5 default "/lib/modules/$(shell,uname -r)/.config"
6 default "/etc/kernel-config"
7 default "/boot/config-$(shell,uname -r)"
9 default "arch/$(ARCH)/defconfig"
12 def_bool $(success,$(CC) --version | head -n 1 | grep -q gcc)
16 default $(shell,$(srctree)/scripts/gcc-version.sh $(CC)) if CC_IS_GCC
20 def_bool $(success,$(CC) --version | head -n 1 | grep -q clang)
24 default $(shell,$(srctree)/scripts/clang-version.sh $(CC))
26 config CC_HAS_ASM_GOTO
27 def_bool $(success,$(srctree)/scripts/gcc-goto.sh $(CC))
29 config CC_HAS_WARN_MAYBE_UNINITIALIZED
30 def_bool $(cc-option,-Wmaybe-uninitialized)
32 GCC >= 4.7 supports this option.
34 config CC_DISABLE_WARN_MAYBE_UNINITIALIZED
36 depends on CC_HAS_WARN_MAYBE_UNINITIALIZED
37 default CC_IS_GCC && GCC_VERSION < 40900 # unreliable for GCC < 4.9
39 GCC's -Wmaybe-uninitialized is not reliable by definition.
40 Lots of false positive warnings are produced in some cases.
42 If this option is enabled, -Wno-maybe-uninitialzed is passed
43 to the compiler to suppress maybe-uninitialized warnings.
52 config BUILDTIME_EXTABLE_SORT
55 config THREAD_INFO_IN_TASK
58 Select this to move thread_info off the stack into task_struct. To
59 make this work, an arch will need to remove all thread_info fields
60 except flags and fix any runtime bugs.
62 One subtle change that will be needed is to use try_get_task_stack()
63 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
72 depends on BROKEN || !SMP
75 config INIT_ENV_ARG_LIMIT
80 Maximum of each of the number of arguments and environment
81 variables passed to init from the kernel command line.
84 bool "Compile also drivers which will not load"
88 Some drivers can be compiled on a different platform than they are
89 intended to be run on. Despite they cannot be loaded there (or even
90 when they load they cannot be used due to missing HW support),
91 developers still, opposing to distributors, might want to build such
92 drivers to compile-test them.
94 If you are a developer and want to build everything available, say Y
95 here. If you are a user/distributor, say N here to exclude useless
96 drivers to be distributed.
99 string "Local version - append to kernel release"
101 Append an extra string to the end of your kernel version.
102 This will show up when you type uname, for example.
103 The string you set here will be appended after the contents of
104 any files with a filename matching localversion* in your
105 object and source tree, in that order. Your total string can
106 be a maximum of 64 characters.
108 config LOCALVERSION_AUTO
109 bool "Automatically append version information to the version string"
111 depends on !COMPILE_TEST
113 This will try to automatically determine if the current tree is a
114 release tree by looking for git tags that belong to the current
115 top of tree revision.
117 A string of the format -gxxxxxxxx will be added to the localversion
118 if a git-based tree is found. The string generated by this will be
119 appended after any matching localversion* files, and after the value
120 set in CONFIG_LOCALVERSION.
122 (The actual string used here is the first eight characters produced
123 by running the command:
125 $ git rev-parse --verify HEAD
127 which is done within the script "scripts/setlocalversion".)
130 string "Build ID Salt"
133 The build ID is used to link binaries and their debug info. Setting
134 this option will use the value in the calculation of the build id.
135 This is mostly useful for distributions which want to ensure the
136 build is unique between builds. It's safe to leave the default.
138 config HAVE_KERNEL_GZIP
141 config HAVE_KERNEL_BZIP2
144 config HAVE_KERNEL_LZMA
147 config HAVE_KERNEL_XZ
150 config HAVE_KERNEL_LZO
153 config HAVE_KERNEL_LZ4
156 config HAVE_KERNEL_UNCOMPRESSED
160 prompt "Kernel compression mode"
162 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
164 The linux kernel is a kind of self-extracting executable.
165 Several compression algorithms are available, which differ
166 in efficiency, compression and decompression speed.
167 Compression speed is only relevant when building a kernel.
168 Decompression speed is relevant at each boot.
170 If you have any problems with bzip2 or lzma compressed
171 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
172 version of this functionality (bzip2 only), for 2.4, was
173 supplied by Christian Ludwig)
175 High compression options are mostly useful for users, who
176 are low on disk space (embedded systems), but for whom ram
179 If in doubt, select 'gzip'
183 depends on HAVE_KERNEL_GZIP
185 The old and tried gzip compression. It provides a good balance
186 between compression ratio and decompression speed.
190 depends on HAVE_KERNEL_BZIP2
192 Its compression ratio and speed is intermediate.
193 Decompression speed is slowest among the choices. The kernel
194 size is about 10% smaller with bzip2, in comparison to gzip.
195 Bzip2 uses a large amount of memory. For modern kernels you
196 will need at least 8MB RAM or more for booting.
200 depends on HAVE_KERNEL_LZMA
202 This compression algorithm's ratio is best. Decompression speed
203 is between gzip and bzip2. Compression is slowest.
204 The kernel size is about 33% smaller with LZMA in comparison to gzip.
208 depends on HAVE_KERNEL_XZ
210 XZ uses the LZMA2 algorithm and instruction set specific
211 BCJ filters which can improve compression ratio of executable
212 code. The size of the kernel is about 30% smaller with XZ in
213 comparison to gzip. On architectures for which there is a BCJ
214 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
215 will create a few percent smaller kernel than plain LZMA.
217 The speed is about the same as with LZMA: The decompression
218 speed of XZ is better than that of bzip2 but worse than gzip
219 and LZO. Compression is slow.
223 depends on HAVE_KERNEL_LZO
225 Its compression ratio is the poorest among the choices. The kernel
226 size is about 10% bigger than gzip; however its speed
227 (both compression and decompression) is the fastest.
231 depends on HAVE_KERNEL_LZ4
233 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
234 A preliminary version of LZ4 de/compression tool is available at
235 <https://code.google.com/p/lz4/>.
237 Its compression ratio is worse than LZO. The size of the kernel
238 is about 8% bigger than LZO. But the decompression speed is
241 config KERNEL_UNCOMPRESSED
243 depends on HAVE_KERNEL_UNCOMPRESSED
245 Produce uncompressed kernel image. This option is usually not what
246 you want. It is useful for debugging the kernel in slow simulation
247 environments, where decompressing and moving the kernel is awfully
248 slow. This option allows early boot code to skip the decompressor
249 and jump right at uncompressed kernel image.
253 config DEFAULT_HOSTNAME
254 string "Default hostname"
257 This option determines the default system hostname before userspace
258 calls sethostname(2). The kernel traditionally uses "(none)" here,
259 but you may wish to use a different default here to make a minimal
260 system more usable with less configuration.
263 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
264 # add proper SWAP support to them, in which case this can be remove.
270 bool "Support for paging of anonymous memory (swap)"
271 depends on MMU && BLOCK && !ARCH_NO_SWAP
274 This option allows you to choose whether you want to have support
275 for so called swap devices or swap files in your kernel that are
276 used to provide more virtual memory than the actual RAM present
277 in your computer. If unsure say Y.
