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
36 depends on BROKEN || !SMP
39 config INIT_ENV_ARG_LIMIT
44 Maximum of each of the number of arguments and environment
45 variables passed to init from the kernel command line.
49 string "Cross-compiler tool prefix"
51 Same as running 'make CROSS_COMPILE=prefix-' but stored for
52 default make runs in this kernel build directory. You don't
53 need to set this unless you want the configured kernel build
54 directory to select the cross-compiler automatically.
57 bool "Compile also drivers which will not load"
60 Some drivers can be compiled on a different platform than they are
61 intended to be run on. Despite they cannot be loaded there (or even
62 when they load they cannot be used due to missing HW support),
63 developers still, opposing to distributors, might want to build such
64 drivers to compile-test them.
66 If you are a developer and want to build everything available, say Y
67 here. If you are a user/distributor, say N here to exclude useless
68 drivers to be distributed.
71 string "Local version - append to kernel release"
73 Append an extra string to the end of your kernel version.
74 This will show up when you type uname, for example.
75 The string you set here will be appended after the contents of
76 any files with a filename matching localversion* in your
77 object and source tree, in that order. Your total string can
78 be a maximum of 64 characters.
80 config LOCALVERSION_AUTO
81 bool "Automatically append version information to the version string"
84 This will try to automatically determine if the current tree is a
85 release tree by looking for git tags that belong to the current
88 A string of the format -gxxxxxxxx will be added to the localversion
89 if a git-based tree is found. The string generated by this will be
90 appended after any matching localversion* files, and after the value
91 set in CONFIG_LOCALVERSION.
93 (The actual string used here is the first eight characters produced
94 by running the command:
96 $ git rev-parse --verify HEAD
98 which is done within the script "scripts/setlocalversion".)
100 config HAVE_KERNEL_GZIP
103 config HAVE_KERNEL_BZIP2
106 config HAVE_KERNEL_LZMA
109 config HAVE_KERNEL_XZ
112 config HAVE_KERNEL_LZO
115 config HAVE_KERNEL_LZ4
119 prompt "Kernel compression mode"
121 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4
123 The linux kernel is a kind of self-extracting executable.
124 Several compression algorithms are available, which differ
125 in efficiency, compression and decompression speed.
126 Compression speed is only relevant when building a kernel.
127 Decompression speed is relevant at each boot.
129 If you have any problems with bzip2 or lzma compressed
130 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
131 version of this functionality (bzip2 only), for 2.4, was
132 supplied by Christian Ludwig)
134 High compression options are mostly useful for users, who
135 are low on disk space (embedded systems), but for whom ram
138 If in doubt, select 'gzip'
142 depends on HAVE_KERNEL_GZIP
144 The old and tried gzip compression. It provides a good balance
145 between compression ratio and decompression speed.
149 depends on HAVE_KERNEL_BZIP2
151 Its compression ratio and speed is intermediate.
152 Decompression speed is slowest among the choices. The kernel
153 size is about 10% smaller with bzip2, in comparison to gzip.
154 Bzip2 uses a large amount of memory. For modern kernels you
155 will need at least 8MB RAM or more for booting.
159 depends on HAVE_KERNEL_LZMA
161 This compression algorithm's ratio is best. Decompression speed
162 is between gzip and bzip2. Compression is slowest.
163 The kernel size is about 33% smaller with LZMA in comparison to gzip.
167 depends on HAVE_KERNEL_XZ
169 XZ uses the LZMA2 algorithm and instruction set specific
170 BCJ filters which can improve compression ratio of executable
171 code. The size of the kernel is about 30% smaller with XZ in
172 comparison to gzip. On architectures for which there is a BCJ
173 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
174 will create a few percent smaller kernel than plain LZMA.
176 The speed is about the same as with LZMA: The decompression
177 speed of XZ is better than that of bzip2 but worse than gzip
178 and LZO. Compression is slow.
182 depends on HAVE_KERNEL_LZO
184 Its compression ratio is the poorest among the choices. The kernel
185 size is about 10% bigger than gzip; however its speed
186 (both compression and decompression) is the fastest.
190 depends on HAVE_KERNEL_LZ4
192 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
193 A preliminary version of LZ4 de/compression tool is available at
194 <https://code.google.com/p/lz4/>.
196 Its compression ratio is worse than LZO. The size of the kernel
197 is about 8% bigger than LZO. But the decompression speed is
202 config DEFAULT_HOSTNAME
203 string "Default hostname"
206 This option determines the default system hostname before userspace
207 calls sethostname(2). The kernel traditionally uses "(none)" here,
208 but you may wish to use a different default here to make a minimal
209 system more usable with less configuration.
212 bool "Support for paging of anonymous memory (swap)"
213 depends on MMU && BLOCK
216 This option allows you to choose whether you want to have support
217 for so called swap devices or swap files in your kernel that are
218 used to provide more virtual memory than the actual RAM present
219 in your computer. If unsure say Y.
224 Inter Process Communication is a suite of library functions and
225 system calls which let processes (running programs) synchronize and
226 exchange information. It is generally considered to be a good thing,
227 and some programs won't run unless you say Y here. In particular, if
228 you want to run the DOS emulator dosemu under Linux (read the
229 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
230 you'll need to say Y here.
232 You can find documentation about IPC with "info ipc" and also in
233 section 6.4 of the Linux Programmer's Guide, available from
234 <http://www.tldp.org/guides.html>.
236 config SYSVIPC_SYSCTL
243 bool "POSIX Message Queues"
246 POSIX variant of message queues is a part of IPC. In POSIX message
247 queues every message has a priority which decides about succession
248 of receiving it by a process. If you want to compile and run
249 programs written e.g. for Solaris with use of its POSIX message
250 queues (functions mq_*) say Y here.
252 POSIX message queues are visible as a filesystem called 'mqueue'
253 and can be mounted somewhere if you want to do filesystem
254 operations on message queues.
258 config POSIX_MQUEUE_SYSCTL
260 depends on POSIX_MQUEUE
264 config CROSS_MEMORY_ATTACH
265 bool "Enable process_vm_readv/writev syscalls"
269 Enabling this option adds the system calls process_vm_readv and
270 process_vm_writev which allow a process with the correct privileges
271 to directly read from or write to another process' address space.
272 See the man page for more details.
275 bool "open by fhandle syscalls" if EXPERT
279 If you say Y here, a user level program will be able to map
280 file names to handle and then later use the handle for
281 different file system operations. This is useful in implementing
282 userspace file servers, which now track files using handles instead
283 of names. The handle would remain the same even if file names
284 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
288 bool "uselib syscall"
289 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
291 This option enables the uselib syscall, a system call used in the
292 dynamic linker from libc5 and earlier. glibc does not use this
293 system call. If you intend to run programs built on libc5 or
294 earlier, you may need to enable this syscall. Current systems
295 running glibc can safely disable this.
298 bool "Auditing support"
301 Enable auditing infrastructure that can be used with another
302 kernel subsystem, such as SELinux (which requires this for
303 logging of avc messages output). System call auditing is included
304 on architectures which support it.
