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
36 bool "Prompt for development and/or incomplete code/drivers"
38 Some of the various things that Linux supports (such as network
39 drivers, file systems, network protocols, etc.) can be in a state
40 of development where the functionality, stability, or the level of
41 testing is not yet high enough for general use. This is usually
42 known as the "alpha-test" phase among developers. If a feature is
43 currently in alpha-test, then the developers usually discourage
44 uninformed widespread use of this feature by the general public to
45 avoid "Why doesn't this work?" type mail messages. However, active
46 testing and use of these systems is welcomed. Just be aware that it
47 may not meet the normal level of reliability or it may fail to work
48 in some special cases. Detailed bug reports from people familiar
49 with the kernel internals are usually welcomed by the developers
50 (before submitting bug reports, please read the documents
51 <file:README>, <file:MAINTAINERS>, <file:REPORTING-BUGS>,
52 <file:Documentation/BUG-HUNTING>, and
53 <file:Documentation/oops-tracing.txt> in the kernel source).
55 This option will also make obsoleted drivers available. These are
56 drivers that have been replaced by something else, and/or are
57 scheduled to be removed in a future kernel release.
59 Unless you intend to help test and develop a feature or driver that
60 falls into this category, or you have a situation that requires
61 using these features, you should probably say N here, which will
62 cause the configurator to present you with fewer choices. If
63 you say Y here, you will be offered the choice of using features or
64 drivers that are currently considered to be in the alpha-test phase.
71 depends on BROKEN || !SMP
74 config INIT_ENV_ARG_LIMIT
79 Maximum of each of the number of arguments and environment
80 variables passed to init from the kernel command line.
84 string "Cross-compiler tool prefix"
86 Same as running 'make CROSS_COMPILE=prefix-' but stored for
87 default make runs in this kernel build directory. You don't
88 need to set this unless you want the configured kernel build
89 directory to select the cross-compiler automatically.
92 string "Local version - append to kernel release"
94 Append an extra string to the end of your kernel version.
95 This will show up when you type uname, for example.
96 The string you set here will be appended after the contents of
97 any files with a filename matching localversion* in your
98 object and source tree, in that order. Your total string can
99 be a maximum of 64 characters.
101 config LOCALVERSION_AUTO
102 bool "Automatically append version information to the version string"
105 This will try to automatically determine if the current tree is a
106 release tree by looking for git tags that belong to the current
107 top of tree revision.
109 A string of the format -gxxxxxxxx will be added to the localversion
110 if a git-based tree is found. The string generated by this will be
111 appended after any matching localversion* files, and after the value
112 set in CONFIG_LOCALVERSION.
114 (The actual string used here is the first eight characters produced
115 by running the command:
117 $ git rev-parse --verify HEAD
119 which is done within the script "scripts/setlocalversion".)
121 config HAVE_KERNEL_GZIP
124 config HAVE_KERNEL_BZIP2
127 config HAVE_KERNEL_LZMA
130 config HAVE_KERNEL_XZ
133 config HAVE_KERNEL_LZO
137 prompt "Kernel compression mode"
139 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO
141 The linux kernel is a kind of self-extracting executable.
142 Several compression algorithms are available, which differ
143 in efficiency, compression and decompression speed.
144 Compression speed is only relevant when building a kernel.
145 Decompression speed is relevant at each boot.
147 If you have any problems with bzip2 or lzma compressed
148 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
149 version of this functionality (bzip2 only), for 2.4, was
150 supplied by Christian Ludwig)
152 High compression options are mostly useful for users, who
153 are low on disk space (embedded systems), but for whom ram
156 If in doubt, select 'gzip'
160 depends on HAVE_KERNEL_GZIP
162 The old and tried gzip compression. It provides a good balance
163 between compression ratio and decompression speed.
167 depends on HAVE_KERNEL_BZIP2
169 Its compression ratio and speed is intermediate.
170 Decompression speed is slowest among the choices. The kernel
171 size is about 10% smaller with bzip2, in comparison to gzip.
172 Bzip2 uses a large amount of memory. For modern kernels you
173 will need at least 8MB RAM or more for booting.
177 depends on HAVE_KERNEL_LZMA
179 This compression algorithm's ratio is best. Decompression speed
180 is between gzip and bzip2. Compression is slowest.
181 The kernel size is about 33% smaller with LZMA in comparison to gzip.
185 depends on HAVE_KERNEL_XZ
187 XZ uses the LZMA2 algorithm and instruction set specific
188 BCJ filters which can improve compression ratio of executable
189 code. The size of the kernel is about 30% smaller with XZ in
190 comparison to gzip. On architectures for which there is a BCJ
191 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
192 will create a few percent smaller kernel than plain LZMA.
194 The speed is about the same as with LZMA: The decompression
195 speed of XZ is better than that of bzip2 but worse than gzip
196 and LZO. Compression is slow.
200 depends on HAVE_KERNEL_LZO
202 Its compression ratio is the poorest among the choices. The kernel
203 size is about 10% bigger than gzip; however its speed
204 (both compression and decompression) is the fastest.
208 config DEFAULT_HOSTNAME
209 string "Default hostname"
212 This option determines the default system hostname before userspace
213 calls sethostname(2). The kernel traditionally uses "(none)" here,
214 but you may wish to use a different default here to make a minimal
215 system more usable with less configuration.
218 bool "Support for paging of anonymous memory (swap)"
219 depends on MMU && BLOCK
222 This option allows you to choose whether you want to have support
223 for so called swap devices or swap files in your kernel that are
224 used to provide more virtual memory than the actual RAM present
225 in your computer. If unsure say Y.
230 Inter Process Communication is a suite of library functions and
231 system calls which let processes (running programs) synchronize and
232 exchange information. It is generally considered to be a good thing,
233 and some programs won't run unless you say Y here. In particular, if
234 you want to run the DOS emulator dosemu under Linux (read the
235 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
236 you'll need to say Y here.
