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"
26 config BUILDTIME_EXTABLE_SORT
29 config THREAD_INFO_IN_TASK
32 Select this to move thread_info off the stack into task_struct. To
33 make this work, an arch will need to remove all thread_info fields
34 except flags and fix any runtime bugs.
36 One subtle change that will be needed is to use try_get_task_stack()
37 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
46 depends on BROKEN || !SMP
49 config INIT_ENV_ARG_LIMIT
54 Maximum of each of the number of arguments and environment
55 variables passed to init from the kernel command line.
59 string "Cross-compiler tool prefix"
61 Same as running 'make CROSS_COMPILE=prefix-' but stored for
62 default make runs in this kernel build directory. You don't
63 need to set this unless you want the configured kernel build
64 directory to select the cross-compiler automatically.
67 bool "Compile also drivers which will not load"
70 Some drivers can be compiled on a different platform than they are
71 intended to be run on. Despite they cannot be loaded there (or even
72 when they load they cannot be used due to missing HW support),
73 developers still, opposing to distributors, might want to build such
74 drivers to compile-test them.
76 If you are a developer and want to build everything available, say Y
77 here. If you are a user/distributor, say N here to exclude useless
78 drivers to be distributed.
81 string "Local version - append to kernel release"
83 Append an extra string to the end of your kernel version.
84 This will show up when you type uname, for example.
85 The string you set here will be appended after the contents of
86 any files with a filename matching localversion* in your
87 object and source tree, in that order. Your total string can
88 be a maximum of 64 characters.
90 config LOCALVERSION_AUTO
91 bool "Automatically append version information to the version string"
94 This will try to automatically determine if the current tree is a
95 release tree by looking for git tags that belong to the current
98 A string of the format -gxxxxxxxx will be added to the localversion
99 if a git-based tree is found. The string generated by this will be
100 appended after any matching localversion* files, and after the value
101 set in CONFIG_LOCALVERSION.
103 (The actual string used here is the first eight characters produced
104 by running the command:
106 $ git rev-parse --verify HEAD
108 which is done within the script "scripts/setlocalversion".)
110 config HAVE_KERNEL_GZIP
113 config HAVE_KERNEL_BZIP2
116 config HAVE_KERNEL_LZMA
119 config HAVE_KERNEL_XZ
122 config HAVE_KERNEL_LZO
125 config HAVE_KERNEL_LZ4
129 prompt "Kernel compression mode"
131 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4
133 The linux kernel is a kind of self-extracting executable.
134 Several compression algorithms are available, which differ
135 in efficiency, compression and decompression speed.
136 Compression speed is only relevant when building a kernel.
137 Decompression speed is relevant at each boot.
139 If you have any problems with bzip2 or lzma compressed
140 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
141 version of this functionality (bzip2 only), for 2.4, was
142 supplied by Christian Ludwig)
144 High compression options are mostly useful for users, who
145 are low on disk space (embedded systems), but for whom ram
148 If in doubt, select 'gzip'
152 depends on HAVE_KERNEL_GZIP
154 The old and tried gzip compression. It provides a good balance
155 between compression ratio and decompression speed.
159 depends on HAVE_KERNEL_BZIP2
161 Its compression ratio and speed is intermediate.
162 Decompression speed is slowest among the choices. The kernel
163 size is about 10% smaller with bzip2, in comparison to gzip.
164 Bzip2 uses a large amount of memory. For modern kernels you
165 will need at least 8MB RAM or more for booting.
169 depends on HAVE_KERNEL_LZMA
171 This compression algorithm's ratio is best. Decompression speed
172 is between gzip and bzip2. Compression is slowest.
173 The kernel size is about 33% smaller with LZMA in comparison to gzip.
177 depends on HAVE_KERNEL_XZ
179 XZ uses the LZMA2 algorithm and instruction set specific
180 BCJ filters which can improve compression ratio of executable
181 code. The size of the kernel is about 30% smaller with XZ in
182 comparison to gzip. On architectures for which there is a BCJ
183 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
184 will create a few percent smaller kernel than plain LZMA.
186 The speed is about the same as with LZMA: The decompression
187 speed of XZ is better than that of bzip2 but worse than gzip
188 and LZO. Compression is slow.
192 depends on HAVE_KERNEL_LZO
194 Its compression ratio is the poorest among the choices. The kernel
195 size is about 10% bigger than gzip; however its speed
196 (both compression and decompression) is the fastest.
200 depends on HAVE_KERNEL_LZ4
202 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
203 A preliminary version of LZ4 de/compression tool is available at
204 <https://code.google.com/p/lz4/>.
206 Its compression ratio is worse than LZO. The size of the kernel
207 is about 8% bigger than LZO. But the decompression speed is
212 config DEFAULT_HOSTNAME
213 string "Default hostname"
216 This option determines the default system hostname before userspace
217 calls sethostname(2). The kernel traditionally uses "(none)" here,
218 but you may wish to use a different default here to make a minimal
219 system more usable with less configuration.
222 bool "Support for paging of anonymous memory (swap)"
223 depends on MMU && BLOCK
226 This option allows you to choose whether you want to have support
227 for so called swap devices or swap files in your kernel that are
228 used to provide more virtual memory than the actual RAM present
229 in your computer. If unsure say Y.
234 Inter Process Communication is a suite of library functions and
235 system calls which let processes (running programs) synchronize and
236 exchange information. It is generally considered to be a good thing,
237 and some programs won't run unless you say Y here. In particular, if
238 you want to run the DOS emulator dosemu under Linux (read the
239 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
240 you'll need to say Y here.
242 You can find documentation about IPC with "info ipc" and also in
243 section 6.4 of the Linux Programmer's Guide, available from
244 <http://www.tldp.org/guides.html>.
246 config SYSVIPC_SYSCTL
253 bool "POSIX Message Queues"
256 POSIX variant of message queues is a part of IPC. In POSIX message
257 queues every message has a priority which decides about succession
258 of receiving it by a process. If you want to compile and run
259 programs written e.g. for Solaris with use of its POSIX message
260 queues (functions mq_*) say Y here.
262 POSIX message queues are visible as a filesystem called 'mqueue'
263 and can be mounted somewhere if you want to do filesystem
264 operations on message queues.
268 config POSIX_MQUEUE_SYSCTL
270 depends on POSIX_MQUEUE
274 config CROSS_MEMORY_ATTACH
275 bool "Enable process_vm_readv/writev syscalls"
279 Enabling this option adds the system calls process_vm_readv and
280 process_vm_writev which allow a process with the correct privileges
281 to directly read from or write to another process' address space.
282 See the man page for more details.
285 bool "open by fhandle syscalls"
288 If you say Y here, a user level program will be able to map
289 file names to handle and then later use the handle for
290 different file system operations. This is useful in implementing
291 userspace file servers, which now track files using handles instead
292 of names. The handle would remain the same even if file names
293 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
297 bool "uselib syscall"
300 This option enables the uselib syscall, a system call used in the
301 dynamic linker from libc5 and earlier. glibc does not use this
302 system call. If you intend to run programs built on libc5 or
303 earlier, you may need to enable this syscall. Current systems
304 running glibc can safely disable this.
307 bool "Auditing support"
310 Enable auditing infrastructure that can be used with another
311 kernel subsystem, such as SELinux (which requires this for
312 logging of avc messages output). Does not do system-call
313 auditing without CONFIG_AUDITSYSCALL.
