERF_USE_VMALLOC
1765 Use vmalloc memory to back perf mmap() buffers.
1767 Mostly useful for debugging the vmalloc code on platforms
1768 that don't require it.
1774 config VM_EVENT_COUNTERS
1776 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1778 VM event counters are needed for event counts to be shown.
1779 This option allows the disabling of the VM event counters
1780 on EXPERT systems. /proc/vmstat will only show page counts
1781 if VM event counters are disabled.
1785 bool "Enable SLUB debugging support" if EXPERT
1786 depends on SLUB && SYSFS
1788 SLUB has extensive debug support features. Disabling these can
1789 result in significant savings in code size. This also disables
1790 SLUB sysfs support. /sys/slab will not exist and there will be
1791 no support for cache validation etc.
1793 config SLUB_MEMCG_SYSFS_ON
1795 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1796 depends on SLUB && SYSFS && MEMCG
1798 SLUB creates a directory under /sys/kernel/slab for each
1799 allocation cache to host info and debug files. If memory
1800 cgroup is enabled, each cache can have per memory cgroup
1801 caches. SLUB can create the same sysfs directories for these
1802 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1803 to a very high number of debug files being created. This is
1804 controlled by slub_memcg_sysfs boot parameter and this
1805 config option determines the parameter's default value.
1808 bool "Disable heap randomization"
1811 Randomizing heap placement makes heap exploits harder, but it
1812 also breaks ancient binaries (including anything libc5 based).
1813 This option changes the bootup default to heap randomization
1814 disabled, and can be overridden at runtime by setting
1815 /proc/sys/kernel/randomize_va_space to 2.
1817 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1820 prompt "Choose SLAB allocator"
1823 This option allows to select a slab allocator.
1827 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1829 The regular slab allocator that is established and known to work
1830 well in all environments. It organizes cache hot objects in
1831 per cpu and per node queues.
1834 bool "SLUB (Unqueued Allocator)"
1835 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1837 SLUB is a slab allocator that minimizes cache line usage
1838 instead of managing queues of cached objects (SLAB approach).
1839 Per cpu caching is realized using slabs of objects instead
1840 of queues of objects. SLUB can use memory efficiently
1841 and has enhanced diagnostics. SLUB is the default choice for
1846 bool "SLOB (Simple Allocator)"
1848 SLOB replaces the stock allocator with a drastically simpler
1849 allocator. SLOB is generally more space efficient but
1850 does not perform as well on large systems.
1854 config SLAB_MERGE_DEFAULT
1855 bool "Allow slab caches to be merged"
1858 For reduced kernel memory fragmentation, slab caches can be
1859 merged when they share the same size and other characteristics.
1860 This carries a risk of kernel heap overflows being able to
1861 overwrite objects from merged caches (and more easily control
1862 cache layout), which makes such heap attacks easier to exploit
1863 by attackers. By keeping caches unmerged, these kinds of exploits
1864 can usually only damage objects in the same cache. To disable
1865 merging at runtime, "slab_nomerge" can be passed on the kernel
1868 config SLAB_FREELIST_RANDOM
1870 depends on SLAB || SLUB
1871 bool "SLAB freelist randomization"
1873 Randomizes the freelist order used on creating new pages. This
1874 security feature reduces the predictability of the kernel slab
1875 allocator against heap overflows.
1877 config SLAB_FREELIST_HARDENED
1878 bool "Harden slab freelist metadata"
1881 Many kernel heap attacks try to target slab cache metadata and
1882 other infrastructure. This options makes minor performance
1883 sacrifices to harden the kernel slab allocator against common
1884 freelist exploit methods.
1886 config SHUFFLE_PAGE_ALLOCATOR
1887 bool "Page allocator randomization"
1888 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1890 Randomization of the page allocator improves the average
1891 utilization of a direct-mapped memory-side-cache. See section
1892 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1893 6.2a specification for an example of how a platform advertises
1894 the presence of a memory-side-cache. There are also incidental
1895 security benefits as it reduces the predictability of page
1896 allocations to compliment SLAB_FREELIST_RANDOM, but the
1897 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1898 10th order of pages is selected based on cache utilization
1901 While the randomization improves cache utilization it may
1902 negatively impact workloads on platforms without a cache. For
1903 this reason, by default, the randomization is enabled only
1904 after runtime detection of a direct-mapped memory-side-cache.