238 You can find documentation about IPC with "info ipc" and also in
239 section 6.4 of the Linux Programmer's Guide, available from
240 <http://www.tldp.org/guides.html>.
242 config SYSVIPC_SYSCTL
249 bool "POSIX Message Queues"
250 depends on NET && EXPERIMENTAL
252 POSIX variant of message queues is a part of IPC. In POSIX message
253 queues every message has a priority which decides about succession
254 of receiving it by a process. If you want to compile and run
255 programs written e.g. for Solaris with use of its POSIX message
256 queues (functions mq_*) say Y here.
258 POSIX message queues are visible as a filesystem called 'mqueue'
259 and can be mounted somewhere if you want to do filesystem
260 operations on message queues.
264 config POSIX_MQUEUE_SYSCTL
266 depends on POSIX_MQUEUE
271 bool "open by fhandle syscalls"
274 If you say Y here, a user level program will be able to map
275 file names to handle and then later use the handle for
276 different file system operations. This is useful in implementing
277 userspace file servers, which now track files using handles instead
278 of names. The handle would remain the same even if file names
279 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
283 bool "Auditing support"
286 Enable auditing infrastructure that can be used with another
287 kernel subsystem, such as SELinux (which requires this for
288 logging of avc messages output). Does not do system-call
289 auditing without CONFIG_AUDITSYSCALL.
292 bool "Enable system-call auditing support"
293 depends on AUDIT && (X86 || PPC || S390 || IA64 || UML || SPARC64 || SUPERH || (ARM && AEABI && !OABI_COMPAT))
294 default y if SECURITY_SELINUX
296 Enable low-overhead system-call auditing infrastructure that
297 can be used independently or with another kernel subsystem,
302 depends on AUDITSYSCALL
307 depends on AUDITSYSCALL
310 config AUDIT_LOGINUID_IMMUTABLE
311 bool "Make audit loginuid immutable"
314 The config option toggles if a task setting its loginuid requires
315 CAP_SYS_AUDITCONTROL or if that task should require no special permissions
316 but should instead only allow setting its loginuid if it was never
317 previously set. On systems which use systemd or a similar central
318 process to restart login services this should be set to true. On older
319 systems in which an admin would typically have to directly stop and
320 start processes this should be set to false. Setting this to true allows
321 one to drop potentially dangerous capabilites from the login tasks,
322 but may not be backwards compatible with older init systems.
324 source "kernel/irq/Kconfig"
325 source "kernel/time/Kconfig"
327 menu "CPU/Task time and stats accounting"
330 prompt "Cputime accounting"
331 default TICK_CPU_ACCOUNTING if !PPC64
332 default VIRT_CPU_ACCOUNTING if PPC64
334 # Kind of a stub config for the pure tick based cputime accounting
335 config TICK_CPU_ACCOUNTING
336 bool "Simple tick based cputime accounting"
339 This is the basic tick based cputime accounting that maintains
340 statistics about user, system and idle time spent on per jiffies
345 config VIRT_CPU_ACCOUNTING
346 bool "Deterministic task and CPU time accounting"
347 depends on HAVE_VIRT_CPU_ACCOUNTING
349 Select this option to enable more accurate task and CPU time
350 accounting. This is done by reading a CPU counter on each
351 kernel entry and exit and on transitions within the kernel
352 between system, softirq and hardirq state, so there is a
353 small performance impact. In the case of s390 or IBM POWER > 5,
354 this also enables accounting of stolen time on logically-partitioned
357 config IRQ_TIME_ACCOUNTING
358 bool "Fine granularity task level IRQ time accounting"
359 depends on HAVE_IRQ_TIME_ACCOUNTING
361 Select this option to enable fine granularity task irq time
362 accounting. This is done by reading a timestamp on each
363 transitions between softirq and hardirq state, so there can be a
364 small performance impact.
366 If in doubt, say N here.
370 config BSD_PROCESS_ACCT
371 bool "BSD Process Accounting"
373 If you say Y here, a user level program will be able to instruct the
374 kernel (via a special system call) to write process accounting
375 information to a file: whenever a process exits, information about
376 that process will be appended to the file by the kernel. The
377 information includes things such as creation time, owning user,
378 command name, memory usage, controlling terminal etc. (the complete
379 list is in the struct acct in <file:include/linux/acct.h>). It is
380 up to the user level program to do useful things with this
381 information. This is generally a good idea, so say Y.
383 config BSD_PROCESS_ACCT_V3
384 bool "BSD Process Accounting version 3 file format"
385 depends on BSD_PROCESS_ACCT
388 If you say Y here, the process accounting information is written
389 in a new file format that also logs the process IDs of each
390 process and it's parent. Note that this file format is incompatible
391 with previous v0/v1/v2 file formats, so you will need updated tools
392 for processing it. A preliminary version of these tools is available
393 at <http://www.gnu.org/software/acct/>.
396 bool "Export task/process statistics through netlink (EXPERIMENTAL)"
400 Export selected statistics for tasks/processes through the
401 generic netlink interface. Unlike BSD process accounting, the
402 statistics are available during the lifetime of tasks/processes as
403 responses to commands. Like BSD accounting, they are sent to user
408 config TASK_DELAY_ACCT
409 bool "Enable per-task delay accounting (EXPERIMENTAL)"
412 Collect information on time spent by a task waiting for system
413 resources like cpu, synchronous block I/O completion and swapping
414 in pages. Such statistics can help in setting a task's priorities
415 relative to other tasks for cpu, io, rss limits etc.
420 bool "Enable extended accounting over taskstats (EXPERIMENTAL)"
423 Collect extended task accounting data and send the data
424 to userland for processing over the taskstats interface.
