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)"
11 config CC_VERSION_TEXT
13 default "$(CC_VERSION_TEXT)"
15 This is used in unclear ways:
17 - Re-run Kconfig when the compiler is updated
18 The 'default' property references the environment variable,
19 CC_VERSION_TEXT so it is recorded in include/config/auto.conf.cmd.
20 When the compiler is updated, Kconfig will be invoked.
22 - Ensure full rebuild when the compier is updated
23 include/linux/kconfig.h contains this option in the comment line so
24 fixdep adds include/config/cc/version/text.h into the auto-generated
25 dependency. When the compiler is updated, syncconfig will touch it
26 and then every file will be rebuilt.
29 def_bool $(success,echo "$(CC_VERSION_TEXT)" | grep -q gcc)
33 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
38 default $(shell,$(LD) --version | $(srctree)/scripts/ld-version.sh)
41 def_bool $(success,echo "$(CC_VERSION_TEXT)" | grep -q clang)
44 def_bool $(success,$(LD) -v | head -n 1 | grep -q LLD)
48 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
52 default $(shell,$(srctree)/scripts/lld-version.sh $(LD))
56 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
57 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
59 config CC_CAN_LINK_STATIC
61 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
62 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
64 config CC_HAS_ASM_GOTO
65 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
67 config CC_HAS_ASM_GOTO_OUTPUT
68 depends on CC_HAS_ASM_GOTO
69 def_bool $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
71 config TOOLS_SUPPORT_RELR
72 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
74 config CC_HAS_ASM_INLINE
75 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
83 config BUILDTIME_TABLE_SORT
86 config THREAD_INFO_IN_TASK
89 Select this to move thread_info off the stack into task_struct. To
90 make this work, an arch will need to remove all thread_info fields
91 except flags and fix any runtime bugs.
93 One subtle change that will be needed is to use try_get_task_stack()
94 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
103 depends on BROKEN || !SMP
106 config INIT_ENV_ARG_LIMIT
111 Maximum of each of the number of arguments and environment
112 variables passed to init from the kernel command line.
115 bool "Compile also drivers which will not load"
116 depends on !UML && !S390
119 Some drivers can be compiled on a different platform than they are
120 intended to be run on. Despite they cannot be loaded there (or even
121 when they load they cannot be used due to missing HW support),
122 developers still, opposing to distributors, might want to build such
123 drivers to compile-test them.
125 If you are a developer and want to build everything available, say Y
126 here. If you are a user/distributor, say N here to exclude useless
127 drivers to be distributed.
129 config UAPI_HEADER_TEST
130 bool "Compile test UAPI headers"
131 depends on HEADERS_INSTALL && CC_CAN_LINK
133 Compile test headers exported to user-space to ensure they are
134 self-contained, i.e. compilable as standalone units.
136 If you are a developer or tester and want to ensure the exported
137 headers are self-contained, say Y here. Otherwise, choose N.
140 string "Local version - append to kernel release"
142 Append an extra string to the end of your kernel version.
143 This will show up when you type uname, for example.
144 The string you set here will be appended after the contents of
145 any files with a filename matching localversion* in your
146 object and source tree, in that order. Your total string can
147 be a maximum of 64 characters.
149 config LOCALVERSION_AUTO
150 bool "Automatically append version information to the version string"
152 depends on !COMPILE_TEST
154 This will try to automatically determine if the current tree is a
155 release tree by looking for git tags that belong to the current
156 top of tree revision.
158 A string of the format -gxxxxxxxx will be added to the localversion
159 if a git-based tree is found. The string generated by this will be
160 appended after any matching localversion* files, and after the value
161 set in CONFIG_LOCALVERSION.
163 (The actual string used here is the first eight characters produced
164 by running the command:
166 $ git rev-parse --verify HEAD
168 which is done within the script "scripts/setlocalversion".)
171 string "Build ID Salt"
174 The build ID is used to link binaries and their debug info. Setting
175 this option will use the value in the calculation of the build id.
176 This is mostly useful for distributions which want to ensure the
177 build is unique between builds. It's safe to leave the default.
179 config HAVE_KERNEL_GZIP
182 config HAVE_KERNEL_BZIP2
185 config HAVE_KERNEL_LZMA
188 config HAVE_KERNEL_XZ
191 config HAVE_KERNEL_LZO
194 config HAVE_KERNEL_LZ4
197 config HAVE_KERNEL_ZSTD
200 config HAVE_KERNEL_UNCOMPRESSED
204 prompt "Kernel compression mode"
206 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_ZSTD || HAVE_KERNEL_UNCOMPRESSED
208 The linux kernel is a kind of self-extracting executable.
209 Several compression algorithms are available, which differ
210 in efficiency, compression and decompression speed.
211 Compression speed is only relevant when building a kernel.
212 Decompression speed is relevant at each boot.
214 If you have any problems with bzip2 or lzma compressed
215 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
216 version of this functionality (bzip2 only), for 2.4, was
217 supplied by Christian Ludwig)
219 High compression options are mostly useful for users, who
220 are low on disk space (embedded systems), but for whom ram
223 If in doubt, select 'gzip'
227 depends on HAVE_KERNEL_GZIP
229 The old and tried gzip compression. It provides a good balance
230 between compression ratio and decompression speed.
234 depends on HAVE_KERNEL_BZIP2
236 Its compression ratio and speed is intermediate.
237 Decompression speed is slowest among the choices. The kernel
238 size is about 10% smaller with bzip2, in comparison to gzip.
239 Bzip2 uses a large amount of memory. For modern kernels you
240 will need at least 8MB RAM or more for booting.
244 depends on HAVE_KERNEL_LZMA
246 This compression algorithm's ratio is best. Decompression speed
247 is between gzip and bzip2. Compression is slowest.
248 The kernel size is about 33% smaller with LZMA in comparison to gzip.
252 depends on HAVE_KERNEL_XZ
254 XZ uses the LZMA2 algorithm and instruction set specific
255 BCJ filters which can improve compression ratio of executable
256 code. The size of the kernel is about 30% smaller with XZ in
257 comparison to gzip. On architectures for which there is a BCJ
258 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
259 will create a few percent smaller kernel than plain LZMA.
261 The speed is about the same as with LZMA: The decompression
262 speed of XZ is better than that of bzip2 but worse than gzip
263 and LZO. Compression is slow.
267 depends on HAVE_KERNEL_LZO
269 Its compression ratio is the poorest among the choices. The kernel
270 size is about 10% bigger than gzip; however its speed
271 (both compression and decompression) is the fastest.
275 depends on HAVE_KERNEL_LZ4
277 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
278 A preliminary version of LZ4 de/compression tool is available at
279 <https://code.google.com/p/lz4/>.
281 Its compression ratio is worse than LZO. The size of the kernel
282 is about 8% bigger than LZO. But the decompression speed is
287 depends on HAVE_KERNEL_ZSTD
289 ZSTD is a compression algorithm targeting intermediate compression
290 with fast decompression speed. It will compress better than GZIP and
291 decompress around the same speed as LZO, but slower than LZ4. You
292 will need at least 192 KB RAM or more for booting. The zstd command
293 line tool is required for compression.
295 config KERNEL_UNCOMPRESSED
297 depends on HAVE_KERNEL_UNCOMPRESSED
299 Produce uncompressed kernel image. This option is usually not what
300 you want. It is useful for debugging the kernel in slow simulation
301 environments, where decompressing and moving the kernel is awfully
302 slow. This option allows early boot code to skip the decompressor
303 and jump right at uncompressed kernel image.
308 string "Default init path"
311 This option determines the default init for the system if no init=
312 option is passed on the kernel command line. If the requested path is
313 not present, we will still then move on to attempting further
314 locations (e.g. /sbin/init, etc). If this is empty, we will just use
315 the fallback list when init= is not passed.
