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))
27 config CC_HAS_ASM_GOTO
28 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
30 config CC_HAS_WARN_MAYBE_UNINITIALIZED
31 def_bool $(cc-option,-Wmaybe-uninitialized)
33 GCC >= 4.7 supports this option.
35 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
37 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
38 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
40 GCC's -Wmaybe-uninitialized is not reliable by definition.
41 Lots of false positive warnings are produced in some cases.
43 If this option is enabled, -Wno-maybe-uninitialzed is passed
44 to the compiler to suppress maybe-uninitialized warnings.
53 config BUILDTIME_EXTABLE_SORT
56 config THREAD_INFO_IN_TASK
59 Select this to move thread_info off the stack into task_struct. To
60 make this work, an arch will need to remove all thread_info fields
61 except flags and fix any runtime bugs.
63 One subtle change that will be needed is to use try_get_task_stack()
64 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
73 depends on BROKEN || !SMP
76 config INIT_ENV_ARG_LIMIT
81 Maximum of each of the number of arguments and environment
82 variables passed to init from the kernel command line.
85 bool "Compile also drivers which will not load"
89 Some drivers can be compiled on a different platform than they are
90 intended to be run on. Despite they cannot be loaded there (or even
91 when they load they cannot be used due to missing HW support),
92 developers still, opposing to distributors, might want to build such
93 drivers to compile-test them.
95 If you are a developer and want to build everything available, say Y
96 here. If you are a user/distributor, say N here to exclude useless
97 drivers to be distributed.
100 string "Local version - append to kernel release"
102 Append an extra string to the end of your kernel version.
103 This will show up when you type uname, for example.
104 The string you set here will be appended after the contents of
105 any files with a filename matching localversion* in your
106 object and source tree, in that order. Your total string can
107 be a maximum of 64 characters.
109 config LOCALVERSION_AUTO
110 bool "Automatically append version information to the version string"
112 depends on !COMPILE_TEST
114 This will try to automatically determine if the current tree is a
115 release tree by looking for git tags that belong to the current
116 top of tree revision.
118 A string of the format -gxxxxxxxx will be added to the localversion
119 if a git-based tree is found. The string generated by this will be
120 appended after any matching localversion* files, and after the value
121 set in CONFIG_LOCALVERSION.
123 (The actual string used here is the first eight characters produced
124 by running the command:
126 $ git rev-parse --verify HEAD
128 which is done within the script "scripts/setlocalversion".)
131 string "Build ID Salt"
134 The build ID is used to link binaries and their debug info. Setting
135 this option will use the value in the calculation of the build id.
136 This is mostly useful for distributions which want to ensure the
137 build is unique between builds. It's safe to leave the default.
139 config HAVE_KERNEL_GZIP
142 config HAVE_KERNEL_BZIP2
145 config HAVE_KERNEL_LZMA
148 config HAVE_KERNEL_XZ
151 config HAVE_KERNEL_LZO
154 config HAVE_KERNEL_LZ4
157 config HAVE_KERNEL_UNCOMPRESSED
161 prompt "Kernel compression mode"
163 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
165 The linux kernel is a kind of self-extracting executable.
166 Several compression algorithms are available, which differ
167 in efficiency, compression and decompression speed.
168 Compression speed is only relevant when building a kernel.
169 Decompression speed is relevant at each boot.
171 If you have any problems with bzip2 or lzma compressed
172 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
173 version of this functionality (bzip2 only), for 2.4, was
174 supplied by Christian Ludwig)
176 High compression options are mostly useful for users, who
177 are low on disk space (embedded systems), but for whom ram
180 If in doubt, select 'gzip'
184 depends on HAVE_KERNEL_GZIP
186 The old and tried gzip compression. It provides a good balance
187 between compression ratio and decompression speed.
191 depends on HAVE_KERNEL_BZIP2
193 Its compression ratio and speed is intermediate.
194 Decompression speed is slowest among the choices. The kernel
195 size is about 10% smaller with bzip2, in comparison to gzip.
196 Bzip2 uses a large amount of memory. For modern kernels you
197 will need at least 8MB RAM or more for booting.
201 depends on HAVE_KERNEL_LZMA
203 This compression algorithm's ratio is best. Decompression speed
204 is between gzip and bzip2. Compression is slowest.
205 The kernel size is about 33% smaller with LZMA in comparison to gzip.
209 depends on HAVE_KERNEL_XZ
211 XZ uses the LZMA2 algorithm and instruction set specific
212 BCJ filters which can improve compression ratio of executable
213 code. The size of the kernel is about 30% smaller with XZ in
214 comparison to gzip. On architectures for which there is a BCJ
215 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
216 will create a few percent smaller kernel than plain LZMA.
218 The speed is about the same as with LZMA: The decompression
219 speed of XZ is better than that of bzip2 but worse than gzip
220 and LZO. Compression is slow.
224 depends on HAVE_KERNEL_LZO
226 Its compression ratio is the poorest among the choices. The kernel
227 size is about 10% bigger than gzip; however its speed
228 (both compression and decompression) is the fastest.
232 depends on HAVE_KERNEL_LZ4
234 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
235 A preliminary version of LZ4 de/compression tool is available at
236 <https://code.google.com/p/lz4/>.
238 Its compression ratio is worse than LZO. The size of the kernel
239 is about 8% bigger than LZO. But the decompression speed is
242 config KERNEL_UNCOMPRESSED
244 depends on HAVE_KERNEL_UNCOMPRESSED
246 Produce uncompressed kernel image. This option is usually not what
247 you want. It is useful for debugging the kernel in slow simulation
248 environments, where decompressing and moving the kernel is awfully
249 slow. This option allows early boot code to skip the decompressor
250 and jump right at uncompressed kernel image.
254 config DEFAULT_HOSTNAME
255 string "Default hostname"
258 This option determines the default system hostname before userspace
259 calls sethostname(2). The kernel traditionally uses "(none)" here,
260 but you may wish to use a different default here to make a minimal
261 system more usable with less configuration.
264 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
265 # add proper SWAP support to them, in which case this can be remove.
271 bool "Support for paging of anonymous memory (swap)"
272 depends on MMU && BLOCK && !ARCH_NO_SWAP
275 This option allows you to choose whether you want to have support
276 for so called swap devices or swap files in your kernel that are
277 used to provide more virtual memory than the actual RAM present
278 in your computer. If unsure say Y.
283 Inter Process Communication is a suite of library functions and
284 system calls which let processes (running programs) synchronize and
285 exchange information. It is generally considered to be a good thing,
286 and some programs won't run unless you say Y here. In particular, if
287 you want to run the DOS emulator dosemu under Linux (read the
288 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
289 you'll need to say Y here.
291 You can find documentation about IPC with "info ipc" and also in
292 section 6.4 of the Linux Programmer's Guide, available from
293 <http://www.tldp.org/guides.html>.
295 config SYSVIPC_SYSCTL
302 bool "POSIX Message Queues"
305 POSIX variant of message queues is a part of IPC. In POSIX message
306 queues every message has a priority which decides about succession
307 of receiving it by a process. If you want to compile and run
308 programs written e.g. for Solaris with use of its POSIX message
309 queues (functions mq_*) say Y here.
311 POSIX message queues are visible as a filesystem called 'mqueue'
312 and can be mounted somewhere if you want to do filesystem
313 operations on message queues.
