1 # SPDX-License-Identifier: GPL-2.0-only
6 default "/lib/modules/$(shell,uname -r)/.config"
7 default "/etc/kernel-config"
8 default "/boot/config-$(shell,uname -r)"
10 default "arch/$(ARCH)/defconfig"
13 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
17 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
21 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
25 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
28 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
30 config CC_HAS_ASM_GOTO
31 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
33 config TOOLS_SUPPORT_RELR
34 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
36 config CC_HAS_WARN_MAYBE_UNINITIALIZED
37 def_bool $(cc-option,-Wmaybe-uninitialized)
39 GCC >= 4.7 supports this option.
41 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
43 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
44 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
46 GCC's -Wmaybe-uninitialized is not reliable by definition.
47 Lots of false positive warnings are produced in some cases.
49 If this option is enabled, -Wno-maybe-uninitialzed is passed
50 to the compiler to suppress maybe-uninitialized warnings.
59 config BUILDTIME_EXTABLE_SORT
62 config THREAD_INFO_IN_TASK
65 Select this to move thread_info off the stack into task_struct. To
66 make this work, an arch will need to remove all thread_info fields
67 except flags and fix any runtime bugs.
69 One subtle change that will be needed is to use try_get_task_stack()
70 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
79 depends on BROKEN || !SMP
82 config INIT_ENV_ARG_LIMIT
87 Maximum of each of the number of arguments and environment
88 variables passed to init from the kernel command line.
91 bool "Compile also drivers which will not load"
95 Some drivers can be compiled on a different platform than they are
96 intended to be run on. Despite they cannot be loaded there (or even
97 when they load they cannot be used due to missing HW support),
98 developers still, opposing to distributors, might want to build such
99 drivers to compile-test them.
101 If you are a developer and want to build everything available, say Y
102 here. If you are a user/distributor, say N here to exclude useless
103 drivers to be distributed.
106 bool "Compile test headers that should be standalone compilable"
108 Compile test headers listed in header-test-y target to ensure they are
109 self-contained, i.e. compilable as standalone units.
111 If you are a developer or tester and want to ensure the requested
112 headers are self-contained, say Y here. Otherwise, choose N.
114 config KERNEL_HEADER_TEST
115 bool "Compile test kernel headers"
116 depends on HEADER_TEST
118 Headers in include/ are used to build external moduls.
119 Compile test them to ensure they are self-contained, i.e.
120 compilable as standalone units.
122 If you are a developer or tester and want to ensure the headers
123 in include/ are self-contained, say Y here. Otherwise, choose N.
125 config UAPI_HEADER_TEST
126 bool "Compile test UAPI headers"
127 depends on HEADER_TEST && HEADERS_INSTALL && CC_CAN_LINK
129 Compile test headers exported to user-space to ensure they are
130 self-contained, i.e. compilable as standalone units.
132 If you are a developer or tester and want to ensure the exported
133 headers are self-contained, say Y here. Otherwise, choose N.
136 string "Local version - append to kernel release"
138 Append an extra string to the end of your kernel version.
139 This will show up when you type uname, for example.
140 The string you set here will be appended after the contents of
141 any files with a filename matching localversion* in your
142 object and source tree, in that order. Your total string can
143 be a maximum of 64 characters.
145 config LOCALVERSION_AUTO
146 bool "Automatically append version information to the version string"
148 depends on !COMPILE_TEST
150 This will try to automatically determine if the current tree is a
151 release tree by looking for git tags that belong to the current
152 top of tree revision.
154 A string of the format -gxxxxxxxx will be added to the localversion
155 if a git-based tree is found. The string generated by this will be
156 appended after any matching localversion* files, and after the value
157 set in CONFIG_LOCALVERSION.
159 (The actual string used here is the first eight characters produced
160 by running the command:
162 $ git rev-parse --verify HEAD
164 which is done within the script "scripts/setlocalversion".)
167 string "Build ID Salt"
170 The build ID is used to link binaries and their debug info. Setting
171 this option will use the value in the calculation of the build id.
172 This is mostly useful for distributions which want to ensure the
173 build is unique between builds. It's safe to leave the default.
175 config HAVE_KERNEL_GZIP
178 config HAVE_KERNEL_BZIP2
181 config HAVE_KERNEL_LZMA
184 config HAVE_KERNEL_XZ
187 config HAVE_KERNEL_LZO
190 config HAVE_KERNEL_LZ4
193 config HAVE_KERNEL_UNCOMPRESSED
197 prompt "Kernel compression mode"
199 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
201 The linux kernel is a kind of self-extracting executable.
202 Several compression algorithms are available, which differ
203 in efficiency, compression and decompression speed.
204 Compression speed is only relevant when building a kernel.
205 Decompression speed is relevant at each boot.
207 If you have any problems with bzip2 or lzma compressed
208 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
209 version of this functionality (bzip2 only), for 2.4, was
210 supplied by Christian Ludwig)
212 High compression options are mostly useful for users, who
213 are low on disk space (embedded systems), but for whom ram
216 If in doubt, select 'gzip'
220 depends on HAVE_KERNEL_GZIP
222 The old and tried gzip compression. It provides a good balance
223 between compression ratio and decompression speed.
227 depends on HAVE_KERNEL_BZIP2
229 Its compression ratio and speed is intermediate.
230 Decompression speed is slowest among the choices. The kernel
231 size is about 10% smaller with bzip2, in comparison to gzip.
232 Bzip2 uses a large amount of memory. For modern kernels you
233 will need at least 8MB RAM or more for booting.
237 depends on HAVE_KERNEL_LZMA
239 This compression algorithm's ratio is best. Decompression speed
240 is between gzip and bzip2. Compression is slowest.
241 The kernel size is about 33% smaller with LZMA in comparison to gzip.
245 depends on HAVE_KERNEL_XZ
247 XZ uses the LZMA2 algorithm and instruction set specific
248 BCJ filters which can improve compression ratio of executable
249 code. The size of the kernel is about 30% smaller with XZ in
250 comparison to gzip. On architectures for which there is a BCJ
251 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
252 will create a few percent smaller kernel than plain LZMA.
254 The speed is about the same as with LZMA: The decompression
255 speed of XZ is better than that of bzip2 but worse than gzip
256 and LZO. Compression is slow.
260 depends on HAVE_KERNEL_LZO
262 Its compression ratio is the poorest among the choices. The kernel
263 size is about 10% bigger than gzip; however its speed
264 (both compression and decompression) is the fastest.
268 depends on HAVE_KERNEL_LZ4
270 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
271 A preliminary version of LZ4 de/compression tool is available at
272 <https://code.google.com/p/lz4/>.
274 Its compression ratio is worse than LZO. The size of the kernel
275 is about 8% bigger than LZO. But the decompression speed is
278 config KERNEL_UNCOMPRESSED
280 depends on HAVE_KERNEL_UNCOMPRESSED
282 Produce uncompressed kernel image. This option is usually not what
283 you want. It is useful for debugging the kernel in slow simulation
284 environments, where decompressing and moving the kernel is awfully
285 slow. This option allows early boot code to skip the decompressor
286 and jump right at uncompressed kernel image.
290 config DEFAULT_HOSTNAME
291 string "Default hostname"
294 This option determines the default system hostname before userspace
295 calls sethostname(2). The kernel traditionally uses "(none)" here,
296 but you may wish to use a different default here to make a minimal
297 system more usable with less configuration.
300 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
301 # add proper SWAP support to them, in which case this can be remove.