315 config HAVE_ARCH_AUDITSYSCALL
319 bool "Enable system-call auditing support"
320 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
321 default y if SECURITY_SELINUX
323 Enable low-overhead system-call auditing infrastructure that
324 can be used independently or with another kernel subsystem,
329 depends on AUDITSYSCALL
334 depends on AUDITSYSCALL
337 source "kernel/irq/Kconfig"
338 source "kernel/time/Kconfig"
340 menu "CPU/Task time and stats accounting"
342 config VIRT_CPU_ACCOUNTING
346 prompt "Cputime accounting"
347 default TICK_CPU_ACCOUNTING if !PPC64
348 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
350 # Kind of a stub config for the pure tick based cputime accounting
351 config TICK_CPU_ACCOUNTING
352 bool "Simple tick based cputime accounting"
353 depends on !S390 && !NO_HZ_FULL
355 This is the basic tick based cputime accounting that maintains
356 statistics about user, system and idle time spent on per jiffies
361 config VIRT_CPU_ACCOUNTING_NATIVE
362 bool "Deterministic task and CPU time accounting"
363 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
364 select VIRT_CPU_ACCOUNTING
366 Select this option to enable more accurate task and CPU time
367 accounting. This is done by reading a CPU counter on each
368 kernel entry and exit and on transitions within the kernel
369 between system, softirq and hardirq state, so there is a
370 small performance impact. In the case of s390 or IBM POWER > 5,
371 this also enables accounting of stolen time on logically-partitioned
374 config VIRT_CPU_ACCOUNTING_GEN
375 bool "Full dynticks CPU time accounting"
376 depends on HAVE_CONTEXT_TRACKING
377 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
378 select VIRT_CPU_ACCOUNTING
379 select CONTEXT_TRACKING
381 Select this option to enable task and CPU time accounting on full
382 dynticks systems. This accounting is implemented by watching every
383 kernel-user boundaries using the context tracking subsystem.
384 The accounting is thus performed at the expense of some significant
387 For now this is only useful if you are working on the full
388 dynticks subsystem development.
392 config IRQ_TIME_ACCOUNTING
393 bool "Fine granularity task level IRQ time accounting"
394 depends on HAVE_IRQ_TIME_ACCOUNTING && !NO_HZ_FULL
396 Select this option to enable fine granularity task irq time
397 accounting. This is done by reading a timestamp on each
398 transitions between softirq and hardirq state, so there can be a
399 small performance impact.
401 If in doubt, say N here.
406 bool "Support window based load tracking"
408 depends on FAIR_GROUP_SCHED
410 This feature will allow the scheduler to maintain a tunable window
411 based set of metrics for tasks and runqueues. These metrics can be
412 used to guide task placement as well as task frequency requirements
413 for cpufreq governors.
415 config BSD_PROCESS_ACCT
416 bool "BSD Process Accounting"
419 If you say Y here, a user level program will be able to instruct the
420 kernel (via a special system call) to write process accounting
421 information to a file: whenever a process exits, information about
422 that process will be appended to the file by the kernel. The
423 information includes things such as creation time, owning user,
424 command name, memory usage, controlling terminal etc. (the complete
425 list is in the struct acct in <file:include/linux/acct.h>). It is
426 up to the user level program to do useful things with this
427 information. This is generally a good idea, so say Y.
429 config BSD_PROCESS_ACCT_V3
430 bool "BSD Process Accounting version 3 file format"
431 depends on BSD_PROCESS_ACCT
434 If you say Y here, the process accounting information is written
435 in a new file format that also logs the process IDs of each
436 process and it's parent. Note that this file format is incompatible
437 with previous v0/v1/v2 file formats, so you will need updated tools
438 for processing it. A preliminary version of these tools is available
439 at <http://www.gnu.org/software/acct/>.
442 bool "Export task/process statistics through netlink"
447 Export selected statistics for tasks/processes through the
448 generic netlink interface. Unlike BSD process accounting, the
449 statistics are available during the lifetime of tasks/processes as
450 responses to commands. Like BSD accounting, they are sent to user
455 config TASK_DELAY_ACCT
456 bool "Enable per-task delay accounting"
460 Collect information on time spent by a task waiting for system
461 resources like cpu, synchronous block I/O completion and swapping
462 in pages. Such statistics can help in setting a task's priorities
463 relative to other tasks for cpu, io, rss limits etc.
468 bool "Enable extended accounting over taskstats"
471 Collect extended task accounting data and send the data
472 to userland for processing over the taskstats interface.
476 config TASK_IO_ACCOUNTING
477 bool "Enable per-task storage I/O accounting"
478 depends on TASK_XACCT
480 Collect information on the number of bytes of storage I/O which this
485 endmenu # "CPU/Task time and stats accounting"
491 default y if !PREEMPT && SMP
493 This option selects the RCU implementation that is
494 designed for very large SMP system with hundreds or
495 thousands of CPUs. It also scales down nicely to
502 This option selects the RCU implementation that is
503 designed for very large SMP systems with hundreds or
504 thousands of CPUs, but for which real-time response
505 is also required. It also scales down nicely to
508 Select this option if you are unsure.
512 default y if !PREEMPT && !SMP
514 This option selects the RCU implementation that is
515 designed for UP systems from which real-time response
516 is not required. This option greatly reduces the
517 memory footprint of RCU.
520 bool "Make expert-level adjustments to RCU configuration"
523 This option needs to be enabled if you wish to make
524 expert-level adjustments to RCU configuration. By default,
525 no such adjustments can be made, which has the often-beneficial
526 side-effect of preventing "make oldconfig" from asking you all
527 sorts of detailed questions about how you would like numerous
528 obscure RCU options to be set up.
530 Say Y if you need to make expert-level adjustments to RCU.
532 Say N if you are unsure.
537 This option selects the sleepable version of RCU. This version
538 permits arbitrary sleeping or blocking within RCU read-side critical
546 This option enables a task-based RCU implementation that uses
547 only voluntary context switch (not preemption!), idle, and
548 user-mode execution as quiescent states.
550 config RCU_STALL_COMMON
551 def_bool ( TREE_RCU || PREEMPT_RCU || RCU_TRACE )
553 This option enables RCU CPU stall code that is common between
554 the TINY and TREE variants of RCU. The purpose is to allow
555 the tiny variants to disable RCU CPU stall warnings, while
556 making these warnings mandatory for the tree variants.
558 config CONTEXT_TRACKING
561 config CONTEXT_TRACKING_FORCE
562 bool "Force context tracking"
563 depends on CONTEXT_TRACKING
564 default y if !NO_HZ_FULL
566 The major pre-requirement for full dynticks to work is to
567 support the context tracking subsystem. But there are also
568 other dependencies to provide in order to make the full
571 This option stands for testing when an arch implements the
572 context tracking backend but doesn't yet fullfill all the
573 requirements to make the full dynticks feature working.
574 Without the full dynticks, there is no way to test the support
575 for context tracking and the subsystems that rely on it: RCU
576 userspace extended quiescent state and tickless cputime
577 accounting. This option copes with the absence of the full
578 dynticks subsystem by forcing the context tracking on all
581 Say Y only if you're working on the development of an
582 architecture backend for the context tracking.
584 Say N otherwise, this option brings an overhead that you
585 don't want in production.
589 int "Tree-based hierarchical RCU fanout value"
592 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
596 This option controls the fanout of hierarchical implementations
597 of RCU, allowing RCU to work efficiently on machines with
598 large numbers of CPUs. This value must be at least the fourth
599 root of NR_CPUS, which allows NR_CPUS to be insanely large.
600 The default value of RCU_FANOUT should be used for production
601 systems, but if you are stress-testing the RCU implementation
602 itself, small RCU_FANOUT values allow you to test large-system
603 code paths on small(er) systems.
