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 bool "Compile test headers that should be standalone compilable"
102 Compile test headers listed in header-test-y target to ensure they are
103 self-contained, i.e. compilable as standalone units.
105 If you are a developer or tester and want to ensure the requested
106 headers are self-contained, say Y here. Otherwise, choose N.
109 string "Local version - append to kernel release"
111 Append an extra string to the end of your kernel version.
112 This will show up when you type uname, for example.
113 The string you set here will be appended after the contents of
114 any files with a filename matching localversion* in your
115 object and source tree, in that order. Your total string can
116 be a maximum of 64 characters.
118 config LOCALVERSION_AUTO
119 bool "Automatically append version information to the version string"
121 depends on !COMPILE_TEST
123 This will try to automatically determine if the current tree is a
124 release tree by looking for git tags that belong to the current
125 top of tree revision.
127 A string of the format -gxxxxxxxx will be added to the localversion
128 if a git-based tree is found. The string generated by this will be
129 appended after any matching localversion* files, and after the value
130 set in CONFIG_LOCALVERSION.
132 (The actual string used here is the first eight characters produced
133 by running the command:
135 $ git rev-parse --verify HEAD
137 which is done within the script "scripts/setlocalversion".)
140 string "Build ID Salt"
143 The build ID is used to link binaries and their debug info. Setting
144 this option will use the value in the calculation of the build id.
145 This is mostly useful for distributions which want to ensure the
146 build is unique between builds. It's safe to leave the default.
148 config HAVE_KERNEL_GZIP
151 config HAVE_KERNEL_BZIP2
154 config HAVE_KERNEL_LZMA
157 config HAVE_KERNEL_XZ
160 config HAVE_KERNEL_LZO
163 config HAVE_KERNEL_LZ4
166 config HAVE_KERNEL_UNCOMPRESSED
170 prompt "Kernel compression mode"
172 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4 || HAVE_KERNEL_UNCOMPRESSED
174 The linux kernel is a kind of self-extracting executable.
175 Several compression algorithms are available, which differ
176 in efficiency, compression and decompression speed.
177 Compression speed is only relevant when building a kernel.
178 Decompression speed is relevant at each boot.
180 If you have any problems with bzip2 or lzma compressed
181 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
182 version of this functionality (bzip2 only), for 2.4, was
183 supplied by Christian Ludwig)
185 High compression options are mostly useful for users, who
186 are low on disk space (embedded systems), but for whom ram
189 If in doubt, select 'gzip'
193 depends on HAVE_KERNEL_GZIP
195 The old and tried gzip compression. It provides a good balance
196 between compression ratio and decompression speed.
200 depends on HAVE_KERNEL_BZIP2
202 Its compression ratio and speed is intermediate.
203 Decompression speed is slowest among the choices. The kernel
204 size is about 10% smaller with bzip2, in comparison to gzip.
205 Bzip2 uses a large amount of memory. For modern kernels you
206 will need at least 8MB RAM or more for booting.
210 depends on HAVE_KERNEL_LZMA
212 This compression algorithm's ratio is best. Decompression speed
213 is between gzip and bzip2. Compression is slowest.
214 The kernel size is about 33% smaller with LZMA in comparison to gzip.
218 depends on HAVE_KERNEL_XZ
220 XZ uses the LZMA2 algorithm and instruction set specific
221 BCJ filters which can improve compression ratio of executable
222 code. The size of the kernel is about 30% smaller with XZ in
223 comparison to gzip. On architectures for which there is a BCJ
224 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
225 will create a few percent smaller kernel than plain LZMA.
227 The speed is about the same as with LZMA: The decompression
228 speed of XZ is better than that of bzip2 but worse than gzip
229 and LZO. Compression is slow.
233 depends on HAVE_KERNEL_LZO
235 Its compression ratio is the poorest among the choices. The kernel
236 size is about 10% bigger than gzip; however its speed
237 (both compression and decompression) is the fastest.
241 depends on HAVE_KERNEL_LZ4
243 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
244 A preliminary version of LZ4 de/compression tool is available at
245 <https://code.google.com/p/lz4/>.
247 Its compression ratio is worse than LZO. The size of the kernel
248 is about 8% bigger than LZO. But the decompression speed is
251 config KERNEL_UNCOMPRESSED
253 depends on HAVE_KERNEL_UNCOMPRESSED
255 Produce uncompressed kernel image. This option is usually not what
256 you want. It is useful for debugging the kernel in slow simulation
257 environments, where decompressing and moving the kernel is awfully
258 slow. This option allows early boot code to skip the decompressor
259 and jump right at uncompressed kernel image.
263 config DEFAULT_HOSTNAME
264 string "Default hostname"
267 This option determines the default system hostname before userspace
268 calls sethostname(2). The kernel traditionally uses "(none)" here,
269 but you may wish to use a different default here to make a minimal
270 system more usable with less configuration.
273 # For some reason microblaze and nios2 hard code SWAP=n. Hopefully we can
274 # add proper SWAP support to them, in which case this can be remove.
280 bool "Support for paging of anonymous memory (swap)"
281 depends on MMU && BLOCK && !ARCH_NO_SWAP
284 This option allows you to choose whether you want to have support
285 for so called swap devices or swap files in your kernel that are
286 used to provide more virtual memory than the actual RAM present
287 in your computer. If unsure say Y.
292 Inter Process Communication is a suite of library functions and
293 system calls which let processes (running programs) synchronize and
294 exchange information. It is generally considered to be a good thing,
295 and some programs won't run unless you say Y here. In particular, if
296 you want to run the DOS emulator dosemu under Linux (read the
297 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
298 you'll need to say Y here.
300 You can find documentation about IPC with "info ipc" and also in
301 section 6.4 of the Linux Programmer's Guide, available from
302 <http://www.tldp.org/guides.html>.
304 config SYSVIPC_SYSCTL
311 bool "POSIX Message Queues"
314 POSIX variant of message queues is a part of IPC. In POSIX message
315 queues every message has a priority which decides about succession
316 of receiving it by a process. If you want to compile and run
317 programs written e.g. for Solaris with use of its POSIX message
318 queues (functions mq_*) say Y here.
320 POSIX message queues are visible as a filesystem called 'mqueue'
321 and can be mounted somewhere if you want to do filesystem
322 operations on message queues.
326 config POSIX_MQUEUE_SYSCTL
328 depends on POSIX_MQUEUE
332 config CROSS_MEMORY_ATTACH
333 bool "Enable process_vm_readv/writev syscalls"
337 Enabling this option adds the system calls process_vm_readv and
338 process_vm_writev which allow a process with the correct privileges
339 to directly read from or write to another process' address space.
340 See the man page for more details.
343 bool "uselib syscall"
344 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
346 This option enables the uselib syscall, a system call used in the
347 dynamic linker from libc5 and earlier. glibc does not use this
348 system call. If you intend to run programs built on libc5 or
349 earlier, you may need to enable this syscall. Current systems
350 running glibc can safely disable this.