428 config TASK_IO_ACCOUNTING
429 bool "Enable per-task storage I/O accounting (EXPERIMENTAL)"
430 depends on TASK_XACCT
432 Collect information on the number of bytes of storage I/O which this
437 endmenu # "CPU/Task time and stats accounting"
442 prompt "RCU Implementation"
446 bool "Tree-based hierarchical RCU"
447 depends on !PREEMPT && SMP
449 This option selects the RCU implementation that is
450 designed for very large SMP system with hundreds or
451 thousands of CPUs. It also scales down nicely to
454 config TREE_PREEMPT_RCU
455 bool "Preemptible tree-based hierarchical RCU"
456 depends on PREEMPT && SMP
458 This option selects the RCU implementation that is
459 designed for very large SMP systems with hundreds or
460 thousands of CPUs, but for which real-time response
461 is also required. It also scales down nicely to
465 bool "UP-only small-memory-footprint RCU"
466 depends on !PREEMPT && !SMP
468 This option selects the RCU implementation that is
469 designed for UP systems from which real-time response
470 is not required. This option greatly reduces the
471 memory footprint of RCU.
473 config TINY_PREEMPT_RCU
474 bool "Preemptible UP-only small-memory-footprint RCU"
475 depends on PREEMPT && !SMP
477 This option selects the RCU implementation that is designed
478 for real-time UP systems. This option greatly reduces the
479 memory footprint of RCU.
484 def_bool ( TREE_PREEMPT_RCU || TINY_PREEMPT_RCU )
486 This option enables preemptible-RCU code that is common between
487 the TREE_PREEMPT_RCU and TINY_PREEMPT_RCU implementations.
489 config CONTEXT_TRACKING
493 bool "Consider userspace as in RCU extended quiescent state"
494 depends on HAVE_CONTEXT_TRACKING && SMP
495 select CONTEXT_TRACKING
497 This option sets hooks on kernel / userspace boundaries and
498 puts RCU in extended quiescent state when the CPU runs in
499 userspace. It means that when a CPU runs in userspace, it is
500 excluded from the global RCU state machine and thus doesn't
501 try to keep the timer tick on for RCU.
503 Unless you want to hack and help the development of the full
504 dynticks mode, you shouldn't enable this option. It also
505 adds unnecessary overhead.
509 config CONTEXT_TRACKING_FORCE
510 bool "Force context tracking"
511 depends on CONTEXT_TRACKING
513 Probe on user/kernel boundaries by default in order to
514 test the features that rely on it such as userspace RCU extended
516 This test is there for debugging until we have a real user like the
520 int "Tree-based hierarchical RCU fanout value"
523 depends on TREE_RCU || TREE_PREEMPT_RCU
527 This option controls the fanout of hierarchical implementations
528 of RCU, allowing RCU to work efficiently on machines with
529 large numbers of CPUs. This value must be at least the fourth
530 root of NR_CPUS, which allows NR_CPUS to be insanely large.
531 The default value of RCU_FANOUT should be used for production
532 systems, but if you are stress-testing the RCU implementation
533 itself, small RCU_FANOUT values allow you to test large-system
534 code paths on small(er) systems.
536 Select a specific number if testing RCU itself.
537 Take the default if unsure.
539 config RCU_FANOUT_LEAF
540 int "Tree-based hierarchical RCU leaf-level fanout value"
541 range 2 RCU_FANOUT if 64BIT
542 range 2 RCU_FANOUT if !64BIT
543 depends on TREE_RCU || TREE_PREEMPT_RCU
546 This option controls the leaf-level fanout of hierarchical
547 implementations of RCU, and allows trading off cache misses
548 against lock contention. Systems that synchronize their
549 scheduling-clock interrupts for energy-efficiency reasons will
550 want the default because the smaller leaf-level fanout keeps
551 lock contention levels acceptably low. Very large systems
552 (hundreds or thousands of CPUs) will instead want to set this
553 value to the maximum value possible in order to reduce the
554 number of cache misses incurred during RCU's grace-period
555 initialization. These systems tend to run CPU-bound, and thus
556 are not helped by synchronized interrupts, and thus tend to
557 skew them, which reduces lock contention enough that large
558 leaf-level fanouts work well.
560 Select a specific number if testing RCU itself.
562 Select the maximum permissible value for large systems.
564 Take the default if unsure.
566 config RCU_FANOUT_EXACT
567 bool "Disable tree-based hierarchical RCU auto-balancing"
568 depends on TREE_RCU || TREE_PREEMPT_RCU
571 This option forces use of the exact RCU_FANOUT value specified,
572 regardless of imbalances in the hierarchy. This is useful for
573 testing RCU itself, and might one day be useful on systems with
574 strong NUMA behavior.
576 Without RCU_FANOUT_EXACT, the code will balance the hierarchy.
580 config RCU_FAST_NO_HZ
581 bool "Accelerate last non-dyntick-idle CPU's grace periods"
582 depends on NO_HZ && SMP
585 This option causes RCU to attempt to accelerate grace periods in
586 order to allow CPUs to enter dynticks-idle state more quickly.
587 On the other hand, this option increases the overhead of the
588 dynticks-idle checking, thus degrading scheduling latency.
590 Say Y if energy efficiency is critically important, and you don't
591 care about real-time response.
593 Say N if you are unsure.
595 config TREE_RCU_TRACE
596 def_bool RCU_TRACE && ( TREE_RCU || TREE_PREEMPT_RCU )
599 This option provides tracing for the TREE_RCU and
600 TREE_PREEMPT_RCU implementations, permitting Makefile to
601 trivially select kernel/rcutree_trace.c.
604 bool "Enable RCU priority boosting"
605 depends on RT_MUTEXES && PREEMPT_RCU
608 This option boosts the priority of preempted RCU readers that
609 block the current preemptible RCU grace period for too long.
610 This option also prevents heavy loads from blocking RCU
611 callback invocation for all flavors of RCU.
613 Say Y here if you are working with real-time apps or heavy loads
614 Say N here if you are unsure.
616 config RCU_BOOST_PRIO
617 int "Real-time priority to boost RCU readers to"
622 This option specifies the real-time priority to which long-term
623 preempted RCU readers are to be boosted. If you are working
624 with a real-time application that has one or more CPU-bound
625 threads running at a real-time priority level, you should set
626 RCU_BOOST_PRIO to a priority higher then the highest-priority
627 real-time CPU-bound thread. The default RCU_BOOST_PRIO value
628 of 1 is appropriate in the common case, which is real-time
629 applications that do not have any CPU-bound threads.