282 Inter Process Communication is a suite of library functions and
283 system calls which let processes (running programs) synchronize and
284 exchange information. It is generally considered to be a good thing,
285 and some programs won't run unless you say Y here. In particular, if
286 you want to run the DOS emulator dosemu under Linux (read the
287 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
288 you'll need to say Y here.
290 You can find documentation about IPC with "info ipc" and also in
291 section 6.4 of the Linux Programmer's Guide, available from
292 <http://www.tldp.org/guides.html>.
294 config SYSVIPC_SYSCTL
301 bool "POSIX Message Queues"
304 POSIX variant of message queues is a part of IPC. In POSIX message
305 queues every message has a priority which decides about succession
306 of receiving it by a process. If you want to compile and run
307 programs written e.g. for Solaris with use of its POSIX message
308 queues (functions mq_*) say Y here.
310 POSIX message queues are visible as a filesystem called 'mqueue'
311 and can be mounted somewhere if you want to do filesystem
312 operations on message queues.
316 config POSIX_MQUEUE_SYSCTL
318 depends on POSIX_MQUEUE
322 config CROSS_MEMORY_ATTACH
323 bool "Enable process_vm_readv/writev syscalls"
327 Enabling this option adds the system calls process_vm_readv and
328 process_vm_writev which allow a process with the correct privileges
329 to directly read from or write to another process' address space.
330 See the man page for more details.
333 bool "uselib syscall"
334 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
336 This option enables the uselib syscall, a system call used in the
337 dynamic linker from libc5 and earlier. glibc does not use this
338 system call. If you intend to run programs built on libc5 or
339 earlier, you may need to enable this syscall. Current systems
340 running glibc can safely disable this.
343 bool "Auditing support"
346 Enable auditing infrastructure that can be used with another
347 kernel subsystem, such as SELinux (which requires this for
348 logging of avc messages output). System call auditing is included
349 on architectures which support it.
351 config HAVE_ARCH_AUDITSYSCALL
356 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
359 source "kernel/irq/Kconfig"
360 source "kernel/time/Kconfig"
361 source "kernel/Kconfig.preempt"
363 menu "CPU/Task time and stats accounting"
365 config VIRT_CPU_ACCOUNTING
369 prompt "Cputime accounting"
370 default TICK_CPU_ACCOUNTING if !PPC64
371 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
373 # Kind of a stub config for the pure tick based cputime accounting
374 config TICK_CPU_ACCOUNTING
375 bool "Simple tick based cputime accounting"
376 depends on !S390 && !NO_HZ_FULL
378 This is the basic tick based cputime accounting that maintains
379 statistics about user, system and idle time spent on per jiffies
384 config VIRT_CPU_ACCOUNTING_NATIVE
385 bool "Deterministic task and CPU time accounting"
386 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
387 select VIRT_CPU_ACCOUNTING
389 Select this option to enable more accurate task and CPU time
390 accounting. This is done by reading a CPU counter on each
391 kernel entry and exit and on transitions within the kernel
392 between system, softirq and hardirq state, so there is a
393 small performance impact. In the case of s390 or IBM POWER > 5,
394 this also enables accounting of stolen time on logically-partitioned
397 config VIRT_CPU_ACCOUNTING_GEN
398 bool "Full dynticks CPU time accounting"
399 depends on HAVE_CONTEXT_TRACKING
400 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
401 depends on GENERIC_CLOCKEVENTS
402 select VIRT_CPU_ACCOUNTING
403 select CONTEXT_TRACKING
405 Select this option to enable task and CPU time accounting on full
406 dynticks systems. This accounting is implemented by watching every
407 kernel-user boundaries using the context tracking subsystem.
408 The accounting is thus performed at the expense of some significant
411 For now this is only useful if you are working on the full
412 dynticks subsystem development.
418 config IRQ_TIME_ACCOUNTING
419 bool "Fine granularity task level IRQ time accounting"
420 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
422 Select this option to enable fine granularity task irq time
423 accounting. This is done by reading a timestamp on each
424 transitions between softirq and hardirq state, so there can be a
425 small performance impact.
427 If in doubt, say N here.
429 config HAVE_SCHED_AVG_IRQ
431 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
434 config BSD_PROCESS_ACCT
435 bool "BSD Process Accounting"
438 If you say Y here, a user level program will be able to instruct the
439 kernel (via a special system call) to write process accounting
440 information to a file: whenever a process exits, information about
441 that process will be appended to the file by the kernel. The
442 information includes things such as creation time, owning user,
443 command name, memory usage, controlling terminal etc. (the complete
444 list is in the struct acct in <file:include/linux/acct.h>). It is
445 up to the user level program to do useful things with this
446 information. This is generally a good idea, so say Y.
448 config BSD_PROCESS_ACCT_V3
449 bool "BSD Process Accounting version 3 file format"
450 depends on BSD_PROCESS_ACCT
453 If you say Y here, the process accounting information is written
454 in a new file format that also logs the process IDs of each
455 process and its parent. Note that this file format is incompatible
456 with previous v0/v1/v2 file formats, so you will need updated tools
457 for processing it. A preliminary version of these tools is available
458 at <http://www.gnu.org/software/acct/>.
461 bool "Export task/process statistics through netlink"
466 Export selected statistics for tasks/processes through the
467 generic netlink interface. Unlike BSD process accounting, the
468 statistics are available during the lifetime of tasks/processes as
469 responses to commands. Like BSD accounting, they are sent to user
474 config TASK_DELAY_ACCT
475 bool "Enable per-task delay accounting"
479 Collect information on time spent by a task waiting for system
480 resources like cpu, synchronous block I/O completion and swapping
481 in pages. Such statistics can help in setting a task's priorities
482 relative to other tasks for cpu, io, rss limits etc.
487 bool "Enable extended accounting over taskstats"
490 Collect extended task accounting data and send the data
491 to userland for processing over the taskstats interface.
495 config TASK_IO_ACCOUNTING
496 bool "Enable per-task storage I/O accounting"
497 depends on TASK_XACCT
499 Collect information on the number of bytes of storage I/O which this
505 bool "Pressure stall information tracking"
507 Collect metrics that indicate how overcommitted the CPU, memory,
508 and IO capacity are in the system.
510 If you say Y here, the kernel will create /proc/pressure/ with the
511 pressure statistics files cpu, memory, and io. These will indicate
512 the share of walltime in which some or all tasks in the system are
513 delayed due to contention of the respective resource.
515 In kernels with cgroup support, cgroups (cgroup2 only) will
516 have cpu.pressure, memory.pressure, and io.pressure files,
517 which aggregate pressure stalls for the grouped tasks only.
519 For more details see Documentation/accounting/psi.txt.
523 config PSI_DEFAULT_DISABLED
524 bool "Require boot parameter to enable pressure stall information tracking"
528 If set, pressure stall information tracking will be disabled
529 per default but can be enabled through passing psi=1 on the
530 kernel commandline during boot.