353 bool "Auditing support"
356 Enable auditing infrastructure that can be used with another
357 kernel subsystem, such as SELinux (which requires this for
358 logging of avc messages output). System call auditing is included
359 on architectures which support it.
361 config HAVE_ARCH_AUDITSYSCALL
366 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
369 source "kernel/irq/Kconfig"
370 source "kernel/time/Kconfig"
371 source "kernel/Kconfig.preempt"
373 menu "CPU/Task time and stats accounting"
375 config VIRT_CPU_ACCOUNTING
379 prompt "Cputime accounting"
380 default TICK_CPU_ACCOUNTING if !PPC64
381 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
383 # Kind of a stub config for the pure tick based cputime accounting
384 config TICK_CPU_ACCOUNTING
385 bool "Simple tick based cputime accounting"
386 depends on !S390 && !NO_HZ_FULL
388 This is the basic tick based cputime accounting that maintains
389 statistics about user, system and idle time spent on per jiffies
394 config VIRT_CPU_ACCOUNTING_NATIVE
395 bool "Deterministic task and CPU time accounting"
396 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
397 select VIRT_CPU_ACCOUNTING
399 Select this option to enable more accurate task and CPU time
400 accounting. This is done by reading a CPU counter on each
401 kernel entry and exit and on transitions within the kernel
402 between system, softirq and hardirq state, so there is a
403 small performance impact. In the case of s390 or IBM POWER > 5,
404 this also enables accounting of stolen time on logically-partitioned
407 config VIRT_CPU_ACCOUNTING_GEN
408 bool "Full dynticks CPU time accounting"
409 depends on HAVE_CONTEXT_TRACKING
410 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
411 depends on GENERIC_CLOCKEVENTS
412 select VIRT_CPU_ACCOUNTING
413 select CONTEXT_TRACKING
415 Select this option to enable task and CPU time accounting on full
416 dynticks systems. This accounting is implemented by watching every
417 kernel-user boundaries using the context tracking subsystem.
418 The accounting is thus performed at the expense of some significant
421 For now this is only useful if you are working on the full
422 dynticks subsystem development.
428 config IRQ_TIME_ACCOUNTING
429 bool "Fine granularity task level IRQ time accounting"
430 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
432 Select this option to enable fine granularity task irq time
433 accounting. This is done by reading a timestamp on each
434 transitions between softirq and hardirq state, so there can be a
435 small performance impact.
437 If in doubt, say N here.
439 config HAVE_SCHED_AVG_IRQ
441 depends on IRQ_TIME_ACCOUNTING || PARAVIRT_TIME_ACCOUNTING
444 config BSD_PROCESS_ACCT
445 bool "BSD Process Accounting"
448 If you say Y here, a user level program will be able to instruct the
449 kernel (via a special system call) to write process accounting
450 information to a file: whenever a process exits, information about
451 that process will be appended to the file by the kernel. The
452 information includes things such as creation time, owning user,
453 command name, memory usage, controlling terminal etc. (the complete
454 list is in the struct acct in <file:include/linux/acct.h>). It is
455 up to the user level program to do useful things with this
456 information. This is generally a good idea, so say Y.
458 config BSD_PROCESS_ACCT_V3
459 bool "BSD Process Accounting version 3 file format"
460 depends on BSD_PROCESS_ACCT
463 If you say Y here, the process accounting information is written
464 in a new file format that also logs the process IDs of each
465 process and its parent. Note that this file format is incompatible
466 with previous v0/v1/v2 file formats, so you will need updated tools
467 for processing it. A preliminary version of these tools is available
468 at <http://www.gnu.org/software/acct/>.
471 bool "Export task/process statistics through netlink"
476 Export selected statistics for tasks/processes through the
477 generic netlink interface. Unlike BSD process accounting, the
478 statistics are available during the lifetime of tasks/processes as
479 responses to commands. Like BSD accounting, they are sent to user
484 config TASK_DELAY_ACCT
485 bool "Enable per-task delay accounting"
489 Collect information on time spent by a task waiting for system
490 resources like cpu, synchronous block I/O completion and swapping
491 in pages. Such statistics can help in setting a task's priorities
492 relative to other tasks for cpu, io, rss limits etc.
497 bool "Enable extended accounting over taskstats"
500 Collect extended task accounting data and send the data
501 to userland for processing over the taskstats interface.
505 config TASK_IO_ACCOUNTING
506 bool "Enable per-task storage I/O accounting"
507 depends on TASK_XACCT
509 Collect information on the number of bytes of storage I/O which this
515 bool "Pressure stall information tracking"
517 Collect metrics that indicate how overcommitted the CPU, memory,
518 and IO capacity are in the system.
520 If you say Y here, the kernel will create /proc/pressure/ with the
521 pressure statistics files cpu, memory, and io. These will indicate
522 the share of walltime in which some or all tasks in the system are
523 delayed due to contention of the respective resource.
525 In kernels with cgroup support, cgroups (cgroup2 only) will
526 have cpu.pressure, memory.pressure, and io.pressure files,
527 which aggregate pressure stalls for the grouped tasks only.
529 For more details see Documentation/accounting/psi.txt.
533 config PSI_DEFAULT_DISABLED
534 bool "Require boot parameter to enable pressure stall information tracking"
538 If set, pressure stall information tracking will be disabled
539 per default but can be enabled through passing psi=1 on the
540 kernel commandline during boot.
542 This feature adds some code to the task wakeup and sleep
543 paths of the scheduler. The overhead is too low to affect
544 common scheduling-intense workloads in practice (such as
545 webservers, memcache), but it does show up in artificial
546 scheduler stress tests, such as hackbench.
548 If you are paranoid and not sure what the kernel will be
553 endmenu # "CPU/Task time and stats accounting"
557 depends on SMP || COMPILE_TEST
560 Make sure that CPUs running critical tasks are not disturbed by
561 any source of "noise" such as unbound workqueues, timers, kthreads...
562 Unbound jobs get offloaded to housekeeping CPUs. This is driven by
563 the "isolcpus=" boot parameter.
567 source "kernel/rcu/Kconfig"
574 tristate "Kernel .config support"
576 This option enables the complete Linux kernel ".config" file
577 contents to be saved in the kernel. It provides documentation
578 of which kernel options are used in a running kernel or in an
579 on-disk kernel. This information can be extracted from the kernel
580 image file with the script scripts/extract-ikconfig and used as
581 input to rebuild the current kernel or to build another kernel.
582 It can also be extracted from a running kernel by reading
583 /proc/config.gz if enabled (below).
586 bool "Enable access to .config through /proc/config.gz"
587 depends on IKCONFIG && PROC_FS
589 This option enables access to the kernel configuration file
590 through /proc/config.gz.
593 tristate "Enable kernel headers through /sys/kernel/kheaders.tar.xz"
596 This option enables access to the in-kernel headers that are generated during
597 the build process. These can be used to build eBPF tracing programs,
598 or similar programs. If you build the headers as a module, a module called
599 kheaders.ko is built which can be loaded on-demand to get access to headers.
602 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
607 Select the minimal kernel log buffer size as a power of 2.
608 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
609 parameter, see below. Any higher size also might be forced
610 by "log_buf_len" boot parameter.
