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.
43 depends on BROKEN || !SMP
46 config INIT_ENV_ARG_LIMIT
51 Maximum of each of the number of arguments and environment
52 variables passed to init from the kernel command line.
56 string "Cross-compiler tool prefix"
58 Same as running 'make CROSS_COMPILE=prefix-' but stored for
59 default make runs in this kernel build directory. You don't
60 need to set this unless you want the configured kernel build
61 directory to select the cross-compiler automatically.
64 bool "Compile also drivers which will not load"
68 Some drivers can be compiled on a different platform than they are
69 intended to be run on. Despite they cannot be loaded there (or even
70 when they load they cannot be used due to missing HW support),
71 developers still, opposing to distributors, might want to build such
72 drivers to compile-test them.
74 If you are a developer and want to build everything available, say Y
75 here. If you are a user/distributor, say N here to exclude useless
76 drivers to be distributed.
79 string "Local version - append to kernel release"
81 Append an extra string to the end of your kernel version.
82 This will show up when you type uname, for example.
83 The string you set here will be appended after the contents of
84 any files with a filename matching localversion* in your
85 object and source tree, in that order. Your total string can
86 be a maximum of 64 characters.
88 config LOCALVERSION_AUTO
89 bool "Automatically append version information to the version string"
91 depends on !COMPILE_TEST
93 This will try to automatically determine if the current tree is a
94 release tree by looking for git tags that belong to the current
97 A string of the format -gxxxxxxxx will be added to the localversion
98 if a git-based tree is found. The string generated by this will be
99 appended after any matching localversion* files, and after the value
100 set in CONFIG_LOCALVERSION.
102 (The actual string used here is the first eight characters produced
103 by running the command:
105 $ git rev-parse --verify HEAD
107 which is done within the script "scripts/setlocalversion".)
109 config HAVE_KERNEL_GZIP
112 config HAVE_KERNEL_BZIP2
115 config HAVE_KERNEL_LZMA
118 config HAVE_KERNEL_XZ
121 config HAVE_KERNEL_LZO
124 config HAVE_KERNEL_LZ4
128 prompt "Kernel compression mode"
130 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4
132 The linux kernel is a kind of self-extracting executable.
133 Several compression algorithms are available, which differ
134 in efficiency, compression and decompression speed.
135 Compression speed is only relevant when building a kernel.
136 Decompression speed is relevant at each boot.
138 If you have any problems with bzip2 or lzma compressed
139 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
140 version of this functionality (bzip2 only), for 2.4, was
141 supplied by Christian Ludwig)
143 High compression options are mostly useful for users, who
144 are low on disk space (embedded systems), but for whom ram
147 If in doubt, select 'gzip'
151 depends on HAVE_KERNEL_GZIP
153 The old and tried gzip compression. It provides a good balance
154 between compression ratio and decompression speed.
158 depends on HAVE_KERNEL_BZIP2
160 Its compression ratio and speed is intermediate.
161 Decompression speed is slowest among the choices. The kernel
162 size is about 10% smaller with bzip2, in comparison to gzip.
163 Bzip2 uses a large amount of memory. For modern kernels you
164 will need at least 8MB RAM or more for booting.
168 depends on HAVE_KERNEL_LZMA
170 This compression algorithm's ratio is best. Decompression speed
171 is between gzip and bzip2. Compression is slowest.
172 The kernel size is about 33% smaller with LZMA in comparison to gzip.
176 depends on HAVE_KERNEL_XZ
178 XZ uses the LZMA2 algorithm and instruction set specific
179 BCJ filters which can improve compression ratio of executable
180 code. The size of the kernel is about 30% smaller with XZ in
181 comparison to gzip. On architectures for which there is a BCJ
182 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
183 will create a few percent smaller kernel than plain LZMA.
185 The speed is about the same as with LZMA: The decompression
186 speed of XZ is better than that of bzip2 but worse than gzip
187 and LZO. Compression is slow.
191 depends on HAVE_KERNEL_LZO
193 Its compression ratio is the poorest among the choices. The kernel
194 size is about 10% bigger than gzip; however its speed
195 (both compression and decompression) is the fastest.
199 depends on HAVE_KERNEL_LZ4
201 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
202 A preliminary version of LZ4 de/compression tool is available at
203 <https://code.google.com/p/lz4/>.
205 Its compression ratio is worse than LZO. The size of the kernel
206 is about 8% bigger than LZO. But the decompression speed is
211 config DEFAULT_HOSTNAME
212 string "Default hostname"
215 This option determines the default system hostname before userspace
216 calls sethostname(2). The kernel traditionally uses "(none)" here,
217 but you may wish to use a different default here to make a minimal
218 system more usable with less configuration.
221 bool "Support for paging of anonymous memory (swap)"
222 depends on MMU && BLOCK
225 This option allows you to choose whether you want to have support
226 for so called swap devices or swap files in your kernel that are
227 used to provide more virtual memory than the actual RAM present
228 in your computer. If unsure say Y.
233 Inter Process Communication is a suite of library functions and
234 system calls which let processes (running programs) synchronize and
235 exchange information. It is generally considered to be a good thing,
236 and some programs won't run unless you say Y here. In particular, if
237 you want to run the DOS emulator dosemu under Linux (read the
238 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
239 you'll need to say Y here.
241 You can find documentation about IPC with "info ipc" and also in
242 section 6.4 of the Linux Programmer's Guide, available from
243 <http://www.tldp.org/guides.html>.
245 config SYSVIPC_SYSCTL
252 bool "POSIX Message Queues"
255 POSIX variant of message queues is a part of IPC. In POSIX message
256 queues every message has a priority which decides about succession
257 of receiving it by a process. If you want to compile and run
258 programs written e.g. for Solaris with use of its POSIX message
259 queues (functions mq_*) say Y here.
261 POSIX message queues are visible as a filesystem called 'mqueue'
262 and can be mounted somewhere if you want to do filesystem
263 operations on message queues.
267 config POSIX_MQUEUE_SYSCTL
269 depends on POSIX_MQUEUE
273 config CROSS_MEMORY_ATTACH
274 bool "Enable process_vm_readv/writev syscalls"
278 Enabling this option adds the system calls process_vm_readv and
279 process_vm_writev which allow a process with the correct privileges
280 to directly read from or write to another process' address space.
281 See the man page for more details.
284 bool "open by fhandle syscalls" if EXPERT
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"
298 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
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). System call auditing is included
313 on architectures which support it.