317 config DEFAULT_HOSTNAME
318 string "Default hostname"
321 This option determines the default system hostname before userspace
322 calls sethostname(2). The kernel traditionally uses "(none)" here,
323 but you may wish to use a different default here to make a minimal
324 system more usable with less configuration.
327 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
328 # add proper SWAP support to them, in which case this can be remove.
334 bool "Support for paging of anonymous memory (swap)"
335 depends on MMU && BLOCK && !ARCH_NO_SWAP
338 This option allows you to choose whether you want to have support
339 for so called swap devices or swap files in your kernel that are
340 used to provide more virtual memory than the actual RAM present
341 in your computer. If unsure say Y.
346 Inter Process Communication is a suite of library functions and
347 system calls which let processes (running programs) synchronize and
348 exchange information. It is generally considered to be a good thing,
349 and some programs won't run unless you say Y here. In particular, if
350 you want to run the DOS emulator dosemu under Linux (read the
351 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
352 you'll need to say Y here.
354 You can find documentation about IPC with "info ipc" and also in
355 section 6.4 of the Linux Programmer's Guide, available from
356 <http://www.tldp.org/guides.html>.
358 config SYSVIPC_SYSCTL
365 bool "POSIX Message Queues"
368 POSIX variant of message queues is a part of IPC. In POSIX message
369 queues every message has a priority which decides about succession
370 of receiving it by a process. If you want to compile and run
371 programs written e.g. for Solaris with use of its POSIX message
372 queues (functions mq_*) say Y here.
374 POSIX message queues are visible as a filesystem called 'mqueue'
375 and can be mounted somewhere if you want to do filesystem
376 operations on message queues.
380 config POSIX_MQUEUE_SYSCTL
382 depends on POSIX_MQUEUE
387 bool "General notification queue"
391 This is a general notification queue for the kernel to pass events to
392 userspace by splicing them into pipes. It can be used in conjunction
393 with watches for key/keyring change notifications and device
396 See Documentation/watch_queue.rst
398 config CROSS_MEMORY_ATTACH
399 bool "Enable process_vm_readv/writev syscalls"
403 Enabling this option adds the system calls process_vm_readv and
404 process_vm_writev which allow a process with the correct privileges
405 to directly read from or write to another process' address space.
406 See the man page for more details.
409 bool "uselib syscall"
410 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
412 This option enables the uselib syscall, a system call used in the
413 dynamic linker from libc5 and earlier. glibc does not use this
414 system call. If you intend to run programs built on libc5 or
415 earlier, you may need to enable this syscall. Current systems
416 running glibc can safely disable this.
419 bool "Auditing support"
422 Enable auditing infrastructure that can be used with another
423 kernel subsystem, such as SELinux (which requires this for
424 logging of avc messages output). System call auditing is included
425 on architectures which support it.
427 config HAVE_ARCH_AUDITSYSCALL
432 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
435 source "kernel/irq/Kconfig"
436 source "kernel/time/Kconfig"
437 source "kernel/Kconfig.preempt"
439 menu "CPU/Task time and stats accounting"
441 config VIRT_CPU_ACCOUNTING
445 prompt "Cputime accounting"
446 default TICK_CPU_ACCOUNTING if !PPC64
447 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
449 # Kind of a stub config for the pure tick based cputime accounting
450 config TICK_CPU_ACCOUNTING
451 bool "Simple tick based cputime accounting"
452 depends on !S390 && !NO_HZ_FULL
454 This is the basic tick based cputime accounting that maintains
455 statistics about user, system and idle time spent on per jiffies
460 config VIRT_CPU_ACCOUNTING_NATIVE
461 bool "Deterministic task and CPU time accounting"
462 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
463 select VIRT_CPU_ACCOUNTING
465 Select this option to enable more accurate task and CPU time
466 accounting. This is done by reading a CPU counter on each
467 kernel entry and exit and on transitions within the kernel
468 between system, softirq and hardirq state, so there is a
469 small performance impact. In the case of s390 or IBM POWER > 5,
470 this also enables accounting of stolen time on logically-partitioned
473 config VIRT_CPU_ACCOUNTING_GEN
474 bool "Full dynticks CPU time accounting"
475 depends on HAVE_CONTEXT_TRACKING
476 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
477 depends on GENERIC_CLOCKEVENTS
478 select VIRT_CPU_ACCOUNTING
479 select CONTEXT_TRACKING
481 Select this option to enable task and CPU time accounting on full
482 dynticks systems. This accounting is implemented by watching every
483 kernel-user boundaries using the context tracking subsystem.
484 The accounting is thus performed at the expense of some significant
487 For now this is only useful if you are working on the full
488 dynticks subsystem development.
494 config IRQ_TIME_ACCOUNTING
495 bool "Fine granularity task level IRQ time accounting"
496 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
498 Select this option to enable fine granularity task irq time
499 accounting. This is done by reading a timestamp on each
500 transitions between softirq and hardirq state, so there can be a
501 small performance impact.
503 If in doubt, say N here.
505 config HAVE_SCHED_AVG_IRQ
507 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
510 config SCHED_THERMAL_PRESSURE
512 default y if ARM && ARM_CPU_TOPOLOGY
515 depends on CPU_FREQ_THERMAL
517 Select this option to enable thermal pressure accounting in the
518 scheduler. Thermal pressure is the value conveyed to the scheduler
519 that reflects the reduction in CPU compute capacity resulted from
520 thermal throttling. Thermal throttling occurs when the performance of
521 a CPU is capped due to high operating temperatures.
523 If selected, the scheduler will be able to balance tasks accordingly,
524 i.e. put less load on throttled CPUs than on non/less throttled ones.
526 This requires the architecture to implement
527 arch_set_thermal_pressure() and arch_scale_thermal_pressure().
529 config BSD_PROCESS_ACCT
530 bool "BSD Process Accounting"
533 If you say Y here, a user level program will be able to instruct the
534 kernel (via a special system call) to write process accounting
535 information to a file: whenever a process exits, information about
536 that process will be appended to the file by the kernel. The
537 information includes things such as creation time, owning user,
538 command name, memory usage, controlling terminal etc. (the complete
539 list is in the struct acct in <file:include/linux/acct.h>). It is
540 up to the user level program to do useful things with this
541 information. This is generally a good idea, so say Y.
543 config BSD_PROCESS_ACCT_V3
544 bool "BSD Process Accounting version 3 file format"
545 depends on BSD_PROCESS_ACCT
548 If you say Y here, the process accounting information is written
549 in a new file format that also logs the process IDs of each
550 process and its parent. Note that this file format is incompatible
551 with previous v0/v1/v2 file formats, so you will need updated tools
552 for processing it. A preliminary version of these tools is available
553 at <http://www.gnu.org/software/acct/>.
556 bool "Export task/process statistics through netlink"
561 Export selected statistics for tasks/processes through the
562 generic netlink interface. Unlike BSD process accounting, the
563 statistics are available during the lifetime of tasks/processes as
564 responses to commands. Like BSD accounting, they are sent to user
569 config TASK_DELAY_ACCT
570 bool "Enable per-task delay accounting"
574 Collect information on time spent by a task waiting for system
575 resources like cpu, synchronous block I/O completion and swapping
576 in pages. Such statistics can help in setting a task's priorities
577 relative to other tasks for cpu, io, rss limits etc.
582 bool "Enable extended accounting over taskstats"
585 Collect extended task accounting data and send the data
586 to userland for processing over the taskstats interface.
590 config TASK_IO_ACCOUNTING
591 bool "Enable per-task storage I/O accounting"
592 depends on TASK_XACCT
594 Collect information on the number of bytes of storage I/O which this
600 bool "Pressure stall information tracking"
602 Collect metrics that indicate how overcommitted the CPU, memory,
603 and IO capacity are in the system.
605 If you say Y here, the kernel will create /proc/pressure/ with the
606 pressure statistics files cpu, memory, and io. These will indicate
607 the share of walltime in which some or all tasks in the system are
608 delayed due to contention of the respective resource.