306 config HAVE_ARCH_AUDITSYSCALL
311 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
315 depends on AUDITSYSCALL
320 depends on AUDITSYSCALL
323 source "kernel/irq/Kconfig"
324 source "kernel/time/Kconfig"
326 menu "CPU/Task time and stats accounting"
328 config VIRT_CPU_ACCOUNTING
332 prompt "Cputime accounting"
333 default TICK_CPU_ACCOUNTING if !PPC64
334 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
336 # Kind of a stub config for the pure tick based cputime accounting
337 config TICK_CPU_ACCOUNTING
338 bool "Simple tick based cputime accounting"
339 depends on !S390 && !NO_HZ_FULL
341 This is the basic tick based cputime accounting that maintains
342 statistics about user, system and idle time spent on per jiffies
347 config VIRT_CPU_ACCOUNTING_NATIVE
348 bool "Deterministic task and CPU time accounting"
349 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
350 select VIRT_CPU_ACCOUNTING
352 Select this option to enable more accurate task and CPU time
353 accounting. This is done by reading a CPU counter on each
354 kernel entry and exit and on transitions within the kernel
355 between system, softirq and hardirq state, so there is a
356 small performance impact. In the case of s390 or IBM POWER > 5,
357 this also enables accounting of stolen time on logically-partitioned
360 config VIRT_CPU_ACCOUNTING_GEN
361 bool "Full dynticks CPU time accounting"
362 depends on HAVE_CONTEXT_TRACKING
363 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
364 select VIRT_CPU_ACCOUNTING
365 select CONTEXT_TRACKING
367 Select this option to enable task and CPU time accounting on full
368 dynticks systems. This accounting is implemented by watching every
369 kernel-user boundaries using the context tracking subsystem.
370 The accounting is thus performed at the expense of some significant
373 For now this is only useful if you are working on the full
374 dynticks subsystem development.
378 config IRQ_TIME_ACCOUNTING
379 bool "Fine granularity task level IRQ time accounting"
380 depends on HAVE_IRQ_TIME_ACCOUNTING && !NO_HZ_FULL
382 Select this option to enable fine granularity task irq time
383 accounting. This is done by reading a timestamp on each
384 transitions between softirq and hardirq state, so there can be a
385 small performance impact.
387 If in doubt, say N here.
391 config BSD_PROCESS_ACCT
392 bool "BSD Process Accounting"
395 If you say Y here, a user level program will be able to instruct the
396 kernel (via a special system call) to write process accounting
397 information to a file: whenever a process exits, information about
398 that process will be appended to the file by the kernel. The
399 information includes things such as creation time, owning user,
400 command name, memory usage, controlling terminal etc. (the complete
401 list is in the struct acct in <file:include/linux/acct.h>). It is
402 up to the user level program to do useful things with this
403 information. This is generally a good idea, so say Y.
405 config BSD_PROCESS_ACCT_V3
406 bool "BSD Process Accounting version 3 file format"
407 depends on BSD_PROCESS_ACCT
410 If you say Y here, the process accounting information is written
411 in a new file format that also logs the process IDs of each
412 process and it's parent. Note that this file format is incompatible
413 with previous v0/v1/v2 file formats, so you will need updated tools
414 for processing it. A preliminary version of these tools is available
415 at <http://www.gnu.org/software/acct/>.
418 bool "Export task/process statistics through netlink"
423 Export selected statistics for tasks/processes through the
424 generic netlink interface. Unlike BSD process accounting, the
425 statistics are available during the lifetime of tasks/processes as
426 responses to commands. Like BSD accounting, they are sent to user
431 config TASK_DELAY_ACCT
432 bool "Enable per-task delay accounting"
436 Collect information on time spent by a task waiting for system
437 resources like cpu, synchronous block I/O completion and swapping
438 in pages. Such statistics can help in setting a task's priorities
439 relative to other tasks for cpu, io, rss limits etc.
444 bool "Enable extended accounting over taskstats"
447 Collect extended task accounting data and send the data
448 to userland for processing over the taskstats interface.
452 config TASK_IO_ACCOUNTING
453 bool "Enable per-task storage I/O accounting"
454 depends on TASK_XACCT
456 Collect information on the number of bytes of storage I/O which this
461 endmenu # "CPU/Task time and stats accounting"
467 default y if !PREEMPT && SMP
469 This option selects the RCU implementation that is
470 designed for very large SMP system with hundreds or
471 thousands of CPUs. It also scales down nicely to
478 This option selects the RCU implementation that is
479 designed for very large SMP systems with hundreds or
480 thousands of CPUs, but for which real-time response
481 is also required. It also scales down nicely to
484 Select this option if you are unsure.
488 default y if !PREEMPT && !SMP
490 This option selects the RCU implementation that is
491 designed for UP systems from which real-time response
492 is not required. This option greatly reduces the
493 memory footprint of RCU.
496 bool "Make expert-level adjustments to RCU configuration"
499 This option needs to be enabled if you wish to make
500 expert-level adjustments to RCU configuration. By default,
501 no such adjustments can be made, which has the often-beneficial
502 side-effect of preventing "make oldconfig" from asking you all
503 sorts of detailed questions about how you would like numerous
504 obscure RCU options to be set up.
506 Say Y if you need to make expert-level adjustments to RCU.
508 Say N if you are unsure.
513 This option selects the sleepable version of RCU. This version
514 permits arbitrary sleeping or blocking within RCU read-side critical
522 This option enables a task-based RCU implementation that uses
523 only voluntary context switch (not preemption!), idle, and
524 user-mode execution as quiescent states.
526 config RCU_STALL_COMMON
527 def_bool ( TREE_RCU || PREEMPT_RCU || RCU_TRACE )
529 This option enables RCU CPU stall code that is common between
530 the TINY and TREE variants of RCU. The purpose is to allow
531 the tiny variants to disable RCU CPU stall warnings, while
532 making these warnings mandatory for the tree variants.
534 config CONTEXT_TRACKING
537 config CONTEXT_TRACKING_FORCE
538 bool "Force context tracking"
539 depends on CONTEXT_TRACKING
540 default y if !NO_HZ_FULL
542 The major pre-requirement for full dynticks to work is to
543 support the context tracking subsystem. But there are also
544 other dependencies to provide in order to make the full
547 This option stands for testing when an arch implements the
548 context tracking backend but doesn't yet fullfill all the
549 requirements to make the full dynticks feature working.
550 Without the full dynticks, there is no way to test the support
551 for context tracking and the subsystems that rely on it: RCU
552 userspace extended quiescent state and tickless cputime
553 accounting. This option copes with the absence of the full
554 dynticks subsystem by forcing the context tracking on all
557 Say Y only if you're working on the development of an
558 architecture backend for the context tracking.
560 Say N otherwise, this option brings an overhead that you
561 don't want in production.
565 int "Tree-based hierarchical RCU fanout value"
568 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
572 This option controls the fanout of hierarchical implementations
573 of RCU, allowing RCU to work efficiently on machines with
574 large numbers of CPUs. This value must be at least the fourth
575 root of NR_CPUS, which allows NR_CPUS to be insanely large.
576 The default value of RCU_FANOUT should be used for production
577 systems, but if you are stress-testing the RCU implementation
578 itself, small RCU_FANOUT values allow you to test large-system
579 code paths on small(er) systems.