620 config LOG_CPU_MAX_BUF_SHIFT
621 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
624 default 12 if !BASE_SMALL
625 default 0 if BASE_SMALL
628 This option allows to increase the default ring buffer size
629 according to the number of CPUs. The value defines the contribution
630 of each CPU as a power of 2. The used space is typically only few
631 lines however it might be much more when problems are reported,
634 The increased size means that a new buffer has to be allocated and
635 the original static one is unused. It makes sense only on systems
636 with more CPUs. Therefore this value is used only when the sum of
637 contributions is greater than the half of the default kernel ring
638 buffer as defined by LOG_BUF_SHIFT. The default values are set
639 so that more than 64 CPUs are needed to trigger the allocation.
641 Also this option is ignored when "log_buf_len" kernel parameter is
642 used as it forces an exact (power of two) size of the ring buffer.
644 The number of possible CPUs is used for this computation ignoring
645 hotplugging making the computation optimal for the worst case
646 scenario while allowing a simple algorithm to be used from bootup.
648 Examples shift values and their meaning:
649 17 => 128 KB for each CPU
650 16 => 64 KB for each CPU
651 15 => 32 KB for each CPU
652 14 => 16 KB for each CPU
653 13 => 8 KB for each CPU
654 12 => 4 KB for each CPU
656 config PRINTK_SAFE_LOG_BUF_SHIFT
657 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
662 Select the size of an alternate printk per-CPU buffer where messages
663 printed from usafe contexts are temporary stored. One example would
664 be NMI messages, another one - printk recursion. The messages are
665 copied to the main log buffer in a safe context to avoid a deadlock.
666 The value defines the size as a power of 2.
668 Those messages are rare and limited. The largest one is when
669 a backtrace is printed. It usually fits into 4KB. Select
670 8KB if you want to be on the safe side.
673 17 => 128 KB for each CPU
674 16 => 64 KB for each CPU
675 15 => 32 KB for each CPU
676 14 => 16 KB for each CPU
677 13 => 8 KB for each CPU
678 12 => 4 KB for each CPU
681 # Architectures with an unreliable sched_clock() should select this:
683 config HAVE_UNSTABLE_SCHED_CLOCK
686 config GENERIC_SCHED_CLOCK
690 # For architectures that want to enable the support for NUMA-affine scheduler
693 config ARCH_SUPPORTS_NUMA_BALANCING
697 # For architectures that prefer to flush all TLBs after a number of pages
698 # are unmapped instead of sending one IPI per page to flush. The architecture
699 # must provide guarantees on what happens if a clean TLB cache entry is
700 # written after the unmap. Details are in mm/rmap.c near the check for
701 # should_defer_flush. The architecture should also consider if the full flush
702 # and the refill costs are offset by the savings of sending fewer IPIs.
703 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
707 # For architectures that know their GCC __int128 support is sound
709 config ARCH_SUPPORTS_INT128
712 # For architectures that (ab)use NUMA to represent different memory regions
713 # all cpu-local but of different latencies, such as SuperH.
715 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
718 config NUMA_BALANCING
719 bool "Memory placement aware NUMA scheduler"
720 depends on ARCH_SUPPORTS_NUMA_BALANCING
721 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
722 depends on SMP && NUMA && MIGRATION
724 This option adds support for automatic NUMA aware memory/task placement.
725 The mechanism is quite primitive and is based on migrating memory when
726 it has references to the node the task is running on.
728 This system will be inactive on UMA systems.
730 config NUMA_BALANCING_DEFAULT_ENABLED
731 bool "Automatically enable NUMA aware memory/task placement"
733 depends on NUMA_BALANCING
735 If set, automatic NUMA balancing will be enabled if running on a NUMA
739 bool "Control Group support"
742 This option adds support for grouping sets of processes together, for
743 use with process control subsystems such as Cpusets, CFS, memory
744 controls or device isolation.
746 - Documentation/scheduler/sched-design-CFS.txt (CFS)
747 - Documentation/cgroup-v1/ (features for grouping, isolation
748 and resource control)
758 bool "Memory controller"
762 Provides control over the memory footprint of tasks in a cgroup.
765 bool "Swap controller"
766 depends on MEMCG && SWAP
768 Provides control over the swap space consumed by tasks in a cgroup.
770 config MEMCG_SWAP_ENABLED
771 bool "Swap controller enabled by default"
772 depends on MEMCG_SWAP
775 Memory Resource Controller Swap Extension comes with its price in
776 a bigger memory consumption. General purpose distribution kernels
777 which want to enable the feature but keep it disabled by default
778 and let the user enable it by swapaccount=1 boot command line
779 parameter should have this option unselected.
780 For those who want to have the feature enabled by default should
781 select this option (if, for some reason, they need to disable it
782 then swapaccount=0 does the trick).
786 depends on MEMCG && !SLOB
794 Generic block IO controller cgroup interface. This is the common
795 cgroup interface which should be used by various IO controlling
798 Currently, CFQ IO scheduler uses it to recognize task groups and
799 control disk bandwidth allocation (proportional time slice allocation)
800 to such task groups. It is also used by bio throttling logic in
801 block layer to implement upper limit in IO rates on a device.
803 This option only enables generic Block IO controller infrastructure.
804 One needs to also enable actual IO controlling logic/policy. For
805 enabling proportional weight division of disk bandwidth in CFQ, set
806 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
807 CONFIG_BLK_DEV_THROTTLING=y.
809 See Documentation/cgroup-v1/blkio-controller.txt for more information.
811 config DEBUG_BLK_CGROUP
812 bool "IO controller debugging"
813 depends on BLK_CGROUP
816 Enable some debugging help. Currently it exports additional stat
817 files in a cgroup which can be useful for debugging.
819 config CGROUP_WRITEBACK
821 depends on MEMCG && BLK_CGROUP
824 menuconfig CGROUP_SCHED
825 bool "CPU controller"
828 This feature lets CPU scheduler recognize task groups and control CPU
829 bandwidth allocation to such task groups. It uses cgroups to group
833 config FAIR_GROUP_SCHED
834 bool "Group scheduling for SCHED_OTHER"
835 depends on CGROUP_SCHED
839 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
840 depends on FAIR_GROUP_SCHED
843 This option allows users to define CPU bandwidth rates (limits) for
844 tasks running within the fair group scheduler. Groups with no limit
845 set are considered to be unconstrained and will run with no
847 See Documentation/scheduler/sched-bwc.txt for more information.
849 config RT_GROUP_SCHED
850 bool "Group scheduling for SCHED_RR/FIFO"
851 depends on CGROUP_SCHED
854 This feature lets you explicitly allocate real CPU bandwidth
855 to task groups. If enabled, it will also make it impossible to
856 schedule realtime tasks for non-root users until you allocate
857 realtime bandwidth for them.