610 In kernels with cgroup support, cgroups (cgroup2 only) will
611 have cpu.pressure, memory.pressure, and io.pressure files,
612 which aggregate pressure stalls for the grouped tasks only.
614 For more details see Documentation/accounting/psi.rst.
618 config PSI_DEFAULT_DISABLED
619 bool "Require boot parameter to enable pressure stall information tracking"
623 If set, pressure stall information tracking will be disabled
624 per default but can be enabled through passing psi=1 on the
625 kernel commandline during boot.
627 This feature adds some code to the task wakeup and sleep
628 paths of the scheduler. The overhead is too low to affect
629 common scheduling-intense workloads in practice (such as
630 webservers, memcache), but it does show up in artificial
631 scheduler stress tests, such as hackbench.
633 If you are paranoid and not sure what the kernel will be
638 endmenu # "CPU/Task time and stats accounting"
642 depends on SMP || COMPILE_TEST
645 Make sure that CPUs running critical tasks are not disturbed by
646 any source of "noise" such as unbound workqueues, timers, kthreads...
647 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
648 the "isolcpus=" boot parameter.
652 source "kernel/rcu/Kconfig"
659 tristate "Kernel .config support"
661 This option enables the complete Linux kernel ".config" file
662 contents to be saved in the kernel. It provides documentation
663 of which kernel options are used in a running kernel or in an
664 on-disk kernel. This information can be extracted from the kernel
665 image file with the script scripts/extract-ikconfig and used as
666 input to rebuild the current kernel or to build another kernel.
667 It can also be extracted from a running kernel by reading
668 /proc/config.gz if enabled (below).
671 bool "Enable access to .config through /proc/config.gz"
672 depends on IKCONFIG && PROC_FS
674 This option enables access to the kernel configuration file
675 through /proc/config.gz.
678 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
681 This option enables access to the in-kernel headers that are generated during
682 the build process. These can be used to build eBPF tracing programs,
683 or similar programs. If you build the headers as a module, a module called
684 kheaders.ko is built which can be loaded on-demand to get access to headers.
687 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
688 range 12 25 if !H8300
693 Select the minimal kernel log buffer size as a power of 2.
694 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
695 parameter, see below. Any higher size also might be forced
696 by "log_buf_len" boot parameter.
706 config LOG_CPU_MAX_BUF_SHIFT
707 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
710 default 12 if !BASE_SMALL
711 default 0 if BASE_SMALL
714 This option allows to increase the default ring buffer size
715 according to the number of CPUs. The value defines the contribution
716 of each CPU as a power of 2. The used space is typically only few
717 lines however it might be much more when problems are reported,
720 The increased size means that a new buffer has to be allocated and
721 the original static one is unused. It makes sense only on systems
722 with more CPUs. Therefore this value is used only when the sum of
723 contributions is greater than the half of the default kernel ring
724 buffer as defined by LOG_BUF_SHIFT. The default values are set
725 so that more than 16 CPUs are needed to trigger the allocation.
727 Also this option is ignored when "log_buf_len" kernel parameter is
728 used as it forces an exact (power of two) size of the ring buffer.
730 The number of possible CPUs is used for this computation ignoring
731 hotplugging making the computation optimal for the worst case
732 scenario while allowing a simple algorithm to be used from bootup.
734 Examples shift values and their meaning:
735 17 => 128 KB for each CPU
736 16 => 64 KB for each CPU
737 15 => 32 KB for each CPU
738 14 => 16 KB for each CPU
739 13 => 8 KB for each CPU
740 12 => 4 KB for each CPU
742 config PRINTK_SAFE_LOG_BUF_SHIFT
743 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
748 Select the size of an alternate printk per-CPU buffer where messages
749 printed from usafe contexts are temporary stored. One example would
750 be NMI messages, another one - printk recursion. The messages are
751 copied to the main log buffer in a safe context to avoid a deadlock.
752 The value defines the size as a power of 2.
754 Those messages are rare and limited. The largest one is when
755 a backtrace is printed. It usually fits into 4KB. Select
756 8KB if you want to be on the safe side.
759 17 => 128 KB for each CPU
760 16 => 64 KB for each CPU
761 15 => 32 KB for each CPU
762 14 => 16 KB for each CPU
763 13 => 8 KB for each CPU
764 12 => 4 KB for each CPU
767 # Architectures with an unreliable sched_clock() should select this:
769 config HAVE_UNSTABLE_SCHED_CLOCK
772 config GENERIC_SCHED_CLOCK
775 menu "Scheduler features"
778 bool "Enable utilization clamping for RT/FAIR tasks"
779 depends on CPU_FREQ_GOV_SCHEDUTIL
781 This feature enables the scheduler to track the clamped utilization
782 of each CPU based on RUNNABLE tasks scheduled on that CPU.
784 With this option, the user can specify the min and max CPU
785 utilization allowed for RUNNABLE tasks. The max utilization defines
786 the maximum frequency a task should use while the min utilization
787 defines the minimum frequency it should use.
789 Both min and max utilization clamp values are hints to the scheduler,
790 aiming at improving its frequency selection policy, but they do not
791 enforce or grant any specific bandwidth for tasks.
795 config UCLAMP_BUCKETS_COUNT
796 int "Number of supported utilization clamp buckets"
799 depends on UCLAMP_TASK
801 Defines the number of clamp buckets to use. The range of each bucket
802 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
803 number of clamp buckets the finer their granularity and the higher
804 the precision of clamping aggregation and tracking at run-time.
806 For example, with the minimum configuration value we will have 5
807 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
808 be refcounted in the [20..39]% bucket and will set the bucket clamp
809 effective value to 25%.
810 If a second 30% boosted task should be co-scheduled on the same CPU,
811 that task will be refcounted in the same bucket of the first task and
812 it will boost the bucket clamp effective value to 30%.
813 The clamp effective value of a bucket is reset to its nominal value
814 (20% in the example above) when there are no more tasks refcounted in
817 An additional boost/capping margin can be added to some tasks. In the
818 example above the 25% task will be boosted to 30% until it exits the
819 CPU. If that should be considered not acceptable on certain systems,
820 it's always possible to reduce the margin by increasing the number of
821 clamp buckets to trade off used memory for run-time tracking
824 If in doubt, use the default value.
829 # For architectures that want to enable the support for NUMA-affine scheduler
832 config ARCH_SUPPORTS_NUMA_BALANCING
836 # For architectures that prefer to flush all TLBs after a number of pages
837 # are unmapped instead of sending one IPI per page to flush. The architecture
838 # must provide guarantees on what happens if a clean TLB cache entry is
839 # written after the unmap. Details are in mm/rmap.c near the check for
840 # should_defer_flush. The architecture should also consider if the full flush
841 # and the refill costs are offset by the savings of sending fewer IPIs.
842 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
846 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
849 # For architectures that know their GCC __int128 support is sound
851 config ARCH_SUPPORTS_INT128
854 # For architectures that (ab)use NUMA to represent different memory regions
855 # all cpu-local but of different latencies, such as SuperH.
857 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
860 config NUMA_BALANCING
861 bool "Memory placement aware NUMA scheduler"
862 depends on ARCH_SUPPORTS_NUMA_BALANCING
863 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
864 depends on SMP && NUMA && MIGRATION
866 This option adds support for automatic NUMA aware memory/task placement.
867 The mechanism is quite primitive and is based on migrating memory when
868 it has references to the node the task is running on.
870 This system will be inactive on UMA systems.
872 config NUMA_BALANCING_DEFAULT_ENABLED
873 bool "Automatically enable NUMA aware memory/task placement"
875 depends on NUMA_BALANCING
877 If set, automatic NUMA balancing will be enabled if running on a NUMA
881 bool "Control Group support"
884 This option adds support for grouping sets of processes together, for
885 use with process control subsystems such as Cpusets, CFS, memory
886 controls or device isolation.