307 bool "Support for paging of anonymous memory (swap)"
308 depends on MMU && BLOCK && !ARCH_NO_SWAP
311 This option allows you to choose whether you want to have support
312 for so called swap devices or swap files in your kernel that are
313 used to provide more virtual memory than the actual RAM present
314 in your computer. If unsure say Y.
319 Inter Process Communication is a suite of library functions and
320 system calls which let processes (running programs) synchronize and
321 exchange information. It is generally considered to be a good thing,
322 and some programs won't run unless you say Y here. In particular, if
323 you want to run the DOS emulator dosemu under Linux (read the
324 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
325 you'll need to say Y here.
327 You can find documentation about IPC with "info ipc" and also in
328 section 6.4 of the Linux Programmer's Guide, available from
329 <http://www.tldp.org/guides.html>.
331 config SYSVIPC_SYSCTL
338 bool "POSIX Message Queues"
341 POSIX variant of message queues is a part of IPC. In POSIX message
342 queues every message has a priority which decides about succession
343 of receiving it by a process. If you want to compile and run
344 programs written e.g. for Solaris with use of its POSIX message
345 queues (functions mq_*) say Y here.
347 POSIX message queues are visible as a filesystem called 'mqueue'
348 and can be mounted somewhere if you want to do filesystem
349 operations on message queues.
353 config POSIX_MQUEUE_SYSCTL
355 depends on POSIX_MQUEUE
359 config CROSS_MEMORY_ATTACH
360 bool "Enable process_vm_readv/writev syscalls"
364 Enabling this option adds the system calls process_vm_readv and
365 process_vm_writev which allow a process with the correct privileges
366 to directly read from or write to another process' address space.
367 See the man page for more details.
370 bool "uselib syscall"
371 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
373 This option enables the uselib syscall, a system call used in the
374 dynamic linker from libc5 and earlier. glibc does not use this
375 system call. If you intend to run programs built on libc5 or
376 earlier, you may need to enable this syscall. Current systems
377 running glibc can safely disable this.
380 bool "Auditing support"
383 Enable auditing infrastructure that can be used with another
384 kernel subsystem, such as SELinux (which requires this for
385 logging of avc messages output). System call auditing is included
386 on architectures which support it.
388 config HAVE_ARCH_AUDITSYSCALL
393 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
396 source "kernel/irq/Kconfig"
397 source "kernel/time/Kconfig"
398 source "kernel/Kconfig.preempt"
400 menu "CPU/Task time and stats accounting"
402 config VIRT_CPU_ACCOUNTING
406 prompt "Cputime accounting"
407 default TICK_CPU_ACCOUNTING if !PPC64
408 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
410 # Kind of a stub config for the pure tick based cputime accounting
411 config TICK_CPU_ACCOUNTING
412 bool "Simple tick based cputime accounting"
413 depends on !S390 && !NO_HZ_FULL
415 This is the basic tick based cputime accounting that maintains
416 statistics about user, system and idle time spent on per jiffies
421 config VIRT_CPU_ACCOUNTING_NATIVE
422 bool "Deterministic task and CPU time accounting"
423 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
424 select VIRT_CPU_ACCOUNTING
426 Select this option to enable more accurate task and CPU time
427 accounting. This is done by reading a CPU counter on each
428 kernel entry and exit and on transitions within the kernel
429 between system, softirq and hardirq state, so there is a
430 small performance impact. In the case of s390 or IBM POWER > 5,
431 this also enables accounting of stolen time on logically-partitioned
434 config VIRT_CPU_ACCOUNTING_GEN
435 bool "Full dynticks CPU time accounting"
436 depends on HAVE_CONTEXT_TRACKING
437 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
438 depends on GENERIC_CLOCKEVENTS
439 select VIRT_CPU_ACCOUNTING
440 select CONTEXT_TRACKING
442 Select this option to enable task and CPU time accounting on full
443 dynticks systems. This accounting is implemented by watching every
444 kernel-user boundaries using the context tracking subsystem.
445 The accounting is thus performed at the expense of some significant
448 For now this is only useful if you are working on the full
449 dynticks subsystem development.
455 config IRQ_TIME_ACCOUNTING
456 bool "Fine granularity task level IRQ time accounting"
457 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
459 Select this option to enable fine granularity task irq time
460 accounting. This is done by reading a timestamp on each
461 transitions between softirq and hardirq state, so there can be a
462 small performance impact.
464 If in doubt, say N here.
466 config HAVE_SCHED_AVG_IRQ
468 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
471 config BSD_PROCESS_ACCT
472 bool "BSD Process Accounting"
475 If you say Y here, a user level program will be able to instruct the
476 kernel (via a special system call) to write process accounting
477 information to a file: whenever a process exits, information about
478 that process will be appended to the file by the kernel. The
479 information includes things such as creation time, owning user,
480 command name, memory usage, controlling terminal etc. (the complete
481 list is in the struct acct in <file:include/linux/acct.h>). It is
482 up to the user level program to do useful things with this
483 information. This is generally a good idea, so say Y.
485 config BSD_PROCESS_ACCT_V3
486 bool "BSD Process Accounting version 3 file format"
487 depends on BSD_PROCESS_ACCT
490 If you say Y here, the process accounting information is written
491 in a new file format that also logs the process IDs of each
492 process and its parent. Note that this file format is incompatible
493 with previous v0/v1/v2 file formats, so you will need updated tools
494 for processing it. A preliminary version of these tools is available
495 at <http://www.gnu.org/software/acct/>.
498 bool "Export task/process statistics through netlink"
503 Export selected statistics for tasks/processes through the
504 generic netlink interface. Unlike BSD process accounting, the
505 statistics are available during the lifetime of tasks/processes as
506 responses to commands. Like BSD accounting, they are sent to user
511 config TASK_DELAY_ACCT
512 bool "Enable per-task delay accounting"
516 Collect information on time spent by a task waiting for system
517 resources like cpu, synchronous block I/O completion and swapping
518 in pages. Such statistics can help in setting a task's priorities
519 relative to other tasks for cpu, io, rss limits etc.
524 bool "Enable extended accounting over taskstats"
527 Collect extended task accounting data and send the data
528 to userland for processing over the taskstats interface.
532 config TASK_IO_ACCOUNTING
533 bool "Enable per-task storage I/O accounting"
534 depends on TASK_XACCT
536 Collect information on the number of bytes of storage I/O which this
542 bool "Pressure stall information tracking"
544 Collect metrics that indicate how overcommitted the CPU, memory,
545 and IO capacity are in the system.
547 If you say Y here, the kernel will create /proc/pressure/ with the
548 pressure statistics files cpu, memory, and io. These will indicate
549 the share of walltime in which some or all tasks in the system are
550 delayed due to contention of the respective resource.
552 In kernels with cgroup support, cgroups (cgroup2 only) will
553 have cpu.pressure, memory.pressure, and io.pressure files,
554 which aggregate pressure stalls for the grouped tasks only.
556 For more details see Documentation/accounting/psi.rst.
560 config PSI_DEFAULT_DISABLED
561 bool "Require boot parameter to enable pressure stall information tracking"
565 If set, pressure stall information tracking will be disabled
566 per default but can be enabled through passing psi=1 on the
567 kernel commandline during boot.
569 This feature adds some code to the task wakeup and sleep
570 paths of the scheduler. The overhead is too low to affect
571 common scheduling-intense workloads in practice (such as
572 webservers, memcache), but it does show up in artificial
573 scheduler stress tests, such as hackbench.