315 config HAVE_ARCH_AUDITSYSCALL
320 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
324 depends on AUDITSYSCALL
329 depends on AUDITSYSCALL
332 source "kernel/irq/Kconfig"
333 source "kernel/time/Kconfig"
335 menu "CPU/Task time and stats accounting"
337 config VIRT_CPU_ACCOUNTING
341 prompt "Cputime accounting"
342 default TICK_CPU_ACCOUNTING if !PPC64
343 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
345 # Kind of a stub config for the pure tick based cputime accounting
346 config TICK_CPU_ACCOUNTING
347 bool "Simple tick based cputime accounting"
348 depends on !S390 && !NO_HZ_FULL
350 This is the basic tick based cputime accounting that maintains
351 statistics about user, system and idle time spent on per jiffies
356 config VIRT_CPU_ACCOUNTING_NATIVE
357 bool "Deterministic task and CPU time accounting"
358 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
359 select VIRT_CPU_ACCOUNTING
361 Select this option to enable more accurate task and CPU time
362 accounting. This is done by reading a CPU counter on each
363 kernel entry and exit and on transitions within the kernel
364 between system, softirq and hardirq state, so there is a
365 small performance impact. In the case of s390 or IBM POWER > 5,
366 this also enables accounting of stolen time on logically-partitioned
369 config VIRT_CPU_ACCOUNTING_GEN
370 bool "Full dynticks CPU time accounting"
371 depends on HAVE_CONTEXT_TRACKING
372 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
373 select VIRT_CPU_ACCOUNTING
374 select CONTEXT_TRACKING
376 Select this option to enable task and CPU time accounting on full
377 dynticks systems. This accounting is implemented by watching every
378 kernel-user boundaries using the context tracking subsystem.
379 The accounting is thus performed at the expense of some significant
382 For now this is only useful if you are working on the full
383 dynticks subsystem development.
389 config IRQ_TIME_ACCOUNTING
390 bool "Fine granularity task level IRQ time accounting"
391 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
393 Select this option to enable fine granularity task irq time
394 accounting. This is done by reading a timestamp on each
395 transitions between softirq and hardirq state, so there can be a
396 small performance impact.
398 If in doubt, say N here.
400 config BSD_PROCESS_ACCT
401 bool "BSD Process Accounting"
404 If you say Y here, a user level program will be able to instruct the
405 kernel (via a special system call) to write process accounting
406 information to a file: whenever a process exits, information about
407 that process will be appended to the file by the kernel. The
408 information includes things such as creation time, owning user,
409 command name, memory usage, controlling terminal etc. (the complete
410 list is in the struct acct in <file:include/linux/acct.h>). It is
411 up to the user level program to do useful things with this
412 information. This is generally a good idea, so say Y.
414 config BSD_PROCESS_ACCT_V3
415 bool "BSD Process Accounting version 3 file format"
416 depends on BSD_PROCESS_ACCT
419 If you say Y here, the process accounting information is written
420 in a new file format that also logs the process IDs of each
421 process and it's parent. Note that this file format is incompatible
422 with previous v0/v1/v2 file formats, so you will need updated tools
423 for processing it. A preliminary version of these tools is available
424 at <http://www.gnu.org/software/acct/>.
427 bool "Export task/process statistics through netlink"
432 Export selected statistics for tasks/processes through the
433 generic netlink interface. Unlike BSD process accounting, the
434 statistics are available during the lifetime of tasks/processes as
435 responses to commands. Like BSD accounting, they are sent to user
440 config TASK_DELAY_ACCT
441 bool "Enable per-task delay accounting"
445 Collect information on time spent by a task waiting for system
446 resources like cpu, synchronous block I/O completion and swapping
447 in pages. Such statistics can help in setting a task's priorities
448 relative to other tasks for cpu, io, rss limits etc.
453 bool "Enable extended accounting over taskstats"
456 Collect extended task accounting data and send the data
457 to userland for processing over the taskstats interface.
461 config TASK_IO_ACCOUNTING
462 bool "Enable per-task storage I/O accounting"
463 depends on TASK_XACCT
465 Collect information on the number of bytes of storage I/O which this
470 endmenu # "CPU/Task time and stats accounting"
476 default y if !PREEMPT && SMP
478 This option selects the RCU implementation that is
479 designed for very large SMP system with hundreds or
480 thousands of CPUs. It also scales down nicely to
487 This option selects the RCU implementation that is
488 designed for very large SMP systems with hundreds or
489 thousands of CPUs, but for which real-time response
490 is also required. It also scales down nicely to
493 Select this option if you are unsure.
497 default y if !PREEMPT && !SMP
499 This option selects the RCU implementation that is
500 designed for UP systems from which real-time response
501 is not required. This option greatly reduces the
502 memory footprint of RCU.
505 bool "Make expert-level adjustments to RCU configuration"
508 This option needs to be enabled if you wish to make
509 expert-level adjustments to RCU configuration. By default,
510 no such adjustments can be made, which has the often-beneficial
511 side-effect of preventing "make oldconfig" from asking you all
512 sorts of detailed questions about how you would like numerous
513 obscure RCU options to be set up.
515 Say Y if you need to make expert-level adjustments to RCU.
517 Say N if you are unsure.
522 This option selects the sleepable version of RCU. This version
523 permits arbitrary sleeping or blocking within RCU read-side critical
532 This option enables a task-based RCU implementation that uses
533 only voluntary context switch (not preemption!), idle, and
534 user-mode execution as quiescent states.
536 config RCU_STALL_COMMON
537 def_bool ( TREE_RCU || PREEMPT_RCU || RCU_TRACE )
539 This option enables RCU CPU stall code that is common between
540 the TINY and TREE variants of RCU. The purpose is to allow
541 the tiny variants to disable RCU CPU stall warnings, while
542 making these warnings mandatory for the tree variants.
544 config CONTEXT_TRACKING
547 config CONTEXT_TRACKING_FORCE
548 bool "Force context tracking"
549 depends on CONTEXT_TRACKING
550 default y if !NO_HZ_FULL
552 The major pre-requirement for full dynticks to work is to
553 support the context tracking subsystem. But there are also
554 other dependencies to provide in order to make the full
557 This option stands for testing when an arch implements the
558 context tracking backend but doesn't yet fullfill all the
559 requirements to make the full dynticks feature working.
560 Without the full dynticks, there is no way to test the support
561 for context tracking and the subsystems that rely on it: RCU
562 userspace extended quiescent state and tickless cputime
563 accounting. This option copes with the absence of the full
564 dynticks subsystem by forcing the context tracking on all
567 Say Y only if you're working on the development of an
568 architecture backend for the context tracking.
570 Say N otherwise, this option brings an overhead that you
571 don't want in production.
575 int "Tree-based hierarchical RCU fanout value"
578 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
582 This option controls the fanout of hierarchical implementations
583 of RCU, allowing RCU to work efficiently on machines with
584 large numbers of CPUs. This value must be at least the fourth
585 root of NR_CPUS, which allows NR_CPUS to be insanely large.
586 The default value of RCU_FANOUT should be used for production
587 systems, but if you are stress-testing the RCU implementation
588 itself, small RCU_FANOUT values allow you to test large-system
589 code paths on small(er) systems.
591 Select a specific number if testing RCU itself.
592 Take the default if unsure.
594 config RCU_FANOUT_LEAF
595 int "Tree-based hierarchical RCU leaf-level fanout value"
598 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
601 This option controls the leaf-level fanout of hierarchical
602 implementations of RCU, and allows trading off cache misses
603 against lock contention. Systems that synchronize their
604 scheduling-clock interrupts for energy-efficiency reasons will
605 want the default because the smaller leaf-level fanout keeps
606 lock contention levels acceptably low. Very large systems
607 (hundreds or thousands of CPUs) will instead want to set this
608 value to the maximum value possible in order to reduce the
609 number of cache misses incurred during RCU's grace-period
610 initialization. These systems tend to run CPU-bound, and thus
611 are not helped by synchronized interrupts, and thus tend to
612 skew them, which reduces lock contention enough that large
613 leaf-level fanouts work well.
