7 option env="KERNELVERSION"
13 default "/lib/modules/$UNAME_RELEASE/.config"
14 default "/etc/kernel-config"
15 default "/boot/config-$UNAME_RELEASE"
16 default "$ARCH_DEFCONFIG"
17 default "arch/$ARCH/defconfig"
26 config BUILDTIME_EXTABLE_SORT
29 config THREAD_INFO_IN_TASK
32 Select this to move thread_info off the stack into task_struct. To
33 make this work, an arch will need to remove all thread_info fields
34 except flags and fix any runtime bugs.
36 One subtle change that will be needed is to use try_get_task_stack()
37 and put_task_stack() in save_thread_stack_tsk() and get_wchan().
46 depends on BROKEN || !SMP
49 config INIT_ENV_ARG_LIMIT
54 Maximum of each of the number of arguments and environment
55 variables passed to init from the kernel command line.
59 string "Cross-compiler tool prefix"
61 Same as running 'make CROSS_COMPILE=prefix-' but stored for
62 default make runs in this kernel build directory. You don't
63 need to set this unless you want the configured kernel build
64 directory to select the cross-compiler automatically.
67 bool "Compile also drivers which will not load"
71 Some drivers can be compiled on a different platform than they are
72 intended to be run on. Despite they cannot be loaded there (or even
73 when they load they cannot be used due to missing HW support),
74 developers still, opposing to distributors, might want to build such
75 drivers to compile-test them.
77 If you are a developer and want to build everything available, say Y
78 here. If you are a user/distributor, say N here to exclude useless
79 drivers to be distributed.
82 string "Local version - append to kernel release"
84 Append an extra string to the end of your kernel version.
85 This will show up when you type uname, for example.
86 The string you set here will be appended after the contents of
87 any files with a filename matching localversion* in your
88 object and source tree, in that order. Your total string can
89 be a maximum of 64 characters.
91 config LOCALVERSION_AUTO
92 bool "Automatically append version information to the version string"
94 depends on !COMPILE_TEST
96 This will try to automatically determine if the current tree is a
97 release tree by looking for git tags that belong to the current
100 A string of the format -gxxxxxxxx will be added to the localversion
101 if a git-based tree is found. The string generated by this will be
102 appended after any matching localversion* files, and after the value
103 set in CONFIG_LOCALVERSION.
105 (The actual string used here is the first eight characters produced
106 by running the command:
108 $ git rev-parse --verify HEAD
110 which is done within the script "scripts/setlocalversion".)
112 config HAVE_KERNEL_GZIP
115 config HAVE_KERNEL_BZIP2
118 config HAVE_KERNEL_LZMA
121 config HAVE_KERNEL_XZ
124 config HAVE_KERNEL_LZO
127 config HAVE_KERNEL_LZ4
131 prompt "Kernel compression mode"
133 depends on HAVE_KERNEL_GZIP || HAVE_KERNEL_BZIP2 || HAVE_KERNEL_LZMA || HAVE_KERNEL_XZ || HAVE_KERNEL_LZO || HAVE_KERNEL_LZ4
135 The linux kernel is a kind of self-extracting executable.
136 Several compression algorithms are available, which differ
137 in efficiency, compression and decompression speed.
138 Compression speed is only relevant when building a kernel.
139 Decompression speed is relevant at each boot.
141 If you have any problems with bzip2 or lzma compressed
142 kernels, mail me (Alain Knaff) <alain@knaff.lu>. (An older
143 version of this functionality (bzip2 only), for 2.4, was
144 supplied by Christian Ludwig)
146 High compression options are mostly useful for users, who
147 are low on disk space (embedded systems), but for whom ram
150 If in doubt, select 'gzip'
154 depends on HAVE_KERNEL_GZIP
156 The old and tried gzip compression. It provides a good balance
157 between compression ratio and decompression speed.
161 depends on HAVE_KERNEL_BZIP2
163 Its compression ratio and speed is intermediate.
164 Decompression speed is slowest among the choices. The kernel
165 size is about 10% smaller with bzip2, in comparison to gzip.
166 Bzip2 uses a large amount of memory. For modern kernels you
167 will need at least 8MB RAM or more for booting.
171 depends on HAVE_KERNEL_LZMA
173 This compression algorithm's ratio is best. Decompression speed
174 is between gzip and bzip2. Compression is slowest.
175 The kernel size is about 33% smaller with LZMA in comparison to gzip.
179 depends on HAVE_KERNEL_XZ
181 XZ uses the LZMA2 algorithm and instruction set specific
182 BCJ filters which can improve compression ratio of executable
183 code. The size of the kernel is about 30% smaller with XZ in
184 comparison to gzip. On architectures for which there is a BCJ
185 filter (i386, x86_64, ARM, IA-64, PowerPC, and SPARC), XZ
186 will create a few percent smaller kernel than plain LZMA.
188 The speed is about the same as with LZMA: The decompression
189 speed of XZ is better than that of bzip2 but worse than gzip
190 and LZO. Compression is slow.
194 depends on HAVE_KERNEL_LZO
196 Its compression ratio is the poorest among the choices. The kernel
197 size is about 10% bigger than gzip; however its speed
198 (both compression and decompression) is the fastest.
202 depends on HAVE_KERNEL_LZ4
204 LZ4 is an LZ77-type compressor with a fixed, byte-oriented encoding.
205 A preliminary version of LZ4 de/compression tool is available at
206 <https://code.google.com/p/lz4/>.
208 Its compression ratio is worse than LZO. The size of the kernel
209 is about 8% bigger than LZO. But the decompression speed is
214 config DEFAULT_HOSTNAME
215 string "Default hostname"
218 This option determines the default system hostname before userspace
219 calls sethostname(2). The kernel traditionally uses "(none)" here,
220 but you may wish to use a different default here to make a minimal
221 system more usable with less configuration.
224 bool "Support for paging of anonymous memory (swap)"
225 depends on MMU && BLOCK
228 This option allows you to choose whether you want to have support
229 for so called swap devices or swap files in your kernel that are
230 used to provide more virtual memory than the actual RAM present
231 in your computer. If unsure say Y.
236 Inter Process Communication is a suite of library functions and
237 system calls which let processes (running programs) synchronize and
238 exchange information. It is generally considered to be a good thing,
239 and some programs won't run unless you say Y here. In particular, if
240 you want to run the DOS emulator dosemu under Linux (read the
241 DOSEMU-HOWTO, available from <http://www.tldp.org/docs.html#howto>),
242 you'll need to say Y here.
244 You can find documentation about IPC with "info ipc" and also in
245 section 6.4 of the Linux Programmer's Guide, available from
246 <http://www.tldp.org/guides.html>.
248 config SYSVIPC_SYSCTL
255 bool "POSIX Message Queues"
258 POSIX variant of message queues is a part of IPC. In POSIX message
259 queues every message has a priority which decides about succession
260 of receiving it by a process. If you want to compile and run
261 programs written e.g. for Solaris with use of its POSIX message
262 queues (functions mq_*) say Y here.
264 POSIX message queues are visible as a filesystem called 'mqueue'
265 and can be mounted somewhere if you want to do filesystem
266 operations on message queues.
270 config POSIX_MQUEUE_SYSCTL
272 depends on POSIX_MQUEUE
276 config CROSS_MEMORY_ATTACH
277 bool "Enable process_vm_readv/writev syscalls"
281 Enabling this option adds the system calls process_vm_readv and
282 process_vm_writev which allow a process with the correct privileges
283 to directly read from or write to another process' address space.
284 See the man page for more details.
287 bool "open by fhandle syscalls" if EXPERT
291 If you say Y here, a user level program will be able to map
292 file names to handle and then later use the handle for
293 different file system operations. This is useful in implementing
294 userspace file servers, which now track files using handles instead
295 of names. The handle would remain the same even if file names
296 get renamed. Enables open_by_handle_at(2) and name_to_handle_at(2)
300 bool "uselib syscall"
301 def_bool ALPHA || M68K || SPARC || X86_32 || IA32_EMULATION
303 This option enables the uselib syscall, a system call used in the
304 dynamic linker from libc5 and earlier. glibc does not use this
305 system call. If you intend to run programs built on libc5 or
306 earlier, you may need to enable this syscall. Current systems
307 running glibc can safely disable this.