575 If you are paranoid and not sure what the kernel will be
580 endmenu # "CPU/Task time and stats accounting"
584 depends on SMP || COMPILE_TEST
587 Make sure that CPUs running critical tasks are not disturbed by
588 any source of "noise" such as unbound workqueues, timers, kthreads...
589 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
590 the "isolcpus=" boot parameter.
594 source "kernel/rcu/Kconfig"
601 tristate "Kernel .config support"
603 This option enables the complete Linux kernel ".config" file
604 contents to be saved in the kernel. It provides documentation
605 of which kernel options are used in a running kernel or in an
606 on-disk kernel. This information can be extracted from the kernel
607 image file with the script scripts/extract-ikconfig and used as
608 input to rebuild the current kernel or to build another kernel.
609 It can also be extracted from a running kernel by reading
610 /proc/config.gz if enabled (below).
613 bool "Enable access to .config through /proc/config.gz"
614 depends on IKCONFIG && PROC_FS
616 This option enables access to the kernel configuration file
617 through /proc/config.gz.
620 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
623 This option enables access to the in-kernel headers that are generated during
624 the build process. These can be used to build eBPF tracing programs,
625 or similar programs. If you build the headers as a module, a module called
626 kheaders.ko is built which can be loaded on-demand to get access to headers.
629 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
634 Select the minimal kernel log buffer size as a power of 2.
635 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
636 parameter, see below. Any higher size also might be forced
637 by "log_buf_len" boot parameter.
647 config LOG_CPU_MAX_BUF_SHIFT
648 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
651 default 12 if !BASE_SMALL
652 default 0 if BASE_SMALL
655 This option allows to increase the default ring buffer size
656 according to the number of CPUs. The value defines the contribution
657 of each CPU as a power of 2. The used space is typically only few
658 lines however it might be much more when problems are reported,
661 The increased size means that a new buffer has to be allocated and
662 the original static one is unused. It makes sense only on systems
663 with more CPUs. Therefore this value is used only when the sum of
664 contributions is greater than the half of the default kernel ring
665 buffer as defined by LOG_BUF_SHIFT. The default values are set
666 so that more than 64 CPUs are needed to trigger the allocation.
668 Also this option is ignored when "log_buf_len" kernel parameter is
669 used as it forces an exact (power of two) size of the ring buffer.
671 The number of possible CPUs is used for this computation ignoring
672 hotplugging making the computation optimal for the worst case
673 scenario while allowing a simple algorithm to be used from bootup.
675 Examples shift values and their meaning:
676 17 => 128 KB for each CPU
677 16 => 64 KB for each CPU
678 15 => 32 KB for each CPU
679 14 => 16 KB for each CPU
680 13 => 8 KB for each CPU
681 12 => 4 KB for each CPU
683 config PRINTK_SAFE_LOG_BUF_SHIFT
684 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
689 Select the size of an alternate printk per-CPU buffer where messages
690 printed from usafe contexts are temporary stored. One example would
691 be NMI messages, another one - printk recursion. The messages are
692 copied to the main log buffer in a safe context to avoid a deadlock.
693 The value defines the size as a power of 2.
695 Those messages are rare and limited. The largest one is when
696 a backtrace is printed. It usually fits into 4KB. Select
697 8KB if you want to be on the safe side.
700 17 => 128 KB for each CPU
701 16 => 64 KB for each CPU
702 15 => 32 KB for each CPU
703 14 => 16 KB for each CPU
704 13 => 8 KB for each CPU
705 12 => 4 KB for each CPU
708 # Architectures with an unreliable sched_clock() should select this:
710 config HAVE_UNSTABLE_SCHED_CLOCK
713 config GENERIC_SCHED_CLOCK
716 menu "Scheduler features"
719 bool "Enable utilization clamping for RT/FAIR tasks"
720 depends on CPU_FREQ_GOV_SCHEDUTIL
722 This feature enables the scheduler to track the clamped utilization
723 of each CPU based on RUNNABLE tasks scheduled on that CPU.
725 With this option, the user can specify the min and max CPU
726 utilization allowed for RUNNABLE tasks. The max utilization defines
727 the maximum frequency a task should use while the min utilization
728 defines the minimum frequency it should use.
730 Both min and max utilization clamp values are hints to the scheduler,
731 aiming at improving its frequency selection policy, but they do not
732 enforce or grant any specific bandwidth for tasks.
736 config UCLAMP_BUCKETS_COUNT
737 int "Number of supported utilization clamp buckets"
740 depends on UCLAMP_TASK
742 Defines the number of clamp buckets to use. The range of each bucket
743 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
744 number of clamp buckets the finer their granularity and the higher
745 the precision of clamping aggregation and tracking at run-time.
747 For example, with the minimum configuration value we will have 5
748 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
749 be refcounted in the [20..39]% bucket and will set the bucket clamp
750 effective value to 25%.
751 If a second 30% boosted task should be co-scheduled on the same CPU,
752 that task will be refcounted in the same bucket of the first task and
753 it will boost the bucket clamp effective value to 30%.
754 The clamp effective value of a bucket is reset to its nominal value
755 (20% in the example above) when there are no more tasks refcounted in
758 An additional boost/capping margin can be added to some tasks. In the
759 example above the 25% task will be boosted to 30% until it exits the
760 CPU. If that should be considered not acceptable on certain systems,
761 it's always possible to reduce the margin by increasing the number of
762 clamp buckets to trade off used memory for run-time tracking
765 If in doubt, use the default value.
770 # For architectures that want to enable the support for NUMA-affine scheduler
773 config ARCH_SUPPORTS_NUMA_BALANCING
777 # For architectures that prefer to flush all TLBs after a number of pages
778 # are unmapped instead of sending one IPI per page to flush. The architecture
779 # must provide guarantees on what happens if a clean TLB cache entry is
780 # written after the unmap. Details are in mm/rmap.c near the check for
781 # should_defer_flush. The architecture should also consider if the full flush
782 # and the refill costs are offset by the savings of sending fewer IPIs.
783 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
787 # For architectures that know their GCC __int128 support is sound
789 config ARCH_SUPPORTS_INT128
792 # For architectures that (ab)use NUMA to represent different memory regions
793 # all cpu-local but of different latencies, such as SuperH.
795 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
798 config NUMA_BALANCING
799 bool "Memory placement aware NUMA scheduler"
800 depends on ARCH_SUPPORTS_NUMA_BALANCING
801 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
802 depends on SMP && NUMA && MIGRATION
804 This option adds support for automatic NUMA aware memory/task placement.
805 The mechanism is quite primitive and is based on migrating memory when
806 it has references to the node the task is running on.
808 This system will be inactive on UMA systems.
810 config NUMA_BALANCING_DEFAULT_ENABLED
811 bool "Automatically enable NUMA aware memory/task placement"
813 depends on NUMA_BALANCING
815 If set, automatic NUMA balancing will be enabled if running on a NUMA
819 bool "Control Group support"
822 This option adds support for grouping sets of processes together, for
823 use with process control subsystems such as Cpusets, CFS, memory
824 controls or device isolation.
826 - Documentation/scheduler/sched-design-CFS.rst (CFS)
827 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
828 and resource control)
838 bool "Memory controller"
842 Provides control over the memory footprint of tasks in a cgroup.
845 bool "Swap controller"
846 depends on MEMCG && SWAP
848 Provides control over the swap space consumed by tasks in a cgroup.
850 config MEMCG_SWAP_ENABLED
851 bool "Swap controller enabled by default"
852 depends on MEMCG_SWAP
855 Memory Resource Controller Swap Extension comes with its price in
856 a bigger memory consumption. General purpose distribution kernels
857 which want to enable the feature but keep it disabled by default
858 and let the user enable it by swapaccount=1 boot command line
859 parameter should have this option unselected.
