1 # SPDX-License-Identifier: GPL-2.0-only
6 default "/lib/modules/$(shell,uname -r)/.config"
7 default "/etc/kernel-config"
8 default "/boot/config-$(shell,uname -r)"
10 default "arch/$(ARCH)/defconfig"
13 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
17 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
21 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
25 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
28 def_bool $(success,$(srctree)/scripts/cc-can-link.sh $(CC))
30 config CC_HAS_ASM_GOTO
31 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
33 config CC_HAS_WARN_MAYBE_UNINITIALIZED
34 def_bool $(cc-option,-Wmaybe-uninitialized)
36 GCC >= 4.7 supports this option.
38 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
40 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
41 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
43 GCC's -Wmaybe-uninitialized is not reliable by definition.
44 Lots of false positive warnings are produced in some cases.
46 If this option is enabled, -Wno-maybe-uninitialzed is passed
47 to the compiler to suppress maybe-uninitialized warnings.
56 config BUILDTIME_EXTABLE_SORT
59 config THREAD_INFO_IN_TASK
62 Select this to move thread_info off the stack into task_struct. To
63 make this work, an arch will need to remove all thread_info fields
64 except flags and fix any runtime bugs.
66 One subtle change that will be needed is to use try_get_task_stack()
67 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
76 depends on BROKEN || !SMP
79 config INIT_ENV_ARG_LIMIT
84 Maximum of each of the number of arguments and environment
85 variables passed to init from the kernel command line.
88 bool "Compile also drivers which will not load"
92 Some drivers can be compiled on a different platform than they are
93 intended to be run on. Despite they cannot be loaded there (or even
94 when they load they cannot be used due to missing HW support),
95 developers still, opposing to distributors, might want to build such
96 drivers to compile-test them.
98 If you are a developer and want to build everything available, say Y
99 here. If you are a user/distributor, say N here to exclude useless
100 drivers to be distributed.
103 bool "Compile test headers that should be standalone compilable"
105 Compile test headers listed in header-test-y target to ensure they are
106 self-contained, i.e. compilable as standalone units.
108 If you are a developer or tester and want to ensure the requested
109 headers are self-contained, say Y here. Otherwise, choose N.
111 config KERNEL_HEADER_TEST
112 bool "Compile test kernel headers"
113 depends on HEADER_TEST
115 Headers in include/ are used to build external moduls.
116 Compile test them to ensure they are self-contained, i.e.
117 compilable as standalone units.
119 If you are a developer or tester and want to ensure the headers
120 in include/ are self-contained, say Y here. Otherwise, choose N.
122 config UAPI_HEADER_TEST
123 bool "Compile test UAPI headers"
124 depends on HEADER_TEST && HEADERS_INSTALL && CC_CAN_LINK
126 Compile test headers exported to user-space to ensure they are
127 self-contained, i.e. compilable as standalone units.
129 If you are a developer or tester and want to ensure the exported
130 headers are self-contained, say Y here. Otherwise, choose N.
133 string "Local version - append to kernel release"
135 Append an extra string to the end of your kernel version.
136 This will show up when you type uname, for example.
137 The string you set here will be appended after the contents of
138 any files with a filename matching localversion* in your
139 object and source tree, in that order. Your total string can
140 be a maximum of 64 characters.
142 config LOCALVERSION_AUTO
143 bool "Automatically append version information to the version string"
145 depends on !COMPILE_TEST
147 This will try to automatically determine if the current tree is a
148 release tree by looking for git tags that belong to the current
149 top of tree revision.
151 A string of the format -gxxxxxxxx will be added to the localversion
152 if a git-based tree is found. The string generated by this will be
153 appended after any matching localversion* files, and after the value
154 set in CONFIG_LOCALVERSION.
156 (The actual string used here is the first eight characters produced
157 by running the command:
159 $ git rev-parse --verify HEAD
161 which is done within the script "scripts/setlocalversion".)
164 string "Build ID Salt"
167 The build ID is used to link binaries and their debug info. Setting
168 this option will use the value in the calculation of the build id.
169 This is mostly useful for distributions which want to ensure the
170 build is unique between builds. It's safe to leave the default.
172 config HAVE_KERNEL_GZIP
175 config HAVE_KERNEL_BZIP2
178 config HAVE_KERNEL_LZMA
181 config HAVE_KERNEL_XZ
184 config HAVE_KERNEL_LZO
187 config HAVE_KERNEL_LZ4
190 config HAVE_KERNEL_UNCOMPRESSED
194 prompt "Kernel compression mode"
196 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
198 The linux kernel is a kind of self-extracting executable.
199 Several compression algorithms are available, which differ
200 in efficiency, compression and decompression speed.
201 Compression speed is only relevant when building a kernel.
202 Decompression speed is relevant at each boot.
204 If you have any problems with bzip2 or lzma compressed
205 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
206 version of this functionality (bzip2 only), for 2.4, was
207 supplied by Christian Ludwig)
209 High compression options are mostly useful for users, who
210 are low on disk space (embedded systems), but for whom ram
213 If in doubt, select 'gzip'
217 depends on HAVE_KERNEL_GZIP
219 The old and tried gzip compression. It provides a good balance
220 between compression ratio and decompression speed.
224 depends on HAVE_KERNEL_BZIP2
226 Its compression ratio and speed is intermediate.
227 Decompression speed is slowest among the choices. The kernel
228 size is about 10% smaller with bzip2, in comparison to gzip.
229 Bzip2 uses a large amount of memory. For modern kernels you
230 will need at least 8MB RAM or more for booting.
234 depends on HAVE_KERNEL_LZMA
236 This compression algorithm's ratio is best. Decompression speed
237 is between gzip and bzip2. Compression is slowest.
238 The kernel size is about 33% smaller with LZMA in comparison to gzip.
242 depends on HAVE_KERNEL_XZ
244 XZ uses the LZMA2 algorithm and instruction set specific
245 BCJ filters which can improve compression ratio of executable
246 code. The size of the kernel is about 30% smaller with XZ in
247 comparison to gzip. On architectures for which there is a BCJ
248 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
249 will create a few percent smaller kernel than plain LZMA.
251 The speed is about the same as with LZMA: The decompression
252 speed of XZ is better than that of bzip2 but worse than gzip
253 and LZO. Compression is slow.
257 depends on HAVE_KERNEL_LZO
259 Its compression ratio is the poorest among the choices. The kernel
260 size is about 10% bigger than gzip; however its speed
261 (both compression and decompression) is the fastest.
265 depends on HAVE_KERNEL_LZ4
267 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
268 A preliminary version of LZ4 de/compression tool is available at
269 <https://code.google.com/p/lz4/>.
271 Its compression ratio is worse than LZO. The size of the kernel
272 is about 8% bigger than LZO. But the decompression speed is
275 config KERNEL_UNCOMPRESSED
277 depends on HAVE_KERNEL_UNCOMPRESSED
279 Produce uncompressed kernel image. This option is usually not what
280 you want. It is useful for debugging the kernel in slow simulation
281 environments, where decompressing and moving the kernel is awfully
282 slow. This option allows early boot code to skip the decompressor
283 and jump right at uncompressed kernel image.
287 config DEFAULT_HOSTNAME
288 string "Default hostname"
291 This option determines the default system hostname before userspace
292 calls sethostname(2). The kernel traditionally uses "(none)" here,
293 but you may wish to use a different default here to make a minimal
294 system more usable with less configuration.