615 Select a specific number if testing RCU itself.
617 Select the maximum permissible value for large systems.
619 Take the default if unsure.
621 config RCU_FAST_NO_HZ
622 bool "Accelerate last non-dyntick-idle CPU's grace periods"
623 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
626 This option permits CPUs to enter dynticks-idle state even if
627 they have RCU callbacks queued, and prevents RCU from waking
628 these CPUs up more than roughly once every four jiffies (by
629 default, you can adjust this using the rcutree.rcu_idle_gp_delay
630 parameter), thus improving energy efficiency. On the other
631 hand, this option increases the duration of RCU grace periods,
632 for example, slowing down synchronize_rcu().
634 Say Y if energy efficiency is critically important, and you
635 don't care about increased grace-period durations.
637 Say N if you are unsure.
639 config TREE_RCU_TRACE
640 def_bool RCU_TRACE && ( TREE_RCU || PREEMPT_RCU )
643 This option provides tracing for the TREE_RCU and
644 PREEMPT_RCU implementations, permitting Makefile to
645 trivially select kernel/rcutree_trace.c.
648 bool "Enable RCU priority boosting"
649 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
652 This option boosts the priority of preempted RCU readers that
653 block the current preemptible RCU grace period for too long.
654 This option also prevents heavy loads from blocking RCU
655 callback invocation for all flavors of RCU.
657 Say Y here if you are working with real-time apps or heavy loads
658 Say N here if you are unsure.
660 config RCU_KTHREAD_PRIO
661 int "Real-time priority to use for RCU worker threads"
662 range 1 99 if RCU_BOOST
663 range 0 99 if !RCU_BOOST
664 default 1 if RCU_BOOST
665 default 0 if !RCU_BOOST
666 depends on RCU_EXPERT
668 This option specifies the SCHED_FIFO priority value that will be
669 assigned to the rcuc/n and rcub/n threads and is also the value
670 used for RCU_BOOST (if enabled). If you are working with a
671 real-time application that has one or more CPU-bound threads
672 running at a real-time priority level, you should set
673 RCU_KTHREAD_PRIO to a priority higher than the highest-priority
674 real-time CPU-bound application thread. The default RCU_KTHREAD_PRIO
675 value of 1 is appropriate in the common case, which is real-time
676 applications that do not have any CPU-bound threads.
678 Some real-time applications might not have a single real-time
679 thread that saturates a given CPU, but instead might have
680 multiple real-time threads that, taken together, fully utilize
681 that CPU. In this case, you should set RCU_KTHREAD_PRIO to
682 a priority higher than the lowest-priority thread that is
683 conspiring to prevent the CPU from running any non-real-time
684 tasks. For example, if one thread at priority 10 and another
685 thread at priority 5 are between themselves fully consuming
686 the CPU time on a given CPU, then RCU_KTHREAD_PRIO should be
687 set to priority 6 or higher.
689 Specify the real-time priority, or take the default if unsure.
691 config RCU_BOOST_DELAY
692 int "Milliseconds to delay boosting after RCU grace-period start"
697 This option specifies the time to wait after the beginning of
698 a given grace period before priority-boosting preempted RCU
699 readers blocking that grace period. Note that any RCU reader
700 blocking an expedited RCU grace period is boosted immediately.
702 Accept the default if unsure.
705 bool "Offload RCU callback processing from boot-selected CPUs"
706 depends on TREE_RCU || PREEMPT_RCU
707 depends on RCU_EXPERT || NO_HZ_FULL
710 Use this option to reduce OS jitter for aggressive HPC or
711 real-time workloads. It can also be used to offload RCU
712 callback invocation to energy-efficient CPUs in battery-powered
713 asymmetric multiprocessors.
715 This option offloads callback invocation from the set of
716 CPUs specified at boot time by the rcu_nocbs parameter.
717 For each such CPU, a kthread ("rcuox/N") will be created to
718 invoke callbacks, where the "N" is the CPU being offloaded,
719 and where the "x" is "b" for RCU-bh, "p" for RCU-preempt, and
720 "s" for RCU-sched. Nothing prevents this kthread from running
721 on the specified CPUs, but (1) the kthreads may be preempted
722 between each callback, and (2) affinity or cgroups can be used
723 to force the kthreads to run on whatever set of CPUs is desired.
725 Say Y here if you want to help to debug reduced OS jitter.
726 Say N here if you are unsure.
729 prompt "Build-forced no-CBs CPUs"
730 default RCU_NOCB_CPU_NONE
731 depends on RCU_NOCB_CPU
733 This option allows no-CBs CPUs (whose RCU callbacks are invoked
734 from kthreads rather than from softirq context) to be specified
735 at build time. Additional no-CBs CPUs may be specified by
736 the rcu_nocbs= boot parameter.
738 config RCU_NOCB_CPU_NONE
739 bool "No build_forced no-CBs CPUs"
741 This option does not force any of the CPUs to be no-CBs CPUs.
742 Only CPUs designated by the rcu_nocbs= boot parameter will be
743 no-CBs CPUs, whose RCU callbacks will be invoked by per-CPU
744 kthreads whose names begin with "rcuo". All other CPUs will
745 invoke their own RCU callbacks in softirq context.
747 Select this option if you want to choose no-CBs CPUs at
748 boot time, for example, to allow testing of different no-CBs
749 configurations without having to rebuild the kernel each time.
751 config RCU_NOCB_CPU_ZERO
752 bool "CPU 0 is a build_forced no-CBs CPU"
754 This option forces CPU 0 to be a no-CBs CPU, so that its RCU
755 callbacks are invoked by a per-CPU kthread whose name begins
756 with "rcuo". Additional CPUs may be designated as no-CBs
757 CPUs using the rcu_nocbs= boot parameter will be no-CBs CPUs.
758 All other CPUs will invoke their own RCU callbacks in softirq
761 Select this if CPU 0 needs to be a no-CBs CPU for real-time
762 or energy-efficiency reasons, but the real reason it exists
763 is to ensure that randconfig testing covers mixed systems.
765 config RCU_NOCB_CPU_ALL
766 bool "All CPUs are build_forced no-CBs CPUs"
768 This option forces all CPUs to be no-CBs CPUs. The rcu_nocbs=
769 boot parameter will be ignored. All CPUs' RCU callbacks will
770 be executed in the context of per-CPU rcuo kthreads created for
771 this purpose. Assuming that the kthreads whose names start with
772 "rcuo" are bound to "housekeeping" CPUs, this reduces OS jitter
773 on the remaining CPUs, but might decrease memory locality during
774 RCU-callback invocation, thus potentially degrading throughput.
776 Select this if all CPUs need to be no-CBs CPUs for real-time
777 or energy-efficiency reasons.
781 config RCU_EXPEDITE_BOOT
785 This option enables expedited grace periods at boot time,
786 as if rcu_expedite_gp() had been invoked early in boot.
787 The corresponding rcu_unexpedite_gp() is invoked from
788 rcu_end_inkernel_boot(), which is intended to be invoked
789 at the end of the kernel-only boot sequence, just before
792 Accept the default if unsure.
794 endmenu # "RCU Subsystem"
801 tristate "Kernel .config support"
804 This option enables the complete Linux kernel ".config" file
805 contents to be saved in the kernel. It provides documentation
806 of which kernel options are used in a running kernel or in an
807 on-disk kernel. This information can be extracted from the kernel
808 image file with the script scripts/extract-ikconfig and used as
809 input to rebuild the current kernel or to build another kernel.