310 bool "Auditing support"
313 Enable auditing infrastructure that can be used with another
314 kernel subsystem, such as SELinux (which requires this for
315 logging of avc messages output). System call auditing is included
316 on architectures which support it.
318 config HAVE_ARCH_AUDITSYSCALL
323 depends on AUDIT && HAVE_ARCH_AUDITSYSCALL
327 depends on AUDITSYSCALL
332 depends on AUDITSYSCALL
335 source "kernel/irq/Kconfig"
336 source "kernel/time/Kconfig"
338 menu "CPU/Task time and stats accounting"
340 config VIRT_CPU_ACCOUNTING
344 prompt "Cputime accounting"
345 default TICK_CPU_ACCOUNTING if !PPC64
346 default VIRT_CPU_ACCOUNTING_NATIVE if PPC64
348 # Kind of a stub config for the pure tick based cputime accounting
349 config TICK_CPU_ACCOUNTING
350 bool "Simple tick based cputime accounting"
351 depends on !S390 && !NO_HZ_FULL
353 This is the basic tick based cputime accounting that maintains
354 statistics about user, system and idle time spent on per jiffies
359 config VIRT_CPU_ACCOUNTING_NATIVE
360 bool "Deterministic task and CPU time accounting"
361 depends on HAVE_VIRT_CPU_ACCOUNTING && !NO_HZ_FULL
362 select VIRT_CPU_ACCOUNTING
364 Select this option to enable more accurate task and CPU time
365 accounting. This is done by reading a CPU counter on each
366 kernel entry and exit and on transitions within the kernel
367 between system, softirq and hardirq state, so there is a
368 small performance impact. In the case of s390 or IBM POWER > 5,
369 this also enables accounting of stolen time on logically-partitioned
372 config VIRT_CPU_ACCOUNTING_GEN
373 bool "Full dynticks CPU time accounting"
374 depends on HAVE_CONTEXT_TRACKING
375 depends on HAVE_VIRT_CPU_ACCOUNTING_GEN
376 select VIRT_CPU_ACCOUNTING
377 select CONTEXT_TRACKING
379 Select this option to enable task and CPU time accounting on full
380 dynticks systems. This accounting is implemented by watching every
381 kernel-user boundaries using the context tracking subsystem.
382 The accounting is thus performed at the expense of some significant
385 For now this is only useful if you are working on the full
386 dynticks subsystem development.
392 config IRQ_TIME_ACCOUNTING
393 bool "Fine granularity task level IRQ time accounting"
394 depends on HAVE_IRQ_TIME_ACCOUNTING && !VIRT_CPU_ACCOUNTING_NATIVE
396 Select this option to enable fine granularity task irq time
397 accounting. This is done by reading a timestamp on each
398 transitions between softirq and hardirq state, so there can be a
399 small performance impact.
401 If in doubt, say N here.
403 config BSD_PROCESS_ACCT
404 bool "BSD Process Accounting"
407 If you say Y here, a user level program will be able to instruct the
408 kernel (via a special system call) to write process accounting
409 information to a file: whenever a process exits, information about
410 that process will be appended to the file by the kernel. The
411 information includes things such as creation time, owning user,
412 command name, memory usage, controlling terminal etc. (the complete
413 list is in the struct acct in <file:include/linux/acct.h>). It is
414 up to the user level program to do useful things with this
415 information. This is generally a good idea, so say Y.
417 config BSD_PROCESS_ACCT_V3
418 bool "BSD Process Accounting version 3 file format"
419 depends on BSD_PROCESS_ACCT
422 If you say Y here, the process accounting information is written
423 in a new file format that also logs the process IDs of each
424 process and it's parent. Note that this file format is incompatible
425 with previous v0/v1/v2 file formats, so you will need updated tools
426 for processing it. A preliminary version of these tools is available
427 at <http://www.gnu.org/software/acct/>.
430 bool "Export task/process statistics through netlink"
435 Export selected statistics for tasks/processes through the
436 generic netlink interface. Unlike BSD process accounting, the
437 statistics are available during the lifetime of tasks/processes as
438 responses to commands. Like BSD accounting, they are sent to user
443 config TASK_DELAY_ACCT
444 bool "Enable per-task delay accounting"
448 Collect information on time spent by a task waiting for system
449 resources like cpu, synchronous block I/O completion and swapping
450 in pages. Such statistics can help in setting a task's priorities
451 relative to other tasks for cpu, io, rss limits etc.
456 bool "Enable extended accounting over taskstats"
459 Collect extended task accounting data and send the data
460 to userland for processing over the taskstats interface.
464 config TASK_IO_ACCOUNTING
465 bool "Enable per-task storage I/O accounting"
466 depends on TASK_XACCT
468 Collect information on the number of bytes of storage I/O which this
473 endmenu # "CPU/Task time and stats accounting"
479 default y if !PREEMPT && SMP
481 This option selects the RCU implementation that is
482 designed for very large SMP system with hundreds or
483 thousands of CPUs. It also scales down nicely to
490 This option selects the RCU implementation that is
491 designed for very large SMP systems with hundreds or
492 thousands of CPUs, but for which real-time response
493 is also required. It also scales down nicely to
496 Select this option if you are unsure.
500 default y if !PREEMPT && !SMP
502 This option selects the RCU implementation that is
503 designed for UP systems from which real-time response
504 is not required. This option greatly reduces the
505 memory footprint of RCU.
508 bool "Make expert-level adjustments to RCU configuration"
511 This option needs to be enabled if you wish to make
512 expert-level adjustments to RCU configuration. By default,
513 no such adjustments can be made, which has the often-beneficial
514 side-effect of preventing "make oldconfig" from asking you all
515 sorts of detailed questions about how you would like numerous
516 obscure RCU options to be set up.
518 Say Y if you need to make expert-level adjustments to RCU.
520 Say N if you are unsure.
525 This option selects the sleepable version of RCU. This version
526 permits arbitrary sleeping or blocking within RCU read-side critical
534 This option enables a task-based RCU implementation that uses
535 only voluntary context switch (not preemption!), idle, and
536 user-mode execution as quiescent states.
538 config RCU_STALL_COMMON
539 def_bool ( TREE_RCU || PREEMPT_RCU || RCU_TRACE )
541 This option enables RCU CPU stall code that is common between
542 the TINY and TREE variants of RCU. The purpose is to allow
543 the tiny variants to disable RCU CPU stall warnings, while
544 making these warnings mandatory for the tree variants.
546 config CONTEXT_TRACKING
549 config CONTEXT_TRACKING_FORCE
550 bool "Force context tracking"
551 depends on CONTEXT_TRACKING
552 default y if !NO_HZ_FULL
554 The major pre-requirement for full dynticks to work is to
555 support the context tracking subsystem. But there are also
556 other dependencies to provide in order to make the full
559 This option stands for testing when an arch implements the
560 context tracking backend but doesn't yet fullfill all the
561 requirements to make the full dynticks feature working.
562 Without the full dynticks, there is no way to test the support
563 for context tracking and the subsystems that rely on it: RCU
564 userspace extended quiescent state and tickless cputime
565 accounting. This option copes with the absence of the full
566 dynticks subsystem by forcing the context tracking on all
569 Say Y only if you're working on the development of an
570 architecture backend for the context tracking.
572 Say N otherwise, this option brings an overhead that you
573 don't want in production.
577 int "Tree-based hierarchical RCU fanout value"
580 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
584 This option controls the fanout of hierarchical implementations
585 of RCU, allowing RCU to work efficiently on machines with
586 large numbers of CPUs. This value must be at least the fourth
587 root of NR_CPUS, which allows NR_CPUS to be insanely large.
588 The default value of RCU_FANOUT should be used for production
589 systems, but if you are stress-testing the RCU implementation
590 itself, small RCU_FANOUT values allow you to test large-system
591 code paths on small(er) systems.