317 config POSIX_MQUEUE_SYSCTL
319 depends on POSIX_MQUEUE
323 config CROSS_MEMORY_ATTACH
324 bool "Enable process_vm_readv/writev syscalls"
328 Enabling this option adds the system calls process_vm_readv and
329 process_vm_writev which allow a process with the correct privileges
330 to directly read from or write to another process' address space.
331 See the man page for more details.
334 bool "uselib syscall"
335 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
337 This option enables the uselib syscall, a system call used in the
338 dynamic linker from libc5 and earlier. glibc does not use this
339 system call. If you intend to run programs built on libc5 or
340 earlier, you may need to enable this syscall. Current systems
341 running glibc can safely disable this.
344 bool "Auditing support"
347 Enable auditing infrastructure that can be used with another
348 kernel subsystem, such as SELinux (which requires this for
349 logging of avc messages output). System call auditing is included
350 on architectures which support it.
352 config HAVE_ARCH_AUDITSYSCALL
357 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
360 source "kernel/irq/Kconfig"
361 source "kernel/time/Kconfig"
362 source "kernel/Kconfig.preempt"
364 menu "CPU/Task time and stats accounting"
366 config VIRT_CPU_ACCOUNTING
370 prompt "Cputime accounting"
371 default TICK_CPU_ACCOUNTING if !PPC64
372 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
374 # Kind of a stub config for the pure tick based cputime accounting
375 config TICK_CPU_ACCOUNTING
376 bool "Simple tick based cputime accounting"
377 depends on !S390 && !NO_HZ_FULL
379 This is the basic tick based cputime accounting that maintains
380 statistics about user, system and idle time spent on per jiffies
385 config VIRT_CPU_ACCOUNTING_NATIVE
386 bool "Deterministic task and CPU time accounting"
387 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
388 select VIRT_CPU_ACCOUNTING
390 Select this option to enable more accurate task and CPU time
391 accounting. This is done by reading a CPU counter on each
392 kernel entry and exit and on transitions within the kernel
393 between system, softirq and hardirq state, so there is a
394 small performance impact. In the case of s390 or IBM POWER > 5,
395 this also enables accounting of stolen time on logically-partitioned
398 config VIRT_CPU_ACCOUNTING_GEN
399 bool "Full dynticks CPU time accounting"
400 depends on HAVE_CONTEXT_TRACKING
401 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
402 depends on GENERIC_CLOCKEVENTS
403 select VIRT_CPU_ACCOUNTING
404 select CONTEXT_TRACKING
406 Select this option to enable task and CPU time accounting on full
407 dynticks systems. This accounting is implemented by watching every
408 kernel-user boundaries using the context tracking subsystem.
409 The accounting is thus performed at the expense of some significant
412 For now this is only useful if you are working on the full
413 dynticks subsystem development.
419 config IRQ_TIME_ACCOUNTING
420 bool "Fine granularity task level IRQ time accounting"
421 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
423 Select this option to enable fine granularity task irq time
424 accounting. This is done by reading a timestamp on each
425 transitions between softirq and hardirq state, so there can be a
426 small performance impact.
428 If in doubt, say N here.
430 config HAVE_SCHED_AVG_IRQ
432 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
435 config BSD_PROCESS_ACCT
436 bool "BSD Process Accounting"
439 If you say Y here, a user level program will be able to instruct the
440 kernel (via a special system call) to write process accounting
441 information to a file: whenever a process exits, information about
442 that process will be appended to the file by the kernel. The
443 information includes things such as creation time, owning user,
444 command name, memory usage, controlling terminal etc. (the complete
445 list is in the struct acct in <file:include/linux/acct.h>). It is
446 up to the user level program to do useful things with this
447 information. This is generally a good idea, so say Y.
449 config BSD_PROCESS_ACCT_V3
450 bool "BSD Process Accounting version 3 file format"
451 depends on BSD_PROCESS_ACCT
454 If you say Y here, the process accounting information is written
455 in a new file format that also logs the process IDs of each
456 process and its parent. Note that this file format is incompatible
457 with previous v0/v1/v2 file formats, so you will need updated tools
458 for processing it. A preliminary version of these tools is available
459 at <http://www.gnu.org/software/acct/>.
462 bool "Export task/process statistics through netlink"
467 Export selected statistics for tasks/processes through the
468 generic netlink interface. Unlike BSD process accounting, the
469 statistics are available during the lifetime of tasks/processes as
470 responses to commands. Like BSD accounting, they are sent to user
475 config TASK_DELAY_ACCT
476 bool "Enable per-task delay accounting"
480 Collect information on time spent by a task waiting for system
481 resources like cpu, synchronous block I/O completion and swapping
482 in pages. Such statistics can help in setting a task's priorities
483 relative to other tasks for cpu, io, rss limits etc.
488 bool "Enable extended accounting over taskstats"
491 Collect extended task accounting data and send the data
492 to userland for processing over the taskstats interface.
496 config TASK_IO_ACCOUNTING
497 bool "Enable per-task storage I/O accounting"
498 depends on TASK_XACCT
500 Collect information on the number of bytes of storage I/O which this
506 bool "Pressure stall information tracking"
508 Collect metrics that indicate how overcommitted the CPU, memory,
509 and IO capacity are in the system.
511 If you say Y here, the kernel will create /proc/pressure/ with the
512 pressure statistics files cpu, memory, and io. These will indicate
513 the share of walltime in which some or all tasks in the system are
514 delayed due to contention of the respective resource.
516 In kernels with cgroup support, cgroups (cgroup2 only) will
517 have cpu.pressure, memory.pressure, and io.pressure files,
518 which aggregate pressure stalls for the grouped tasks only.
520 For more details see Documentation/accounting/psi.txt.
524 config PSI_DEFAULT_DISABLED
525 bool "Require boot parameter to enable pressure stall information tracking"
529 If set, pressure stall information tracking will be disabled
530 per default but can be enabled through passing psi=1 on the
531 kernel commandline during boot.
533 This feature adds some code to the task wakeup and sleep
534 paths of the scheduler. The overhead is too low to affect
535 common scheduling-intense workloads in practice (such as
536 webservers, memcache), but it does show up in artificial
537 scheduler stress tests, such as hackbench.
539 If you are paranoid and not sure what the kernel will be
544 endmenu # "CPU/Task time and stats accounting"
548 depends on SMP || COMPILE_TEST
551 Make sure that CPUs running critical tasks are not disturbed by
552 any source of "noise" such as unbound workqueues, timers, kthreads...
553 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
554 the "isolcpus=" boot parameter.
558 source "kernel/rcu/Kconfig"
565 tristate "Kernel .config support"
567 This option enables the complete Linux kernel ".config" file
568 contents to be saved in the kernel. It provides documentation
569 of which kernel options are used in a running kernel or in an
570 on-disk kernel. This information can be extracted from the kernel
571 image file with the script scripts/extract-ikconfig and used as
572 input to rebuild the current kernel or to build another kernel.
573 It can also be extracted from a running kernel by reading
574 /proc/config.gz if enabled (below).
577 bool "Enable access to .config through /proc/config.gz"
578 depends on IKCONFIG && PROC_FS
580 This option enables access to the kernel configuration file
581 through /proc/config.gz.
