1 # SPDX-License-Identifier: GPL-2.0-only
4 default "$(CC_VERSION_TEXT)"
6 This is used in unclear ways:
8 - Re-run Kconfig when the compiler is updated
9 The 'default' property references the environment variable,
10 CC_VERSION_TEXT so it is recorded in include/config/auto.conf.cmd.
11 When the compiler is updated, Kconfig will be invoked.
13 - Ensure full rebuild when the compiler is updated
14 include/linux/compiler-version.h contains this option in the comment
15 line so fixdep adds include/config/CC_VERSION_TEXT into the
16 auto-generated dependency. When the compiler is updated, syncconfig
17 will touch it and then every file will be rebuilt.
20 def_bool $(success,test "$(cc-name)" = GCC)
24 default $(cc-version) if CC_IS_GCC
28 def_bool $(success,test "$(cc-name)" = Clang)
32 default $(cc-version) if CC_IS_CLANG
36 def_bool $(success,test "$(as-name)" = GNU)
39 def_bool $(success,test "$(as-name)" = LLVM)
43 # Use clang version if this is the integrated assembler
44 default CLANG_VERSION if AS_IS_LLVM
48 def_bool $(success,test "$(ld-name)" = BFD)
52 default $(ld-version) if LD_IS_BFD
56 def_bool $(success,test "$(ld-name)" = LLD)
60 default $(ld-version) if LD_IS_LLD
65 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag)) if 64BIT
66 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag))
68 config CC_CAN_LINK_STATIC
70 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m64-flag) -static) if 64BIT
71 default $(success,$(srctree)/scripts/cc-can-link.sh $(CC) $(CLANG_FLAGS) $(m32-flag) -static)
73 config CC_HAS_ASM_GOTO
74 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
76 config CC_HAS_ASM_GOTO_OUTPUT
77 depends on CC_HAS_ASM_GOTO
78 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)
80 config CC_HAS_ASM_GOTO_TIED_OUTPUT
81 depends on CC_HAS_ASM_GOTO_OUTPUT
82 # Detect buggy gcc and clang, fixed in gcc-11 clang-14.
83 def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
85 config TOOLS_SUPPORT_RELR
86 def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
88 config CC_HAS_ASM_INLINE
89 def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
91 config CC_HAS_NO_PROFILE_FN_ATTR
92 def_bool $(success,echo '__attribute__((no_profile_instrument_function)) int x();' | $(CC) -x c - -c -o /dev/null -Werror)
100 config BUILDTIME_TABLE_SORT
103 config THREAD_INFO_IN_TASK
106 Select this to move thread_info off the stack into task_struct. To
107 make this work, an arch will need to remove all thread_info fields
108 except flags and fix any runtime bugs.
110 One subtle change that will be needed is to use try_get_task_stack()
111 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
120 depends on BROKEN || !SMP
123 config INIT_ENV_ARG_LIMIT
128 Maximum of each of the number of arguments and environment
129 variables passed to init from the kernel command line.
132 bool "Compile also drivers which will not load"
135 Some drivers can be compiled on a different platform than they are
136 intended to be run on. Despite they cannot be loaded there (or even
137 when they load they cannot be used due to missing HW support),
138 developers still, opposing to distributors, might want to build such
139 drivers to compile-test them.
141 If you are a developer and want to build everything available, say Y
142 here. If you are a user/distributor, say N here to exclude useless
143 drivers to be distributed.
146 bool "Compile the kernel with warnings as errors"
149 A kernel build should not cause any compiler warnings, and this
150 enables the '-Werror' flag to enforce that rule by default.
152 However, if you have a new (or very old) compiler with odd and
153 unusual warnings, or you have some architecture with problems,
154 you may need to disable this config option in order to
155 successfully build the kernel.
159 config UAPI_HEADER_TEST
160 bool "Compile test UAPI headers"
161 depends on HEADERS_INSTALL && CC_CAN_LINK
163 Compile test headers exported to user-space to ensure they are
164 self-contained, i.e. compilable as standalone units.
166 If you are a developer or tester and want to ensure the exported
167 headers are self-contained, say Y here. Otherwise, choose N.
170 string "Local version - append to kernel release"
172 Append an extra string to the end of your kernel version.
173 This will show up when you type uname, for example.
174 The string you set here will be appended after the contents of
175 any files with a filename matching localversion* in your
176 object and source tree, in that order. Your total string can
177 be a maximum of 64 characters.
179 config LOCALVERSION_AUTO
180 bool "Automatically append version information to the version string"
182 depends on !COMPILE_TEST
184 This will try to automatically determine if the current tree is a
185 release tree by looking for git tags that belong to the current
186 top of tree revision.
188 A string of the format -gxxxxxxxx will be added to the localversion
189 if a git-based tree is found. The string generated by this will be
190 appended after any matching localversion* files, and after the value
191 set in CONFIG_LOCALVERSION.
193 (The actual string used here is the first eight characters produced
194 by running the command:
196 $ git rev-parse --verify HEAD
198 which is done within the script "scripts/setlocalversion".)
201 string "Build ID Salt"
204 The build ID is used to link binaries and their debug info. Setting
205 this option will use the value in the calculation of the build id.
206 This is mostly useful for distributions which want to ensure the
207 build is unique between builds. It's safe to leave the default.
209 config HAVE_KERNEL_GZIP
212 config HAVE_KERNEL_BZIP2
215 config HAVE_KERNEL_LZMA
218 config HAVE_KERNEL_XZ
221 config HAVE_KERNEL_LZO
224 config HAVE_KERNEL_LZ4
227 config HAVE_KERNEL_ZSTD
230 config HAVE_KERNEL_UNCOMPRESSED
234 prompt "Kernel compression mode"
236 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
238 The linux kernel is a kind of self-extracting executable.
239 Several compression algorithms are available, which differ
240 in efficiency, compression and decompression speed.
241 Compression speed is only relevant when building a kernel.
242 Decompression speed is relevant at each boot.
244 If you have any problems with bzip2 or lzma compressed
245 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
246 version of this functionality (bzip2 only), for 2.4, was
247 supplied by Christian Ludwig)
249 High compression options are mostly useful for users, who
250 are low on disk space (embedded systems), but for whom ram
253 If in doubt, select 'gzip'
257 depends on HAVE_KERNEL_GZIP
259 The old and tried gzip compression. It provides a good balance
260 between compression ratio and decompression speed.
264 depends on HAVE_KERNEL_BZIP2
266 Its compression ratio and speed is intermediate.
267 Decompression speed is slowest among the choices. The kernel
268 size is about 10% smaller with bzip2, in comparison to gzip.
269 Bzip2 uses a large amount of memory. For modern kernels you
270 will need at least 8MB RAM or more for booting.
274 depends on HAVE_KERNEL_LZMA
276 This compression algorithm's ratio is best. Decompression speed
277 is between gzip and bzip2. Compression is slowest.
278 The kernel size is about 33% smaller with LZMA in comparison to gzip.
282 depends on HAVE_KERNEL_XZ
284 XZ uses the LZMA2 algorithm and instruction set specific
285 BCJ filters which can improve compression ratio of executable
286 code. The size of the kernel is about 30% smaller with XZ in
287 comparison to gzip. On architectures for which there is a BCJ
288 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
289 will create a few percent smaller kernel than plain LZMA.
291 The speed is about the same as with LZMA: The decompression
292 speed of XZ is better than that of bzip2 but worse than gzip
293 and LZO. Compression is slow.
297 depends on HAVE_KERNEL_LZO
299 Its compression ratio is the poorest among the choices. The kernel
300 size is about 10% bigger than gzip; however its speed
301 (both compression and decompression) is the fastest.
305 depends on HAVE_KERNEL_LZ4
307 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
308 A preliminary version of LZ4 de/compression tool is available at
309 <https://code.google.com/p/lz4/>.
311 Its compression ratio is worse than LZO. The size of the kernel
312 is about 8% bigger than LZO. But the decompression speed is
317 depends on HAVE_KERNEL_ZSTD
319 ZSTD is a compression algorithm targeting intermediate compression
320 with fast decompression speed. It will compress better than GZIP and
321 decompress around the same speed as LZO, but slower than LZ4. You
322 will need at least 192 KB RAM or more for booting. The zstd command
323 line tool is required for compression.