593 Select a specific number if testing RCU itself.
594 Take the default if unsure.
596 config RCU_FANOUT_LEAF
597 int "Tree-based hierarchical RCU leaf-level fanout value"
600 depends on (TREE_RCU || PREEMPT_RCU) && RCU_EXPERT
603 This option controls the leaf-level fanout of hierarchical
604 implementations of RCU, and allows trading off cache misses
605 against lock contention. Systems that synchronize their
606 scheduling-clock interrupts for energy-efficiency reasons will
607 want the default because the smaller leaf-level fanout keeps
608 lock contention levels acceptably low. Very large systems
609 (hundreds or thousands of CPUs) will instead want to set this
610 value to the maximum value possible in order to reduce the
611 number of cache misses incurred during RCU's grace-period
612 initialization. These systems tend to run CPU-bound, and thus
613 are not helped by synchronized interrupts, and thus tend to
614 skew them, which reduces lock contention enough that large
615 leaf-level fanouts work well.
617 Select a specific number if testing RCU itself.
619 Select the maximum permissible value for large systems.
621 Take the default if unsure.
623 config RCU_FAST_NO_HZ
624 bool "Accelerate last non-dyntick-idle CPU's grace periods"
625 depends on NO_HZ_COMMON && SMP && RCU_EXPERT
628 This option permits CPUs to enter dynticks-idle state even if
629 they have RCU callbacks queued, and prevents RCU from waking
630 these CPUs up more than roughly once every four jiffies (by
631 default, you can adjust this using the rcutree.rcu_idle_gp_delay
632 parameter), thus improving energy efficiency. On the other
633 hand, this option increases the duration of RCU grace periods,
634 for example, slowing down synchronize_rcu().
636 Say Y if energy efficiency is critically important, and you
637 don't care about increased grace-period durations.
639 Say N if you are unsure.
641 config TREE_RCU_TRACE
642 def_bool RCU_TRACE && ( TREE_RCU || PREEMPT_RCU )
645 This option provides tracing for the TREE_RCU and
646 PREEMPT_RCU implementations, permitting Makefile to
647 trivially select kernel/rcutree_trace.c.
650 bool "Enable RCU priority boosting"
651 depends on RT_MUTEXES && PREEMPT_RCU && RCU_EXPERT
654 This option boosts the priority of preempted RCU readers that
655 block the current preemptible RCU grace period for too long.
656 This option also prevents heavy loads from blocking RCU
657 callback invocation for all flavors of RCU.
659 Say Y here if you are working with real-time apps or heavy loads
660 Say N here if you are unsure.
662 config RCU_KTHREAD_PRIO
663 int "Real-time priority to use for RCU worker threads"
664 range 1 99 if RCU_BOOST
665 range 0 99 if !RCU_BOOST
666 default 1 if RCU_BOOST
667 default 0 if !RCU_BOOST
668 depends on RCU_EXPERT
670 This option specifies the SCHED_FIFO priority value that will be
671 assigned to the rcuc/n and rcub/n threads and is also the value
672 used for RCU_BOOST (if enabled). If you are working with a
673 real-time application that has one or more CPU-bound threads
674 running at a real-time priority level, you should set
675 RCU_KTHREAD_PRIO to a priority higher than the highest-priority
676 real-time CPU-bound application thread. The default RCU_KTHREAD_PRIO
677 value of 1 is appropriate in the common case, which is real-time
678 applications that do not have any CPU-bound threads.
680 Some real-time applications might not have a single real-time
681 thread that saturates a given CPU, but instead might have
682 multiple real-time threads that, taken together, fully utilize
683 that CPU. In this case, you should set RCU_KTHREAD_PRIO to
684 a priority higher than the lowest-priority thread that is
685 conspiring to prevent the CPU from running any non-real-time
686 tasks. For example, if one thread at priority 10 and another
687 thread at priority 5 are between themselves fully consuming
688 the CPU time on a given CPU, then RCU_KTHREAD_PRIO should be
689 set to priority 6 or higher.
691 Specify the real-time priority, or take the default if unsure.
693 config RCU_BOOST_DELAY
694 int "Milliseconds to delay boosting after RCU grace-period start"
699 This option specifies the time to wait after the beginning of
700 a given grace period before priority-boosting preempted RCU
701 readers blocking that grace period. Note that any RCU reader
702 blocking an expedited RCU grace period is boosted immediately.
704 Accept the default if unsure.
707 bool "Offload RCU callback processing from boot-selected CPUs"
708 depends on TREE_RCU || PREEMPT_RCU
709 depends on RCU_EXPERT || NO_HZ_FULL
712 Use this option to reduce OS jitter for aggressive HPC or
713 real-time workloads. It can also be used to offload RCU
714 callback invocation to energy-efficient CPUs in battery-powered
715 asymmetric multiprocessors.
717 This option offloads callback invocation from the set of
718 CPUs specified at boot time by the rcu_nocbs parameter.
719 For each such CPU, a kthread ("rcuox/N") will be created to
720 invoke callbacks, where the "N" is the CPU being offloaded,
721 and where the "x" is "b" for RCU-bh, "p" for RCU-preempt, and
722 "s" for RCU-sched. Nothing prevents this kthread from running
723 on the specified CPUs, but (1) the kthreads may be preempted
724 between each callback, and (2) affinity or cgroups can be used
725 to force the kthreads to run on whatever set of CPUs is desired.
727 Say Y here if you want to help to debug reduced OS jitter.
728 Say N here if you are unsure.
731 prompt "Build-forced no-CBs CPUs"
732 default RCU_NOCB_CPU_NONE
733 depends on RCU_NOCB_CPU
735 This option allows no-CBs CPUs (whose RCU callbacks are invoked
736 from kthreads rather than from softirq context) to be specified
737 at build time. Additional no-CBs CPUs may be specified by
738 the rcu_nocbs= boot parameter.
740 config RCU_NOCB_CPU_NONE
741 bool "No build_forced no-CBs CPUs"
743 This option does not force any of the CPUs to be no-CBs CPUs.
744 Only CPUs designated by the rcu_nocbs= boot parameter will be
745 no-CBs CPUs, whose RCU callbacks will be invoked by per-CPU
746 kthreads whose names begin with "rcuo". All other CPUs will
747 invoke their own RCU callbacks in softirq context.
749 Select this option if you want to choose no-CBs CPUs at
750 boot time, for example, to allow testing of different no-CBs
751 configurations without having to rebuild the kernel each time.
753 config RCU_NOCB_CPU_ZERO
754 bool "CPU 0 is a build_forced no-CBs CPU"
756 This option forces CPU 0 to be a no-CBs CPU, so that its RCU
757 callbacks are invoked by a per-CPU kthread whose name begins
758 with "rcuo". Additional CPUs may be designated as no-CBs
759 CPUs using the rcu_nocbs= boot parameter will be no-CBs CPUs.
760 All other CPUs will invoke their own RCU callbacks in softirq
763 Select this if CPU 0 needs to be a no-CBs CPU for real-time
764 or energy-efficiency reasons, but the real reason it exists
765 is to ensure that randconfig testing covers mixed systems.
767 config RCU_NOCB_CPU_ALL
768 bool "All CPUs are build_forced no-CBs CPUs"
770 This option forces all CPUs to be no-CBs CPUs. The rcu_nocbs=
771 boot parameter will be ignored. All CPUs' RCU callbacks will
772 be executed in the context of per-CPU rcuo kthreads created for
773 this purpose. Assuming that the kthreads whose names start with
774 "rcuo" are bound to "housekeeping" CPUs, this reduces OS jitter
775 on the remaining CPUs, but might decrease memory locality during
776 RCU-callback invocation, thus potentially degrading throughput.
778 Select this if all CPUs need to be no-CBs CPUs for real-time
779 or energy-efficiency reasons.
783 endmenu # "RCU Subsystem"
790 tristate "Kernel .config support"
793 This option enables the complete Linux kernel ".config" file
794 contents to be saved in the kernel. It provides documentation
795 of which kernel options are used in a running kernel or in an
796 on-disk kernel. This information can be extracted from the kernel
797 image file with the script scripts/extract-ikconfig and used as
798 input to rebuild the current kernel or to build another kernel.