605 Select a specific number if testing RCU itself.
606 Take the default if unsure.
608 config RCU_FANOUT_LEAF
609 int "Tree-based hierarchical RCU leaf-level fanout value"
612 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
615 This option controls the leaf-level fanout of hierarchical
616 implementations of RCU, and allows trading off cache misses
617 against lock contention. Systems that synchronize their
618 scheduling-clock interrupts for energy-efficiency reasons will
619 want the default because the smaller leaf-level fanout keeps
620 lock contention levels acceptably low. Very large systems
621 (hundreds or thousands of CPUs) will instead want to set this
622 value to the maximum value possible in order to reduce the
623 number of cache misses incurred during RCU's grace-period
624 initialization. These systems tend to run CPU-bound, and thus
625 are not helped by synchronized interrupts, and thus tend to
626 skew them, which reduces lock contention enough that large
627 leaf-level fanouts work well.
629 Select a specific number if testing RCU itself.
631 Select the maximum permissible value for large systems.
633 Take the default if unsure.
635 config RCU_FAST_NO_HZ
636 bool "Accelerate last non-dyntick-idle CPU's grace periods"
637 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
640 This option permits CPUs to enter dynticks-idle state even if
641 they have RCU callbacks queued, and prevents RCU from waking
642 these CPUs up more than roughly once every four jiffies (by
643 default, you can adjust this using the rcutree.rcu_idle_gp_delay
644 parameter), thus improving energy efficiency. On the other
645 hand, this option increases the duration of RCU grace periods,
646 for example, slowing down synchronize_rcu().
648 Say Y if energy efficiency is critically important, and you
649 don't care about increased grace-period durations.
651 Say N if you are unsure.
653 config TREE_RCU_TRACE
654 def_bool RCU_TRACE && ( TREE_RCU || PREEMPT_RCU )
657 This option provides tracing for the TREE_RCU and
658 PREEMPT_RCU implementations, permitting Makefile to
659 trivially select kernel/rcutree_trace.c.
662 bool "Enable RCU priority boosting"
663 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
666 This option boosts the priority of preempted RCU readers that
667 block the current preemptible RCU grace period for too long.
668 This option also prevents heavy loads from blocking RCU
669 callback invocation for all flavors of RCU.
671 Say Y here if you are working with real-time apps or heavy loads
672 Say N here if you are unsure.
674 config RCU_KTHREAD_PRIO
675 int "Real-time priority to use for RCU worker threads"
676 range 1 99 if RCU_BOOST
677 range 0 99 if !RCU_BOOST
678 default 1 if RCU_BOOST
679 default 0 if !RCU_BOOST
680 depends on RCU_EXPERT
682 This option specifies the SCHED_FIFO priority value that will be
683 assigned to the rcuc/n and rcub/n threads and is also the value
684 used for RCU_BOOST (if enabled). If you are working with a
685 real-time application that has one or more CPU-bound threads
686 running at a real-time priority level, you should set
687 RCU_KTHREAD_PRIO to a priority higher than the highest-priority
688 real-time CPU-bound application thread. The default RCU_KTHREAD_PRIO
689 value of 1 is appropriate in the common case, which is real-time
690 applications that do not have any CPU-bound threads.
692 Some real-time applications might not have a single real-time
693 thread that saturates a given CPU, but instead might have
694 multiple real-time threads that, taken together, fully utilize
695 that CPU. In this case, you should set RCU_KTHREAD_PRIO to
696 a priority higher than the lowest-priority thread that is
697 conspiring to prevent the CPU from running any non-real-time
698 tasks. For example, if one thread at priority 10 and another
699 thread at priority 5 are between themselves fully consuming
700 the CPU time on a given CPU, then RCU_KTHREAD_PRIO should be
701 set to priority 6 or higher.
703 Specify the real-time priority, or take the default if unsure.
705 config RCU_BOOST_DELAY
706 int "Milliseconds to delay boosting after RCU grace-period start"
711 This option specifies the time to wait after the beginning of
712 a given grace period before priority-boosting preempted RCU
713 readers blocking that grace period. Note that any RCU reader
714 blocking an expedited RCU grace period is boosted immediately.
716 Accept the default if unsure.
719 bool "Offload RCU callback processing from boot-selected CPUs"
720 depends on TREE_RCU || PREEMPT_RCU
721 depends on RCU_EXPERT || NO_HZ_FULL
724 Use this option to reduce OS jitter for aggressive HPC or
725 real-time workloads. It can also be used to offload RCU
726 callback invocation to energy-efficient CPUs in battery-powered
727 asymmetric multiprocessors.
729 This option offloads callback invocation from the set of
730 CPUs specified at boot time by the rcu_nocbs parameter.
731 For each such CPU, a kthread ("rcuox/N") will be created to
732 invoke callbacks, where the "N" is the CPU being offloaded,
733 and where the "x" is "b" for RCU-bh, "p" for RCU-preempt, and
734 "s" for RCU-sched. Nothing prevents this kthread from running
735 on the specified CPUs, but (1) the kthreads may be preempted
736 between each callback, and (2) affinity or cgroups can be used
737 to force the kthreads to run on whatever set of CPUs is desired.
739 Say Y here if you want to help to debug reduced OS jitter.
740 Say N here if you are unsure.
743 prompt "Build-forced no-CBs CPUs"
744 default RCU_NOCB_CPU_NONE
745 depends on RCU_NOCB_CPU
747 This option allows no-CBs CPUs (whose RCU callbacks are invoked
748 from kthreads rather than from softirq context) to be specified
749 at build time. Additional no-CBs CPUs may be specified by
750 the rcu_nocbs= boot parameter.
752 config RCU_NOCB_CPU_NONE
753 bool "No build_forced no-CBs CPUs"
755 This option does not force any of the CPUs to be no-CBs CPUs.
756 Only CPUs designated by the rcu_nocbs= boot parameter will be
757 no-CBs CPUs, whose RCU callbacks will be invoked by per-CPU
758 kthreads whose names begin with "rcuo". All other CPUs will
759 invoke their own RCU callbacks in softirq context.
761 Select this option if you want to choose no-CBs CPUs at
762 boot time, for example, to allow testing of different no-CBs
763 configurations without having to rebuild the kernel each time.
765 config RCU_NOCB_CPU_ZERO
766 bool "CPU 0 is a build_forced no-CBs CPU"
768 This option forces CPU 0 to be a no-CBs CPU, so that its RCU
769 callbacks are invoked by a per-CPU kthread whose name begins
770 with "rcuo". Additional CPUs may be designated as no-CBs
771 CPUs using the rcu_nocbs= boot parameter will be no-CBs CPUs.
772 All other CPUs will invoke their own RCU callbacks in softirq
775 Select this if CPU 0 needs to be a no-CBs CPU for real-time
776 or energy-efficiency reasons, but the real reason it exists
777 is to ensure that randconfig testing covers mixed systems.
779 config RCU_NOCB_CPU_ALL
780 bool "All CPUs are build_forced no-CBs CPUs"
782 This option forces all CPUs to be no-CBs CPUs. The rcu_nocbs=
783 boot parameter will be ignored. All CPUs' RCU callbacks will
784 be executed in the context of per-CPU rcuo kthreads created for
785 this purpose. Assuming that the kthreads whose names start with
786 "rcuo" are bound to "housekeeping" CPUs, this reduces OS jitter
787 on the remaining CPUs, but might decrease memory locality during
788 RCU-callback invocation, thus potentially degrading throughput.
790 Select this if all CPUs need to be no-CBs CPUs for real-time
791 or energy-efficiency reasons.