584 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
587 This option enables access to the in-kernel headers that are generated during
588 the build process. These can be used to build eBPF tracing programs,
589 or similar programs. If you build the headers as a module, a module called
590 kheaders.ko is built which can be loaded on-demand to get access to headers.
593 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
598 Select the minimal kernel log buffer size as a power of 2.
599 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
600 parameter, see below. Any higher size also might be forced
601 by "log_buf_len" boot parameter.
611 config LOG_CPU_MAX_BUF_SHIFT
612 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
615 default 12 if !BASE_SMALL
616 default 0 if BASE_SMALL
619 This option allows to increase the default ring buffer size
620 according to the number of CPUs. The value defines the contribution
621 of each CPU as a power of 2. The used space is typically only few
622 lines however it might be much more when problems are reported,
625 The increased size means that a new buffer has to be allocated and
626 the original static one is unused. It makes sense only on systems
627 with more CPUs. Therefore this value is used only when the sum of
628 contributions is greater than the half of the default kernel ring
629 buffer as defined by LOG_BUF_SHIFT. The default values are set
630 so that more than 64 CPUs are needed to trigger the allocation.
632 Also this option is ignored when "log_buf_len" kernel parameter is
633 used as it forces an exact (power of two) size of the ring buffer.
635 The number of possible CPUs is used for this computation ignoring
636 hotplugging making the computation optimal for the worst case
637 scenario while allowing a simple algorithm to be used from bootup.
639 Examples shift values and their meaning:
640 17 => 128 KB for each CPU
641 16 => 64 KB for each CPU
642 15 => 32 KB for each CPU
643 14 => 16 KB for each CPU
644 13 => 8 KB for each CPU
645 12 => 4 KB for each CPU
647 config PRINTK_SAFE_LOG_BUF_SHIFT
648 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
653 Select the size of an alternate printk per-CPU buffer where messages
654 printed from usafe contexts are temporary stored. One example would
655 be NMI messages, another one - printk recursion. The messages are
656 copied to the main log buffer in a safe context to avoid a deadlock.
657 The value defines the size as a power of 2.
659 Those messages are rare and limited. The largest one is when
660 a backtrace is printed. It usually fits into 4KB. Select
661 8KB if you want to be on the safe side.
664 17 => 128 KB for each CPU
665 16 => 64 KB for each CPU
666 15 => 32 KB for each CPU
667 14 => 16 KB for each CPU
668 13 => 8 KB for each CPU
669 12 => 4 KB for each CPU
672 # Architectures with an unreliable sched_clock() should select this:
674 config HAVE_UNSTABLE_SCHED_CLOCK
677 config GENERIC_SCHED_CLOCK
680 menu "Scheduler features"
683 bool "Enable utilization clamping for RT/FAIR tasks"
684 depends on CPU_FREQ_GOV_SCHEDUTIL
686 This feature enables the scheduler to track the clamped utilization
687 of each CPU based on RUNNABLE tasks scheduled on that CPU.
689 With this option, the user can specify the min and max CPU
690 utilization allowed for RUNNABLE tasks. The max utilization defines
691 the maximum frequency a task should use while the min utilization
692 defines the minimum frequency it should use.
694 Both min and max utilization clamp values are hints to the scheduler,
695 aiming at improving its frequency selection policy, but they do not
696 enforce or grant any specific bandwidth for tasks.
700 config UCLAMP_BUCKETS_COUNT
701 int "Number of supported utilization clamp buckets"
704 depends on UCLAMP_TASK
706 Defines the number of clamp buckets to use. The range of each bucket
707 will be SCHED_CAPACITY_SCALE/UCLAMP_BUCKETS_COUNT. The higher the
708 number of clamp buckets the finer their granularity and the higher
709 the precision of clamping aggregation and tracking at run-time.
711 For example, with the minimum configuration value we will have 5
712 clamp buckets tracking 20% utilization each. A 25% boosted tasks will
713 be refcounted in the [20..39]% bucket and will set the bucket clamp
714 effective value to 25%.
715 If a second 30% boosted task should be co-scheduled on the same CPU,
716 that task will be refcounted in the same bucket of the first task and
717 it will boost the bucket clamp effective value to 30%.
718 The clamp effective value of a bucket is reset to its nominal value
719 (20% in the example above) when there are no more tasks refcounted in
722 An additional boost/capping margin can be added to some tasks. In the
723 example above the 25% task will be boosted to 30% until it exits the
724 CPU. If that should be considered not acceptable on certain systems,
725 it's always possible to reduce the margin by increasing the number of
726 clamp buckets to trade off used memory for run-time tracking
729 If in doubt, use the default value.
734 # For architectures that want to enable the support for NUMA-affine scheduler
737 config ARCH_SUPPORTS_NUMA_BALANCING
741 # For architectures that prefer to flush all TLBs after a number of pages
742 # are unmapped instead of sending one IPI per page to flush. The architecture
743 # must provide guarantees on what happens if a clean TLB cache entry is
744 # written after the unmap. Details are in mm/rmap.c near the check for
745 # should_defer_flush. The architecture should also consider if the full flush
746 # and the refill costs are offset by the savings of sending fewer IPIs.
747 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
751 # For architectures that know their GCC __int128 support is sound
753 config ARCH_SUPPORTS_INT128
756 # For architectures that (ab)use NUMA to represent different memory regions
757 # all cpu-local but of different latencies, such as SuperH.
759 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
762 config NUMA_BALANCING
763 bool "Memory placement aware NUMA scheduler"
764 depends on ARCH_SUPPORTS_NUMA_BALANCING
765 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
766 depends on SMP && NUMA && MIGRATION
768 This option adds support for automatic NUMA aware memory/task placement.
769 The mechanism is quite primitive and is based on migrating memory when
770 it has references to the node the task is running on.
772 This system will be inactive on UMA systems.
774 config NUMA_BALANCING_DEFAULT_ENABLED
775 bool "Automatically enable NUMA aware memory/task placement"
777 depends on NUMA_BALANCING
779 If set, automatic NUMA balancing will be enabled if running on a NUMA
783 bool "Control Group support"
786 This option adds support for grouping sets of processes together, for
787 use with process control subsystems such as Cpusets, CFS, memory
788 controls or device isolation.
790 - Documentation/scheduler/sched-design-CFS.txt (CFS)
791 - Documentation/cgroup-v1/ (features for grouping, isolation
792 and resource control)
802 bool "Memory controller"
806 Provides control over the memory footprint of tasks in a cgroup.
809 bool "Swap controller"
810 depends on MEMCG && SWAP
812 Provides control over the swap space consumed by tasks in a cgroup.
814 config MEMCG_SWAP_ENABLED
815 bool "Swap controller enabled by default"
816 depends on MEMCG_SWAP
819 Memory Resource Controller Swap Extension comes with its price in
820 a bigger memory consumption. General purpose distribution kernels
821 which want to enable the feature but keep it disabled by default
822 and let the user enable it by swapaccount=1 boot command line
823 parameter should have this option unselected.
824 For those who want to have the feature enabled by default should
825 select this option (if, for some reason, they need to disable it
826 then swapaccount=0 does the trick).