631 Some real-time applications might not have a single real-time
632 thread that saturates a given CPU, but instead might have
633 multiple real-time threads that, taken together, fully utilize
634 that CPU. In this case, you should set RCU_BOOST_PRIO to
635 a priority higher than the lowest-priority thread that is
636 conspiring to prevent the CPU from running any non-real-time
637 tasks. For example, if one thread at priority 10 and another
638 thread at priority 5 are between themselves fully consuming
639 the CPU time on a given CPU, then RCU_BOOST_PRIO should be
640 set to priority 6 or higher.
642 Specify the real-time priority, or take the default if unsure.
644 config RCU_BOOST_DELAY
645 int "Milliseconds to delay boosting after RCU grace-period start"
650 This option specifies the time to wait after the beginning of
651 a given grace period before priority-boosting preempted RCU
652 readers blocking that grace period. Note that any RCU reader
653 blocking an expedited RCU grace period is boosted immediately.
655 Accept the default if unsure.
658 bool "Offload RCU callback processing from boot-selected CPUs"
659 depends on TREE_RCU || TREE_PREEMPT_RCU
662 Use this option to reduce OS jitter for aggressive HPC or
663 real-time workloads. It can also be used to offload RCU
664 callback invocation to energy-efficient CPUs in battery-powered
665 asymmetric multiprocessors.
667 This option offloads callback invocation from the set of
668 CPUs specified at boot time by the rcu_nocbs parameter.
669 For each such CPU, a kthread ("rcuoN") will be created to
670 invoke callbacks, where the "N" is the CPU being offloaded.
671 Nothing prevents this kthread from running on the specified
672 CPUs, but (1) the kthreads may be preempted between each
673 callback, and (2) affinity or cgroups can be used to force
674 the kthreads to run on whatever set of CPUs is desired.
676 Say Y here if you want reduced OS jitter on selected CPUs.
677 Say N here if you are unsure.
679 endmenu # "RCU Subsystem"
682 tristate "Kernel .config support"
684 This option enables the complete Linux kernel ".config" file
685 contents to be saved in the kernel. It provides documentation
686 of which kernel options are used in a running kernel or in an
687 on-disk kernel. This information can be extracted from the kernel
688 image file with the script scripts/extract-ikconfig and used as
689 input to rebuild the current kernel or to build another kernel.
690 It can also be extracted from a running kernel by reading
691 /proc/config.gz if enabled (below).
694 bool "Enable access to .config through /proc/config.gz"
695 depends on IKCONFIG && PROC_FS
697 This option enables access to the kernel configuration file
698 through /proc/config.gz.
701 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
705 Select kernel log buffer size as a power of 2.
715 # Architectures with an unreliable sched_clock() should select this:
717 config HAVE_UNSTABLE_SCHED_CLOCK
721 boolean "Control Group support"
724 This option adds support for grouping sets of processes together, for
725 use with process control subsystems such as Cpusets, CFS, memory
726 controls or device isolation.
728 - Documentation/scheduler/sched-design-CFS.txt (CFS)
729 - Documentation/cgroups/ (features for grouping, isolation
730 and resource control)
737 bool "Example debug cgroup subsystem"
740 This option enables a simple cgroup subsystem that
741 exports useful debugging information about the cgroups
746 config CGROUP_FREEZER
747 bool "Freezer cgroup subsystem"
749 Provides a way to freeze and unfreeze all tasks in a
753 bool "Device controller for cgroups"
755 Provides a cgroup implementing whitelists for devices which
756 a process in the cgroup can mknod or open.
759 bool "Cpuset support"
761 This option will let you create and manage CPUSETs which
762 allow dynamically partitioning a system into sets of CPUs and
763 Memory Nodes and assigning tasks to run only within those sets.
764 This is primarily useful on large SMP or NUMA systems.
768 config PROC_PID_CPUSET
769 bool "Include legacy /proc/<pid>/cpuset file"
773 config CGROUP_CPUACCT
774 bool "Simple CPU accounting cgroup subsystem"
776 Provides a simple Resource Controller for monitoring the
777 total CPU consumed by the tasks in a cgroup.
779 config RESOURCE_COUNTERS
780 bool "Resource counters"
782 This option enables controller independent resource accounting
783 infrastructure that works with cgroups.
786 bool "Memory Resource Controller for Control Groups"
787 depends on RESOURCE_COUNTERS
790 Provides a memory resource controller that manages both anonymous
791 memory and page cache. (See Documentation/cgroups/memory.txt)
793 Note that setting this option increases fixed memory overhead
794 associated with each page of memory in the system. By this,
795 20(40)bytes/PAGE_SIZE on 32(64)bit system will be occupied by memory
796 usage tracking struct at boot. Total amount of this is printed out
799 Only enable when you're ok with these trade offs and really
800 sure you need the memory resource controller. Even when you enable
801 this, you can set "cgroup_disable=memory" at your boot option to
802 disable memory resource controller and you can avoid overheads.
803 (and lose benefits of memory resource controller)
805 This config option also selects MM_OWNER config option, which
806 could in turn add some fork/exit overhead.
809 bool "Memory Resource Controller Swap Extension"
810 depends on MEMCG && SWAP
812 Add swap management feature to memory resource controller. When you
813 enable this, you can limit mem+swap usage per cgroup. In other words,
814 when you disable this, memory resource controller has no cares to
815 usage of swap...a process can exhaust all of the swap. This extension
816 is useful when you want to avoid exhaustion swap but this itself
817 adds more overheads and consumes memory for remembering information.
818 Especially if you use 32bit system or small memory system, please
819 be careful about enabling this. When memory resource controller
820 is disabled by boot option, this will be automatically disabled and
821 there will be no overhead from this. Even when you set this config=y,
822 if boot option "swapaccount=0" is set, swap will not be accounted.