532 This feature adds some code to the task wakeup and sleep
533 paths of the scheduler. The overhead is too low to affect
534 common scheduling-intense workloads in practice (such as
535 webservers, memcache), but it does show up in artificial
536 scheduler stress tests, such as hackbench.
538 If you are paranoid and not sure what the kernel will be
543 endmenu # "CPU/Task time and stats accounting"
547 depends on SMP || COMPILE_TEST
550 Make sure that CPUs running critical tasks are not disturbed by
551 any source of "noise" such as unbound workqueues, timers, kthreads...
552 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
553 the "isolcpus=" boot parameter.
557 source "kernel/rcu/Kconfig"
564 tristate "Kernel .config support"
566 This option enables the complete Linux kernel ".config" file
567 contents to be saved in the kernel. It provides documentation
568 of which kernel options are used in a running kernel or in an
569 on-disk kernel. This information can be extracted from the kernel
570 image file with the script scripts/extract-ikconfig and used as
571 input to rebuild the current kernel or to build another kernel.
572 It can also be extracted from a running kernel by reading
573 /proc/config.gz if enabled (below).
576 bool "Enable access to .config through /proc/config.gz"
577 depends on IKCONFIG && PROC_FS
579 This option enables access to the kernel configuration file
580 through /proc/config.gz.
583 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
586 This option enables access to the in-kernel headers that are generated during
587 the build process. These can be used to build eBPF tracing programs,
588 or similar programs. If you build the headers as a module, a module called
589 kheaders.ko is built which can be loaded on-demand to get access to headers.
592 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
597 Select the minimal kernel log buffer size as a power of 2.
598 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
599 parameter, see below. Any higher size also might be forced
600 by "log_buf_len" boot parameter.
610 config LOG_CPU_MAX_BUF_SHIFT
611 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
614 default 12 if !BASE_SMALL
615 default 0 if BASE_SMALL
618 This option allows to increase the default ring buffer size
619 according to the number of CPUs. The value defines the contribution
620 of each CPU as a power of 2. The used space is typically only few
621 lines however it might be much more when problems are reported,
624 The increased size means that a new buffer has to be allocated and
625 the original static one is unused. It makes sense only on systems
626 with more CPUs. Therefore this value is used only when the sum of
627 contributions is greater than the half of the default kernel ring
628 buffer as defined by LOG_BUF_SHIFT. The default values are set
629 so that more than 64 CPUs are needed to trigger the allocation.
631 Also this option is ignored when "log_buf_len" kernel parameter is
632 used as it forces an exact (power of two) size of the ring buffer.
634 The number of possible CPUs is used for this computation ignoring
635 hotplugging making the computation optimal for the worst case
636 scenario while allowing a simple algorithm to be used from bootup.
638 Examples shift values and their meaning:
639 17 => 128 KB for each CPU
640 16 => 64 KB for each CPU
641 15 => 32 KB for each CPU
642 14 => 16 KB for each CPU
643 13 => 8 KB for each CPU
644 12 => 4 KB for each CPU
646 config PRINTK_SAFE_LOG_BUF_SHIFT
647 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
652 Select the size of an alternate printk per-CPU buffer where messages
653 printed from usafe contexts are temporary stored. One example would
654 be NMI messages, another one - printk recursion. The messages are
655 copied to the main log buffer in a safe context to avoid a deadlock.
656 The value defines the size as a power of 2.
658 Those messages are rare and limited. The largest one is when
659 a backtrace is printed. It usually fits into 4KB. Select
660 8KB if you want to be on the safe side.
663 17 => 128 KB for each CPU
664 16 => 64 KB for each CPU
665 15 => 32 KB for each CPU
666 14 => 16 KB for each CPU
667 13 => 8 KB for each CPU
668 12 => 4 KB for each CPU
671 # Architectures with an unreliable sched_clock() should select this:
673 config HAVE_UNSTABLE_SCHED_CLOCK
676 config GENERIC_SCHED_CLOCK
680 # For architectures that want to enable the support for NUMA-affine scheduler
683 config ARCH_SUPPORTS_NUMA_BALANCING
687 # For architectures that prefer to flush all TLBs after a number of pages
688 # are unmapped instead of sending one IPI per page to flush. The architecture
689 # must provide guarantees on what happens if a clean TLB cache entry is
690 # written after the unmap. Details are in mm/rmap.c near the check for
691 # should_defer_flush. The architecture should also consider if the full flush
692 # and the refill costs are offset by the savings of sending fewer IPIs.
693 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
697 # For architectures that know their GCC __int128 support is sound
699 config ARCH_SUPPORTS_INT128
702 # For architectures that (ab)use NUMA to represent different memory regions
703 # all cpu-local but of different latencies, such as SuperH.
705 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
708 config NUMA_BALANCING
709 bool "Memory placement aware NUMA scheduler"
710 depends on ARCH_SUPPORTS_NUMA_BALANCING
711 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
712 depends on SMP && NUMA && MIGRATION
714 This option adds support for automatic NUMA aware memory/task placement.
715 The mechanism is quite primitive and is based on migrating memory when
716 it has references to the node the task is running on.
718 This system will be inactive on UMA systems.
720 config NUMA_BALANCING_DEFAULT_ENABLED
721 bool "Automatically enable NUMA aware memory/task placement"
723 depends on NUMA_BALANCING
725 If set, automatic NUMA balancing will be enabled if running on a NUMA
729 bool "Control Group support"
732 This option adds support for grouping sets of processes together, for
733 use with process control subsystems such as Cpusets, CFS, memory
734 controls or device isolation.
736 - Documentation/scheduler/sched-design-CFS.txt (CFS)
737 - Documentation/cgroup-v1/ (features for grouping, isolation
738 and resource control)
748 bool "Memory controller"
752 Provides control over the memory footprint of tasks in a cgroup.
755 bool "Swap controller"
756 depends on MEMCG && SWAP
758 Provides control over the swap space consumed by tasks in a cgroup.
760 config MEMCG_SWAP_ENABLED
761 bool "Swap controller enabled by default"
762 depends on MEMCG_SWAP
765 Memory Resource Controller Swap Extension comes with its price in
766 a bigger memory consumption. General purpose distribution kernels
767 which want to enable the feature but keep it disabled by default
768 and let the user enable it by swapaccount=1 boot command line
769 parameter should have this option unselected.
770 For those who want to have the feature enabled by default should
771 select this option (if, for some reason, they need to disable it
772 then swapaccount=0 does the trick).
776 depends on MEMCG && !SLOB
784 Generic block IO controller cgroup interface. This is the common
785 cgroup interface which should be used by various IO controlling
788 Currently, CFQ IO scheduler uses it to recognize task groups and
789 control disk bandwidth allocation (proportional time slice allocation)
790 to such task groups. It is also used by bio throttling logic in
791 block layer to implement upper limit in IO rates on a device.
793 This option only enables generic Block IO controller infrastructure.
794 One needs to also enable actual IO controlling logic/policy. For
795 enabling proportional weight division of disk bandwidth in CFQ, set
796 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
797 CONFIG_BLK_DEV_THROTTLING=y.
799 See Documentation/cgroup-v1/blkio-controller.txt for more information.
801 config DEBUG_BLK_CGROUP
802 bool "IO controller debugging"
803 depends on BLK_CGROUP
806 Enable some debugging help. Currently it exports additional stat
807 files in a cgroup which can be useful for debugging.