799 It can also be extracted from a running kernel by reading
800 /proc/config.gz if enabled (below).
803 bool "Enable access to .config through /proc/config.gz"
804 depends on IKCONFIG && PROC_FS
806 This option enables access to the kernel configuration file
807 through /proc/config.gz.
810 int "Kernel log buffer size (16 => 64KB, 17 => 128KB)"
815 Select the minimal kernel log buffer size as a power of 2.
816 The final size is affected by LOG_CPU_MAX_BUF_SHIFT config
817 parameter, see below. Any higher size also might be forced
818 by "log_buf_len" boot parameter.
828 config LOG_CPU_MAX_BUF_SHIFT
829 int "CPU kernel log buffer size contribution (13 => 8 KB, 17 => 128KB)"
832 default 12 if !BASE_SMALL
833 default 0 if BASE_SMALL
836 This option allows to increase the default ring buffer size
837 according to the number of CPUs. The value defines the contribution
838 of each CPU as a power of 2. The used space is typically only few
839 lines however it might be much more when problems are reported,
842 The increased size means that a new buffer has to be allocated and
843 the original static one is unused. It makes sense only on systems
844 with more CPUs. Therefore this value is used only when the sum of
845 contributions is greater than the half of the default kernel ring
846 buffer as defined by LOG_BUF_SHIFT. The default values are set
847 so that more than 64 CPUs are needed to trigger the allocation.
849 Also this option is ignored when "log_buf_len" kernel parameter is
850 used as it forces an exact (power of two) size of the ring buffer.
852 The number of possible CPUs is used for this computation ignoring
853 hotplugging making the computation optimal for the worst case
854 scenario while allowing a simple algorithm to be used from bootup.
856 Examples shift values and their meaning:
857 17 => 128 KB for each CPU
858 16 => 64 KB for each CPU
859 15 => 32 KB for each CPU
860 14 => 16 KB for each CPU
861 13 => 8 KB for each CPU
862 12 => 4 KB for each CPU
864 config PRINTK_SAFE_LOG_BUF_SHIFT
865 int "Temporary per-CPU printk log buffer size (12 => 4KB, 13 => 8KB)"
870 Select the size of an alternate printk per-CPU buffer where messages
871 printed from usafe contexts are temporary stored. One example would
872 be NMI messages, another one - printk recursion. The messages are
873 copied to the main log buffer in a safe context to avoid a deadlock.
874 The value defines the size as a power of 2.
876 Those messages are rare and limited. The largest one is when
877 a backtrace is printed. It usually fits into 4KB. Select
878 8KB if you want to be on the safe side.
881 17 => 128 KB for each CPU
882 16 => 64 KB for each CPU
883 15 => 32 KB for each CPU
884 14 => 16 KB for each CPU
885 13 => 8 KB for each CPU
886 12 => 4 KB for each CPU
889 # Architectures with an unreliable sched_clock() should select this:
891 config HAVE_UNSTABLE_SCHED_CLOCK
894 config GENERIC_SCHED_CLOCK
898 # For architectures that want to enable the support for NUMA-affine scheduler
901 config ARCH_SUPPORTS_NUMA_BALANCING
905 # For architectures that prefer to flush all TLBs after a number of pages
906 # are unmapped instead of sending one IPI per page to flush. The architecture
907 # must provide guarantees on what happens if a clean TLB cache entry is
908 # written after the unmap. Details are in mm/rmap.c near the check for
909 # should_defer_flush. The architecture should also consider if the full flush
910 # and the refill costs are offset by the savings of sending fewer IPIs.
911 config ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
915 # For architectures that know their GCC __int128 support is sound
917 config ARCH_SUPPORTS_INT128
920 # For architectures that (ab)use NUMA to represent different memory regions
921 # all cpu-local but of different latencies, such as SuperH.
923 config ARCH_WANT_NUMA_VARIABLE_LOCALITY
926 config NUMA_BALANCING
927 bool "Memory placement aware NUMA scheduler"
928 depends on ARCH_SUPPORTS_NUMA_BALANCING
929 depends on !ARCH_WANT_NUMA_VARIABLE_LOCALITY
930 depends on SMP && NUMA && MIGRATION
932 This option adds support for automatic NUMA aware memory/task placement.
933 The mechanism is quite primitive and is based on migrating memory when
934 it has references to the node the task is running on.
936 This system will be inactive on UMA systems.
938 config NUMA_BALANCING_DEFAULT_ENABLED
939 bool "Automatically enable NUMA aware memory/task placement"
941 depends on NUMA_BALANCING
943 If set, automatic NUMA balancing will be enabled if running on a NUMA
947 bool "Control Group support"
950 This option adds support for grouping sets of processes together, for
951 use with process control subsystems such as Cpusets, CFS, memory
952 controls or device isolation.
954 - Documentation/scheduler/sched-design-CFS.txt (CFS)
955 - Documentation/cgroup-v1/ (features for grouping, isolation
956 and resource control)
966 bool "Memory controller"
970 Provides control over the memory footprint of tasks in a cgroup.
973 bool "Swap controller"
974 depends on MEMCG && SWAP
976 Provides control over the swap space consumed by tasks in a cgroup.
978 config MEMCG_SWAP_ENABLED
979 bool "Swap controller enabled by default"
980 depends on MEMCG_SWAP
983 Memory Resource Controller Swap Extension comes with its price in
984 a bigger memory consumption. General purpose distribution kernels
985 which want to enable the feature but keep it disabled by default
986 and let the user enable it by swapaccount=1 boot command line
987 parameter should have this option unselected.
988 For those who want to have the feature enabled by default should
989 select this option (if, for some reason, they need to disable it
990 then swapaccount=0 does the trick).
997 Generic block IO controller cgroup interface. This is the common
998 cgroup interface which should be used by various IO controlling
1001 Currently, CFQ IO scheduler uses it to recognize task groups and
1002 control disk bandwidth allocation (proportional time slice allocation)
1003 to such task groups. It is also used by bio throttling logic in
1004 block layer to implement upper limit in IO rates on a device.
1006 This option only enables generic Block IO controller infrastructure.
1007 One needs to also enable actual IO controlling logic/policy. For
1008 enabling proportional weight division of disk bandwidth in CFQ, set
1009 CONFIG_CFQ_GROUP_IOSCHED=y; for enabling throttling policy, set
1010 CONFIG_BLK_DEV_THROTTLING=y.
1012 See Documentation/cgroup-v1/blkio-controller.txt for more information.
1014 config DEBUG_BLK_CGROUP
1015 bool "IO controller debugging"
1016 depends on BLK_CGROUP
1019 Enable some debugging help. Currently it exports additional stat
1020 files in a cgroup which can be useful for debugging.
1022 config CGROUP_WRITEBACK
1024 depends on MEMCG && BLK_CGROUP
1027 menuconfig CGROUP_SCHED
1028 bool "CPU controller"
1031 This feature lets CPU scheduler recognize task groups and control CPU
1032 bandwidth allocation to such task groups. It uses cgroups to group
1036 config FAIR_GROUP_SCHED
1037 bool "Group scheduling for SCHED_OTHER"
1038 depends on CGROUP_SCHED
1039 default CGROUP_SCHED
1041 config CFS_BANDWIDTH
1042 bool "CPU bandwidth provisioning for FAIR_GROUP_SCHED"
1043 depends on FAIR_GROUP_SCHED
1046 This option allows users to define CPU bandwidth rates (limits) for
1047 tasks running within the fair group scheduler. Groups with no limit
1048 set are considered to be unconstrained and will run with no
1050 See tip/Documentation/scheduler/sched-bwc.txt for more information.
1052 config RT_GROUP_SCHED
1053 bool "Group scheduling for SCHED_RR/FIFO"
1054 depends on CGROUP_SCHED
1057 This feature lets you explicitly allocate real CPU bandwidth
1058 to task groups. If enabled, it will also make it impossible to
1059 schedule realtime tasks for non-root users until you allocate
1060 realtime bandwidth for them.
