7 option env="KERNELVERSION"
13 default "/lib/modules/$UNAME_RELEASE/.config"
14 default "/etc/kernel-config"
15 default "/boot/config-$UNAME_RELEASE"
16 default "$ARCH_DEFCONFIG"
17 default "arch/$ARCH/defconfig"
28 depends on HAVE_IRQ_WORK
30 config BUILDTIME_EXTABLE_SORT
44 depends on BROKEN || !SMP
47 config INIT_ENV_ARG_LIMIT
52 Maximum of each of the number of arguments and environment
53 variables passed to init from the kernel command line.
57 string "Cross-compiler tool prefix"
59 Same as running 'make CROSS_COMPILE=prefix-' but stored for
60 default make runs in this kernel build directory. You don't
61 need to set this unless you want the configured kernel build
62 directory to select the cross-compiler automatically.
65 string "Local version - append to kernel release"
67 Append an extra string to the end of your kernel version.
68 This will show up when you type uname, for example.
69 The string you set here will be appended after the contents of
70 any files with a filename matching localversion* in your
71 object and source tree, in that order. Your total string can
72 be a maximum of 64 characters.
74 config LOCALVERSION_AUTO
75 bool "Automatically append version information to the version string"
78 This will try to automatically determine if the current tree is a
79 release tree by looking for git tags that belong to the current
82 A string of the format -gxxxxxxxx will be added to the localversion
83 if a git-based tree is found. The string generated by this will be
84 appended after any matching localversion* files, and after the value
85 set in CONFIG_LOCALVERSION.
87 (The actual string used here is the first eight characters produced
88 by running the command:
90 $ git rev-parse --verify HEAD
92 which is done within the script "scripts/setlocalversion".)
94 config HAVE_KERNEL_GZIP
97 config HAVE_KERNEL_BZIP2
100 config HAVE_KERNEL_LZMA
103 config HAVE_KERNEL_XZ
106 config HAVE_KERNEL_LZO
110 prompt "Kernel compression mode"
112 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO
114 The linux kernel is a kind of self-extracting executable.
115 Several compression algorithms are available, which differ
116 in efficiency, compression and decompression speed.
117 Compression speed is only relevant when building a kernel.
118 Decompression speed is relevant at each boot.
120 If you have any problems with bzip2 or lzma compressed
121 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
122 version of this functionality (bzip2 only), for 2.4, was
123 supplied by Christian Ludwig)
125 High compression options are mostly useful for users, who
126 are low on disk space (embedded systems), but for whom ram
129 If in doubt, select 'gzip'
133 depends on HAVE_KERNEL_GZIP
135 The old and tried gzip compression. It provides a good balance
136 between compression ratio and decompression speed.
140 depends on HAVE_KERNEL_BZIP2
142 Its compression ratio and speed is intermediate.
143 Decompression speed is slowest among the choices. The kernel
144 size is about 10% smaller with bzip2, in comparison to gzip.
145 Bzip2 uses a large amount of memory. For modern kernels you
146 will need at least 8MB RAM or more for booting.
150 depends on HAVE_KERNEL_LZMA
152 This compression algorithm's ratio is best. Decompression speed
153 is between gzip and bzip2. Compression is slowest.
154 The kernel size is about 33% smaller with LZMA in comparison to gzip.
158 depends on HAVE_KERNEL_XZ
160 XZ uses the LZMA2 algorithm and instruction set specific
161 BCJ filters which can improve compression ratio of executable
162 code. The size of the kernel is about 30% smaller with XZ in
163 comparison to gzip. On architectures for which there is a BCJ
164 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
165 will create a few percent smaller kernel than plain LZMA.
167 The speed is about the same as with LZMA: The decompression
168 speed of XZ is better than that of bzip2 but worse than gzip
169 and LZO. Compression is slow.
173 depends on HAVE_KERNEL_LZO
175 Its compression ratio is the poorest among the choices. The kernel
176 size is about 10% bigger than gzip; however its speed
177 (both compression and decompression) is the fastest.
181 config DEFAULT_HOSTNAME
182 string "Default hostname"
185 This option determines the default system hostname before userspace
186 calls sethostname(2). The kernel traditionally uses "(none)" here,
187 but you may wish to use a different default here to make a minimal
188 system more usable with less configuration.
191 bool "Support for paging of anonymous memory (swap)"
192 depends on MMU && BLOCK
195 This option allows you to choose whether you want to have support
196 for so called swap devices or swap files in your kernel that are
197 used to provide more virtual memory than the actual RAM present
198 in your computer. If unsure say Y.
203 Inter Process Communication is a suite of library functions and
204 system calls which let processes (running programs) synchronize and
205 exchange information. It is generally considered to be a good thing,
206 and some programs won't run unless you say Y here. In particular, if
207 you want to run the DOS emulator dosemu under Linux (read the
208 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
209 you'll need to say Y here.
211 You can find documentation about IPC with "info ipc" and also in
212 section 6.4 of the Linux Programmer's Guide, available from
213 <http://www.tldp.org/guides.html>.
215 config SYSVIPC_SYSCTL
222 bool "POSIX Message Queues"
223 depends on NET && EXPERIMENTAL
225 POSIX variant of message queues is a part of IPC. In POSIX message
226 queues every message has a priority which decides about succession
227 of receiving it by a process. If you want to compile and run
228 programs written e.g. for Solaris with use of its POSIX message
229 queues (functions mq_*) say Y here.
231 POSIX message queues are visible as a filesystem called 'mqueue'
232 and can be mounted somewhere if you want to do filesystem
233 operations on message queues.
237 config POSIX_MQUEUE_SYSCTL
239 depends on POSIX_MQUEUE
244 bool "open by fhandle syscalls"
247 If you say Y here, a user level program will be able to map
248 file names to handle and then later use the handle for
249 different file system operations. This is useful in implementing
250 userspace file servers, which now track files using handles instead
251 of names. The handle would remain the same even if file names
252 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
256 bool "Auditing support"
259 Enable auditing infrastructure that can be used with another
260 kernel subsystem, such as SELinux (which requires this for
261 logging of avc messages output). Does not do system-call
262 auditing without CONFIG_AUDITSYSCALL.