297 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
298 # add proper SWAP support to them, in which case this can be remove.
304 bool "Support for paging of anonymous memory (swap)"
305 depends on MMU && BLOCK && !ARCH_NO_SWAP
308 This option allows you to choose whether you want to have support
309 for so called swap devices or swap files in your kernel that are
310 used to provide more virtual memory than the actual RAM present
311 in your computer. If unsure say Y.
316 Inter Process Communication is a suite of library functions and
317 system calls which let processes (running programs) synchronize and
318 exchange information. It is generally considered to be a good thing,
319 and some programs won't run unless you say Y here. In particular, if
320 you want to run the DOS emulator dosemu under Linux (read the
321 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
322 you'll need to say Y here.
324 You can find documentation about IPC with "info ipc" and also in
325 section 6.4 of the Linux Programmer's Guide, available from
326 <http://www.tldp.org/guides.html>.
328 config SYSVIPC_SYSCTL
335 bool "POSIX Message Queues"
338 POSIX variant of message queues is a part of IPC. In POSIX message
339 queues every message has a priority which decides about succession
340 of receiving it by a process. If you want to compile and run
341 programs written e.g. for Solaris with use of its POSIX message
342 queues (functions mq_*) say Y here.
344 POSIX message queues are visible as a filesystem called 'mqueue'
345 and can be mounted somewhere if you want to do filesystem
346 operations on message queues.
350 config POSIX_MQUEUE_SYSCTL
352 depends on POSIX_MQUEUE
356 config CROSS_MEMORY_ATTACH
357 bool "Enable process_vm_readv/writev syscalls"
361 Enabling this option adds the system calls process_vm_readv and
362 process_vm_writev which allow a process with the correct privileges
363 to directly read from or write to another process' address space.
364 See the man page for more details.
367 bool "uselib syscall"
368 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
370 This option enables the uselib syscall, a system call used in the
371 dynamic linker from libc5 and earlier. glibc does not use this
372 system call. If you intend to run programs built on libc5 or
373 earlier, you may need to enable this syscall. Current systems
374 running glibc can safely disable this.
377 bool "Auditing support"
380 Enable auditing infrastructure that can be used with another
381 kernel subsystem, such as SELinux (which requires this for
382 logging of avc messages output). System call auditing is included
383 on architectures which support it.
385 config HAVE_ARCH_AUDITSYSCALL
390 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
393 source "kernel/irq/Kconfig"
394 source "kernel/time/Kconfig"
395 source "kernel/Kconfig.preempt"
397 menu "CPU/Task time and stats accounting"
399 config VIRT_CPU_ACCOUNTING
403 prompt "Cputime accounting"
404 default TICK_CPU_ACCOUNTING if !PPC64
405 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
407 # Kind of a stub config for the pure tick based cputime accounting
408 config TICK_CPU_ACCOUNTING
409 bool "Simple tick based cputime accounting"
410 depends on !S390 && !NO_HZ_FULL
412 This is the basic tick based cputime accounting that maintains
413 statistics about user, system and idle time spent on per jiffies
418 config VIRT_CPU_ACCOUNTING_NATIVE
419 bool "Deterministic task and CPU time accounting"
420 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
421 select VIRT_CPU_ACCOUNTING
423 Select this option to enable more accurate task and CPU time
424 accounting. This is done by reading a CPU counter on each
425 kernel entry and exit and on transitions within the kernel
426 between system, softirq and hardirq state, so there is a
427 small performance impact. In the case of s390 or IBM POWER > 5,
428 this also enables accounting of stolen time on logically-partitioned
431 config VIRT_CPU_ACCOUNTING_GEN
432 bool "Full dynticks CPU time accounting"
433 depends on HAVE_CONTEXT_TRACKING
434 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
435 depends on GENERIC_CLOCKEVENTS
436 select VIRT_CPU_ACCOUNTING
437 select CONTEXT_TRACKING
439 Select this option to enable task and CPU time accounting on full
440 dynticks systems. This accounting is implemented by watching every
441 kernel-user boundaries using the context tracking subsystem.
442 The accounting is thus performed at the expense of some significant
445 For now this is only useful if you are working on the full
446 dynticks subsystem development.
452 config IRQ_TIME_ACCOUNTING
453 bool "Fine granularity task level IRQ time accounting"
454 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
456 Select this option to enable fine granularity task irq time
457 accounting. This is done by reading a timestamp on each
458 transitions between softirq and hardirq state, so there can be a
459 small performance impact.
461 If in doubt, say N here.
463 config HAVE_SCHED_AVG_IRQ
465 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
468 config BSD_PROCESS_ACCT
469 bool "BSD Process Accounting"
472 If you say Y here, a user level program will be able to instruct the
473 kernel (via a special system call) to write process accounting
474 information to a file: whenever a process exits, information about
475 that process will be appended to the file by the kernel. The
476 information includes things such as creation time, owning user,
477 command name, memory usage, controlling terminal etc. (the complete
478 list is in the struct acct in <file:include/linux/acct.h>). It is
479 up to the user level program to do useful things with this
480 information. This is generally a good idea, so say Y.
482 config BSD_PROCESS_ACCT_V3
483 bool "BSD Process Accounting version 3 file format"
484 depends on BSD_PROCESS_ACCT
487 If you say Y here, the process accounting information is written
488 in a new file format that also logs the process IDs of each
489 process and its parent. Note that this file format is incompatible
490 with previous v0/v1/v2 file formats, so you will need updated tools
491 for processing it. A preliminary version of these tools is available
492 at <http://www.gnu.org/software/acct/>.
495 bool "Export task/process statistics through netlink"
500 Export selected statistics for tasks/processes through the
501 generic netlink interface. Unlike BSD process accounting, the
502 statistics are available during the lifetime of tasks/processes as
503 responses to commands. Like BSD accounting, they are sent to user
508 config TASK_DELAY_ACCT
509 bool "Enable per-task delay accounting"
513 Collect information on time spent by a task waiting for system
514 resources like cpu, synchronous block I/O completion and swapping
515 in pages. Such statistics can help in setting a task's priorities
516 relative to other tasks for cpu, io, rss limits etc.
521 bool "Enable extended accounting over taskstats"
524 Collect extended task accounting data and send the data
525 to userland for processing over the taskstats interface.
529 config TASK_IO_ACCOUNTING
530 bool "Enable per-task storage I/O accounting"
531 depends on TASK_XACCT
533 Collect information on the number of bytes of storage I/O which this
539 bool "Pressure stall information tracking"
541 Collect metrics that indicate how overcommitted the CPU, memory,
542 and IO capacity are in the system.
544 If you say Y here, the kernel will create /proc/pressure/ with the
545 pressure statistics files cpu, memory, and io. These will indicate
546 the share of walltime in which some or all tasks in the system are
547 delayed due to contention of the respective resource.
549 In kernels with cgroup support, cgroups (cgroup2 only) will
550 have cpu.pressure, memory.pressure, and io.pressure files,
551 which aggregate pressure stalls for the grouped tasks only.
553 For more details see Documentation/accounting/psi.rst.
557 config PSI_DEFAULT_DISABLED
558 bool "Require boot parameter to enable pressure stall information tracking"
562 If set, pressure stall information tracking will be disabled
563 per default but can be enabled through passing psi=1 on the
564 kernel commandline during boot.
566 This feature adds some code to the task wakeup and sleep
567 paths of the scheduler. The overhead is too low to affect
568 common scheduling-intense workloads in practice (such as
569 webservers, memcache), but it does show up in artificial
570 scheduler stress tests, such as hackbench.