795 config RCU_EXPEDITE_BOOT
799 This option enables expedited grace periods at boot time,
800 as if rcu_expedite_gp() had been invoked early in boot.
801 The corresponding rcu_unexpedite_gp() is invoked from
802 rcu_end_inkernel_boot(), which is intended to be invoked
803 at the end of the kernel-only boot sequence, just before
806 Accept the default if unsure.
808 endmenu # "RCU Subsystem"
815 tristate "Kernel .config support"
818 This option enables the complete Linux kernel ".config" file
819 contents to be saved in the kernel. It provides documentation
820 of which kernel options are used in a running kernel or in an
821 on-disk kernel. This information can be extracted from the kernel
822 image file with the script scripts/extract-ikconfig and used as
823 input to rebuild the current kernel or to build another kernel.
824 It can also be extracted from a running kernel by reading
825 /proc/config.gz if enabled (below).
828 bool "Enable access to .config through /proc/config.gz"
829 depends on IKCONFIG && PROC_FS
831 This option enables access to the kernel configuration file
832 through /proc/config.gz.
835 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
840 Select the minimal kernel log buffer size as a power of 2.
841 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
842 parameter, see below. Any higher size also might be forced
843 by "log_buf_len" boot parameter.
853 config LOG_CPU_MAX_BUF_SHIFT
854 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
857 default 12 if !BASE_SMALL
858 default 0 if BASE_SMALL
861 This option allows to increase the default ring buffer size
862 according to the number of CPUs. The value defines the contribution
863 of each CPU as a power of 2. The used space is typically only few
864 lines however it might be much more when problems are reported,
867 The increased size means that a new buffer has to be allocated and
868 the original static one is unused. It makes sense only on systems
869 with more CPUs. Therefore this value is used only when the sum of
870 contributions is greater than the half of the default kernel ring
871 buffer as defined by LOG_BUF_SHIFT. The default values are set
872 so that more than 64 CPUs are needed to trigger the allocation.
874 Also this option is ignored when "log_buf_len" kernel parameter is
875 used as it forces an exact (power of two) size of the ring buffer.
877 The number of possible CPUs is used for this computation ignoring
878 hotplugging making the compuation optimal for the the worst case
879 scenerio while allowing a simple algorithm to be used from bootup.
881 Examples shift values and their meaning:
882 17 => 128 KB for each CPU
883 16 => 64 KB for each CPU
884 15 => 32 KB for each CPU
885 14 => 16 KB for each CPU
886 13 => 8 KB for each CPU
887 12 => 4 KB for each CPU
890 # Architectures with an unreliable sched_clock() should select this:
892 config HAVE_UNSTABLE_SCHED_CLOCK
895 config GENERIC_SCHED_CLOCK
899 # For architectures that want to enable the support for NUMA-affine scheduler
902 config ARCH_SUPPORTS_NUMA_BALANCING
906 # For architectures that prefer to flush all TLBs after a number of pages
907 # are unmapped instead of sending one IPI per page to flush. The architecture
908 # must provide guarantees on what happens if a clean TLB cache entry is
909 # written after the unmap. Details are in mm/rmap.c near the check for
910 # should_defer_flush. The architecture should also consider if the full flush
911 # and the refill costs are offset by the savings of sending fewer IPIs.
912 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
916 # For architectures that know their GCC __int128 support is sound
918 config ARCH_SUPPORTS_INT128
921 # For architectures that (ab)use NUMA to represent different memory regions
922 # all cpu-local but of different latencies, such as SuperH.
924 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
927 config NUMA_BALANCING
928 bool "Memory placement aware NUMA scheduler"
929 depends on ARCH_SUPPORTS_NUMA_BALANCING
930 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
931 depends on SMP && NUMA && MIGRATION
933 This option adds support for automatic NUMA aware memory/task placement.
934 The mechanism is quite primitive and is based on migrating memory when
935 it has references to the node the task is running on.
937 This system will be inactive on UMA systems.
939 config NUMA_BALANCING_DEFAULT_ENABLED
940 bool "Automatically enable NUMA aware memory/task placement"
942 depends on NUMA_BALANCING
944 If set, automatic NUMA balancing will be enabled if running on a NUMA
948 bool "Control Group support"
951 This option adds support for grouping sets of processes together, for
952 use with process control subsystems such as Cpusets, CFS, memory
953 controls or device isolation.
955 - Documentation/scheduler/sched-design-CFS.txt (CFS)
956 - Documentation/cgroups/ (features for grouping, isolation
957 and resource control)
964 bool "Example debug cgroup subsystem"
967 This option enables a simple cgroup subsystem that
968 exports useful debugging information about the cgroups
973 config CGROUP_FREEZER
974 bool "Freezer cgroup subsystem"
976 Provides a way to freeze and unfreeze all tasks in a
980 bool "PIDs cgroup subsystem"
982 Provides enforcement of process number limits in the scope of a
983 cgroup. Any attempt to fork more processes than is allowed in the
984 cgroup will fail. PIDs are fundamentally a global resource because it
985 is fairly trivial to reach PID exhaustion before you reach even a
986 conservative kmemcg limit. As a result, it is possible to grind a
987 system to halt without being limited by other cgroup policies. The
988 PIDs cgroup subsystem is designed to stop this from happening.
990 It should be noted that organisational operations (such as attaching
991 to a cgroup hierarchy will *not* be blocked by the PIDs subsystem),
992 since the PIDs limit only affects a process's ability to fork, not to
996 bool "Device controller for cgroups"
998 Provides a cgroup implementing whitelists for devices which
999 a process in the cgroup can mknod or open.
1002 bool "Cpuset support"
1004 This option will let you create and manage CPUSETs which
1005 allow dynamically partitioning a system into sets of CPUs and
1006 Memory Nodes and assigning tasks to run only within those sets.
1007 This is primarily useful on large SMP or NUMA systems.
1011 config PROC_PID_CPUSET
1012 bool "Include legacy /proc/<pid>/cpuset file"
1016 config CGROUP_CPUACCT
1017 bool "Simple CPU accounting cgroup subsystem"
1019 Provides a simple Resource Controller for monitoring the
1020 total CPU consumed by the tasks in a cgroup.
1022 config CGROUP_SCHEDTUNE
1023 bool "CFS tasks boosting cgroup subsystem (EXPERIMENTAL)"
1024 depends on SCHED_TUNE
1026 This option provides the "schedtune" controller which improves the
1027 flexibility of the task boosting mechanism by introducing the support
1028 to define "per task" boost values.
1030 This new controller:
1031 1. allows only a two layers hierarchy, where the root defines the
1032 system-wide boost value and its direct childrens define each one a
1033 different "class of tasks" to be boosted with a different value
1034 2. supports up to 16 different task classes, each one which could be
1035 configured with a different boost value
1043 bool "Memory Resource Controller for Control Groups"
1047 Provides a memory resource controller that manages both anonymous
1048 memory and page cache. (See Documentation/cgroups/memory.txt)
1051 bool "Memory Resource Controller Swap Extension"
1052 depends on MEMCG && SWAP
1054 Add swap management feature to memory resource controller. When you
1055 enable this, you can limit mem+swap usage per cgroup. In other words,
1056 when you disable this, memory resource controller has no cares to
1057 usage of swap...a process can exhaust all of the swap. This extension
1058 is useful when you want to avoid exhaustion swap but this itself
1059 adds more overheads and consumes memory for remembering information.