325 config KERNEL_UNCOMPRESSED
327 depends on HAVE_KERNEL_UNCOMPRESSED
329 Produce uncompressed kernel image. This option is usually not what
330 you want. It is useful for debugging the kernel in slow simulation
331 environments, where decompressing and moving the kernel is awfully
332 slow. This option allows early boot code to skip the decompressor
333 and jump right at uncompressed kernel image.
338 string "Default init path"
341 This option determines the default init for the system if no init=
342 option is passed on the kernel command line. If the requested path is
343 not present, we will still then move on to attempting further
344 locations (e.g. /sbin/init, etc). If this is empty, we will just use
345 the fallback list when init= is not passed.
347 config DEFAULT_HOSTNAME
348 string "Default hostname"
351 This option determines the default system hostname before userspace
352 calls sethostname(2). The kernel traditionally uses "(none)" here,
353 but you may wish to use a different default here to make a minimal
354 system more usable with less configuration.
357 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
358 # add proper SWAP support to them, in which case this can be remove.
364 bool "Support for paging of anonymous memory (swap)"
365 depends on MMU && BLOCK && !ARCH_NO_SWAP
368 This option allows you to choose whether you want to have support
369 for so called swap devices or swap files in your kernel that are
370 used to provide more virtual memory than the actual RAM present
371 in your computer. If unsure say Y.
376 Inter Process Communication is a suite of library functions and
377 system calls which let processes (running programs) synchronize and
378 exchange information. It is generally considered to be a good thing,
379 and some programs won't run unless you say Y here. In particular, if
380 you want to run the DOS emulator dosemu under Linux (read the
381 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
382 you'll need to say Y here.
384 You can find documentation about IPC with "info ipc" and also in
385 section 6.4 of the Linux Programmer's Guide, available from
386 <http://www.tldp.org/guides.html>.
388 config SYSVIPC_SYSCTL
395 bool "POSIX Message Queues"
398 POSIX variant of message queues is a part of IPC. In POSIX message
399 queues every message has a priority which decides about succession
400 of receiving it by a process. If you want to compile and run
401 programs written e.g. for Solaris with use of its POSIX message
402 queues (functions mq_*) say Y here.
404 POSIX message queues are visible as a filesystem called 'mqueue'
405 and can be mounted somewhere if you want to do filesystem
406 operations on message queues.
410 config POSIX_MQUEUE_SYSCTL
412 depends on POSIX_MQUEUE
417 bool "General notification queue"
421 This is a general notification queue for the kernel to pass events to
422 userspace by splicing them into pipes. It can be used in conjunction
423 with watches for key/keyring change notifications and device
426 See Documentation/watch_queue.rst
428 config CROSS_MEMORY_ATTACH
429 bool "Enable process_vm_readv/writev syscalls"
433 Enabling this option adds the system calls process_vm_readv and
434 process_vm_writev which allow a process with the correct privileges
435 to directly read from or write to another process' address space.
436 See the man page for more details.
439 bool "uselib syscall"
440 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
442 This option enables the uselib syscall, a system call used in the
443 dynamic linker from libc5 and earlier. glibc does not use this
444 system call. If you intend to run programs built on libc5 or
445 earlier, you may need to enable this syscall. Current systems
446 running glibc can safely disable this.
449 bool "Auditing support"
452 Enable auditing infrastructure that can be used with another
453 kernel subsystem, such as SELinux (which requires this for
454 logging of avc messages output). System call auditing is included
455 on architectures which support it.
457 config HAVE_ARCH_AUDITSYSCALL
462 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
465 source "kernel/irq/Kconfig"
466 source "kernel/time/Kconfig"
467 source "kernel/bpf/Kconfig"
468 source "kernel/Kconfig.preempt"
470 menu "CPU/Task time and stats accounting"
472 config VIRT_CPU_ACCOUNTING
476 prompt "Cputime accounting"
477 default TICK_CPU_ACCOUNTING if !PPC64
478 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
480 # Kind of a stub config for the pure tick based cputime accounting
481 config TICK_CPU_ACCOUNTING
482 bool "Simple tick based cputime accounting"
483 depends on !S390 && !NO_HZ_FULL
485 This is the basic tick based cputime accounting that maintains
486 statistics about user, system and idle time spent on per jiffies
491 config VIRT_CPU_ACCOUNTING_NATIVE
492 bool "Deterministic task and CPU time accounting"
493 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
494 select VIRT_CPU_ACCOUNTING
496 Select this option to enable more accurate task and CPU time
497 accounting. This is done by reading a CPU counter on each
498 kernel entry and exit and on transitions within the kernel
499 between system, softirq and hardirq state, so there is a
500 small performance impact. In the case of s390 or IBM POWER > 5,
501 this also enables accounting of stolen time on logically-partitioned
504 config VIRT_CPU_ACCOUNTING_GEN
505 bool "Full dynticks CPU time accounting"
506 depends on HAVE_CONTEXT_TRACKING
507 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
508 depends on GENERIC_CLOCKEVENTS
509 select VIRT_CPU_ACCOUNTING
510 select CONTEXT_TRACKING
512 Select this option to enable task and CPU time accounting on full
513 dynticks systems. This accounting is implemented by watching every
514 kernel-user boundaries using the context tracking subsystem.
515 The accounting is thus performed at the expense of some significant
518 For now this is only useful if you are working on the full
519 dynticks subsystem development.
525 config IRQ_TIME_ACCOUNTING
526 bool "Fine granularity task level IRQ time accounting"
527 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
529 Select this option to enable fine granularity task irq time
530 accounting. This is done by reading a timestamp on each
531 transitions between softirq and hardirq state, so there can be a
532 small performance impact.
534 If in doubt, say N here.
536 config HAVE_SCHED_AVG_IRQ
538 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
541 config SCHED_THERMAL_PRESSURE
543 default y if ARM && ARM_CPU_TOPOLOGY
546 depends on CPU_FREQ_THERMAL
548 Select this option to enable thermal pressure accounting in the
549 scheduler. Thermal pressure is the value conveyed to the scheduler
550 that reflects the reduction in CPU compute capacity resulted from
551 thermal throttling. Thermal throttling occurs when the performance of
552 a CPU is capped due to high operating temperatures.
554 If selected, the scheduler will be able to balance tasks accordingly,
555 i.e. put less load on throttled CPUs than on non/less throttled ones.
557 This requires the architecture to implement
558 arch_set_thermal_pressure() and arch_scale_thermal_pressure().
560 config BSD_PROCESS_ACCT
561 bool "BSD Process Accounting"
564 If you say Y here, a user level program will be able to instruct the
565 kernel (via a special system call) to write process accounting
566 information to a file: whenever a process exits, information about
567 that process will be appended to the file by the kernel. The
568 information includes things such as creation time, owning user,
569 command name, memory usage, controlling terminal etc. (the complete
570 list is in the struct acct in <file:include/linux/acct.h>). It is
571 up to the user level program to do useful things with this
572 information. This is generally a good idea, so say Y.
574 config BSD_PROCESS_ACCT_V3
575 bool "BSD Process Accounting version 3 file format"
576 depends on BSD_PROCESS_ACCT
579 If you say Y here, the process accounting information is written
580 in a new file format that also logs the process IDs of each
581 process and its parent. Note that this file format is incompatible
582 with previous v0/v1/v2 file formats, so you will need updated tools
583 for processing it. A preliminary version of these tools is available
584 at <http://www.gnu.org/software/acct/>.
587 bool "Export task/process statistics through netlink"
592 Export selected statistics for tasks/processes through the
593 generic netlink interface. Unlike BSD process accounting, the
594 statistics are available during the lifetime of tasks/processes as
595 responses to commands. Like BSD accounting, they are sent to user
600 config TASK_DELAY_ACCT
601 bool "Enable per-task delay accounting"
605 Collect information on time spent by a task waiting for system
606 resources like cpu, synchronous block I/O completion and swapping
607 in pages. Such statistics can help in setting a task's priorities
608 relative to other tasks for cpu, io, rss limits etc.
613 bool "Enable extended accounting over taskstats"
616 Collect extended task accounting data and send the data
617 to userland for processing over the taskstats interface.