572 If you are paranoid and not sure what the kernel will be
577 endmenu # "CPU/Task time and stats accounting"
581 depends on SMP || COMPILE_TEST
584 Make sure that CPUs running critical tasks are not disturbed by
585 any source of "noise" such as unbound workqueues, timers, kthreads...
586 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
587 the "isolcpus=" boot parameter.
591 source "kernel/rcu/Kconfig"
598 tristate "Kernel .config support"
600 This option enables the complete Linux kernel ".config" file
601 contents to be saved in the kernel. It provides documentation
602 of which kernel options are used in a running kernel or in an
603 on-disk kernel. This information can be extracted from the kernel
604 image file with the script scripts/extract-ikconfig and used as
605 input to rebuild the current kernel or to build another kernel.
606 It can also be extracted from a running kernel by reading
607 /proc/config.gz if enabled (below).
610 bool "Enable access to .config through /proc/config.gz"
611 depends on IKCONFIG && PROC_FS
613 This option enables access to the kernel configuration file
614 through /proc/config.gz.
617 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
620 This option enables access to the in-kernel headers that are generated during
621 the build process. These can be used to build eBPF tracing programs,
622 or similar programs. If you build the headers as a module, a module called
623 kheaders.ko is built which can be loaded on-demand to get access to headers.
626 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
631 Select the minimal kernel log buffer size as a power of 2.
632 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
633 parameter, see below. Any higher size also might be forced
634 by "log_buf_len" boot parameter.
644 config LOG_CPU_MAX_BUF_SHIFT
645 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
648 default 12 if !BASE_SMALL
649 default 0 if BASE_SMALL
652 This option allows to increase the default ring buffer size
653 according to the number of CPUs. The value defines the contribution
654 of each CPU as a power of 2. The used space is typically only few
655 lines however it might be much more when problems are reported,
658 The increased size means that a new buffer has to be allocated and
659 the original static one is unused. It makes sense only on systems
660 with more CPUs. Therefore this value is used only when the sum of
661 contributions is greater than the half of the default kernel ring
662 buffer as defined by LOG_BUF_SHIFT. The default values are set
663 so that more than 64 CPUs are needed to trigger the allocation.
665 Also this option is ignored when "log_buf_len" kernel parameter is
666 used as it forces an exact (power of two) size of the ring buffer.
668 The number of possible CPUs is used for this computation ignoring
669 hotplugging making the computation optimal for the worst case
670 scenario while allowing a simple algorithm to be used from bootup.
672 Examples shift values and their meaning:
673 17 => 128 KB for each CPU
674 16 => 64 KB for each CPU
675 15 => 32 KB for each CPU
676 14 => 16 KB for each CPU
677 13 => 8 KB for each CPU
678 12 => 4 KB for each CPU
680 config PRINTK_SAFE_LOG_BUF_SHIFT
681 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
686 Select the size of an alternate printk per-CPU buffer where messages
687 printed from usafe contexts are temporary stored. One example would
688 be NMI messages, another one - printk recursion. The messages are
689 copied to the main log buffer in a safe context to avoid a deadlock.
690 The value defines the size as a power of 2.
692 Those messages are rare and limited. The largest one is when
693 a backtrace is printed. It usually fits into 4KB. Select
694 8KB if you want to be on the safe side.
697 17 => 128 KB for each CPU
698 16 => 64 KB for each CPU
699 15 => 32 KB for each CPU
700 14 => 16 KB for each CPU
701 13 => 8 KB for each CPU
702 12 => 4 KB for each CPU
705 # Architectures with an unreliable sched_clock() should select this:
707 config HAVE_UNSTABLE_SCHED_CLOCK
710 config GENERIC_SCHED_CLOCK
713 menu "Scheduler features"
716 bool "Enable utilization clamping for RT/FAIR tasks"
717 depends on CPU_FREQ_GOV_SCHEDUTIL
719 This feature enables the scheduler to track the clamped utilization
720 of each CPU based on RUNNABLE tasks scheduled on that CPU.
722 With this option, the user can specify the min and max CPU
723 utilization allowed for RUNNABLE tasks. The max utilization defines
724 the maximum frequency a task should use while the min utilization
725 defines the minimum frequency it should use.
727 Both min and max utilization clamp values are hints to the scheduler,
728 aiming at improving its frequency selection policy, but they do not
729 enforce or grant any specific bandwidth for tasks.
733 config UCLAMP_BUCKETS_COUNT
734 int "Number of supported utilization clamp buckets"
737 depends on UCLAMP_TASK
739 Defines the number of clamp buckets to use. The range of each bucket
740 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
741 number of clamp buckets the finer their granularity and the higher
742 the precision of clamping aggregation and tracking at run-time.
744 For example, with the minimum configuration value we will have 5
745 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
746 be refcounted in the [20..39]% bucket and will set the bucket clamp
747 effective value to 25%.
748 If a second 30% boosted task should be co-scheduled on the same CPU,
749 that task will be refcounted in the same bucket of the first task and
750 it will boost the bucket clamp effective value to 30%.
751 The clamp effective value of a bucket is reset to its nominal value
752 (20% in the example above) when there are no more tasks refcounted in
755 An additional boost/capping margin can be added to some tasks. In the
756 example above the 25% task will be boosted to 30% until it exits the
757 CPU. If that should be considered not acceptable on certain systems,
758 it's always possible to reduce the margin by increasing the number of
759 clamp buckets to trade off used memory for run-time tracking
762 If in doubt, use the default value.
767 # For architectures that want to enable the support for NUMA-affine scheduler
770 config ARCH_SUPPORTS_NUMA_BALANCING
774 # For architectures that prefer to flush all TLBs after a number of pages
775 # are unmapped instead of sending one IPI per page to flush. The architecture
776 # must provide guarantees on what happens if a clean TLB cache entry is
777 # written after the unmap. Details are in mm/rmap.c near the check for
778 # should_defer_flush. The architecture should also consider if the full flush
779 # and the refill costs are offset by the savings of sending fewer IPIs.
780 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
784 # For architectures that know their GCC __int128 support is sound
786 config ARCH_SUPPORTS_INT128
789 # For architectures that (ab)use NUMA to represent different memory regions
790 # all cpu-local but of different latencies, such as SuperH.
792 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
795 config NUMA_BALANCING
796 bool "Memory placement aware NUMA scheduler"
797 depends on ARCH_SUPPORTS_NUMA_BALANCING
798 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
799 depends on SMP && NUMA && MIGRATION
801 This option adds support for automatic NUMA aware memory/task placement.
802 The mechanism is quite primitive and is based on migrating memory when
803 it has references to the node the task is running on.
805 This system will be inactive on UMA systems.
807 config NUMA_BALANCING_DEFAULT_ENABLED
808 bool "Automatically enable NUMA aware memory/task placement"
810 depends on NUMA_BALANCING
812 If set, automatic NUMA balancing will be enabled if running on a NUMA
816 bool "Control Group support"
819 This option adds support for grouping sets of processes together, for
820 use with process control subsystems such as Cpusets, CFS, memory
821 controls or device isolation.