1061 See Documentation/scheduler/sched-rt-group.txt for more information.
1066 bool "PIDs controller"
1068 Provides enforcement of process number limits in the scope of a
1069 cgroup. Any attempt to fork more processes than is allowed in the
1070 cgroup will fail. PIDs are fundamentally a global resource because it
1071 is fairly trivial to reach PID exhaustion before you reach even a
1072 conservative kmemcg limit. As a result, it is possible to grind a
1073 system to halt without being limited by other cgroup policies. The
1074 PIDs controller is designed to stop this from happening.
1076 It should be noted that organisational operations (such as attaching
1077 to a cgroup hierarchy will *not* be blocked by the PIDs controller),
1078 since the PIDs limit only affects a process's ability to fork, not to
1082 bool "RDMA controller"
1084 Provides enforcement of RDMA resources defined by IB stack.
1085 It is fairly easy for consumers to exhaust RDMA resources, which
1086 can result into resource unavailability to other consumers.
1087 RDMA controller is designed to stop this from happening.
1088 Attaching processes with active RDMA resources to the cgroup
1089 hierarchy is allowed even if can cross the hierarchy's limit.
1091 config CGROUP_FREEZER
1092 bool "Freezer controller"
1094 Provides a way to freeze and unfreeze all tasks in a
1097 This option affects the ORIGINAL cgroup interface. The cgroup2 memory
1098 controller includes important in-kernel memory consumers per default.
1100 If you're using cgroup2, say N.
1102 config CGROUP_HUGETLB
1103 bool "HugeTLB controller"
1104 depends on HUGETLB_PAGE
1108 Provides a cgroup controller for HugeTLB pages.
1109 When you enable this, you can put a per cgroup limit on HugeTLB usage.
1110 The limit is enforced during page fault. Since HugeTLB doesn't
1111 support page reclaim, enforcing the limit at page fault time implies
1112 that, the application will get SIGBUS signal if it tries to access
1113 HugeTLB pages beyond its limit. This requires the application to know
1114 beforehand how much HugeTLB pages it would require for its use. The
1115 control group is tracked in the third page lru pointer. This means
1116 that we cannot use the controller with huge page less than 3 pages.
1119 bool "Cpuset controller"
1121 This option will let you create and manage CPUSETs which
1122 allow dynamically partitioning a system into sets of CPUs and
1123 Memory Nodes and assigning tasks to run only within those sets.
1124 This is primarily useful on large SMP or NUMA systems.
1128 config PROC_PID_CPUSET
1129 bool "Include legacy /proc/<pid>/cpuset file"
1133 config CGROUP_DEVICE
1134 bool "Device controller"
1136 Provides a cgroup controller implementing whitelists for
1137 devices which a process in the cgroup can mknod or open.
1139 config CGROUP_CPUACCT
1140 bool "Simple CPU accounting controller"
1142 Provides a simple controller for monitoring the
1143 total CPU consumed by the tasks in a cgroup.
1146 bool "Perf controller"
1147 depends on PERF_EVENTS
1149 This option extends the perf per-cpu mode to restrict monitoring
1150 to threads which belong to the cgroup specified and run on the
1156 bool "Support for eBPF programs attached to cgroups"
1157 depends on BPF_SYSCALL
1158 select SOCK_CGROUP_DATA
1160 Allow attaching eBPF programs to a cgroup using the bpf(2)
1161 syscall command BPF_PROG_ATTACH.
1163 In which context these programs are accessed depends on the type
1164 of attachment. For instance, programs that are attached using
1165 BPF_CGROUP_INET_INGRESS will be executed on the ingress path of
1169 bool "Example controller"
1172 This option enables a simple controller that exports
1173 debugging information about the cgroups framework.
1177 config SOCK_CGROUP_DATA
1183 config CHECKPOINT_RESTORE
1184 bool "Checkpoint/restore support" if EXPERT
1185 select PROC_CHILDREN
1188 Enables additional kernel features in a sake of checkpoint/restore.
1189 In particular it adds auxiliary prctl codes to setup process text,
1190 data and heap segment sizes, and a few additional /proc filesystem
1193 If unsure, say N here.
1195 menuconfig NAMESPACES
1196 bool "Namespaces support" if EXPERT
1197 depends on MULTIUSER
1200 Provides the way to make tasks work with different objects using
1201 the same id. For example same IPC id may refer to different objects
1202 or same user id or pid may refer to different tasks when used in
1203 different namespaces.
1208 bool "UTS namespace"
1211 In this namespace tasks see different info provided with the
1215 bool "IPC namespace"
1216 depends on (SYSVIPC || POSIX_MQUEUE)
1219 In this namespace tasks work with IPC ids which correspond to
1220 different IPC objects in different namespaces.
1223 bool "User namespace"
1226 This allows containers, i.e. vservers, to use user namespaces
1227 to provide different user info for different servers.
1229 When user namespaces are enabled in the kernel it is
1230 recommended that the MEMCG option also be enabled and that
1231 user-space use the memory control groups to limit the amount
1232 of memory a memory unprivileged users can use.
1237 bool "PID Namespaces"
1240 Support process id namespaces. This allows having multiple
1241 processes with the same pid as long as they are in different
1242 pid namespaces. This is a building block of containers.
1245 bool "Network namespace"
1249 Allow user space to create what appear to be multiple instances
1250 of the network stack.
1254 config SCHED_AUTOGROUP
1255 bool "Automatic process group scheduling"
1258 select FAIR_GROUP_SCHED
1260 This option optimizes the scheduler for common desktop workloads by
1261 automatically creating and populating task groups. This separation
1262 of workloads isolates aggressive CPU burners (like build jobs) from
1263 desktop applications. Task group autogeneration is currently based
1266 config SYSFS_DEPRECATED
1267 bool "Enable deprecated sysfs features to support old userspace tools"
1271 This option adds code that switches the layout of the "block" class
1272 devices, to not show up in /sys/class/block/, but only in
1275 This switch is only active when the sysfs.deprecated=1 boot option is
1276 passed or the SYSFS_DEPRECATED_V2 option is set.
1278 This option allows new kernels to run on old distributions and tools,
1279 which might get confused by /sys/class/block/. Since 2007/2008 all
1280 major distributions and tools handle this just fine.
1282 Recent distributions and userspace tools after 2009/2010 depend on
1283 the existence of /sys/class/block/, and will not work with this
1286 Only if you are using a new kernel on an old distribution, you might
1289 config SYSFS_DEPRECATED_V2
1290 bool "Enable deprecated sysfs features by default"
1293 depends on SYSFS_DEPRECATED
1295 Enable deprecated sysfs by default.
1297 See the CONFIG_SYSFS_DEPRECATED option for more details about this
1300 Only if you are using a new kernel on an old distribution, you might
1301 need to say Y here. Even then, odds are you would not need it
1302 enabled, you can always pass the boot option if absolutely necessary.
1305 bool "Kernel->user space relay support (formerly relayfs)"
1308 This option enables support for relay interface support in
1309 certain file systems (such as debugfs).
1310 It is designed to provide an efficient mechanism for tools and
1311 facilities to relay large amounts of data from kernel space to
1316 config BLK_DEV_INITRD
1317 bool "Initial RAM filesystem and RAM disk (initramfs/initrd) support"
1318 depends on BROKEN || !FRV
1320 The initial RAM filesystem is a ramfs which is loaded by the
1321 boot loader (loadlin or lilo) and that is mounted as root
1322 before the normal boot procedure. It is typically used to
1323 load modules needed to mount the "real" root file system,
1324 etc. See <file:Documentation/admin-guide/initrd.rst> for details.
1326 If RAM disk support (BLK_DEV_RAM) is also included, this
1327 also enables initial RAM disk (initrd) support and adds
1328 15 Kbytes (more on some other architectures) to the kernel size.
1334 source "usr/Kconfig"
1339 prompt "Compiler optimization level"
1340 default CONFIG_CC_OPTIMIZE_FOR_PERFORMANCE
1342 config CC_OPTIMIZE_FOR_PERFORMANCE
1343 bool "Optimize for performance"
1345 This is the default optimization level for the kernel, building
1346 with the "-O2" compiler flag for best performance and most
1347 helpful compile-time warnings.