810 It can also be extracted from a running kernel by reading
811 /proc/config.gz if enabled (below).
814 bool "Enable access to .config through /proc/config.gz"
815 depends on IKCONFIG && PROC_FS
817 This option enables access to the kernel configuration file
818 through /proc/config.gz.
821 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
826 Select the minimal kernel log buffer size as a power of 2.
827 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
828 parameter, see below. Any higher size also might be forced
829 by "log_buf_len" boot parameter.
839 config LOG_CPU_MAX_BUF_SHIFT
840 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
843 default 12 if !BASE_SMALL
844 default 0 if BASE_SMALL
847 This option allows to increase the default ring buffer size
848 according to the number of CPUs. The value defines the contribution
849 of each CPU as a power of 2. The used space is typically only few
850 lines however it might be much more when problems are reported,
853 The increased size means that a new buffer has to be allocated and
854 the original static one is unused. It makes sense only on systems
855 with more CPUs. Therefore this value is used only when the sum of
856 contributions is greater than the half of the default kernel ring
857 buffer as defined by LOG_BUF_SHIFT. The default values are set
858 so that more than 64 CPUs are needed to trigger the allocation.
860 Also this option is ignored when "log_buf_len" kernel parameter is
861 used as it forces an exact (power of two) size of the ring buffer.
863 The number of possible CPUs is used for this computation ignoring
864 hotplugging making the computation optimal for the worst case
865 scenario while allowing a simple algorithm to be used from bootup.
867 Examples shift values and their meaning:
868 17 => 128 KB for each CPU
869 16 => 64 KB for each CPU
870 15 => 32 KB for each CPU
871 14 => 16 KB for each CPU
872 13 => 8 KB for each CPU
873 12 => 4 KB for each CPU
875 config NMI_LOG_BUF_SHIFT
876 int "Temporary per-CPU NMI log buffer size (12 => 4KB, 13 => 8KB)"
879 depends on PRINTK_NMI
881 Select the size of a per-CPU buffer where NMI messages are temporary
882 stored. They are copied to the main log buffer in a safe context
883 to avoid a deadlock. The value defines the size as a power of 2.
885 NMI messages are rare and limited. The largest one is when
886 a backtrace is printed. It usually fits into 4KB. Select
887 8KB if you want to be on the safe side.
890 17 => 128 KB for each CPU
891 16 => 64 KB for each CPU
892 15 => 32 KB for each CPU
893 14 => 16 KB for each CPU
894 13 => 8 KB for each CPU
895 12 => 4 KB for each CPU
898 # Architectures with an unreliable sched_clock() should select this:
900 config HAVE_UNSTABLE_SCHED_CLOCK
903 config GENERIC_SCHED_CLOCK
907 # For architectures that want to enable the support for NUMA-affine scheduler
910 config ARCH_SUPPORTS_NUMA_BALANCING
914 # For architectures that prefer to flush all TLBs after a number of pages
915 # are unmapped instead of sending one IPI per page to flush. The architecture
916 # must provide guarantees on what happens if a clean TLB cache entry is
917 # written after the unmap. Details are in mm/rmap.c near the check for
918 # should_defer_flush. The architecture should also consider if the full flush
919 # and the refill costs are offset by the savings of sending fewer IPIs.
920 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
924 # For architectures that know their GCC __int128 support is sound
926 config ARCH_SUPPORTS_INT128
929 # For architectures that (ab)use NUMA to represent different memory regions
930 # all cpu-local but of different latencies, such as SuperH.
932 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
935 config NUMA_BALANCING
936 bool "Memory placement aware NUMA scheduler"
937 depends on ARCH_SUPPORTS_NUMA_BALANCING
938 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
939 depends on SMP && NUMA && MIGRATION
941 This option adds support for automatic NUMA aware memory/task placement.
942 The mechanism is quite primitive and is based on migrating memory when
943 it has references to the node the task is running on.
945 This system will be inactive on UMA systems.
947 config NUMA_BALANCING_DEFAULT_ENABLED
948 bool "Automatically enable NUMA aware memory/task placement"
950 depends on NUMA_BALANCING
952 If set, automatic NUMA balancing will be enabled if running on a NUMA
956 bool "Control Group support"
959 This option adds support for grouping sets of processes together, for
960 use with process control subsystems such as Cpusets, CFS, memory
961 controls or device isolation.
963 - Documentation/scheduler/sched-design-CFS.txt (CFS)
964 - Documentation/cgroup-v1/ (features for grouping, isolation
965 and resource control)
975 bool "Memory controller"
979 Provides control over the memory footprint of tasks in a cgroup.
982 bool "Swap controller"
983 depends on MEMCG && SWAP
985 Provides control over the swap space consumed by tasks in a cgroup.
987 config MEMCG_SWAP_ENABLED
988 bool "Swap controller enabled by default"
989 depends on MEMCG_SWAP
992 Memory Resource Controller Swap Extension comes with its price in
993 a bigger memory consumption. General purpose distribution kernels
994 which want to enable the feature but keep it disabled by default
995 and let the user enable it by swapaccount=1 boot command line
996 parameter should have this option unselected.
997 For those who want to have the feature enabled by default should
998 select this option (if, for some reason, they need to disable it
999 then swapaccount=0 does the trick).
1002 bool "IO controller"
1006 Generic block IO controller cgroup interface. This is the common
1007 cgroup interface which should be used by various IO controlling
1010 Currently, CFQ IO scheduler uses it to recognize task groups and
1011 control disk bandwidth allocation (proportional time slice allocation)
1012 to such task groups. It is also used by bio throttling logic in
1013 block layer to implement upper limit in IO rates on a device.
1015 This option only enables generic Block IO controller infrastructure.
1016 One needs to also enable actual IO controlling logic/policy. For
1017 enabling proportional weight division of disk bandwidth in CFQ, set
1018 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
1019 CONFIG_BLK_DEV_THROTTLING=y.
1021 See Documentation/cgroup-v1/blkio-controller.txt for more information.
1023 config DEBUG_BLK_CGROUP
1024 bool "IO controller debugging"
1025 depends on BLK_CGROUP
1028 Enable some debugging help. Currently it exports additional stat
1029 files in a cgroup which can be useful for debugging.
1031 config CGROUP_WRITEBACK
1033 depends on MEMCG && BLK_CGROUP
1036 menuconfig CGROUP_SCHED
1037 bool "CPU controller"
1040 This feature lets CPU scheduler recognize task groups and control CPU
1041 bandwidth allocation to such task groups. It uses cgroups to group
1045 config FAIR_GROUP_SCHED
1046 bool "Group scheduling for SCHED_OTHER"
1047 depends on CGROUP_SCHED
1048 default CGROUP_SCHED
1050 config CFS_BANDWIDTH
1051 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
1052 depends on FAIR_GROUP_SCHED
1055 This option allows users to define CPU bandwidth rates (limits) for
1056 tasks running within the fair group scheduler. Groups with no limit
1057 set are considered to be unconstrained and will run with no
1059 See tip/Documentation/scheduler/sched-bwc.txt for more information.
1061 config RT_GROUP_SCHED
1062 bool "Group scheduling for SCHED_RR/FIFO"
1063 depends on CGROUP_SCHED
1066 This feature lets you explicitly allocate real CPU bandwidth
1067 to task groups. If enabled, it will also make it impossible to
1068 schedule realtime tasks for non-root users until you allocate
1069 realtime bandwidth for them.