809 config CGROUP_WRITEBACK
811 depends on MEMCG && BLK_CGROUP
814 menuconfig CGROUP_SCHED
815 bool "CPU controller"
818 This feature lets CPU scheduler recognize task groups and control CPU
819 bandwidth allocation to such task groups. It uses cgroups to group
823 config FAIR_GROUP_SCHED
824 bool "Group scheduling for SCHED_OTHER"
825 depends on CGROUP_SCHED
829 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
830 depends on FAIR_GROUP_SCHED
833 This option allows users to define CPU bandwidth rates (limits) for
834 tasks running within the fair group scheduler. Groups with no limit
835 set are considered to be unconstrained and will run with no
837 See Documentation/scheduler/sched-bwc.txt for more information.
839 config RT_GROUP_SCHED
840 bool "Group scheduling for SCHED_RR/FIFO"
841 depends on CGROUP_SCHED
844 This feature lets you explicitly allocate real CPU bandwidth
845 to task groups. If enabled, it will also make it impossible to
846 schedule realtime tasks for non-root users until you allocate
847 realtime bandwidth for them.
848 See Documentation/scheduler/sched-rt-group.txt for more information.
853 bool "PIDs controller"
855 Provides enforcement of process number limits in the scope of a
856 cgroup. Any attempt to fork more processes than is allowed in the
857 cgroup will fail. PIDs are fundamentally a global resource because it
858 is fairly trivial to reach PID exhaustion before you reach even a
859 conservative kmemcg limit. As a result, it is possible to grind a
860 system to halt without being limited by other cgroup policies. The
861 PIDs controller is designed to stop this from happening.
863 It should be noted that organisational operations (such as attaching
864 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
865 since the PIDs limit only affects a process's ability to fork, not to
869 bool "RDMA controller"
871 Provides enforcement of RDMA resources defined by IB stack.
872 It is fairly easy for consumers to exhaust RDMA resources, which
873 can result into resource unavailability to other consumers.
874 RDMA controller is designed to stop this from happening.
875 Attaching processes with active RDMA resources to the cgroup
876 hierarchy is allowed even if can cross the hierarchy's limit.
878 config CGROUP_FREEZER
879 bool "Freezer controller"
881 Provides a way to freeze and unfreeze all tasks in a
884 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
885 controller includes important in-kernel memory consumers per default.
887 If you're using cgroup2, say N.
889 config CGROUP_HUGETLB
890 bool "HugeTLB controller"
891 depends on HUGETLB_PAGE
895 Provides a cgroup controller for HugeTLB pages.
896 When you enable this, you can put a per cgroup limit on HugeTLB usage.
897 The limit is enforced during page fault. Since HugeTLB doesn't
898 support page reclaim, enforcing the limit at page fault time implies
899 that, the application will get SIGBUS signal if it tries to access
900 HugeTLB pages beyond its limit. This requires the application to know
901 beforehand how much HugeTLB pages it would require for its use. The
902 control group is tracked in the third page lru pointer. This means
903 that we cannot use the controller with huge page less than 3 pages.
906 bool "Cpuset controller"
909 This option will let you create and manage CPUSETs which
910 allow dynamically partitioning a system into sets of CPUs and
911 Memory Nodes and assigning tasks to run only within those sets.
912 This is primarily useful on large SMP or NUMA systems.
916 config PROC_PID_CPUSET
917 bool "Include legacy /proc/<pid>/cpuset file"
922 bool "Device controller"
924 Provides a cgroup controller implementing whitelists for
925 devices which a process in the cgroup can mknod or open.
927 config CGROUP_CPUACCT
928 bool "Simple CPU accounting controller"
930 Provides a simple controller for monitoring the
931 total CPU consumed by the tasks in a cgroup.
934 bool "Perf controller"
935 depends on PERF_EVENTS
937 This option extends the perf per-cpu mode to restrict monitoring
938 to threads which belong to the cgroup specified and run on the
944 bool "Support for eBPF programs attached to cgroups"
945 depends on BPF_SYSCALL
946 select SOCK_CGROUP_DATA
948 Allow attaching eBPF programs to a cgroup using the bpf(2)
949 syscall command BPF_PROG_ATTACH.
951 In which context these programs are accessed depends on the type
952 of attachment. For instance, programs that are attached using
953 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
957 bool "Debug controller"
959 depends on DEBUG_KERNEL
961 This option enables a simple controller that exports
962 debugging information about the cgroups framework. This
963 controller is for control cgroup debugging only. Its
964 interfaces are not stable.
968 config SOCK_CGROUP_DATA
974 menuconfig NAMESPACES
975 bool "Namespaces support" if EXPERT
979 Provides the way to make tasks work with different objects using
980 the same id. For example same IPC id may refer to different objects
981 or same user id or pid may refer to different tasks when used in
982 different namespaces.
990 In this namespace tasks see different info provided with the
995 depends on (SYSVIPC || POSIX_MQUEUE)
998 In this namespace tasks work with IPC ids which correspond to
999 different IPC objects in different namespaces.
1002 bool "User namespace"
1005 This allows containers, i.e. vservers, to use user namespaces
1006 to provide different user info for different servers.
1008 When user namespaces are enabled in the kernel it is
1009 recommended that the MEMCG option also be enabled and that
1010 user-space use the memory control groups to limit the amount
1011 of memory a memory unprivileged users can use.
1016 bool "PID Namespaces"
1019 Support process id namespaces. This allows having multiple
1020 processes with the same pid as long as they are in different
1021 pid namespaces. This is a building block of containers.
1024 bool "Network namespace"
1028 Allow user space to create what appear to be multiple instances
1029 of the network stack.
1033 config CHECKPOINT_RESTORE
1034 bool "Checkpoint/restore support"
1035 select PROC_CHILDREN
1038 Enables additional kernel features in a sake of checkpoint/restore.
1039 In particular it adds auxiliary prctl codes to setup process text,
1040 data and heap segment sizes, and a few additional /proc filesystem
1043 If unsure, say N here.
1045 config SCHED_AUTOGROUP
1046 bool "Automatic process group scheduling"
1049 select FAIR_GROUP_SCHED
1051 This option optimizes the scheduler for common desktop workloads by
1052 automatically creating and populating task groups. This separation
1053 of workloads isolates aggressive CPU burners (like build jobs) from
1054 desktop applications. Task group autogeneration is currently based
1057 config SYSFS_DEPRECATED
1058 bool "Enable deprecated sysfs features to support old userspace tools"
1062 This option adds code that switches the layout of the "block" class
1063 devices, to not show up in /sys/class/block/, but only in
1066 This switch is only active when the sysfs.deprecated=1 boot option is
1067 passed or the SYSFS_DEPRECATED_V2 option is set.
1069 This option allows new kernels to run on old distributions and tools,
1070 which might get confused by /sys/class/block/. Since 2007/2008 all
1071 major distributions and tools handle this just fine.