858 See Documentation/scheduler/sched-rt-group.txt for more information.
863 bool "PIDs controller"
865 Provides enforcement of process number limits in the scope of a
866 cgroup. Any attempt to fork more processes than is allowed in the
867 cgroup will fail. PIDs are fundamentally a global resource because it
868 is fairly trivial to reach PID exhaustion before you reach even a
869 conservative kmemcg limit. As a result, it is possible to grind a
870 system to halt without being limited by other cgroup policies. The
871 PIDs controller is designed to stop this from happening.
873 It should be noted that organisational operations (such as attaching
874 to a cgroup hierarchy) will *not* be blocked by the PIDs controller,
875 since the PIDs limit only affects a process's ability to fork, not to
879 bool "RDMA controller"
881 Provides enforcement of RDMA resources defined by IB stack.
882 It is fairly easy for consumers to exhaust RDMA resources, which
883 can result into resource unavailability to other consumers.
884 RDMA controller is designed to stop this from happening.
885 Attaching processes with active RDMA resources to the cgroup
886 hierarchy is allowed even if can cross the hierarchy's limit.
888 config CGROUP_FREEZER
889 bool "Freezer controller"
891 Provides a way to freeze and unfreeze all tasks in a
894 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
895 controller includes important in-kernel memory consumers per default.
897 If you're using cgroup2, say N.
899 config CGROUP_HUGETLB
900 bool "HugeTLB controller"
901 depends on HUGETLB_PAGE
905 Provides a cgroup controller for HugeTLB pages.
906 When you enable this, you can put a per cgroup limit on HugeTLB usage.
907 The limit is enforced during page fault. Since HugeTLB doesn't
908 support page reclaim, enforcing the limit at page fault time implies
909 that, the application will get SIGBUS signal if it tries to access
910 HugeTLB pages beyond its limit. This requires the application to know
911 beforehand how much HugeTLB pages it would require for its use. The
912 control group is tracked in the third page lru pointer. This means
913 that we cannot use the controller with huge page less than 3 pages.
916 bool "Cpuset controller"
919 This option will let you create and manage CPUSETs which
920 allow dynamically partitioning a system into sets of CPUs and
921 Memory Nodes and assigning tasks to run only within those sets.
922 This is primarily useful on large SMP or NUMA systems.
926 config PROC_PID_CPUSET
927 bool "Include legacy /proc/<pid>/cpuset file"
932 bool "Device controller"
934 Provides a cgroup controller implementing whitelists for
935 devices which a process in the cgroup can mknod or open.
937 config CGROUP_CPUACCT
938 bool "Simple CPU accounting controller"
940 Provides a simple controller for monitoring the
941 total CPU consumed by the tasks in a cgroup.
944 bool "Perf controller"
945 depends on PERF_EVENTS
947 This option extends the perf per-cpu mode to restrict monitoring
948 to threads which belong to the cgroup specified and run on the
954 bool "Support for eBPF programs attached to cgroups"
955 depends on BPF_SYSCALL
956 select SOCK_CGROUP_DATA
958 Allow attaching eBPF programs to a cgroup using the bpf(2)
959 syscall command BPF_PROG_ATTACH.
961 In which context these programs are accessed depends on the type
962 of attachment. For instance, programs that are attached using
963 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
967 bool "Debug controller"
969 depends on DEBUG_KERNEL
971 This option enables a simple controller that exports
972 debugging information about the cgroups framework. This
973 controller is for control cgroup debugging only. Its
974 interfaces are not stable.
978 config SOCK_CGROUP_DATA
984 menuconfig NAMESPACES
985 bool "Namespaces support" if EXPERT
989 Provides the way to make tasks work with different objects using
990 the same id. For example same IPC id may refer to different objects
991 or same user id or pid may refer to different tasks when used in
992 different namespaces.
1000 In this namespace tasks see different info provided with the
1004 bool "IPC namespace"
1005 depends on (SYSVIPC || POSIX_MQUEUE)
1008 In this namespace tasks work with IPC ids which correspond to
1009 different IPC objects in different namespaces.
1012 bool "User namespace"
1015 This allows containers, i.e. vservers, to use user namespaces
1016 to provide different user info for different servers.
1018 When user namespaces are enabled in the kernel it is
1019 recommended that the MEMCG option also be enabled and that
1020 user-space use the memory control groups to limit the amount
1021 of memory a memory unprivileged users can use.
1026 bool "PID Namespaces"
1029 Support process id namespaces. This allows having multiple
1030 processes with the same pid as long as they are in different
1031 pid namespaces. This is a building block of containers.
1034 bool "Network namespace"
1038 Allow user space to create what appear to be multiple instances
1039 of the network stack.
1043 config CHECKPOINT_RESTORE
1044 bool "Checkpoint/restore support"
1045 select PROC_CHILDREN
1048 Enables additional kernel features in a sake of checkpoint/restore.
1049 In particular it adds auxiliary prctl codes to setup process text,
1050 data and heap segment sizes, and a few additional /proc filesystem
1053 If unsure, say N here.
1055 config SCHED_AUTOGROUP
1056 bool "Automatic process group scheduling"
1059 select FAIR_GROUP_SCHED
1061 This option optimizes the scheduler for common desktop workloads by
1062 automatically creating and populating task groups. This separation
1063 of workloads isolates aggressive CPU burners (like build jobs) from
1064 desktop applications. Task group autogeneration is currently based
1067 config SYSFS_DEPRECATED
1068 bool "Enable deprecated sysfs features to support old userspace tools"
1072 This option adds code that switches the layout of the "block" class
1073 devices, to not show up in /sys/class/block/, but only in
1076 This switch is only active when the sysfs.deprecated=1 boot option is
1077 passed or the SYSFS_DEPRECATED_V2 option is set.
1079 This option allows new kernels to run on old distributions and tools,
1080 which might get confused by /sys/class/block/. Since 2007/2008 all
1081 major distributions and tools handle this just fine.
1083 Recent distributions and userspace tools after 2009/2010 depend on
1084 the existence of /sys/class/block/, and will not work with this
1087 Only if you are using a new kernel on an old distribution, you might
1090 config SYSFS_DEPRECATED_V2
1091 bool "Enable deprecated sysfs features by default"
1094 depends on SYSFS_DEPRECATED
1096 Enable deprecated sysfs by default.
1098 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1101 Only if you are using a new kernel on an old distribution, you might
1102 need to say Y here. Even then, odds are you would not need it
1103 enabled, you can always pass the boot option if absolutely necessary.
1106 bool "Kernel->user space relay support (formerly relayfs)"
1109 This option enables support for relay interface support in
1110 certain file systems (such as debugfs).
1111 It is designed to provide an efficient mechanism for tools and
1112 facilities to relay large amounts of data from kernel space to
1117 config BLK_DEV_INITRD
1118 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1120 The initial RAM filesystem is a ramfs which is loaded by the
1121 boot loader (loadlin or lilo) and that is mounted as root
1122 before the normal boot procedure. It is typically used to
1123 load modules needed to mount the "real" root file system,
1124 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1126 If RAM disk support (BLK_DEV_RAM) is also included, this
1127 also enables initial RAM disk (initrd) support and adds
1128 15 Kbytes (more on some other architectures) to the kernel size.