823 Now, memory usage of swap_cgroup is 2 bytes per entry. If swap page
824 size is 4096bytes, 512k per 1Gbytes of swap.
825 config MEMCG_SWAP_ENABLED
826 bool "Memory Resource Controller Swap Extension enabled by default"
827 depends on MEMCG_SWAP
830 Memory Resource Controller Swap Extension comes with its price in
831 a bigger memory consumption. General purpose distribution kernels
832 which want to enable the feature but keep it disabled by default
833 and let the user enable it by swapaccount boot command line
834 parameter should have this option unselected.
835 For those who want to have the feature enabled by default should
836 select this option (if, for some reason, they need to disable it
837 then swapaccount=0 does the trick).
839 bool "Memory Resource Controller Kernel Memory accounting (EXPERIMENTAL)"
840 depends on MEMCG && EXPERIMENTAL
843 The Kernel Memory extension for Memory Resource Controller can limit
844 the amount of memory used by kernel objects in the system. Those are
845 fundamentally different from the entities handled by the standard
846 Memory Controller, which are page-based, and can be swapped. Users of
847 the kmem extension can use it to guarantee that no group of processes
848 will ever exhaust kernel resources alone.
850 config CGROUP_HUGETLB
851 bool "HugeTLB Resource Controller for Control Groups"
852 depends on RESOURCE_COUNTERS && HUGETLB_PAGE && EXPERIMENTAL
855 Provides a cgroup Resource Controller for HugeTLB pages.
856 When you enable this, you can put a per cgroup limit on HugeTLB usage.
857 The limit is enforced during page fault. Since HugeTLB doesn't
858 support page reclaim, enforcing the limit at page fault time implies
859 that, the application will get SIGBUS signal if it tries to access
860 HugeTLB pages beyond its limit. This requires the application to know
861 beforehand how much HugeTLB pages it would require for its use. The
862 control group is tracked in the third page lru pointer. This means
863 that we cannot use the controller with huge page less than 3 pages.
866 bool "Enable perf_event per-cpu per-container group (cgroup) monitoring"
867 depends on PERF_EVENTS && CGROUPS
869 This option extends the per-cpu mode to restrict monitoring to
870 threads which belong to the cgroup specified and run on the
875 menuconfig CGROUP_SCHED
876 bool "Group CPU scheduler"
879 This feature lets CPU scheduler recognize task groups and control CPU
880 bandwidth allocation to such task groups. It uses cgroups to group
884 config FAIR_GROUP_SCHED
885 bool "Group scheduling for SCHED_OTHER"
886 depends on CGROUP_SCHED
890 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
891 depends on EXPERIMENTAL
892 depends on FAIR_GROUP_SCHED
895 This option allows users to define CPU bandwidth rates (limits) for
896 tasks running within the fair group scheduler. Groups with no limit
897 set are considered to be unconstrained and will run with no
899 See tip/Documentation/scheduler/sched-bwc.txt for more information.
901 config RT_GROUP_SCHED
902 bool "Group scheduling for SCHED_RR/FIFO"
903 depends on EXPERIMENTAL
904 depends on CGROUP_SCHED
907 This feature lets you explicitly allocate real CPU bandwidth
908 to task groups. If enabled, it will also make it impossible to
909 schedule realtime tasks for non-root users until you allocate
910 realtime bandwidth for them.
911 See Documentation/scheduler/sched-rt-group.txt for more information.
916 bool "Block IO controller"
920 Generic block IO controller cgroup interface. This is the common
921 cgroup interface which should be used by various IO controlling
924 Currently, CFQ IO scheduler uses it to recognize task groups and
925 control disk bandwidth allocation (proportional time slice allocation)
926 to such task groups. It is also used by bio throttling logic in
927 block layer to implement upper limit in IO rates on a device.
929 This option only enables generic Block IO controller infrastructure.
930 One needs to also enable actual IO controlling logic/policy. For
931 enabling proportional weight division of disk bandwidth in CFQ, set
932 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
933 CONFIG_BLK_DEV_THROTTLING=y.
935 See Documentation/cgroups/blkio-controller.txt for more information.
937 config DEBUG_BLK_CGROUP
938 bool "Enable Block IO controller debugging"
939 depends on BLK_CGROUP
942 Enable some debugging help. Currently it exports additional stat
943 files in a cgroup which can be useful for debugging.
947 config CHECKPOINT_RESTORE
948 bool "Checkpoint/restore support" if EXPERT
951 Enables additional kernel features in a sake of checkpoint/restore.
952 In particular it adds auxiliary prctl codes to setup process text,
953 data and heap segment sizes, and a few additional /proc filesystem
956 If unsure, say N here.
958 menuconfig NAMESPACES
959 bool "Namespaces support" if EXPERT
962 Provides the way to make tasks work with different objects using
963 the same id. For example same IPC id may refer to different objects
964 or same user id or pid may refer to different tasks when used in
965 different namespaces.
973 In this namespace tasks see different info provided with the
978 depends on (SYSVIPC || POSIX_MQUEUE)
981 In this namespace tasks work with IPC ids which correspond to
982 different IPC objects in different namespaces.
985 bool "User namespace (EXPERIMENTAL)"
986 depends on EXPERIMENTAL
987 depends on UIDGID_CONVERTED
988 select UIDGID_STRICT_TYPE_CHECKS
992 This allows containers, i.e. vservers, to use user namespaces
993 to provide different user info for different servers.
997 bool "PID Namespaces"
1000 Support process id namespaces. This allows having multiple
1001 processes with the same pid as long as they are in different
1002 pid namespaces. This is a building block of containers.
1005 bool "Network namespace"
1009 Allow user space to create what appear to be multiple instances
1010 of the network stack.
1014 config UIDGID_CONVERTED
1015 # True if all of the selected software conmponents are known
1016 # to have uid_t and gid_t converted to kuid_t and kgid_t
1017 # where appropriate and are otherwise safe to use with
1018 # the user namespace.