830 depends on MEMCG && !SLOB
838 Generic block IO controller cgroup interface. This is the common
839 cgroup interface which should be used by various IO controlling
842 Currently, CFQ IO scheduler uses it to recognize task groups and
843 control disk bandwidth allocation (proportional time slice allocation)
844 to such task groups. It is also used by bio throttling logic in
845 block layer to implement upper limit in IO rates on a device.
847 This option only enables generic Block IO controller infrastructure.
848 One needs to also enable actual IO controlling logic/policy. For
849 enabling proportional weight division of disk bandwidth in CFQ, set
850 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
851 CONFIG_BLK_DEV_THROTTLING=y.
853 See Documentation/cgroup-v1/blkio-controller.txt for more information.
855 config DEBUG_BLK_CGROUP
856 bool "IO controller debugging"
857 depends on BLK_CGROUP
860 Enable some debugging help. Currently it exports additional stat
861 files in a cgroup which can be useful for debugging.
863 config CGROUP_WRITEBACK
865 depends on MEMCG && BLK_CGROUP
868 menuconfig CGROUP_SCHED
869 bool "CPU controller"
872 This feature lets CPU scheduler recognize task groups and control CPU
873 bandwidth allocation to such task groups. It uses cgroups to group
877 config FAIR_GROUP_SCHED
878 bool "Group scheduling for SCHED_OTHER"
879 depends on CGROUP_SCHED
883 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
884 depends on FAIR_GROUP_SCHED
887 This option allows users to define CPU bandwidth rates (limits) for
888 tasks running within the fair group scheduler. Groups with no limit
889 set are considered to be unconstrained and will run with no
891 See Documentation/scheduler/sched-bwc.txt for more information.
893 config RT_GROUP_SCHED
894 bool "Group scheduling for SCHED_RR/FIFO"
895 depends on CGROUP_SCHED
898 This feature lets you explicitly allocate real CPU bandwidth
899 to task groups. If enabled, it will also make it impossible to
900 schedule realtime tasks for non-root users until you allocate
901 realtime bandwidth for them.
902 See Documentation/scheduler/sched-rt-group.txt for more information.
907 bool "PIDs controller"
909 Provides enforcement of process number limits in the scope of a
910 cgroup. Any attempt to fork more processes than is allowed in the
911 cgroup will fail. PIDs are fundamentally a global resource because it
912 is fairly trivial to reach PID exhaustion before you reach even a
913 conservative kmemcg limit. As a result, it is possible to grind a
914 system to halt without being limited by other cgroup policies. The
915 PIDs controller is designed to stop this from happening.
917 It should be noted that organisational operations (such as attaching
918 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
919 since the PIDs limit only affects a process's ability to fork, not to
923 bool "RDMA controller"
925 Provides enforcement of RDMA resources defined by IB stack.
926 It is fairly easy for consumers to exhaust RDMA resources, which
927 can result into resource unavailability to other consumers.
928 RDMA controller is designed to stop this from happening.
929 Attaching processes with active RDMA resources to the cgroup
930 hierarchy is allowed even if can cross the hierarchy's limit.
932 config CGROUP_FREEZER
933 bool "Freezer controller"
935 Provides a way to freeze and unfreeze all tasks in a
938 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
939 controller includes important in-kernel memory consumers per default.
941 If you're using cgroup2, say N.
943 config CGROUP_HUGETLB
944 bool "HugeTLB controller"
945 depends on HUGETLB_PAGE
949 Provides a cgroup controller for HugeTLB pages.
950 When you enable this, you can put a per cgroup limit on HugeTLB usage.
951 The limit is enforced during page fault. Since HugeTLB doesn't
952 support page reclaim, enforcing the limit at page fault time implies
953 that, the application will get SIGBUS signal if it tries to access
954 HugeTLB pages beyond its limit. This requires the application to know
955 beforehand how much HugeTLB pages it would require for its use. The
956 control group is tracked in the third page lru pointer. This means
957 that we cannot use the controller with huge page less than 3 pages.
960 bool "Cpuset controller"
963 This option will let you create and manage CPUSETs which
964 allow dynamically partitioning a system into sets of CPUs and
965 Memory Nodes and assigning tasks to run only within those sets.
966 This is primarily useful on large SMP or NUMA systems.
970 config PROC_PID_CPUSET
971 bool "Include legacy /proc/<pid>/cpuset file"
976 bool "Device controller"
978 Provides a cgroup controller implementing whitelists for
979 devices which a process in the cgroup can mknod or open.
981 config CGROUP_CPUACCT
982 bool "Simple CPU accounting controller"
984 Provides a simple controller for monitoring the
985 total CPU consumed by the tasks in a cgroup.
988 bool "Perf controller"
989 depends on PERF_EVENTS
991 This option extends the perf per-cpu mode to restrict monitoring
992 to threads which belong to the cgroup specified and run on the
998 bool "Support for eBPF programs attached to cgroups"
999 depends on BPF_SYSCALL
1000 select SOCK_CGROUP_DATA
1002 Allow attaching eBPF programs to a cgroup using the bpf(2)
1003 syscall command BPF_PROG_ATTACH.
1005 In which context these programs are accessed depends on the type
1006 of attachment. For instance, programs that are attached using
1007 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1011 bool "Debug controller"
1013 depends on DEBUG_KERNEL
1015 This option enables a simple controller that exports
1016 debugging information about the cgroups framework. This
1017 controller is for control cgroup debugging only. Its
1018 interfaces are not stable.
1022 config SOCK_CGROUP_DATA
1028 menuconfig NAMESPACES
1029 bool "Namespaces support" if EXPERT
1030 depends on MULTIUSER
1033 Provides the way to make tasks work with different objects using
1034 the same id. For example same IPC id may refer to different objects
1035 or same user id or pid may refer to different tasks when used in
1036 different namespaces.
1041 bool "UTS namespace"
1044 In this namespace tasks see different info provided with the
1048 bool "IPC namespace"
1049 depends on (SYSVIPC || POSIX_MQUEUE)
1052 In this namespace tasks work with IPC ids which correspond to
1053 different IPC objects in different namespaces.
1056 bool "User namespace"
1059 This allows containers, i.e. vservers, to use user namespaces
1060 to provide different user info for different servers.
1062 When user namespaces are enabled in the kernel it is
1063 recommended that the MEMCG option also be enabled and that
1064 user-space use the memory control groups to limit the amount
1065 of memory a memory unprivileged users can use.
1070 bool "PID Namespaces"
1073 Support process id namespaces. This allows having multiple
1074 processes with the same pid as long as they are in different
1075 pid namespaces. This is a building block of containers.
1078 bool "Network namespace"
1082 Allow user space to create what appear to be multiple instances
1083 of the network stack.
1087 config CHECKPOINT_RESTORE
1088 bool "Checkpoint/restore support"
1089 select PROC_CHILDREN
1092 Enables additional kernel features in a sake of checkpoint/restore.
1093 In particular it adds auxiliary prctl codes to setup process text,
1094 data and heap segment sizes, and a few additional /proc filesystem
1097 If unsure, say N here.
1099 config SCHED_AUTOGROUP
1100 bool "Automatic process group scheduling"
1103 select FAIR_GROUP_SCHED
1105 This option optimizes the scheduler for common desktop workloads by
1106 automatically creating and populating task groups. This separation
1107 of workloads isolates aggressive CPU burners (like build jobs) from
1108 desktop applications. Task group autogeneration is currently based
1111 config SYSFS_DEPRECATED
1112 bool "Enable deprecated sysfs features to support old userspace tools"
1116 This option adds code that switches the layout of the "block" class
1117 devices, to not show up in /sys/class/block/, but only in
1120 This switch is only active when the sysfs.deprecated=1 boot option is
1121 passed or the SYSFS_DEPRECATED_V2 option is set.