621 config TASK_IO_ACCOUNTING
622 bool "Enable per-task storage I/O accounting"
623 depends on TASK_XACCT
625 Collect information on the number of bytes of storage I/O which this
631 bool "Pressure stall information tracking"
633 Collect metrics that indicate how overcommitted the CPU, memory,
634 and IO capacity are in the system.
636 If you say Y here, the kernel will create /proc/pressure/ with the
637 pressure statistics files cpu, memory, and io. These will indicate
638 the share of walltime in which some or all tasks in the system are
639 delayed due to contention of the respective resource.
641 In kernels with cgroup support, cgroups (cgroup2 only) will
642 have cpu.pressure, memory.pressure, and io.pressure files,
643 which aggregate pressure stalls for the grouped tasks only.
645 For more details see Documentation/accounting/psi.rst.
649 config PSI_DEFAULT_DISABLED
650 bool "Require boot parameter to enable pressure stall information tracking"
654 If set, pressure stall information tracking will be disabled
655 per default but can be enabled through passing psi=1 on the
656 kernel commandline during boot.
658 This feature adds some code to the task wakeup and sleep
659 paths of the scheduler. The overhead is too low to affect
660 common scheduling-intense workloads in practice (such as
661 webservers, memcache), but it does show up in artificial
662 scheduler stress tests, such as hackbench.
664 If you are paranoid and not sure what the kernel will be
669 endmenu # "CPU/Task time and stats accounting"
673 depends on SMP || COMPILE_TEST
676 Make sure that CPUs running critical tasks are not disturbed by
677 any source of "noise" such as unbound workqueues, timers, kthreads...
678 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
679 the "isolcpus=" boot parameter.
683 source "kernel/rcu/Kconfig"
690 tristate "Kernel .config support"
692 This option enables the complete Linux kernel ".config" file
693 contents to be saved in the kernel. It provides documentation
694 of which kernel options are used in a running kernel or in an
695 on-disk kernel. This information can be extracted from the kernel
696 image file with the script scripts/extract-ikconfig and used as
697 input to rebuild the current kernel or to build another kernel.
698 It can also be extracted from a running kernel by reading
699 /proc/config.gz if enabled (below).
702 bool "Enable access to .config through /proc/config.gz"
703 depends on IKCONFIG && PROC_FS
705 This option enables access to the kernel configuration file
706 through /proc/config.gz.
709 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
712 This option enables access to the in-kernel headers that are generated during
713 the build process. These can be used to build eBPF tracing programs,
714 or similar programs. If you build the headers as a module, a module called
715 kheaders.ko is built which can be loaded on-demand to get access to headers.
718 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
719 range 12 25 if !H8300
724 Select the minimal kernel log buffer size as a power of 2.
725 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
726 parameter, see below. Any higher size also might be forced
727 by "log_buf_len" boot parameter.
737 config LOG_CPU_MAX_BUF_SHIFT
738 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
741 default 12 if !BASE_SMALL
742 default 0 if BASE_SMALL
745 This option allows to increase the default ring buffer size
746 according to the number of CPUs. The value defines the contribution
747 of each CPU as a power of 2. The used space is typically only few
748 lines however it might be much more when problems are reported,
751 The increased size means that a new buffer has to be allocated and
752 the original static one is unused. It makes sense only on systems
753 with more CPUs. Therefore this value is used only when the sum of
754 contributions is greater than the half of the default kernel ring
755 buffer as defined by LOG_BUF_SHIFT. The default values are set
756 so that more than 16 CPUs are needed to trigger the allocation.
758 Also this option is ignored when "log_buf_len" kernel parameter is
759 used as it forces an exact (power of two) size of the ring buffer.
761 The number of possible CPUs is used for this computation ignoring
762 hotplugging making the computation optimal for the worst case
763 scenario while allowing a simple algorithm to be used from bootup.
765 Examples shift values and their meaning:
766 17 => 128 KB for each CPU
767 16 => 64 KB for each CPU
768 15 => 32 KB for each CPU
769 14 => 16 KB for each CPU
770 13 => 8 KB for each CPU
771 12 => 4 KB for each CPU
773 config PRINTK_SAFE_LOG_BUF_SHIFT
774 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
779 Select the size of an alternate printk per-CPU buffer where messages
780 printed from usafe contexts are temporary stored. One example would
781 be NMI messages, another one - printk recursion. The messages are
782 copied to the main log buffer in a safe context to avoid a deadlock.
783 The value defines the size as a power of 2.
785 Those messages are rare and limited. The largest one is when
786 a backtrace is printed. It usually fits into 4KB. Select
787 8KB if you want to be on the safe side.
790 17 => 128 KB for each CPU
791 16 => 64 KB for each CPU
792 15 => 32 KB for each CPU
793 14 => 16 KB for each CPU
794 13 => 8 KB for each CPU
795 12 => 4 KB for each CPU
798 bool "Printk indexing debugfs interface"
799 depends on PRINTK && DEBUG_FS
801 Add support for indexing of all printk formats known at compile time
802 at <debugfs>/printk/index/<module>.
804 This can be used as part of maintaining daemons which monitor
805 /dev/kmsg, as it permits auditing the printk formats present in a
806 kernel, allowing detection of cases where monitored printks are
807 changed or no longer present.
809 There is no additional runtime cost to printk with this enabled.
812 # Architectures with an unreliable sched_clock() should select this:
814 config HAVE_UNSTABLE_SCHED_CLOCK
817 config GENERIC_SCHED_CLOCK
820 menu "Scheduler features"
823 bool "Enable utilization clamping for RT/FAIR tasks"
824 depends on CPU_FREQ_GOV_SCHEDUTIL
826 This feature enables the scheduler to track the clamped utilization
827 of each CPU based on RUNNABLE tasks scheduled on that CPU.
829 With this option, the user can specify the min and max CPU
830 utilization allowed for RUNNABLE tasks. The max utilization defines
831 the maximum frequency a task should use while the min utilization
832 defines the minimum frequency it should use.
834 Both min and max utilization clamp values are hints to the scheduler,
835 aiming at improving its frequency selection policy, but they do not
836 enforce or grant any specific bandwidth for tasks.
840 config UCLAMP_BUCKETS_COUNT
841 int "Number of supported utilization clamp buckets"
844 depends on UCLAMP_TASK
846 Defines the number of clamp buckets to use. The range of each bucket
847 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
848 number of clamp buckets the finer their granularity and the higher
849 the precision of clamping aggregation and tracking at run-time.
851 For example, with the minimum configuration value we will have 5
852 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
853 be refcounted in the [20..39]% bucket and will set the bucket clamp
854 effective value to 25%.
855 If a second 30% boosted task should be co-scheduled on the same CPU,
856 that task will be refcounted in the same bucket of the first task and
857 it will boost the bucket clamp effective value to 30%.
858 The clamp effective value of a bucket is reset to its nominal value
859 (20% in the example above) when there are no more tasks refcounted in
862 An additional boost/capping margin can be added to some tasks. In the
863 example above the 25% task will be boosted to 30% until it exits the
864 CPU. If that should be considered not acceptable on certain systems,
865 it's always possible to reduce the margin by increasing the number of
866 clamp buckets to trade off used memory for run-time tracking
869 If in doubt, use the default value.
874 # For architectures that want to enable the support for NUMA-affine scheduler
877 config ARCH_SUPPORTS_NUMA_BALANCING
881 # For architectures that prefer to flush all TLBs after a number of pages
882 # are unmapped instead of sending one IPI per page to flush. The architecture
883 # must provide guarantees on what happens if a clean TLB cache entry is
884 # written after the unmap. Details are in mm/rmap.c near the check for
885 # should_defer_flush. The architecture should also consider if the full flush
886 # and the refill costs are offset by the savings of sending fewer IPIs.
887 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
891 def_bool !$(cc-option,$(m64-flag) -D__SIZEOF_INT128__=0) && 64BIT
894 # For architectures that know their GCC __int128 support is sound
896 config ARCH_SUPPORTS_INT128
899 # For architectures that (ab)use NUMA to represent different memory regions
900 # all cpu-local but of different latencies, such as SuperH.
902 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
905 config NUMA_BALANCING
906 bool "Memory placement aware NUMA scheduler"
907 depends on ARCH_SUPPORTS_NUMA_BALANCING
908 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
909 depends on SMP && NUMA && MIGRATION
911 This option adds support for automatic NUMA aware memory/task placement.
912 The mechanism is quite primitive and is based on migrating memory when
913 it has references to the node the task is running on.