860 For those who want to have the feature enabled by default should
861 select this option (if, for some reason, they need to disable it
862 then swapaccount=0 does the trick).
866 depends on MEMCG && !SLOB
874 Generic block IO controller cgroup interface. This is the common
875 cgroup interface which should be used by various IO controlling
878 Currently, CFQ IO scheduler uses it to recognize task groups and
879 control disk bandwidth allocation (proportional time slice allocation)
880 to such task groups. It is also used by bio throttling logic in
881 block layer to implement upper limit in IO rates on a device.
883 This option only enables generic Block IO controller infrastructure.
884 One needs to also enable actual IO controlling logic/policy. For
885 enabling proportional weight division of disk bandwidth in CFQ, set
886 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
887 CONFIG_BLK_DEV_THROTTLING=y.
889 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
891 config CGROUP_WRITEBACK
893 depends on MEMCG && BLK_CGROUP
896 menuconfig CGROUP_SCHED
897 bool "CPU controller"
900 This feature lets CPU scheduler recognize task groups and control CPU
901 bandwidth allocation to such task groups. It uses cgroups to group
905 config FAIR_GROUP_SCHED
906 bool "Group scheduling for SCHED_OTHER"
907 depends on CGROUP_SCHED
911 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
912 depends on FAIR_GROUP_SCHED
915 This option allows users to define CPU bandwidth rates (limits) for
916 tasks running within the fair group scheduler. Groups with no limit
917 set are considered to be unconstrained and will run with no
919 See Documentation/scheduler/sched-bwc.rst for more information.
921 config RT_GROUP_SCHED
922 bool "Group scheduling for SCHED_RR/FIFO"
923 depends on CGROUP_SCHED
926 This feature lets you explicitly allocate real CPU bandwidth
927 to task groups. If enabled, it will also make it impossible to
928 schedule realtime tasks for non-root users until you allocate
929 realtime bandwidth for them.
930 See Documentation/scheduler/sched-rt-group.rst for more information.
934 config UCLAMP_TASK_GROUP
935 bool "Utilization clamping per group of tasks"
936 depends on CGROUP_SCHED
937 depends on UCLAMP_TASK
940 This feature enables the scheduler to track the clamped utilization
941 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
943 When this option is enabled, the user can specify a min and max
944 CPU bandwidth which is allowed for each single task in a group.
945 The max bandwidth allows to clamp the maximum frequency a task
946 can use, while the min bandwidth allows to define a minimum
947 frequency a task will always use.
949 When task group based utilization clamping is enabled, an eventually
950 specified task-specific clamp value is constrained by the cgroup
951 specified clamp value. Both minimum and maximum task clamping cannot
952 be bigger than the corresponding clamping defined at task group level.
957 bool "PIDs controller"
959 Provides enforcement of process number limits in the scope of a
960 cgroup. Any attempt to fork more processes than is allowed in the
961 cgroup will fail. PIDs are fundamentally a global resource because it
962 is fairly trivial to reach PID exhaustion before you reach even a
963 conservative kmemcg limit. As a result, it is possible to grind a
964 system to halt without being limited by other cgroup policies. The
965 PIDs controller is designed to stop this from happening.
967 It should be noted that organisational operations (such as attaching
968 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
969 since the PIDs limit only affects a process's ability to fork, not to
973 bool "RDMA controller"
975 Provides enforcement of RDMA resources defined by IB stack.
976 It is fairly easy for consumers to exhaust RDMA resources, which
977 can result into resource unavailability to other consumers.
978 RDMA controller is designed to stop this from happening.
979 Attaching processes with active RDMA resources to the cgroup
980 hierarchy is allowed even if can cross the hierarchy's limit.
982 config CGROUP_FREEZER
983 bool "Freezer controller"
985 Provides a way to freeze and unfreeze all tasks in a
988 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
989 controller includes important in-kernel memory consumers per default.
991 If you're using cgroup2, say N.
993 config CGROUP_HUGETLB
994 bool "HugeTLB controller"
995 depends on HUGETLB_PAGE
999 Provides a cgroup controller for HugeTLB pages.
1000 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1001 The limit is enforced during page fault. Since HugeTLB doesn't
1002 support page reclaim, enforcing the limit at page fault time implies
1003 that, the application will get SIGBUS signal if it tries to access
1004 HugeTLB pages beyond its limit. This requires the application to know
1005 beforehand how much HugeTLB pages it would require for its use. The
1006 control group is tracked in the third page lru pointer. This means
1007 that we cannot use the controller with huge page less than 3 pages.
1010 bool "Cpuset controller"
1013 This option will let you create and manage CPUSETs which
1014 allow dynamically partitioning a system into sets of CPUs and
1015 Memory Nodes and assigning tasks to run only within those sets.
1016 This is primarily useful on large SMP or NUMA systems.
1020 config PROC_PID_CPUSET
1021 bool "Include legacy /proc/<pid>/cpuset file"
1025 config CGROUP_DEVICE
1026 bool "Device controller"
1028 Provides a cgroup controller implementing whitelists for
1029 devices which a process in the cgroup can mknod or open.
1031 config CGROUP_CPUACCT
1032 bool "Simple CPU accounting controller"
1034 Provides a simple controller for monitoring the
1035 total CPU consumed by the tasks in a cgroup.
1038 bool "Perf controller"
1039 depends on PERF_EVENTS
1041 This option extends the perf per-cpu mode to restrict monitoring
1042 to threads which belong to the cgroup specified and run on the
1048 bool "Support for eBPF programs attached to cgroups"
1049 depends on BPF_SYSCALL
1050 select SOCK_CGROUP_DATA
1052 Allow attaching eBPF programs to a cgroup using the bpf(2)
1053 syscall command BPF_PROG_ATTACH.
1055 In which context these programs are accessed depends on the type
1056 of attachment. For instance, programs that are attached using
1057 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1061 bool "Debug controller"
1063 depends on DEBUG_KERNEL
1065 This option enables a simple controller that exports
1066 debugging information about the cgroups framework. This
1067 controller is for control cgroup debugging only. Its
1068 interfaces are not stable.
1072 config SOCK_CGROUP_DATA
1078 menuconfig NAMESPACES
1079 bool "Namespaces support" if EXPERT
1080 depends on MULTIUSER
1083 Provides the way to make tasks work with different objects using
1084 the same id. For example same IPC id may refer to different objects
1085 or same user id or pid may refer to different tasks when used in
1086 different namespaces.
1091 bool "UTS namespace"
1094 In this namespace tasks see different info provided with the
1098 bool "IPC namespace"
1099 depends on (SYSVIPC || POSIX_MQUEUE)
1102 In this namespace tasks work with IPC ids which correspond to
1103 different IPC objects in different namespaces.
1106 bool "User namespace"
1109 This allows containers, i.e. vservers, to use user namespaces
1110 to provide different user info for different servers.
1112 When user namespaces are enabled in the kernel it is
1113 recommended that the MEMCG option also be enabled and that
1114 user-space use the memory control groups to limit the amount
1115 of memory a memory unprivileged users can use.
1120 bool "PID Namespaces"
1123 Support process id namespaces. This allows having multiple
1124 processes with the same pid as long as they are in different
1125 pid namespaces. This is a building block of containers.
1128 bool "Network namespace"
1132 Allow user space to create what appear to be multiple instances
1133 of the network stack.