888 - Documentation/scheduler/sched-design-CFS.rst (CFS)
889 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
890 and resource control)
900 bool "Memory controller"
904 Provides control over the memory footprint of tasks in a cgroup.
908 depends on MEMCG && SWAP
913 depends on MEMCG && !SLOB
921 Generic block IO controller cgroup interface. This is the common
922 cgroup interface which should be used by various IO controlling
925 Currently, CFQ IO scheduler uses it to recognize task groups and
926 control disk bandwidth allocation (proportional time slice allocation)
927 to such task groups. It is also used by bio throttling logic in
928 block layer to implement upper limit in IO rates on a device.
930 This option only enables generic Block IO controller infrastructure.
931 One needs to also enable actual IO controlling logic/policy. For
932 enabling proportional weight division of disk bandwidth in CFQ, set
933 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
934 CONFIG_BLK_DEV_THROTTLING=y.
936 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
938 config CGROUP_WRITEBACK
940 depends on MEMCG && BLK_CGROUP
943 menuconfig CGROUP_SCHED
944 bool "CPU controller"
947 This feature lets CPU scheduler recognize task groups and control CPU
948 bandwidth allocation to such task groups. It uses cgroups to group
952 config FAIR_GROUP_SCHED
953 bool "Group scheduling for SCHED_OTHER"
954 depends on CGROUP_SCHED
958 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
959 depends on FAIR_GROUP_SCHED
962 This option allows users to define CPU bandwidth rates (limits) for
963 tasks running within the fair group scheduler. Groups with no limit
964 set are considered to be unconstrained and will run with no
966 See Documentation/scheduler/sched-bwc.rst for more information.
968 config RT_GROUP_SCHED
969 bool "Group scheduling for SCHED_RR/FIFO"
970 depends on CGROUP_SCHED
973 This feature lets you explicitly allocate real CPU bandwidth
974 to task groups. If enabled, it will also make it impossible to
975 schedule realtime tasks for non-root users until you allocate
976 realtime bandwidth for them.
977 See Documentation/scheduler/sched-rt-group.rst for more information.
981 config UCLAMP_TASK_GROUP
982 bool "Utilization clamping per group of tasks"
983 depends on CGROUP_SCHED
984 depends on UCLAMP_TASK
987 This feature enables the scheduler to track the clamped utilization
988 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
990 When this option is enabled, the user can specify a min and max
991 CPU bandwidth which is allowed for each single task in a group.
992 The max bandwidth allows to clamp the maximum frequency a task
993 can use, while the min bandwidth allows to define a minimum
994 frequency a task will always use.
996 When task group based utilization clamping is enabled, an eventually
997 specified task-specific clamp value is constrained by the cgroup
998 specified clamp value. Both minimum and maximum task clamping cannot
999 be bigger than the corresponding clamping defined at task group level.
1004 bool "PIDs controller"
1006 Provides enforcement of process number limits in the scope of a
1007 cgroup. Any attempt to fork more processes than is allowed in the
1008 cgroup will fail. PIDs are fundamentally a global resource because it
1009 is fairly trivial to reach PID exhaustion before you reach even a
1010 conservative kmemcg limit. As a result, it is possible to grind a
1011 system to halt without being limited by other cgroup policies. The
1012 PIDs controller is designed to stop this from happening.
1014 It should be noted that organisational operations (such as attaching
1015 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1016 since the PIDs limit only affects a process's ability to fork, not to
1020 bool "RDMA controller"
1022 Provides enforcement of RDMA resources defined by IB stack.
1023 It is fairly easy for consumers to exhaust RDMA resources, which
1024 can result into resource unavailability to other consumers.
1025 RDMA controller is designed to stop this from happening.
1026 Attaching processes with active RDMA resources to the cgroup
1027 hierarchy is allowed even if can cross the hierarchy's limit.
1029 config CGROUP_FREEZER
1030 bool "Freezer controller"
1032 Provides a way to freeze and unfreeze all tasks in a
1035 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1036 controller includes important in-kernel memory consumers per default.
1038 If you're using cgroup2, say N.
1040 config CGROUP_HUGETLB
1041 bool "HugeTLB controller"
1042 depends on HUGETLB_PAGE
1046 Provides a cgroup controller for HugeTLB pages.
1047 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1048 The limit is enforced during page fault. Since HugeTLB doesn't
1049 support page reclaim, enforcing the limit at page fault time implies
1050 that, the application will get SIGBUS signal if it tries to access
1051 HugeTLB pages beyond its limit. This requires the application to know
1052 beforehand how much HugeTLB pages it would require for its use. The
1053 control group is tracked in the third page lru pointer. This means
1054 that we cannot use the controller with huge page less than 3 pages.
1057 bool "Cpuset controller"
1060 This option will let you create and manage CPUSETs which
1061 allow dynamically partitioning a system into sets of CPUs and
1062 Memory Nodes and assigning tasks to run only within those sets.
1063 This is primarily useful on large SMP or NUMA systems.
1067 config PROC_PID_CPUSET
1068 bool "Include legacy /proc/<pid>/cpuset file"
1072 config CGROUP_DEVICE
1073 bool "Device controller"
1075 Provides a cgroup controller implementing whitelists for
1076 devices which a process in the cgroup can mknod or open.
1078 config CGROUP_CPUACCT
1079 bool "Simple CPU accounting controller"
1081 Provides a simple controller for monitoring the
1082 total CPU consumed by the tasks in a cgroup.
1085 bool "Perf controller"
1086 depends on PERF_EVENTS
1088 This option extends the perf per-cpu mode to restrict monitoring
1089 to threads which belong to the cgroup specified and run on the
1090 designated cpu. Or this can be used to have cgroup ID in samples
1091 so that it can monitor performance events among cgroups.
1096 bool "Support for eBPF programs attached to cgroups"
1097 depends on BPF_SYSCALL
1098 select SOCK_CGROUP_DATA
1100 Allow attaching eBPF programs to a cgroup using the bpf(2)
1101 syscall command BPF_PROG_ATTACH.
1103 In which context these programs are accessed depends on the type
1104 of attachment. For instance, programs that are attached using
1105 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1109 bool "Debug controller"
1111 depends on DEBUG_KERNEL
1113 This option enables a simple controller that exports
1114 debugging information about the cgroups framework. This
1115 controller is for control cgroup debugging only. Its
1116 interfaces are not stable.
1120 config SOCK_CGROUP_DATA
1126 menuconfig NAMESPACES
1127 bool "Namespaces support" if EXPERT
1128 depends on MULTIUSER
1131 Provides the way to make tasks work with different objects using
1132 the same id. For example same IPC id may refer to different objects
1133 or same user id or pid may refer to different tasks when used in
1134 different namespaces.
1139 bool "UTS namespace"
1142 In this namespace tasks see different info provided with the
1146 bool "TIME namespace"
1147 depends on GENERIC_VDSO_TIME_NS
1150 In this namespace boottime and monotonic clocks can be set.
1151 The time will keep going with the same pace.
1154 bool "IPC namespace"
1155 depends on (SYSVIPC || POSIX_MQUEUE)
1158 In this namespace tasks work with IPC ids which correspond to
1159 different IPC objects in different namespaces.
1162 bool "User namespace"
1165 This allows containers, i.e. vservers, to use user namespaces
1166 to provide different user info for different servers.
1168 When user namespaces are enabled in the kernel it is
1169 recommended that the MEMCG option also be enabled and that
1170 user-space use the memory control groups to limit the amount
1171 of memory a memory unprivileged users can use.
1176 bool "PID Namespaces"
1179 Support process id namespaces. This allows having multiple
1180 processes with the same pid as long as they are in different
1181 pid namespaces. This is a building block of containers.
1184 bool "Network namespace"
1188 Allow user space to create what appear to be multiple instances
1189 of the network stack.