1123 This option allows new kernels to run on old distributions and tools,
1124 which might get confused by /sys/class/block/. Since 2007/2008 all
1125 major distributions and tools handle this just fine.
1127 Recent distributions and userspace tools after 2009/2010 depend on
1128 the existence of /sys/class/block/, and will not work with this
1131 Only if you are using a new kernel on an old distribution, you might
1134 config SYSFS_DEPRECATED_V2
1135 bool "Enable deprecated sysfs features by default"
1138 depends on SYSFS_DEPRECATED
1140 Enable deprecated sysfs by default.
1142 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1145 Only if you are using a new kernel on an old distribution, you might
1146 need to say Y here. Even then, odds are you would not need it
1147 enabled, you can always pass the boot option if absolutely necessary.
1150 bool "Kernel->user space relay support (formerly relayfs)"
1153 This option enables support for relay interface support in
1154 certain file systems (such as debugfs).
1155 It is designed to provide an efficient mechanism for tools and
1156 facilities to relay large amounts of data from kernel space to
1161 config BLK_DEV_INITRD
1162 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1164 The initial RAM filesystem is a ramfs which is loaded by the
1165 boot loader (loadlin or lilo) and that is mounted as root
1166 before the normal boot procedure. It is typically used to
1167 load modules needed to mount the "real" root file system,
1168 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1170 If RAM disk support (BLK_DEV_RAM) is also included, this
1171 also enables initial RAM disk (initrd) support and adds
1172 15 Kbytes (more on some other architectures) to the kernel size.
1178 source "usr/Kconfig"
1183 prompt "Compiler optimization level"
1184 default CC_OPTIMIZE_FOR_PERFORMANCE
1186 config CC_OPTIMIZE_FOR_PERFORMANCE
1187 bool "Optimize for performance"
1189 This is the default optimization level for the kernel, building
1190 with the "-O2" compiler flag for best performance and most
1191 helpful compile-time warnings.
1193 config CC_OPTIMIZE_FOR_SIZE
1194 bool "Optimize for size"
1195 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1197 Enabling this option will pass "-Os" instead of "-O2" to
1198 your compiler resulting in a smaller kernel.
1204 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1207 This requires that the arch annotates or otherwise protects
1208 its external entry points from being discarded. Linker scripts
1209 must also merge .text.*, .data.*, and .bss.* correctly into
1210 output sections. Care must be taken not to pull in unrelated
1211 sections (e.g., '.text.init'). Typically '.' in section names
1212 is used to distinguish them from label names / C identifiers.
1214 config LD_DEAD_CODE_DATA_ELIMINATION
1215 bool "Dead code and data elimination (EXPERIMENTAL)"
1216 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1218 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1219 depends on $(cc-option,-ffunction-sections -fdata-sections)
1220 depends on $(ld-option,--gc-sections)
1222 Enable this if you want to do dead code and data elimination with
1223 the linker by compiling with -ffunction-sections -fdata-sections,
1224 and linking with --gc-sections.
1226 This can reduce on disk and in-memory size of the kernel
1227 code and static data, particularly for small configs and
1228 on small systems. This has the possibility of introducing
1229 silently broken kernel if the required annotations are not
1230 present. This option is not well tested yet, so use at your
1239 config SYSCTL_EXCEPTION_TRACE
1242 Enable support for /proc/sys/debug/exception-trace.
1244 config SYSCTL_ARCH_UNALIGN_NO_WARN
1247 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1248 Allows arch to define/use @no_unaligned_warning to possibly warn
1249 about unaligned access emulation going on under the hood.
1251 config SYSCTL_ARCH_UNALIGN_ALLOW
1254 Enable support for /proc/sys/kernel/unaligned-trap
1255 Allows arches to define/use @unaligned_enabled to runtime toggle
1256 the unaligned access emulation.
1257 see arch/parisc/kernel/unaligned.c for reference
1259 config HAVE_PCSPKR_PLATFORM
1262 # interpreter that classic socket filters depend on
1267 bool "Configure standard kernel features (expert users)"
1268 # Unhide debug options, to make the on-by-default options visible
1271 This option allows certain base kernel options and settings
1272 to be disabled or tweaked. This is for specialized
1273 environments which can tolerate a "non-standard" kernel.
1274 Only use this if you really know what you are doing.
1277 bool "Enable 16-bit UID system calls" if EXPERT
1278 depends on HAVE_UID16 && MULTIUSER
1281 This enables the legacy 16-bit UID syscall wrappers.
1284 bool "Multiple users, groups and capabilities support" if EXPERT
1287 This option enables support for non-root users, groups and
1290 If you say N here, all processes will run with UID 0, GID 0, and all
1291 possible capabilities. Saying N here also compiles out support for
1292 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1295 If unsure, say Y here.
1297 config SGETMASK_SYSCALL
1298 bool "sgetmask/ssetmask syscalls support" if EXPERT
1299 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1301 sys_sgetmask and sys_ssetmask are obsolete system calls
1302 no longer supported in libc but still enabled by default in some
1305 If unsure, leave the default option here.
1307 config SYSFS_SYSCALL
1308 bool "Sysfs syscall support" if EXPERT
1311 sys_sysfs is an obsolete system call no longer supported in libc.
1312 Note that disabling this option is more secure but might break
1313 compatibility with some systems.
1315 If unsure say Y here.
1317 config SYSCTL_SYSCALL
1318 bool "Sysctl syscall support" if EXPERT
1319 depends on PROC_SYSCTL
1323 sys_sysctl uses binary paths that have been found challenging
1324 to properly maintain and use. The interface in /proc/sys
1325 using paths with ascii names is now the primary path to this
1328 Almost nothing using the binary sysctl interface so if you are
1329 trying to save some space it is probably safe to disable this,
1330 making your kernel marginally smaller.
1332 If unsure say N here.
1335 bool "open by fhandle syscalls" if EXPERT
1339 If you say Y here, a user level program will be able to map
1340 file names to handle and then later use the handle for
1341 different file system operations. This is useful in implementing
1342 userspace file servers, which now track files using handles instead
1343 of names. The handle would remain the same even if file names
1344 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1348 bool "Posix Clocks & timers" if EXPERT
1351 This includes native support for POSIX timers to the kernel.
1352 Some embedded systems have no use for them and therefore they
1353 can be configured out to reduce the size of the kernel image.
1355 When this option is disabled, the following syscalls won't be
1356 available: timer_create, timer_gettime: timer_getoverrun,
1357 timer_settime, timer_delete, clock_adjtime, getitimer,
1358 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1359 clock_getres and clock_nanosleep syscalls will be limited to
1360 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1366 bool "Enable support for printk" if EXPERT
1369 This option enables normal printk support. Removing it
1370 eliminates most of the message strings from the kernel image
1371 and makes the kernel more or less silent. As this makes it
1372 very difficult to diagnose system problems, saying N here is
1373 strongly discouraged.