1137 config CHECKPOINT_RESTORE
1138 bool "Checkpoint/restore support"
1139 select PROC_CHILDREN
1142 Enables additional kernel features in a sake of checkpoint/restore.
1143 In particular it adds auxiliary prctl codes to setup process text,
1144 data and heap segment sizes, and a few additional /proc filesystem
1147 If unsure, say N here.
1149 config SCHED_AUTOGROUP
1150 bool "Automatic process group scheduling"
1153 select FAIR_GROUP_SCHED
1155 This option optimizes the scheduler for common desktop workloads by
1156 automatically creating and populating task groups. This separation
1157 of workloads isolates aggressive CPU burners (like build jobs) from
1158 desktop applications. Task group autogeneration is currently based
1161 config SYSFS_DEPRECATED
1162 bool "Enable deprecated sysfs features to support old userspace tools"
1166 This option adds code that switches the layout of the "block" class
1167 devices, to not show up in /sys/class/block/, but only in
1170 This switch is only active when the sysfs.deprecated=1 boot option is
1171 passed or the SYSFS_DEPRECATED_V2 option is set.
1173 This option allows new kernels to run on old distributions and tools,
1174 which might get confused by /sys/class/block/. Since 2007/2008 all
1175 major distributions and tools handle this just fine.
1177 Recent distributions and userspace tools after 2009/2010 depend on
1178 the existence of /sys/class/block/, and will not work with this
1181 Only if you are using a new kernel on an old distribution, you might
1184 config SYSFS_DEPRECATED_V2
1185 bool "Enable deprecated sysfs features by default"
1188 depends on SYSFS_DEPRECATED
1190 Enable deprecated sysfs by default.
1192 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1195 Only if you are using a new kernel on an old distribution, you might
1196 need to say Y here. Even then, odds are you would not need it
1197 enabled, you can always pass the boot option if absolutely necessary.
1200 bool "Kernel->user space relay support (formerly relayfs)"
1203 This option enables support for relay interface support in
1204 certain file systems (such as debugfs).
1205 It is designed to provide an efficient mechanism for tools and
1206 facilities to relay large amounts of data from kernel space to
1211 config BLK_DEV_INITRD
1212 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1214 The initial RAM filesystem is a ramfs which is loaded by the
1215 boot loader (loadlin or lilo) and that is mounted as root
1216 before the normal boot procedure. It is typically used to
1217 load modules needed to mount the "real" root file system,
1218 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1220 If RAM disk support (BLK_DEV_RAM) is also included, this
1221 also enables initial RAM disk (initrd) support and adds
1222 15 Kbytes (more on some other architectures) to the kernel size.
1228 source "usr/Kconfig"
1233 prompt "Compiler optimization level"
1234 default CC_OPTIMIZE_FOR_PERFORMANCE
1236 config CC_OPTIMIZE_FOR_PERFORMANCE
1237 bool "Optimize for performance (-O2)"
1239 This is the default optimization level for the kernel, building
1240 with the "-O2" compiler flag for best performance and most
1241 helpful compile-time warnings.
1243 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1244 bool "Optimize more for performance (-O3)"
1246 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1248 Choosing this option will pass "-O3" to your compiler to optimize
1249 the kernel yet more for performance.
1251 config CC_OPTIMIZE_FOR_SIZE
1252 bool "Optimize for size (-Os)"
1253 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1255 Choosing this option will pass "-Os" to your compiler resulting
1256 in a smaller kernel.
1260 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1263 This requires that the arch annotates or otherwise protects
1264 its external entry points from being discarded. Linker scripts
1265 must also merge .text.*, .data.*, and .bss.* correctly into
1266 output sections. Care must be taken not to pull in unrelated
1267 sections (e.g., '.text.init'). Typically '.' in section names
1268 is used to distinguish them from label names / C identifiers.
1270 config LD_DEAD_CODE_DATA_ELIMINATION
1271 bool "Dead code and data elimination (EXPERIMENTAL)"
1272 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1274 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1275 depends on $(cc-option,-ffunction-sections -fdata-sections)
1276 depends on $(ld-option,--gc-sections)
1278 Enable this if you want to do dead code and data elimination with
1279 the linker by compiling with -ffunction-sections -fdata-sections,
1280 and linking with --gc-sections.
1282 This can reduce on disk and in-memory size of the kernel
1283 code and static data, particularly for small configs and
1284 on small systems. This has the possibility of introducing
1285 silently broken kernel if the required annotations are not
1286 present. This option is not well tested yet, so use at your
1295 config SYSCTL_EXCEPTION_TRACE
1298 Enable support for /proc/sys/debug/exception-trace.
1300 config SYSCTL_ARCH_UNALIGN_NO_WARN
1303 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1304 Allows arch to define/use @no_unaligned_warning to possibly warn
1305 about unaligned access emulation going on under the hood.
1307 config SYSCTL_ARCH_UNALIGN_ALLOW
1310 Enable support for /proc/sys/kernel/unaligned-trap
1311 Allows arches to define/use @unaligned_enabled to runtime toggle
1312 the unaligned access emulation.
1313 see arch/parisc/kernel/unaligned.c for reference
1315 config HAVE_PCSPKR_PLATFORM
1318 # interpreter that classic socket filters depend on
1323 bool "Configure standard kernel features (expert users)"
1324 # Unhide debug options, to make the on-by-default options visible
1327 This option allows certain base kernel options and settings
1328 to be disabled or tweaked. This is for specialized
1329 environments which can tolerate a "non-standard" kernel.
1330 Only use this if you really know what you are doing.
1333 bool "Enable 16-bit UID system calls" if EXPERT
1334 depends on HAVE_UID16 && MULTIUSER
1337 This enables the legacy 16-bit UID syscall wrappers.
1340 bool "Multiple users, groups and capabilities support" if EXPERT
1343 This option enables support for non-root users, groups and
1346 If you say N here, all processes will run with UID 0, GID 0, and all
1347 possible capabilities. Saying N here also compiles out support for
1348 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1351 If unsure, say Y here.
1353 config SGETMASK_SYSCALL
1354 bool "sgetmask/ssetmask syscalls support" if EXPERT
1355 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1357 sys_sgetmask and sys_ssetmask are obsolete system calls
1358 no longer supported in libc but still enabled by default in some
1361 If unsure, leave the default option here.
1363 config SYSFS_SYSCALL
1364 bool "Sysfs syscall support" if EXPERT
1367 sys_sysfs is an obsolete system call no longer supported in libc.
1368 Note that disabling this option is more secure but might break
1369 compatibility with some systems.
1371 If unsure say Y here.
1373 config SYSCTL_SYSCALL
1374 bool "Sysctl syscall support" if EXPERT
1375 depends on PROC_SYSCTL
1379 sys_sysctl uses binary paths that have been found challenging
1380 to properly maintain and use. The interface in /proc/sys
1381 using paths with ascii names is now the primary path to this
1384 Almost nothing using the binary sysctl interface so if you are
1385 trying to save some space it is probably safe to disable this,
1386 making your kernel marginally smaller.
1388 If unsure say N here.
1391 bool "open by fhandle syscalls" if EXPERT
1395 If you say Y here, a user level program will be able to map
1396 file names to handle and then later use the handle for
1397 different file system operations. This is useful in implementing
1398 userspace file servers, which now track files using handles instead
1399 of names. The handle would remain the same even if file names
1400 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1404 bool "Posix Clocks & timers" if EXPERT
1407 This includes native support for POSIX timers to the kernel.
1408 Some embedded systems have no use for them and therefore they
1409 can be configured out to reduce the size of the kernel image.