1060 Especially if you use 32bit system or small memory system, please
1061 be careful about enabling this. When memory resource controller
1062 is disabled by boot option, this will be automatically disabled and
1063 there will be no overhead from this. Even when you set this config=y,
1064 if boot option "swapaccount=0" is set, swap will not be accounted.
1065 Now, memory usage of swap_cgroup is 2 bytes per entry. If swap page
1066 size is 4096bytes, 512k per 1Gbytes of swap.
1067 config MEMCG_SWAP_ENABLED
1068 bool "Memory Resource Controller Swap Extension enabled by default"
1069 depends on MEMCG_SWAP
1072 Memory Resource Controller Swap Extension comes with its price in
1073 a bigger memory consumption. General purpose distribution kernels
1074 which want to enable the feature but keep it disabled by default
1075 and let the user enable it by swapaccount=1 boot command line
1076 parameter should have this option unselected.
1077 For those who want to have the feature enabled by default should
1078 select this option (if, for some reason, they need to disable it
1079 then swapaccount=0 does the trick).
1081 bool "Memory Resource Controller Kernel Memory accounting"
1083 depends on SLUB || SLAB
1085 The Kernel Memory extension for Memory Resource Controller can limit
1086 the amount of memory used by kernel objects in the system. Those are
1087 fundamentally different from the entities handled by the standard
1088 Memory Controller, which are page-based, and can be swapped. Users of
1089 the kmem extension can use it to guarantee that no group of processes
1090 will ever exhaust kernel resources alone.
1092 config CGROUP_HUGETLB
1093 bool "HugeTLB Resource Controller for Control Groups"
1094 depends on HUGETLB_PAGE
1098 Provides a cgroup Resource Controller for HugeTLB pages.
1099 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1100 The limit is enforced during page fault. Since HugeTLB doesn't
1101 support page reclaim, enforcing the limit at page fault time implies
1102 that, the application will get SIGBUS signal if it tries to access
1103 HugeTLB pages beyond its limit. This requires the application to know
1104 beforehand how much HugeTLB pages it would require for its use. The
1105 control group is tracked in the third page lru pointer. This means
1106 that we cannot use the controller with huge page less than 3 pages.
1109 bool "Enable perf_event per-cpu per-container group (cgroup) monitoring"
1110 depends on PERF_EVENTS && CGROUPS
1112 This option extends the per-cpu mode to restrict monitoring to
1113 threads which belong to the cgroup specified and run on the
1118 menuconfig CGROUP_SCHED
1119 bool "Group CPU scheduler"
1122 This feature lets CPU scheduler recognize task groups and control CPU
1123 bandwidth allocation to such task groups. It uses cgroups to group
1127 config FAIR_GROUP_SCHED
1128 bool "Group scheduling for SCHED_OTHER"
1129 depends on CGROUP_SCHED
1130 default CGROUP_SCHED
1132 config CFS_BANDWIDTH
1133 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
1134 depends on FAIR_GROUP_SCHED
1137 This option allows users to define CPU bandwidth rates (limits) for
1138 tasks running within the fair group scheduler. Groups with no limit
1139 set are considered to be unconstrained and will run with no
1141 See tip/Documentation/scheduler/sched-bwc.txt for more information.
1143 config RT_GROUP_SCHED
1144 bool "Group scheduling for SCHED_RR/FIFO"
1145 depends on CGROUP_SCHED
1148 This feature lets you explicitly allocate real CPU bandwidth
1149 to task groups. If enabled, it will also make it impossible to
1150 schedule realtime tasks for non-root users until you allocate
1151 realtime bandwidth for them.
1152 See Documentation/scheduler/sched-rt-group.txt for more information.
1157 bool "Block IO controller"
1161 Generic block IO controller cgroup interface. This is the common
1162 cgroup interface which should be used by various IO controlling
1165 Currently, CFQ IO scheduler uses it to recognize task groups and
1166 control disk bandwidth allocation (proportional time slice allocation)
1167 to such task groups. It is also used by bio throttling logic in
1168 block layer to implement upper limit in IO rates on a device.
1170 This option only enables generic Block IO controller infrastructure.
1171 One needs to also enable actual IO controlling logic/policy. For
1172 enabling proportional weight division of disk bandwidth in CFQ, set
1173 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
1174 CONFIG_BLK_DEV_THROTTLING=y.
1176 See Documentation/cgroups/blkio-controller.txt for more information.
1178 config DEBUG_BLK_CGROUP
1179 bool "Enable Block IO controller debugging"
1180 depends on BLK_CGROUP
1183 Enable some debugging help. Currently it exports additional stat
1184 files in a cgroup which can be useful for debugging.
1186 config CGROUP_WRITEBACK
1188 depends on MEMCG && BLK_CGROUP
1193 config CHECKPOINT_RESTORE
1194 bool "Checkpoint/restore support" if EXPERT
1195 select PROC_CHILDREN
1198 Enables additional kernel features in a sake of checkpoint/restore.
1199 In particular it adds auxiliary prctl codes to setup process text,
1200 data and heap segment sizes, and a few additional /proc filesystem
1203 If unsure, say N here.
1205 menuconfig NAMESPACES
1206 bool "Namespaces support" if EXPERT
1207 depends on MULTIUSER
1210 Provides the way to make tasks work with different objects using
1211 the same id. For example same IPC id may refer to different objects
1212 or same user id or pid may refer to different tasks when used in
1213 different namespaces.
1218 bool "UTS namespace"
1221 In this namespace tasks see different info provided with the
1225 bool "IPC namespace"
1226 depends on (SYSVIPC || POSIX_MQUEUE)
1229 In this namespace tasks work with IPC ids which correspond to
1230 different IPC objects in different namespaces.
1233 bool "User namespace"
1236 This allows containers, i.e. vservers, to use user namespaces
1237 to provide different user info for different servers.
1239 When user namespaces are enabled in the kernel it is
1240 recommended that the MEMCG and MEMCG_KMEM options also be
1241 enabled and that user-space use the memory control groups to
1242 limit the amount of memory a memory unprivileged users can
1248 bool "PID Namespaces"
1251 Support process id namespaces. This allows having multiple
1252 processes with the same pid as long as they are in different
1253 pid namespaces. This is a building block of containers.
1256 bool "Network namespace"
1260 Allow user space to create what appear to be multiple instances
1261 of the network stack.
1265 config SCHED_AUTOGROUP
1266 bool "Automatic process group scheduling"
1269 select FAIR_GROUP_SCHED
1271 This option optimizes the scheduler for common desktop workloads by
1272 automatically creating and populating task groups. This separation
1273 of workloads isolates aggressive CPU burners (like build jobs) from
1274 desktop applications. Task group autogeneration is currently based
1278 bool "Boosting for CFS tasks (EXPERIMENTAL)"
1281 This option enables the system-wide support for task boosting.
1282 When this support is enabled a new sysctl interface is exposed to
1284 /proc/sys/kernel/sched_cfs_boost
1285 which allows to set a system-wide boost value in range [0..100].
1287 The currently boosting strategy is implemented in such a way that:
1288 - a 0% boost value requires to operate in "standard" mode by
1289 scheduling all tasks at the minimum capacities required by their
1291 - a 100% boost value requires to push at maximum the task
1292 performances, "regardless" of the incurred energy consumption
1294 A boost value in between these two boundaries is used to bias the
1295 power/performance trade-off, the higher the boost value the more the
1296 scheduler is biased toward performance boosting instead of energy
1299 Since this support exposes a single system-wide knob, the specified
1300 boost value is applied to all (CFS) tasks in the system.