265 bool "Enable system-call auditing support"
266 depends on AUDIT && (X86 || PPC || S390 || IA64 || UML || SPARC64 || SUPERH || (ARM && AEABI && !OABI_COMPAT))
267 default y if SECURITY_SELINUX
269 Enable low-overhead system-call auditing infrastructure that
270 can be used independently or with another kernel subsystem,
275 depends on AUDITSYSCALL
280 depends on AUDITSYSCALL
283 config AUDIT_LOGINUID_IMMUTABLE
284 bool "Make audit loginuid immutable"
287 The config option toggles if a task setting its loginuid requires
288 CAP_SYS_AUDITCONTROL or if that task should require no special permissions
289 but should instead only allow setting its loginuid if it was never
290 previously set. On systems which use systemd or a similar central
291 process to restart login services this should be set to true. On older
292 systems in which an admin would typically have to directly stop and
293 start processes this should be set to false. Setting this to true allows
294 one to drop potentially dangerous capabilites from the login tasks,
295 but may not be backwards compatible with older init systems.
297 source "kernel/irq/Kconfig"
298 source "kernel/time/Kconfig"
300 menu "CPU/Task time and stats accounting"
303 prompt "Cputime accounting"
304 default TICK_CPU_ACCOUNTING if !PPC64
305 default VIRT_CPU_ACCOUNTING if PPC64
307 # Kind of a stub config for the pure tick based cputime accounting
308 config TICK_CPU_ACCOUNTING
309 bool "Simple tick based cputime accounting"
312 This is the basic tick based cputime accounting that maintains
313 statistics about user, system and idle time spent on per jiffies
318 config VIRT_CPU_ACCOUNTING
319 bool "Deterministic task and CPU time accounting"
320 depends on HAVE_VIRT_CPU_ACCOUNTING
322 Select this option to enable more accurate task and CPU time
323 accounting. This is done by reading a CPU counter on each
324 kernel entry and exit and on transitions within the kernel
325 between system, softirq and hardirq state, so there is a
326 small performance impact. In the case of s390 or IBM POWER > 5,
327 this also enables accounting of stolen time on logically-partitioned
330 config IRQ_TIME_ACCOUNTING
331 bool "Fine granularity task level IRQ time accounting"
332 depends on HAVE_IRQ_TIME_ACCOUNTING
334 Select this option to enable fine granularity task irq time
335 accounting. This is done by reading a timestamp on each
336 transitions between softirq and hardirq state, so there can be a
337 small performance impact.
339 If in doubt, say N here.
343 config BSD_PROCESS_ACCT
344 bool "BSD Process Accounting"
346 If you say Y here, a user level program will be able to instruct the
347 kernel (via a special system call) to write process accounting
348 information to a file: whenever a process exits, information about
349 that process will be appended to the file by the kernel. The
350 information includes things such as creation time, owning user,
351 command name, memory usage, controlling terminal etc. (the complete
352 list is in the struct acct in <file:include/linux/acct.h>). It is
353 up to the user level program to do useful things with this
354 information. This is generally a good idea, so say Y.
356 config BSD_PROCESS_ACCT_V3
357 bool "BSD Process Accounting version 3 file format"
358 depends on BSD_PROCESS_ACCT
361 If you say Y here, the process accounting information is written
362 in a new file format that also logs the process IDs of each
363 process and it's parent. Note that this file format is incompatible
364 with previous v0/v1/v2 file formats, so you will need updated tools
365 for processing it. A preliminary version of these tools is available
366 at <http://www.gnu.org/software/acct/>.
369 bool "Export task/process statistics through netlink (EXPERIMENTAL)"
373 Export selected statistics for tasks/processes through the
374 generic netlink interface. Unlike BSD process accounting, the
375 statistics are available during the lifetime of tasks/processes as
376 responses to commands. Like BSD accounting, they are sent to user
381 config TASK_DELAY_ACCT
382 bool "Enable per-task delay accounting (EXPERIMENTAL)"
385 Collect information on time spent by a task waiting for system
386 resources like cpu, synchronous block I/O completion and swapping
387 in pages. Such statistics can help in setting a task's priorities
388 relative to other tasks for cpu, io, rss limits etc.
393 bool "Enable extended accounting over taskstats (EXPERIMENTAL)"
396 Collect extended task accounting data and send the data
397 to userland for processing over the taskstats interface.
401 config TASK_IO_ACCOUNTING
402 bool "Enable per-task storage I/O accounting (EXPERIMENTAL)"
403 depends on TASK_XACCT
405 Collect information on the number of bytes of storage I/O which this
410 endmenu # "CPU/Task time and stats accounting"
415 prompt "RCU Implementation"
419 bool "Tree-based hierarchical RCU"
420 depends on !PREEMPT && SMP
422 This option selects the RCU implementation that is
423 designed for very large SMP system with hundreds or
424 thousands of CPUs. It also scales down nicely to
427 config TREE_PREEMPT_RCU
428 bool "Preemptible tree-based hierarchical RCU"
429 depends on PREEMPT && SMP
431 This option selects the RCU implementation that is
432 designed for very large SMP systems with hundreds or
433 thousands of CPUs, but for which real-time response
434 is also required. It also scales down nicely to
438 bool "UP-only small-memory-footprint RCU"
439 depends on !PREEMPT && !SMP
441 This option selects the RCU implementation that is
442 designed for UP systems from which real-time response
443 is not required. This option greatly reduces the
444 memory footprint of RCU.
446 config TINY_PREEMPT_RCU
447 bool "Preemptible UP-only small-memory-footprint RCU"
448 depends on PREEMPT && !SMP
450 This option selects the RCU implementation that is designed
451 for real-time UP systems. This option greatly reduces the
452 memory footprint of RCU.