1349 config CC_OPTIMIZE_FOR_SIZE
1350 bool "Optimize for size"
1352 Enabling this option will pass "-Os" instead of "-O2" to
1353 your compiler resulting in a smaller kernel.
1368 config SYSCTL_EXCEPTION_TRACE
1371 Enable support for /proc/sys/debug/exception-trace.
1373 config SYSCTL_ARCH_UNALIGN_NO_WARN
1376 Enable support for /proc/sys/kernel/ignore-unaligned-usertrap
1377 Allows arch to define/use @no_unaligned_warning to possibly warn
1378 about unaligned access emulation going on under the hood.
1380 config SYSCTL_ARCH_UNALIGN_ALLOW
1383 Enable support for /proc/sys/kernel/unaligned-trap
1384 Allows arches to define/use @unaligned_enabled to runtime toggle
1385 the unaligned access emulation.
1386 see arch/parisc/kernel/unaligned.c for reference
1388 config HAVE_PCSPKR_PLATFORM
1391 # interpreter that classic socket filters depend on
1396 bool "Configure standard kernel features (expert users)"
1397 # Unhide debug options, to make the on-by-default options visible
1400 This option allows certain base kernel options and settings
1401 to be disabled or tweaked. This is for specialized
1402 environments which can tolerate a "non-standard" kernel.
1403 Only use this if you really know what you are doing.
1406 bool "Enable 16-bit UID system calls" if EXPERT
1407 depends on HAVE_UID16 && MULTIUSER
1410 This enables the legacy 16-bit UID syscall wrappers.
1413 bool "Multiple users, groups and capabilities support" if EXPERT
1416 This option enables support for non-root users, groups and
1419 If you say N here, all processes will run with UID 0, GID 0, and all
1420 possible capabilities. Saying N here also compiles out support for
1421 system calls related to UIDs, GIDs, and capabilities, such as setuid,
1424 If unsure, say Y here.
1426 config SGETMASK_SYSCALL
1427 bool "sgetmask/ssetmask syscalls support" if EXPERT
1428 def_bool PARISC || MN10300 || BLACKFIN || M68K || PPC || MIPS || X86 || SPARC || CRIS || MICROBLAZE || SUPERH
1430 sys_sgetmask and sys_ssetmask are obsolete system calls
1431 no longer supported in libc but still enabled by default in some
1434 If unsure, leave the default option here.
1436 config SYSFS_SYSCALL
1437 bool "Sysfs syscall support" if EXPERT
1440 sys_sysfs is an obsolete system call no longer supported in libc.
1441 Note that disabling this option is more secure but might break
1442 compatibility with some systems.
1444 If unsure say Y here.
1446 config SYSCTL_SYSCALL
1447 bool "Sysctl syscall support" if EXPERT
1448 depends on PROC_SYSCTL
1452 sys_sysctl uses binary paths that have been found challenging
1453 to properly maintain and use. The interface in /proc/sys
1454 using paths with ascii names is now the primary path to this
1457 Almost nothing using the binary sysctl interface so if you are
1458 trying to save some space it is probably safe to disable this,
1459 making your kernel marginally smaller.
1461 If unsure say N here.
1464 bool "Posix Clocks & timers" if EXPERT
1467 This includes native support for POSIX timers to the kernel.
1468 Some embedded systems have no use for them and therefore they
1469 can be configured out to reduce the size of the kernel image.
1471 When this option is disabled, the following syscalls won't be
1472 available: timer_create, timer_gettime: timer_getoverrun,
1473 timer_settime, timer_delete, clock_adjtime, getitimer,
1474 setitimer, alarm. Furthermore, the clock_settime, clock_gettime,
1475 clock_getres and clock_nanosleep syscalls will be limited to
1476 CLOCK_REALTIME, CLOCK_MONOTONIC and CLOCK_BOOTTIME only.
1481 bool "Load all symbols for debugging/ksymoops" if EXPERT
1484 Say Y here to let the kernel print out symbolic crash information and
1485 symbolic stack backtraces. This increases the size of the kernel
1486 somewhat, as all symbols have to be loaded into the kernel image.
1489 bool "Include all symbols in kallsyms"
1490 depends on DEBUG_KERNEL && KALLSYMS
1492 Normally kallsyms only contains the symbols of functions for nicer
1493 OOPS messages and backtraces (i.e., symbols from the text and inittext
1494 sections). This is sufficient for most cases. And only in very rare
1495 cases (e.g., when a debugger is used) all symbols are required (e.g.,
1496 names of variables from the data sections, etc).
1498 This option makes sure that all symbols are loaded into the kernel
1499 image (i.e., symbols from all sections) in cost of increased kernel
1500 size (depending on the kernel configuration, it may be 300KiB or
1501 something like this).
1503 Say N unless you really need all symbols.
1505 config KALLSYMS_ABSOLUTE_PERCPU
1508 default X86_64 && SMP
1510 config KALLSYMS_BASE_RELATIVE
1513 default !IA64 && !(TILE && 64BIT)
1515 Instead of emitting them as absolute values in the native word size,
1516 emit the symbol references in the kallsyms table as 32-bit entries,
1517 each containing a relative value in the range [base, base + U32_MAX]
1518 or, when KALLSYMS_ABSOLUTE_PERCPU is in effect, each containing either
1519 an absolute value in the range [0, S32_MAX] or a relative value in the
1520 range [base, base + S32_MAX], where base is the lowest relative symbol
1521 address encountered in the image.
1523 On 64-bit builds, this reduces the size of the address table by 50%,
1524 but more importantly, it results in entries whose values are build
1525 time constants, and no relocation pass is required at runtime to fix
1526 up the entries based on the runtime load address of the kernel.
1530 bool "Enable support for printk" if EXPERT
1533 This option enables normal printk support. Removing it
1534 eliminates most of the message strings from the kernel image
1535 and makes the kernel more or less silent. As this makes it
1536 very difficult to diagnose system problems, saying N here is
1537 strongly discouraged.
1545 bool "BUG() support" if EXPERT
1548 Disabling this option eliminates support for BUG and WARN, reducing
1549 the size of your kernel image and potentially quietly ignoring
1550 numerous fatal conditions. You should only consider disabling this
1551 option for embedded systems with no facilities for reporting errors.
1557 bool "Enable ELF core dumps" if EXPERT
1559 Enable support for generating core dumps. Disabling saves about 4k.
1562 config PCSPKR_PLATFORM
1563 bool "Enable PC-Speaker support" if EXPERT
1564 depends on HAVE_PCSPKR_PLATFORM
1568 This option allows to disable the internal PC-Speaker
1569 support, saving some memory.
1573 bool "Enable full-sized data structures for core" if EXPERT
1575 Disabling this option reduces the size of miscellaneous core
1576 kernel data structures. This saves memory on small machines,
1577 but may reduce performance.
1580 bool "Enable futex support" if EXPERT
1584 Disabling this option will cause the kernel to be built without
1585 support for "fast userspace mutexes". The resulting kernel may not
1586 run glibc-based applications correctly.
1588 config HAVE_FUTEX_CMPXCHG
1592 Architectures should select this if futex_atomic_cmpxchg_inatomic()
1593 is implemented and always working. This removes a couple of runtime
1597 bool "Enable eventpoll support" if EXPERT
1601 Disabling this option will cause the kernel to be built without
1602 support for epoll family of system calls.
1605 bool "Enable signalfd() system call" if EXPERT
1609 Enable the signalfd() system call that allows to receive signals
1610 on a file descriptor.
1615 bool "Enable timerfd() system call" if EXPERT
1619 Enable the timerfd() system call that allows to receive timer
1620 events on a file descriptor.
1625 bool "Enable eventfd() system call" if EXPERT
1629 Enable the eventfd() system call that allows to receive both
1630 kernel notification (ie. KAIO) or userspace notifications.
1634 # syscall, maps, verifier
1636 bool "Enable bpf() system call"
1641 Enable the bpf() system call that allows to manipulate eBPF
1642 programs and maps via file descriptors.