1193 config CHECKPOINT_RESTORE
1194 bool "Checkpoint/restore support"
1195 select PROC_CHILDREN
1199 Enables additional kernel features in a sake of checkpoint/restore.
1200 In particular it adds auxiliary prctl codes to setup process text,
1201 data and heap segment sizes, and a few additional /proc filesystem
1204 If unsure, say N here.
1206 config SCHED_AUTOGROUP
1207 bool "Automatic process group scheduling"
1210 select FAIR_GROUP_SCHED
1212 This option optimizes the scheduler for common desktop workloads by
1213 automatically creating and populating task groups. This separation
1214 of workloads isolates aggressive CPU burners (like build jobs) from
1215 desktop applications. Task group autogeneration is currently based
1218 config SYSFS_DEPRECATED
1219 bool "Enable deprecated sysfs features to support old userspace tools"
1223 This option adds code that switches the layout of the "block" class
1224 devices, to not show up in /sys/class/block/, but only in
1227 This switch is only active when the sysfs.deprecated=1 boot option is
1228 passed or the SYSFS_DEPRECATED_V2 option is set.
1230 This option allows new kernels to run on old distributions and tools,
1231 which might get confused by /sys/class/block/. Since 2007/2008 all
1232 major distributions and tools handle this just fine.
1234 Recent distributions and userspace tools after 2009/2010 depend on
1235 the existence of /sys/class/block/, and will not work with this
1238 Only if you are using a new kernel on an old distribution, you might
1241 config SYSFS_DEPRECATED_V2
1242 bool "Enable deprecated sysfs features by default"
1245 depends on SYSFS_DEPRECATED
1247 Enable deprecated sysfs by default.
1249 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1252 Only if you are using a new kernel on an old distribution, you might
1253 need to say Y here. Even then, odds are you would not need it
1254 enabled, you can always pass the boot option if absolutely necessary.
1257 bool "Kernel->user space relay support (formerly relayfs)"
1260 This option enables support for relay interface support in
1261 certain file systems (such as debugfs).
1262 It is designed to provide an efficient mechanism for tools and
1263 facilities to relay large amounts of data from kernel space to
1268 config BLK_DEV_INITRD
1269 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1271 The initial RAM filesystem is a ramfs which is loaded by the
1272 boot loader (loadlin or lilo) and that is mounted as root
1273 before the normal boot procedure. It is typically used to
1274 load modules needed to mount the "real" root file system,
1275 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1277 If RAM disk support (BLK_DEV_RAM) is also included, this
1278 also enables initial RAM disk (initrd) support and adds
1279 15 Kbytes (more on some other architectures) to the kernel size.
1285 source "usr/Kconfig"
1290 bool "Boot config support"
1291 select BLK_DEV_INITRD
1293 Extra boot config allows system admin to pass a config file as
1294 complemental extension of kernel cmdline when booting.
1295 The boot config file must be attached at the end of initramfs
1296 with checksum, size and magic word.
1297 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1302 prompt "Compiler optimization level"
1303 default CC_OPTIMIZE_FOR_PERFORMANCE
1305 config CC_OPTIMIZE_FOR_PERFORMANCE
1306 bool "Optimize for performance (-O2)"
1308 This is the default optimization level for the kernel, building
1309 with the "-O2" compiler flag for best performance and most
1310 helpful compile-time warnings.
1312 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1313 bool "Optimize more for performance (-O3)"
1316 Choosing this option will pass "-O3" to your compiler to optimize
1317 the kernel yet more for performance.
1319 config CC_OPTIMIZE_FOR_SIZE
1320 bool "Optimize for size (-Os)"
1322 Choosing this option will pass "-Os" to your compiler resulting
1323 in a smaller kernel.
1327 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1330 This requires that the arch annotates or otherwise protects
1331 its external entry points from being discarded. Linker scripts
1332 must also merge .text.*, .data.*, and .bss.* correctly into
1333 output sections. Care must be taken not to pull in unrelated
1334 sections (e.g., '.text.init'). Typically '.' in section names
1335 is used to distinguish them from label names / C identifiers.
1337 config LD_DEAD_CODE_DATA_ELIMINATION
1338 bool "Dead code and data elimination (EXPERIMENTAL)"
1339 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1341 depends on $(cc-option,-ffunction-sections -fdata-sections)
1342 depends on $(ld-option,--gc-sections)
1344 Enable this if you want to do dead code and data elimination with
1345 the linker by compiling with -ffunction-sections -fdata-sections,
1346 and linking with --gc-sections.
1348 This can reduce on disk and in-memory size of the kernel
1349 code and static data, particularly for small configs and
1350 on small systems. This has the possibility of introducing
1351 silently broken kernel if the required annotations are not
1352 present. This option is not well tested yet, so use at your
1355 config LD_ORPHAN_WARN
1357 depends on ARCH_WANT_LD_ORPHAN_WARN
1358 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1359 depends on $(ld-option,--orphan-handling=warn)
1367 config SYSCTL_EXCEPTION_TRACE
1370 Enable support for /proc/sys/debug/exception-trace.
1372 config SYSCTL_ARCH_UNALIGN_NO_WARN
1375 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1376 Allows arch to define/use @no_unaligned_warning to possibly warn
1377 about unaligned access emulation going on under the hood.
1379 config SYSCTL_ARCH_UNALIGN_ALLOW
1382 Enable support for /proc/sys/kernel/unaligned-trap
1383 Allows arches to define/use @unaligned_enabled to runtime toggle
1384 the unaligned access emulation.
1385 see arch/parisc/kernel/unaligned.c for reference
1387 config HAVE_PCSPKR_PLATFORM
1390 # interpreter that classic socket filters depend on
1395 bool "Configure standard kernel features (expert users)"
1396 # Unhide debug options, to make the on-by-default options visible
1399 This option allows certain base kernel options and settings
1400 to be disabled or tweaked. This is for specialized
1401 environments which can tolerate a "non-standard" kernel.
1402 Only use this if you really know what you are doing.
1405 bool "Enable 16-bit UID system calls" if EXPERT
1406 depends on HAVE_UID16 && MULTIUSER
1409 This enables the legacy 16-bit UID syscall wrappers.
1412 bool "Multiple users, groups and capabilities support" if EXPERT
1415 This option enables support for non-root users, groups and
1418 If you say N here, all processes will run with UID 0, GID 0, and all
1419 possible capabilities. Saying N here also compiles out support for
1420 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1423 If unsure, say Y here.
1425 config SGETMASK_SYSCALL
1426 bool "sgetmask/ssetmask syscalls support" if EXPERT
1427 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1429 sys_sgetmask and sys_ssetmask are obsolete system calls
1430 no longer supported in libc but still enabled by default in some
1433 If unsure, leave the default option here.
1435 config SYSFS_SYSCALL
1436 bool "Sysfs syscall support" if EXPERT
1439 sys_sysfs is an obsolete system call no longer supported in libc.
1440 Note that disabling this option is more secure but might break
1441 compatibility with some systems.
1443 If unsure say Y here.
1446 bool "open by fhandle syscalls" if EXPERT
1450 If you say Y here, a user level program will be able to map
1451 file names to handle and then later use the handle for
1452 different file system operations. This is useful in implementing
1453 userspace file servers, which now track files using handles instead
1454 of names. The handle would remain the same even if file names
1455 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1459 bool "Posix Clocks & timers" if EXPERT
1462 This includes native support for POSIX timers to the kernel.
1463 Some embedded systems have no use for them and therefore they
1464 can be configured out to reduce the size of the kernel image.
1466 When this option is disabled, the following syscalls won't be
1467 available: timer_create, timer_gettime: timer_getoverrun,
1468 timer_settime, timer_delete, clock_adjtime, getitimer,
1469 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1470 clock_getres and clock_nanosleep syscalls will be limited to
1471 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1477 bool "Enable support for printk" if EXPERT
1480 This option enables normal printk support. Removing it
1481 eliminates most of the message strings from the kernel image
1482 and makes the kernel more or less silent. As this makes it
1483 very difficult to diagnose system problems, saying N here is
1484 strongly discouraged.