581 Select a specific number if testing RCU itself.
582 Take the default if unsure.
584 config RCU_FANOUT_LEAF
585 int "Tree-based hierarchical RCU leaf-level fanout value"
588 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
591 This option controls the leaf-level fanout of hierarchical
592 implementations of RCU, and allows trading off cache misses
593 against lock contention. Systems that synchronize their
594 scheduling-clock interrupts for energy-efficiency reasons will
595 want the default because the smaller leaf-level fanout keeps
596 lock contention levels acceptably low. Very large systems
597 (hundreds or thousands of CPUs) will instead want to set this
598 value to the maximum value possible in order to reduce the
599 number of cache misses incurred during RCU's grace-period
600 initialization. These systems tend to run CPU-bound, and thus
601 are not helped by synchronized interrupts, and thus tend to
602 skew them, which reduces lock contention enough that large
603 leaf-level fanouts work well.
605 Select a specific number if testing RCU itself.
607 Select the maximum permissible value for large systems.
609 Take the default if unsure.
611 config RCU_FAST_NO_HZ
612 bool "Accelerate last non-dyntick-idle CPU's grace periods"
613 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
616 This option permits CPUs to enter dynticks-idle state even if
617 they have RCU callbacks queued, and prevents RCU from waking
618 these CPUs up more than roughly once every four jiffies (by
619 default, you can adjust this using the rcutree.rcu_idle_gp_delay
620 parameter), thus improving energy efficiency. On the other
621 hand, this option increases the duration of RCU grace periods,
622 for example, slowing down synchronize_rcu().
624 Say Y if energy efficiency is critically important, and you
625 don't care about increased grace-period durations.
627 Say N if you are unsure.
629 config TREE_RCU_TRACE
630 def_bool RCU_TRACE && ( TREE_RCU || PREEMPT_RCU )
633 This option provides tracing for the TREE_RCU and
634 PREEMPT_RCU implementations, permitting Makefile to
635 trivially select kernel/rcutree_trace.c.
638 bool "Enable RCU priority boosting"
639 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
642 This option boosts the priority of preempted RCU readers that
643 block the current preemptible RCU grace period for too long.
644 This option also prevents heavy loads from blocking RCU
645 callback invocation for all flavors of RCU.
647 Say Y here if you are working with real-time apps or heavy loads
648 Say N here if you are unsure.
650 config RCU_KTHREAD_PRIO
651 int "Real-time priority to use for RCU worker threads"
652 range 1 99 if RCU_BOOST
653 range 0 99 if !RCU_BOOST
654 default 1 if RCU_BOOST
655 default 0 if !RCU_BOOST
656 depends on RCU_EXPERT
658 This option specifies the SCHED_FIFO priority value that will be
659 assigned to the rcuc/n and rcub/n threads and is also the value
660 used for RCU_BOOST (if enabled). If you are working with a
661 real-time application that has one or more CPU-bound threads
662 running at a real-time priority level, you should set
663 RCU_KTHREAD_PRIO to a priority higher than the highest-priority
664 real-time CPU-bound application thread. The default RCU_KTHREAD_PRIO
665 value of 1 is appropriate in the common case, which is real-time
666 applications that do not have any CPU-bound threads.
668 Some real-time applications might not have a single real-time
669 thread that saturates a given CPU, but instead might have
670 multiple real-time threads that, taken together, fully utilize
671 that CPU. In this case, you should set RCU_KTHREAD_PRIO to
672 a priority higher than the lowest-priority thread that is
673 conspiring to prevent the CPU from running any non-real-time
674 tasks. For example, if one thread at priority 10 and another
675 thread at priority 5 are between themselves fully consuming
676 the CPU time on a given CPU, then RCU_KTHREAD_PRIO should be
677 set to priority 6 or higher.
679 Specify the real-time priority, or take the default if unsure.
681 config RCU_BOOST_DELAY
682 int "Milliseconds to delay boosting after RCU grace-period start"
687 This option specifies the time to wait after the beginning of
688 a given grace period before priority-boosting preempted RCU
689 readers blocking that grace period. Note that any RCU reader
690 blocking an expedited RCU grace period is boosted immediately.
692 Accept the default if unsure.
695 bool "Offload RCU callback processing from boot-selected CPUs"
696 depends on TREE_RCU || PREEMPT_RCU
697 depends on RCU_EXPERT || NO_HZ_FULL
700 Use this option to reduce OS jitter for aggressive HPC or
701 real-time workloads. It can also be used to offload RCU
702 callback invocation to energy-efficient CPUs in battery-powered
703 asymmetric multiprocessors.
705 This option offloads callback invocation from the set of
706 CPUs specified at boot time by the rcu_nocbs parameter.
707 For each such CPU, a kthread ("rcuox/N") will be created to
708 invoke callbacks, where the "N" is the CPU being offloaded,
709 and where the "x" is "b" for RCU-bh, "p" for RCU-preempt, and
710 "s" for RCU-sched. Nothing prevents this kthread from running
711 on the specified CPUs, but (1) the kthreads may be preempted
712 between each callback, and (2) affinity or cgroups can be used
713 to force the kthreads to run on whatever set of CPUs is desired.
715 Say Y here if you want to help to debug reduced OS jitter.
716 Say N here if you are unsure.
719 prompt "Build-forced no-CBs CPUs"
720 default RCU_NOCB_CPU_NONE
721 depends on RCU_NOCB_CPU
723 This option allows no-CBs CPUs (whose RCU callbacks are invoked
724 from kthreads rather than from softirq context) to be specified
725 at build time. Additional no-CBs CPUs may be specified by
726 the rcu_nocbs= boot parameter.
728 config RCU_NOCB_CPU_NONE
729 bool "No build_forced no-CBs CPUs"
731 This option does not force any of the CPUs to be no-CBs CPUs.
732 Only CPUs designated by the rcu_nocbs= boot parameter will be
733 no-CBs CPUs, whose RCU callbacks will be invoked by per-CPU
734 kthreads whose names begin with "rcuo". All other CPUs will
735 invoke their own RCU callbacks in softirq context.
737 Select this option if you want to choose no-CBs CPUs at
738 boot time, for example, to allow testing of different no-CBs
739 configurations without having to rebuild the kernel each time.
741 config RCU_NOCB_CPU_ZERO
742 bool "CPU 0 is a build_forced no-CBs CPU"
744 This option forces CPU 0 to be a no-CBs CPU, so that its RCU
745 callbacks are invoked by a per-CPU kthread whose name begins
746 with "rcuo". Additional CPUs may be designated as no-CBs
747 CPUs using the rcu_nocbs= boot parameter will be no-CBs CPUs.
748 All other CPUs will invoke their own RCU callbacks in softirq
751 Select this if CPU 0 needs to be a no-CBs CPU for real-time
752 or energy-efficiency reasons, but the real reason it exists
753 is to ensure that randconfig testing covers mixed systems.
755 config RCU_NOCB_CPU_ALL
756 bool "All CPUs are build_forced no-CBs CPUs"
758 This option forces all CPUs to be no-CBs CPUs. The rcu_nocbs=
759 boot parameter will be ignored. All CPUs' RCU callbacks will
760 be executed in the context of per-CPU rcuo kthreads created for
761 this purpose. Assuming that the kthreads whose names start with
762 "rcuo" are bound to "housekeeping" CPUs, this reduces OS jitter
763 on the remaining CPUs, but might decrease memory locality during
764 RCU-callback invocation, thus potentially degrading throughput.
766 Select this if all CPUs need to be no-CBs CPUs for real-time
767 or energy-efficiency reasons.