457 def_bool ( TREE_PREEMPT_RCU || TINY_PREEMPT_RCU )
459 This option enables preemptible-RCU code that is common between
460 the TREE_PREEMPT_RCU and TINY_PREEMPT_RCU implementations.
462 config CONTEXT_TRACKING
466 bool "Consider userspace as in RCU extended quiescent state"
467 depends on HAVE_CONTEXT_TRACKING && SMP
468 select CONTEXT_TRACKING
470 This option sets hooks on kernel / userspace boundaries and
471 puts RCU in extended quiescent state when the CPU runs in
472 userspace. It means that when a CPU runs in userspace, it is
473 excluded from the global RCU state machine and thus doesn't
474 try to keep the timer tick on for RCU.
476 Unless you want to hack and help the development of the full
477 dynticks mode, you shouldn't enable this option. It also
478 adds unnecessary overhead.
482 config CONTEXT_TRACKING_FORCE
483 bool "Force context tracking"
484 depends on CONTEXT_TRACKING
486 Probe on user/kernel boundaries by default in order to
487 test the features that rely on it such as userspace RCU extended
489 This test is there for debugging until we have a real user like the
493 int "Tree-based hierarchical RCU fanout value"
496 depends on TREE_RCU || TREE_PREEMPT_RCU
500 This option controls the fanout of hierarchical implementations
501 of RCU, allowing RCU to work efficiently on machines with
502 large numbers of CPUs. This value must be at least the fourth
503 root of NR_CPUS, which allows NR_CPUS to be insanely large.
504 The default value of RCU_FANOUT should be used for production
505 systems, but if you are stress-testing the RCU implementation
506 itself, small RCU_FANOUT values allow you to test large-system
507 code paths on small(er) systems.
509 Select a specific number if testing RCU itself.
510 Take the default if unsure.
512 config RCU_FANOUT_LEAF
513 int "Tree-based hierarchical RCU leaf-level fanout value"
514 range 2 RCU_FANOUT if 64BIT
515 range 2 RCU_FANOUT if !64BIT
516 depends on TREE_RCU || TREE_PREEMPT_RCU
519 This option controls the leaf-level fanout of hierarchical
520 implementations of RCU, and allows trading off cache misses
521 against lock contention. Systems that synchronize their
522 scheduling-clock interrupts for energy-efficiency reasons will
523 want the default because the smaller leaf-level fanout keeps
524 lock contention levels acceptably low. Very large systems
525 (hundreds or thousands of CPUs) will instead want to set this
526 value to the maximum value possible in order to reduce the
527 number of cache misses incurred during RCU's grace-period
528 initialization. These systems tend to run CPU-bound, and thus
529 are not helped by synchronized interrupts, and thus tend to
530 skew them, which reduces lock contention enough that large
531 leaf-level fanouts work well.
533 Select a specific number if testing RCU itself.
535 Select the maximum permissible value for large systems.
537 Take the default if unsure.
539 config RCU_FANOUT_EXACT
540 bool "Disable tree-based hierarchical RCU auto-balancing"
541 depends on TREE_RCU || TREE_PREEMPT_RCU
544 This option forces use of the exact RCU_FANOUT value specified,
545 regardless of imbalances in the hierarchy. This is useful for
546 testing RCU itself, and might one day be useful on systems with
547 strong NUMA behavior.
549 Without RCU_FANOUT_EXACT, the code will balance the hierarchy.
553 config RCU_FAST_NO_HZ
554 bool "Accelerate last non-dyntick-idle CPU's grace periods"
555 depends on NO_HZ && SMP
558 This option causes RCU to attempt to accelerate grace periods in
559 order to allow CPUs to enter dynticks-idle state more quickly.
560 On the other hand, this option increases the overhead of the
561 dynticks-idle checking, thus degrading scheduling latency.
563 Say Y if energy efficiency is critically important, and you don't
564 care about real-time response.
566 Say N if you are unsure.
568 config TREE_RCU_TRACE
569 def_bool RCU_TRACE && ( TREE_RCU || TREE_PREEMPT_RCU )
572 This option provides tracing for the TREE_RCU and
573 TREE_PREEMPT_RCU implementations, permitting Makefile to
574 trivially select kernel/rcutree_trace.c.
577 bool "Enable RCU priority boosting"
578 depends on RT_MUTEXES && PREEMPT_RCU
581 This option boosts the priority of preempted RCU readers that
582 block the current preemptible RCU grace period for too long.
583 This option also prevents heavy loads from blocking RCU
584 callback invocation for all flavors of RCU.
586 Say Y here if you are working with real-time apps or heavy loads
587 Say N here if you are unsure.
589 config RCU_BOOST_PRIO
590 int "Real-time priority to boost RCU readers to"
595 This option specifies the real-time priority to which long-term
596 preempted RCU readers are to be boosted. If you are working
597 with a real-time application that has one or more CPU-bound
598 threads running at a real-time priority level, you should set
599 RCU_BOOST_PRIO to a priority higher then the highest-priority
600 real-time CPU-bound thread. The default RCU_BOOST_PRIO value
601 of 1 is appropriate in the common case, which is real-time
602 applications that do not have any CPU-bound threads.
604 Some real-time applications might not have a single real-time
605 thread that saturates a given CPU, but instead might have
606 multiple real-time threads that, taken together, fully utilize
607 that CPU. In this case, you should set RCU_BOOST_PRIO to
608 a priority higher than the lowest-priority thread that is
609 conspiring to prevent the CPU from running any non-real-time
610 tasks. For example, if one thread at priority 10 and another
611 thread at priority 5 are between themselves fully consuming
612 the CPU time on a given CPU, then RCU_BOOST_PRIO should be
613 set to priority 6 or higher.
615 Specify the real-time priority, or take the default if unsure.
617 config RCU_BOOST_DELAY
618 int "Milliseconds to delay boosting after RCU grace-period start"
623 This option specifies the time to wait after the beginning of
624 a given grace period before priority-boosting preempted RCU
625 readers blocking that grace period. Note that any RCU reader
626 blocking an expedited RCU grace period is boosted immediately.