1070 See Documentation/scheduler/sched-rt-group.txt for more information.
1075 bool "PIDs controller"
1077 Provides enforcement of process number limits in the scope of a
1078 cgroup. Any attempt to fork more processes than is allowed in the
1079 cgroup will fail. PIDs are fundamentally a global resource because it
1080 is fairly trivial to reach PID exhaustion before you reach even a
1081 conservative kmemcg limit. As a result, it is possible to grind a
1082 system to halt without being limited by other cgroup policies. The
1083 PIDs controller is designed to stop this from happening.
1085 It should be noted that organisational operations (such as attaching
1086 to a cgroup hierarchy will *not* be blocked by the PIDs controller),
1087 since the PIDs limit only affects a process's ability to fork, not to
1090 config CGROUP_FREEZER
1091 bool "Freezer controller"
1093 Provides a way to freeze and unfreeze all tasks in a
1096 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1097 controller includes important in-kernel memory consumers per default.
1099 If you're using cgroup2, say N.
1101 config CGROUP_HUGETLB
1102 bool "HugeTLB controller"
1103 depends on HUGETLB_PAGE
1107 Provides a cgroup controller for HugeTLB pages.
1108 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1109 The limit is enforced during page fault. Since HugeTLB doesn't
1110 support page reclaim, enforcing the limit at page fault time implies
1111 that, the application will get SIGBUS signal if it tries to access
1112 HugeTLB pages beyond its limit. This requires the application to know
1113 beforehand how much HugeTLB pages it would require for its use. The
1114 control group is tracked in the third page lru pointer. This means
1115 that we cannot use the controller with huge page less than 3 pages.
1118 bool "Cpuset controller"
1120 This option will let you create and manage CPUSETs which
1121 allow dynamically partitioning a system into sets of CPUs and
1122 Memory Nodes and assigning tasks to run only within those sets.
1123 This is primarily useful on large SMP or NUMA systems.
1127 config PROC_PID_CPUSET
1128 bool "Include legacy /proc/<pid>/cpuset file"
1132 config CGROUP_DEVICE
1133 bool "Device controller"
1135 Provides a cgroup controller implementing whitelists for
1136 devices which a process in the cgroup can mknod or open.
1138 config CGROUP_CPUACCT
1139 bool "Simple CPU accounting controller"
1141 Provides a simple controller for monitoring the
1142 total CPU consumed by the tasks in a cgroup.
1145 bool "Perf controller"
1146 depends on PERF_EVENTS
1148 This option extends the perf per-cpu mode to restrict monitoring
1149 to threads which belong to the cgroup specified and run on the
1155 bool "Example controller"
1158 This option enables a simple controller that exports
1159 debugging information about the cgroups framework.
1165 config CHECKPOINT_RESTORE
1166 bool "Checkpoint/restore support" if EXPERT
1167 select PROC_CHILDREN
1170 Enables additional kernel features in a sake of checkpoint/restore.
1171 In particular it adds auxiliary prctl codes to setup process text,
1172 data and heap segment sizes, and a few additional /proc filesystem
1175 If unsure, say N here.
1177 menuconfig NAMESPACES
1178 bool "Namespaces support" if EXPERT
1179 depends on MULTIUSER
1182 Provides the way to make tasks work with different objects using
1183 the same id. For example same IPC id may refer to different objects
1184 or same user id or pid may refer to different tasks when used in
1185 different namespaces.
1190 bool "UTS namespace"
1193 In this namespace tasks see different info provided with the
1197 bool "IPC namespace"
1198 depends on (SYSVIPC || POSIX_MQUEUE)
1201 In this namespace tasks work with IPC ids which correspond to
1202 different IPC objects in different namespaces.
1205 bool "User namespace"
1208 This allows containers, i.e. vservers, to use user namespaces
1209 to provide different user info for different servers.
1211 When user namespaces are enabled in the kernel it is
1212 recommended that the MEMCG option also be enabled and that
1213 user-space use the memory control groups to limit the amount
1214 of memory a memory unprivileged users can use.
1219 bool "PID Namespaces"
1222 Support process id namespaces. This allows having multiple
1223 processes with the same pid as long as they are in different
1224 pid namespaces. This is a building block of containers.
1227 bool "Network namespace"
1231 Allow user space to create what appear to be multiple instances
1232 of the network stack.
1236 config SCHED_AUTOGROUP
1237 bool "Automatic process group scheduling"
1240 select FAIR_GROUP_SCHED
1242 This option optimizes the scheduler for common desktop workloads by
1243 automatically creating and populating task groups. This separation
1244 of workloads isolates aggressive CPU burners (like build jobs) from
1245 desktop applications. Task group autogeneration is currently based
1248 config SYSFS_DEPRECATED
1249 bool "Enable deprecated sysfs features to support old userspace tools"
1253 This option adds code that switches the layout of the "block" class
1254 devices, to not show up in /sys/class/block/, but only in
1257 This switch is only active when the sysfs.deprecated=1 boot option is
1258 passed or the SYSFS_DEPRECATED_V2 option is set.
1260 This option allows new kernels to run on old distributions and tools,
1261 which might get confused by /sys/class/block/. Since 2007/2008 all
1262 major distributions and tools handle this just fine.
1264 Recent distributions and userspace tools after 2009/2010 depend on
1265 the existence of /sys/class/block/, and will not work with this
1268 Only if you are using a new kernel on an old distribution, you might
1271 config SYSFS_DEPRECATED_V2
1272 bool "Enable deprecated sysfs features by default"
1275 depends on SYSFS_DEPRECATED
1277 Enable deprecated sysfs by default.
1279 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1282 Only if you are using a new kernel on an old distribution, you might
1283 need to say Y here. Even then, odds are you would not need it
1284 enabled, you can always pass the boot option if absolutely necessary.
1287 bool "Kernel->user space relay support (formerly relayfs)"
1289 This option enables support for relay interface support in
1290 certain file systems (such as debugfs).
1291 It is designed to provide an efficient mechanism for tools and
1292 facilities to relay large amounts of data from kernel space to
1297 config BLK_DEV_INITRD
1298 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1299 depends on BROKEN || !FRV
1301 The initial RAM filesystem is a ramfs which is loaded by the
1302 boot loader (loadlin or lilo) and that is mounted as root
1303 before the normal boot procedure. It is typically used to
1304 load modules needed to mount the "real" root file system,
1305 etc. See <file:Documentation/initrd.txt> for details.
1307 If RAM disk support (BLK_DEV_RAM) is also included, this
1308 also enables initial RAM disk (initrd) support and adds
1309 15 Kbytes (more on some other architectures) to the kernel size.
1315 source "usr/Kconfig"
1320 prompt "Compiler optimization level"
1321 default CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
1323 config CC_OPTIMIZE_FOR_PERFORMANCE
1324 bool "Optimize for performance"
1326 This is the default optimization level for the kernel, building
1327 with the "-O2" compiler flag for best performance and most
1328 helpful compile-time warnings.