771 config RCU_EXPEDITE_BOOT
775 This option enables expedited grace periods at boot time,
776 as if rcu_expedite_gp() had been invoked early in boot.
777 The corresponding rcu_unexpedite_gp() is invoked from
778 rcu_end_inkernel_boot(), which is intended to be invoked
779 at the end of the kernel-only boot sequence, just before
782 Accept the default if unsure.
784 endmenu # "RCU Subsystem"
791 tristate "Kernel .config support"
794 This option enables the complete Linux kernel ".config" file
795 contents to be saved in the kernel. It provides documentation
796 of which kernel options are used in a running kernel or in an
797 on-disk kernel. This information can be extracted from the kernel
798 image file with the script scripts/extract-ikconfig and used as
799 input to rebuild the current kernel or to build another kernel.
800 It can also be extracted from a running kernel by reading
801 /proc/config.gz if enabled (below).
804 bool "Enable access to .config through /proc/config.gz"
805 depends on IKCONFIG && PROC_FS
807 This option enables access to the kernel configuration file
808 through /proc/config.gz.
811 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
816 Select the minimal kernel log buffer size as a power of 2.
817 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
818 parameter, see below. Any higher size also might be forced
819 by "log_buf_len" boot parameter.
829 config LOG_CPU_MAX_BUF_SHIFT
830 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
833 default 12 if !BASE_SMALL
834 default 0 if BASE_SMALL
837 This option allows to increase the default ring buffer size
838 according to the number of CPUs. The value defines the contribution
839 of each CPU as a power of 2. The used space is typically only few
840 lines however it might be much more when problems are reported,
843 The increased size means that a new buffer has to be allocated and
844 the original static one is unused. It makes sense only on systems
845 with more CPUs. Therefore this value is used only when the sum of
846 contributions is greater than the half of the default kernel ring
847 buffer as defined by LOG_BUF_SHIFT. The default values are set
848 so that more than 64 CPUs are needed to trigger the allocation.
850 Also this option is ignored when "log_buf_len" kernel parameter is
851 used as it forces an exact (power of two) size of the ring buffer.
853 The number of possible CPUs is used for this computation ignoring
854 hotplugging making the compuation optimal for the the worst case
855 scenerio while allowing a simple algorithm to be used from bootup.
857 Examples shift values and their meaning:
858 17 => 128 KB for each CPU
859 16 => 64 KB for each CPU
860 15 => 32 KB for each CPU
861 14 => 16 KB for each CPU
862 13 => 8 KB for each CPU
863 12 => 4 KB for each CPU
865 config NMI_LOG_BUF_SHIFT
866 int "Temporary per-CPU NMI log buffer size (12 => 4KB, 13 => 8KB)"
869 depends on PRINTK_NMI
871 Select the size of a per-CPU buffer where NMI messages are temporary
872 stored. They are copied to the main log buffer in a safe context
873 to avoid a deadlock. The value defines the size as a power of 2.
875 NMI messages are rare and limited. The largest one is when
876 a backtrace is printed. It usually fits into 4KB. Select
877 8KB if you want to be on the safe side.
880 17 => 128 KB for each CPU
881 16 => 64 KB for each CPU
882 15 => 32 KB for each CPU
883 14 => 16 KB for each CPU
884 13 => 8 KB for each CPU
885 12 => 4 KB for each CPU
888 # Architectures with an unreliable sched_clock() should select this:
890 config HAVE_UNSTABLE_SCHED_CLOCK
893 config GENERIC_SCHED_CLOCK
897 # For architectures that want to enable the support for NUMA-affine scheduler
900 config ARCH_SUPPORTS_NUMA_BALANCING
904 # For architectures that prefer to flush all TLBs after a number of pages
905 # are unmapped instead of sending one IPI per page to flush. The architecture
906 # must provide guarantees on what happens if a clean TLB cache entry is
907 # written after the unmap. Details are in mm/rmap.c near the check for
908 # should_defer_flush. The architecture should also consider if the full flush
909 # and the refill costs are offset by the savings of sending fewer IPIs.
910 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
914 # For architectures that know their GCC __int128 support is sound
916 config ARCH_SUPPORTS_INT128
919 # For architectures that (ab)use NUMA to represent different memory regions
920 # all cpu-local but of different latencies, such as SuperH.
922 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
925 config NUMA_BALANCING
926 bool "Memory placement aware NUMA scheduler"
927 depends on ARCH_SUPPORTS_NUMA_BALANCING
928 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
929 depends on SMP && NUMA && MIGRATION
931 This option adds support for automatic NUMA aware memory/task placement.
932 The mechanism is quite primitive and is based on migrating memory when
933 it has references to the node the task is running on.
935 This system will be inactive on UMA systems.
937 config NUMA_BALANCING_DEFAULT_ENABLED
938 bool "Automatically enable NUMA aware memory/task placement"
940 depends on NUMA_BALANCING
942 If set, automatic NUMA balancing will be enabled if running on a NUMA
946 bool "Control Group support"
949 This option adds support for grouping sets of processes together, for
950 use with process control subsystems such as Cpusets, CFS, memory
951 controls or device isolation.
953 - Documentation/scheduler/sched-design-CFS.txt (CFS)
954 - Documentation/cgroups/ (features for grouping, isolation
955 and resource control)
965 bool "Memory controller"
969 Provides control over the memory footprint of tasks in a cgroup.
972 bool "Swap controller"
973 depends on MEMCG && SWAP
975 Provides control over the swap space consumed by tasks in a cgroup.
977 config MEMCG_SWAP_ENABLED
978 bool "Swap controller enabled by default"
979 depends on MEMCG_SWAP
982 Memory Resource Controller Swap Extension comes with its price in
983 a bigger memory consumption. General purpose distribution kernels
984 which want to enable the feature but keep it disabled by default
985 and let the user enable it by swapaccount=1 boot command line
986 parameter should have this option unselected.
987 For those who want to have the feature enabled by default should
988 select this option (if, for some reason, they need to disable it
989 then swapaccount=0 does the trick).
996 Generic block IO controller cgroup interface. This is the common
997 cgroup interface which should be used by various IO controlling
1000 Currently, CFQ IO scheduler uses it to recognize task groups and
1001 control disk bandwidth allocation (proportional time slice allocation)
1002 to such task groups. It is also used by bio throttling logic in
1003 block layer to implement upper limit in IO rates on a device.
1005 This option only enables generic Block IO controller infrastructure.
1006 One needs to also enable actual IO controlling logic/policy. For
1007 enabling proportional weight division of disk bandwidth in CFQ, set
1008 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
1009 CONFIG_BLK_DEV_THROTTLING=y.
1011 See Documentation/cgroups/blkio-controller.txt for more information.
1013 config DEBUG_BLK_CGROUP
1014 bool "IO controller debugging"
1015 depends on BLK_CGROUP
1018 Enable some debugging help. Currently it exports additional stat
1019 files in a cgroup which can be useful for debugging.