915 This system will be inactive on UMA systems.
917 config NUMA_BALANCING_DEFAULT_ENABLED
918 bool "Automatically enable NUMA aware memory/task placement"
920 depends on NUMA_BALANCING
922 If set, automatic NUMA balancing will be enabled if running on a NUMA
926 bool "Control Group support"
929 This option adds support for grouping sets of processes together, for
930 use with process control subsystems such as Cpusets, CFS, memory
931 controls or device isolation.
933 - Documentation/scheduler/sched-design-CFS.rst (CFS)
934 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
935 and resource control)
945 bool "Memory controller"
949 Provides control over the memory footprint of tasks in a cgroup.
953 depends on MEMCG && SWAP
958 depends on MEMCG && !SLOB
966 Generic block IO controller cgroup interface. This is the common
967 cgroup interface which should be used by various IO controlling
970 Currently, CFQ IO scheduler uses it to recognize task groups and
971 control disk bandwidth allocation (proportional time slice allocation)
972 to such task groups. It is also used by bio throttling logic in
973 block layer to implement upper limit in IO rates on a device.
975 This option only enables generic Block IO controller infrastructure.
976 One needs to also enable actual IO controlling logic/policy. For
977 enabling proportional weight division of disk bandwidth in CFQ, set
978 CONFIG_BFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
979 CONFIG_BLK_DEV_THROTTLING=y.
981 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
983 config CGROUP_WRITEBACK
985 depends on MEMCG && BLK_CGROUP
988 menuconfig CGROUP_SCHED
989 bool "CPU controller"
992 This feature lets CPU scheduler recognize task groups and control CPU
993 bandwidth allocation to such task groups. It uses cgroups to group
997 config FAIR_GROUP_SCHED
998 bool "Group scheduling for SCHED_OTHER"
999 depends on CGROUP_SCHED
1000 default CGROUP_SCHED
1002 config CFS_BANDWIDTH
1003 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
1004 depends on FAIR_GROUP_SCHED
1007 This option allows users to define CPU bandwidth rates (limits) for
1008 tasks running within the fair group scheduler. Groups with no limit
1009 set are considered to be unconstrained and will run with no
1011 See Documentation/scheduler/sched-bwc.rst for more information.
1013 config RT_GROUP_SCHED
1014 bool "Group scheduling for SCHED_RR/FIFO"
1015 depends on CGROUP_SCHED
1018 This feature lets you explicitly allocate real CPU bandwidth
1019 to task groups. If enabled, it will also make it impossible to
1020 schedule realtime tasks for non-root users until you allocate
1021 realtime bandwidth for them.
1022 See Documentation/scheduler/sched-rt-group.rst for more information.
1026 config UCLAMP_TASK_GROUP
1027 bool "Utilization clamping per group of tasks"
1028 depends on CGROUP_SCHED
1029 depends on UCLAMP_TASK
1032 This feature enables the scheduler to track the clamped utilization
1033 of each CPU based on RUNNABLE tasks currently scheduled on that CPU.
1035 When this option is enabled, the user can specify a min and max
1036 CPU bandwidth which is allowed for each single task in a group.
1037 The max bandwidth allows to clamp the maximum frequency a task
1038 can use, while the min bandwidth allows to define a minimum
1039 frequency a task will always use.
1041 When task group based utilization clamping is enabled, an eventually
1042 specified task-specific clamp value is constrained by the cgroup
1043 specified clamp value. Both minimum and maximum task clamping cannot
1044 be bigger than the corresponding clamping defined at task group level.
1049 bool "PIDs controller"
1051 Provides enforcement of process number limits in the scope of a
1052 cgroup. Any attempt to fork more processes than is allowed in the
1053 cgroup will fail. PIDs are fundamentally a global resource because it
1054 is fairly trivial to reach PID exhaustion before you reach even a
1055 conservative kmemcg limit. As a result, it is possible to grind a
1056 system to halt without being limited by other cgroup policies. The
1057 PIDs controller is designed to stop this from happening.
1059 It should be noted that organisational operations (such as attaching
1060 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
1061 since the PIDs limit only affects a process's ability to fork, not to
1065 bool "RDMA controller"
1067 Provides enforcement of RDMA resources defined by IB stack.
1068 It is fairly easy for consumers to exhaust RDMA resources, which
1069 can result into resource unavailability to other consumers.
1070 RDMA controller is designed to stop this from happening.
1071 Attaching processes with active RDMA resources to the cgroup
1072 hierarchy is allowed even if can cross the hierarchy's limit.
1074 config CGROUP_FREEZER
1075 bool "Freezer controller"
1077 Provides a way to freeze and unfreeze all tasks in a
1080 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1081 controller includes important in-kernel memory consumers per default.
1083 If you're using cgroup2, say N.
1085 config CGROUP_HUGETLB
1086 bool "HugeTLB controller"
1087 depends on HUGETLB_PAGE
1091 Provides a cgroup controller for HugeTLB pages.
1092 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1093 The limit is enforced during page fault. Since HugeTLB doesn't
1094 support page reclaim, enforcing the limit at page fault time implies
1095 that, the application will get SIGBUS signal if it tries to access
1096 HugeTLB pages beyond its limit. This requires the application to know
1097 beforehand how much HugeTLB pages it would require for its use. The
1098 control group is tracked in the third page lru pointer. This means
1099 that we cannot use the controller with huge page less than 3 pages.
1102 bool "Cpuset controller"
1105 This option will let you create and manage CPUSETs which
1106 allow dynamically partitioning a system into sets of CPUs and
1107 Memory Nodes and assigning tasks to run only within those sets.
1108 This is primarily useful on large SMP or NUMA systems.
1112 config PROC_PID_CPUSET
1113 bool "Include legacy /proc/<pid>/cpuset file"
1117 config CGROUP_DEVICE
1118 bool "Device controller"
1120 Provides a cgroup controller implementing whitelists for
1121 devices which a process in the cgroup can mknod or open.
1123 config CGROUP_CPUACCT
1124 bool "Simple CPU accounting controller"
1126 Provides a simple controller for monitoring the
1127 total CPU consumed by the tasks in a cgroup.
1130 bool "Perf controller"
1131 depends on PERF_EVENTS
1133 This option extends the perf per-cpu mode to restrict monitoring
1134 to threads which belong to the cgroup specified and run on the
1135 designated cpu. Or this can be used to have cgroup ID in samples
1136 so that it can monitor performance events among cgroups.
1141 bool "Support for eBPF programs attached to cgroups"
1142 depends on BPF_SYSCALL
1143 select SOCK_CGROUP_DATA
1145 Allow attaching eBPF programs to a cgroup using the bpf(2)
1146 syscall command BPF_PROG_ATTACH.
1148 In which context these programs are accessed depends on the type
1149 of attachment. For instance, programs that are attached using
1150 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1154 bool "Misc resource controller"
1157 Provides a controller for miscellaneous resources on a host.
1159 Miscellaneous scalar resources are the resources on the host system
1160 which cannot be abstracted like the other cgroups. This controller
1161 tracks and limits the miscellaneous resources used by a process
1162 attached to a cgroup hierarchy.
1164 For more information, please check misc cgroup section in
1165 /Documentation/admin-guide/cgroup-v2.rst.
1168 bool "Debug controller"
1170 depends on DEBUG_KERNEL
1172 This option enables a simple controller that exports
1173 debugging information about the cgroups framework. This
1174 controller is for control cgroup debugging only. Its
1175 interfaces are not stable.
1179 config SOCK_CGROUP_DATA
1185 menuconfig NAMESPACES
1186 bool "Namespaces support" if EXPERT
1187 depends on MULTIUSER
1190 Provides the way to make tasks work with different objects using
1191 the same id. For example same IPC id may refer to different objects
1192 or same user id or pid may refer to different tasks when used in
1193 different namespaces.
1198 bool "UTS namespace"
1201 In this namespace tasks see different info provided with the
1205 bool "TIME namespace"
1206 depends on GENERIC_VDSO_TIME_NS
1209 In this namespace boottime and monotonic clocks can be set.
1210 The time will keep going with the same pace.
1213 bool "IPC namespace"
1214 depends on (SYSVIPC || POSIX_MQUEUE)
1217 In this namespace tasks work with IPC ids which correspond to
1218 different IPC objects in different namespaces.