1645 bool "Use full shmem filesystem" if EXPERT
1649 The shmem is an internal filesystem used to manage shared memory.
1650 It is backed by swap and manages resource limits. It is also exported
1651 to userspace as tmpfs if TMPFS is enabled. Disabling this
1652 option replaces shmem and tmpfs with the much simpler ramfs code,
1653 which may be appropriate on small systems without swap.
1656 bool "Enable AIO support" if EXPERT
1659 This option enables POSIX asynchronous I/O which may by used
1660 by some high performance threaded applications. Disabling
1661 this option saves about 7k.
1663 config ADVISE_SYSCALLS
1664 bool "Enable madvise/fadvise syscalls" if EXPERT
1667 This option enables the madvise and fadvise syscalls, used by
1668 applications to advise the kernel about their future memory or file
1669 usage, improving performance. If building an embedded system where no
1670 applications use these syscalls, you can disable this option to save
1674 bool "Enable userfaultfd() system call"
1678 Enable the userfaultfd() system call that allows to intercept and
1679 handle page faults in userland.
1683 bool "Enable PCI quirk workarounds" if EXPERT
1686 This enables workarounds for various PCI chipset
1687 bugs/quirks. Disable this only if your target machine is
1688 unaffected by PCI quirks.
1691 bool "Enable membarrier() system call" if EXPERT
1694 Enable the membarrier() system call that allows issuing memory
1695 barriers across all running threads, which can be used to distribute
1696 the cost of user-space memory barriers asymmetrically by transforming
1697 pairs of memory barriers into pairs consisting of membarrier() and a
1703 bool "Embedded system"
1704 option allnoconfig_y
1707 This option should be enabled if compiling the kernel for
1708 an embedded system so certain expert options are available
1711 config HAVE_PERF_EVENTS
1714 See tools/perf/design.txt for details.
1716 config PERF_USE_VMALLOC
1719 See tools/perf/design.txt for details
1722 bool "PC/104 support"
1724 Expose PC/104 form factor device drivers and options available for
1725 selection and configuration. Enable this option if your target
1726 machine has a PC/104 bus.
1728 menu "Kernel Performance Events And Counters"
1731 bool "Kernel performance events and counters"
1732 default y if PROFILING
1733 depends on HAVE_PERF_EVENTS
1738 Enable kernel support for various performance events provided
1739 by software and hardware.
1741 Software events are supported either built-in or via the
1742 use of generic tracepoints.
1744 Most modern CPUs support performance events via performance
1745 counter registers. These registers count the number of certain
1746 types of hw events: such as instructions executed, cachemisses
1747 suffered, or branches mis-predicted - without slowing down the
1748 kernel or applications. These registers can also trigger interrupts
1749 when a threshold number of events have passed - and can thus be
1750 used to profile the code that runs on that CPU.
1752 The Linux Performance Event subsystem provides an abstraction of
1753 these software and hardware event capabilities, available via a
1754 system call and used by the "perf" utility in tools/perf/. It
1755 provides per task and per CPU counters, and it provides event
1756 capabilities on top of those.
1760 config DEBUG_PERF_USE_VMALLOC
1762 bool "Debug: use vmalloc to back perf mmap() buffers"
1763 depends on PERF_EVENTS && DEBUG_KERNEL && !PPC
1764 select PERF_USE_VMALLOC
1766 Use vmalloc memory to back perf mmap() buffers.
1768 Mostly useful for debugging the vmalloc code on platforms
1769 that don't require it.
1775 config VM_EVENT_COUNTERS
1777 bool "Enable VM event counters for /proc/vmstat" if EXPERT
1779 VM event counters are needed for event counts to be shown.
1780 This option allows the disabling of the VM event counters
1781 on EXPERT systems. /proc/vmstat will only show page counts
1782 if VM event counters are disabled.
1786 bool "Enable SLUB debugging support" if EXPERT
1787 depends on SLUB && SYSFS
1789 SLUB has extensive debug support features. Disabling these can
1790 result in significant savings in code size. This also disables
1791 SLUB sysfs support. /sys/slab will not exist and there will be
1792 no support for cache validation etc.
1794 config SLUB_MEMCG_SYSFS_ON
1796 bool "Enable memcg SLUB sysfs support by default" if EXPERT
1797 depends on SLUB && SYSFS && MEMCG
1799 SLUB creates a directory under /sys/kernel/slab for each
1800 allocation cache to host info and debug files. If memory
1801 cgroup is enabled, each cache can have per memory cgroup
1802 caches. SLUB can create the same sysfs directories for these
1803 caches under /sys/kernel/slab/CACHE/cgroup but it can lead
1804 to a very high number of debug files being created. This is
1805 controlled by slub_memcg_sysfs boot parameter and this
1806 config option determines the parameter's default value.
1809 bool "Disable heap randomization"
1812 Randomizing heap placement makes heap exploits harder, but it
1813 also breaks ancient binaries (including anything libc5 based).
1814 This option changes the bootup default to heap randomization
1815 disabled, and can be overridden at runtime by setting
1816 /proc/sys/kernel/randomize_va_space to 2.
1818 On non-ancient distros (post-2000 ones) N is usually a safe choice.
1821 prompt "Choose SLAB allocator"
1824 This option allows to select a slab allocator.
1828 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1830 The regular slab allocator that is established and known to work
1831 well in all environments. It organizes cache hot objects in
1832 per cpu and per node queues.
1835 bool "SLUB (Unqueued Allocator)"
1836 select HAVE_HARDENED_USERCOPY_ALLOCATOR
1838 SLUB is a slab allocator that minimizes cache line usage
1839 instead of managing queues of cached objects (SLAB approach).
1840 Per cpu caching is realized using slabs of objects instead
1841 of queues of objects. SLUB can use memory efficiently
1842 and has enhanced diagnostics. SLUB is the default choice for
1847 bool "SLOB (Simple Allocator)"
1849 SLOB replaces the stock allocator with a drastically simpler
1850 allocator. SLOB is generally more space efficient but
1851 does not perform as well on large systems.
1855 config SLAB_FREELIST_RANDOM
1857 depends on SLAB || SLUB
1858 bool "SLAB freelist randomization"
1860 Randomizes the freelist order used on creating new pages. This
1861 security feature reduces the predictability of the kernel slab
1862 allocator against heap overflows.
1864 config SLUB_CPU_PARTIAL
1866 depends on SLUB && SMP
1867 bool "SLUB per cpu partial cache"
1869 Per cpu partial caches accellerate objects allocation and freeing
1870 that is local to a processor at the price of more indeterminism
1871 in the latency of the free. On overflow these caches will be cleared
1872 which requires the taking of locks that may cause latency spikes.
1873 Typically one would choose no for a realtime system.
1875 config MMAP_ALLOW_UNINITIALIZED
1876 bool "Allow mmapped anonymous memory to be uninitialized"
1877 depends on EXPERT && !MMU
1880 Normally, and according to the Linux spec, anonymous memory obtained
1881 from mmap() has it's contents cleared before it is passed to
1882 userspace. Enabling this config option allows you to request that
1883 mmap() skip that if it is given an MAP_UNINITIALIZED flag, thus
1884 providing a huge performance boost. If this option is not enabled,
1885 then the flag will be ignored.
1887 This is taken advantage of by uClibc's malloc(), and also by
1888 ELF-FDPIC binfmt's brk and stack allocator.
1890 Because of the obvious security issues, this option should only be
1891 enabled on embedded devices where you control what is run in
1892 userspace. Since that isn't generally a problem on no-MMU systems,
1893 it is normally safe to say Y here.
1895 See Documentation/nommu-mmap.txt for more information.
1897 config SYSTEM_DATA_VERIFICATION
1899 select SYSTEM_TRUSTED_KEYRING
1903 select ASYMMETRIC_KEY_TYPE
1904 select ASYMMETRIC_PUBLIC_KEY_SUBTYPE
1907 select X509_CERTIFICATE_PARSER
1908 select PKCS7_MESSAGE_PARSER
1910 Provide PKCS#7 message verification using the contents of the system
1911 trusted keyring to provide public keys. This then can be used for
1912 module verification, kexec image verification and firmware blob
1916 bool "Profiling support"
1918 Say Y here to enable the extended profiling support mechanisms used
1919 by profilers such as OProfile.