628 Accept the default if unsure.
631 bool "Offload RCU callback processing from boot-selected CPUs"
632 depends on TREE_RCU || TREE_PREEMPT_RCU
635 Use this option to reduce OS jitter for aggressive HPC or
636 real-time workloads. It can also be used to offload RCU
637 callback invocation to energy-efficient CPUs in battery-powered
638 asymmetric multiprocessors.
640 This option offloads callback invocation from the set of
641 CPUs specified at boot time by the rcu_nocbs parameter.
642 For each such CPU, a kthread ("rcuoN") will be created to
643 invoke callbacks, where the "N" is the CPU being offloaded.
644 Nothing prevents this kthread from running on the specified
645 CPUs, but (1) the kthreads may be preempted between each
646 callback, and (2) affinity or cgroups can be used to force
647 the kthreads to run on whatever set of CPUs is desired.
649 Say Y here if you want reduced OS jitter on selected CPUs.
650 Say N here if you are unsure.
652 endmenu # "RCU Subsystem"
655 tristate "Kernel .config support"
657 This option enables the complete Linux kernel ".config" file
658 contents to be saved in the kernel. It provides documentation
659 of which kernel options are used in a running kernel or in an
660 on-disk kernel. This information can be extracted from the kernel
661 image file with the script scripts/extract-ikconfig and used as
662 input to rebuild the current kernel or to build another kernel.
663 It can also be extracted from a running kernel by reading
664 /proc/config.gz if enabled (below).
667 bool "Enable access to .config through /proc/config.gz"
668 depends on IKCONFIG && PROC_FS
670 This option enables access to the kernel configuration file
671 through /proc/config.gz.
674 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
678 Select kernel log buffer size as a power of 2.
688 # Architectures with an unreliable sched_clock() should select this:
690 config HAVE_UNSTABLE_SCHED_CLOCK
694 # For architectures that want to enable the support for NUMA-affine scheduler
697 config ARCH_SUPPORTS_NUMA_BALANCING
700 # For architectures that (ab)use NUMA to represent different memory regions
701 # all cpu-local but of different latencies, such as SuperH.
703 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
707 # For architectures that are willing to define _PAGE_NUMA as _PAGE_PROTNONE
708 config ARCH_WANTS_PROT_NUMA_PROT_NONE
711 config ARCH_USES_NUMA_PROT_NONE
714 depends on ARCH_WANTS_PROT_NUMA_PROT_NONE
715 depends on NUMA_BALANCING
717 config NUMA_BALANCING_DEFAULT_ENABLED
718 bool "Automatically enable NUMA aware memory/task placement"
720 depends on NUMA_BALANCING
722 If set, autonumic NUMA balancing will be enabled if running on a NUMA
725 config NUMA_BALANCING
726 bool "Memory placement aware NUMA scheduler"
727 depends on ARCH_SUPPORTS_NUMA_BALANCING
728 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
729 depends on SMP && NUMA && MIGRATION
731 This option adds support for automatic NUMA aware memory/task placement.
732 The mechanism is quite primitive and is based on migrating memory when
733 it is references to the node the task is running on.
735 This system will be inactive on UMA systems.
738 boolean "Control Group support"
741 This option adds support for grouping sets of processes together, for
742 use with process control subsystems such as Cpusets, CFS, memory
743 controls or device isolation.
745 - Documentation/scheduler/sched-design-CFS.txt (CFS)
746 - Documentation/cgroups/ (features for grouping, isolation
747 and resource control)
754 bool "Example debug cgroup subsystem"
757 This option enables a simple cgroup subsystem that
758 exports useful debugging information about the cgroups
763 config CGROUP_FREEZER
764 bool "Freezer cgroup subsystem"
766 Provides a way to freeze and unfreeze all tasks in a
770 bool "Device controller for cgroups"
772 Provides a cgroup implementing whitelists for devices which
773 a process in the cgroup can mknod or open.
776 bool "Cpuset support"
778 This option will let you create and manage CPUSETs which
779 allow dynamically partitioning a system into sets of CPUs and
780 Memory Nodes and assigning tasks to run only within those sets.
781 This is primarily useful on large SMP or NUMA systems.
785 config PROC_PID_CPUSET
786 bool "Include legacy /proc/<pid>/cpuset file"
790 config CGROUP_CPUACCT
791 bool "Simple CPU accounting cgroup subsystem"
793 Provides a simple Resource Controller for monitoring the
794 total CPU consumed by the tasks in a cgroup.
796 config RESOURCE_COUNTERS
797 bool "Resource counters"
799 This option enables controller independent resource accounting
800 infrastructure that works with cgroups.
803 bool "Memory Resource Controller for Control Groups"
804 depends on RESOURCE_COUNTERS
807 Provides a memory resource controller that manages both anonymous
808 memory and page cache. (See Documentation/cgroups/memory.txt)
810 Note that setting this option increases fixed memory overhead
811 associated with each page of memory in the system. By this,
812 20(40)bytes/PAGE_SIZE on 32(64)bit system will be occupied by memory
813 usage tracking struct at boot. Total amount of this is printed out
816 Only enable when you're ok with these trade offs and really
817 sure you need the memory resource controller. Even when you enable
818 this, you can set "cgroup_disable=memory" at your boot option to
819 disable memory resource controller and you can avoid overheads.
820 (and lose benefits of memory resource controller)
822 This config option also selects MM_OWNER config option, which
823 could in turn add some fork/exit overhead.
826 bool "Memory Resource Controller Swap Extension"
827 depends on MEMCG && SWAP
829 Add swap management feature to memory resource controller. When you
830 enable this, you can limit mem+swap usage per cgroup. In other words,
831 when you disable this, memory resource controller has no cares to
832 usage of swap...a process can exhaust all of the swap. This extension
833 is useful when you want to avoid exhaustion swap but this itself
834 adds more overheads and consumes memory for remembering information.