1221 bool "User namespace"
1224 This allows containers, i.e. vservers, to use user namespaces
1225 to provide different user info for different servers.
1227 When user namespaces are enabled in the kernel it is
1228 recommended that the MEMCG option also be enabled and that
1229 user-space use the memory control groups to limit the amount
1230 of memory a memory unprivileged users can use.
1235 bool "PID Namespaces"
1238 Support process id namespaces. This allows having multiple
1239 processes with the same pid as long as they are in different
1240 pid namespaces. This is a building block of containers.
1243 bool "Network namespace"
1247 Allow user space to create what appear to be multiple instances
1248 of the network stack.
1252 config CHECKPOINT_RESTORE
1253 bool "Checkpoint/restore support"
1254 select PROC_CHILDREN
1258 Enables additional kernel features in a sake of checkpoint/restore.
1259 In particular it adds auxiliary prctl codes to setup process text,
1260 data and heap segment sizes, and a few additional /proc filesystem
1263 If unsure, say N here.
1265 config SCHED_AUTOGROUP
1266 bool "Automatic process group scheduling"
1269 select FAIR_GROUP_SCHED
1271 This option optimizes the scheduler for common desktop workloads by
1272 automatically creating and populating task groups. This separation
1273 of workloads isolates aggressive CPU burners (like build jobs) from
1274 desktop applications. Task group autogeneration is currently based
1277 config SYSFS_DEPRECATED
1278 bool "Enable deprecated sysfs features to support old userspace tools"
1282 This option adds code that switches the layout of the "block" class
1283 devices, to not show up in /sys/class/block/, but only in
1286 This switch is only active when the sysfs.deprecated=1 boot option is
1287 passed or the SYSFS_DEPRECATED_V2 option is set.
1289 This option allows new kernels to run on old distributions and tools,
1290 which might get confused by /sys/class/block/. Since 2007/2008 all
1291 major distributions and tools handle this just fine.
1293 Recent distributions and userspace tools after 2009/2010 depend on
1294 the existence of /sys/class/block/, and will not work with this
1297 Only if you are using a new kernel on an old distribution, you might
1300 config SYSFS_DEPRECATED_V2
1301 bool "Enable deprecated sysfs features by default"
1304 depends on SYSFS_DEPRECATED
1306 Enable deprecated sysfs by default.
1308 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1311 Only if you are using a new kernel on an old distribution, you might
1312 need to say Y here. Even then, odds are you would not need it
1313 enabled, you can always pass the boot option if absolutely necessary.
1316 bool "Kernel->user space relay support (formerly relayfs)"
1319 This option enables support for relay interface support in
1320 certain file systems (such as debugfs).
1321 It is designed to provide an efficient mechanism for tools and
1322 facilities to relay large amounts of data from kernel space to
1327 config BLK_DEV_INITRD
1328 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1330 The initial RAM filesystem is a ramfs which is loaded by the
1331 boot loader (loadlin or lilo) and that is mounted as root
1332 before the normal boot procedure. It is typically used to
1333 load modules needed to mount the "real" root file system,
1334 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1336 If RAM disk support (BLK_DEV_RAM) is also included, this
1337 also enables initial RAM disk (initrd) support and adds
1338 15 Kbytes (more on some other architectures) to the kernel size.
1344 source "usr/Kconfig"
1349 bool "Boot config support"
1350 select BLK_DEV_INITRD
1352 Extra boot config allows system admin to pass a config file as
1353 complemental extension of kernel cmdline when booting.
1354 The boot config file must be attached at the end of initramfs
1355 with checksum, size and magic word.
1356 See <file:Documentation/admin-guide/bootconfig.rst> for details.
1361 prompt "Compiler optimization level"
1362 default CC_OPTIMIZE_FOR_PERFORMANCE
1364 config CC_OPTIMIZE_FOR_PERFORMANCE
1365 bool "Optimize for performance (-O2)"
1367 This is the default optimization level for the kernel, building
1368 with the "-O2" compiler flag for best performance and most
1369 helpful compile-time warnings.
1371 config CC_OPTIMIZE_FOR_PERFORMANCE_O3
1372 bool "Optimize more for performance (-O3)"
1375 Choosing this option will pass "-O3" to your compiler to optimize
1376 the kernel yet more for performance.
1378 config CC_OPTIMIZE_FOR_SIZE
1379 bool "Optimize for size (-Os)"
1381 Choosing this option will pass "-Os" to your compiler resulting
1382 in a smaller kernel.
1386 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1389 This requires that the arch annotates or otherwise protects
1390 its external entry points from being discarded. Linker scripts
1391 must also merge .text.*, .data.*, and .bss.* correctly into
1392 output sections. Care must be taken not to pull in unrelated
1393 sections (e.g., '.text.init'). Typically '.' in section names
1394 is used to distinguish them from label names / C identifiers.
1396 config LD_DEAD_CODE_DATA_ELIMINATION
1397 bool "Dead code and data elimination (EXPERIMENTAL)"
1398 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1400 depends on $(cc-option,-ffunction-sections -fdata-sections)
1401 depends on $(ld-option,--gc-sections)
1403 Enable this if you want to do dead code and data elimination with
1404 the linker by compiling with -ffunction-sections -fdata-sections,
1405 and linking with --gc-sections.
1407 This can reduce on disk and in-memory size of the kernel
1408 code and static data, particularly for small configs and
1409 on small systems. This has the possibility of introducing
1410 silently broken kernel if the required annotations are not
1411 present. This option is not well tested yet, so use at your
1414 config LD_ORPHAN_WARN
1416 depends on ARCH_WANT_LD_ORPHAN_WARN
1417 depends on !LD_IS_LLD || LLD_VERSION >= 110000
1418 depends on $(ld-option,--orphan-handling=warn)
1426 config SYSCTL_EXCEPTION_TRACE
1429 Enable support for /proc/sys/debug/exception-trace.
1431 config SYSCTL_ARCH_UNALIGN_NO_WARN
1434 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1435 Allows arch to define/use @no_unaligned_warning to possibly warn
1436 about unaligned access emulation going on under the hood.
1438 config SYSCTL_ARCH_UNALIGN_ALLOW
1441 Enable support for /proc/sys/kernel/unaligned-trap
1442 Allows arches to define/use @unaligned_enabled to runtime toggle
1443 the unaligned access emulation.
1444 see arch/parisc/kernel/unaligned.c for reference
1446 config HAVE_PCSPKR_PLATFORM
1449 # interpreter that classic socket filters depend on
1454 bool "Configure standard kernel features (expert users)"
1455 # Unhide debug options, to make the on-by-default options visible
1458 This option allows certain base kernel options and settings
1459 to be disabled or tweaked. This is for specialized
1460 environments which can tolerate a "non-standard" kernel.
1461 Only use this if you really know what you are doing.
1464 bool "Enable 16-bit UID system calls" if EXPERT
1465 depends on HAVE_UID16 && MULTIUSER
1468 This enables the legacy 16-bit UID syscall wrappers.
1471 bool "Multiple users, groups and capabilities support" if EXPERT
1474 This option enables support for non-root users, groups and
1477 If you say N here, all processes will run with UID 0, GID 0, and all
1478 possible capabilities. Saying N here also compiles out support for
1479 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1482 If unsure, say Y here.
1484 config SGETMASK_SYSCALL
1485 bool "sgetmask/ssetmask syscalls support" if EXPERT
1486 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1488 sys_sgetmask and sys_ssetmask are obsolete system calls
1489 no longer supported in libc but still enabled by default in some
1492 If unsure, leave the default option here.
1494 config SYSFS_SYSCALL
1495 bool "Sysfs syscall support" if EXPERT
1498 sys_sysfs is an obsolete system call no longer supported in libc.
1499 Note that disabling this option is more secure but might break
1500 compatibility with some systems.
1502 If unsure say Y here.
1505 bool "open by fhandle syscalls" if EXPERT
1509 If you say Y here, a user level program will be able to map
1510 file names to handle and then later use the handle for
1511 different file system operations. This is useful in implementing
1512 userspace file servers, which now track files using handles instead
1513 of names. The handle would remain the same even if file names
1514 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1518 bool "Posix Clocks & timers" if EXPERT
1521 This includes native support for POSIX timers to the kernel.
1522 Some embedded systems have no use for them and therefore they
1523 can be configured out to reduce the size of the kernel image.