1073 Recent distributions and userspace tools after 2009/2010 depend on
1074 the existence of /sys/class/block/, and will not work with this
1077 Only if you are using a new kernel on an old distribution, you might
1080 config SYSFS_DEPRECATED_V2
1081 bool "Enable deprecated sysfs features by default"
1084 depends on SYSFS_DEPRECATED
1086 Enable deprecated sysfs by default.
1088 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1091 Only if you are using a new kernel on an old distribution, you might
1092 need to say Y here. Even then, odds are you would not need it
1093 enabled, you can always pass the boot option if absolutely necessary.
1096 bool "Kernel->user space relay support (formerly relayfs)"
1099 This option enables support for relay interface support in
1100 certain file systems (such as debugfs).
1101 It is designed to provide an efficient mechanism for tools and
1102 facilities to relay large amounts of data from kernel space to
1107 config BLK_DEV_INITRD
1108 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1110 The initial RAM filesystem is a ramfs which is loaded by the
1111 boot loader (loadlin or lilo) and that is mounted as root
1112 before the normal boot procedure. It is typically used to
1113 load modules needed to mount the "real" root file system,
1114 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1116 If RAM disk support (BLK_DEV_RAM) is also included, this
1117 also enables initial RAM disk (initrd) support and adds
1118 15 Kbytes (more on some other architectures) to the kernel size.
1124 source "usr/Kconfig"
1129 prompt "Compiler optimization level"
1130 default CC_OPTIMIZE_FOR_PERFORMANCE
1132 config CC_OPTIMIZE_FOR_PERFORMANCE
1133 bool "Optimize for performance"
1135 This is the default optimization level for the kernel, building
1136 with the "-O2" compiler flag for best performance and most
1137 helpful compile-time warnings.
1139 config CC_OPTIMIZE_FOR_SIZE
1140 bool "Optimize for size"
1141 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1143 Enabling this option will pass "-Os" instead of "-O2" to
1144 your compiler resulting in a smaller kernel.
1150 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1153 This requires that the arch annotates or otherwise protects
1154 its external entry points from being discarded. Linker scripts
1155 must also merge .text.*, .data.*, and .bss.* correctly into
1156 output sections. Care must be taken not to pull in unrelated
1157 sections (e.g., '.text.init'). Typically '.' in section names
1158 is used to distinguish them from label names / C identifiers.
1160 config LD_DEAD_CODE_DATA_ELIMINATION
1161 bool "Dead code and data elimination (EXPERIMENTAL)"
1162 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1164 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1165 depends on $(cc-option,-ffunction-sections -fdata-sections)
1166 depends on $(ld-option,--gc-sections)
1168 Enable this if you want to do dead code and data elimination with
1169 the linker by compiling with -ffunction-sections -fdata-sections,
1170 and linking with --gc-sections.
1172 This can reduce on disk and in-memory size of the kernel
1173 code and static data, particularly for small configs and
1174 on small systems. This has the possibility of introducing
1175 silently broken kernel if the required annotations are not
1176 present. This option is not well tested yet, so use at your
1185 config SYSCTL_EXCEPTION_TRACE
1188 Enable support for /proc/sys/debug/exception-trace.
1190 config SYSCTL_ARCH_UNALIGN_NO_WARN
1193 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1194 Allows arch to define/use @no_unaligned_warning to possibly warn
1195 about unaligned access emulation going on under the hood.
1197 config SYSCTL_ARCH_UNALIGN_ALLOW
1200 Enable support for /proc/sys/kernel/unaligned-trap
1201 Allows arches to define/use @unaligned_enabled to runtime toggle
1202 the unaligned access emulation.
1203 see arch/parisc/kernel/unaligned.c for reference
1205 config HAVE_PCSPKR_PLATFORM
1208 # interpreter that classic socket filters depend on
1213 bool "Configure standard kernel features (expert users)"
1214 # Unhide debug options, to make the on-by-default options visible
1217 This option allows certain base kernel options and settings
1218 to be disabled or tweaked. This is for specialized
1219 environments which can tolerate a "non-standard" kernel.
1220 Only use this if you really know what you are doing.
1223 bool "Enable 16-bit UID system calls" if EXPERT
1224 depends on HAVE_UID16 && MULTIUSER
1227 This enables the legacy 16-bit UID syscall wrappers.
1230 bool "Multiple users, groups and capabilities support" if EXPERT
1233 This option enables support for non-root users, groups and
1236 If you say N here, all processes will run with UID 0, GID 0, and all
1237 possible capabilities. Saying N here also compiles out support for
1238 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1241 If unsure, say Y here.
1243 config SGETMASK_SYSCALL
1244 bool "sgetmask/ssetmask syscalls support" if EXPERT
1245 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1247 sys_sgetmask and sys_ssetmask are obsolete system calls
1248 no longer supported in libc but still enabled by default in some
1251 If unsure, leave the default option here.
1253 config SYSFS_SYSCALL
1254 bool "Sysfs syscall support" if EXPERT
1257 sys_sysfs is an obsolete system call no longer supported in libc.
1258 Note that disabling this option is more secure but might break
1259 compatibility with some systems.
1261 If unsure say Y here.
1263 config SYSCTL_SYSCALL
1264 bool "Sysctl syscall support" if EXPERT
1265 depends on PROC_SYSCTL
1269 sys_sysctl uses binary paths that have been found challenging
1270 to properly maintain and use. The interface in /proc/sys
1271 using paths with ascii names is now the primary path to this
1274 Almost nothing using the binary sysctl interface so if you are
1275 trying to save some space it is probably safe to disable this,
1276 making your kernel marginally smaller.
1278 If unsure say N here.
1281 bool "open by fhandle syscalls" if EXPERT
1285 If you say Y here, a user level program will be able to map
1286 file names to handle and then later use the handle for
1287 different file system operations. This is useful in implementing
1288 userspace file servers, which now track files using handles instead
1289 of names. The handle would remain the same even if file names
1290 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1294 bool "Posix Clocks & timers" if EXPERT
1297 This includes native support for POSIX timers to the kernel.
1298 Some embedded systems have no use for them and therefore they
1299 can be configured out to reduce the size of the kernel image.
1301 When this option is disabled, the following syscalls won't be
1302 available: timer_create, timer_gettime: timer_getoverrun,
1303 timer_settime, timer_delete, clock_adjtime, getitimer,
1304 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1305 clock_getres and clock_nanosleep syscalls will be limited to
1306 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1312 bool "Enable support for printk" if EXPERT
1315 This option enables normal printk support. Removing it
1316 eliminates most of the message strings from the kernel image
1317 and makes the kernel more or less silent. As this makes it
1318 very difficult to diagnose system problems, saying N here is
1319 strongly discouraged.
1327 bool "BUG() support" if EXPERT
1330 Disabling this option eliminates support for BUG and WARN, reducing
1331 the size of your kernel image and potentially quietly ignoring
1332 numerous fatal conditions. You should only consider disabling this
1333 option for embedded systems with no facilities for reporting errors.
1339 bool "Enable ELF core dumps" if EXPERT
1341 Enable support for generating core dumps. Disabling saves about 4k.
1344 config PCSPKR_PLATFORM
1345 bool "Enable PC-Speaker support" if EXPERT
1346 depends on HAVE_PCSPKR_PLATFORM
1350 This option allows to disable the internal PC-Speaker
1351 support, saving some memory.