1492 bool "BUG() support" if EXPERT
1495 Disabling this option eliminates support for BUG and WARN, reducing
1496 the size of your kernel image and potentially quietly ignoring
1497 numerous fatal conditions. You should only consider disabling this
1498 option for embedded systems with no facilities for reporting errors.
1504 bool "Enable ELF core dumps" if EXPERT
1506 Enable support for generating core dumps. Disabling saves about 4k.
1509 config PCSPKR_PLATFORM
1510 bool "Enable PC-Speaker support" if EXPERT
1511 depends on HAVE_PCSPKR_PLATFORM
1515 This option allows to disable the internal PC-Speaker
1516 support, saving some memory.
1520 bool "Enable full-sized data structures for core" if EXPERT
1522 Disabling this option reduces the size of miscellaneous core
1523 kernel data structures. This saves memory on small machines,
1524 but may reduce performance.
1527 bool "Enable futex support" if EXPERT
1531 Disabling this option will cause the kernel to be built without
1532 support for "fast userspace mutexes". The resulting kernel may not
1533 run glibc-based applications correctly.
1537 depends on FUTEX && RT_MUTEXES
1540 config HAVE_FUTEX_CMPXCHG
1544 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1545 is implemented and always working. This removes a couple of runtime
1549 bool "Enable eventpoll support" if EXPERT
1552 Disabling this option will cause the kernel to be built without
1553 support for epoll family of system calls.
1556 bool "Enable signalfd() system call" if EXPERT
1559 Enable the signalfd() system call that allows to receive signals
1560 on a file descriptor.
1565 bool "Enable timerfd() system call" if EXPERT
1568 Enable the timerfd() system call that allows to receive timer
1569 events on a file descriptor.
1574 bool "Enable eventfd() system call" if EXPERT
1577 Enable the eventfd() system call that allows to receive both
1578 kernel notification (ie. KAIO) or userspace notifications.
1583 bool "Use full shmem filesystem" if EXPERT
1587 The shmem is an internal filesystem used to manage shared memory.
1588 It is backed by swap and manages resource limits. It is also exported
1589 to userspace as tmpfs if TMPFS is enabled. Disabling this
1590 option replaces shmem and tmpfs with the much simpler ramfs code,
1591 which may be appropriate on small systems without swap.
1594 bool "Enable AIO support" if EXPERT
1597 This option enables POSIX asynchronous I/O which may by used
1598 by some high performance threaded applications. Disabling
1599 this option saves about 7k.
1602 bool "Enable IO uring support" if EXPERT
1606 This option enables support for the io_uring interface, enabling
1607 applications to submit and complete IO through submission and
1608 completion rings that are shared between the kernel and application.
1610 config ADVISE_SYSCALLS
1611 bool "Enable madvise/fadvise syscalls" if EXPERT
1614 This option enables the madvise and fadvise syscalls, used by
1615 applications to advise the kernel about their future memory or file
1616 usage, improving performance. If building an embedded system where no
1617 applications use these syscalls, you can disable this option to save
1620 config HAVE_ARCH_USERFAULTFD_WP
1623 Arch has userfaultfd write protection support
1626 bool "Enable membarrier() system call" if EXPERT
1629 Enable the membarrier() system call that allows issuing memory
1630 barriers across all running threads, which can be used to distribute
1631 the cost of user-space memory barriers asymmetrically by transforming
1632 pairs of memory barriers into pairs consisting of membarrier() and a
1638 bool "Load all symbols for debugging/ksymoops" if EXPERT
1641 Say Y here to let the kernel print out symbolic crash information and
1642 symbolic stack backtraces. This increases the size of the kernel
1643 somewhat, as all symbols have to be loaded into the kernel image.
1646 bool "Include all symbols in kallsyms"
1647 depends on DEBUG_KERNEL && KALLSYMS
1649 Normally kallsyms only contains the symbols of functions for nicer
1650 OOPS messages and backtraces (i.e., symbols from the text and inittext
1651 sections). This is sufficient for most cases. And only in very rare
1652 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1653 names of variables from the data sections, etc).
1655 This option makes sure that all symbols are loaded into the kernel
1656 image (i.e., symbols from all sections) in cost of increased kernel
1657 size (depending on the kernel configuration, it may be 300KiB or
1658 something like this).
1660 Say N unless you really need all symbols.
1662 config KALLSYMS_ABSOLUTE_PERCPU
1665 default X86_64 && SMP
1667 config KALLSYMS_BASE_RELATIVE
1672 Instead of emitting them as absolute values in the native word size,
1673 emit the symbol references in the kallsyms table as 32-bit entries,
1674 each containing a relative value in the range [base, base + U32_MAX]
1675 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1676 an absolute value in the range [0, S32_MAX] or a relative value in the
1677 range [base, base + S32_MAX], where base is the lowest relative symbol
1678 address encountered in the image.
1680 On 64-bit builds, this reduces the size of the address table by 50%,
1681 but more importantly, it results in entries whose values are build
1682 time constants, and no relocation pass is required at runtime to fix
1683 up the entries based on the runtime load address of the kernel.
1685 # end of the "standard kernel features (expert users)" menu
1687 # syscall, maps, verifier
1690 bool "LSM Instrumentation with BPF"
1691 depends on BPF_EVENTS
1692 depends on BPF_SYSCALL
1696 Enables instrumentation of the security hooks with eBPF programs for
1697 implementing dynamic MAC and Audit Policies.
1699 If you are unsure how to answer this question, answer N.
1702 bool "Enable bpf() system call"
1705 select TASKS_TRACE_RCU
1708 Enable the bpf() system call that allows to manipulate eBPF
1709 programs and maps via file descriptors.
1711 config ARCH_WANT_DEFAULT_BPF_JIT
1714 config BPF_JIT_ALWAYS_ON
1715 bool "Permanently enable BPF JIT and remove BPF interpreter"
1716 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1718 Enables BPF JIT and removes BPF interpreter to avoid
1719 speculative execution of BPF instructions by the interpreter
1721 config BPF_JIT_DEFAULT_ON
1722 def_bool ARCH_WANT_DEFAULT_BPF_JIT || BPF_JIT_ALWAYS_ON
1723 depends on HAVE_EBPF_JIT && BPF_JIT
1725 source "kernel/bpf/preload/Kconfig"
1728 bool "Enable userfaultfd() system call"
1731 Enable the userfaultfd() system call that allows to intercept and
1732 handle page faults in userland.
1734 config ARCH_HAS_MEMBARRIER_CALLBACKS
1737 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1741 bool "Enable kcmp() system call" if EXPERT
1743 Enable the kernel resource comparison system call. It provides
1744 user-space with the ability to compare two processes to see if they
1745 share a common resource, such as a file descriptor or even virtual
1751 bool "Enable rseq() system call" if EXPERT
1753 depends on HAVE_RSEQ
1756 Enable the restartable sequences system call. It provides a
1757 user-space cache for the current CPU number value, which
1758 speeds up getting the current CPU number from user-space,
1759 as well as an ABI to speed up user-space operations on
1766 bool "Enabled debugging of rseq() system call" if EXPERT
1767 depends on RSEQ && DEBUG_KERNEL
1769 Enable extra debugging checks for the rseq system call.
1774 bool "Embedded system"
1775 option allnoconfig_y
1778 This option should be enabled if compiling the kernel for
1779 an embedded system so certain expert options are available
1782 config HAVE_PERF_EVENTS
1785 See tools/perf/design.txt for details.
1787 config PERF_USE_VMALLOC
1790 See tools/perf/design.txt for details
1793 bool "PC/104 support" if EXPERT
1795 Expose PC/104 form factor device drivers and options available for
1796 selection and configuration. Enable this option if your target
1797 machine has a PC/104 bus.
1799 menu "Kernel Performance Events And Counters"
1802 bool "Kernel performance events and counters"
1803 default y if PROFILING
1804 depends on HAVE_PERF_EVENTS
1808 Enable kernel support for various performance events provided
1809 by software and hardware.