1021 config CGROUP_WRITEBACK
1023 depends on MEMCG && BLK_CGROUP
1026 menuconfig CGROUP_SCHED
1027 bool "CPU controller"
1030 This feature lets CPU scheduler recognize task groups and control CPU
1031 bandwidth allocation to such task groups. It uses cgroups to group
1035 config FAIR_GROUP_SCHED
1036 bool "Group scheduling for SCHED_OTHER"
1037 depends on CGROUP_SCHED
1038 default CGROUP_SCHED
1040 config CFS_BANDWIDTH
1041 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
1042 depends on FAIR_GROUP_SCHED
1045 This option allows users to define CPU bandwidth rates (limits) for
1046 tasks running within the fair group scheduler. Groups with no limit
1047 set are considered to be unconstrained and will run with no
1049 See tip/Documentation/scheduler/sched-bwc.txt for more information.
1051 config RT_GROUP_SCHED
1052 bool "Group scheduling for SCHED_RR/FIFO"
1053 depends on CGROUP_SCHED
1056 This feature lets you explicitly allocate real CPU bandwidth
1057 to task groups. If enabled, it will also make it impossible to
1058 schedule realtime tasks for non-root users until you allocate
1059 realtime bandwidth for them.
1060 See Documentation/scheduler/sched-rt-group.txt for more information.
1065 bool "PIDs controller"
1067 Provides enforcement of process number limits in the scope of a
1068 cgroup. Any attempt to fork more processes than is allowed in the
1069 cgroup will fail. PIDs are fundamentally a global resource because it
1070 is fairly trivial to reach PID exhaustion before you reach even a
1071 conservative kmemcg limit. As a result, it is possible to grind a
1072 system to halt without being limited by other cgroup policies. The
1073 PIDs controller is designed to stop this from happening.
1075 It should be noted that organisational operations (such as attaching
1076 to a cgroup hierarchy will *not* be blocked by the PIDs controller),
1077 since the PIDs limit only affects a process's ability to fork, not to
1080 config CGROUP_FREEZER
1081 bool "Freezer controller"
1083 Provides a way to freeze and unfreeze all tasks in a
1086 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1087 controller includes important in-kernel memory consumers per default.
1089 If you're using cgroup2, say N.
1091 config CGROUP_HUGETLB
1092 bool "HugeTLB controller"
1093 depends on HUGETLB_PAGE
1097 Provides a cgroup controller for HugeTLB pages.
1098 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1099 The limit is enforced during page fault. Since HugeTLB doesn't
1100 support page reclaim, enforcing the limit at page fault time implies
1101 that, the application will get SIGBUS signal if it tries to access
1102 HugeTLB pages beyond its limit. This requires the application to know
1103 beforehand how much HugeTLB pages it would require for its use. The
1104 control group is tracked in the third page lru pointer. This means
1105 that we cannot use the controller with huge page less than 3 pages.
1108 bool "Cpuset controller"
1110 This option will let you create and manage CPUSETs which
1111 allow dynamically partitioning a system into sets of CPUs and
1112 Memory Nodes and assigning tasks to run only within those sets.
1113 This is primarily useful on large SMP or NUMA systems.
1117 config PROC_PID_CPUSET
1118 bool "Include legacy /proc/<pid>/cpuset file"
1122 config CGROUP_DEVICE
1123 bool "Device controller"
1125 Provides a cgroup controller implementing whitelists for
1126 devices which a process in the cgroup can mknod or open.
1128 config CGROUP_CPUACCT
1129 bool "Simple CPU accounting controller"
1131 Provides a simple controller for monitoring the
1132 total CPU consumed by the tasks in a cgroup.
1135 bool "Perf controller"
1136 depends on PERF_EVENTS
1138 This option extends the perf per-cpu mode to restrict monitoring
1139 to threads which belong to the cgroup specified and run on the
1145 bool "Example controller"
1148 This option enables a simple controller that exports
1149 debugging information about the cgroups framework.
1155 config CHECKPOINT_RESTORE
1156 bool "Checkpoint/restore support" if EXPERT
1157 select PROC_CHILDREN
1160 Enables additional kernel features in a sake of checkpoint/restore.
1161 In particular it adds auxiliary prctl codes to setup process text,
1162 data and heap segment sizes, and a few additional /proc filesystem
1165 If unsure, say N here.
1167 menuconfig NAMESPACES
1168 bool "Namespaces support" if EXPERT
1169 depends on MULTIUSER
1172 Provides the way to make tasks work with different objects using
1173 the same id. For example same IPC id may refer to different objects
1174 or same user id or pid may refer to different tasks when used in
1175 different namespaces.
1180 bool "UTS namespace"
1183 In this namespace tasks see different info provided with the
1187 bool "IPC namespace"
1188 depends on (SYSVIPC || POSIX_MQUEUE)
1191 In this namespace tasks work with IPC ids which correspond to
1192 different IPC objects in different namespaces.
1195 bool "User namespace"
1198 This allows containers, i.e. vservers, to use user namespaces
1199 to provide different user info for different servers.
1201 When user namespaces are enabled in the kernel it is
1202 recommended that the MEMCG option also be enabled and that
1203 user-space use the memory control groups to limit the amount
1204 of memory a memory unprivileged users can use.
1209 bool "PID Namespaces"
1212 Support process id namespaces. This allows having multiple
1213 processes with the same pid as long as they are in different
1214 pid namespaces. This is a building block of containers.
1217 bool "Network namespace"
1221 Allow user space to create what appear to be multiple instances
1222 of the network stack.
1226 config SCHED_AUTOGROUP
1227 bool "Automatic process group scheduling"
1230 select FAIR_GROUP_SCHED
1232 This option optimizes the scheduler for common desktop workloads by
1233 automatically creating and populating task groups. This separation
1234 of workloads isolates aggressive CPU burners (like build jobs) from
1235 desktop applications. Task group autogeneration is currently based
1238 config SYSFS_DEPRECATED
1239 bool "Enable deprecated sysfs features to support old userspace tools"
1243 This option adds code that switches the layout of the "block" class
1244 devices, to not show up in /sys/class/block/, but only in
1247 This switch is only active when the sysfs.deprecated=1 boot option is
1248 passed or the SYSFS_DEPRECATED_V2 option is set.
1250 This option allows new kernels to run on old distributions and tools,
1251 which might get confused by /sys/class/block/. Since 2007/2008 all
1252 major distributions and tools handle this just fine.
1254 Recent distributions and userspace tools after 2009/2010 depend on
1255 the existence of /sys/class/block/, and will not work with this
1258 Only if you are using a new kernel on an old distribution, you might
1261 config SYSFS_DEPRECATED_V2
1262 bool "Enable deprecated sysfs features by default"
1265 depends on SYSFS_DEPRECATED
1267 Enable deprecated sysfs by default.
1269 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1272 Only if you are using a new kernel on an old distribution, you might
1273 need to say Y here. Even then, odds are you would not need it
1274 enabled, you can always pass the boot option if absolutely necessary.
1277 bool "Kernel->user space relay support (formerly relayfs)"
1279 This option enables support for relay interface support in
1280 certain file systems (such as debugfs).
1281 It is designed to provide an efficient mechanism for tools and
1282 facilities to relay large amounts of data from kernel space to
1287 config BLK_DEV_INITRD
1288 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1289 depends on BROKEN || !FRV
1291 The initial RAM filesystem is a ramfs which is loaded by the
1292 boot loader (loadlin or lilo) and that is mounted as root
1293 before the normal boot procedure. It is typically used to
1294 load modules needed to mount the "real" root file system,
1295 etc. See <file:Documentation/initrd.txt> for details.