1411 When this option is disabled, the following syscalls won't be
1412 available: timer_create, timer_gettime: timer_getoverrun,
1413 timer_settime, timer_delete, clock_adjtime, getitimer,
1414 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1415 clock_getres and clock_nanosleep syscalls will be limited to
1416 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1422 bool "Enable support for printk" if EXPERT
1425 This option enables normal printk support. Removing it
1426 eliminates most of the message strings from the kernel image
1427 and makes the kernel more or less silent. As this makes it
1428 very difficult to diagnose system problems, saying N here is
1429 strongly discouraged.
1437 bool "BUG() support" if EXPERT
1440 Disabling this option eliminates support for BUG and WARN, reducing
1441 the size of your kernel image and potentially quietly ignoring
1442 numerous fatal conditions. You should only consider disabling this
1443 option for embedded systems with no facilities for reporting errors.
1449 bool "Enable ELF core dumps" if EXPERT
1451 Enable support for generating core dumps. Disabling saves about 4k.
1454 config PCSPKR_PLATFORM
1455 bool "Enable PC-Speaker support" if EXPERT
1456 depends on HAVE_PCSPKR_PLATFORM
1460 This option allows to disable the internal PC-Speaker
1461 support, saving some memory.
1465 bool "Enable full-sized data structures for core" if EXPERT
1467 Disabling this option reduces the size of miscellaneous core
1468 kernel data structures. This saves memory on small machines,
1469 but may reduce performance.
1472 bool "Enable futex support" if EXPERT
1476 Disabling this option will cause the kernel to be built without
1477 support for "fast userspace mutexes". The resulting kernel may not
1478 run glibc-based applications correctly.
1482 depends on FUTEX && RT_MUTEXES
1485 config HAVE_FUTEX_CMPXCHG
1489 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1490 is implemented and always working. This removes a couple of runtime
1494 bool "Enable eventpoll support" if EXPERT
1497 Disabling this option will cause the kernel to be built without
1498 support for epoll family of system calls.
1501 bool "Enable signalfd() system call" if EXPERT
1504 Enable the signalfd() system call that allows to receive signals
1505 on a file descriptor.
1510 bool "Enable timerfd() system call" if EXPERT
1513 Enable the timerfd() system call that allows to receive timer
1514 events on a file descriptor.
1519 bool "Enable eventfd() system call" if EXPERT
1522 Enable the eventfd() system call that allows to receive both
1523 kernel notification (ie. KAIO) or userspace notifications.
1528 bool "Use full shmem filesystem" if EXPERT
1532 The shmem is an internal filesystem used to manage shared memory.
1533 It is backed by swap and manages resource limits. It is also exported
1534 to userspace as tmpfs if TMPFS is enabled. Disabling this
1535 option replaces shmem and tmpfs with the much simpler ramfs code,
1536 which may be appropriate on small systems without swap.
1539 bool "Enable AIO support" if EXPERT
1542 This option enables POSIX asynchronous I/O which may by used
1543 by some high performance threaded applications. Disabling
1544 this option saves about 7k.
1547 bool "Enable IO uring support" if EXPERT
1551 This option enables support for the io_uring interface, enabling
1552 applications to submit and complete IO through submission and
1553 completion rings that are shared between the kernel and application.
1555 config ADVISE_SYSCALLS
1556 bool "Enable madvise/fadvise syscalls" if EXPERT
1559 This option enables the madvise and fadvise syscalls, used by
1560 applications to advise the kernel about their future memory or file
1561 usage, improving performance. If building an embedded system where no
1562 applications use these syscalls, you can disable this option to save
1566 bool "Enable membarrier() system call" if EXPERT
1569 Enable the membarrier() system call that allows issuing memory
1570 barriers across all running threads, which can be used to distribute
1571 the cost of user-space memory barriers asymmetrically by transforming
1572 pairs of memory barriers into pairs consisting of membarrier() and a
1578 bool "Load all symbols for debugging/ksymoops" if EXPERT
1581 Say Y here to let the kernel print out symbolic crash information and
1582 symbolic stack backtraces. This increases the size of the kernel
1583 somewhat, as all symbols have to be loaded into the kernel image.
1586 bool "Include all symbols in kallsyms"
1587 depends on DEBUG_KERNEL && KALLSYMS
1589 Normally kallsyms only contains the symbols of functions for nicer
1590 OOPS messages and backtraces (i.e., symbols from the text and inittext
1591 sections). This is sufficient for most cases. And only in very rare
1592 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1593 names of variables from the data sections, etc).
1595 This option makes sure that all symbols are loaded into the kernel
1596 image (i.e., symbols from all sections) in cost of increased kernel
1597 size (depending on the kernel configuration, it may be 300KiB or
1598 something like this).
1600 Say N unless you really need all symbols.
1602 config KALLSYMS_ABSOLUTE_PERCPU
1605 default X86_64 && SMP
1607 config KALLSYMS_BASE_RELATIVE
1612 Instead of emitting them as absolute values in the native word size,
1613 emit the symbol references in the kallsyms table as 32-bit entries,
1614 each containing a relative value in the range [base, base + U32_MAX]
1615 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1616 an absolute value in the range [0, S32_MAX] or a relative value in the
1617 range [base, base + S32_MAX], where base is the lowest relative symbol
1618 address encountered in the image.
1620 On 64-bit builds, this reduces the size of the address table by 50%,
1621 but more importantly, it results in entries whose values are build
1622 time constants, and no relocation pass is required at runtime to fix
1623 up the entries based on the runtime load address of the kernel.
1625 # end of the "standard kernel features (expert users)" menu
1627 # syscall, maps, verifier
1629 bool "Enable bpf() system call"
1634 Enable the bpf() system call that allows to manipulate eBPF
1635 programs and maps via file descriptors.
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
1645 bool "Enable userfaultfd() system call"
1648 Enable the userfaultfd() system call that allows to intercept and
1649 handle page faults in userland.
1651 config ARCH_HAS_MEMBARRIER_CALLBACKS
1654 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1658 bool "Enable rseq() system call" if EXPERT
1660 depends on HAVE_RSEQ
1663 Enable the restartable sequences system call. It provides a
1664 user-space cache for the current CPU number value, which
1665 speeds up getting the current CPU number from user-space,
1666 as well as an ABI to speed up user-space operations on
1673 bool "Enabled debugging of rseq() system call" if EXPERT
1674 depends on RSEQ && DEBUG_KERNEL
1676 Enable extra debugging checks for the rseq system call.
1681 bool "Embedded system"
1682 option allnoconfig_y
1685 This option should be enabled if compiling the kernel for
1686 an embedded system so certain expert options are available
1689 config HAVE_PERF_EVENTS
1692 See tools/perf/design.txt for details.
1694 config PERF_USE_VMALLOC
1697 See tools/perf/design.txt for details
1700 bool "PC/104 support" if EXPERT
1702 Expose PC/104 form factor device drivers and options available for
1703 selection and configuration. Enable this option if your target
1704 machine has a PC/104 bus.
1706 menu "Kernel Performance Events And Counters"
1709 bool "Kernel performance events and counters"
1710 default y if PROFILING
1711 depends on HAVE_PERF_EVENTS
1715 Enable kernel support for various performance events provided
1716 by software and hardware.
1718 Software events are supported either built-in or via the
1719 use of generic tracepoints.