1922 # Place an empty function call at each tracepoint site. Can be
1923 # dynamically changed for a probe function.
1928 source "arch/Kconfig"
1930 endmenu # General setup
1932 config HAVE_GENERIC_DMA_COHERENT
1939 depends on SLAB || SLUB_DEBUG
1947 default 0 if BASE_FULL
1948 default 1 if !BASE_FULL
1951 bool "Enable loadable module support"
1954 Kernel modules are small pieces of compiled code which can
1955 be inserted in the running kernel, rather than being
1956 permanently built into the kernel. You use the "modprobe"
1957 tool to add (and sometimes remove) them. If you say Y here,
1958 many parts of the kernel can be built as modules (by
1959 answering M instead of Y where indicated): this is most
1960 useful for infrequently used options which are not required
1961 for booting. For more information, see the man pages for
1962 modprobe, lsmod, modinfo, insmod and rmmod.
1964 If you say Y here, you will need to run "make
1965 modules_install" to put the modules under /lib/modules/
1966 where modprobe can find them (you may need to be root to do
1973 config MODULE_FORCE_LOAD
1974 bool "Forced module loading"
1977 Allow loading of modules without version information (ie. modprobe
1978 --force). Forced module loading sets the 'F' (forced) taint flag and
1979 is usually a really bad idea.
1981 config MODULE_UNLOAD
1982 bool "Module unloading"
1984 Without this option you will not be able to unload any
1985 modules (note that some modules may not be unloadable
1986 anyway), which makes your kernel smaller, faster
1987 and simpler. If unsure, say Y.
1989 config MODULE_FORCE_UNLOAD
1990 bool "Forced module unloading"
1991 depends on MODULE_UNLOAD
1993 This option allows you to force a module to unload, even if the
1994 kernel believes it is unsafe: the kernel will remove the module
1995 without waiting for anyone to stop using it (using the -f option to
1996 rmmod). This is mainly for kernel developers and desperate users.
2000 bool "Module versioning support"
2002 Usually, you have to use modules compiled with your kernel.
2003 Saying Y here makes it sometimes possible to use modules
2004 compiled for different kernels, by adding enough information
2005 to the modules to (hopefully) spot any changes which would
2006 make them incompatible with the kernel you are running. If
2009 config MODULE_REL_CRCS
2011 depends on MODVERSIONS
2013 config MODULE_SRCVERSION_ALL
2014 bool "Source checksum for all modules"
2016 Modules which contain a MODULE_VERSION get an extra "srcversion"
2017 field inserted into their modinfo section, which contains a
2018 sum of the source files which made it. This helps maintainers
2019 see exactly which source was used to build a module (since
2020 others sometimes change the module source without updating
2021 the version). With this option, such a "srcversion" field
2022 will be created for all modules. If unsure, say N.
2025 bool "Module signature verification"
2027 select SYSTEM_DATA_VERIFICATION
2029 Check modules for valid signatures upon load: the signature
2030 is simply appended to the module. For more information see
2031 Documentation/module-signing.txt.
2033 Note that this option adds the OpenSSL development packages as a
2034 kernel build dependency so that the signing tool can use its crypto
2037 !!!WARNING!!! If you enable this option, you MUST make sure that the
2038 module DOES NOT get stripped after being signed. This includes the
2039 debuginfo strip done by some packagers (such as rpmbuild) and
2040 inclusion into an initramfs that wants the module size reduced.
2042 config MODULE_SIG_FORCE
2043 bool "Require modules to be validly signed"
2044 depends on MODULE_SIG
2046 Reject unsigned modules or signed modules for which we don't have a
2047 key. Without this, such modules will simply taint the kernel.
2049 config MODULE_SIG_ALL
2050 bool "Automatically sign all modules"
2052 depends on MODULE_SIG
2054 Sign all modules during make modules_install. Without this option,
2055 modules must be signed manually, using the scripts/sign-file tool.
2057 comment "Do not forget to sign required modules with scripts/sign-file"
2058 depends on MODULE_SIG_FORCE && !MODULE_SIG_ALL
2061 prompt "Which hash algorithm should modules be signed with?"
2062 depends on MODULE_SIG
2064 This determines which sort of hashing algorithm will be used during
2065 signature generation. This algorithm _must_ be built into the kernel
2066 directly so that signature verification can take place. It is not
2067 possible to load a signed module containing the algorithm to check
2068 the signature on that module.
2070 config MODULE_SIG_SHA1
2071 bool "Sign modules with SHA-1"
2074 config MODULE_SIG_SHA224
2075 bool "Sign modules with SHA-224"
2076 select CRYPTO_SHA256
2078 config MODULE_SIG_SHA256
2079 bool "Sign modules with SHA-256"
2080 select CRYPTO_SHA256
2082 config MODULE_SIG_SHA384
2083 bool "Sign modules with SHA-384"
2084 select CRYPTO_SHA512
2086 config MODULE_SIG_SHA512
2087 bool "Sign modules with SHA-512"
2088 select CRYPTO_SHA512
2092 config MODULE_SIG_HASH
2094 depends on MODULE_SIG
2095 default "sha1" if MODULE_SIG_SHA1
2096 default "sha224" if MODULE_SIG_SHA224
2097 default "sha256" if MODULE_SIG_SHA256
2098 default "sha384" if MODULE_SIG_SHA384
2099 default "sha512" if MODULE_SIG_SHA512
2101 config MODULE_COMPRESS
2102 bool "Compress modules on installation"
2106 Compresses kernel modules when 'make modules_install' is run; gzip or
2107 xz depending on "Compression algorithm" below.
2109 module-init-tools MAY support gzip, and kmod MAY support gzip and xz.
2111 Out-of-tree kernel modules installed using Kbuild will also be
2112 compressed upon installation.
2114 Note: for modules inside an initrd or initramfs, it's more efficient
2115 to compress the whole initrd or initramfs instead.
2117 Note: This is fully compatible with signed modules.
2122 prompt "Compression algorithm"
2123 depends on MODULE_COMPRESS
2124 default MODULE_COMPRESS_GZIP
2126 This determines which sort of compression will be used during
2127 'make modules_install'.
2129 GZIP (default) and XZ are supported.
2131 config MODULE_COMPRESS_GZIP
2134 config MODULE_COMPRESS_XZ
2139 config TRIM_UNUSED_KSYMS
2140 bool "Trim unused exported kernel symbols"
2141 depends on MODULES && !UNUSED_SYMBOLS
2143 The kernel and some modules make many symbols available for
2144 other modules to use via EXPORT_SYMBOL() and variants. Depending
2145 on the set of modules being selected in your kernel configuration,
2146 many of those exported symbols might never be used.
2148 This option allows for unused exported symbols to be dropped from
2149 the build. In turn, this provides the compiler more opportunities
2150 (especially when using LTO) for optimizing the code and reducing
2151 binary size. This might have some security advantages as well.
2153 If unsure, or if you need to build out-of-tree modules, say N.
2157 config MODULES_TREE_LOOKUP
2159 depends on PERF_EVENTS || TRACING
2161 config INIT_ALL_POSSIBLE
2164 Back when each arch used to define their own cpu_online_mask and
2165 cpu_possible_mask, some of them chose to initialize cpu_possible_mask
2166 with all 1s, and others with all 0s. When they were centralised,
2167 it was better to provide this option than to break all the archs
2168 and have several arch maintainers pursuing me down dark alleys.
2170 source "block/Kconfig"
2172 config PREEMPT_NOTIFIERS
2182 Build a simple ASN.1 grammar compiler that produces a bytecode output
2183 that can be interpreted by the ASN.1 stream decoder and used to
2184 inform it as to what tags are to be expected in a stream and what
2185 functions to call on what tags.
2187 source "kernel/Kconfig.locks"
git.samba.org / sfrench / cifs-2.6.git / blob