1297 If RAM disk support (BLK_DEV_RAM) is also included, this
1298 also enables initial RAM disk (initrd) support and adds
1299 15 Kbytes (more on some other architectures) to the kernel size.
1305 source "usr/Kconfig"
1310 prompt "Compiler optimization level"
1311 default CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
1313 config CC_OPTIMIZE_FOR_PERFORMANCE
1314 bool "Optimize for performance"
1316 This is the default optimization level for the kernel, building
1317 with the "-O2" compiler flag for best performance and most
1318 helpful compile-time warnings.
1320 config CC_OPTIMIZE_FOR_SIZE
1321 bool "Optimize for size"
1323 Enabling this option will pass "-Os" instead of "-O2" to
1324 your compiler resulting in a smaller kernel.
1339 config SYSCTL_EXCEPTION_TRACE
1342 Enable support for /proc/sys/debug/exception-trace.
1344 config SYSCTL_ARCH_UNALIGN_NO_WARN
1347 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1348 Allows arch to define/use @no_unaligned_warning to possibly warn
1349 about unaligned access emulation going on under the hood.
1351 config SYSCTL_ARCH_UNALIGN_ALLOW
1354 Enable support for /proc/sys/kernel/unaligned-trap
1355 Allows arches to define/use @unaligned_enabled to runtime toggle
1356 the unaligned access emulation.
1357 see arch/parisc/kernel/unaligned.c for reference
1359 config HAVE_PCSPKR_PLATFORM
1362 # interpreter that classic socket filters depend on
1367 bool "Configure standard kernel features (expert users)"
1368 # Unhide debug options, to make the on-by-default options visible
1371 This option allows certain base kernel options and settings
1372 to be disabled or tweaked. This is for specialized
1373 environments which can tolerate a "non-standard" kernel.
1374 Only use this if you really know what you are doing.
1377 bool "Enable 16-bit UID system calls" if EXPERT
1378 depends on HAVE_UID16 && MULTIUSER
1381 This enables the legacy 16-bit UID syscall wrappers.
1384 bool "Multiple users, groups and capabilities support" if EXPERT
1387 This option enables support for non-root users, groups and
1390 If you say N here, all processes will run with UID 0, GID 0, and all
1391 possible capabilities. Saying N here also compiles out support for
1392 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1395 If unsure, say Y here.
1397 config SGETMASK_SYSCALL
1398 bool "sgetmask/ssetmask syscalls support" if EXPERT
1399 def_bool PARISC || MN10300 || BLACKFIN || M68K || PPC || MIPS || X86 || SPARC || CRIS || MICROBLAZE || SUPERH
1401 sys_sgetmask and sys_ssetmask are obsolete system calls
1402 no longer supported in libc but still enabled by default in some
1405 If unsure, leave the default option here.
1407 config SYSFS_SYSCALL
1408 bool "Sysfs syscall support" if EXPERT
1411 sys_sysfs is an obsolete system call no longer supported in libc.
1412 Note that disabling this option is more secure but might break
1413 compatibility with some systems.
1415 If unsure say Y here.
1417 config SYSCTL_SYSCALL
1418 bool "Sysctl syscall support" if EXPERT
1419 depends on PROC_SYSCTL
1423 sys_sysctl uses binary paths that have been found challenging
1424 to properly maintain and use. The interface in /proc/sys
1425 using paths with ascii names is now the primary path to this
1428 Almost nothing using the binary sysctl interface so if you are
1429 trying to save some space it is probably safe to disable this,
1430 making your kernel marginally smaller.
1432 If unsure say N here.
1435 bool "Load all symbols for debugging/ksymoops" if EXPERT
1438 Say Y here to let the kernel print out symbolic crash information and
1439 symbolic stack backtraces. This increases the size of the kernel
1440 somewhat, as all symbols have to be loaded into the kernel image.
1443 bool "Include all symbols in kallsyms"
1444 depends on DEBUG_KERNEL && KALLSYMS
1446 Normally kallsyms only contains the symbols of functions for nicer
1447 OOPS messages and backtraces (i.e., symbols from the text and inittext
1448 sections). This is sufficient for most cases. And only in very rare
1449 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1450 names of variables from the data sections, etc).
1452 This option makes sure that all symbols are loaded into the kernel
1453 image (i.e., symbols from all sections) in cost of increased kernel
1454 size (depending on the kernel configuration, it may be 300KiB or
1455 something like this).
1457 Say N unless you really need all symbols.
1459 config KALLSYMS_ABSOLUTE_PERCPU
1461 default X86_64 && SMP
1463 config KALLSYMS_BASE_RELATIVE
1466 default !IA64 && !(TILE && 64BIT)
1468 Instead of emitting them as absolute values in the native word size,
1469 emit the symbol references in the kallsyms table as 32-bit entries,
1470 each containing a relative value in the range [base, base + U32_MAX]
1471 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1472 an absolute value in the range [0, S32_MAX] or a relative value in the
1473 range [base, base + S32_MAX], where base is the lowest relative symbol
1474 address encountered in the image.
1476 On 64-bit builds, this reduces the size of the address table by 50%,
1477 but more importantly, it results in entries whose values are build
1478 time constants, and no relocation pass is required at runtime to fix
1479 up the entries based on the runtime load address of the kernel.
1483 bool "Enable support for printk" if EXPERT
1486 This option enables normal printk support. Removing it
1487 eliminates most of the message strings from the kernel image
1488 and makes the kernel more or less silent. As this makes it
1489 very difficult to diagnose system problems, saying N here is
1490 strongly discouraged.
1498 bool "BUG() support" if EXPERT
1501 Disabling this option eliminates support for BUG and WARN, reducing
1502 the size of your kernel image and potentially quietly ignoring
1503 numerous fatal conditions. You should only consider disabling this
1504 option for embedded systems with no facilities for reporting errors.
1510 bool "Enable ELF core dumps" if EXPERT
1512 Enable support for generating core dumps. Disabling saves about 4k.
1515 config PCSPKR_PLATFORM
1516 bool "Enable PC-Speaker support" if EXPERT
1517 depends on HAVE_PCSPKR_PLATFORM
1521 This option allows to disable the internal PC-Speaker
1522 support, saving some memory.
1526 bool "Enable full-sized data structures for core" if EXPERT
1528 Disabling this option reduces the size of miscellaneous core
1529 kernel data structures. This saves memory on small machines,
1530 but may reduce performance.
1533 bool "Enable futex support" if EXPERT
1537 Disabling this option will cause the kernel to be built without
1538 support for "fast userspace mutexes". The resulting kernel may not
1539 run glibc-based applications correctly.
1541 config HAVE_FUTEX_CMPXCHG
1545 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1546 is implemented and always working. This removes a couple of runtime
1550 bool "Enable eventpoll support" if EXPERT
1554 Disabling this option will cause the kernel to be built without
1555 support for epoll family of system calls.
1558 bool "Enable signalfd() system call" if EXPERT
1562 Enable the signalfd() system call that allows to receive signals
1563 on a file descriptor.
1568 bool "Enable timerfd() system call" if EXPERT
1572 Enable the timerfd() system call that allows to receive timer
1573 events on a file descriptor.
1578 bool "Enable eventfd() system call" if EXPERT
1582 Enable the eventfd() system call that allows to receive both
1583 kernel notification (ie. KAIO) or userspace notifications.