1381 bool "BUG() support" if EXPERT
1384 Disabling this option eliminates support for BUG and WARN, reducing
1385 the size of your kernel image and potentially quietly ignoring
1386 numerous fatal conditions. You should only consider disabling this
1387 option for embedded systems with no facilities for reporting errors.
1393 bool "Enable ELF core dumps" if EXPERT
1395 Enable support for generating core dumps. Disabling saves about 4k.
1398 config PCSPKR_PLATFORM
1399 bool "Enable PC-Speaker support" if EXPERT
1400 depends on HAVE_PCSPKR_PLATFORM
1404 This option allows to disable the internal PC-Speaker
1405 support, saving some memory.
1409 bool "Enable full-sized data structures for core" if EXPERT
1411 Disabling this option reduces the size of miscellaneous core
1412 kernel data structures. This saves memory on small machines,
1413 but may reduce performance.
1416 bool "Enable futex support" if EXPERT
1420 Disabling this option will cause the kernel to be built without
1421 support for "fast userspace mutexes". The resulting kernel may not
1422 run glibc-based applications correctly.
1426 depends on FUTEX && RT_MUTEXES
1429 config HAVE_FUTEX_CMPXCHG
1433 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1434 is implemented and always working. This removes a couple of runtime
1438 bool "Enable eventpoll support" if EXPERT
1441 Disabling this option will cause the kernel to be built without
1442 support for epoll family of system calls.
1445 bool "Enable signalfd() system call" if EXPERT
1448 Enable the signalfd() system call that allows to receive signals
1449 on a file descriptor.
1454 bool "Enable timerfd() system call" if EXPERT
1457 Enable the timerfd() system call that allows to receive timer
1458 events on a file descriptor.
1463 bool "Enable eventfd() system call" if EXPERT
1466 Enable the eventfd() system call that allows to receive both
1467 kernel notification (ie. KAIO) or userspace notifications.
1472 bool "Use full shmem filesystem" if EXPERT
1476 The shmem is an internal filesystem used to manage shared memory.
1477 It is backed by swap and manages resource limits. It is also exported
1478 to userspace as tmpfs if TMPFS is enabled. Disabling this
1479 option replaces shmem and tmpfs with the much simpler ramfs code,
1480 which may be appropriate on small systems without swap.
1483 bool "Enable AIO support" if EXPERT
1486 This option enables POSIX asynchronous I/O which may by used
1487 by some high performance threaded applications. Disabling
1488 this option saves about 7k.
1491 bool "Enable IO uring support" if EXPERT
1495 This option enables support for the io_uring interface, enabling
1496 applications to submit and complete IO through submission and
1497 completion rings that are shared between the kernel and application.
1499 config ADVISE_SYSCALLS
1500 bool "Enable madvise/fadvise syscalls" if EXPERT
1503 This option enables the madvise and fadvise syscalls, used by
1504 applications to advise the kernel about their future memory or file
1505 usage, improving performance. If building an embedded system where no
1506 applications use these syscalls, you can disable this option to save
1510 bool "Enable membarrier() system call" if EXPERT
1513 Enable the membarrier() system call that allows issuing memory
1514 barriers across all running threads, which can be used to distribute
1515 the cost of user-space memory barriers asymmetrically by transforming
1516 pairs of memory barriers into pairs consisting of membarrier() and a
1522 bool "Load all symbols for debugging/ksymoops" if EXPERT
1525 Say Y here to let the kernel print out symbolic crash information and
1526 symbolic stack backtraces. This increases the size of the kernel
1527 somewhat, as all symbols have to be loaded into the kernel image.
1530 bool "Include all symbols in kallsyms"
1531 depends on DEBUG_KERNEL && KALLSYMS
1533 Normally kallsyms only contains the symbols of functions for nicer
1534 OOPS messages and backtraces (i.e., symbols from the text and inittext
1535 sections). This is sufficient for most cases. And only in very rare
1536 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1537 names of variables from the data sections, etc).
1539 This option makes sure that all symbols are loaded into the kernel
1540 image (i.e., symbols from all sections) in cost of increased kernel
1541 size (depending on the kernel configuration, it may be 300KiB or
1542 something like this).
1544 Say N unless you really need all symbols.
1546 config KALLSYMS_ABSOLUTE_PERCPU
1549 default X86_64 && SMP
1551 config KALLSYMS_BASE_RELATIVE
1556 Instead of emitting them as absolute values in the native word size,
1557 emit the symbol references in the kallsyms table as 32-bit entries,
1558 each containing a relative value in the range [base, base + U32_MAX]
1559 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1560 an absolute value in the range [0, S32_MAX] or a relative value in the
1561 range [base, base + S32_MAX], where base is the lowest relative symbol
1562 address encountered in the image.
1564 On 64-bit builds, this reduces the size of the address table by 50%,
1565 but more importantly, it results in entries whose values are build
1566 time constants, and no relocation pass is required at runtime to fix
1567 up the entries based on the runtime load address of the kernel.
1569 # end of the "standard kernel features (expert users)" menu
1571 # syscall, maps, verifier
1573 bool "Enable bpf() system call"
1578 Enable the bpf() system call that allows to manipulate eBPF
1579 programs and maps via file descriptors.
1581 config BPF_JIT_ALWAYS_ON
1582 bool "Permanently enable BPF JIT and remove BPF interpreter"
1583 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1585 Enables BPF JIT and removes BPF interpreter to avoid
1586 speculative execution of BPF instructions by the interpreter
1589 bool "Enable userfaultfd() system call"
1592 Enable the userfaultfd() system call that allows to intercept and
1593 handle page faults in userland.
1595 config ARCH_HAS_MEMBARRIER_CALLBACKS
1598 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1602 bool "Enable rseq() system call" if EXPERT
1604 depends on HAVE_RSEQ
1607 Enable the restartable sequences system call. It provides a
1608 user-space cache for the current CPU number value, which
1609 speeds up getting the current CPU number from user-space,
1610 as well as an ABI to speed up user-space operations on
1617 bool "Enabled debugging of rseq() system call" if EXPERT
1618 depends on RSEQ && DEBUG_KERNEL
1620 Enable extra debugging checks for the rseq system call.
1625 bool "Embedded system"
1626 option allnoconfig_y
1629 This option should be enabled if compiling the kernel for
1630 an embedded system so certain expert options are available
1633 config HAVE_PERF_EVENTS
1636 See tools/perf/design.txt for details.
1638 config PERF_USE_VMALLOC
1641 See tools/perf/design.txt for details
1644 bool "PC/104 support" if EXPERT
1646 Expose PC/104 form factor device drivers and options available for
1647 selection and configuration. Enable this option if your target
1648 machine has a PC/104 bus.
1650 menu "Kernel Performance Events And Counters"
1653 bool "Kernel performance events and counters"
1654 default y if PROFILING
1655 depends on HAVE_PERF_EVENTS
1659 Enable kernel support for various performance events provided
1660 by software and hardware.
1662 Software events are supported either built-in or via the
1663 use of generic tracepoints.
1665 Most modern CPUs support performance events via performance
1666 counter registers. These registers count the number of certain
1667 types of hw events: such as instructions executed, cachemisses
1668 suffered, or branches mis-predicted - without slowing down the
1669 kernel or applications. These registers can also trigger interrupts
1670 when a threshold number of events have passed - and can thus be
1671 used to profile the code that runs on that CPU.