1023 depends on NET_9P = n
1026 depends on 9P_FS = n
1027 depends on AFS_FS = n
1028 depends on AUTOFS4_FS = n
1029 depends on CEPH_FS = n
1031 depends on CODA_FS = n
1032 depends on FUSE_FS = n
1033 depends on GFS2_FS = n
1034 depends on NCP_FS = n
1036 depends on NFS_FS = n
1037 depends on OCFS2_FS = n
1038 depends on XFS_FS = n
1040 config UIDGID_STRICT_TYPE_CHECKS
1041 bool "Require conversions between uid/gids and their internal representation"
1042 depends on UIDGID_CONVERTED
1045 While the nececessary conversions are being added to all subsystems this option allows
1046 the code to continue to build for unconverted subsystems.
1048 Say Y here if you want the strict type checking enabled
1050 config SCHED_AUTOGROUP
1051 bool "Automatic process group scheduling"
1055 select FAIR_GROUP_SCHED
1057 This option optimizes the scheduler for common desktop workloads by
1058 automatically creating and populating task groups. This separation
1059 of workloads isolates aggressive CPU burners (like build jobs) from
1060 desktop applications. Task group autogeneration is currently based
1066 config SYSFS_DEPRECATED
1067 bool "Enable deprecated sysfs features to support old userspace tools"
1071 This option adds code that switches the layout of the "block" class
1072 devices, to not show up in /sys/class/block/, but only in
1075 This switch is only active when the sysfs.deprecated=1 boot option is
1076 passed or the SYSFS_DEPRECATED_V2 option is set.
1078 This option allows new kernels to run on old distributions and tools,
1079 which might get confused by /sys/class/block/. Since 2007/2008 all
1080 major distributions and tools handle this just fine.
1082 Recent distributions and userspace tools after 2009/2010 depend on
1083 the existence of /sys/class/block/, and will not work with this
1086 Only if you are using a new kernel on an old distribution, you might
1089 config SYSFS_DEPRECATED_V2
1090 bool "Enable deprecated sysfs features by default"
1093 depends on SYSFS_DEPRECATED
1095 Enable deprecated sysfs by default.
1097 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1100 Only if you are using a new kernel on an old distribution, you might
1101 need to say Y here. Even then, odds are you would not need it
1102 enabled, you can always pass the boot option if absolutely necessary.
1105 bool "Kernel->user space relay support (formerly relayfs)"
1107 This option enables support for relay interface support in
1108 certain file systems (such as debugfs).
1109 It is designed to provide an efficient mechanism for tools and
1110 facilities to relay large amounts of data from kernel space to
1115 config BLK_DEV_INITRD
1116 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1117 depends on BROKEN || !FRV
1119 The initial RAM filesystem is a ramfs which is loaded by the
1120 boot loader (loadlin or lilo) and that is mounted as root
1121 before the normal boot procedure. It is typically used to
1122 load modules needed to mount the "real" root file system,
1123 etc. See <file:Documentation/initrd.txt> for details.
1125 If RAM disk support (BLK_DEV_RAM) is also included, this
1126 also enables initial RAM disk (initrd) support and adds
1127 15 Kbytes (more on some other architectures) to the kernel size.
1133 source "usr/Kconfig"
1137 config CC_OPTIMIZE_FOR_SIZE
1138 bool "Optimize for size"
1140 Enabling this option will pass "-Os" instead of "-O2" to gcc
1141 resulting in a smaller kernel.
1152 bool "Configure standard kernel features (expert users)"
1153 # Unhide debug options, to make the on-by-default options visible
1156 This option allows certain base kernel options and settings
1157 to be disabled or tweaked. This is for specialized
1158 environments which can tolerate a "non-standard" kernel.
1159 Only use this if you really know what you are doing.
1165 bool "Enable 16-bit UID system calls" if EXPERT
1166 depends on HAVE_UID16
1169 This enables the legacy 16-bit UID syscall wrappers.
1171 config SYSCTL_SYSCALL
1172 bool "Sysctl syscall support" if EXPERT
1173 depends on PROC_SYSCTL
1177 sys_sysctl uses binary paths that have been found challenging
1178 to properly maintain and use. The interface in /proc/sys
1179 using paths with ascii names is now the primary path to this
1182 Almost nothing using the binary sysctl interface so if you are
1183 trying to save some space it is probably safe to disable this,
1184 making your kernel marginally smaller.
1186 If unsure say N here.
1188 config SYSCTL_EXCEPTION_TRACE
1191 Enable support for /proc/sys/debug/exception-trace.
1194 bool "Load all symbols for debugging/ksymoops" if EXPERT
1197 Say Y here to let the kernel print out symbolic crash information and
1198 symbolic stack backtraces. This increases the size of the kernel
1199 somewhat, as all symbols have to be loaded into the kernel image.
1202 bool "Include all symbols in kallsyms"
1203 depends on DEBUG_KERNEL && KALLSYMS
1205 Normally kallsyms only contains the symbols of functions for nicer
1206 OOPS messages and backtraces (i.e., symbols from the text and inittext
1207 sections). This is sufficient for most cases. And only in very rare
1208 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1209 names of variables from the data sections, etc).
1211 This option makes sure that all symbols are loaded into the kernel
1212 image (i.e., symbols from all sections) in cost of increased kernel
1213 size (depending on the kernel configuration, it may be 300KiB or
1214 something like this).
1216 Say N unless you really need all symbols.
1223 bool "Enable support for printk" if EXPERT
1225 This option enables normal printk support. Removing it
1226 eliminates most of the message strings from the kernel image
1227 and makes the kernel more or less silent. As this makes it
1228 very difficult to diagnose system problems, saying N here is
1229 strongly discouraged.
1232 bool "BUG() support" if EXPERT
1235 Disabling this option eliminates support for BUG and WARN, reducing
1236 the size of your kernel image and potentially quietly ignoring
1237 numerous fatal conditions. You should only consider disabling this
1238 option for embedded systems with no facilities for reporting errors.