1134 source "usr/Kconfig"
1139 prompt "Compiler optimization level"
1140 default CC_OPTIMIZE_FOR_PERFORMANCE
1142 config CC_OPTIMIZE_FOR_PERFORMANCE
1143 bool "Optimize for performance"
1145 This is the default optimization level for the kernel, building
1146 with the "-O2" compiler flag for best performance and most
1147 helpful compile-time warnings.
1149 config CC_OPTIMIZE_FOR_SIZE
1150 bool "Optimize for size"
1151 imply CC_DISABLE_WARN_MAYBE_UNINITIALIZED # avoid false positives
1153 Enabling this option will pass "-Os" instead of "-O2" to
1154 your compiler resulting in a smaller kernel.
1160 config HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1163 This requires that the arch annotates or otherwise protects
1164 its external entry points from being discarded. Linker scripts
1165 must also merge .text.*, .data.*, and .bss.* correctly into
1166 output sections. Care must be taken not to pull in unrelated
1167 sections (e.g., '.text.init'). Typically '.' in section names
1168 is used to distinguish them from label names / C identifiers.
1170 config LD_DEAD_CODE_DATA_ELIMINATION
1171 bool "Dead code and data elimination (EXPERIMENTAL)"
1172 depends on HAVE_LD_DEAD_CODE_DATA_ELIMINATION
1174 depends on !(FUNCTION_TRACER && CC_IS_GCC && GCC_VERSION < 40800)
1175 depends on $(cc-option,-ffunction-sections -fdata-sections)
1176 depends on $(ld-option,--gc-sections)
1178 Enable this if you want to do dead code and data elimination with
1179 the linker by compiling with -ffunction-sections -fdata-sections,
1180 and linking with --gc-sections.
1182 This can reduce on disk and in-memory size of the kernel
1183 code and static data, particularly for small configs and
1184 on small systems. This has the possibility of introducing
1185 silently broken kernel if the required annotations are not
1186 present. This option is not well tested yet, so use at your
1195 config SYSCTL_EXCEPTION_TRACE
1198 Enable support for /proc/sys/debug/exception-trace.
1200 config SYSCTL_ARCH_UNALIGN_NO_WARN
1203 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1204 Allows arch to define/use @no_unaligned_warning to possibly warn
1205 about unaligned access emulation going on under the hood.
1207 config SYSCTL_ARCH_UNALIGN_ALLOW
1210 Enable support for /proc/sys/kernel/unaligned-trap
1211 Allows arches to define/use @unaligned_enabled to runtime toggle
1212 the unaligned access emulation.
1213 see arch/parisc/kernel/unaligned.c for reference
1215 config HAVE_PCSPKR_PLATFORM
1218 # interpreter that classic socket filters depend on
1223 bool "Configure standard kernel features (expert users)"
1224 # Unhide debug options, to make the on-by-default options visible
1227 This option allows certain base kernel options and settings
1228 to be disabled or tweaked. This is for specialized
1229 environments which can tolerate a "non-standard" kernel.
1230 Only use this if you really know what you are doing.
1233 bool "Enable 16-bit UID system calls" if EXPERT
1234 depends on HAVE_UID16 && MULTIUSER
1237 This enables the legacy 16-bit UID syscall wrappers.
1240 bool "Multiple users, groups and capabilities support" if EXPERT
1243 This option enables support for non-root users, groups and
1246 If you say N here, all processes will run with UID 0, GID 0, and all
1247 possible capabilities. Saying N here also compiles out support for
1248 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1251 If unsure, say Y here.
1253 config SGETMASK_SYSCALL
1254 bool "sgetmask/ssetmask syscalls support" if EXPERT
1255 def_bool PARISC || M68K || PPC || MIPS || X86 || SPARC || MICROBLAZE || SUPERH
1257 sys_sgetmask and sys_ssetmask are obsolete system calls
1258 no longer supported in libc but still enabled by default in some
1261 If unsure, leave the default option here.
1263 config SYSFS_SYSCALL
1264 bool "Sysfs syscall support" if EXPERT
1267 sys_sysfs is an obsolete system call no longer supported in libc.
1268 Note that disabling this option is more secure but might break
1269 compatibility with some systems.
1271 If unsure say Y here.
1273 config SYSCTL_SYSCALL
1274 bool "Sysctl syscall support" if EXPERT
1275 depends on PROC_SYSCTL
1279 sys_sysctl uses binary paths that have been found challenging
1280 to properly maintain and use. The interface in /proc/sys
1281 using paths with ascii names is now the primary path to this
1284 Almost nothing using the binary sysctl interface so if you are
1285 trying to save some space it is probably safe to disable this,
1286 making your kernel marginally smaller.
1288 If unsure say N here.
1291 bool "open by fhandle syscalls" if EXPERT
1295 If you say Y here, a user level program will be able to map
1296 file names to handle and then later use the handle for
1297 different file system operations. This is useful in implementing
1298 userspace file servers, which now track files using handles instead
1299 of names. The handle would remain the same even if file names
1300 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
1304 bool "Posix Clocks & timers" if EXPERT
1307 This includes native support for POSIX timers to the kernel.
1308 Some embedded systems have no use for them and therefore they
1309 can be configured out to reduce the size of the kernel image.
1311 When this option is disabled, the following syscalls won't be
1312 available: timer_create, timer_gettime: timer_getoverrun,
1313 timer_settime, timer_delete, clock_adjtime, getitimer,
1314 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1315 clock_getres and clock_nanosleep syscalls will be limited to
1316 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1322 bool "Enable support for printk" if EXPERT
1325 This option enables normal printk support. Removing it
1326 eliminates most of the message strings from the kernel image
1327 and makes the kernel more or less silent. As this makes it
1328 very difficult to diagnose system problems, saying N here is
1329 strongly discouraged.
1337 bool "BUG() support" if EXPERT
1340 Disabling this option eliminates support for BUG and WARN, reducing
1341 the size of your kernel image and potentially quietly ignoring
1342 numerous fatal conditions. You should only consider disabling this
1343 option for embedded systems with no facilities for reporting errors.
1349 bool "Enable ELF core dumps" if EXPERT
1351 Enable support for generating core dumps. Disabling saves about 4k.
1354 config PCSPKR_PLATFORM
1355 bool "Enable PC-Speaker support" if EXPERT
1356 depends on HAVE_PCSPKR_PLATFORM
1360 This option allows to disable the internal PC-Speaker
1361 support, saving some memory.
1365 bool "Enable full-sized data structures for core" if EXPERT
1367 Disabling this option reduces the size of miscellaneous core
1368 kernel data structures. This saves memory on small machines,
1369 but may reduce performance.
1372 bool "Enable futex support" if EXPERT
1376 Disabling this option will cause the kernel to be built without
1377 support for "fast userspace mutexes". The resulting kernel may not
1378 run glibc-based applications correctly.
1382 depends on FUTEX && RT_MUTEXES
1385 config HAVE_FUTEX_CMPXCHG
1389 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1390 is implemented and always working. This removes a couple of runtime
1394 bool "Enable eventpoll support" if EXPERT
1397 Disabling this option will cause the kernel to be built without
1398 support for epoll family of system calls.