1811 Software events are supported either built-in or via the
1812 use of generic tracepoints.
1814 Most modern CPUs support performance events via performance
1815 counter registers. These registers count the number of certain
1816 types of hw events: such as instructions executed, cachemisses
1817 suffered, or branches mis-predicted - without slowing down the
1818 kernel or applications. These registers can also trigger interrupts
1819 when a threshold number of events have passed - and can thus be
1820 used to profile the code that runs on that CPU.
1822 The Linux Performance Event subsystem provides an abstraction of
1823 these software and hardware event capabilities, available via a
1824 system call and used by the "perf" utility in tools/perf/. It
1825 provides per task and per CPU counters, and it provides event
1826 capabilities on top of those.
1830 config DEBUG_PERF_USE_VMALLOC
1832 bool "Debug: use vmalloc to back perf mmap() buffers"
1833 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1834 select PERF_USE_VMALLOC
1836 Use vmalloc memory to back perf mmap() buffers.
1838 Mostly useful for debugging the vmalloc code on platforms
1839 that don't require it.
1845 config VM_EVENT_COUNTERS
1847 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1849 VM event counters are needed for event counts to be shown.
1850 This option allows the disabling of the VM event counters
1851 on EXPERT systems. /proc/vmstat will only show page counts
1852 if VM event counters are disabled.
1856 bool "Enable SLUB debugging support" if EXPERT
1857 depends on SLUB && SYSFS
1859 SLUB has extensive debug support features. Disabling these can
1860 result in significant savings in code size. This also disables
1861 SLUB sysfs support. /sys/slab will not exist and there will be
1862 no support for cache validation etc.
1865 bool "Disable heap randomization"
1868 Randomizing heap placement makes heap exploits harder, but it
1869 also breaks ancient binaries (including anything libc5 based).
1870 This option changes the bootup default to heap randomization
1871 disabled, and can be overridden at runtime by setting
1872 /proc/sys/kernel/randomize_va_space to 2.
1874 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1877 prompt "Choose SLAB allocator"
1880 This option allows to select a slab allocator.
1884 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1886 The regular slab allocator that is established and known to work
1887 well in all environments. It organizes cache hot objects in
1888 per cpu and per node queues.
1891 bool "SLUB (Unqueued Allocator)"
1892 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1894 SLUB is a slab allocator that minimizes cache line usage
1895 instead of managing queues of cached objects (SLAB approach).
1896 Per cpu caching is realized using slabs of objects instead
1897 of queues of objects. SLUB can use memory efficiently
1898 and has enhanced diagnostics. SLUB is the default choice for
1903 bool "SLOB (Simple Allocator)"
1905 SLOB replaces the stock allocator with a drastically simpler
1906 allocator. SLOB is generally more space efficient but
1907 does not perform as well on large systems.
1911 config SLAB_MERGE_DEFAULT
1912 bool "Allow slab caches to be merged"
1915 For reduced kernel memory fragmentation, slab caches can be
1916 merged when they share the same size and other characteristics.
1917 This carries a risk of kernel heap overflows being able to
1918 overwrite objects from merged caches (and more easily control
1919 cache layout), which makes such heap attacks easier to exploit
1920 by attackers. By keeping caches unmerged, these kinds of exploits
1921 can usually only damage objects in the same cache. To disable
1922 merging at runtime, "slab_nomerge" can be passed on the kernel
1925 config SLAB_FREELIST_RANDOM
1926 bool "Randomize slab freelist"
1927 depends on SLAB || SLUB
1929 Randomizes the freelist order used on creating new pages. This
1930 security feature reduces the predictability of the kernel slab
1931 allocator against heap overflows.
1933 config SLAB_FREELIST_HARDENED
1934 bool "Harden slab freelist metadata"
1935 depends on SLAB || SLUB
1937 Many kernel heap attacks try to target slab cache metadata and
1938 other infrastructure. This options makes minor performance
1939 sacrifices to harden the kernel slab allocator against common
1940 freelist exploit methods. Some slab implementations have more
1941 sanity-checking than others. This option is most effective with
1944 config SHUFFLE_PAGE_ALLOCATOR
1945 bool "Page allocator randomization"
1946 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1948 Randomization of the page allocator improves the average
1949 utilization of a direct-mapped memory-side-cache. See section
1950 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1951 6.2a specification for an example of how a platform advertises
1952 the presence of a memory-side-cache. There are also incidental
1953 security benefits as it reduces the predictability of page
1954 allocations to compliment SLAB_FREELIST_RANDOM, but the
1955 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1956 10th order of pages is selected based on cache utilization
1959 While the randomization improves cache utilization it may
1960 negatively impact workloads on platforms without a cache. For
1961 this reason, by default, the randomization is enabled only
1962 after runtime detection of a direct-mapped memory-side-cache.
1963 Otherwise, the randomization may be force enabled with the
1964 'page_alloc.shuffle' kernel command line parameter.
1968 config SLUB_CPU_PARTIAL
1970 depends on SLUB && SMP
1971 bool "SLUB per cpu partial cache"
1973 Per cpu partial caches accelerate objects allocation and freeing
1974 that is local to a processor at the price of more indeterminism
1975 in the latency of the free. On overflow these caches will be cleared
1976 which requires the taking of locks that may cause latency spikes.
1977 Typically one would choose no for a realtime system.
1979 config MMAP_ALLOW_UNINITIALIZED
1980 bool "Allow mmapped anonymous memory to be uninitialized"
1981 depends on EXPERT && !MMU
1984 Normally, and according to the Linux spec, anonymous memory obtained
1985 from mmap() has its contents cleared before it is passed to
1986 userspace. Enabling this config option allows you to request that
1987 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1988 providing a huge performance boost. If this option is not enabled,
1989 then the flag will be ignored.
1991 This is taken advantage of by uClibc's malloc(), and also by
1992 ELF-FDPIC binfmt's brk and stack allocator.
1994 Because of the obvious security issues, this option should only be
1995 enabled on embedded devices where you control what is run in
1996 userspace. Since that isn't generally a problem on no-MMU systems,
1997 it is normally safe to say Y here.
1999 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2001 config SYSTEM_DATA_VERIFICATION
2003 select SYSTEM_TRUSTED_KEYRING
2007 select ASYMMETRIC_KEY_TYPE
2008 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2011 select X509_CERTIFICATE_PARSER
2012 select PKCS7_MESSAGE_PARSER
2014 Provide PKCS#7 message verification using the contents of the system
2015 trusted keyring to provide public keys. This then can be used for
2016 module verification, kexec image verification and firmware blob
2020 bool "Profiling support"
2022 Say Y here to enable the extended profiling support mechanisms used
2026 # Place an empty function call at each tracepoint site. Can be
2027 # dynamically changed for a probe function.
2032 endmenu # General setup
2034 source "arch/Kconfig"
2041 default 0 if BASE_FULL
2042 default 1 if !BASE_FULL
2044 config MODULE_SIG_FORMAT
2046 select SYSTEM_DATA_VERIFICATION
2049 bool "Enable loadable module support"
2052 Kernel modules are small pieces of compiled code which can
2053 be inserted in the running kernel, rather than being
2054 permanently built into the kernel. You use the "modprobe"
2055 tool to add (and sometimes remove) them. If you say Y here,
2056 many parts of the kernel can be built as modules (by
2057 answering M instead of Y where indicated): this is most
2058 useful for infrequently used options which are not required
2059 for booting. For more information, see the man pages for
2060 modprobe, lsmod, modinfo, insmod and rmmod.
2062 If you say Y here, you will need to run "make
2063 modules_install" to put the modules under /lib/modules/
2064 where modprobe can find them (you may need to be root to do
2071 config MODULE_FORCE_LOAD
2072 bool "Forced module loading"
2075 Allow loading of modules without version information (ie. modprobe
2076 --force). Forced module loading sets the 'F' (forced) taint flag and
2077 is usually a really bad idea.