1721 Most modern CPUs support performance events via performance
1722 counter registers. These registers count the number of certain
1723 types of hw events: such as instructions executed, cachemisses
1724 suffered, or branches mis-predicted - without slowing down the
1725 kernel or applications. These registers can also trigger interrupts
1726 when a threshold number of events have passed - and can thus be
1727 used to profile the code that runs on that CPU.
1729 The Linux Performance Event subsystem provides an abstraction of
1730 these software and hardware event capabilities, available via a
1731 system call and used by the "perf" utility in tools/perf/. It
1732 provides per task and per CPU counters, and it provides event
1733 capabilities on top of those.
1737 config DEBUG_PERF_USE_VMALLOC
1739 bool "Debug: use vmalloc to back perf mmap() buffers"
1740 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1741 select PERF_USE_VMALLOC
1743 Use vmalloc memory to back perf mmap() buffers.
1745 Mostly useful for debugging the vmalloc code on platforms
1746 that don't require it.
1752 config VM_EVENT_COUNTERS
1754 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1756 VM event counters are needed for event counts to be shown.
1757 This option allows the disabling of the VM event counters
1758 on EXPERT systems. /proc/vmstat will only show page counts
1759 if VM event counters are disabled.
1763 bool "Enable SLUB debugging support" if EXPERT
1764 depends on SLUB && SYSFS
1766 SLUB has extensive debug support features. Disabling these can
1767 result in significant savings in code size. This also disables
1768 SLUB sysfs support. /sys/slab will not exist and there will be
1769 no support for cache validation etc.
1771 config SLUB_MEMCG_SYSFS_ON
1773 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1774 depends on SLUB && SYSFS && MEMCG
1776 SLUB creates a directory under /sys/kernel/slab for each
1777 allocation cache to host info and debug files. If memory
1778 cgroup is enabled, each cache can have per memory cgroup
1779 caches. SLUB can create the same sysfs directories for these
1780 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1781 to a very high number of debug files being created. This is
1782 controlled by slub_memcg_sysfs boot parameter and this
1783 config option determines the parameter's default value.
1786 bool "Disable heap randomization"
1789 Randomizing heap placement makes heap exploits harder, but it
1790 also breaks ancient binaries (including anything libc5 based).
1791 This option changes the bootup default to heap randomization
1792 disabled, and can be overridden at runtime by setting
1793 /proc/sys/kernel/randomize_va_space to 2.
1795 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1798 prompt "Choose SLAB allocator"
1801 This option allows to select a slab allocator.
1805 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1807 The regular slab allocator that is established and known to work
1808 well in all environments. It organizes cache hot objects in
1809 per cpu and per node queues.
1812 bool "SLUB (Unqueued Allocator)"
1813 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1815 SLUB is a slab allocator that minimizes cache line usage
1816 instead of managing queues of cached objects (SLAB approach).
1817 Per cpu caching is realized using slabs of objects instead
1818 of queues of objects. SLUB can use memory efficiently
1819 and has enhanced diagnostics. SLUB is the default choice for
1824 bool "SLOB (Simple Allocator)"
1826 SLOB replaces the stock allocator with a drastically simpler
1827 allocator. SLOB is generally more space efficient but
1828 does not perform as well on large systems.
1832 config SLAB_MERGE_DEFAULT
1833 bool "Allow slab caches to be merged"
1836 For reduced kernel memory fragmentation, slab caches can be
1837 merged when they share the same size and other characteristics.
1838 This carries a risk of kernel heap overflows being able to
1839 overwrite objects from merged caches (and more easily control
1840 cache layout), which makes such heap attacks easier to exploit
1841 by attackers. By keeping caches unmerged, these kinds of exploits
1842 can usually only damage objects in the same cache. To disable
1843 merging at runtime, "slab_nomerge" can be passed on the kernel
1846 config SLAB_FREELIST_RANDOM
1848 depends on SLAB || SLUB
1849 bool "SLAB freelist randomization"
1851 Randomizes the freelist order used on creating new pages. This
1852 security feature reduces the predictability of the kernel slab
1853 allocator against heap overflows.
1855 config SLAB_FREELIST_HARDENED
1856 bool "Harden slab freelist metadata"
1859 Many kernel heap attacks try to target slab cache metadata and
1860 other infrastructure. This options makes minor performance
1861 sacrifices to harden the kernel slab allocator against common
1862 freelist exploit methods.
1864 config SHUFFLE_PAGE_ALLOCATOR
1865 bool "Page allocator randomization"
1866 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1868 Randomization of the page allocator improves the average
1869 utilization of a direct-mapped memory-side-cache. See section
1870 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1871 6.2a specification for an example of how a platform advertises
1872 the presence of a memory-side-cache. There are also incidental
1873 security benefits as it reduces the predictability of page
1874 allocations to compliment SLAB_FREELIST_RANDOM, but the
1875 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1876 10th order of pages is selected based on cache utilization
1879 While the randomization improves cache utilization it may
1880 negatively impact workloads on platforms without a cache. For
1881 this reason, by default, the randomization is enabled only
1882 after runtime detection of a direct-mapped memory-side-cache.
1883 Otherwise, the randomization may be force enabled with the
1884 'page_alloc.shuffle' kernel command line parameter.
1888 config SLUB_CPU_PARTIAL
1890 depends on SLUB && SMP
1891 bool "SLUB per cpu partial cache"
1893 Per cpu partial caches accelerate objects allocation and freeing
1894 that is local to a processor at the price of more indeterminism
1895 in the latency of the free. On overflow these caches will be cleared
1896 which requires the taking of locks that may cause latency spikes.
1897 Typically one would choose no for a realtime system.
1899 config MMAP_ALLOW_UNINITIALIZED
1900 bool "Allow mmapped anonymous memory to be uninitialized"
1901 depends on EXPERT && !MMU
1904 Normally, and according to the Linux spec, anonymous memory obtained
1905 from mmap() has its contents cleared before it is passed to
1906 userspace. Enabling this config option allows you to request that
1907 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1908 providing a huge performance boost. If this option is not enabled,
1909 then the flag will be ignored.
1911 This is taken advantage of by uClibc's malloc(), and also by
1912 ELF-FDPIC binfmt's brk and stack allocator.
1914 Because of the obvious security issues, this option should only be
1915 enabled on embedded devices where you control what is run in
1916 userspace. Since that isn't generally a problem on no-MMU systems,
1917 it is normally safe to say Y here.
1919 See Documentation/nommu-mmap.txt for more information.
1921 config SYSTEM_DATA_VERIFICATION
1923 select SYSTEM_TRUSTED_KEYRING
1927 select ASYMMETRIC_KEY_TYPE
1928 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1931 select X509_CERTIFICATE_PARSER
1932 select PKCS7_MESSAGE_PARSER
1934 Provide PKCS#7 message verification using the contents of the system
1935 trusted keyring to provide public keys. This then can be used for
1936 module verification, kexec image verification and firmware blob
1940 bool "Profiling support"
1942 Say Y here to enable the extended profiling support mechanisms used
1943 by profilers such as OProfile.
1946 # Place an empty function call at each tracepoint site. Can be
1947 # dynamically changed for a probe function.
1952 endmenu # General setup
1954 source "arch/Kconfig"
1961 default 0 if BASE_FULL
1962 default 1 if !BASE_FULL
1965 bool "Enable loadable module support"
1968 Kernel modules are small pieces of compiled code which can
1969 be inserted in the running kernel, rather than being
1970 permanently built into the kernel. You use the "modprobe"
1971 tool to add (and sometimes remove) them. If you say Y here,
1972 many parts of the kernel can be built as modules (by
1973 answering M instead of Y where indicated): this is most
1974 useful for infrequently used options which are not required
1975 for booting. For more information, see the man pages for
1976 modprobe, lsmod, modinfo, insmod and rmmod.