1330 config CC_OPTIMIZE_FOR_SIZE
1331 bool "Optimize for size"
1333 Enabling this option will pass "-Os" instead of "-O2" to
1334 your compiler resulting in a smaller kernel.
1349 config SYSCTL_EXCEPTION_TRACE
1352 Enable support for /proc/sys/debug/exception-trace.
1354 config SYSCTL_ARCH_UNALIGN_NO_WARN
1357 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1358 Allows arch to define/use @no_unaligned_warning to possibly warn
1359 about unaligned access emulation going on under the hood.
1361 config SYSCTL_ARCH_UNALIGN_ALLOW
1364 Enable support for /proc/sys/kernel/unaligned-trap
1365 Allows arches to define/use @unaligned_enabled to runtime toggle
1366 the unaligned access emulation.
1367 see arch/parisc/kernel/unaligned.c for reference
1369 config HAVE_PCSPKR_PLATFORM
1372 # interpreter that classic socket filters depend on
1377 bool "Configure standard kernel features (expert users)"
1378 # Unhide debug options, to make the on-by-default options visible
1381 This option allows certain base kernel options and settings
1382 to be disabled or tweaked. This is for specialized
1383 environments which can tolerate a "non-standard" kernel.
1384 Only use this if you really know what you are doing.
1387 bool "Enable 16-bit UID system calls" if EXPERT
1388 depends on HAVE_UID16 && MULTIUSER
1391 This enables the legacy 16-bit UID syscall wrappers.
1394 bool "Multiple users, groups and capabilities support" if EXPERT
1397 This option enables support for non-root users, groups and
1400 If you say N here, all processes will run with UID 0, GID 0, and all
1401 possible capabilities. Saying N here also compiles out support for
1402 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1405 If unsure, say Y here.
1407 config SGETMASK_SYSCALL
1408 bool "sgetmask/ssetmask syscalls support" if EXPERT
1409 def_bool PARISC || MN10300 || BLACKFIN || M68K || PPC || MIPS || X86 || SPARC || CRIS || MICROBLAZE || SUPERH
1411 sys_sgetmask and sys_ssetmask are obsolete system calls
1412 no longer supported in libc but still enabled by default in some
1415 If unsure, leave the default option here.
1417 config SYSFS_SYSCALL
1418 bool "Sysfs syscall support" if EXPERT
1421 sys_sysfs is an obsolete system call no longer supported in libc.
1422 Note that disabling this option is more secure but might break
1423 compatibility with some systems.
1425 If unsure say Y here.
1427 config SYSCTL_SYSCALL
1428 bool "Sysctl syscall support" if EXPERT
1429 depends on PROC_SYSCTL
1433 sys_sysctl uses binary paths that have been found challenging
1434 to properly maintain and use. The interface in /proc/sys
1435 using paths with ascii names is now the primary path to this
1438 Almost nothing using the binary sysctl interface so if you are
1439 trying to save some space it is probably safe to disable this,
1440 making your kernel marginally smaller.
1442 If unsure say N here.
1445 bool "Load all symbols for debugging/ksymoops" if EXPERT
1448 Say Y here to let the kernel print out symbolic crash information and
1449 symbolic stack backtraces. This increases the size of the kernel
1450 somewhat, as all symbols have to be loaded into the kernel image.
1453 bool "Include all symbols in kallsyms"
1454 depends on DEBUG_KERNEL && KALLSYMS
1456 Normally kallsyms only contains the symbols of functions for nicer
1457 OOPS messages and backtraces (i.e., symbols from the text and inittext
1458 sections). This is sufficient for most cases. And only in very rare
1459 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1460 names of variables from the data sections, etc).
1462 This option makes sure that all symbols are loaded into the kernel
1463 image (i.e., symbols from all sections) in cost of increased kernel
1464 size (depending on the kernel configuration, it may be 300KiB or
1465 something like this).
1467 Say N unless you really need all symbols.
1469 config KALLSYMS_ABSOLUTE_PERCPU
1472 default X86_64 && SMP
1474 config KALLSYMS_BASE_RELATIVE
1477 default !IA64 && !(TILE && 64BIT)
1479 Instead of emitting them as absolute values in the native word size,
1480 emit the symbol references in the kallsyms table as 32-bit entries,
1481 each containing a relative value in the range [base, base + U32_MAX]
1482 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1483 an absolute value in the range [0, S32_MAX] or a relative value in the
1484 range [base, base + S32_MAX], where base is the lowest relative symbol
1485 address encountered in the image.
1487 On 64-bit builds, this reduces the size of the address table by 50%,
1488 but more importantly, it results in entries whose values are build
1489 time constants, and no relocation pass is required at runtime to fix
1490 up the entries based on the runtime load address of the kernel.
1494 bool "Enable support for printk" if EXPERT
1497 This option enables normal printk support. Removing it
1498 eliminates most of the message strings from the kernel image
1499 and makes the kernel more or less silent. As this makes it
1500 very difficult to diagnose system problems, saying N here is
1501 strongly discouraged.
1509 bool "BUG() support" if EXPERT
1512 Disabling this option eliminates support for BUG and WARN, reducing
1513 the size of your kernel image and potentially quietly ignoring
1514 numerous fatal conditions. You should only consider disabling this
1515 option for embedded systems with no facilities for reporting errors.
1521 bool "Enable ELF core dumps" if EXPERT
1523 Enable support for generating core dumps. Disabling saves about 4k.
1526 config PCSPKR_PLATFORM
1527 bool "Enable PC-Speaker support" if EXPERT
1528 depends on HAVE_PCSPKR_PLATFORM
1532 This option allows to disable the internal PC-Speaker
1533 support, saving some memory.
1537 bool "Enable full-sized data structures for core" if EXPERT
1539 Disabling this option reduces the size of miscellaneous core
1540 kernel data structures. This saves memory on small machines,
1541 but may reduce performance.
1544 bool "Enable futex support" if EXPERT
1548 Disabling this option will cause the kernel to be built without
1549 support for "fast userspace mutexes". The resulting kernel may not
1550 run glibc-based applications correctly.
1552 config HAVE_FUTEX_CMPXCHG
1556 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1557 is implemented and always working. This removes a couple of runtime
1561 bool "Enable eventpoll support" if EXPERT
1565 Disabling this option will cause the kernel to be built without
1566 support for epoll family of system calls.
1569 bool "Enable signalfd() system call" if EXPERT
1573 Enable the signalfd() system call that allows to receive signals
1574 on a file descriptor.
1579 bool "Enable timerfd() system call" if EXPERT
1583 Enable the timerfd() system call that allows to receive timer
1584 events on a file descriptor.
1589 bool "Enable eventfd() system call" if EXPERT
1593 Enable the eventfd() system call that allows to receive both
1594 kernel notification (ie. KAIO) or userspace notifications.
1598 # syscall, maps, verifier
1600 bool "Enable bpf() system call"
1605 Enable the bpf() system call that allows to manipulate eBPF
1606 programs and maps via file descriptors.