1587 # syscall, maps, verifier
1589 bool "Enable bpf() system call"
1594 Enable the bpf() system call that allows to manipulate eBPF
1595 programs and maps via file descriptors.
1598 bool "Use full shmem filesystem" if EXPERT
1602 The shmem is an internal filesystem used to manage shared memory.
1603 It is backed by swap and manages resource limits. It is also exported
1604 to userspace as tmpfs if TMPFS is enabled. Disabling this
1605 option replaces shmem and tmpfs with the much simpler ramfs code,
1606 which may be appropriate on small systems without swap.
1609 bool "Enable AIO support" if EXPERT
1612 This option enables POSIX asynchronous I/O which may by used
1613 by some high performance threaded applications. Disabling
1614 this option saves about 7k.
1616 config ADVISE_SYSCALLS
1617 bool "Enable madvise/fadvise syscalls" if EXPERT
1620 This option enables the madvise and fadvise syscalls, used by
1621 applications to advise the kernel about their future memory or file
1622 usage, improving performance. If building an embedded system where no
1623 applications use these syscalls, you can disable this option to save
1627 bool "Enable userfaultfd() system call"
1631 Enable the userfaultfd() system call that allows to intercept and
1632 handle page faults in userland.
1636 bool "Enable PCI quirk workarounds" if EXPERT
1639 This enables workarounds for various PCI chipset
1640 bugs/quirks. Disable this only if your target machine is
1641 unaffected by PCI quirks.
1644 bool "Enable membarrier() system call" if EXPERT
1647 Enable the membarrier() system call that allows issuing memory
1648 barriers across all running threads, which can be used to distribute
1649 the cost of user-space memory barriers asymmetrically by transforming
1650 pairs of memory barriers into pairs consisting of membarrier() and a
1656 bool "Embedded system"
1657 option allnoconfig_y
1660 This option should be enabled if compiling the kernel for
1661 an embedded system so certain expert options are available
1664 config HAVE_PERF_EVENTS
1667 See tools/perf/design.txt for details.
1669 config PERF_USE_VMALLOC
1672 See tools/perf/design.txt for details
1674 menu "Kernel Performance Events And Counters"
1677 bool "Kernel performance events and counters"
1678 default y if PROFILING
1679 depends on HAVE_PERF_EVENTS
1684 Enable kernel support for various performance events provided
1685 by software and hardware.
1687 Software events are supported either built-in or via the
1688 use of generic tracepoints.
1690 Most modern CPUs support performance events via performance
1691 counter registers. These registers count the number of certain
1692 types of hw events: such as instructions executed, cachemisses
1693 suffered, or branches mis-predicted - without slowing down the
1694 kernel or applications. These registers can also trigger interrupts
1695 when a threshold number of events have passed - and can thus be
1696 used to profile the code that runs on that CPU.
1698 The Linux Performance Event subsystem provides an abstraction of
1699 these software and hardware event capabilities, available via a
1700 system call and used by the "perf" utility in tools/perf/. It
1701 provides per task and per CPU counters, and it provides event
1702 capabilities on top of those.
1706 config DEBUG_PERF_USE_VMALLOC
1708 bool "Debug: use vmalloc to back perf mmap() buffers"
1709 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1710 select PERF_USE_VMALLOC
1712 Use vmalloc memory to back perf mmap() buffers.
1714 Mostly useful for debugging the vmalloc code on platforms
1715 that don't require it.
1721 config VM_EVENT_COUNTERS
1723 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1725 VM event counters are needed for event counts to be shown.
1726 This option allows the disabling of the VM event counters
1727 on EXPERT systems. /proc/vmstat will only show page counts
1728 if VM event counters are disabled.
1732 bool "Enable SLUB debugging support" if EXPERT
1733 depends on SLUB && SYSFS
1735 SLUB has extensive debug support features. Disabling these can
1736 result in significant savings in code size. This also disables
1737 SLUB sysfs support. /sys/slab will not exist and there will be
1738 no support for cache validation etc.
1741 bool "Disable heap randomization"
1744 Randomizing heap placement makes heap exploits harder, but it
1745 also breaks ancient binaries (including anything libc5 based).
1746 This option changes the bootup default to heap randomization
1747 disabled, and can be overridden at runtime by setting
1748 /proc/sys/kernel/randomize_va_space to 2.
1750 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1753 prompt "Choose SLAB allocator"
1756 This option allows to select a slab allocator.
1761 The regular slab allocator that is established and known to work
1762 well in all environments. It organizes cache hot objects in
1763 per cpu and per node queues.
1766 bool "SLUB (Unqueued Allocator)"
1768 SLUB is a slab allocator that minimizes cache line usage
1769 instead of managing queues of cached objects (SLAB approach).
1770 Per cpu caching is realized using slabs of objects instead
1771 of queues of objects. SLUB can use memory efficiently
1772 and has enhanced diagnostics. SLUB is the default choice for
1777 bool "SLOB (Simple Allocator)"
1779 SLOB replaces the stock allocator with a drastically simpler
1780 allocator. SLOB is generally more space efficient but
1781 does not perform as well on large systems.
1785 config SLAB_FREELIST_RANDOM
1788 bool "SLAB freelist randomization"
1790 Randomizes the freelist order used on creating new SLABs. This
1791 security feature reduces the predictability of the kernel slab
1792 allocator against heap overflows.
1794 config SLUB_CPU_PARTIAL
1796 depends on SLUB && SMP
1797 bool "SLUB per cpu partial cache"
1799 Per cpu partial caches accellerate objects allocation and freeing
1800 that is local to a processor at the price of more indeterminism
1801 in the latency of the free. On overflow these caches will be cleared
1802 which requires the taking of locks that may cause latency spikes.
1803 Typically one would choose no for a realtime system.
1805 config MMAP_ALLOW_UNINITIALIZED
1806 bool "Allow mmapped anonymous memory to be uninitialized"
1807 depends on EXPERT && !MMU
1810 Normally, and according to the Linux spec, anonymous memory obtained
1811 from mmap() has it's contents cleared before it is passed to
1812 userspace. Enabling this config option allows you to request that
1813 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1814 providing a huge performance boost. If this option is not enabled,
1815 then the flag will be ignored.
1817 This is taken advantage of by uClibc's malloc(), and also by
1818 ELF-FDPIC binfmt's brk and stack allocator.
1820 Because of the obvious security issues, this option should only be
1821 enabled on embedded devices where you control what is run in
1822 userspace. Since that isn't generally a problem on no-MMU systems,
1823 it is normally safe to say Y here.
1825 See Documentation/nommu-mmap.txt for more information.
1827 config SYSTEM_DATA_VERIFICATION
1829 select SYSTEM_TRUSTED_KEYRING
1833 select ASYMMETRIC_KEY_TYPE
1834 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1837 select X509_CERTIFICATE_PARSER
1838 select PKCS7_MESSAGE_PARSER
1840 Provide PKCS#7 message verification using the contents of the system
1841 trusted keyring to provide public keys. This then can be used for
1842 module verification, kexec image verification and firmware blob
1846 bool "Profiling support"
1848 Say Y here to enable the extended profiling support mechanisms used
1849 by profilers such as OProfile.
1852 # Place an empty function call at each tracepoint site. Can be
1853 # dynamically changed for a probe function.