1401 bool "Enable signalfd() system call" if EXPERT
1404 Enable the signalfd() system call that allows to receive signals
1405 on a file descriptor.
1410 bool "Enable timerfd() system call" if EXPERT
1413 Enable the timerfd() system call that allows to receive timer
1414 events on a file descriptor.
1419 bool "Enable eventfd() system call" if EXPERT
1422 Enable the eventfd() system call that allows to receive both
1423 kernel notification (ie. KAIO) or userspace notifications.
1428 bool "Use full shmem filesystem" if EXPERT
1432 The shmem is an internal filesystem used to manage shared memory.
1433 It is backed by swap and manages resource limits. It is also exported
1434 to userspace as tmpfs if TMPFS is enabled. Disabling this
1435 option replaces shmem and tmpfs with the much simpler ramfs code,
1436 which may be appropriate on small systems without swap.
1439 bool "Enable AIO support" if EXPERT
1442 This option enables POSIX asynchronous I/O which may by used
1443 by some high performance threaded applications. Disabling
1444 this option saves about 7k.
1447 bool "Enable IO uring support" if EXPERT
1451 This option enables support for the io_uring interface, enabling
1452 applications to submit and complete IO through submission and
1453 completion rings that are shared between the kernel and application.
1455 config ADVISE_SYSCALLS
1456 bool "Enable madvise/fadvise syscalls" if EXPERT
1459 This option enables the madvise and fadvise syscalls, used by
1460 applications to advise the kernel about their future memory or file
1461 usage, improving performance. If building an embedded system where no
1462 applications use these syscalls, you can disable this option to save
1466 bool "Enable membarrier() system call" if EXPERT
1469 Enable the membarrier() system call that allows issuing memory
1470 barriers across all running threads, which can be used to distribute
1471 the cost of user-space memory barriers asymmetrically by transforming
1472 pairs of memory barriers into pairs consisting of membarrier() and a
1478 bool "Load all symbols for debugging/ksymoops" if EXPERT
1481 Say Y here to let the kernel print out symbolic crash information and
1482 symbolic stack backtraces. This increases the size of the kernel
1483 somewhat, as all symbols have to be loaded into the kernel image.
1486 bool "Include all symbols in kallsyms"
1487 depends on DEBUG_KERNEL && KALLSYMS
1489 Normally kallsyms only contains the symbols of functions for nicer
1490 OOPS messages and backtraces (i.e., symbols from the text and inittext
1491 sections). This is sufficient for most cases. And only in very rare
1492 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1493 names of variables from the data sections, etc).
1495 This option makes sure that all symbols are loaded into the kernel
1496 image (i.e., symbols from all sections) in cost of increased kernel
1497 size (depending on the kernel configuration, it may be 300KiB or
1498 something like this).
1500 Say N unless you really need all symbols.
1502 config KALLSYMS_ABSOLUTE_PERCPU
1505 default X86_64 && SMP
1507 config KALLSYMS_BASE_RELATIVE
1512 Instead of emitting them as absolute values in the native word size,
1513 emit the symbol references in the kallsyms table as 32-bit entries,
1514 each containing a relative value in the range [base, base + U32_MAX]
1515 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1516 an absolute value in the range [0, S32_MAX] or a relative value in the
1517 range [base, base + S32_MAX], where base is the lowest relative symbol
1518 address encountered in the image.
1520 On 64-bit builds, this reduces the size of the address table by 50%,
1521 but more importantly, it results in entries whose values are build
1522 time constants, and no relocation pass is required at runtime to fix
1523 up the entries based on the runtime load address of the kernel.
1525 # end of the "standard kernel features (expert users)" menu
1527 # syscall, maps, verifier
1529 bool "Enable bpf() system call"
1534 Enable the bpf() system call that allows to manipulate eBPF
1535 programs and maps via file descriptors.
1537 config BPF_JIT_ALWAYS_ON
1538 bool "Permanently enable BPF JIT and remove BPF interpreter"
1539 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1541 Enables BPF JIT and removes BPF interpreter to avoid
1542 speculative execution of BPF instructions by the interpreter
1545 bool "Enable userfaultfd() system call"
1548 Enable the userfaultfd() system call that allows to intercept and
1549 handle page faults in userland.
1551 config ARCH_HAS_MEMBARRIER_CALLBACKS
1554 config ARCH_HAS_MEMBARRIER_SYNC_CORE
1558 bool "Enable rseq() system call" if EXPERT
1560 depends on HAVE_RSEQ
1563 Enable the restartable sequences system call. It provides a
1564 user-space cache for the current CPU number value, which
1565 speeds up getting the current CPU number from user-space,
1566 as well as an ABI to speed up user-space operations on
1573 bool "Enabled debugging of rseq() system call" if EXPERT
1574 depends on RSEQ && DEBUG_KERNEL
1576 Enable extra debugging checks for the rseq system call.
1581 bool "Embedded system"
1582 option allnoconfig_y
1585 This option should be enabled if compiling the kernel for
1586 an embedded system so certain expert options are available
1589 config HAVE_PERF_EVENTS
1592 See tools/perf/design.txt for details.
1594 config PERF_USE_VMALLOC
1597 See tools/perf/design.txt for details
1600 bool "PC/104 support" if EXPERT
1602 Expose PC/104 form factor device drivers and options available for
1603 selection and configuration. Enable this option if your target
1604 machine has a PC/104 bus.
1606 menu "Kernel Performance Events And Counters"
1609 bool "Kernel performance events and counters"
1610 default y if PROFILING
1611 depends on HAVE_PERF_EVENTS
1615 Enable kernel support for various performance events provided
1616 by software and hardware.
1618 Software events are supported either built-in or via the
1619 use of generic tracepoints.
1621 Most modern CPUs support performance events via performance
1622 counter registers. These registers count the number of certain
1623 types of hw events: such as instructions executed, cachemisses
1624 suffered, or branches mis-predicted - without slowing down the
1625 kernel or applications. These registers can also trigger interrupts
1626 when a threshold number of events have passed - and can thus be
1627 used to profile the code that runs on that CPU.
1629 The Linux Performance Event subsystem provides an abstraction of
1630 these software and hardware event capabilities, available via a
1631 system call and used by the "perf" utility in tools/perf/. It
1632 provides per task and per CPU counters, and it provides event
1633 capabilities on top of those.
1637 config DEBUG_PERF_USE_VMALLOC
1639 bool "Debug: use vmalloc to back perf mmap() buffers"
1640 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1641 select PERF_USE_VMALLOC
1643 Use vmalloc memory to back perf mmap() buffers.
1645 Mostly useful for debugging the vmalloc code on platforms
1646 that don't require it.
1652 config VM_EVENT_COUNTERS
1654 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1656 VM event counters are needed for event counts to be shown.
1657 This option allows the disabling of the VM event counters
1658 on EXPERT systems. /proc/vmstat will only show page counts
1659 if VM event counters are disabled.