1355 bool "Enable full-sized data structures for core" if EXPERT
1357 Disabling this option reduces the size of miscellaneous core
1358 kernel data structures. This saves memory on small machines,
1359 but may reduce performance.
1362 bool "Enable futex support" if EXPERT
1366 Disabling this option will cause the kernel to be built without
1367 support for "fast userspace mutexes". The resulting kernel may not
1368 run glibc-based applications correctly.
1372 depends on FUTEX && RT_MUTEXES
1375 config HAVE_FUTEX_CMPXCHG
1379 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1380 is implemented and always working. This removes a couple of runtime
1384 bool "Enable eventpoll support" if EXPERT
1387 Disabling this option will cause the kernel to be built without
1388 support for epoll family of system calls.
1391 bool "Enable signalfd() system call" if EXPERT
1394 Enable the signalfd() system call that allows to receive signals
1395 on a file descriptor.
1400 bool "Enable timerfd() system call" if EXPERT
1403 Enable the timerfd() system call that allows to receive timer
1404 events on a file descriptor.
1409 bool "Enable eventfd() system call" if EXPERT
1412 Enable the eventfd() system call that allows to receive both
1413 kernel notification (ie. KAIO) or userspace notifications.
1418 bool "Use full shmem filesystem" if EXPERT
1422 The shmem is an internal filesystem used to manage shared memory.
1423 It is backed by swap and manages resource limits. It is also exported
1424 to userspace as tmpfs if TMPFS is enabled. Disabling this
1425 option replaces shmem and tmpfs with the much simpler ramfs code,
1426 which may be appropriate on small systems without swap.
1429 bool "Enable AIO support" if EXPERT
1432 This option enables POSIX asynchronous I/O which may by used
1433 by some high performance threaded applications. Disabling
1434 this option saves about 7k.
1437 bool "Enable IO uring support" if EXPERT
1441 This option enables support for the io_uring interface, enabling
1442 applications to submit and complete IO through submission and
1443 completion rings that are shared between the kernel and application.
1445 config ADVISE_SYSCALLS
1446 bool "Enable madvise/fadvise syscalls" if EXPERT
1449 This option enables the madvise and fadvise syscalls, used by
1450 applications to advise the kernel about their future memory or file
1451 usage, improving performance. If building an embedded system where no
1452 applications use these syscalls, you can disable this option to save
1456 bool "Enable membarrier() system call" if EXPERT
1459 Enable the membarrier() system call that allows issuing memory
1460 barriers across all running threads, which can be used to distribute
1461 the cost of user-space memory barriers asymmetrically by transforming
1462 pairs of memory barriers into pairs consisting of membarrier() and a
1468 bool "Load all symbols for debugging/ksymoops" if EXPERT
1471 Say Y here to let the kernel print out symbolic crash information and
1472 symbolic stack backtraces. This increases the size of the kernel
1473 somewhat, as all symbols have to be loaded into the kernel image.
1476 bool "Include all symbols in kallsyms"
1477 depends on DEBUG_KERNEL && KALLSYMS
1479 Normally kallsyms only contains the symbols of functions for nicer
1480 OOPS messages and backtraces (i.e., symbols from the text and inittext
1481 sections). This is sufficient for most cases. And only in very rare
1482 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1483 names of variables from the data sections, etc).
1485 This option makes sure that all symbols are loaded into the kernel
1486 image (i.e., symbols from all sections) in cost of increased kernel
1487 size (depending on the kernel configuration, it may be 300KiB or
1488 something like this).
1490 Say N unless you really need all symbols.
1492 config KALLSYMS_ABSOLUTE_PERCPU
1495 default X86_64 && SMP
1497 config KALLSYMS_BASE_RELATIVE
1502 Instead of emitting them as absolute values in the native word size,
1503 emit the symbol references in the kallsyms table as 32-bit entries,
1504 each containing a relative value in the range [base, base + U32_MAX]
1505 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1506 an absolute value in the range [0, S32_MAX] or a relative value in the
1507 range [base, base + S32_MAX], where base is the lowest relative symbol
1508 address encountered in the image.
1510 On 64-bit builds, this reduces the size of the address table by 50%,
1511 but more importantly, it results in entries whose values are build
1512 time constants, and no relocation pass is required at runtime to fix
1513 up the entries based on the runtime load address of the kernel.
1515 # end of the "standard kernel features (expert users)" menu
1517 # syscall, maps, verifier
1519 bool "Enable bpf() system call"
1524 Enable the bpf() system call that allows to manipulate eBPF
1525 programs and maps via file descriptors.
1527 config BPF_JIT_ALWAYS_ON
1528 bool "Permanently enable BPF JIT and remove BPF interpreter"
1529 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1531 Enables BPF JIT and removes BPF interpreter to avoid
1532 speculative execution of BPF instructions by the interpreter
1535 bool "Enable userfaultfd() system call"
1538 Enable the userfaultfd() system call that allows to intercept and
1539 handle page faults in userland.
1541 config ARCH_HAS_MEMBARRIER_CALLBACKS
1544 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1548 bool "Enable rseq() system call" if EXPERT
1550 depends on HAVE_RSEQ
1553 Enable the restartable sequences system call. It provides a
1554 user-space cache for the current CPU number value, which
1555 speeds up getting the current CPU number from user-space,
1556 as well as an ABI to speed up user-space operations on
1563 bool "Enabled debugging of rseq() system call" if EXPERT
1564 depends on RSEQ && DEBUG_KERNEL
1566 Enable extra debugging checks for the rseq system call.
1571 bool "Embedded system"
1572 option allnoconfig_y
1575 This option should be enabled if compiling the kernel for
1576 an embedded system so certain expert options are available
1579 config HAVE_PERF_EVENTS
1582 See tools/perf/design.txt for details.
1584 config PERF_USE_VMALLOC
1587 See tools/perf/design.txt for details
1590 bool "PC/104 support" if EXPERT
1592 Expose PC/104 form factor device drivers and options available for
1593 selection and configuration. Enable this option if your target
1594 machine has a PC/104 bus.
1596 menu "Kernel Performance Events And Counters"
1599 bool "Kernel performance events and counters"
1600 default y if PROFILING
1601 depends on HAVE_PERF_EVENTS
1605 Enable kernel support for various performance events provided
1606 by software and hardware.
1608 Software events are supported either built-in or via the
1609 use of generic tracepoints.
1611 Most modern CPUs support performance events via performance
1612 counter registers. These registers count the number of certain
1613 types of hw events: such as instructions executed, cachemisses
1614 suffered, or branches mis-predicted - without slowing down the
1615 kernel or applications. These registers can also trigger interrupts
1616 when a threshold number of events have passed - and can thus be
1617 used to profile the code that runs on that CPU.
1619 The Linux Performance Event subsystem provides an abstraction of
1620 these software and hardware event capabilities, available via a
1621 system call and used by the "perf" utility in tools/perf/. It
1622 provides per task and per CPU counters, and it provides event
1623 capabilities on top of those.