1978 If you say Y here, you will need to run "make
1979 modules_install" to put the modules under /lib/modules/
1980 where modprobe can find them (you may need to be root to do
1987 config MODULE_FORCE_LOAD
1988 bool "Forced module loading"
1991 Allow loading of modules without version information (ie. modprobe
1992 --force). Forced module loading sets the 'F' (forced) taint flag and
1993 is usually a really bad idea.
1995 config MODULE_UNLOAD
1996 bool "Module unloading"
1998 Without this option you will not be able to unload any
1999 modules (note that some modules may not be unloadable
2000 anyway), which makes your kernel smaller, faster
2001 and simpler. If unsure, say Y.
2003 config MODULE_FORCE_UNLOAD
2004 bool "Forced module unloading"
2005 depends on MODULE_UNLOAD
2007 This option allows you to force a module to unload, even if the
2008 kernel believes it is unsafe: the kernel will remove the module
2009 without waiting for anyone to stop using it (using the -f option to
2010 rmmod). This is mainly for kernel developers and desperate users.
2014 bool "Module versioning support"
2016 Usually, you have to use modules compiled with your kernel.
2017 Saying Y here makes it sometimes possible to use modules
2018 compiled for different kernels, by adding enough information
2019 to the modules to (hopefully) spot any changes which would
2020 make them incompatible with the kernel you are running. If
2023 config ASM_MODVERSIONS
2025 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2027 This enables module versioning for exported symbols also from
2028 assembly. This can be enabled only when the target architecture
2031 config MODULE_REL_CRCS
2033 depends on MODVERSIONS
2035 config MODULE_SRCVERSION_ALL
2036 bool "Source checksum for all modules"
2038 Modules which contain a MODULE_VERSION get an extra "srcversion"
2039 field inserted into their modinfo section, which contains a
2040 sum of the source files which made it. This helps maintainers
2041 see exactly which source was used to build a module (since
2042 others sometimes change the module source without updating
2043 the version). With this option, such a "srcversion" field
2044 will be created for all modules. If unsure, say N.
2047 bool "Module signature verification"
2049 select SYSTEM_DATA_VERIFICATION
2051 Check modules for valid signatures upon load: the signature
2052 is simply appended to the module. For more information see
2053 <file:Documentation/admin-guide/module-signing.rst>.
2055 Note that this option adds the OpenSSL development packages as a
2056 kernel build dependency so that the signing tool can use its crypto
2059 !!!WARNING!!! If you enable this option, you MUST make sure that the
2060 module DOES NOT get stripped after being signed. This includes the
2061 debuginfo strip done by some packagers (such as rpmbuild) and
2062 inclusion into an initramfs that wants the module size reduced.
2064 config MODULE_SIG_FORCE
2065 bool "Require modules to be validly signed"
2066 depends on MODULE_SIG
2068 Reject unsigned modules or signed modules for which we don't have a
2069 key. Without this, such modules will simply taint the kernel.
2071 config MODULE_SIG_ALL
2072 bool "Automatically sign all modules"
2074 depends on MODULE_SIG
2076 Sign all modules during make modules_install. Without this option,
2077 modules must be signed manually, using the scripts/sign-file tool.
2079 comment "Do not forget to sign required modules with scripts/sign-file"
2080 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2083 prompt "Which hash algorithm should modules be signed with?"
2084 depends on MODULE_SIG
2086 This determines which sort of hashing algorithm will be used during
2087 signature generation. This algorithm _must_ be built into the kernel
2088 directly so that signature verification can take place. It is not
2089 possible to load a signed module containing the algorithm to check
2090 the signature on that module.
2092 config MODULE_SIG_SHA1
2093 bool "Sign modules with SHA-1"
2096 config MODULE_SIG_SHA224
2097 bool "Sign modules with SHA-224"
2098 select CRYPTO_SHA256
2100 config MODULE_SIG_SHA256
2101 bool "Sign modules with SHA-256"
2102 select CRYPTO_SHA256
2104 config MODULE_SIG_SHA384
2105 bool "Sign modules with SHA-384"
2106 select CRYPTO_SHA512
2108 config MODULE_SIG_SHA512
2109 bool "Sign modules with SHA-512"
2110 select CRYPTO_SHA512
2114 config MODULE_SIG_HASH
2116 depends on MODULE_SIG
2117 default "sha1" if MODULE_SIG_SHA1
2118 default "sha224" if MODULE_SIG_SHA224
2119 default "sha256" if MODULE_SIG_SHA256
2120 default "sha384" if MODULE_SIG_SHA384
2121 default "sha512" if MODULE_SIG_SHA512
2123 config MODULE_COMPRESS
2124 bool "Compress modules on installation"
2128 Compresses kernel modules when 'make modules_install' is run; gzip or
2129 xz depending on "Compression algorithm" below.
2131 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2133 Out-of-tree kernel modules installed using Kbuild will also be
2134 compressed upon installation.
2136 Note: for modules inside an initrd or initramfs, it's more efficient
2137 to compress the whole initrd or initramfs instead.
2139 Note: This is fully compatible with signed modules.
2144 prompt "Compression algorithm"
2145 depends on MODULE_COMPRESS
2146 default MODULE_COMPRESS_GZIP
2148 This determines which sort of compression will be used during
2149 'make modules_install'.
2151 GZIP (default) and XZ are supported.
2153 config MODULE_COMPRESS_GZIP
2156 config MODULE_COMPRESS_XZ
2161 config TRIM_UNUSED_KSYMS
2162 bool "Trim unused exported kernel symbols"
2163 depends on MODULES && !UNUSED_SYMBOLS
2165 The kernel and some modules make many symbols available for
2166 other modules to use via EXPORT_SYMBOL() and variants. Depending
2167 on the set of modules being selected in your kernel configuration,
2168 many of those exported symbols might never be used.
2170 This option allows for unused exported symbols to be dropped from
2171 the build. In turn, this provides the compiler more opportunities
2172 (especially when using LTO) for optimizing the code and reducing
2173 binary size. This might have some security advantages as well.
2175 If unsure, or if you need to build out-of-tree modules, say N.
2179 config MODULES_TREE_LOOKUP
2181 depends on PERF_EVENTS || TRACING
2183 config INIT_ALL_POSSIBLE
2186 Back when each arch used to define their own cpu_online_mask and
2187 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2188 with all 1s, and others with all 0s. When they were centralised,
2189 it was better to provide this option than to break all the archs
2190 and have several arch maintainers pursuing me down dark alleys.
2192 source "block/Kconfig"
2194 config PREEMPT_NOTIFIERS
2204 Build a simple ASN.1 grammar compiler that produces a bytecode output
2205 that can be interpreted by the ASN.1 stream decoder and used to
2206 inform it as to what tags are to be expected in a stream and what
2207 functions to call on what tags.
2209 source "kernel/Kconfig.locks"
2211 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2214 # It may be useful for an architecture to override the definitions of the
2215 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2216 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2217 # different calling convention for syscalls. They can also override the
2218 # macros for not-implemented syscalls in kernel/sys_ni.c and
2219 # kernel/time/posix-stubs.c. All these overrides need to be available in
2220 # <asm/syscall_wrapper.h>.
2221 config ARCH_HAS_SYSCALL_WRAPPER
git.samba.org / sfrench / cifs-2.6.git / blob