1663 bool "Enable SLUB debugging support" if EXPERT
1664 depends on SLUB && SYSFS
1666 SLUB has extensive debug support features. Disabling these can
1667 result in significant savings in code size. This also disables
1668 SLUB sysfs support. /sys/slab will not exist and there will be
1669 no support for cache validation etc.
1671 config SLUB_MEMCG_SYSFS_ON
1673 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1674 depends on SLUB && SYSFS && MEMCG
1676 SLUB creates a directory under /sys/kernel/slab for each
1677 allocation cache to host info and debug files. If memory
1678 cgroup is enabled, each cache can have per memory cgroup
1679 caches. SLUB can create the same sysfs directories for these
1680 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1681 to a very high number of debug files being created. This is
1682 controlled by slub_memcg_sysfs boot parameter and this
1683 config option determines the parameter's default value.
1686 bool "Disable heap randomization"
1689 Randomizing heap placement makes heap exploits harder, but it
1690 also breaks ancient binaries (including anything libc5 based).
1691 This option changes the bootup default to heap randomization
1692 disabled, and can be overridden at runtime by setting
1693 /proc/sys/kernel/randomize_va_space to 2.
1695 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1698 prompt "Choose SLAB allocator"
1701 This option allows to select a slab allocator.
1705 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1707 The regular slab allocator that is established and known to work
1708 well in all environments. It organizes cache hot objects in
1709 per cpu and per node queues.
1712 bool "SLUB (Unqueued Allocator)"
1713 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1715 SLUB is a slab allocator that minimizes cache line usage
1716 instead of managing queues of cached objects (SLAB approach).
1717 Per cpu caching is realized using slabs of objects instead
1718 of queues of objects. SLUB can use memory efficiently
1719 and has enhanced diagnostics. SLUB is the default choice for
1724 bool "SLOB (Simple Allocator)"
1726 SLOB replaces the stock allocator with a drastically simpler
1727 allocator. SLOB is generally more space efficient but
1728 does not perform as well on large systems.
1732 config SLAB_MERGE_DEFAULT
1733 bool "Allow slab caches to be merged"
1736 For reduced kernel memory fragmentation, slab caches can be
1737 merged when they share the same size and other characteristics.
1738 This carries a risk of kernel heap overflows being able to
1739 overwrite objects from merged caches (and more easily control
1740 cache layout), which makes such heap attacks easier to exploit
1741 by attackers. By keeping caches unmerged, these kinds of exploits
1742 can usually only damage objects in the same cache. To disable
1743 merging at runtime, "slab_nomerge" can be passed on the kernel
1746 config SLAB_FREELIST_RANDOM
1748 depends on SLAB || SLUB
1749 bool "SLAB freelist randomization"
1751 Randomizes the freelist order used on creating new pages. This
1752 security feature reduces the predictability of the kernel slab
1753 allocator against heap overflows.
1755 config SLAB_FREELIST_HARDENED
1756 bool "Harden slab freelist metadata"
1759 Many kernel heap attacks try to target slab cache metadata and
1760 other infrastructure. This options makes minor performance
1761 sacrifies to harden the kernel slab allocator against common
1762 freelist exploit methods.
1764 config SHUFFLE_PAGE_ALLOCATOR
1765 bool "Page allocator randomization"
1766 default SLAB_FREELIST_RANDOM && ACPI_NUMA
1768 Randomization of the page allocator improves the average
1769 utilization of a direct-mapped memory-side-cache. See section
1770 5.2.27 Heterogeneous Memory Attribute Table (HMAT) in the ACPI
1771 6.2a specification for an example of how a platform advertises
1772 the presence of a memory-side-cache. There are also incidental
1773 security benefits as it reduces the predictability of page
1774 allocations to compliment SLAB_FREELIST_RANDOM, but the
1775 default granularity of shuffling on the "MAX_ORDER - 1" i.e,
1776 10th order of pages is selected based on cache utilization
1779 While the randomization improves cache utilization it may
1780 negatively impact workloads on platforms without a cache. For
1781 this reason, by default, the randomization is enabled only
1782 after runtime detection of a direct-mapped memory-side-cache.
1783 Otherwise, the randomization may be force enabled with the
1784 'page_alloc.shuffle' kernel command line parameter.
1788 config SLUB_CPU_PARTIAL
1790 depends on SLUB && SMP
1791 bool "SLUB per cpu partial cache"
1793 Per cpu partial caches accellerate objects allocation and freeing
1794 that is local to a processor at the price of more indeterminism
1795 in the latency of the free. On overflow these caches will be cleared
1796 which requires the taking of locks that may cause latency spikes.
1797 Typically one would choose no for a realtime system.
1799 config MMAP_ALLOW_UNINITIALIZED
1800 bool "Allow mmapped anonymous memory to be uninitialized"
1801 depends on EXPERT && !MMU
1804 Normally, and according to the Linux spec, anonymous memory obtained
1805 from mmap() has its contents cleared before it is passed to
1806 userspace. Enabling this config option allows you to request that
1807 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1808 providing a huge performance boost. If this option is not enabled,
1809 then the flag will be ignored.
1811 This is taken advantage of by uClibc's malloc(), and also by
1812 ELF-FDPIC binfmt's brk and stack allocator.
1814 Because of the obvious security issues, this option should only be
1815 enabled on embedded devices where you control what is run in
1816 userspace. Since that isn't generally a problem on no-MMU systems,
1817 it is normally safe to say Y here.
1819 See Documentation/nommu-mmap.txt for more information.
1821 config SYSTEM_DATA_VERIFICATION
1823 select SYSTEM_TRUSTED_KEYRING
1827 select ASYMMETRIC_KEY_TYPE
1828 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1831 select X509_CERTIFICATE_PARSER
1832 select PKCS7_MESSAGE_PARSER
1834 Provide PKCS#7 message verification using the contents of the system
1835 trusted keyring to provide public keys. This then can be used for
1836 module verification, kexec image verification and firmware blob
1840 bool "Profiling support"
1842 Say Y here to enable the extended profiling support mechanisms used
1843 by profilers such as OProfile.
1846 # Place an empty function call at each tracepoint site. Can be
1847 # dynamically changed for a probe function.
1852 endmenu # General setup
1854 source "arch/Kconfig"
1861 default 0 if BASE_FULL
1862 default 1 if !BASE_FULL
1865 bool "Enable loadable module support"
1868 Kernel modules are small pieces of compiled code which can
1869 be inserted in the running kernel, rather than being
1870 permanently built into the kernel. You use the "modprobe"
1871 tool to add (and sometimes remove) them. If you say Y here,
1872 many parts of the kernel can be built as modules (by
1873 answering M instead of Y where indicated): this is most
1874 useful for infrequently used options which are not required
1875 for booting. For more information, see the man pages for
1876 modprobe, lsmod, modinfo, insmod and rmmod.
1878 If you say Y here, you will need to run "make
1879 modules_install" to put the modules under /lib/modules/
1880 where modprobe can find them (you may need to be root to do
1887 config MODULE_FORCE_LOAD
1888 bool "Forced module loading"
1891 Allow loading of modules without version information (ie. modprobe
1892 --force). Forced module loading sets the 'F' (forced) taint flag and
1893 is usually a really bad idea.