1304 config DEFAULT_USE_ENERGY_AWARE
1305 bool "Default to enabling the Energy Aware Scheduler feature"
1308 This option defaults the ENERGY_AWARE scheduling feature to true,
1309 as without SCHED_DEBUG set this feature can't be enabled or disabled
1314 config SYSFS_DEPRECATED
1315 bool "Enable deprecated sysfs features to support old userspace tools"
1319 This option adds code that switches the layout of the "block" class
1320 devices, to not show up in /sys/class/block/, but only in
1323 This switch is only active when the sysfs.deprecated=1 boot option is
1324 passed or the SYSFS_DEPRECATED_V2 option is set.
1326 This option allows new kernels to run on old distributions and tools,
1327 which might get confused by /sys/class/block/. Since 2007/2008 all
1328 major distributions and tools handle this just fine.
1330 Recent distributions and userspace tools after 2009/2010 depend on
1331 the existence of /sys/class/block/, and will not work with this
1334 Only if you are using a new kernel on an old distribution, you might
1337 config SYSFS_DEPRECATED_V2
1338 bool "Enable deprecated sysfs features by default"
1341 depends on SYSFS_DEPRECATED
1343 Enable deprecated sysfs by default.
1345 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1348 Only if you are using a new kernel on an old distribution, you might
1349 need to say Y here. Even then, odds are you would not need it
1350 enabled, you can always pass the boot option if absolutely necessary.
1353 bool "Kernel->user space relay support (formerly relayfs)"
1355 This option enables support for relay interface support in
1356 certain file systems (such as debugfs).
1357 It is designed to provide an efficient mechanism for tools and
1358 facilities to relay large amounts of data from kernel space to
1363 config BLK_DEV_INITRD
1364 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1365 depends on BROKEN || !FRV
1367 The initial RAM filesystem is a ramfs which is loaded by the
1368 boot loader (loadlin or lilo) and that is mounted as root
1369 before the normal boot procedure. It is typically used to
1370 load modules needed to mount the "real" root file system,
1371 etc. See <file:Documentation/initrd.txt> for details.
1373 If RAM disk support (BLK_DEV_RAM) is also included, this
1374 also enables initial RAM disk (initrd) support and adds
1375 15 Kbytes (more on some other architectures) to the kernel size.
1381 source "usr/Kconfig"
1386 prompt "Compiler optimization level"
1387 default CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
1389 config CC_OPTIMIZE_FOR_PERFORMANCE
1390 bool "Optimize for performance"
1392 This is the default optimization level for the kernel, building
1393 with the "-O2" compiler flag for best performance and most
1394 helpful compile-time warnings.
1396 config CC_OPTIMIZE_FOR_SIZE
1397 bool "Optimize for size"
1399 Enabling this option will pass "-Os" instead of "-O2" to
1400 your compiler resulting in a smaller kernel.
1415 config SYSCTL_EXCEPTION_TRACE
1418 Enable support for /proc/sys/debug/exception-trace.
1420 config SYSCTL_ARCH_UNALIGN_NO_WARN
1423 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1424 Allows arch to define/use @no_unaligned_warning to possibly warn
1425 about unaligned access emulation going on under the hood.
1427 config SYSCTL_ARCH_UNALIGN_ALLOW
1430 Enable support for /proc/sys/kernel/unaligned-trap
1431 Allows arches to define/use @unaligned_enabled to runtime toggle
1432 the unaligned access emulation.
1433 see arch/parisc/kernel/unaligned.c for reference
1435 config HAVE_PCSPKR_PLATFORM
1438 # interpreter that classic socket filters depend on
1443 bool "Configure standard kernel features (expert users)"
1444 # Unhide debug options, to make the on-by-default options visible
1447 This option allows certain base kernel options and settings
1448 to be disabled or tweaked. This is for specialized
1449 environments which can tolerate a "non-standard" kernel.
1450 Only use this if you really know what you are doing.
1453 bool "Enable 16-bit UID system calls" if EXPERT
1454 depends on HAVE_UID16 && MULTIUSER
1457 This enables the legacy 16-bit UID syscall wrappers.
1460 bool "Multiple users, groups and capabilities support" if EXPERT
1463 This option enables support for non-root users, groups and
1466 If you say N here, all processes will run with UID 0, GID 0, and all
1467 possible capabilities. Saying N here also compiles out support for
1468 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1471 If unsure, say Y here.
1473 config SGETMASK_SYSCALL
1474 bool "sgetmask/ssetmask syscalls support" if EXPERT
1475 def_bool PARISC || MN10300 || BLACKFIN || M68K || PPC || MIPS || X86 || SPARC || CRIS || MICROBLAZE || SUPERH
1477 sys_sgetmask and sys_ssetmask are obsolete system calls
1478 no longer supported in libc but still enabled by default in some
1481 If unsure, leave the default option here.
1483 config SYSFS_SYSCALL
1484 bool "Sysfs syscall support" if EXPERT
1487 sys_sysfs is an obsolete system call no longer supported in libc.
1488 Note that disabling this option is more secure but might break
1489 compatibility with some systems.
1491 If unsure say Y here.
1493 config SYSCTL_SYSCALL
1494 bool "Sysctl syscall support" if EXPERT
1495 depends on PROC_SYSCTL
1499 sys_sysctl uses binary paths that have been found challenging
1500 to properly maintain and use. The interface in /proc/sys
1501 using paths with ascii names is now the primary path to this
1504 Almost nothing using the binary sysctl interface so if you are
1505 trying to save some space it is probably safe to disable this,
1506 making your kernel marginally smaller.
1508 If unsure say N here.
1511 bool "Load all symbols for debugging/ksymoops" if EXPERT
1514 Say Y here to let the kernel print out symbolic crash information and
1515 symbolic stack backtraces. This increases the size of the kernel
1516 somewhat, as all symbols have to be loaded into the kernel image.
1519 bool "Include all symbols in kallsyms"
1520 depends on DEBUG_KERNEL && KALLSYMS
1522 Normally kallsyms only contains the symbols of functions for nicer
1523 OOPS messages and backtraces (i.e., symbols from the text and inittext
1524 sections). This is sufficient for most cases. And only in very rare
1525 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1526 names of variables from the data sections, etc).
1528 This option makes sure that all symbols are loaded into the kernel
1529 image (i.e., symbols from all sections) in cost of increased kernel
1530 size (depending on the kernel configuration, it may be 300KiB or
1531 something like this).
1533 Say N unless you really need all symbols.
1537 bool "Enable support for printk" if EXPERT
1540 This option enables normal printk support. Removing it
1541 eliminates most of the message strings from the kernel image
1542 and makes the kernel more or less silent. As this makes it
1543 very difficult to diagnose system problems, saying N here is
1544 strongly discouraged.
1547 bool "BUG() support" if EXPERT
1550 Disabling this option eliminates support for BUG and WARN, reducing
1551 the size of your kernel image and potentially quietly ignoring
1552 numerous fatal conditions. You should only consider disabling this
1553 option for embedded systems with no facilities for reporting errors.
1559 bool "Enable ELF core dumps" if EXPERT
1561 Enable support for generating core dumps. Disabling saves about 4k.
1564 config PCSPKR_PLATFORM
1565 bool "Enable PC-Speaker support" if EXPERT
1566 depends on HAVE_PCSPKR_PLATFORM
1570 This option allows to disable the internal PC-Speaker
1571 support, saving some memory.
1575 bool "Enable full-sized data structures for core" if EXPERT
1577 Disabling this option reduces the size of miscellaneous core
1578 kernel data structures. This saves memory on small machines,
1579 but may reduce performance.