835 Especially if you use 32bit system or small memory system, please
836 be careful about enabling this. When memory resource controller
837 is disabled by boot option, this will be automatically disabled and
838 there will be no overhead from this. Even when you set this config=y,
839 if boot option "swapaccount=0" is set, swap will not be accounted.
840 Now, memory usage of swap_cgroup is 2 bytes per entry. If swap page
841 size is 4096bytes, 512k per 1Gbytes of swap.
842 config MEMCG_SWAP_ENABLED
843 bool "Memory Resource Controller Swap Extension enabled by default"
844 depends on MEMCG_SWAP
847 Memory Resource Controller Swap Extension comes with its price in
848 a bigger memory consumption. General purpose distribution kernels
849 which want to enable the feature but keep it disabled by default
850 and let the user enable it by swapaccount boot command line
851 parameter should have this option unselected.
852 For those who want to have the feature enabled by default should
853 select this option (if, for some reason, they need to disable it
854 then swapaccount=0 does the trick).
856 bool "Memory Resource Controller Kernel Memory accounting (EXPERIMENTAL)"
857 depends on MEMCG && EXPERIMENTAL
858 depends on SLUB || SLAB
860 The Kernel Memory extension for Memory Resource Controller can limit
861 the amount of memory used by kernel objects in the system. Those are
862 fundamentally different from the entities handled by the standard
863 Memory Controller, which are page-based, and can be swapped. Users of
864 the kmem extension can use it to guarantee that no group of processes
865 will ever exhaust kernel resources alone.
867 config CGROUP_HUGETLB
868 bool "HugeTLB Resource Controller for Control Groups"
869 depends on RESOURCE_COUNTERS && HUGETLB_PAGE && EXPERIMENTAL
872 Provides a cgroup Resource Controller for HugeTLB pages.
873 When you enable this, you can put a per cgroup limit on HugeTLB usage.
874 The limit is enforced during page fault. Since HugeTLB doesn't
875 support page reclaim, enforcing the limit at page fault time implies
876 that, the application will get SIGBUS signal if it tries to access
877 HugeTLB pages beyond its limit. This requires the application to know
878 beforehand how much HugeTLB pages it would require for its use. The
879 control group is tracked in the third page lru pointer. This means
880 that we cannot use the controller with huge page less than 3 pages.
883 bool "Enable perf_event per-cpu per-container group (cgroup) monitoring"
884 depends on PERF_EVENTS && CGROUPS
886 This option extends the per-cpu mode to restrict monitoring to
887 threads which belong to the cgroup specified and run on the
892 menuconfig CGROUP_SCHED
893 bool "Group CPU scheduler"
896 This feature lets CPU scheduler recognize task groups and control CPU
897 bandwidth allocation to such task groups. It uses cgroups to group
901 config FAIR_GROUP_SCHED
902 bool "Group scheduling for SCHED_OTHER"
903 depends on CGROUP_SCHED
907 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
908 depends on EXPERIMENTAL
909 depends on FAIR_GROUP_SCHED
912 This option allows users to define CPU bandwidth rates (limits) for
913 tasks running within the fair group scheduler. Groups with no limit
914 set are considered to be unconstrained and will run with no
916 See tip/Documentation/scheduler/sched-bwc.txt for more information.
918 config RT_GROUP_SCHED
919 bool "Group scheduling for SCHED_RR/FIFO"
920 depends on EXPERIMENTAL
921 depends on CGROUP_SCHED
924 This feature lets you explicitly allocate real CPU bandwidth
925 to task groups. If enabled, it will also make it impossible to
926 schedule realtime tasks for non-root users until you allocate
927 realtime bandwidth for them.
928 See Documentation/scheduler/sched-rt-group.txt for more information.
933 bool "Block IO controller"
937 Generic block IO controller cgroup interface. This is the common
938 cgroup interface which should be used by various IO controlling
941 Currently, CFQ IO scheduler uses it to recognize task groups and
942 control disk bandwidth allocation (proportional time slice allocation)
943 to such task groups. It is also used by bio throttling logic in
944 block layer to implement upper limit in IO rates on a device.
946 This option only enables generic Block IO controller infrastructure.
947 One needs to also enable actual IO controlling logic/policy. For
948 enabling proportional weight division of disk bandwidth in CFQ, set
949 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
950 CONFIG_BLK_DEV_THROTTLING=y.
952 See Documentation/cgroups/blkio-controller.txt for more information.
954 config DEBUG_BLK_CGROUP
955 bool "Enable Block IO controller debugging"
956 depends on BLK_CGROUP
959 Enable some debugging help. Currently it exports additional stat
960 files in a cgroup which can be useful for debugging.
964 config CHECKPOINT_RESTORE
965 bool "Checkpoint/restore support" if EXPERT
968 Enables additional kernel features in a sake of checkpoint/restore.
969 In particular it adds auxiliary prctl codes to setup process text,
970 data and heap segment sizes, and a few additional /proc filesystem
973 If unsure, say N here.
975 menuconfig NAMESPACES
976 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 (EXPERIMENTAL)"
1003 depends on EXPERIMENTAL
1004 depends on UIDGID_CONVERTED
1005 select UIDGID_STRICT_TYPE_CHECKS
1009 This allows containers, i.e. vservers, to use user namespaces
1010 to provide different user info for different servers.
1014 bool "PID Namespaces"
1017 Support process id namespaces. This allows having multiple
1018 processes with the same pid as long as they are in different
1019 pid namespaces. This is a building block of containers.
1022 bool "Network namespace"
1026 Allow user space to create what appear to be multiple instances
1027 of the network stack.
1031 config UIDGID_CONVERTED
1032 # True if all of the selected software conmponents are known
1033 # to have uid_t and gid_t converted to kuid_t and kgid_t
1034 # where appropriate and are otherwise safe to use with
1035 # the user namespace.