2079 config MODULE_UNLOAD
2080 bool "Module unloading"
2082 Without this option you will not be able to unload any
2083 modules (note that some modules may not be unloadable
2084 anyway), which makes your kernel smaller, faster
2085 and simpler. If unsure, say Y.
2087 config MODULE_FORCE_UNLOAD
2088 bool "Forced module unloading"
2089 depends on MODULE_UNLOAD
2091 This option allows you to force a module to unload, even if the
2092 kernel believes it is unsafe: the kernel will remove the module
2093 without waiting for anyone to stop using it (using the -f option to
2094 rmmod). This is mainly for kernel developers and desperate users.
2098 bool "Module versioning support"
2100 Usually, you have to use modules compiled with your kernel.
2101 Saying Y here makes it sometimes possible to use modules
2102 compiled for different kernels, by adding enough information
2103 to the modules to (hopefully) spot any changes which would
2104 make them incompatible with the kernel you are running. If
2107 config ASM_MODVERSIONS
2109 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2111 This enables module versioning for exported symbols also from
2112 assembly. This can be enabled only when the target architecture
2115 config MODULE_REL_CRCS
2117 depends on MODVERSIONS
2119 config MODULE_SRCVERSION_ALL
2120 bool "Source checksum for all modules"
2122 Modules which contain a MODULE_VERSION get an extra "srcversion"
2123 field inserted into their modinfo section, which contains a
2124 sum of the source files which made it. This helps maintainers
2125 see exactly which source was used to build a module (since
2126 others sometimes change the module source without updating
2127 the version). With this option, such a "srcversion" field
2128 will be created for all modules. If unsure, say N.
2131 bool "Module signature verification"
2132 select MODULE_SIG_FORMAT
2134 Check modules for valid signatures upon load: the signature
2135 is simply appended to the module. For more information see
2136 <file:Documentation/admin-guide/module-signing.rst>.
2138 Note that this option adds the OpenSSL development packages as a
2139 kernel build dependency so that the signing tool can use its crypto
2142 You should enable this option if you wish to use either
2143 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2144 another LSM - otherwise unsigned modules will be loadable regardless
2145 of the lockdown policy.
2147 !!!WARNING!!! If you enable this option, you MUST make sure that the
2148 module DOES NOT get stripped after being signed. This includes the
2149 debuginfo strip done by some packagers (such as rpmbuild) and
2150 inclusion into an initramfs that wants the module size reduced.
2152 config MODULE_SIG_FORCE
2153 bool "Require modules to be validly signed"
2154 depends on MODULE_SIG
2156 Reject unsigned modules or signed modules for which we don't have a
2157 key. Without this, such modules will simply taint the kernel.
2159 config MODULE_SIG_ALL
2160 bool "Automatically sign all modules"
2162 depends on MODULE_SIG
2164 Sign all modules during make modules_install. Without this option,
2165 modules must be signed manually, using the scripts/sign-file tool.
2167 comment "Do not forget to sign required modules with scripts/sign-file"
2168 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2171 prompt "Which hash algorithm should modules be signed with?"
2172 depends on MODULE_SIG
2174 This determines which sort of hashing algorithm will be used during
2175 signature generation. This algorithm _must_ be built into the kernel
2176 directly so that signature verification can take place. It is not
2177 possible to load a signed module containing the algorithm to check
2178 the signature on that module.
2180 config MODULE_SIG_SHA1
2181 bool "Sign modules with SHA-1"
2184 config MODULE_SIG_SHA224
2185 bool "Sign modules with SHA-224"
2186 select CRYPTO_SHA256
2188 config MODULE_SIG_SHA256
2189 bool "Sign modules with SHA-256"
2190 select CRYPTO_SHA256
2192 config MODULE_SIG_SHA384
2193 bool "Sign modules with SHA-384"
2194 select CRYPTO_SHA512
2196 config MODULE_SIG_SHA512
2197 bool "Sign modules with SHA-512"
2198 select CRYPTO_SHA512
2202 config MODULE_SIG_HASH
2204 depends on MODULE_SIG
2205 default "sha1" if MODULE_SIG_SHA1
2206 default "sha224" if MODULE_SIG_SHA224
2207 default "sha256" if MODULE_SIG_SHA256
2208 default "sha384" if MODULE_SIG_SHA384
2209 default "sha512" if MODULE_SIG_SHA512
2211 config MODULE_COMPRESS
2212 bool "Compress modules on installation"
2215 Compresses kernel modules when 'make modules_install' is run; gzip or
2216 xz depending on "Compression algorithm" below.
2218 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2220 Out-of-tree kernel modules installed using Kbuild will also be
2221 compressed upon installation.
2223 Note: for modules inside an initrd or initramfs, it's more efficient
2224 to compress the whole initrd or initramfs instead.
2226 Note: This is fully compatible with signed modules.
2231 prompt "Compression algorithm"
2232 depends on MODULE_COMPRESS
2233 default MODULE_COMPRESS_GZIP
2235 This determines which sort of compression will be used during
2236 'make modules_install'.
2238 GZIP (default) and XZ are supported.
2240 config MODULE_COMPRESS_GZIP
2243 config MODULE_COMPRESS_XZ
2248 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2249 bool "Allow loading of modules with missing namespace imports"
2251 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2252 a namespace. A module that makes use of a symbol exported with such a
2253 namespace is required to import the namespace via MODULE_IMPORT_NS().
2254 There is no technical reason to enforce correct namespace imports,
2255 but it creates consistency between symbols defining namespaces and
2256 users importing namespaces they make use of. This option relaxes this
2257 requirement and lifts the enforcement when loading a module.
2261 config TRIM_UNUSED_KSYMS
2262 bool "Trim unused exported kernel symbols"
2265 The kernel and some modules make many symbols available for
2266 other modules to use via EXPORT_SYMBOL() and variants. Depending
2267 on the set of modules being selected in your kernel configuration,
2268 many of those exported symbols might never be used.
2270 This option allows for unused exported symbols to be dropped from
2271 the build. In turn, this provides the compiler more opportunities
2272 (especially when using LTO) for optimizing the code and reducing
2273 binary size. This might have some security advantages as well.
2275 If unsure, or if you need to build out-of-tree modules, say N.
2277 config UNUSED_KSYMS_WHITELIST
2278 string "Whitelist of symbols to keep in ksymtab"
2279 depends on TRIM_UNUSED_KSYMS
2280 default "scripts/lto-used-symbollist.txt" if LTO_CLANG
2282 By default, all unused exported symbols will be un-exported from the
2283 build when TRIM_UNUSED_KSYMS is selected.
2285 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2286 exported at all times, even in absence of in-tree users. The value to
2287 set here is the path to a text file containing the list of symbols,
2288 one per line. The path can be absolute, or relative to the kernel
2293 config MODULES_TREE_LOOKUP
2295 depends on PERF_EVENTS || TRACING
2297 config INIT_ALL_POSSIBLE
2300 Back when each arch used to define their own cpu_online_mask and
2301 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2302 with all 1s, and others with all 0s. When they were centralised,
2303 it was better to provide this option than to break all the archs
2304 and have several arch maintainers pursuing me down dark alleys.
2306 source "block/Kconfig"
2308 config PREEMPT_NOTIFIERS
2318 Build a simple ASN.1 grammar compiler that produces a bytecode output
2319 that can be interpreted by the ASN.1 stream decoder and used to
2320 inform it as to what tags are to be expected in a stream and what
2321 functions to call on what tags.
2323 source "kernel/Kconfig.locks"
2325 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2328 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2331 # It may be useful for an architecture to override the definitions of the
2332 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2333 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2334 # different calling convention for syscalls. They can also override the
2335 # macros for not-implemented syscalls in kernel/sys_ni.c and
2336 # kernel/time/posix-stubs.c. All these overrides need to be available in
2337 # <asm/syscall_wrapper.h>.
2338 config ARCH_HAS_SYSCALL_WRAPPER
projects / platform / kernel / linux-rpi.git / blob