1905 Otherwise, the randomization may be force enabled with the
1906 'page_alloc.shuffle' kernel command line parameter.
1910 config SLUB_CPU_PARTIAL
1912 depends on SLUB && SMP
1913 bool "SLUB per cpu partial cache"
1915 Per cpu partial caches accelerate objects allocation and freeing
1916 that is local to a processor at the price of more indeterminism
1917 in the latency of the free. On overflow these caches will be cleared
1918 which requires the taking of locks that may cause latency spikes.
1919 Typically one would choose no for a realtime system.
1921 config MMAP_ALLOW_UNINITIALIZED
1922 bool "Allow mmapped anonymous memory to be uninitialized"
1923 depends on EXPERT && !MMU
1926 Normally, and according to the Linux spec, anonymous memory obtained
1927 from mmap() has its contents cleared before it is passed to
1928 userspace. Enabling this config option allows you to request that
1929 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1930 providing a huge performance boost. If this option is not enabled,
1931 then the flag will be ignored.
1933 This is taken advantage of by uClibc's malloc(), and also by
1934 ELF-FDPIC binfmt's brk and stack allocator.
1936 Because of the obvious security issues, this option should only be
1937 enabled on embedded devices where you control what is run in
1938 userspace. Since that isn't generally a problem on no-MMU systems,
1939 it is normally safe to say Y here.
1941 See Documentation/nommu-mmap.txt for more information.
1943 config SYSTEM_DATA_VERIFICATION
1945 select SYSTEM_TRUSTED_KEYRING
1949 select ASYMMETRIC_KEY_TYPE
1950 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1953 select X509_CERTIFICATE_PARSER
1954 select PKCS7_MESSAGE_PARSER
1956 Provide PKCS#7 message verification using the contents of the system
1957 trusted keyring to provide public keys. This then can be used for
1958 module verification, kexec image verification and firmware blob
1962 bool "Profiling support"
1964 Say Y here to enable the extended profiling support mechanisms used
1965 by profilers such as OProfile.
1968 # Place an empty function call at each tracepoint site. Can be
1969 # dynamically changed for a probe function.
1974 endmenu # General setup
1976 source "arch/Kconfig"
1983 default 0 if BASE_FULL
1984 default 1 if !BASE_FULL
1986 config MODULE_SIG_FORMAT
1988 select SYSTEM_DATA_VERIFICATION
1991 bool "Enable loadable module support"
1994 Kernel modules are small pieces of compiled code which can
1995 be inserted in the running kernel, rather than being
1996 permanently built into the kernel. You use the "modprobe"
1997 tool to add (and sometimes remove) them. If you say Y here,
1998 many parts of the kernel can be built as modules (by
1999 answering M instead of Y where indicated): this is most
2000 useful for infrequently used options which are not required
2001 for booting. For more information, see the man pages for
2002 modprobe, lsmod, modinfo, insmod and rmmod.
2004 If you say Y here, you will need to run "make
2005 modules_install" to put the modules under /lib/modules/
2006 where modprobe can find them (you may need to be root to do
2013 config MODULE_FORCE_LOAD
2014 bool "Forced module loading"
2017 Allow loading of modules without version information (ie. modprobe
2018 --force). Forced module loading sets the 'F' (forced) taint flag and
2019 is usually a really bad idea.
2021 config MODULE_UNLOAD
2022 bool "Module unloading"
2024 Without this option you will not be able to unload any
2025 modules (note that some modules may not be unloadable
2026 anyway), which makes your kernel smaller, faster
2027 and simpler. If unsure, say Y.
2029 config MODULE_FORCE_UNLOAD
2030 bool "Forced module unloading"
2031 depends on MODULE_UNLOAD
2033 This option allows you to force a module to unload, even if the
2034 kernel believes it is unsafe: the kernel will remove the module
2035 without waiting for anyone to stop using it (using the -f option to
2036 rmmod). This is mainly for kernel developers and desperate users.