1040 depends on NET_9P = n
1043 depends on 9P_FS = n
1044 depends on AFS_FS = n
1045 depends on CEPH_FS = n
1047 depends on CODA_FS = n
1048 depends on GFS2_FS = n
1049 depends on NCP_FS = n
1051 depends on NFS_FS = n
1052 depends on OCFS2_FS = n
1053 depends on XFS_FS = n
1055 config UIDGID_STRICT_TYPE_CHECKS
1056 bool "Require conversions between uid/gids and their internal representation"
1057 depends on UIDGID_CONVERTED
1060 While the nececessary conversions are being added to all subsystems this option allows
1061 the code to continue to build for unconverted subsystems.
1063 Say Y here if you want the strict type checking enabled
1065 config SCHED_AUTOGROUP
1066 bool "Automatic process group scheduling"
1070 select FAIR_GROUP_SCHED
1072 This option optimizes the scheduler for common desktop workloads by
1073 automatically creating and populating task groups. This separation
1074 of workloads isolates aggressive CPU burners (like build jobs) from
1075 desktop applications. Task group autogeneration is currently based
1081 config SYSFS_DEPRECATED
1082 bool "Enable deprecated sysfs features to support old userspace tools"
1086 This option adds code that switches the layout of the "block" class
1087 devices, to not show up in /sys/class/block/, but only in
1090 This switch is only active when the sysfs.deprecated=1 boot option is
1091 passed or the SYSFS_DEPRECATED_V2 option is set.
1093 This option allows new kernels to run on old distributions and tools,
1094 which might get confused by /sys/class/block/. Since 2007/2008 all
1095 major distributions and tools handle this just fine.
1097 Recent distributions and userspace tools after 2009/2010 depend on
1098 the existence of /sys/class/block/, and will not work with this
1101 Only if you are using a new kernel on an old distribution, you might
1104 config SYSFS_DEPRECATED_V2
1105 bool "Enable deprecated sysfs features by default"
1108 depends on SYSFS_DEPRECATED
1110 Enable deprecated sysfs by default.
1112 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1115 Only if you are using a new kernel on an old distribution, you might
1116 need to say Y here. Even then, odds are you would not need it
1117 enabled, you can always pass the boot option if absolutely necessary.
1120 bool "Kernel->user space relay support (formerly relayfs)"
1122 This option enables support for relay interface support in
1123 certain file systems (such as debugfs).
1124 It is designed to provide an efficient mechanism for tools and
1125 facilities to relay large amounts of data from kernel space to
1130 config BLK_DEV_INITRD
1131 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1132 depends on BROKEN || !FRV
1134 The initial RAM filesystem is a ramfs which is loaded by the
1135 boot loader (loadlin or lilo) and that is mounted as root
1136 before the normal boot procedure. It is typically used to
1137 load modules needed to mount the "real" root file system,
1138 etc. See <file:Documentation/initrd.txt> for details.
1140 If RAM disk support (BLK_DEV_RAM) is also included, this
1141 also enables initial RAM disk (initrd) support and adds
1142 15 Kbytes (more on some other architectures) to the kernel size.
1148 source "usr/Kconfig"
1152 config CC_OPTIMIZE_FOR_SIZE
1153 bool "Optimize for size"
1155 Enabling this option will pass "-Os" instead of "-O2" to gcc
1156 resulting in a smaller kernel.
1167 bool "Configure standard kernel features (expert users)"
1168 # Unhide debug options, to make the on-by-default options visible
1171 This option allows certain base kernel options and settings
1172 to be disabled or tweaked. This is for specialized
1173 environments which can tolerate a "non-standard" kernel.
1174 Only use this if you really know what you are doing.
1180 bool "Enable 16-bit UID system calls" if EXPERT
1181 depends on HAVE_UID16
1184 This enables the legacy 16-bit UID syscall wrappers.
1186 config SYSCTL_SYSCALL
1187 bool "Sysctl syscall support" if EXPERT
1188 depends on PROC_SYSCTL
1192 sys_sysctl uses binary paths that have been found challenging
1193 to properly maintain and use. The interface in /proc/sys
1194 using paths with ascii names is now the primary path to this
1197 Almost nothing using the binary sysctl interface so if you are
1198 trying to save some space it is probably safe to disable this,
1199 making your kernel marginally smaller.
1201 If unsure say N here.
1203 config SYSCTL_EXCEPTION_TRACE
1206 Enable support for /proc/sys/debug/exception-trace.
1209 bool "Load all symbols for debugging/ksymoops" if EXPERT
1212 Say Y here to let the kernel print out symbolic crash information and
1213 symbolic stack backtraces. This increases the size of the kernel
1214 somewhat, as all symbols have to be loaded into the kernel image.
1217 bool "Include all symbols in kallsyms"
1218 depends on DEBUG_KERNEL && KALLSYMS
1220 Normally kallsyms only contains the symbols of functions for nicer
1221 OOPS messages and backtraces (i.e., symbols from the text and inittext
1222 sections). This is sufficient for most cases. And only in very rare
1223 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1224 names of variables from the data sections, etc).
1226 This option makes sure that all symbols are loaded into the kernel
1227 image (i.e., symbols from all sections) in cost of increased kernel
1228 size (depending on the kernel configuration, it may be 300KiB or
1229 something like this).
1231 Say N unless you really need all symbols.
1238 bool "Enable support for printk" if EXPERT
1240 This option enables normal printk support. Removing it
1241 eliminates most of the message strings from the kernel image
1242 and makes the kernel more or less silent. As this makes it
1243 very difficult to diagnose system problems, saying N here is
1244 strongly discouraged.
1247 bool "BUG() support" if EXPERT
1250 Disabling this option eliminates support for BUG and WARN, reducing
1251 the size of your kernel image and potentially quietly ignoring
1252 numerous fatal conditions. You should only consider disabling this
1253 option for embedded systems with no facilities for reporting errors.