1525 When this option is disabled, the following syscalls won't be
1526 available: timer_create, timer_gettime: timer_getoverrun,
1527 timer_settime, timer_delete, clock_adjtime, getitimer,
1528 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1529 clock_getres and clock_nanosleep syscalls will be limited to
1530 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1536 bool "Enable support for printk" if EXPERT
1539 This option enables normal printk support. Removing it
1540 eliminates most of the message strings from the kernel image
1541 and makes the kernel more or less silent. As this makes it
1542 very difficult to diagnose system problems, saying N here is
1543 strongly discouraged.
1546 bool "BUG() support" if EXPERT
1549 Disabling this option eliminates support for BUG and WARN, reducing
1550 the size of your kernel image and potentially quietly ignoring
1551 numerous fatal conditions. You should only consider disabling this
1552 option for embedded systems with no facilities for reporting errors.
1558 bool "Enable ELF core dumps" if EXPERT
1560 Enable support for generating core dumps. Disabling saves about 4k.
1563 config PCSPKR_PLATFORM
1564 bool "Enable PC-Speaker support" if EXPERT
1565 depends on HAVE_PCSPKR_PLATFORM
1569 This option allows to disable the internal PC-Speaker
1570 support, saving some memory.
1574 bool "Enable full-sized data structures for core" if EXPERT
1576 Disabling this option reduces the size of miscellaneous core
1577 kernel data structures. This saves memory on small machines,
1578 but may reduce performance.
1581 bool "Enable futex support" if EXPERT
1585 Disabling this option will cause the kernel to be built without
1586 support for "fast userspace mutexes". The resulting kernel may not
1587 run glibc-based applications correctly.
1591 depends on FUTEX && RT_MUTEXES
1594 config HAVE_FUTEX_CMPXCHG
1598 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1599 is implemented and always working. This removes a couple of runtime
1603 bool "Enable eventpoll support" if EXPERT
1606 Disabling this option will cause the kernel to be built without
1607 support for epoll family of system calls.
1610 bool "Enable signalfd() system call" if EXPERT
1613 Enable the signalfd() system call that allows to receive signals
1614 on a file descriptor.
1619 bool "Enable timerfd() system call" if EXPERT
1622 Enable the timerfd() system call that allows to receive timer
1623 events on a file descriptor.
1628 bool "Enable eventfd() system call" if EXPERT
1631 Enable the eventfd() system call that allows to receive both
1632 kernel notification (ie. KAIO) or userspace notifications.
1637 bool "Use full shmem filesystem" if EXPERT
1641 The shmem is an internal filesystem used to manage shared memory.
1642 It is backed by swap and manages resource limits. It is also exported
1643 to userspace as tmpfs if TMPFS is enabled. Disabling this
1644 option replaces shmem and tmpfs with the much simpler ramfs code,
1645 which may be appropriate on small systems without swap.
1648 bool "Enable AIO support" if EXPERT
1651 This option enables POSIX asynchronous I/O which may by used
1652 by some high performance threaded applications. Disabling
1653 this option saves about 7k.
1656 bool "Enable IO uring support" if EXPERT
1660 This option enables support for the io_uring interface, enabling
1661 applications to submit and complete IO through submission and
1662 completion rings that are shared between the kernel and application.
1664 config ADVISE_SYSCALLS
1665 bool "Enable madvise/fadvise syscalls" if EXPERT
1668 This option enables the madvise and fadvise syscalls, used by
1669 applications to advise the kernel about their future memory or file
1670 usage, improving performance. If building an embedded system where no
1671 applications use these syscalls, you can disable this option to save
1674 config HAVE_ARCH_USERFAULTFD_WP
1677 Arch has userfaultfd write protection support
1679 config HAVE_ARCH_USERFAULTFD_MINOR
1682 Arch has userfaultfd minor fault support
1685 bool "Enable membarrier() system call" if EXPERT
1688 Enable the membarrier() system call that allows issuing memory
1689 barriers across all running threads, which can be used to distribute
1690 the cost of user-space memory barriers asymmetrically by transforming
1691 pairs of memory barriers into pairs consisting of membarrier() and a
1697 bool "Load all symbols for debugging/ksymoops" if EXPERT
1700 Say Y here to let the kernel print out symbolic crash information and
1701 symbolic stack backtraces. This increases the size of the kernel
1702 somewhat, as all symbols have to be loaded into the kernel image.
1705 bool "Include all symbols in kallsyms"
1706 depends on DEBUG_KERNEL && KALLSYMS
1708 Normally kallsyms only contains the symbols of functions for nicer
1709 OOPS messages and backtraces (i.e., symbols from the text and inittext
1710 sections). This is sufficient for most cases. And only in very rare
1711 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1712 names of variables from the data sections, etc).
1714 This option makes sure that all symbols are loaded into the kernel
1715 image (i.e., symbols from all sections) in cost of increased kernel
1716 size (depending on the kernel configuration, it may be 300KiB or
1717 something like this).
1719 Say N unless you really need all symbols.
1721 config KALLSYMS_ABSOLUTE_PERCPU
1724 default X86_64 && SMP
1726 config KALLSYMS_BASE_RELATIVE
1731 Instead of emitting them as absolute values in the native word size,
1732 emit the symbol references in the kallsyms table as 32-bit entries,
1733 each containing a relative value in the range [base, base + U32_MAX]
1734 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1735 an absolute value in the range [0, S32_MAX] or a relative value in the
1736 range [base, base + S32_MAX], where base is the lowest relative symbol
1737 address encountered in the image.
1739 On 64-bit builds, this reduces the size of the address table by 50%,
1740 but more importantly, it results in entries whose values are build
1741 time constants, and no relocation pass is required at runtime to fix
1742 up the entries based on the runtime load address of the kernel.
1744 # end of the "standard kernel features (expert users)" menu
1746 # syscall, maps, verifier
1749 bool "Enable userfaultfd() system call"
1752 Enable the userfaultfd() system call that allows to intercept and
1753 handle page faults in userland.
1755 config ARCH_HAS_MEMBARRIER_CALLBACKS
1758 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1762 bool "Enable kcmp() system call" if EXPERT
1764 Enable the kernel resource comparison system call. It provides
1765 user-space with the ability to compare two processes to see if they
1766 share a common resource, such as a file descriptor or even virtual
1772 bool "Enable rseq() system call" if EXPERT
1774 depends on HAVE_RSEQ
1777 Enable the restartable sequences system call. It provides a
1778 user-space cache for the current CPU number value, which
1779 speeds up getting the current CPU number from user-space,
1780 as well as an ABI to speed up user-space operations on
1787 bool "Enabled debugging of rseq() system call" if EXPERT
1788 depends on RSEQ && DEBUG_KERNEL
1790 Enable extra debugging checks for the rseq system call.
1795 bool "Embedded system"
1798 This option should be enabled if compiling the kernel for
1799 an embedded system so certain expert options are available
1802 config HAVE_PERF_EVENTS
1805 See tools/perf/design.txt for details.
1807 config PERF_USE_VMALLOC
1810 See tools/perf/design.txt for details
1813 bool "PC/104 support" if EXPERT
1815 Expose PC/104 form factor device drivers and options available for
1816 selection and configuration. Enable this option if your target
1817 machine has a PC/104 bus.
1819 menu "Kernel Performance Events And Counters"
1822 bool "Kernel performance events and counters"
1823 default y if PROFILING
1824 depends on HAVE_PERF_EVENTS
1828 Enable kernel support for various performance events provided
1829 by software and hardware.
1831 Software events are supported either built-in or via the
1832 use of generic tracepoints.
1834 Most modern CPUs support performance events via performance
1835 counter registers. These registers count the number of certain
1836 types of hw events: such as instructions executed, cachemisses
1837 suffered, or branches mis-predicted - without slowing down the
1838 kernel or applications. These registers can also trigger interrupts
1839 when a threshold number of events have passed - and can thus be
1840 used to profile the code that runs on that CPU.