1673 The Linux Performance Event subsystem provides an abstraction of
1674 these software and hardware event capabilities, available via a
1675 system call and used by the "perf" utility in tools/perf/. It
1676 provides per task and per CPU counters, and it provides event
1677 capabilities on top of those.
1681 config DEBUG_PERF_USE_VMALLOC
1683 bool "Debug: use vmalloc to back perf mmap() buffers"
1684 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1685 select PERF_USE_VMALLOC
1687 Use vmalloc memory to back perf mmap() buffers.
1689 Mostly useful for debugging the vmalloc code on platforms
1690 that don't require it.
1696 config VM_EVENT_COUNTERS
1698 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1700 VM event counters are needed for event counts to be shown.
1701 This option allows the disabling of the VM event counters
1702 on EXPERT systems. /proc/vmstat will only show page counts
1703 if VM event counters are disabled.
1707 bool "Enable SLUB debugging support" if EXPERT
1708 depends on SLUB && SYSFS
1710 SLUB has extensive debug support features. Disabling these can
1711 result in significant savings in code size. This also disables
1712 SLUB sysfs support. /sys/slab will not exist and there will be
1713 no support for cache validation etc.
1715 config SLUB_MEMCG_SYSFS_ON
1717 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1718 depends on SLUB && SYSFS && MEMCG
1720 SLUB creates a directory under /sys/kernel/slab for each
1721 allocation cache to host info and debug files. If memory
1722 cgroup is enabled, each cache can have per memory cgroup
1723 caches. SLUB can create the same sysfs directories for these
1724 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1725 to a very high number of debug files being created. This is
1726 controlled by slub_memcg_sysfs boot parameter and this
1727 config option determines the parameter's default value.
1730 bool "Disable heap randomization"
1733 Randomizing heap placement makes heap exploits harder, but it
1734 also breaks ancient binaries (including anything libc5 based).
1735 This option changes the bootup default to heap randomization
1736 disabled, and can be overridden at runtime by setting
1737 /proc/sys/kernel/randomize_va_space to 2.
1739 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1742 prompt "Choose SLAB allocator"
1745 This option allows to select a slab allocator.
1749 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1751 The regular slab allocator that is established and known to work
1752 well in all environments. It organizes cache hot objects in
1753 per cpu and per node queues.
1756 bool "SLUB (Unqueued Allocator)"
1757 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1759 SLUB is a slab allocator that minimizes cache line usage
1760 instead of managing queues of cached objects (SLAB approach).
1761 Per cpu caching is realized using slabs of objects instead
1762 of queues of objects. SLUB can use memory efficiently
1763 and has enhanced diagnostics. SLUB is the default choice for
1768 bool "SLOB (Simple Allocator)"
1770 SLOB replaces the stock allocator with a drastically simpler
1771 allocator. SLOB is generally more space efficient but
1772 does not perform as well on large systems.
1776 config SLAB_MERGE_DEFAULT
1777 bool "Allow slab caches to be merged"
1780 For reduced kernel memory fragmentation, slab caches can be
1781 merged when they share the same size and other characteristics.
1782 This carries a risk of kernel heap overflows being able to
1783 overwrite objects from merged caches (and more easily control
1784 cache layout), which makes such heap attacks easier to exploit
1785 by attackers. By keeping caches unmerged, these kinds of exploits
1786 can usually only damage objects in the same cache. To disable
1787 merging at runtime, "slab_nomerge" can be passed on the kernel
1790 config SLAB_FREELIST_RANDOM
1792 depends on SLAB || SLUB
1793 bool "SLAB freelist randomization"
1795 Randomizes the freelist order used on creating new pages. This
1796 security feature reduces the predictability of the kernel slab
1797 allocator against heap overflows.
1799 config SLAB_FREELIST_HARDENED
1800 bool "Harden slab freelist metadata"
1803 Many kernel heap attacks try to target slab cache metadata and
1804 other infrastructure. This options makes minor performance
1805 sacrifies to harden the kernel slab allocator against common
1806 freelist exploit methods.
1808 config SHUFFLE_PAGE_ALLOCATOR
1809 bool "Page allocator randomization"
1810 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1812 Randomization of the page allocator improves the average
1813 utilization of a direct-mapped memory-side-cache. See section
1814 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1815 6.2a specification for an example of how a platform advertises
1816 the presence of a memory-side-cache. There are also incidental
1817 security benefits as it reduces the predictability of page
1818 allocations to compliment SLAB_FREELIST_RANDOM, but the
1819 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1820 10th order of pages is selected based on cache utilization
1823 While the randomization improves cache utilization it may
1824 negatively impact workloads on platforms without a cache. For
1825 this reason, by default, the randomization is enabled only
1826 after runtime detection of a direct-mapped memory-side-cache.
1827 Otherwise, the randomization may be force enabled with the
1828 'page_alloc.shuffle' kernel command line parameter.
1832 config SLUB_CPU_PARTIAL
1834 depends on SLUB && SMP
1835 bool "SLUB per cpu partial cache"
1837 Per cpu partial caches accellerate objects allocation and freeing
1838 that is local to a processor at the price of more indeterminism
1839 in the latency of the free. On overflow these caches will be cleared
1840 which requires the taking of locks that may cause latency spikes.
1841 Typically one would choose no for a realtime system.
1843 config MMAP_ALLOW_UNINITIALIZED
1844 bool "Allow mmapped anonymous memory to be uninitialized"
1845 depends on EXPERT && !MMU
1848 Normally, and according to the Linux spec, anonymous memory obtained
1849 from mmap() has its contents cleared before it is passed to
1850 userspace. Enabling this config option allows you to request that
1851 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1852 providing a huge performance boost. If this option is not enabled,
1853 then the flag will be ignored.
1855 This is taken advantage of by uClibc's malloc(), and also by
1856 ELF-FDPIC binfmt's brk and stack allocator.
1858 Because of the obvious security issues, this option should only be
1859 enabled on embedded devices where you control what is run in
1860 userspace. Since that isn't generally a problem on no-MMU systems,
1861 it is normally safe to say Y here.
1863 See Documentation/nommu-mmap.txt for more information.
1865 config SYSTEM_DATA_VERIFICATION
1867 select SYSTEM_TRUSTED_KEYRING
1871 select ASYMMETRIC_KEY_TYPE
1872 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1875 select X509_CERTIFICATE_PARSER
1876 select PKCS7_MESSAGE_PARSER
1878 Provide PKCS#7 message verification using the contents of the system
1879 trusted keyring to provide public keys. This then can be used for
1880 module verification, kexec image verification and firmware blob
1884 bool "Profiling support"
1886 Say Y here to enable the extended profiling support mechanisms used
1887 by profilers such as OProfile.
1890 # Place an empty function call at each tracepoint site. Can be
1891 # dynamically changed for a probe function.
1896 endmenu # General setup
1898 source "arch/Kconfig"
1905 default 0 if BASE_FULL
1906 default 1 if !BASE_FULL
1909 bool "Enable loadable module support"
1912 Kernel modules are small pieces of compiled code which can
1913 be inserted in the running kernel, rather than being
1914 permanently built into the kernel. You use the "modprobe"
1915 tool to add (and sometimes remove) them. If you say Y here,
1916 many parts of the kernel can be built as modules (by
1917 answering M instead of Y where indicated): this is most
1918 useful for infrequently used options which are not required
1919 for booting. For more information, see the man pages for
1920 modprobe, lsmod, modinfo, insmod and rmmod.