1259 bool "Enable ELF core dumps" if EXPERT
1261 Enable support for generating core dumps. Disabling saves about 4k.
1264 config PCSPKR_PLATFORM
1265 bool "Enable PC-Speaker support" if EXPERT
1266 depends on HAVE_PCSPKR_PLATFORM
1270 This option allows to disable the internal PC-Speaker
1271 support, saving some memory.
1273 config HAVE_PCSPKR_PLATFORM
1278 bool "Enable full-sized data structures for core" if EXPERT
1280 Disabling this option reduces the size of miscellaneous core
1281 kernel data structures. This saves memory on small machines,
1282 but may reduce performance.
1285 bool "Enable futex support" if EXPERT
1289 Disabling this option will cause the kernel to be built without
1290 support for "fast userspace mutexes". The resulting kernel may not
1291 run glibc-based applications correctly.
1294 bool "Enable eventpoll support" if EXPERT
1298 Disabling this option will cause the kernel to be built without
1299 support for epoll family of system calls.
1302 bool "Enable signalfd() system call" if EXPERT
1306 Enable the signalfd() system call that allows to receive signals
1307 on a file descriptor.
1312 bool "Enable timerfd() system call" if EXPERT
1316 Enable the timerfd() system call that allows to receive timer
1317 events on a file descriptor.
1322 bool "Enable eventfd() system call" if EXPERT
1326 Enable the eventfd() system call that allows to receive both
1327 kernel notification (ie. KAIO) or userspace notifications.
1332 bool "Use full shmem filesystem" if EXPERT
1336 The shmem is an internal filesystem used to manage shared memory.
1337 It is backed by swap and manages resource limits. It is also exported
1338 to userspace as tmpfs if TMPFS is enabled. Disabling this
1339 option replaces shmem and tmpfs with the much simpler ramfs code,
1340 which may be appropriate on small systems without swap.
1343 bool "Enable AIO support" if EXPERT
1346 This option enables POSIX asynchronous I/O which may by used
1347 by some high performance threaded applications. Disabling
1348 this option saves about 7k.
1351 bool "Embedded system"
1354 This option should be enabled if compiling the kernel for
1355 an embedded system so certain expert options are available
1358 config HAVE_PERF_EVENTS
1361 See tools/perf/design.txt for details.
1363 config PERF_USE_VMALLOC
1366 See tools/perf/design.txt for details
1368 menu "Kernel Performance Events And Counters"
1371 bool "Kernel performance events and counters"
1372 default y if PROFILING
1373 depends on HAVE_PERF_EVENTS
1377 Enable kernel support for various performance events provided
1378 by software and hardware.
1380 Software events are supported either built-in or via the
1381 use of generic tracepoints.
1383 Most modern CPUs support performance events via performance
1384 counter registers. These registers count the number of certain
1385 types of hw events: such as instructions executed, cachemisses
1386 suffered, or branches mis-predicted - without slowing down the
1387 kernel or applications. These registers can also trigger interrupts
1388 when a threshold number of events have passed - and can thus be
1389 used to profile the code that runs on that CPU.
1391 The Linux Performance Event subsystem provides an abstraction of
1392 these software and hardware event capabilities, available via a
1393 system call and used by the "perf" utility in tools/perf/. It
1394 provides per task and per CPU counters, and it provides event
1395 capabilities on top of those.
1399 config DEBUG_PERF_USE_VMALLOC
1401 bool "Debug: use vmalloc to back perf mmap() buffers"
1402 depends on PERF_EVENTS && DEBUG_KERNEL
1403 select PERF_USE_VMALLOC
1405 Use vmalloc memory to back perf mmap() buffers.
1407 Mostly useful for debugging the vmalloc code on platforms
1408 that don't require it.
1414 config VM_EVENT_COUNTERS
1416 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1418 VM event counters are needed for event counts to be shown.
1419 This option allows the disabling of the VM event counters
1420 on EXPERT systems. /proc/vmstat will only show page counts
1421 if VM event counters are disabled.
1425 bool "Enable PCI quirk workarounds" if EXPERT
1428 This enables workarounds for various PCI chipset
1429 bugs/quirks. Disable this only if your target machine is
1430 unaffected by PCI quirks.
1434 bool "Enable SLUB debugging support" if EXPERT
1435 depends on SLUB && SYSFS
1437 SLUB has extensive debug support features. Disabling these can
1438 result in significant savings in code size. This also disables
1439 SLUB sysfs support. /sys/slab will not exist and there will be
1440 no support for cache validation etc.
1443 bool "Disable heap randomization"
1446 Randomizing heap placement makes heap exploits harder, but it
1447 also breaks ancient binaries (including anything libc5 based).
1448 This option changes the bootup default to heap randomization
1449 disabled, and can be overridden at runtime by setting
1450 /proc/sys/kernel/randomize_va_space to 2.
1452 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1455 prompt "Choose SLAB allocator"
1458 This option allows to select a slab allocator.
1463 The regular slab allocator that is established and known to work
1464 well in all environments. It organizes cache hot objects in
1465 per cpu and per node queues.
1468 bool "SLUB (Unqueued Allocator)"
1470 SLUB is a slab allocator that minimizes cache line usage
1471 instead of managing queues of cached objects (SLAB approach).
1472 Per cpu caching is realized using slabs of objects instead
1473 of queues of objects. SLUB can use memory efficiently
1474 and has enhanced diagnostics. SLUB is the default choice for
1479 bool "SLOB (Simple Allocator)"
1481 SLOB replaces the stock allocator with a drastically simpler
1482 allocator. SLOB is generally more space efficient but
1483 does not perform as well on large systems.
1487 config MMAP_ALLOW_UNINITIALIZED
1488 bool "Allow mmapped anonymous memory to be uninitialized"
1489 depends on EXPERT && !MMU
1492 Normally, and according to the Linux spec, anonymous memory obtained
1493 from mmap() has it's contents cleared before it is passed to
1494 userspace. Enabling this config option allows you to request that
1495 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1496 providing a huge performance boost. If this option is not enabled,
1497 then the flag will be ignored.