1842 The Linux Performance Event subsystem provides an abstraction of
1843 these software and hardware event capabilities, available via a
1844 system call and used by the "perf" utility in tools/perf/. It
1845 provides per task and per CPU counters, and it provides event
1846 capabilities on top of those.
1850 config DEBUG_PERF_USE_VMALLOC
1852 bool "Debug: use vmalloc to back perf mmap() buffers"
1853 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1854 select PERF_USE_VMALLOC
1856 Use vmalloc memory to back perf mmap() buffers.
1858 Mostly useful for debugging the vmalloc code on platforms
1859 that don't require it.
1865 config VM_EVENT_COUNTERS
1867 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1869 VM event counters are needed for event counts to be shown.
1870 This option allows the disabling of the VM event counters
1871 on EXPERT systems. /proc/vmstat will only show page counts
1872 if VM event counters are disabled.
1876 bool "Enable SLUB debugging support" if EXPERT
1877 depends on SLUB && SYSFS
1879 SLUB has extensive debug support features. Disabling these can
1880 result in significant savings in code size. This also disables
1881 SLUB sysfs support. /sys/slab will not exist and there will be
1882 no support for cache validation etc.
1885 bool "Disable heap randomization"
1888 Randomizing heap placement makes heap exploits harder, but it
1889 also breaks ancient binaries (including anything libc5 based).
1890 This option changes the bootup default to heap randomization
1891 disabled, and can be overridden at runtime by setting
1892 /proc/sys/kernel/randomize_va_space to 2.
1894 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1897 prompt "Choose SLAB allocator"
1900 This option allows to select a slab allocator.
1904 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1906 The regular slab allocator that is established and known to work
1907 well in all environments. It organizes cache hot objects in
1908 per cpu and per node queues.
1911 bool "SLUB (Unqueued Allocator)"
1912 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1914 SLUB is a slab allocator that minimizes cache line usage
1915 instead of managing queues of cached objects (SLAB approach).
1916 Per cpu caching is realized using slabs of objects instead
1917 of queues of objects. SLUB can use memory efficiently
1918 and has enhanced diagnostics. SLUB is the default choice for
1923 bool "SLOB (Simple Allocator)"
1925 SLOB replaces the stock allocator with a drastically simpler
1926 allocator. SLOB is generally more space efficient but
1927 does not perform as well on large systems.
1931 config SLAB_MERGE_DEFAULT
1932 bool "Allow slab caches to be merged"
1935 For reduced kernel memory fragmentation, slab caches can be
1936 merged when they share the same size and other characteristics.
1937 This carries a risk of kernel heap overflows being able to
1938 overwrite objects from merged caches (and more easily control
1939 cache layout), which makes such heap attacks easier to exploit
1940 by attackers. By keeping caches unmerged, these kinds of exploits
1941 can usually only damage objects in the same cache. To disable
1942 merging at runtime, "slab_nomerge" can be passed on the kernel
1945 config SLAB_FREELIST_RANDOM
1946 bool "Randomize slab freelist"
1947 depends on SLAB || SLUB
1949 Randomizes the freelist order used on creating new pages. This
1950 security feature reduces the predictability of the kernel slab
1951 allocator against heap overflows.
1953 config SLAB_FREELIST_HARDENED
1954 bool "Harden slab freelist metadata"
1955 depends on SLAB || SLUB
1957 Many kernel heap attacks try to target slab cache metadata and
1958 other infrastructure. This options makes minor performance
1959 sacrifices to harden the kernel slab allocator against common
1960 freelist exploit methods. Some slab implementations have more
1961 sanity-checking than others. This option is most effective with
1964 config SHUFFLE_PAGE_ALLOCATOR
1965 bool "Page allocator randomization"
1966 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1968 Randomization of the page allocator improves the average
1969 utilization of a direct-mapped memory-side-cache. See section
1970 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1971 6.2a specification for an example of how a platform advertises
1972 the presence of a memory-side-cache. There are also incidental
1973 security benefits as it reduces the predictability of page
1974 allocations to compliment SLAB_FREELIST_RANDOM, but the
1975 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1976 10th order of pages is selected based on cache utilization
1979 While the randomization improves cache utilization it may
1980 negatively impact workloads on platforms without a cache. For
1981 this reason, by default, the randomization is enabled only
1982 after runtime detection of a direct-mapped memory-side-cache.
1983 Otherwise, the randomization may be force enabled with the
1984 'page_alloc.shuffle' kernel command line parameter.
1988 config SLUB_CPU_PARTIAL
1990 depends on SLUB && SMP
1991 bool "SLUB per cpu partial cache"
1993 Per cpu partial caches accelerate objects allocation and freeing
1994 that is local to a processor at the price of more indeterminism
1995 in the latency of the free. On overflow these caches will be cleared
1996 which requires the taking of locks that may cause latency spikes.
1997 Typically one would choose no for a realtime system.
1999 config MMAP_ALLOW_UNINITIALIZED
2000 bool "Allow mmapped anonymous memory to be uninitialized"
2001 depends on EXPERT && !MMU
2004 Normally, and according to the Linux spec, anonymous memory obtained
2005 from mmap() has its contents cleared before it is passed to
2006 userspace. Enabling this config option allows you to request that
2007 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
2008 providing a huge performance boost. If this option is not enabled,
2009 then the flag will be ignored.
2011 This is taken advantage of by uClibc's malloc(), and also by
2012 ELF-FDPIC binfmt's brk and stack allocator.
2014 Because of the obvious security issues, this option should only be
2015 enabled on embedded devices where you control what is run in
2016 userspace. Since that isn't generally a problem on no-MMU systems,
2017 it is normally safe to say Y here.
2019 See Documentation/admin-guide/mm/nommu-mmap.rst for more information.
2021 config SYSTEM_DATA_VERIFICATION
2023 select SYSTEM_TRUSTED_KEYRING
2027 select ASYMMETRIC_KEY_TYPE
2028 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
2031 select X509_CERTIFICATE_PARSER
2032 select PKCS7_MESSAGE_PARSER
2034 Provide PKCS#7 message verification using the contents of the system
2035 trusted keyring to provide public keys. This then can be used for
2036 module verification, kexec image verification and firmware blob
2040 bool "Profiling support"
2042 Say Y here to enable the extended profiling support mechanisms used
2046 # Place an empty function call at each tracepoint site. Can be
2047 # dynamically changed for a probe function.
2052 endmenu # General setup
2054 source "arch/Kconfig"
2061 default 0 if BASE_FULL
2062 default 1 if !BASE_FULL
2064 config MODULE_SIG_FORMAT
2066 select SYSTEM_DATA_VERIFICATION
2069 bool "Enable loadable module support"
2072 Kernel modules are small pieces of compiled code which can
2073 be inserted in the running kernel, rather than being
2074 permanently built into the kernel. You use the "modprobe"
2075 tool to add (and sometimes remove) them. If you say Y here,
2076 many parts of the kernel can be built as modules (by
2077 answering M instead of Y where indicated): this is most
2078 useful for infrequently used options which are not required
2079 for booting. For more information, see the man pages for
2080 modprobe, lsmod, modinfo, insmod and rmmod.
2082 If you say Y here, you will need to run "make
2083 modules_install" to put the modules under /lib/modules/
2084 where modprobe can find them (you may need to be root to do
2091 config MODULE_FORCE_LOAD
2092 bool "Forced module loading"
2095 Allow loading of modules without version information (ie. modprobe
2096 --force). Forced module loading sets the 'F' (forced) taint flag and
2097 is usually a really bad idea.
2099 config MODULE_UNLOAD
2100 bool "Module unloading"
2102 Without this option you will not be able to unload any
2103 modules (note that some modules may not be unloadable
2104 anyway), which makes your kernel smaller, faster
2105 and simpler. If unsure, say Y.
2107 config MODULE_FORCE_UNLOAD
2108 bool "Forced module unloading"
2109 depends on MODULE_UNLOAD
2111 This option allows you to force a module to unload, even if the
2112 kernel believes it is unsafe: the kernel will remove the module
2113 without waiting for anyone to stop using it (using the -f option to
2114 rmmod). This is mainly for kernel developers and desperate users.