2040 bool "Module versioning support"
2042 Usually, you have to use modules compiled with your kernel.
2043 Saying Y here makes it sometimes possible to use modules
2044 compiled for different kernels, by adding enough information
2045 to the modules to (hopefully) spot any changes which would
2046 make them incompatible with the kernel you are running. If
2049 config ASM_MODVERSIONS
2051 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2053 This enables module versioning for exported symbols also from
2054 assembly. This can be enabled only when the target architecture
2057 config MODULE_REL_CRCS
2059 depends on MODVERSIONS
2061 config MODULE_SRCVERSION_ALL
2062 bool "Source checksum for all modules"
2064 Modules which contain a MODULE_VERSION get an extra "srcversion"
2065 field inserted into their modinfo section, which contains a
2066 sum of the source files which made it. This helps maintainers
2067 see exactly which source was used to build a module (since
2068 others sometimes change the module source without updating
2069 the version). With this option, such a "srcversion" field
2070 will be created for all modules. If unsure, say N.
2073 bool "Module signature verification"
2074 select MODULE_SIG_FORMAT
2076 Check modules for valid signatures upon load: the signature
2077 is simply appended to the module. For more information see
2078 <file:Documentation/admin-guide/module-signing.rst>.
2080 Note that this option adds the OpenSSL development packages as a
2081 kernel build dependency so that the signing tool can use its crypto
2084 You should enable this option if you wish to use either
2085 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2086 another LSM - otherwise unsigned modules will be loadable regardless
2087 of the lockdown policy.
2089 !!!WARNING!!! If you enable this option, you MUST make sure that the
2090 module DOES NOT get stripped after being signed. This includes the
2091 debuginfo strip done by some packagers (such as rpmbuild) and
2092 inclusion into an initramfs that wants the module size reduced.
2094 config MODULE_SIG_FORCE
2095 bool "Require modules to be validly signed"
2096 depends on MODULE_SIG
2098 Reject unsigned modules or signed modules for which we don't have a
2099 key. Without this, such modules will simply taint the kernel.
2101 config MODULE_SIG_ALL
2102 bool "Automatically sign all modules"
2104 depends on MODULE_SIG
2106 Sign all modules during make modules_install. Without this option,
2107 modules must be signed manually, using the scripts/sign-file tool.
2109 comment "Do not forget to sign required modules with scripts/sign-file"
2110 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2113 prompt "Which hash algorithm should modules be signed with?"
2114 depends on MODULE_SIG
2116 This determines which sort of hashing algorithm will be used during
2117 signature generation. This algorithm _must_ be built into the kernel
2118 directly so that signature verification can take place. It is not
2119 possible to load a signed module containing the algorithm to check
2120 the signature on that module.
2122 config MODULE_SIG_SHA1
2123 bool "Sign modules with SHA-1"
2126 config MODULE_SIG_SHA224
2127 bool "Sign modules with SHA-224"
2128 select CRYPTO_SHA256
2130 config MODULE_SIG_SHA256
2131 bool "Sign modules with SHA-256"
2132 select CRYPTO_SHA256
2134 config MODULE_SIG_SHA384
2135 bool "Sign modules with SHA-384"
2136 select CRYPTO_SHA512
2138 config MODULE_SIG_SHA512
2139 bool "Sign modules with SHA-512"
2140 select CRYPTO_SHA512
2144 config MODULE_SIG_HASH
2146 depends on MODULE_SIG
2147 default "sha1" if MODULE_SIG_SHA1
2148 default "sha224" if MODULE_SIG_SHA224
2149 default "sha256" if MODULE_SIG_SHA256
2150 default "sha384" if MODULE_SIG_SHA384
2151 default "sha512" if MODULE_SIG_SHA512
2153 config MODULE_COMPRESS
2154 bool "Compress modules on installation"
2157 Compresses kernel modules when 'make modules_install' is run; gzip or
2158 xz depending on "Compression algorithm" below.
2160 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2162 Out-of-tree kernel modules installed using Kbuild will also be
2163 compressed upon installation.