1895 config MODULE_UNLOAD
1896 bool "Module unloading"
1898 Without this option you will not be able to unload any
1899 modules (note that some modules may not be unloadable
1900 anyway), which makes your kernel smaller, faster
1901 and simpler. If unsure, say Y.
1903 config MODULE_FORCE_UNLOAD
1904 bool "Forced module unloading"
1905 depends on MODULE_UNLOAD
1907 This option allows you to force a module to unload, even if the
1908 kernel believes it is unsafe: the kernel will remove the module
1909 without waiting for anyone to stop using it (using the -f option to
1910 rmmod). This is mainly for kernel developers and desperate users.
1914 bool "Module versioning support"
1916 Usually, you have to use modules compiled with your kernel.
1917 Saying Y here makes it sometimes possible to use modules
1918 compiled for different kernels, by adding enough information
1919 to the modules to (hopefully) spot any changes which would
1920 make them incompatible with the kernel you are running. If
1923 config MODULE_REL_CRCS
1925 depends on MODVERSIONS
1927 config MODULE_SRCVERSION_ALL
1928 bool "Source checksum for all modules"
1930 Modules which contain a MODULE_VERSION get an extra "srcversion"
1931 field inserted into their modinfo section, which contains a
1932 sum of the source files which made it. This helps maintainers
1933 see exactly which source was used to build a module (since
1934 others sometimes change the module source without updating
1935 the version). With this option, such a "srcversion" field
1936 will be created for all modules. If unsure, say N.
1939 bool "Module signature verification"
1941 select SYSTEM_DATA_VERIFICATION
1943 Check modules for valid signatures upon load: the signature
1944 is simply appended to the module. For more information see
1945 <file:Documentation/admin-guide/module-signing.rst>.
1947 Note that this option adds the OpenSSL development packages as a
1948 kernel build dependency so that the signing tool can use its crypto
1951 !!!WARNING!!! If you enable this option, you MUST make sure that the
1952 module DOES NOT get stripped after being signed. This includes the
1953 debuginfo strip done by some packagers (such as rpmbuild) and
1954 inclusion into an initramfs that wants the module size reduced.
1956 config MODULE_SIG_FORCE
1957 bool "Require modules to be validly signed"
1958 depends on MODULE_SIG
1960 Reject unsigned modules or signed modules for which we don't have a
1961 key. Without this, such modules will simply taint the kernel.
1963 config MODULE_SIG_ALL
1964 bool "Automatically sign all modules"
1966 depends on MODULE_SIG
1968 Sign all modules during make modules_install. Without this option,
1969 modules must be signed manually, using the scripts/sign-file tool.
1971 comment "Do not forget to sign required modules with scripts/sign-file"
1972 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
1975 prompt "Which hash algorithm should modules be signed with?"
1976 depends on MODULE_SIG
1978 This determines which sort of hashing algorithm will be used during
1979 signature generation. This algorithm _must_ be built into the kernel
1980 directly so that signature verification can take place. It is not
1981 possible to load a signed module containing the algorithm to check
1982 the signature on that module.
1984 config MODULE_SIG_SHA1
1985 bool "Sign modules with SHA-1"
1988 config MODULE_SIG_SHA224
1989 bool "Sign modules with SHA-224"
1990 select CRYPTO_SHA256
1992 config MODULE_SIG_SHA256
1993 bool "Sign modules with SHA-256"
1994 select CRYPTO_SHA256
1996 config MODULE_SIG_SHA384
1997 bool "Sign modules with SHA-384"
1998 select CRYPTO_SHA512
2000 config MODULE_SIG_SHA512
2001 bool "Sign modules with SHA-512"
2002 select CRYPTO_SHA512
2006 config MODULE_SIG_HASH
2008 depends on MODULE_SIG
2009 default "sha1" if MODULE_SIG_SHA1
2010 default "sha224" if MODULE_SIG_SHA224
2011 default "sha256" if MODULE_SIG_SHA256
2012 default "sha384" if MODULE_SIG_SHA384
2013 default "sha512" if MODULE_SIG_SHA512
2015 config MODULE_COMPRESS
2016 bool "Compress modules on installation"
2020 Compresses kernel modules when 'make modules_install' is run; gzip or
2021 xz depending on "Compression algorithm" below.
2023 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2025 Out-of-tree kernel modules installed using Kbuild will also be
2026 compressed upon installation.
2028 Note: for modules inside an initrd or initramfs, it's more efficient
2029 to compress the whole initrd or initramfs instead.
2031 Note: This is fully compatible with signed modules.
2036 prompt "Compression algorithm"
2037 depends on MODULE_COMPRESS
2038 default MODULE_COMPRESS_GZIP
2040 This determines which sort of compression will be used during
2041 'make modules_install'.
2043 GZIP (default) and XZ are supported.
2045 config MODULE_COMPRESS_GZIP
2048 config MODULE_COMPRESS_XZ
2053 config TRIM_UNUSED_KSYMS
2054 bool "Trim unused exported kernel symbols"
2055 depends on MODULES && !UNUSED_SYMBOLS
2057 The kernel and some modules make many symbols available for
2058 other modules to use via EXPORT_SYMBOL() and variants. Depending
2059 on the set of modules being selected in your kernel configuration,
2060 many of those exported symbols might never be used.
2062 This option allows for unused exported symbols to be dropped from
2063 the build. In turn, this provides the compiler more opportunities
2064 (especially when using LTO) for optimizing the code and reducing
2065 binary size. This might have some security advantages as well.
2067 If unsure, or if you need to build out-of-tree modules, say N.
2071 config MODULES_TREE_LOOKUP
2073 depends on PERF_EVENTS || TRACING
2075 config INIT_ALL_POSSIBLE
2078 Back when each arch used to define their own cpu_online_mask and
2079 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2080 with all 1s, and others with all 0s. When they were centralised,
2081 it was better to provide this option than to break all the archs
2082 and have several arch maintainers pursuing me down dark alleys.
2084 source "block/Kconfig"
2086 config PREEMPT_NOTIFIERS
2096 Build a simple ASN.1 grammar compiler that produces a bytecode output
2097 that can be interpreted by the ASN.1 stream decoder and used to
2098 inform it as to what tags are to be expected in a stream and what
2099 functions to call on what tags.
2101 source "kernel/Kconfig.locks"
2103 config ARCH_HAS_SYNC_CORE_BEFORE_USERMODE
2106 # It may be useful for an architecture to override the definitions of the
2107 # SYSCALL_DEFINE() and __SYSCALL_DEFINEx() macros in <linux/syscalls.h>
2108 # and the COMPAT_ variants in <linux/compat.h>, in particular to use a
2109 # different calling convention for syscalls. They can also override the
2110 # macros for not-implemented syscalls in kernel/sys_ni.c and
2111 # kernel/time/posix-stubs.c. All these overrides need to be available in
2112 # <asm/syscall_wrapper.h>.
2113 config ARCH_HAS_SYSCALL_WRAPPER