2118 bool "Module versioning support"
2120 Usually, you have to use modules compiled with your kernel.
2121 Saying Y here makes it sometimes possible to use modules
2122 compiled for different kernels, by adding enough information
2123 to the modules to (hopefully) spot any changes which would
2124 make them incompatible with the kernel you are running. If
2127 config ASM_MODVERSIONS
2129 default HAVE_ASM_MODVERSIONS && MODVERSIONS
2131 This enables module versioning for exported symbols also from
2132 assembly. This can be enabled only when the target architecture
2135 config MODULE_REL_CRCS
2137 depends on MODVERSIONS
2139 config MODULE_SRCVERSION_ALL
2140 bool "Source checksum for all modules"
2142 Modules which contain a MODULE_VERSION get an extra "srcversion"
2143 field inserted into their modinfo section, which contains a
2144 sum of the source files which made it. This helps maintainers
2145 see exactly which source was used to build a module (since
2146 others sometimes change the module source without updating
2147 the version). With this option, such a "srcversion" field
2148 will be created for all modules. If unsure, say N.
2151 bool "Module signature verification"
2152 select MODULE_SIG_FORMAT
2154 Check modules for valid signatures upon load: the signature
2155 is simply appended to the module. For more information see
2156 <file:Documentation/admin-guide/module-signing.rst>.
2158 Note that this option adds the OpenSSL development packages as a
2159 kernel build dependency so that the signing tool can use its crypto
2162 You should enable this option if you wish to use either
2163 CONFIG_SECURITY_LOCKDOWN_LSM or lockdown functionality imposed via
2164 another LSM - otherwise unsigned modules will be loadable regardless
2165 of the lockdown policy.
2167 !!!WARNING!!! If you enable this option, you MUST make sure that the
2168 module DOES NOT get stripped after being signed. This includes the
2169 debuginfo strip done by some packagers (such as rpmbuild) and
2170 inclusion into an initramfs that wants the module size reduced.
2172 config MODULE_SIG_FORCE
2173 bool "Require modules to be validly signed"
2174 depends on MODULE_SIG
2176 Reject unsigned modules or signed modules for which we don't have a
2177 key. Without this, such modules will simply taint the kernel.
2179 config MODULE_SIG_ALL
2180 bool "Automatically sign all modules"
2182 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2184 Sign all modules during make modules_install. Without this option,
2185 modules must be signed manually, using the scripts/sign-file tool.
2187 comment "Do not forget to sign required modules with scripts/sign-file"
2188 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2191 prompt "Which hash algorithm should modules be signed with?"
2192 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2194 This determines which sort of hashing algorithm will be used during
2195 signature generation. This algorithm _must_ be built into the kernel
2196 directly so that signature verification can take place. It is not
2197 possible to load a signed module containing the algorithm to check
2198 the signature on that module.
2200 config MODULE_SIG_SHA1
2201 bool "Sign modules with SHA-1"
2204 config MODULE_SIG_SHA224
2205 bool "Sign modules with SHA-224"
2206 select CRYPTO_SHA256
2208 config MODULE_SIG_SHA256
2209 bool "Sign modules with SHA-256"
2210 select CRYPTO_SHA256
2212 config MODULE_SIG_SHA384
2213 bool "Sign modules with SHA-384"
2214 select CRYPTO_SHA512
2216 config MODULE_SIG_SHA512
2217 bool "Sign modules with SHA-512"
2218 select CRYPTO_SHA512
2222 config MODULE_SIG_HASH
2224 depends on MODULE_SIG || IMA_APPRAISE_MODSIG
2225 default "sha1" if MODULE_SIG_SHA1
2226 default "sha224" if MODULE_SIG_SHA224
2227 default "sha256" if MODULE_SIG_SHA256
2228 default "sha384" if MODULE_SIG_SHA384
2229 default "sha512" if MODULE_SIG_SHA512
2232 prompt "Module compression mode"
2234 This option allows you to choose the algorithm which will be used to
2235 compress modules when 'make modules_install' is run. (or, you can
2236 choose to not compress modules at all.)
2238 External modules will also be compressed in the same way during the
2241 For modules inside an initrd or initramfs, it's more efficient to
2242 compress the whole initrd or initramfs instead.
2244 This is fully compatible with signed modules.
2246 Please note that the tool used to load modules needs to support the
2247 corresponding algorithm. module-init-tools MAY support gzip, and kmod
2248 MAY support gzip, xz and zstd.
2250 Your build system needs to provide the appropriate compression tool
2251 to compress the modules.
2253 If in doubt, select 'None'.
2255 config MODULE_COMPRESS_NONE
2258 Do not compress modules. The installed modules are suffixed
2261 config MODULE_COMPRESS_GZIP
2264 Compress modules with GZIP. The installed modules are suffixed
2267 config MODULE_COMPRESS_XZ
2270 Compress modules with XZ. The installed modules are suffixed
2273 config MODULE_COMPRESS_ZSTD
2276 Compress modules with ZSTD. The installed modules are suffixed
2281 config MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS
2282 bool "Allow loading of modules with missing namespace imports"
2284 Symbols exported with EXPORT_SYMBOL_NS*() are considered exported in
2285 a namespace. A module that makes use of a symbol exported with such a
2286 namespace is required to import the namespace via MODULE_IMPORT_NS().
2287 There is no technical reason to enforce correct namespace imports,
2288 but it creates consistency between symbols defining namespaces and
2289 users importing namespaces they make use of. This option relaxes this
2290 requirement and lifts the enforcement when loading a module.
2294 config MODPROBE_PATH
2295 string "Path to modprobe binary"
2296 default "/sbin/modprobe"
2298 When kernel code requests a module, it does so by calling
2299 the "modprobe" userspace utility. This option allows you to
2300 set the path where that binary is found. This can be changed
2301 at runtime via the sysctl file
2302 /proc/sys/kernel/modprobe. Setting this to the empty string
2303 removes the kernel's ability to request modules (but
2304 userspace can still load modules explicitly).
2306 config TRIM_UNUSED_KSYMS
2307 bool "Trim unused exported kernel symbols" if EXPERT
2308 depends on !COMPILE_TEST
2310 The kernel and some modules make many symbols available for
2311 other modules to use via EXPORT_SYMBOL() and variants. Depending
2312 on the set of modules being selected in your kernel configuration,
2313 many of those exported symbols might never be used.
2315 This option allows for unused exported symbols to be dropped from
2316 the build. In turn, this provides the compiler more opportunities
2317 (especially when using LTO) for optimizing the code and reducing
2318 binary size. This might have some security advantages as well.
2320 If unsure, or if you need to build out-of-tree modules, say N.
2322 config UNUSED_KSYMS_WHITELIST
2323 string "Whitelist of symbols to keep in ksymtab"
2324 depends on TRIM_UNUSED_KSYMS
2326 By default, all unused exported symbols will be un-exported from the
2327 build when TRIM_UNUSED_KSYMS is selected.
2329 UNUSED_KSYMS_WHITELIST allows to whitelist symbols that must be kept
2330 exported at all times, even in absence of in-tree users. The value to
2331 set here is the path to a text file containing the list of symbols,
2332 one per line. The path can be absolute, or relative to the kernel
2337 config MODULES_TREE_LOOKUP
2339 depends on PERF_EVENTS || TRACING || CFI_CLANG
2341 config INIT_ALL_POSSIBLE
2344 Back when each arch used to define their own cpu_online_mask and
2345 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2346 with all 1s, and others with all 0s. When they were centralised,
2347 it was better to provide this option than to break all the archs
2348 and have several arch maintainers pursuing me down dark alleys.
2350 source "block/Kconfig"
2352 config PREEMPT_NOTIFIERS
2362 Build a simple ASN.1 grammar compiler that produces a bytecode output
2363 that can be interpreted by the ASN.1 stream decoder and used to
2364 inform it as to what tags are to be expected in a stream and what
2365 functions to call on what tags.
2367 source "kernel/Kconfig.locks"
2369 config ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
2372 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2375 # It may be useful for an architecture to override the definitions of the
2376 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2377 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2378 # different calling convention for syscalls. They can also override the
2379 # macros for not-implemented syscalls in kernel/sys_ni.c and
2380 # kernel/time/posix-stubs.c. All these overrides need to be available in
2381 # <asm/syscall_wrapper.h>.
2382 config ARCH_HAS_SYSCALL_WRAPPER
projects / platform / kernel / linux-rpi.git / blob