2165 Note: for modules inside an initrd or initramfs, it's more efficient
2166 to compress the whole initrd or initramfs instead.
2168 Note: This is fully compatible with signed modules.
2173 prompt "Compression algorithm"
2174 depends on MODULE_COMPRESS
2175 default MODULE_COMPRESS_GZIP
2177 This determines which sort of compression will be used during
2178 'make modules_install'.
2180 GZIP (default) and XZ are supported.
2182 config MODULE_COMPRESS_GZIP
2185 config MODULE_COMPRESS_XZ
2190 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2191 bool "Allow loading of modules with missing namespace imports"
2193 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2194 a namespace. A module that makes use of a symbol exported with such a
2195 namespace is required to import the namespace via MODULE_IMPORT_NS().
2196 There is no technical reason to enforce correct namespace imports,
2197 but it creates consistency between symbols defining namespaces and
2198 users importing namespaces they make use of. This option relaxes this
2199 requirement and lifts the enforcement when loading a module.
2203 config UNUSED_SYMBOLS
2204 bool "Enable unused/obsolete exported symbols"
2207 Unused but exported symbols make the kernel needlessly bigger. For
2208 that reason most of these unused exports will soon be removed. This
2209 option is provided temporarily to provide a transition period in case
2210 some external kernel module needs one of these symbols anyway. If you
2211 encounter such a case in your module, consider if you are actually
2212 using the right API. (rationale: since nobody in the kernel is using
2213 this in a module, there is a pretty good chance it's actually the
2214 wrong interface to use). If you really need the symbol, please send a
2215 mail to the linux kernel mailing list mentioning the symbol and why
2216 you really need it, and what the merge plan to the mainline kernel for
2219 config TRIM_UNUSED_KSYMS
2220 bool "Trim unused exported kernel symbols"
2221 depends on !UNUSED_SYMBOLS
2223 The kernel and some modules make many symbols available for
2224 other modules to use via EXPORT_SYMBOL() and variants. Depending
2225 on the set of modules being selected in your kernel configuration,
2226 many of those exported symbols might never be used.
2228 This option allows for unused exported symbols to be dropped from
2229 the build. In turn, this provides the compiler more opportunities
2230 (especially when using LTO) for optimizing the code and reducing
2231 binary size. This might have some security advantages as well.
2233 If unsure, or if you need to build out-of-tree modules, say N.
2235 config UNUSED_KSYMS_WHITELIST
2236 string "Whitelist of symbols to keep in ksymtab"
2237 depends on TRIM_UNUSED_KSYMS
2239 By default, all unused exported symbols will be un-exported from the
2240 build when TRIM_UNUSED_KSYMS is selected.
2242 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2243 exported at all times, even in absence of in-tree users. The value to
2244 set here is the path to a text file containing the list of symbols,
2245 one per line. The path can be absolute, or relative to the kernel
2250 config MODULES_TREE_LOOKUP
2252 depends on PERF_EVENTS || TRACING
2254 config INIT_ALL_POSSIBLE
2257 Back when each arch used to define their own cpu_online_mask and
2258 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2259 with all 1s, and others with all 0s. When they were centralised,
2260 it was better to provide this option than to break all the archs
2261 and have several arch maintainers pursuing me down dark alleys.
2263 source "block/Kconfig"
2265 config PREEMPT_NOTIFIERS
2275 Build a simple ASN.1 grammar compiler that produces a bytecode output
2276 that can be interpreted by the ASN.1 stream decoder and used to
2277 inform it as to what tags are to be expected in a stream and what
2278 functions to call on what tags.
2280 source "kernel/Kconfig.locks"
2282 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2285 # It may be useful for an architecture to override the definitions of the
2286 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2287 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2288 # different calling convention for syscalls. They can also override the
2289 # macros for not-implemented syscalls in kernel/sys_ni.c and
2290 # kernel/time/posix-stubs.c. All these overrides need to be available in
2291 # <asm/syscall_wrapper.h>.
2292 config ARCH_HAS_SYSCALL_WRAPPER
OSDN Git Service