1858 source "arch/Kconfig"
1860 endmenu # General setup
1862 config HAVE_GENERIC_DMA_COHERENT
1869 depends on SLAB || SLUB_DEBUG
1877 default 0 if BASE_FULL
1878 default 1 if !BASE_FULL
1881 bool "Enable loadable module support"
1884 Kernel modules are small pieces of compiled code which can
1885 be inserted in the running kernel, rather than being
1886 permanently built into the kernel. You use the "modprobe"
1887 tool to add (and sometimes remove) them. If you say Y here,
1888 many parts of the kernel can be built as modules (by
1889 answering M instead of Y where indicated): this is most
1890 useful for infrequently used options which are not required
1891 for booting. For more information, see the man pages for
1892 modprobe, lsmod, modinfo, insmod and rmmod.
1894 If you say Y here, you will need to run "make
1895 modules_install" to put the modules under /lib/modules/
1896 where modprobe can find them (you may need to be root to do
1903 config MODULE_FORCE_LOAD
1904 bool "Forced module loading"
1907 Allow loading of modules without version information (ie. modprobe
1908 --force). Forced module loading sets the 'F' (forced) taint flag and
1909 is usually a really bad idea.
1911 config MODULE_UNLOAD
1912 bool "Module unloading"
1914 Without this option you will not be able to unload any
1915 modules (note that some modules may not be unloadable
1916 anyway), which makes your kernel smaller, faster
1917 and simpler. If unsure, say Y.
1919 config MODULE_FORCE_UNLOAD
1920 bool "Forced module unloading"
1921 depends on MODULE_UNLOAD
1923 This option allows you to force a module to unload, even if the
1924 kernel believes it is unsafe: the kernel will remove the module
1925 without waiting for anyone to stop using it (using the -f option to
1926 rmmod). This is mainly for kernel developers and desperate users.
1930 bool "Module versioning support"
1932 Usually, you have to use modules compiled with your kernel.
1933 Saying Y here makes it sometimes possible to use modules
1934 compiled for different kernels, by adding enough information
1935 to the modules to (hopefully) spot any changes which would
1936 make them incompatible with the kernel you are running. If
1939 config MODULE_SRCVERSION_ALL
1940 bool "Source checksum for all modules"
1942 Modules which contain a MODULE_VERSION get an extra "srcversion"
1943 field inserted into their modinfo section, which contains a
1944 sum of the source files which made it. This helps maintainers
1945 see exactly which source was used to build a module (since
1946 others sometimes change the module source without updating
1947 the version). With this option, such a "srcversion" field
1948 will be created for all modules. If unsure, say N.
1951 bool "Module signature verification"
1953 select SYSTEM_DATA_VERIFICATION
1955 Check modules for valid signatures upon load: the signature
1956 is simply appended to the module. For more information see
1957 Documentation/module-signing.txt.
1959 Note that this option adds the OpenSSL development packages as a
1960 kernel build dependency so that the signing tool can use its crypto
1963 !!!WARNING!!! If you enable this option, you MUST make sure that the
1964 module DOES NOT get stripped after being signed. This includes the
1965 debuginfo strip done by some packagers (such as rpmbuild) and
1966 inclusion into an initramfs that wants the module size reduced.
1968 config MODULE_SIG_FORCE
1969 bool "Require modules to be validly signed"
1970 depends on MODULE_SIG
1972 Reject unsigned modules or signed modules for which we don't have a
1973 key. Without this, such modules will simply taint the kernel.
1975 config MODULE_SIG_ALL
1976 bool "Automatically sign all modules"
1978 depends on MODULE_SIG
1980 Sign all modules during make modules_install. Without this option,
1981 modules must be signed manually, using the scripts/sign-file tool.
1983 comment "Do not forget to sign required modules with scripts/sign-file"
1984 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
1987 prompt "Which hash algorithm should modules be signed with?"
1988 depends on MODULE_SIG
1990 This determines which sort of hashing algorithm will be used during
1991 signature generation. This algorithm _must_ be built into the kernel
1992 directly so that signature verification can take place. It is not
1993 possible to load a signed module containing the algorithm to check
1994 the signature on that module.
1996 config MODULE_SIG_SHA1
1997 bool "Sign modules with SHA-1"
2000 config MODULE_SIG_SHA224
2001 bool "Sign modules with SHA-224"
2002 select CRYPTO_SHA256
2004 config MODULE_SIG_SHA256
2005 bool "Sign modules with SHA-256"
2006 select CRYPTO_SHA256
2008 config MODULE_SIG_SHA384
2009 bool "Sign modules with SHA-384"
2010 select CRYPTO_SHA512
2012 config MODULE_SIG_SHA512
2013 bool "Sign modules with SHA-512"
2014 select CRYPTO_SHA512
2018 config MODULE_SIG_HASH
2020 depends on MODULE_SIG
2021 default "sha1" if MODULE_SIG_SHA1
2022 default "sha224" if MODULE_SIG_SHA224
2023 default "sha256" if MODULE_SIG_SHA256
2024 default "sha384" if MODULE_SIG_SHA384
2025 default "sha512" if MODULE_SIG_SHA512
2027 config MODULE_COMPRESS
2028 bool "Compress modules on installation"
2032 Compresses kernel modules when 'make modules_install' is run; gzip or
2033 xz depending on "Compression algorithm" below.
2035 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2037 Out-of-tree kernel modules installed using Kbuild will also be
2038 compressed upon installation.
2040 Note: for modules inside an initrd or initramfs, it's more efficient
2041 to compress the whole initrd or initramfs instead.
2043 Note: This is fully compatible with signed modules.
2048 prompt "Compression algorithm"
2049 depends on MODULE_COMPRESS
2050 default MODULE_COMPRESS_GZIP
2052 This determines which sort of compression will be used during
2053 'make modules_install'.
2055 GZIP (default) and XZ are supported.
2057 config MODULE_COMPRESS_GZIP
2060 config MODULE_COMPRESS_XZ
2065 config TRIM_UNUSED_KSYMS
2066 bool "Trim unused exported kernel symbols"
2067 depends on MODULES && !UNUSED_SYMBOLS
2069 The kernel and some modules make many symbols available for
2070 other modules to use via EXPORT_SYMBOL() and variants. Depending
2071 on the set of modules being selected in your kernel configuration,
2072 many of those exported symbols might never be used.
2074 This option allows for unused exported symbols to be dropped from
2075 the build. In turn, this provides the compiler more opportunities
2076 (especially when using LTO) for optimizing the code and reducing
2077 binary size. This might have some security advantages as well.
2083 config MODULES_TREE_LOOKUP
2085 depends on PERF_EVENTS || TRACING
2087 config INIT_ALL_POSSIBLE
2090 Back when each arch used to define their own cpu_online_mask and
2091 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2092 with all 1s, and others with all 0s. When they were centralised,
2093 it was better to provide this option than to break all the archs
2094 and have several arch maintainers pursuing me down dark alleys.
2096 source "block/Kconfig"
2098 config PREEMPT_NOTIFIERS
2105 # Can be selected by architectures with broken toolchains
2106 # that get confused by correct const<->read_only section
2108 config BROKEN_RODATA
2114 Build a simple ASN.1 grammar compiler that produces a bytecode output
2115 that can be interpreted by the ASN.1 stream decoder and used to
2116 inform it as to what tags are to be expected in a stream and what
2117 functions to call on what tags.
2119 source "kernel/Kconfig.locks"