1499 This is taken advantage of by uClibc's malloc(), and also by
1500 ELF-FDPIC binfmt's brk and stack allocator.
1502 Because of the obvious security issues, this option should only be
1503 enabled on embedded devices where you control what is run in
1504 userspace. Since that isn't generally a problem on no-MMU systems,
1505 it is normally safe to say Y here.
1507 See Documentation/nommu-mmap.txt for more information.
1510 bool "Profiling support"
1512 Say Y here to enable the extended profiling support mechanisms used
1513 by profilers such as OProfile.
1516 # Place an empty function call at each tracepoint site. Can be
1517 # dynamically changed for a probe function.
1522 source "arch/Kconfig"
1524 endmenu # General setup
1526 config HAVE_GENERIC_DMA_COHERENT
1533 depends on SLAB || SLUB_DEBUG
1541 default 0 if BASE_FULL
1542 default 1 if !BASE_FULL
1545 bool "Enable loadable module support"
1547 Kernel modules are small pieces of compiled code which can
1548 be inserted in the running kernel, rather than being
1549 permanently built into the kernel. You use the "modprobe"
1550 tool to add (and sometimes remove) them. If you say Y here,
1551 many parts of the kernel can be built as modules (by
1552 answering M instead of Y where indicated): this is most
1553 useful for infrequently used options which are not required
1554 for booting. For more information, see the man pages for
1555 modprobe, lsmod, modinfo, insmod and rmmod.
1557 If you say Y here, you will need to run "make
1558 modules_install" to put the modules under /lib/modules/
1559 where modprobe can find them (you may need to be root to do
1566 config MODULE_FORCE_LOAD
1567 bool "Forced module loading"
1570 Allow loading of modules without version information (ie. modprobe
1571 --force). Forced module loading sets the 'F' (forced) taint flag and
1572 is usually a really bad idea.
1574 config MODULE_UNLOAD
1575 bool "Module unloading"
1577 Without this option you will not be able to unload any
1578 modules (note that some modules may not be unloadable
1579 anyway), which makes your kernel smaller, faster
1580 and simpler. If unsure, say Y.
1582 config MODULE_FORCE_UNLOAD
1583 bool "Forced module unloading"
1584 depends on MODULE_UNLOAD && EXPERIMENTAL
1586 This option allows you to force a module to unload, even if the
1587 kernel believes it is unsafe: the kernel will remove the module
1588 without waiting for anyone to stop using it (using the -f option to
1589 rmmod). This is mainly for kernel developers and desperate users.
1593 bool "Module versioning support"
1595 Usually, you have to use modules compiled with your kernel.
1596 Saying Y here makes it sometimes possible to use modules
1597 compiled for different kernels, by adding enough information
1598 to the modules to (hopefully) spot any changes which would
1599 make them incompatible with the kernel you are running. If
1602 config MODULE_SRCVERSION_ALL
1603 bool "Source checksum for all modules"
1605 Modules which contain a MODULE_VERSION get an extra "srcversion"
1606 field inserted into their modinfo section, which contains a
1607 sum of the source files which made it. This helps maintainers
1608 see exactly which source was used to build a module (since
1609 others sometimes change the module source without updating
1610 the version). With this option, such a "srcversion" field
1611 will be created for all modules. If unsure, say N.
1614 bool "Module signature verification"
1618 select ASYMMETRIC_KEY_TYPE
1619 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1620 select PUBLIC_KEY_ALGO_RSA
1623 select X509_CERTIFICATE_PARSER
1625 Check modules for valid signatures upon load: the signature
1626 is simply appended to the module. For more information see
1627 Documentation/module-signing.txt.
1629 !!!WARNING!!! If you enable this option, you MUST make sure that the
1630 module DOES NOT get stripped after being signed. This includes the
1631 debuginfo strip done by some packagers (such as rpmbuild) and
1632 inclusion into an initramfs that wants the module size reduced.
1634 config MODULE_SIG_FORCE
1635 bool "Require modules to be validly signed"
1636 depends on MODULE_SIG
1638 Reject unsigned modules or signed modules for which we don't have a
1639 key. Without this, such modules will simply taint the kernel.
1642 prompt "Which hash algorithm should modules be signed with?"
1643 depends on MODULE_SIG
1645 This determines which sort of hashing algorithm will be used during
1646 signature generation. This algorithm _must_ be built into the kernel
1647 directly so that signature verification can take place. It is not
1648 possible to load a signed module containing the algorithm to check
1649 the signature on that module.
1651 config MODULE_SIG_SHA1
1652 bool "Sign modules with SHA-1"
1655 config MODULE_SIG_SHA224
1656 bool "Sign modules with SHA-224"
1657 select CRYPTO_SHA256
1659 config MODULE_SIG_SHA256
1660 bool "Sign modules with SHA-256"
1661 select CRYPTO_SHA256
1663 config MODULE_SIG_SHA384
1664 bool "Sign modules with SHA-384"
1665 select CRYPTO_SHA512
1667 config MODULE_SIG_SHA512
1668 bool "Sign modules with SHA-512"
1669 select CRYPTO_SHA512
1675 config INIT_ALL_POSSIBLE
1678 Back when each arch used to define their own cpu_online_mask and
1679 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
1680 with all 1s, and others with all 0s. When they were centralised,
1681 it was better to provide this option than to break all the archs
1682 and have several arch maintainers pursuing me down dark alleys.
1687 depends on (SMP && MODULE_UNLOAD) || HOTPLUG_CPU
1689 Need stop_machine() primitive.
1691 source "block/Kconfig"
1693 config PREEMPT_NOTIFIERS
1700 # Can be selected by architectures with broken toolchains
1701 # that get confused by correct const<->read_only section
1703 config BROKEN_RODATA
1709 Build a simple ASN.1 grammar compiler that produces a bytecode output
1710 that can be interpreted by the ASN.1 stream decoder and used to
1711 inform it as to what tags are to be expected in a stream and what
1712 functions to call on what tags.
1714 source "kernel/Kconfig.locks"