1922 If you say Y here, you will need to run "make
1923 modules_install" to put the modules under /lib/modules/
1924 where modprobe can find them (you may need to be root to do
1931 config MODULE_FORCE_LOAD
1932 bool "Forced module loading"
1935 Allow loading of modules without version information (ie. modprobe
1936 --force). Forced module loading sets the 'F' (forced) taint flag and
1937 is usually a really bad idea.
1939 config MODULE_UNLOAD
1940 bool "Module unloading"
1942 Without this option you will not be able to unload any
1943 modules (note that some modules may not be unloadable
1944 anyway), which makes your kernel smaller, faster
1945 and simpler. If unsure, say Y.
1947 config MODULE_FORCE_UNLOAD
1948 bool "Forced module unloading"
1949 depends on MODULE_UNLOAD
1951 This option allows you to force a module to unload, even if the
1952 kernel believes it is unsafe: the kernel will remove the module
1953 without waiting for anyone to stop using it (using the -f option to
1954 rmmod). This is mainly for kernel developers and desperate users.
1958 bool "Module versioning support"
1960 Usually, you have to use modules compiled with your kernel.
1961 Saying Y here makes it sometimes possible to use modules
1962 compiled for different kernels, by adding enough information
1963 to the modules to (hopefully) spot any changes which would
1964 make them incompatible with the kernel you are running. If
1967 config MODULE_REL_CRCS
1969 depends on MODVERSIONS
1971 config MODULE_SRCVERSION_ALL
1972 bool "Source checksum for all modules"
1974 Modules which contain a MODULE_VERSION get an extra "srcversion"
1975 field inserted into their modinfo section, which contains a
1976 sum of the source files which made it. This helps maintainers
1977 see exactly which source was used to build a module (since
1978 others sometimes change the module source without updating
1979 the version). With this option, such a "srcversion" field
1980 will be created for all modules. If unsure, say N.
1983 bool "Module signature verification"
1985 select SYSTEM_DATA_VERIFICATION
1987 Check modules for valid signatures upon load: the signature
1988 is simply appended to the module. For more information see
1989 <file:Documentation/admin-guide/module-signing.rst>.
1991 Note that this option adds the OpenSSL development packages as a
1992 kernel build dependency so that the signing tool can use its crypto
1995 !!!WARNING!!! If you enable this option, you MUST make sure that the
1996 module DOES NOT get stripped after being signed. This includes the
1997 debuginfo strip done by some packagers (such as rpmbuild) and
1998 inclusion into an initramfs that wants the module size reduced.
2000 config MODULE_SIG_FORCE
2001 bool "Require modules to be validly signed"
2002 depends on MODULE_SIG
2004 Reject unsigned modules or signed modules for which we don't have a
2005 key. Without this, such modules will simply taint the kernel.
2007 config MODULE_SIG_ALL
2008 bool "Automatically sign all modules"
2010 depends on MODULE_SIG
2012 Sign all modules during make modules_install. Without this option,
2013 modules must be signed manually, using the scripts/sign-file tool.
2015 comment "Do not forget to sign required modules with scripts/sign-file"
2016 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2019 prompt "Which hash algorithm should modules be signed with?"
2020 depends on MODULE_SIG
2022 This determines which sort of hashing algorithm will be used during
2023 signature generation. This algorithm _must_ be built into the kernel
2024 directly so that signature verification can take place. It is not
2025 possible to load a signed module containing the algorithm to check
2026 the signature on that module.
2028 config MODULE_SIG_SHA1
2029 bool "Sign modules with SHA-1"
2032 config MODULE_SIG_SHA224
2033 bool "Sign modules with SHA-224"
2034 select CRYPTO_SHA256
2036 config MODULE_SIG_SHA256
2037 bool "Sign modules with SHA-256"
2038 select CRYPTO_SHA256
2040 config MODULE_SIG_SHA384
2041 bool "Sign modules with SHA-384"
2042 select CRYPTO_SHA512
2044 config MODULE_SIG_SHA512
2045 bool "Sign modules with SHA-512"
2046 select CRYPTO_SHA512
2050 config MODULE_SIG_HASH
2052 depends on MODULE_SIG
2053 default "sha1" if MODULE_SIG_SHA1
2054 default "sha224" if MODULE_SIG_SHA224
2055 default "sha256" if MODULE_SIG_SHA256
2056 default "sha384" if MODULE_SIG_SHA384
2057 default "sha512" if MODULE_SIG_SHA512
2059 config MODULE_COMPRESS
2060 bool "Compress modules on installation"
2064 Compresses kernel modules when 'make modules_install' is run; gzip or
2065 xz depending on "Compression algorithm" below.
2067 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2069 Out-of-tree kernel modules installed using Kbuild will also be
2070 compressed upon installation.
2072 Note: for modules inside an initrd or initramfs, it's more efficient
2073 to compress the whole initrd or initramfs instead.
2075 Note: This is fully compatible with signed modules.
2080 prompt "Compression algorithm"
2081 depends on MODULE_COMPRESS
2082 default MODULE_COMPRESS_GZIP
2084 This determines which sort of compression will be used during
2085 'make modules_install'.
2087 GZIP (default) and XZ are supported.
2089 config MODULE_COMPRESS_GZIP
2092 config MODULE_COMPRESS_XZ
2097 config TRIM_UNUSED_KSYMS
2098 bool "Trim unused exported kernel symbols"
2099 depends on MODULES && !UNUSED_SYMBOLS
2101 The kernel and some modules make many symbols available for
2102 other modules to use via EXPORT_SYMBOL() and variants. Depending
2103 on the set of modules being selected in your kernel configuration,
2104 many of those exported symbols might never be used.
2106 This option allows for unused exported symbols to be dropped from
2107 the build. In turn, this provides the compiler more opportunities
2108 (especially when using LTO) for optimizing the code and reducing
2109 binary size. This might have some security advantages as well.
2111 If unsure, or if you need to build out-of-tree modules, say N.
2115 config MODULES_TREE_LOOKUP
2117 depends on PERF_EVENTS || TRACING
2119 config INIT_ALL_POSSIBLE
2122 Back when each arch used to define their own cpu_online_mask and
2123 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2124 with all 1s, and others with all 0s. When they were centralised,
2125 it was better to provide this option than to break all the archs
2126 and have several arch maintainers pursuing me down dark alleys.
2128 source "block/Kconfig"
2130 config PREEMPT_NOTIFIERS
2140 Build a simple ASN.1 grammar compiler that produces a bytecode output
2141 that can be interpreted by the ASN.1 stream decoder and used to
2142 inform it as to what tags are to be expected in a stream and what
2143 functions to call on what tags.
2145 source "kernel/Kconfig.locks"
2147 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2150 # It may be useful for an architecture to override the definitions of the
2151 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2152 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2153 # different calling convention for syscalls. They can also override the
2154 # macros for not-implemented syscalls in kernel/sys_ni.c and
2155 # kernel/time/posix-stubs.c. All these overrides need to be available in
2156 # <asm/syscall_wrapper.h>.
2157 config ARCH_HAS_SYSCALL_WRAPPER