823 - Documentation/scheduler/sched-design-CFS.rst (CFS)
824 - Documentation/admin-guide/cgroup-v1/ (features for grouping, isolation
825 and resource control)
835 bool "Memory controller"
839 Provides control over the memory footprint of tasks in a cgroup.
842 bool "Swap controller"
843 depends on MEMCG && SWAP
845 Provides control over the swap space consumed by tasks in a cgroup.
847 config MEMCG_SWAP_ENABLED
848 bool "Swap controller enabled by default"
849 depends on MEMCG_SWAP
852 Memory Resource Controller Swap Extension comes with its price in
853 a bigger memory consumption. General purpose distribution kernels
854 which want to enable the feature but keep it disabled by default
855 and let the user enable it by swapaccount=1 boot command line
856 parameter should have this option unselected.
857 For those who want to have the feature enabled by default should
858 select this option (if, for some reason, they need to disable it
859 then swapaccount=0 does the trick).
863 depends on MEMCG && !SLOB
871 Generic block IO controller cgroup interface. This is the common
872 cgroup interface which should be used by various IO controlling
875 Currently, CFQ IO scheduler uses it to recognize task groups and
876 control disk bandwidth allocation (proportional time slice allocation)
877 to such task groups. It is also used by bio throttling logic in
878 block layer to implement upper limit in IO rates on a device.
880 This option only enables generic Block IO controller infrastructure.
881 One needs to also enable actual IO controlling logic/policy. For
882 enabling proportional weight division of disk bandwidth in CFQ, set
883 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
884 CONFIG_BLK_DEV_THROTTLING=y.
886 See Documentation/admin-guide/cgroup-v1/blkio-controller.rst for more information.
888 config CGROUP_WRITEBACK
890 depends on MEMCG && BLK_CGROUP
893 menuconfig CGROUP_SCHED
894 bool "CPU controller"
897 This feature lets CPU scheduler recognize task groups and control CPU
898 bandwidth allocation to such task groups. It uses cgroups to group
902 config FAIR_GROUP_SCHED
903 bool "Group scheduling for SCHED_OTHER"
904 depends on CGROUP_SCHED
908 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
909 depends on FAIR_GROUP_SCHED
912 This option allows users to define CPU bandwidth rates (limits) for
913 tasks running within the fair group scheduler. Groups with no limit
914 set are considered to be unconstrained and will run with no
916 See Documentation/scheduler/sched-bwc.rst for more information.
918 config RT_GROUP_SCHED
919 bool "Group scheduling for SCHED_RR/FIFO"
920 depends on CGROUP_SCHED
923 This feature lets you explicitly allocate real CPU bandwidth
924 to task groups. If enabled, it will also make it impossible to
925 schedule realtime tasks for non-root users until you allocate
926 realtime bandwidth for them.
927 See Documentation/scheduler/sched-rt-group.rst for more information.
932 bool "PIDs controller"
934 Provides enforcement of process number limits in the scope of a
935 cgroup. Any attempt to fork more processes than is allowed in the
936 cgroup will fail. PIDs are fundamentally a global resource because it
937 is fairly trivial to reach PID exhaustion before you reach even a
938 conservative kmemcg limit. As a result, it is possible to grind a
939 system to halt without being limited by other cgroup policies. The
940 PIDs controller is designed to stop this from happening.
942 It should be noted that organisational operations (such as attaching
943 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
944 since the PIDs limit only affects a process's ability to fork, not to
948 bool "RDMA controller"
950 Provides enforcement of RDMA resources defined by IB stack.
951 It is fairly easy for consumers to exhaust RDMA resources, which
952 can result into resource unavailability to other consumers.
953 RDMA controller is designed to stop this from happening.
954 Attaching processes with active RDMA resources to the cgroup
955 hierarchy is allowed even if can cross the hierarchy's limit.
957 config CGROUP_FREEZER
958 bool "Freezer controller"
960 Provides a way to freeze and unfreeze all tasks in a
963 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
964 controller includes important in-kernel memory consumers per default.
966 If you're using cgroup2, say N.
968 config CGROUP_HUGETLB
969 bool "HugeTLB controller"
970 depends on HUGETLB_PAGE
974 Provides a cgroup controller for HugeTLB pages.
975 When you enable this, you can put a per cgroup limit on HugeTLB usage.
976 The limit is enforced during page fault. Since HugeTLB doesn't
977 support page reclaim, enforcing the limit at page fault time implies
978 that, the application will get SIGBUS signal if it tries to access
979 HugeTLB pages beyond its limit. This requires the application to know
980 beforehand how much HugeTLB pages it would require for its use. The
981 control group is tracked in the third page lru pointer. This means
982 that we cannot use the controller with huge page less than 3 pages.
985 bool "Cpuset controller"
988 This option will let you create and manage CPUSETs which
989 allow dynamically partitioning a system into sets of CPUs and
990 Memory Nodes and assigning tasks to run only within those sets.
991 This is primarily useful on large SMP or NUMA systems.
995 config PROC_PID_CPUSET
996 bool "Include legacy /proc/<pid>/cpuset file"
1000 config CGROUP_DEVICE
1001 bool "Device controller"
1003 Provides a cgroup controller implementing whitelists for
1004 devices which a process in the cgroup can mknod or open.
1006 config CGROUP_CPUACCT
1007 bool "Simple CPU accounting controller"
1009 Provides a simple controller for monitoring the
1010 total CPU consumed by the tasks in a cgroup.
1013 bool "Perf controller"
1014 depends on PERF_EVENTS
1016 This option extends the perf per-cpu mode to restrict monitoring
1017 to threads which belong to the cgroup specified and run on the
1023 bool "Support for eBPF programs attached to cgroups"
1024 depends on BPF_SYSCALL
1025 select SOCK_CGROUP_DATA
1027 Allow attaching eBPF programs to a cgroup using the bpf(2)
1028 syscall command BPF_PROG_ATTACH.
1030 In which context these programs are accessed depends on the type
1031 of attachment. For instance, programs that are attached using
1032 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1036 bool "Debug controller"
1038 depends on DEBUG_KERNEL
1040 This option enables a simple controller that exports
1041 debugging information about the cgroups framework. This
1042 controller is for control cgroup debugging only. Its
1043 interfaces are not stable.
1047 config SOCK_CGROUP_DATA
1053 menuconfig NAMESPACES
1054 bool "Namespaces support" if EXPERT
1055 depends on MULTIUSER
1058 Provides the way to make tasks work with different objects using
1059 the same id. For example same IPC id may refer to different objects
1060 or same user id or pid may refer to different tasks when used in
1061 different namespaces.
1066 bool "UTS namespace"
1069 In this namespace tasks see different info provided with the
1073 bool "IPC namespace"
1074 depends on (SYSVIPC || POSIX_MQUEUE)
1077 In this namespace tasks work with IPC ids which correspond to
1078 different IPC objects in different namespaces.
1081 bool "User namespace"
1084 This allows containers, i.e. vservers, to use user namespaces
1085 to provide different user info for different servers.
1087 When user namespaces are enabled in the kernel it is
1088 recommended that the MEMCG option also be enabled and that
1089 user-space use the memory control groups to limit the amount
1090 of memory a memory unprivileged users can use.
1095 bool "PID Namespaces"
1098 Support process id namespaces. This allows having multiple
1099 processes with the same pid as long as they are in different
1100 pid namespaces. This is a building block of containers.
1103 bool "Network namespace"
1107 Allow user space to create what appear to be multiple instances
1108 of the network stack.