1244 bool "Enable ELF core dumps" if EXPERT
1246 Enable support for generating core dumps. Disabling saves about 4k.
1249 config PCSPKR_PLATFORM
1250 bool "Enable PC-Speaker support" if EXPERT
1251 depends on HAVE_PCSPKR_PLATFORM
1255 This option allows to disable the internal PC-Speaker
1256 support, saving some memory.
1258 config HAVE_PCSPKR_PLATFORM
1263 bool "Enable full-sized data structures for core" if EXPERT
1265 Disabling this option reduces the size of miscellaneous core
1266 kernel data structures. This saves memory on small machines,
1267 but may reduce performance.
1270 bool "Enable futex support" if EXPERT
1274 Disabling this option will cause the kernel to be built without
1275 support for "fast userspace mutexes". The resulting kernel may not
1276 run glibc-based applications correctly.
1279 bool "Enable eventpoll support" if EXPERT
1283 Disabling this option will cause the kernel to be built without
1284 support for epoll family of system calls.
1287 bool "Enable signalfd() system call" if EXPERT
1291 Enable the signalfd() system call that allows to receive signals
1292 on a file descriptor.
1297 bool "Enable timerfd() system call" if EXPERT
1301 Enable the timerfd() system call that allows to receive timer
1302 events on a file descriptor.
1307 bool "Enable eventfd() system call" if EXPERT
1311 Enable the eventfd() system call that allows to receive both
1312 kernel notification (ie. KAIO) or userspace notifications.
1317 bool "Use full shmem filesystem" if EXPERT
1321 The shmem is an internal filesystem used to manage shared memory.
1322 It is backed by swap and manages resource limits. It is also exported
1323 to userspace as tmpfs if TMPFS is enabled. Disabling this
1324 option replaces shmem and tmpfs with the much simpler ramfs code,
1325 which may be appropriate on small systems without swap.
1328 bool "Enable AIO support" if EXPERT
1331 This option enables POSIX asynchronous I/O which may by used
1332 by some high performance threaded applications. Disabling
1333 this option saves about 7k.
1336 bool "Embedded system"
1339 This option should be enabled if compiling the kernel for
1340 an embedded system so certain expert options are available
1343 config HAVE_PERF_EVENTS
1346 See tools/perf/design.txt for details.
1348 config PERF_USE_VMALLOC
1351 See tools/perf/design.txt for details
1353 menu "Kernel Performance Events And Counters"
1356 bool "Kernel performance events and counters"
1357 default y if PROFILING
1358 depends on HAVE_PERF_EVENTS
1362 Enable kernel support for various performance events provided
1363 by software and hardware.
1365 Software events are supported either built-in or via the
1366 use of generic tracepoints.
1368 Most modern CPUs support performance events via performance
1369 counter registers. These registers count the number of certain
1370 types of hw events: such as instructions executed, cachemisses
1371 suffered, or branches mis-predicted - without slowing down the
1372 kernel or applications. These registers can also trigger interrupts
1373 when a threshold number of events have passed - and can thus be
1374 used to profile the code that runs on that CPU.
1376 The Linux Performance Event subsystem provides an abstraction of
1377 these software and hardware event capabilities, available via a
1378 system call and used by the "perf" utility in tools/perf/. It
1379 provides per task and per CPU counters, and it provides event
1380 capabilities on top of those.
1384 config DEBUG_PERF_USE_VMALLOC
1386 bool "Debug: use vmalloc to back perf mmap() buffers"
1387 depends on PERF_EVENTS && DEBUG_KERNEL
1388 select PERF_USE_VMALLOC
1390 Use vmalloc memory to back perf mmap() buffers.
1392 Mostly useful for debugging the vmalloc code on platforms
1393 that don't require it.
1399 config VM_EVENT_COUNTERS
1401 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1403 VM event counters are needed for event counts to be shown.
1404 This option allows the disabling of the VM event counters
1405 on EXPERT systems. /proc/vmstat will only show page counts
1406 if VM event counters are disabled.
1410 bool "Enable PCI quirk workarounds" if EXPERT
1413 This enables workarounds for various PCI chipset
1414 bugs/quirks. Disable this only if your target machine is
1415 unaffected by PCI quirks.
1419 bool "Enable SLUB debugging support" if EXPERT
1420 depends on SLUB && SYSFS
1422 SLUB has extensive debug support features. Disabling these can
1423 result in significant savings in code size. This also disables
1424 SLUB sysfs support. /sys/slab will not exist and there will be
1425 no support for cache validation etc.
1428 bool "Disable heap randomization"
1431 Randomizing heap placement makes heap exploits harder, but it
1432 also breaks ancient binaries (including anything libc5 based).
1433 This option changes the bootup default to heap randomization
1434 disabled, and can be overridden at runtime by setting
1435 /proc/sys/kernel/randomize_va_space to 2.
1437 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1440 prompt "Choose SLAB allocator"
1443 This option allows to select a slab allocator.
1448 The regular slab allocator that is established and known to work
1449 well in all environments. It organizes cache hot objects in
1450 per cpu and per node queues.
1453 bool "SLUB (Unqueued Allocator)"
1455 SLUB is a slab allocator that minimizes cache line usage
1456 instead of managing queues of cached objects (SLAB approach).
1457 Per cpu caching is realized using slabs of objects instead
1458 of queues of objects. SLUB can use memory efficiently
1459 and has enhanced diagnostics. SLUB is the default choice for
1464 bool "SLOB (Simple Allocator)"
1466 SLOB replaces the stock allocator with a drastically simpler
1467 allocator. SLOB is generally more space efficient but
1468 does not perform as well on large systems.
1472 config MMAP_ALLOW_UNINITIALIZED
1473 bool "Allow mmapped anonymous memory to be uninitialized"
1474 depends on EXPERT && !MMU
1477 Normally, and according to the Linux spec, anonymous memory obtained
1478 from mmap() has it's contents cleared before it is passed to
1479 userspace. Enabling this config option allows you to request that
1480 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1481 providing a huge performance boost. If this option is not enabled,
1482 then the flag will be ignored.