1609 bool "Use full shmem filesystem" if EXPERT
1613 The shmem is an internal filesystem used to manage shared memory.
1614 It is backed by swap and manages resource limits. It is also exported
1615 to userspace as tmpfs if TMPFS is enabled. Disabling this
1616 option replaces shmem and tmpfs with the much simpler ramfs code,
1617 which may be appropriate on small systems without swap.
1620 bool "Enable AIO support" if EXPERT
1623 This option enables POSIX asynchronous I/O which may by used
1624 by some high performance threaded applications. Disabling
1625 this option saves about 7k.
1627 config ADVISE_SYSCALLS
1628 bool "Enable madvise/fadvise syscalls" if EXPERT
1631 This option enables the madvise and fadvise syscalls, used by
1632 applications to advise the kernel about their future memory or file
1633 usage, improving performance. If building an embedded system where no
1634 applications use these syscalls, you can disable this option to save
1638 bool "Enable userfaultfd() system call"
1642 Enable the userfaultfd() system call that allows to intercept and
1643 handle page faults in userland.
1647 bool "Enable PCI quirk workarounds" if EXPERT
1650 This enables workarounds for various PCI chipset
1651 bugs/quirks. Disable this only if your target machine is
1652 unaffected by PCI quirks.
1655 bool "Enable membarrier() system call" if EXPERT
1658 Enable the membarrier() system call that allows issuing memory
1659 barriers across all running threads, which can be used to distribute
1660 the cost of user-space memory barriers asymmetrically by transforming
1661 pairs of memory barriers into pairs consisting of membarrier() and a
1667 bool "Embedded system"
1668 option allnoconfig_y
1671 This option should be enabled if compiling the kernel for
1672 an embedded system so certain expert options are available
1675 config HAVE_PERF_EVENTS
1678 See tools/perf/design.txt for details.
1680 config PERF_USE_VMALLOC
1683 See tools/perf/design.txt for details
1685 menu "Kernel Performance Events And Counters"
1688 bool "Kernel performance events and counters"
1689 default y if PROFILING
1690 depends on HAVE_PERF_EVENTS
1695 Enable kernel support for various performance events provided
1696 by software and hardware.
1698 Software events are supported either built-in or via the
1699 use of generic tracepoints.
1701 Most modern CPUs support performance events via performance
1702 counter registers. These registers count the number of certain
1703 types of hw events: such as instructions executed, cachemisses
1704 suffered, or branches mis-predicted - without slowing down the
1705 kernel or applications. These registers can also trigger interrupts
1706 when a threshold number of events have passed - and can thus be
1707 used to profile the code that runs on that CPU.
1709 The Linux Performance Event subsystem provides an abstraction of
1710 these software and hardware event capabilities, available via a
1711 system call and used by the "perf" utility in tools/perf/. It
1712 provides per task and per CPU counters, and it provides event
1713 capabilities on top of those.
1717 config DEBUG_PERF_USE_VMALLOC
1719 bool "Debug: use vmalloc to back perf mmap() buffers"
1720 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1721 select PERF_USE_VMALLOC
1723 Use vmalloc memory to back perf mmap() buffers.
1725 Mostly useful for debugging the vmalloc code on platforms
1726 that don't require it.
1732 config VM_EVENT_COUNTERS
1734 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1736 VM event counters are needed for event counts to be shown.
1737 This option allows the disabling of the VM event counters
1738 on EXPERT systems. /proc/vmstat will only show page counts
1739 if VM event counters are disabled.
1743 bool "Enable SLUB debugging support" if EXPERT
1744 depends on SLUB && SYSFS
1746 SLUB has extensive debug support features. Disabling these can
1747 result in significant savings in code size. This also disables
1748 SLUB sysfs support. /sys/slab will not exist and there will be
1749 no support for cache validation etc.
1752 bool "Disable heap randomization"
1755 Randomizing heap placement makes heap exploits harder, but it
1756 also breaks ancient binaries (including anything libc5 based).
1757 This option changes the bootup default to heap randomization
1758 disabled, and can be overridden at runtime by setting
1759 /proc/sys/kernel/randomize_va_space to 2.
1761 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1764 prompt "Choose SLAB allocator"
1767 This option allows to select a slab allocator.
1771 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1773 The regular slab allocator that is established and known to work
1774 well in all environments. It organizes cache hot objects in
1775 per cpu and per node queues.
1778 bool "SLUB (Unqueued Allocator)"
1779 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1781 SLUB is a slab allocator that minimizes cache line usage
1782 instead of managing queues of cached objects (SLAB approach).
1783 Per cpu caching is realized using slabs of objects instead
1784 of queues of objects. SLUB can use memory efficiently
1785 and has enhanced diagnostics. SLUB is the default choice for
1790 bool "SLOB (Simple Allocator)"
1792 SLOB replaces the stock allocator with a drastically simpler
1793 allocator. SLOB is generally more space efficient but
1794 does not perform as well on large systems.
1798 config SLAB_FREELIST_RANDOM
1800 depends on SLAB || SLUB
1801 bool "SLAB freelist randomization"
1803 Randomizes the freelist order used on creating new pages. This
1804 security feature reduces the predictability of the kernel slab
1805 allocator against heap overflows.
1807 config SLUB_CPU_PARTIAL
1809 depends on SLUB && SMP
1810 bool "SLUB per cpu partial cache"
1812 Per cpu partial caches accellerate objects allocation and freeing
1813 that is local to a processor at the price of more indeterminism
1814 in the latency of the free. On overflow these caches will be cleared
1815 which requires the taking of locks that may cause latency spikes.
1816 Typically one would choose no for a realtime system.
1818 config MMAP_ALLOW_UNINITIALIZED
1819 bool "Allow mmapped anonymous memory to be uninitialized"
1820 depends on EXPERT && !MMU
1823 Normally, and according to the Linux spec, anonymous memory obtained
1824 from mmap() has it's contents cleared before it is passed to
1825 userspace. Enabling this config option allows you to request that
1826 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1827 providing a huge performance boost. If this option is not enabled,
1828 then the flag will be ignored.
1830 This is taken advantage of by uClibc's malloc(), and also by
1831 ELF-FDPIC binfmt's brk and stack allocator.
1833 Because of the obvious security issues, this option should only be
1834 enabled on embedded devices where you control what is run in
1835 userspace. Since that isn't generally a problem on no-MMU systems,
1836 it is normally safe to say Y here.
1838 See Documentation/nommu-mmap.txt for more information.
1840 config SYSTEM_DATA_VERIFICATION
1842 select SYSTEM_TRUSTED_KEYRING
1846 select ASYMMETRIC_KEY_TYPE
1847 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1850 select X509_CERTIFICATE_PARSER
1851 select PKCS7_MESSAGE_PARSER
1853 Provide PKCS#7 message verification using the contents of the system
1854 trusted keyring to provide public keys. This then can be used for
1855 module verification, kexec image verification and firmware blob
1859 bool "Profiling support"
1861 Say Y here to enable the extended profiling support mechanisms used
1862 by profilers such as OProfile.
1865 # Place an empty function call at each tracepoint site. Can be
1866 # dynamically changed for a probe function.