1112 config CHECKPOINT_RESTORE
1113 bool "Checkpoint/restore support"
1114 select PROC_CHILDREN
1117 Enables additional kernel features in a sake of checkpoint/restore.
1118 In particular it adds auxiliary prctl codes to setup process text,
1119 data and heap segment sizes, and a few additional /proc filesystem
1122 If unsure, say N here.
1124 config SCHED_AUTOGROUP
1125 bool "Automatic process group scheduling"
1128 select FAIR_GROUP_SCHED
1130 This option optimizes the scheduler for common desktop workloads by
1131 automatically creating and populating task groups. This separation
1132 of workloads isolates aggressive CPU burners (like build jobs) from
1133 desktop applications. Task group autogeneration is currently based
1136 config SYSFS_DEPRECATED
1137 bool "Enable deprecated sysfs features to support old userspace tools"
1141 This option adds code that switches the layout of the "block" class
1142 devices, to not show up in /sys/class/block/, but only in
1145 This switch is only active when the sysfs.deprecated=1 boot option is
1146 passed or the SYSFS_DEPRECATED_V2 option is set.
1148 This option allows new kernels to run on old distributions and tools,
1149 which might get confused by /sys/class/block/. Since 2007/2008 all
1150 major distributions and tools handle this just fine.
1152 Recent distributions and userspace tools after 2009/2010 depend on
1153 the existence of /sys/class/block/, and will not work with this
1156 Only if you are using a new kernel on an old distribution, you might
1159 config SYSFS_DEPRECATED_V2
1160 bool "Enable deprecated sysfs features by default"
1163 depends on SYSFS_DEPRECATED
1165 Enable deprecated sysfs by default.
1167 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1170 Only if you are using a new kernel on an old distribution, you might
1171 need to say Y here. Even then, odds are you would not need it
1172 enabled, you can always pass the boot option if absolutely necessary.
1175 bool "Kernel->user space relay support (formerly relayfs)"
1178 This option enables support for relay interface support in
1179 certain file systems (such as debugfs).
1180 It is designed to provide an efficient mechanism for tools and
1181 facilities to relay large amounts of data from kernel space to
1186 config BLK_DEV_INITRD
1187 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1189 The initial RAM filesystem is a ramfs which is loaded by the
1190 boot loader (loadlin or lilo) and that is mounted as root
1191 before the normal boot procedure. It is typically used to
1192 load modules needed to mount the "real" root file system,
1193 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1195 If RAM disk support (BLK_DEV_RAM) is also included, this
1196 also enables initial RAM disk (initrd) support and adds
1197 15 Kbytes (more on some other architectures) to the kernel size.
1203 source "usr/Kconfig"
1208 prompt "Compiler optimization level"
1209 default CC_OPTIMIZE_FOR_PERFORMANCE
1211 config CC_OPTIMIZE_FOR_PERFORMANCE
1212 bool "Optimize for performance"
1214 This is the default optimization level for the kernel, building
1215 with the "-O2" compiler flag for best performance and most
1216 helpful compile-time warnings.
1218 config CC_OPTIMIZE_FOR_SIZE
1219 bool "Optimize for size"
1220 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1222 Enabling this option will pass "-Os" instead of "-O2" to
1223 your compiler resulting in a smaller kernel.
1229 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1232 This requires that the arch annotates or otherwise protects
1233 its external entry points from being discarded. Linker scripts
1234 must also merge .text.*, .data.*, and .bss.* correctly into
1235 output sections. Care must be taken not to pull in unrelated
1236 sections (e.g., '.text.init'). Typically '.' in section names
1237 is used to distinguish them from label names / C identifiers.
1239 config LD_DEAD_CODE_DATA_ELIMINATION
1240 bool "Dead code and data elimination (EXPERIMENTAL)"
1241 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1243 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1244 depends on $(cc-option,-ffunction-sections -fdata-sections)
1245 depends on $(ld-option,--gc-sections)
1247 Enable this if you want to do dead code and data elimination with
1248 the linker by compiling with -ffunction-sections -fdata-sections,
1249 and linking with --gc-sections.
1251 This can reduce on disk and in-memory size of the kernel
1252 code and static data, particularly for small configs and
1253 on small systems. This has the possibility of introducing
1254 silently broken kernel if the required annotations are not
1255 present. This option is not well tested yet, so use at your
1264 config SYSCTL_EXCEPTION_TRACE
1267 Enable support for /proc/sys/debug/exception-trace.
1269 config SYSCTL_ARCH_UNALIGN_NO_WARN
1272 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1273 Allows arch to define/use @no_unaligned_warning to possibly warn
1274 about unaligned access emulation going on under the hood.
1276 config SYSCTL_ARCH_UNALIGN_ALLOW
1279 Enable support for /proc/sys/kernel/unaligned-trap
1280 Allows arches to define/use @unaligned_enabled to runtime toggle
1281 the unaligned access emulation.
1282 see arch/parisc/kernel/unaligned.c for reference
1284 config HAVE_PCSPKR_PLATFORM
1287 # interpreter that classic socket filters depend on
1292 bool "Configure standard kernel features (expert users)"
1293 # Unhide debug options, to make the on-by-default options visible
1296 This option allows certain base kernel options and settings
1297 to be disabled or tweaked. This is for specialized
1298 environments which can tolerate a "non-standard" kernel.
1299 Only use this if you really know what you are doing.
1302 bool "Enable 16-bit UID system calls" if EXPERT
1303 depends on HAVE_UID16 && MULTIUSER
1306 This enables the legacy 16-bit UID syscall wrappers.
1309 bool "Multiple users, groups and capabilities support" if EXPERT
1312 This option enables support for non-root users, groups and
1315 If you say N here, all processes will run with UID 0, GID 0, and all
1316 possible capabilities. Saying N here also compiles out support for
1317 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1320 If unsure, say Y here.
1322 config SGETMASK_SYSCALL
1323 bool "sgetmask/ssetmask syscalls support" if EXPERT
1324 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1326 sys_sgetmask and sys_ssetmask are obsolete system calls
1327 no longer supported in libc but still enabled by default in some
1330 If unsure, leave the default option here.
1332 config SYSFS_SYSCALL
1333 bool "Sysfs syscall support" if EXPERT
1336 sys_sysfs is an obsolete system call no longer supported in libc.
1337 Note that disabling this option is more secure but might break
1338 compatibility with some systems.
1340 If unsure say Y here.
1342 config SYSCTL_SYSCALL
1343 bool "Sysctl syscall support" if EXPERT
1344 depends on PROC_SYSCTL
1348 sys_sysctl uses binary paths that have been found challenging
1349 to properly maintain and use. The interface in /proc/sys
1350 using paths with ascii names is now the primary path to this
1353 Almost nothing using the binary sysctl interface so if you are
1354 trying to save some space it is probably safe to disable this,
1355 making your kernel marginally smaller.
1357 If unsure say N here.
1360 bool "open by fhandle syscalls" if EXPERT
1364 If you say Y here, a user level program will be able to map
1365 file names to handle and then later use the handle for
1366 different file system operations. This is useful in implementing
1367 userspace file servers, which now track files using handles instead
1368 of names. The handle would remain the same even if file names
1369 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1373 bool "Posix Clocks & timers" if EXPERT
1376 This includes native support for POSIX timers to the kernel.
1377 Some embedded systems have no use for them and therefore they
1378 can be configured out to reduce the size of the kernel image.