1484 This is taken advantage of by uClibc's malloc(), and also by
1485 ELF-FDPIC binfmt's brk and stack allocator.
1487 Because of the obvious security issues, this option should only be
1488 enabled on embedded devices where you control what is run in
1489 userspace. Since that isn't generally a problem on no-MMU systems,
1490 it is normally safe to say Y here.
1492 See Documentation/nommu-mmap.txt for more information.
1495 bool "Profiling support"
1497 Say Y here to enable the extended profiling support mechanisms used
1498 by profilers such as OProfile.
1501 # Place an empty function call at each tracepoint site. Can be
1502 # dynamically changed for a probe function.
1507 source "arch/Kconfig"
1509 endmenu # General setup
1511 config HAVE_GENERIC_DMA_COHERENT
1518 depends on SLAB || SLUB_DEBUG
1526 default 0 if BASE_FULL
1527 default 1 if !BASE_FULL
1530 bool "Enable loadable module support"
1532 Kernel modules are small pieces of compiled code which can
1533 be inserted in the running kernel, rather than being
1534 permanently built into the kernel. You use the "modprobe"
1535 tool to add (and sometimes remove) them. If you say Y here,
1536 many parts of the kernel can be built as modules (by
1537 answering M instead of Y where indicated): this is most
1538 useful for infrequently used options which are not required
1539 for booting. For more information, see the man pages for
1540 modprobe, lsmod, modinfo, insmod and rmmod.
1542 If you say Y here, you will need to run "make
1543 modules_install" to put the modules under /lib/modules/
1544 where modprobe can find them (you may need to be root to do
1551 config MODULE_FORCE_LOAD
1552 bool "Forced module loading"
1555 Allow loading of modules without version information (ie. modprobe
1556 --force). Forced module loading sets the 'F' (forced) taint flag and
1557 is usually a really bad idea.
1559 config MODULE_UNLOAD
1560 bool "Module unloading"
1562 Without this option you will not be able to unload any
1563 modules (note that some modules may not be unloadable
1564 anyway), which makes your kernel smaller, faster
1565 and simpler. If unsure, say Y.
1567 config MODULE_FORCE_UNLOAD
1568 bool "Forced module unloading"
1569 depends on MODULE_UNLOAD && EXPERIMENTAL
1571 This option allows you to force a module to unload, even if the
1572 kernel believes it is unsafe: the kernel will remove the module
1573 without waiting for anyone to stop using it (using the -f option to
1574 rmmod). This is mainly for kernel developers and desperate users.
1578 bool "Module versioning support"
1580 Usually, you have to use modules compiled with your kernel.
1581 Saying Y here makes it sometimes possible to use modules
1582 compiled for different kernels, by adding enough information
1583 to the modules to (hopefully) spot any changes which would
1584 make them incompatible with the kernel you are running. If
1587 config MODULE_SRCVERSION_ALL
1588 bool "Source checksum for all modules"
1590 Modules which contain a MODULE_VERSION get an extra "srcversion"
1591 field inserted into their modinfo section, which contains a
1592 sum of the source files which made it. This helps maintainers
1593 see exactly which source was used to build a module (since
1594 others sometimes change the module source without updating
1595 the version). With this option, such a "srcversion" field
1596 will be created for all modules. If unsure, say N.
1599 bool "Module signature verification"
1603 select ASYMMETRIC_KEY_TYPE
1604 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1605 select PUBLIC_KEY_ALGO_RSA
1608 select X509_CERTIFICATE_PARSER
1610 Check modules for valid signatures upon load: the signature
1611 is simply appended to the module. For more information see
1612 Documentation/module-signing.txt.
1614 !!!WARNING!!! If you enable this option, you MUST make sure that the
1615 module DOES NOT get stripped after being signed. This includes the
1616 debuginfo strip done by some packagers (such as rpmbuild) and
1617 inclusion into an initramfs that wants the module size reduced.
1619 config MODULE_SIG_FORCE
1620 bool "Require modules to be validly signed"
1621 depends on MODULE_SIG
1623 Reject unsigned modules or signed modules for which we don't have a
1624 key. Without this, such modules will simply taint the kernel.
1627 prompt "Which hash algorithm should modules be signed with?"
1628 depends on MODULE_SIG
1630 This determines which sort of hashing algorithm will be used during
1631 signature generation. This algorithm _must_ be built into the kernel
1632 directly so that signature verification can take place. It is not
1633 possible to load a signed module containing the algorithm to check
1634 the signature on that module.
1636 config MODULE_SIG_SHA1
1637 bool "Sign modules with SHA-1"
1640 config MODULE_SIG_SHA224
1641 bool "Sign modules with SHA-224"
1642 select CRYPTO_SHA256
1644 config MODULE_SIG_SHA256
1645 bool "Sign modules with SHA-256"
1646 select CRYPTO_SHA256
1648 config MODULE_SIG_SHA384
1649 bool "Sign modules with SHA-384"
1650 select CRYPTO_SHA512
1652 config MODULE_SIG_SHA512
1653 bool "Sign modules with SHA-512"
1654 select CRYPTO_SHA512
1660 config INIT_ALL_POSSIBLE
1663 Back when each arch used to define their own cpu_online_mask and
1664 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
1665 with all 1s, and others with all 0s. When they were centralised,
1666 it was better to provide this option than to break all the archs
1667 and have several arch maintainers pursuing me down dark alleys.
1672 depends on (SMP && MODULE_UNLOAD) || HOTPLUG_CPU
1674 Need stop_machine() primitive.
1676 source "block/Kconfig"
1678 config PREEMPT_NOTIFIERS
1685 # Can be selected by architectures with broken toolchains
1686 # that get confused by correct const<->read_only section
1688 config BROKEN_RODATA
1694 Build a simple ASN.1 grammar compiler that produces a bytecode output
1695 that can be interpreted by the ASN.1 stream decoder and used to
1696 inform it as to what tags are to be expected in a stream and what
1697 functions to call on what tags.
1699 source "kernel/Kconfig.locks"