1871 source "arch/Kconfig"
1873 endmenu # General setup
1875 config HAVE_GENERIC_DMA_COHERENT
1882 depends on SLAB || SLUB_DEBUG
1890 default 0 if BASE_FULL
1891 default 1 if !BASE_FULL
1894 bool "Enable loadable module support"
1897 Kernel modules are small pieces of compiled code which can
1898 be inserted in the running kernel, rather than being
1899 permanently built into the kernel. You use the "modprobe"
1900 tool to add (and sometimes remove) them. If you say Y here,
1901 many parts of the kernel can be built as modules (by
1902 answering M instead of Y where indicated): this is most
1903 useful for infrequently used options which are not required
1904 for booting. For more information, see the man pages for
1905 modprobe, lsmod, modinfo, insmod and rmmod.
1907 If you say Y here, you will need to run "make
1908 modules_install" to put the modules under /lib/modules/
1909 where modprobe can find them (you may need to be root to do
1916 config MODULE_FORCE_LOAD
1917 bool "Forced module loading"
1920 Allow loading of modules without version information (ie. modprobe
1921 --force). Forced module loading sets the 'F' (forced) taint flag and
1922 is usually a really bad idea.
1924 config MODULE_UNLOAD
1925 bool "Module unloading"
1927 Without this option you will not be able to unload any
1928 modules (note that some modules may not be unloadable
1929 anyway), which makes your kernel smaller, faster
1930 and simpler. If unsure, say Y.
1932 config MODULE_FORCE_UNLOAD
1933 bool "Forced module unloading"
1934 depends on MODULE_UNLOAD
1936 This option allows you to force a module to unload, even if the
1937 kernel believes it is unsafe: the kernel will remove the module
1938 without waiting for anyone to stop using it (using the -f option to
1939 rmmod). This is mainly for kernel developers and desperate users.
1943 bool "Module versioning support"
1945 Usually, you have to use modules compiled with your kernel.
1946 Saying Y here makes it sometimes possible to use modules
1947 compiled for different kernels, by adding enough information
1948 to the modules to (hopefully) spot any changes which would
1949 make them incompatible with the kernel you are running. If
1952 config MODULE_SRCVERSION_ALL
1953 bool "Source checksum for all modules"
1955 Modules which contain a MODULE_VERSION get an extra "srcversion"
1956 field inserted into their modinfo section, which contains a
1957 sum of the source files which made it. This helps maintainers
1958 see exactly which source was used to build a module (since
1959 others sometimes change the module source without updating
1960 the version). With this option, such a "srcversion" field
1961 will be created for all modules. If unsure, say N.
1964 bool "Module signature verification"
1966 select SYSTEM_DATA_VERIFICATION
1968 Check modules for valid signatures upon load: the signature
1969 is simply appended to the module. For more information see
1970 Documentation/module-signing.txt.
1972 Note that this option adds the OpenSSL development packages as a
1973 kernel build dependency so that the signing tool can use its crypto
1976 !!!WARNING!!! If you enable this option, you MUST make sure that the
1977 module DOES NOT get stripped after being signed. This includes the
1978 debuginfo strip done by some packagers (such as rpmbuild) and
1979 inclusion into an initramfs that wants the module size reduced.
1981 config MODULE_SIG_FORCE
1982 bool "Require modules to be validly signed"
1983 depends on MODULE_SIG
1985 Reject unsigned modules or signed modules for which we don't have a
1986 key. Without this, such modules will simply taint the kernel.
1988 config MODULE_SIG_ALL
1989 bool "Automatically sign all modules"
1991 depends on MODULE_SIG
1993 Sign all modules during make modules_install. Without this option,
1994 modules must be signed manually, using the scripts/sign-file tool.
1996 comment "Do not forget to sign required modules with scripts/sign-file"
1997 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2000 prompt "Which hash algorithm should modules be signed with?"
2001 depends on MODULE_SIG
2003 This determines which sort of hashing algorithm will be used during
2004 signature generation. This algorithm _must_ be built into the kernel
2005 directly so that signature verification can take place. It is not
2006 possible to load a signed module containing the algorithm to check
2007 the signature on that module.
2009 config MODULE_SIG_SHA1
2010 bool "Sign modules with SHA-1"
2013 config MODULE_SIG_SHA224
2014 bool "Sign modules with SHA-224"
2015 select CRYPTO_SHA256
2017 config MODULE_SIG_SHA256
2018 bool "Sign modules with SHA-256"
2019 select CRYPTO_SHA256
2021 config MODULE_SIG_SHA384
2022 bool "Sign modules with SHA-384"
2023 select CRYPTO_SHA512
2025 config MODULE_SIG_SHA512
2026 bool "Sign modules with SHA-512"
2027 select CRYPTO_SHA512
2031 config MODULE_SIG_HASH
2033 depends on MODULE_SIG
2034 default "sha1" if MODULE_SIG_SHA1
2035 default "sha224" if MODULE_SIG_SHA224
2036 default "sha256" if MODULE_SIG_SHA256
2037 default "sha384" if MODULE_SIG_SHA384
2038 default "sha512" if MODULE_SIG_SHA512
2040 config MODULE_COMPRESS
2041 bool "Compress modules on installation"
2045 Compresses kernel modules when 'make modules_install' is run; gzip or
2046 xz depending on "Compression algorithm" below.
2048 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2050 Out-of-tree kernel modules installed using Kbuild will also be
2051 compressed upon installation.
2053 Note: for modules inside an initrd or initramfs, it's more efficient
2054 to compress the whole initrd or initramfs instead.
2056 Note: This is fully compatible with signed modules.
2061 prompt "Compression algorithm"
2062 depends on MODULE_COMPRESS
2063 default MODULE_COMPRESS_GZIP
2065 This determines which sort of compression will be used during
2066 'make modules_install'.
2068 GZIP (default) and XZ are supported.
2070 config MODULE_COMPRESS_GZIP
2073 config MODULE_COMPRESS_XZ
2078 config TRIM_UNUSED_KSYMS
2079 bool "Trim unused exported kernel symbols"
2080 depends on MODULES && !UNUSED_SYMBOLS
2082 The kernel and some modules make many symbols available for
2083 other modules to use via EXPORT_SYMBOL() and variants. Depending
2084 on the set of modules being selected in your kernel configuration,
2085 many of those exported symbols might never be used.
2087 This option allows for unused exported symbols to be dropped from
2088 the build. In turn, this provides the compiler more opportunities
2089 (especially when using LTO) for optimizing the code and reducing
2090 binary size. This might have some security advantages as well.
2092 If unsure, or if you need to build out-of-tree modules, say N.
2096 config MODULES_TREE_LOOKUP
2098 depends on PERF_EVENTS || TRACING
2100 config INIT_ALL_POSSIBLE
2103 Back when each arch used to define their own cpu_online_mask and
2104 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2105 with all 1s, and others with all 0s. When they were centralised,
2106 it was better to provide this option than to break all the archs
2107 and have several arch maintainers pursuing me down dark alleys.
2109 source "block/Kconfig"
2111 config PREEMPT_NOTIFIERS
2118 # Can be selected by architectures with broken toolchains
2119 # that get confused by correct const<->read_only section
2121 config BROKEN_RODATA
2127 Build a simple ASN.1 grammar compiler that produces a bytecode output
2128 that can be interpreted by the ASN.1 stream decoder and used to
2129 inform it as to what tags are to be expected in a stream and what
2130 functions to call on what tags.
2132 source "kernel/Kconfig.locks"