1380 When this option is disabled, the following syscalls won't be
1381 available: timer_create, timer_gettime: timer_getoverrun,
1382 timer_settime, timer_delete, clock_adjtime, getitimer,
1383 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1384 clock_getres and clock_nanosleep syscalls will be limited to
1385 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1391 bool "Enable support for printk" if EXPERT
1394 This option enables normal printk support. Removing it
1395 eliminates most of the message strings from the kernel image
1396 and makes the kernel more or less silent. As this makes it
1397 very difficult to diagnose system problems, saying N here is
1398 strongly discouraged.
1406 bool "BUG() support" if EXPERT
1409 Disabling this option eliminates support for BUG and WARN, reducing
1410 the size of your kernel image and potentially quietly ignoring
1411 numerous fatal conditions. You should only consider disabling this
1412 option for embedded systems with no facilities for reporting errors.
1418 bool "Enable ELF core dumps" if EXPERT
1420 Enable support for generating core dumps. Disabling saves about 4k.
1423 config PCSPKR_PLATFORM
1424 bool "Enable PC-Speaker support" if EXPERT
1425 depends on HAVE_PCSPKR_PLATFORM
1429 This option allows to disable the internal PC-Speaker
1430 support, saving some memory.
1434 bool "Enable full-sized data structures for core" if EXPERT
1436 Disabling this option reduces the size of miscellaneous core
1437 kernel data structures. This saves memory on small machines,
1438 but may reduce performance.
1441 bool "Enable futex support" if EXPERT
1445 Disabling this option will cause the kernel to be built without
1446 support for "fast userspace mutexes". The resulting kernel may not
1447 run glibc-based applications correctly.
1451 depends on FUTEX && RT_MUTEXES
1454 config HAVE_FUTEX_CMPXCHG
1458 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1459 is implemented and always working. This removes a couple of runtime
1463 bool "Enable eventpoll support" if EXPERT
1466 Disabling this option will cause the kernel to be built without
1467 support for epoll family of system calls.
1470 bool "Enable signalfd() system call" if EXPERT
1473 Enable the signalfd() system call that allows to receive signals
1474 on a file descriptor.
1479 bool "Enable timerfd() system call" if EXPERT
1482 Enable the timerfd() system call that allows to receive timer
1483 events on a file descriptor.
1488 bool "Enable eventfd() system call" if EXPERT
1491 Enable the eventfd() system call that allows to receive both
1492 kernel notification (ie. KAIO) or userspace notifications.
1497 bool "Use full shmem filesystem" if EXPERT
1501 The shmem is an internal filesystem used to manage shared memory.
1502 It is backed by swap and manages resource limits. It is also exported
1503 to userspace as tmpfs if TMPFS is enabled. Disabling this
1504 option replaces shmem and tmpfs with the much simpler ramfs code,
1505 which may be appropriate on small systems without swap.
1508 bool "Enable AIO support" if EXPERT
1511 This option enables POSIX asynchronous I/O which may by used
1512 by some high performance threaded applications. Disabling
1513 this option saves about 7k.
1516 bool "Enable IO uring support" if EXPERT
1520 This option enables support for the io_uring interface, enabling
1521 applications to submit and complete IO through submission and
1522 completion rings that are shared between the kernel and application.
1524 config ADVISE_SYSCALLS
1525 bool "Enable madvise/fadvise syscalls" if EXPERT
1528 This option enables the madvise and fadvise syscalls, used by
1529 applications to advise the kernel about their future memory or file
1530 usage, improving performance. If building an embedded system where no
1531 applications use these syscalls, you can disable this option to save
1535 bool "Enable membarrier() system call" if EXPERT
1538 Enable the membarrier() system call that allows issuing memory
1539 barriers across all running threads, which can be used to distribute
1540 the cost of user-space memory barriers asymmetrically by transforming
1541 pairs of memory barriers into pairs consisting of membarrier() and a
1547 bool "Load all symbols for debugging/ksymoops" if EXPERT
1550 Say Y here to let the kernel print out symbolic crash information and
1551 symbolic stack backtraces. This increases the size of the kernel
1552 somewhat, as all symbols have to be loaded into the kernel image.
1555 bool "Include all symbols in kallsyms"
1556 depends on DEBUG_KERNEL && KALLSYMS
1558 Normally kallsyms only contains the symbols of functions for nicer
1559 OOPS messages and backtraces (i.e., symbols from the text and inittext
1560 sections). This is sufficient for most cases. And only in very rare
1561 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1562 names of variables from the data sections, etc).
1564 This option makes sure that all symbols are loaded into the kernel
1565 image (i.e., symbols from all sections) in cost of increased kernel
1566 size (depending on the kernel configuration, it may be 300KiB or
1567 something like this).
1569 Say N unless you really need all symbols.
1571 config KALLSYMS_ABSOLUTE_PERCPU
1574 default X86_64 && SMP
1576 config KALLSYMS_BASE_RELATIVE
1581 Instead of emitting them as absolute values in the native word size,
1582 emit the symbol references in the kallsyms table as 32-bit entries,
1583 each containing a relative value in the range [base, base + U32_MAX]
1584 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1585 an absolute value in the range [0, S32_MAX] or a relative value in the
1586 range [base, base + S32_MAX], where base is the lowest relative symbol
1587 address encountered in the image.
1589 On 64-bit builds, this reduces the size of the address table by 50%,
1590 but more importantly, it results in entries whose values are build
1591 time constants, and no relocation pass is required at runtime to fix
1592 up the entries based on the runtime load address of the kernel.
1594 # end of the "standard kernel features (expert users)" menu
1596 # syscall, maps, verifier
1598 bool "Enable bpf() system call"
1603 Enable the bpf() system call that allows to manipulate eBPF
1604 programs and maps via file descriptors.
1606 config BPF_JIT_ALWAYS_ON
1607 bool "Permanently enable BPF JIT and remove BPF interpreter"
1608 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1610 Enables BPF JIT and removes BPF interpreter to avoid
1611 speculative execution of BPF instructions by the interpreter
1614 bool "Enable userfaultfd() system call"
1617 Enable the userfaultfd() system call that allows to intercept and
1618 handle page faults in userland.
1620 config ARCH_HAS_MEMBARRIER_CALLBACKS
1623 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1627 bool "Enable rseq() system call" if EXPERT
1629 depends on HAVE_RSEQ
1632 Enable the restartable sequences system call. It provides a
1633 user-space cache for the current CPU number value, which
1634 speeds up getting the current CPU number from user-space,
1635 as well as an ABI to speed up user-space operations on
1642 bool "Enabled debugging of rseq() system call" if EXPERT
1643 depends on RSEQ && DEBUG_KERNEL
1645 Enable extra debugging checks for the rseq system call.
1650 bool "Embedded system"
1651 option allnoconfig_y
1654 This option should be enabled if compiling the kernel for
1655 an embedded system so certain expert options are available
1658 config HAVE_PERF_EVENTS
1661 See tools/perf/design.txt for details.
1663 config PERF_USE_VMALLOC
1666 See tools/perf/design.txt for details
1669 bool "PC/104 support" if EXPERT
1671 Expose PC/104 form factor device drivers and options available for
1672 selection and configuration. Enable this option if your target
1673 machine has a PC/104 bus.
1675 menu "Kernel Performance Events And Counters"
1678 bool "Kernel performance events and counters"
1679 default y if PROFILING
1680 depends on HAVE_PERF_EVENTS
1684 Enable kernel support for various performance events provided
1685 by software and hardware.
1687 Software events are supported either built-in or via the
1688 use of generic tracepoints.