1582 bool "Enable futex support" if EXPERT
1586 Disabling this option will cause the kernel to be built without
1587 support for "fast userspace mutexes". The resulting kernel may not
1588 run glibc-based applications correctly.
1590 config HAVE_FUTEX_CMPXCHG
1594 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1595 is implemented and always working. This removes a couple of runtime
1599 bool "Enable eventpoll support" if EXPERT
1603 Disabling this option will cause the kernel to be built without
1604 support for epoll family of system calls.
1607 bool "Enable signalfd() system call" if EXPERT
1611 Enable the signalfd() system call that allows to receive signals
1612 on a file descriptor.
1617 bool "Enable timerfd() system call" if EXPERT
1621 Enable the timerfd() system call that allows to receive timer
1622 events on a file descriptor.
1627 bool "Enable eventfd() system call" if EXPERT
1631 Enable the eventfd() system call that allows to receive both
1632 kernel notification (ie. KAIO) or userspace notifications.
1636 # syscall, maps, verifier
1638 bool "Enable bpf() system call"
1643 Enable the bpf() system call that allows to manipulate eBPF
1644 programs and maps via file descriptors.
1646 config BPF_JIT_ALWAYS_ON
1647 bool "Permanently enable BPF JIT and remove BPF interpreter"
1648 depends on BPF_SYSCALL && HAVE_EBPF_JIT && BPF_JIT
1650 Enables BPF JIT and removes BPF interpreter to avoid
1651 speculative execution of BPF instructions by the interpreter
1654 bool "Use full shmem filesystem" if EXPERT
1658 The shmem is an internal filesystem used to manage shared memory.
1659 It is backed by swap and manages resource limits. It is also exported
1660 to userspace as tmpfs if TMPFS is enabled. Disabling this
1661 option replaces shmem and tmpfs with the much simpler ramfs code,
1662 which may be appropriate on small systems without swap.
1665 bool "Enable AIO support" if EXPERT
1668 This option enables POSIX asynchronous I/O which may by used
1669 by some high performance threaded applications. Disabling
1670 this option saves about 7k.
1672 config ADVISE_SYSCALLS
1673 bool "Enable madvise/fadvise syscalls" if EXPERT
1676 This option enables the madvise and fadvise syscalls, used by
1677 applications to advise the kernel about their future memory or file
1678 usage, improving performance. If building an embedded system where no
1679 applications use these syscalls, you can disable this option to save
1683 bool "Enable userfaultfd() system call"
1687 Enable the userfaultfd() system call that allows to intercept and
1688 handle page faults in userland.
1692 bool "Enable PCI quirk workarounds" if EXPERT
1695 This enables workarounds for various PCI chipset
1696 bugs/quirks. Disable this only if your target machine is
1697 unaffected by PCI quirks.
1700 bool "Enable membarrier() system call" if EXPERT
1703 Enable the membarrier() system call that allows issuing memory
1704 barriers across all running threads, which can be used to distribute
1705 the cost of user-space memory barriers asymmetrically by transforming
1706 pairs of memory barriers into pairs consisting of membarrier() and a
1712 bool "Embedded system"
1713 option allnoconfig_y
1716 This option should be enabled if compiling the kernel for
1717 an embedded system so certain expert options are available
1720 config HAVE_PERF_EVENTS
1723 See tools/perf/design.txt for details.
1725 config PERF_USE_VMALLOC
1728 See tools/perf/design.txt for details
1730 menu "Kernel Performance Events And Counters"
1733 bool "Kernel performance events and counters"
1734 default y if PROFILING
1735 depends on HAVE_PERF_EVENTS
1740 Enable kernel support for various performance events provided
1741 by software and hardware.
1743 Software events are supported either built-in or via the
1744 use of generic tracepoints.
1746 Most modern CPUs support performance events via performance
1747 counter registers. These registers count the number of certain
1748 types of hw events: such as instructions executed, cachemisses
1749 suffered, or branches mis-predicted - without slowing down the
1750 kernel or applications. These registers can also trigger interrupts
1751 when a threshold number of events have passed - and can thus be
1752 used to profile the code that runs on that CPU.
1754 The Linux Performance Event subsystem provides an abstraction of
1755 these software and hardware event capabilities, available via a
1756 system call and used by the "perf" utility in tools/perf/. It
1757 provides per task and per CPU counters, and it provides event
1758 capabilities on top of those.
1762 config DEBUG_PERF_USE_VMALLOC
1764 bool "Debug: use vmalloc to back perf mmap() buffers"
1765 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1766 select PERF_USE_VMALLOC
1768 Use vmalloc memory to back perf mmap() buffers.
1770 Mostly useful for debugging the vmalloc code on platforms
1771 that don't require it.
1777 config VM_EVENT_COUNTERS
1779 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1781 VM event counters are needed for event counts to be shown.
1782 This option allows the disabling of the VM event counters
1783 on EXPERT systems. /proc/vmstat will only show page counts
1784 if VM event counters are disabled.
1788 bool "Enable SLUB debugging support" if EXPERT
1789 depends on SLUB && SYSFS
1791 SLUB has extensive debug support features. Disabling these can
1792 result in significant savings in code size. This also disables
1793 SLUB sysfs support. /sys/slab will not exist and there will be
1794 no support for cache validation etc.
1797 bool "Disable heap randomization"
1800 Randomizing heap placement makes heap exploits harder, but it
1801 also breaks ancient binaries (including anything libc5 based).
1802 This option changes the bootup default to heap randomization
1803 disabled, and can be overridden at runtime by setting
1804 /proc/sys/kernel/randomize_va_space to 2.
1806 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1809 prompt "Choose SLAB allocator"
1812 This option allows to select a slab allocator.
1816 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1818 The regular slab allocator that is established and known to work
1819 well in all environments. It organizes cache hot objects in
1820 per cpu and per node queues.
1823 bool "SLUB (Unqueued Allocator)"
1824 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1826 SLUB is a slab allocator that minimizes cache line usage
1827 instead of managing queues of cached objects (SLAB approach).
1828 Per cpu caching is realized using slabs of objects instead
1829 of queues of objects. SLUB can use memory efficiently
1830 and has enhanced diagnostics. SLUB is the default choice for
1835 bool "SLOB (Simple Allocator)"
1837 SLOB replaces the stock allocator with a drastically simpler
1838 allocator. SLOB is generally more space efficient but
1839 does not perform as well on large systems.
1843 config SLUB_CPU_PARTIAL
1845 depends on SLUB && SMP
1846 bool "SLUB per cpu partial cache"
1848 Per cpu partial caches accellerate objects allocation and freeing
1849 that is local to a processor at the price of more indeterminism
1850 in the latency of the free. On overflow these caches will be cleared
1851 which requires the taking of locks that may cause latency spikes.
1852 Typically one would choose no for a realtime system.
1854 config MMAP_ALLOW_UNINITIALIZED
1855 bool "Allow mmapped anonymous memory to be uninitialized"
1856 depends on EXPERT && !MMU
1859 Normally, and according to the Linux spec, anonymous memory obtained
1860 from mmap() has it's contents cleared before it is passed to
1861 userspace. Enabling this config option allows you to request that
1862 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1863 providing a huge performance boost. If this option is not enabled,
1864 then the flag will be ignored.
1866 This is taken advantage of by uClibc's malloc(), and also by
1867 ELF-FDPIC binfmt's brk and stack allocator.
1869 Because of the obvious security issues, this option should only be
1870 enabled on embedded devices where you control what is run in
1871 userspace. Since that isn't generally a problem on no-MMU systems,
1872 it is normally safe to say Y here.