1627 config DEBUG_PERF_USE_VMALLOC
1629 bool "Debug: use vmalloc to back perf mmap() buffers"
1630 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1631 select PERF_USE_VMALLOC
1633 Use vmalloc memory to back perf mmap() buffers.
1635 Mostly useful for debugging the vmalloc code on platforms
1636 that don't require it.
1642 config VM_EVENT_COUNTERS
1644 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1646 VM event counters are needed for event counts to be shown.
1647 This option allows the disabling of the VM event counters
1648 on EXPERT systems. /proc/vmstat will only show page counts
1649 if VM event counters are disabled.
1653 bool "Enable SLUB debugging support" if EXPERT
1654 depends on SLUB && SYSFS
1656 SLUB has extensive debug support features. Disabling these can
1657 result in significant savings in code size. This also disables
1658 SLUB sysfs support. /sys/slab will not exist and there will be
1659 no support for cache validation etc.
1661 config SLUB_MEMCG_SYSFS_ON
1663 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1664 depends on SLUB && SYSFS && MEMCG
1666 SLUB creates a directory under /sys/kernel/slab for each
1667 allocation cache to host info and debug files. If memory
1668 cgroup is enabled, each cache can have per memory cgroup
1669 caches. SLUB can create the same sysfs directories for these
1670 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1671 to a very high number of debug files being created. This is
1672 controlled by slub_memcg_sysfs boot parameter and this
1673 config option determines the parameter's default value.
1676 bool "Disable heap randomization"
1679 Randomizing heap placement makes heap exploits harder, but it
1680 also breaks ancient binaries (including anything libc5 based).
1681 This option changes the bootup default to heap randomization
1682 disabled, and can be overridden at runtime by setting
1683 /proc/sys/kernel/randomize_va_space to 2.
1685 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1688 prompt "Choose SLAB allocator"
1691 This option allows to select a slab allocator.
1695 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1697 The regular slab allocator that is established and known to work
1698 well in all environments. It organizes cache hot objects in
1699 per cpu and per node queues.
1702 bool "SLUB (Unqueued Allocator)"
1703 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1705 SLUB is a slab allocator that minimizes cache line usage
1706 instead of managing queues of cached objects (SLAB approach).
1707 Per cpu caching is realized using slabs of objects instead
1708 of queues of objects. SLUB can use memory efficiently
1709 and has enhanced diagnostics. SLUB is the default choice for
1714 bool "SLOB (Simple Allocator)"
1716 SLOB replaces the stock allocator with a drastically simpler
1717 allocator. SLOB is generally more space efficient but
1718 does not perform as well on large systems.
1722 config SLAB_MERGE_DEFAULT
1723 bool "Allow slab caches to be merged"
1726 For reduced kernel memory fragmentation, slab caches can be
1727 merged when they share the same size and other characteristics.
1728 This carries a risk of kernel heap overflows being able to
1729 overwrite objects from merged caches (and more easily control
1730 cache layout), which makes such heap attacks easier to exploit
1731 by attackers. By keeping caches unmerged, these kinds of exploits
1732 can usually only damage objects in the same cache. To disable
1733 merging at runtime, "slab_nomerge" can be passed on the kernel
1736 config SLAB_FREELIST_RANDOM
1738 depends on SLAB || SLUB
1739 bool "SLAB freelist randomization"
1741 Randomizes the freelist order used on creating new pages. This
1742 security feature reduces the predictability of the kernel slab
1743 allocator against heap overflows.
1745 config SLAB_FREELIST_HARDENED
1746 bool "Harden slab freelist metadata"
1749 Many kernel heap attacks try to target slab cache metadata and
1750 other infrastructure. This options makes minor performance
1751 sacrifies to harden the kernel slab allocator against common
1752 freelist exploit methods.
1754 config SHUFFLE_PAGE_ALLOCATOR
1755 bool "Page allocator randomization"
1756 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1758 Randomization of the page allocator improves the average
1759 utilization of a direct-mapped memory-side-cache. See section
1760 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1761 6.2a specification for an example of how a platform advertises
1762 the presence of a memory-side-cache. There are also incidental
1763 security benefits as it reduces the predictability of page
1764 allocations to compliment SLAB_FREELIST_RANDOM, but the
1765 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1766 10th order of pages is selected based on cache utilization
1769 While the randomization improves cache utilization it may
1770 negatively impact workloads on platforms without a cache. For
1771 this reason, by default, the randomization is enabled only
1772 after runtime detection of a direct-mapped memory-side-cache.
1773 Otherwise, the randomization may be force enabled with the
1774 'page_alloc.shuffle' kernel command line parameter.
1778 config SLUB_CPU_PARTIAL
1780 depends on SLUB && SMP
1781 bool "SLUB per cpu partial cache"
1783 Per cpu partial caches accellerate objects allocation and freeing
1784 that is local to a processor at the price of more indeterminism
1785 in the latency of the free. On overflow these caches will be cleared
1786 which requires the taking of locks that may cause latency spikes.
1787 Typically one would choose no for a realtime system.
1789 config MMAP_ALLOW_UNINITIALIZED
1790 bool "Allow mmapped anonymous memory to be uninitialized"
1791 depends on EXPERT && !MMU
1794 Normally, and according to the Linux spec, anonymous memory obtained
1795 from mmap() has its contents cleared before it is passed to
1796 userspace. Enabling this config option allows you to request that
1797 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1798 providing a huge performance boost. If this option is not enabled,
1799 then the flag will be ignored.
1801 This is taken advantage of by uClibc's malloc(), and also by
1802 ELF-FDPIC binfmt's brk and stack allocator.
1804 Because of the obvious security issues, this option should only be
1805 enabled on embedded devices where you control what is run in
1806 userspace. Since that isn't generally a problem on no-MMU systems,
1807 it is normally safe to say Y here.
1809 See Documentation/nommu-mmap.txt for more information.
1811 config SYSTEM_DATA_VERIFICATION
1813 select SYSTEM_TRUSTED_KEYRING
1817 select ASYMMETRIC_KEY_TYPE
1818 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1821 select X509_CERTIFICATE_PARSER
1822 select PKCS7_MESSAGE_PARSER
1824 Provide PKCS#7 message verification using the contents of the system
1825 trusted keyring to provide public keys. This then can be used for
1826 module verification, kexec image verification and firmware blob
1830 bool "Profiling support"
1832 Say Y here to enable the extended profiling support mechanisms used
1833 by profilers such as OProfile.
1836 # Place an empty function call at each tracepoint site. Can be
1837 # dynamically changed for a probe function.
1842 endmenu # General setup
1844 source "arch/Kconfig"
1851 default 0 if BASE_FULL
1852 default 1 if !BASE_FULL
1855 bool "Enable loadable module support"
1858 Kernel modules are small pieces of compiled code which can
1859 be inserted in the running kernel, rather than being
1860 permanently built into the kernel. You use the "modprobe"
1861 tool to add (and sometimes remove) them. If you say Y here,
1862 many parts of the kernel can be built as modules (by
1863 answering M instead of Y where indicated): this is most
1864 useful for infrequently used options which are not required
1865 for booting. For more information, see the man pages for
1866 modprobe, lsmod, modinfo, insmod and rmmod.