1690 Most modern CPUs support performance events via performance
1691 counter registers. These registers count the number of certain
1692 types of hw events: such as instructions executed, cachemisses
1693 suffered, or branches mis-predicted - without slowing down the
1694 kernel or applications. These registers can also trigger interrupts
1695 when a threshold number of events have passed - and can thus be
1696 used to profile the code that runs on that CPU.
1698 The Linux Performance Event subsystem provides an abstraction of
1699 these software and hardware event capabilities, available via a
1700 system call and used by the "perf" utility in tools/perf/. It
1701 provides per task and per CPU counters, and it provides event
1702 capabilities on top of those.
1706 config DEBUG_PERF_USE_VMALLOC
1708 bool "Debug: use vmalloc to back perf mmap() buffers"
1709 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1710 select PERF_USE_VMALLOC
1712 Use vmalloc memory to back perf mmap() buffers.
1714 Mostly useful for debugging the vmalloc code on platforms
1715 that don't require it.
1721 config VM_EVENT_COUNTERS
1723 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1725 VM event counters are needed for event counts to be shown.
1726 This option allows the disabling of the VM event counters
1727 on EXPERT systems. /proc/vmstat will only show page counts
1728 if VM event counters are disabled.
1732 bool "Enable SLUB debugging support" if EXPERT
1733 depends on SLUB && SYSFS
1735 SLUB has extensive debug support features. Disabling these can
1736 result in significant savings in code size. This also disables
1737 SLUB sysfs support. /sys/slab will not exist and there will be
1738 no support for cache validation etc.
1740 config SLUB_MEMCG_SYSFS_ON
1742 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1743 depends on SLUB && SYSFS && MEMCG
1745 SLUB creates a directory under /sys/kernel/slab for each
1746 allocation cache to host info and debug files. If memory
1747 cgroup is enabled, each cache can have per memory cgroup
1748 caches. SLUB can create the same sysfs directories for these
1749 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1750 to a very high number of debug files being created. This is
1751 controlled by slub_memcg_sysfs boot parameter and this
1752 config option determines the parameter's default value.
1755 bool "Disable heap randomization"
1758 Randomizing heap placement makes heap exploits harder, but it
1759 also breaks ancient binaries (including anything libc5 based).
1760 This option changes the bootup default to heap randomization
1761 disabled, and can be overridden at runtime by setting
1762 /proc/sys/kernel/randomize_va_space to 2.
1764 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1767 prompt "Choose SLAB allocator"
1770 This option allows to select a slab allocator.
1774 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1776 The regular slab allocator that is established and known to work
1777 well in all environments. It organizes cache hot objects in
1778 per cpu and per node queues.
1781 bool "SLUB (Unqueued Allocator)"
1782 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1784 SLUB is a slab allocator that minimizes cache line usage
1785 instead of managing queues of cached objects (SLAB approach).
1786 Per cpu caching is realized using slabs of objects instead
1787 of queues of objects. SLUB can use memory efficiently
1788 and has enhanced diagnostics. SLUB is the default choice for
1793 bool "SLOB (Simple Allocator)"
1795 SLOB replaces the stock allocator with a drastically simpler
1796 allocator. SLOB is generally more space efficient but
1797 does not perform as well on large systems.
1801 config SLAB_MERGE_DEFAULT
1802 bool "Allow slab caches to be merged"
1805 For reduced kernel memory fragmentation, slab caches can be
1806 merged when they share the same size and other characteristics.
1807 This carries a risk of kernel heap overflows being able to
1808 overwrite objects from merged caches (and more easily control
1809 cache layout), which makes such heap attacks easier to exploit
1810 by attackers. By keeping caches unmerged, these kinds of exploits
1811 can usually only damage objects in the same cache. To disable
1812 merging at runtime, "slab_nomerge" can be passed on the kernel
1815 config SLAB_FREELIST_RANDOM
1817 depends on SLAB || SLUB
1818 bool "SLAB freelist randomization"
1820 Randomizes the freelist order used on creating new pages. This
1821 security feature reduces the predictability of the kernel slab
1822 allocator against heap overflows.
1824 config SLAB_FREELIST_HARDENED
1825 bool "Harden slab freelist metadata"
1828 Many kernel heap attacks try to target slab cache metadata and
1829 other infrastructure. This options makes minor performance
1830 sacrifices to harden the kernel slab allocator against common
1831 freelist exploit methods.
1833 config SHUFFLE_PAGE_ALLOCATOR
1834 bool "Page allocator randomization"
1835 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1837 Randomization of the page allocator improves the average
1838 utilization of a direct-mapped memory-side-cache. See section
1839 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1840 6.2a specification for an example of how a platform advertises
1841 the presence of a memory-side-cache. There are also incidental
1842 security benefits as it reduces the predictability of page
1843 allocations to compliment SLAB_FREELIST_RANDOM, but the
1844 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1845 10th order of pages is selected based on cache utilization
1848 While the randomization improves cache utilization it may
1849 negatively impact workloads on platforms without a cache. For
1850 this reason, by default, the randomization is enabled only
1851 after runtime detection of a direct-mapped memory-side-cache.
1852 Otherwise, the randomization may be force enabled with the
1853 'page_alloc.shuffle' kernel command line parameter.
1857 config SLUB_CPU_PARTIAL
1859 depends on SLUB && SMP
1860 bool "SLUB per cpu partial cache"
1862 Per cpu partial caches accelerate objects allocation and freeing
1863 that is local to a processor at the price of more indeterminism
1864 in the latency of the free. On overflow these caches will be cleared
1865 which requires the taking of locks that may cause latency spikes.
1866 Typically one would choose no for a realtime system.
1868 config MMAP_ALLOW_UNINITIALIZED
1869 bool "Allow mmapped anonymous memory to be uninitialized"
1870 depends on EXPERT && !MMU
1873 Normally, and according to the Linux spec, anonymous memory obtained
1874 from mmap() has its contents cleared before it is passed to
1875 userspace. Enabling this config option allows you to request that
1876 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1877 providing a huge performance boost. If this option is not enabled,
1878 then the flag will be ignored.
1880 This is taken advantage of by uClibc's malloc(), and also by
1881 ELF-FDPIC binfmt's brk and stack allocator.
1883 Because of the obvious security issues, this option should only be
1884 enabled on embedded devices where you control what is run in
1885 userspace. Since that isn't generally a problem on no-MMU systems,
1886 it is normally safe to say Y here.
1888 See Documentation/nommu-mmap.txt for more information.
1890 config SYSTEM_DATA_VERIFICATION
1892 select SYSTEM_TRUSTED_KEYRING
1896 select ASYMMETRIC_KEY_TYPE
1897 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1900 select X509_CERTIFICATE_PARSER
1901 select PKCS7_MESSAGE_PARSER
1903 Provide PKCS#7 message verification using the contents of the system
1904 trusted keyring to provide public keys. This then can be used for
1905 module verification, kexec image verification and firmware blob
1909 bool "Profiling support"
1911 Say Y here to enable the extended profiling support mechanisms used
1912 by profilers such as OProfile.
1915 # Place an empty function call at each tracepoint site. Can be
1916 # dynamically changed for a probe function.
1921 endmenu # General setup
1923 source "arch/Kconfig"
1930 default 0 if BASE_FULL
1931 default 1 if !BASE_FULL
1934 bool "Enable loadable module support"
1937 Kernel modules are small pieces of compiled code which can
1938 be inserted in the running kernel, rather than being
1939 permanently built into the kernel. You use the "modprobe"
1940 tool to add (and sometimes remove) them. If you say Y here,
1941 many parts of the kernel can be built as modules (by
1942 answering M instead of Y where indicated): this is most
1943 useful for infrequently used options which are not required
1944 for booting. For more information, see the man pages for
1945 modprobe, lsmod, modinfo, insmod and rmmod.