1874 See Documentation/nommu-mmap.txt for more information.
1876 config SYSTEM_DATA_VERIFICATION
1878 select SYSTEM_TRUSTED_KEYRING
1881 select ASYMMETRIC_KEY_TYPE
1882 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1883 select PUBLIC_KEY_ALGO_RSA
1886 select X509_CERTIFICATE_PARSER
1887 select PKCS7_MESSAGE_PARSER
1889 Provide PKCS#7 message verification using the contents of the system
1890 trusted keyring to provide public keys. This then can be used for
1891 module verification, kexec image verification and firmware blob
1895 bool "Profiling support"
1897 Say Y here to enable the extended profiling support mechanisms used
1898 by profilers such as OProfile.
1901 # Place an empty function call at each tracepoint site. Can be
1902 # dynamically changed for a probe function.
1907 source "arch/Kconfig"
1909 endmenu # General setup
1911 config HAVE_GENERIC_DMA_COHERENT
1918 depends on SLAB || SLUB_DEBUG
1926 default 0 if BASE_FULL
1927 default 1 if !BASE_FULL
1930 bool "Enable loadable module support"
1933 Kernel modules are small pieces of compiled code which can
1934 be inserted in the running kernel, rather than being
1935 permanently built into the kernel. You use the "modprobe"
1936 tool to add (and sometimes remove) them. If you say Y here,
1937 many parts of the kernel can be built as modules (by
1938 answering M instead of Y where indicated): this is most
1939 useful for infrequently used options which are not required
1940 for booting. For more information, see the man pages for
1941 modprobe, lsmod, modinfo, insmod and rmmod.
1943 If you say Y here, you will need to run "make
1944 modules_install" to put the modules under /lib/modules/
1945 where modprobe can find them (you may need to be root to do
1952 config MODULE_FORCE_LOAD
1953 bool "Forced module loading"
1956 Allow loading of modules without version information (ie. modprobe
1957 --force). Forced module loading sets the 'F' (forced) taint flag and
1958 is usually a really bad idea.
1960 config MODULE_UNLOAD
1961 bool "Module unloading"
1963 Without this option you will not be able to unload any
1964 modules (note that some modules may not be unloadable
1965 anyway), which makes your kernel smaller, faster
1966 and simpler. If unsure, say Y.
1968 config MODULE_FORCE_UNLOAD
1969 bool "Forced module unloading"
1970 depends on MODULE_UNLOAD
1972 This option allows you to force a module to unload, even if the
1973 kernel believes it is unsafe: the kernel will remove the module
1974 without waiting for anyone to stop using it (using the -f option to
1975 rmmod). This is mainly for kernel developers and desperate users.
1979 bool "Module versioning support"
1981 Usually, you have to use modules compiled with your kernel.
1982 Saying Y here makes it sometimes possible to use modules
1983 compiled for different kernels, by adding enough information
1984 to the modules to (hopefully) spot any changes which would
1985 make them incompatible with the kernel you are running. If
1988 config MODULE_SRCVERSION_ALL
1989 bool "Source checksum for all modules"
1991 Modules which contain a MODULE_VERSION get an extra "srcversion"
1992 field inserted into their modinfo section, which contains a
1993 sum of the source files which made it. This helps maintainers
1994 see exactly which source was used to build a module (since
1995 others sometimes change the module source without updating
1996 the version). With this option, such a "srcversion" field
1997 will be created for all modules. If unsure, say N.
2000 bool "Module signature verification"
2002 select SYSTEM_DATA_VERIFICATION
2004 Check modules for valid signatures upon load: the signature
2005 is simply appended to the module. For more information see
2006 Documentation/module-signing.txt.
2008 Note that this option adds the OpenSSL development packages as a
2009 kernel build dependency so that the signing tool can use its crypto
2012 !!!WARNING!!! If you enable this option, you MUST make sure that the
2013 module DOES NOT get stripped after being signed. This includes the
2014 debuginfo strip done by some packagers (such as rpmbuild) and
2015 inclusion into an initramfs that wants the module size reduced.
2017 config MODULE_SIG_FORCE
2018 bool "Require modules to be validly signed"
2019 depends on MODULE_SIG
2021 Reject unsigned modules or signed modules for which we don't have a
2022 key. Without this, such modules will simply taint the kernel.
2024 config MODULE_SIG_ALL
2025 bool "Automatically sign all modules"
2027 depends on MODULE_SIG
2029 Sign all modules during make modules_install. Without this option,
2030 modules must be signed manually, using the scripts/sign-file tool.
2032 comment "Do not forget to sign required modules with scripts/sign-file"
2033 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2036 prompt "Which hash algorithm should modules be signed with?"
2037 depends on MODULE_SIG
2039 This determines which sort of hashing algorithm will be used during
2040 signature generation. This algorithm _must_ be built into the kernel
2041 directly so that signature verification can take place. It is not
2042 possible to load a signed module containing the algorithm to check
2043 the signature on that module.
2045 config MODULE_SIG_SHA1
2046 bool "Sign modules with SHA-1"
2049 config MODULE_SIG_SHA224
2050 bool "Sign modules with SHA-224"
2051 select CRYPTO_SHA256
2053 config MODULE_SIG_SHA256
2054 bool "Sign modules with SHA-256"
2055 select CRYPTO_SHA256
2057 config MODULE_SIG_SHA384
2058 bool "Sign modules with SHA-384"
2059 select CRYPTO_SHA512
2061 config MODULE_SIG_SHA512
2062 bool "Sign modules with SHA-512"
2063 select CRYPTO_SHA512
2067 config MODULE_SIG_HASH
2069 depends on MODULE_SIG
2070 default "sha1" if MODULE_SIG_SHA1
2071 default "sha224" if MODULE_SIG_SHA224
2072 default "sha256" if MODULE_SIG_SHA256
2073 default "sha384" if MODULE_SIG_SHA384
2074 default "sha512" if MODULE_SIG_SHA512
2076 config MODULE_COMPRESS
2077 bool "Compress modules on installation"
2081 Compresses kernel modules when 'make modules_install' is run; gzip or
2082 xz depending on "Compression algorithm" below.
2084 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2086 Out-of-tree kernel modules installed using Kbuild will also be
2087 compressed upon installation.
2089 Note: for modules inside an initrd or initramfs, it's more efficient
2090 to compress the whole initrd or initramfs instead.
2092 Note: This is fully compatible with signed modules.
2097 prompt "Compression algorithm"
2098 depends on MODULE_COMPRESS
2099 default MODULE_COMPRESS_GZIP
2101 This determines which sort of compression will be used during
2102 'make modules_install'.
2104 GZIP (default) and XZ are supported.
2106 config MODULE_COMPRESS_GZIP
2109 config MODULE_COMPRESS_XZ
2116 config MODULES_TREE_LOOKUP
2118 depends on PERF_EVENTS || TRACING
2120 config INIT_ALL_POSSIBLE
2123 Back when each arch used to define their own cpu_online_mask and
2124 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2125 with all 1s, and others with all 0s. When they were centralised,
2126 it was better to provide this option than to break all the archs
2127 and have several arch maintainers pursuing me down dark alleys.
2129 source "block/Kconfig"
2131 config PREEMPT_NOTIFIERS
2138 # Can be selected by architectures with broken toolchains
2139 # that get confused by correct const<->read_only section
2141 config BROKEN_RODATA
2147 Build a simple ASN.1 grammar compiler that produces a bytecode output
2148 that can be interpreted by the ASN.1 stream decoder and used to
2149 inform it as to what tags are to be expected in a stream and what
2150 functions to call on what tags.
2152 source "kernel/Kconfig.locks"