1868 If you say Y here, you will need to run "make
1869 modules_install" to put the modules under /lib/modules/
1870 where modprobe can find them (you may need to be root to do
1877 config MODULE_FORCE_LOAD
1878 bool "Forced module loading"
1881 Allow loading of modules without version information (ie. modprobe
1882 --force). Forced module loading sets the 'F' (forced) taint flag and
1883 is usually a really bad idea.
1885 config MODULE_UNLOAD
1886 bool "Module unloading"
1888 Without this option you will not be able to unload any
1889 modules (note that some modules may not be unloadable
1890 anyway), which makes your kernel smaller, faster
1891 and simpler. If unsure, say Y.
1893 config MODULE_FORCE_UNLOAD
1894 bool "Forced module unloading"
1895 depends on MODULE_UNLOAD
1897 This option allows you to force a module to unload, even if the
1898 kernel believes it is unsafe: the kernel will remove the module
1899 without waiting for anyone to stop using it (using the -f option to
1900 rmmod). This is mainly for kernel developers and desperate users.
1904 bool "Module versioning support"
1906 Usually, you have to use modules compiled with your kernel.
1907 Saying Y here makes it sometimes possible to use modules
1908 compiled for different kernels, by adding enough information
1909 to the modules to (hopefully) spot any changes which would
1910 make them incompatible with the kernel you are running. If
1913 config MODULE_REL_CRCS
1915 depends on MODVERSIONS
1917 config MODULE_SRCVERSION_ALL
1918 bool "Source checksum for all modules"
1920 Modules which contain a MODULE_VERSION get an extra "srcversion"
1921 field inserted into their modinfo section, which contains a
1922 sum of the source files which made it. This helps maintainers
1923 see exactly which source was used to build a module (since
1924 others sometimes change the module source without updating
1925 the version). With this option, such a "srcversion" field
1926 will be created for all modules. If unsure, say N.
1929 bool "Module signature verification"
1931 select SYSTEM_DATA_VERIFICATION
1933 Check modules for valid signatures upon load: the signature
1934 is simply appended to the module. For more information see
1935 <file:Documentation/admin-guide/module-signing.rst>.
1937 Note that this option adds the OpenSSL development packages as a
1938 kernel build dependency so that the signing tool can use its crypto
1941 !!!WARNING!!! If you enable this option, you MUST make sure that the
1942 module DOES NOT get stripped after being signed. This includes the
1943 debuginfo strip done by some packagers (such as rpmbuild) and
1944 inclusion into an initramfs that wants the module size reduced.
1946 config MODULE_SIG_FORCE
1947 bool "Require modules to be validly signed"
1948 depends on MODULE_SIG
1950 Reject unsigned modules or signed modules for which we don't have a
1951 key. Without this, such modules will simply taint the kernel.
1953 config MODULE_SIG_ALL
1954 bool "Automatically sign all modules"
1956 depends on MODULE_SIG
1958 Sign all modules during make modules_install. Without this option,
1959 modules must be signed manually, using the scripts/sign-file tool.
1961 comment "Do not forget to sign required modules with scripts/sign-file"
1962 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
1965 prompt "Which hash algorithm should modules be signed with?"
1966 depends on MODULE_SIG
1968 This determines which sort of hashing algorithm will be used during
1969 signature generation. This algorithm _must_ be built into the kernel
1970 directly so that signature verification can take place. It is not
1971 possible to load a signed module containing the algorithm to check
1972 the signature on that module.
1974 config MODULE_SIG_SHA1
1975 bool "Sign modules with SHA-1"
1978 config MODULE_SIG_SHA224
1979 bool "Sign modules with SHA-224"
1980 select CRYPTO_SHA256
1982 config MODULE_SIG_SHA256
1983 bool "Sign modules with SHA-256"
1984 select CRYPTO_SHA256
1986 config MODULE_SIG_SHA384
1987 bool "Sign modules with SHA-384"
1988 select CRYPTO_SHA512
1990 config MODULE_SIG_SHA512
1991 bool "Sign modules with SHA-512"
1992 select CRYPTO_SHA512
1996 config MODULE_SIG_HASH
1998 depends on MODULE_SIG
1999 default "sha1" if MODULE_SIG_SHA1
2000 default "sha224" if MODULE_SIG_SHA224
2001 default "sha256" if MODULE_SIG_SHA256
2002 default "sha384" if MODULE_SIG_SHA384
2003 default "sha512" if MODULE_SIG_SHA512
2005 config MODULE_COMPRESS
2006 bool "Compress modules on installation"
2010 Compresses kernel modules when 'make modules_install' is run; gzip or
2011 xz depending on "Compression algorithm" below.
2013 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2015 Out-of-tree kernel modules installed using Kbuild will also be
2016 compressed upon installation.
2018 Note: for modules inside an initrd or initramfs, it's more efficient
2019 to compress the whole initrd or initramfs instead.
2021 Note: This is fully compatible with signed modules.
2026 prompt "Compression algorithm"
2027 depends on MODULE_COMPRESS
2028 default MODULE_COMPRESS_GZIP
2030 This determines which sort of compression will be used during
2031 'make modules_install'.
2033 GZIP (default) and XZ are supported.
2035 config MODULE_COMPRESS_GZIP
2038 config MODULE_COMPRESS_XZ
2043 config TRIM_UNUSED_KSYMS
2044 bool "Trim unused exported kernel symbols"
2045 depends on MODULES && !UNUSED_SYMBOLS
2047 The kernel and some modules make many symbols available for
2048 other modules to use via EXPORT_SYMBOL() and variants. Depending
2049 on the set of modules being selected in your kernel configuration,
2050 many of those exported symbols might never be used.
2052 This option allows for unused exported symbols to be dropped from
2053 the build. In turn, this provides the compiler more opportunities
2054 (especially when using LTO) for optimizing the code and reducing
2055 binary size. This might have some security advantages as well.
2057 If unsure, or if you need to build out-of-tree modules, say N.
2061 config MODULES_TREE_LOOKUP
2063 depends on PERF_EVENTS || TRACING
2065 config INIT_ALL_POSSIBLE
2068 Back when each arch used to define their own cpu_online_mask and
2069 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2070 with all 1s, and others with all 0s. When they were centralised,
2071 it was better to provide this option than to break all the archs
2072 and have several arch maintainers pursuing me down dark alleys.
2074 source "block/Kconfig"
2076 config PREEMPT_NOTIFIERS
2086 Build a simple ASN.1 grammar compiler that produces a bytecode output
2087 that can be interpreted by the ASN.1 stream decoder and used to
2088 inform it as to what tags are to be expected in a stream and what
2089 functions to call on what tags.
2091 source "kernel/Kconfig.locks"
2093 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2096 # It may be useful for an architecture to override the definitions of the
2097 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2098 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2099 # different calling convention for syscalls. They can also override the
2100 # macros for not-implemented syscalls in kernel/sys_ni.c and
2101 # kernel/time/posix-stubs.c. All these overrides need to be available in
2102 # <asm/syscall_wrapper.h>.
2103 config ARCH_HAS_SYSCALL_WRAPPER