1947 If you say Y here, you will need to run "make
1948 modules_install" to put the modules under /lib/modules/
1949 where modprobe can find them (you may need to be root to do
1956 config MODULE_FORCE_LOAD
1957 bool "Forced module loading"
1960 Allow loading of modules without version information (ie. modprobe
1961 --force). Forced module loading sets the 'F' (forced) taint flag and
1962 is usually a really bad idea.
1964 config MODULE_UNLOAD
1965 bool "Module unloading"
1967 Without this option you will not be able to unload any
1968 modules (note that some modules may not be unloadable
1969 anyway), which makes your kernel smaller, faster
1970 and simpler. If unsure, say Y.
1972 config MODULE_FORCE_UNLOAD
1973 bool "Forced module unloading"
1974 depends on MODULE_UNLOAD
1976 This option allows you to force a module to unload, even if the
1977 kernel believes it is unsafe: the kernel will remove the module
1978 without waiting for anyone to stop using it (using the -f option to
1979 rmmod). This is mainly for kernel developers and desperate users.
1983 bool "Module versioning support"
1985 Usually, you have to use modules compiled with your kernel.
1986 Saying Y here makes it sometimes possible to use modules
1987 compiled for different kernels, by adding enough information
1988 to the modules to (hopefully) spot any changes which would
1989 make them incompatible with the kernel you are running. If
1992 config MODULE_REL_CRCS
1994 depends on MODVERSIONS
1996 config MODULE_SRCVERSION_ALL
1997 bool "Source checksum for all modules"
1999 Modules which contain a MODULE_VERSION get an extra "srcversion"
2000 field inserted into their modinfo section, which contains a
2001 sum of the source files which made it. This helps maintainers
2002 see exactly which source was used to build a module (since
2003 others sometimes change the module source without updating
2004 the version). With this option, such a "srcversion" field
2005 will be created for all modules. If unsure, say N.
2008 bool "Module signature verification"
2010 select SYSTEM_DATA_VERIFICATION
2012 Check modules for valid signatures upon load: the signature
2013 is simply appended to the module. For more information see
2014 <file:Documentation/admin-guide/module-signing.rst>.
2016 Note that this option adds the OpenSSL development packages as a
2017 kernel build dependency so that the signing tool can use its crypto
2020 !!!WARNING!!! If you enable this option, you MUST make sure that the
2021 module DOES NOT get stripped after being signed. This includes the
2022 debuginfo strip done by some packagers (such as rpmbuild) and
2023 inclusion into an initramfs that wants the module size reduced.
2025 config MODULE_SIG_FORCE
2026 bool "Require modules to be validly signed"
2027 depends on MODULE_SIG
2029 Reject unsigned modules or signed modules for which we don't have a
2030 key. Without this, such modules will simply taint the kernel.
2032 config MODULE_SIG_ALL
2033 bool "Automatically sign all modules"
2035 depends on MODULE_SIG
2037 Sign all modules during make modules_install. Without this option,
2038 modules must be signed manually, using the scripts/sign-file tool.
2040 comment "Do not forget to sign required modules with scripts/sign-file"
2041 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2044 prompt "Which hash algorithm should modules be signed with?"
2045 depends on MODULE_SIG
2047 This determines which sort of hashing algorithm will be used during
2048 signature generation. This algorithm _must_ be built into the kernel
2049 directly so that signature verification can take place. It is not
2050 possible to load a signed module containing the algorithm to check
2051 the signature on that module.
2053 config MODULE_SIG_SHA1
2054 bool "Sign modules with SHA-1"
2057 config MODULE_SIG_SHA224
2058 bool "Sign modules with SHA-224"
2059 select CRYPTO_SHA256
2061 config MODULE_SIG_SHA256
2062 bool "Sign modules with SHA-256"
2063 select CRYPTO_SHA256
2065 config MODULE_SIG_SHA384
2066 bool "Sign modules with SHA-384"
2067 select CRYPTO_SHA512
2069 config MODULE_SIG_SHA512
2070 bool "Sign modules with SHA-512"
2071 select CRYPTO_SHA512
2075 config MODULE_SIG_HASH
2077 depends on MODULE_SIG
2078 default "sha1" if MODULE_SIG_SHA1
2079 default "sha224" if MODULE_SIG_SHA224
2080 default "sha256" if MODULE_SIG_SHA256
2081 default "sha384" if MODULE_SIG_SHA384
2082 default "sha512" if MODULE_SIG_SHA512
2084 config MODULE_COMPRESS
2085 bool "Compress modules on installation"
2089 Compresses kernel modules when 'make modules_install' is run; gzip or
2090 xz depending on "Compression algorithm" below.
2092 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2094 Out-of-tree kernel modules installed using Kbuild will also be
2095 compressed upon installation.
2097 Note: for modules inside an initrd or initramfs, it's more efficient
2098 to compress the whole initrd or initramfs instead.
2100 Note: This is fully compatible with signed modules.
2105 prompt "Compression algorithm"
2106 depends on MODULE_COMPRESS
2107 default MODULE_COMPRESS_GZIP
2109 This determines which sort of compression will be used during
2110 'make modules_install'.
2112 GZIP (default) and XZ are supported.
2114 config MODULE_COMPRESS_GZIP
2117 config MODULE_COMPRESS_XZ
2122 config TRIM_UNUSED_KSYMS
2123 bool "Trim unused exported kernel symbols"
2124 depends on MODULES && !UNUSED_SYMBOLS
2126 The kernel and some modules make many symbols available for
2127 other modules to use via EXPORT_SYMBOL() and variants. Depending
2128 on the set of modules being selected in your kernel configuration,
2129 many of those exported symbols might never be used.
2131 This option allows for unused exported symbols to be dropped from
2132 the build. In turn, this provides the compiler more opportunities
2133 (especially when using LTO) for optimizing the code and reducing
2134 binary size. This might have some security advantages as well.
2136 If unsure, or if you need to build out-of-tree modules, say N.
2140 config MODULES_TREE_LOOKUP
2142 depends on PERF_EVENTS || TRACING
2144 config INIT_ALL_POSSIBLE
2147 Back when each arch used to define their own cpu_online_mask and
2148 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2149 with all 1s, and others with all 0s. When they were centralised,
2150 it was better to provide this option than to break all the archs
2151 and have several arch maintainers pursuing me down dark alleys.
2153 source "block/Kconfig"
2155 config PREEMPT_NOTIFIERS
2165 Build a simple ASN.1 grammar compiler that produces a bytecode output
2166 that can be interpreted by the ASN.1 stream decoder and used to
2167 inform it as to what tags are to be expected in a stream and what
2168 functions to call on what tags.
2170 source "kernel/Kconfig.locks"
2172 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2175 # It may be useful for an architecture to override the definitions of the
2176 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2177 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2178 # different calling convention for syscalls. They can also override the
2179 # macros for not-implemented syscalls in kernel/sys_ni.c and
2180 # kernel/time/posix-stubs.c. All these overrides need to be available in
2181 # <asm/syscall_wrapper.h>.
2182 config ARCH_HAS_SYSCALL_WRAPPER
git.samba.org / sfrench / cifs-2.6.git / blob