4 The configuration database is a collection of configuration options
5 organized in a tree structure:
7 +- Code maturity level options
8 | +- Prompt for development and/or incomplete code/drivers
10 | +- Networking support
12 | +- BSD Process Accounting
14 +- Loadable module support
15 | +- Enable loadable module support
16 | +- Set version information on all module symbols
17 | +- Kernel module loader
20 Every entry has its own dependencies. These dependencies are used
21 to determine the visibility of an entry. Any child entry is only
22 visible if its parent entry is also visible.
27 Most entries define a config option; all other entries help to organize
28 them. A single configuration option is defined like this:
31 bool "Set version information on all module symbols"
34 Usually, modules have to be recompiled whenever you switch to a new
37 Every line starts with a key word and can be followed by multiple
38 arguments. "config" starts a new config entry. The following lines
39 define attributes for this config option. Attributes can be the type of
40 the config option, input prompt, dependencies, help text and default
41 values. A config option can be defined multiple times with the same
42 name, but every definition can have only a single input prompt and the
43 type must not conflict.
48 A menu entry can have a number of attributes. Not all of them are
49 applicable everywhere (see syntax).
51 - type definition: "bool"/"tristate"/"string"/"hex"/"int"
52 Every config option must have a type. There are only two basic types:
53 tristate and string; the other types are based on these two. The type
54 definition optionally accepts an input prompt, so these two examples
57 bool "Networking support"
60 prompt "Networking support"
62 - input prompt: "prompt" <prompt> ["if" <expr>]
63 Every menu entry can have at most one prompt, which is used to display
64 to the user. Optionally dependencies only for this prompt can be added
67 - default value: "default" <expr> ["if" <expr>]
68 A config option can have any number of default values. If multiple
69 default values are visible, only the first defined one is active.
70 Default values are not limited to the menu entry where they are
71 defined. This means the default can be defined somewhere else or be
72 overridden by an earlier definition.
73 The default value is only assigned to the config symbol if no other
74 value was set by the user (via the input prompt above). If an input
75 prompt is visible the default value is presented to the user and can
77 Optionally, dependencies only for this default value can be added with
80 - type definition + default value:
81 "def_bool"/"def_tristate" <expr> ["if" <expr>]
82 This is a shorthand notation for a type definition plus a value.
83 Optionally dependencies for this default value can be added with "if".
85 - dependencies: "depends on" <expr>
86 This defines a dependency for this menu entry. If multiple
87 dependencies are defined, they are connected with '&&'. Dependencies
88 are applied to all other options within this menu entry (which also
89 accept an "if" expression), so these two examples are equivalent:
98 - reverse dependencies: "select" <symbol> ["if" <expr>]
99 While normal dependencies reduce the upper limit of a symbol (see
100 below), reverse dependencies can be used to force a lower limit of
101 another symbol. The value of the current menu symbol is used as the
102 minimal value <symbol> can be set to. If <symbol> is selected multiple
103 times, the limit is set to the largest selection.
104 Reverse dependencies can only be used with boolean or tristate
107 select should be used with care. select will force
108 a symbol to a value without visiting the dependencies.
109 By abusing select you are able to select a symbol FOO even
110 if FOO depends on BAR that is not set.
111 In general use select only for non-visible symbols
112 (no prompts anywhere) and for symbols with no dependencies.
113 That will limit the usefulness but on the other hand avoid
114 the illegal configurations all over.
116 - weak reverse dependencies: "imply" <symbol> ["if" <expr>]
117 This is similar to "select" as it enforces a lower limit on another
118 symbol except that the "implied" symbol's value may still be set to n
119 from a direct dependency or with a visible prompt.
121 Given the following example:
131 The following values are possible:
133 FOO BAR BAZ's default choice for BAZ
134 --- --- ------------- --------------
140 This is useful e.g. with multiple drivers that want to indicate their
141 ability to hook into a secondary subsystem while allowing the user to
142 configure that subsystem out without also having to unset these drivers.
144 - limiting menu display: "visible if" <expr>
145 This attribute is only applicable to menu blocks, if the condition is
146 false, the menu block is not displayed to the user (the symbols
147 contained there can still be selected by other symbols, though). It is
148 similar to a conditional "prompt" attribute for individual menu
149 entries. Default value of "visible" is true.
151 - numerical ranges: "range" <symbol> <symbol> ["if" <expr>]
152 This allows to limit the range of possible input values for int
153 and hex symbols. The user can only input a value which is larger than
154 or equal to the first symbol and smaller than or equal to the second
157 - help text: "help" or "---help---"
158 This defines a help text. The end of the help text is determined by
159 the indentation level, this means it ends at the first line which has
160 a smaller indentation than the first line of the help text.
161 "---help---" and "help" do not differ in behaviour, "---help---" is
162 used to help visually separate configuration logic from help within
163 the file as an aid to developers.
165 - misc options: "option" <symbol>[=<value>]
166 Various less common options can be defined via this option syntax,
167 which can modify the behaviour of the menu entry and its config
168 symbol. These options are currently possible:
171 This declares a list of default entries which can be used when
172 looking for the default configuration (which is used when the main
173 .config doesn't exists yet.)
176 This declares the symbol to be used as the MODULES symbol, which
177 enables the third modular state for all config symbols.
178 At most one symbol may have the "modules" option set.
181 This imports the environment variable into Kconfig. It behaves like
182 a default, except that the value comes from the environment, this
183 also means that the behaviour when mixing it with normal defaults is
184 undefined at this point. The symbol is currently not exported back
185 to the build environment (if this is desired, it can be done via
189 This declares the symbol as one that should have the value y when
190 using "allnoconfig". Used for symbols that hide other symbols.
195 Dependencies define the visibility of a menu entry and can also reduce
196 the input range of tristate symbols. The tristate logic used in the
197 expressions uses one more state than normal boolean logic to express the
198 module state. Dependency expressions have the following syntax:
200 <expr> ::= <symbol> (1)
201 <symbol> '=' <symbol> (2)
202 <symbol> '!=' <symbol> (3)
205 <expr> '&&' <expr> (6)
206 <expr> '||' <expr> (7)
208 Expressions are listed in decreasing order of precedence.
210 (1) Convert the symbol into an expression. Boolean and tristate symbols
211 are simply converted into the respective expression values. All
212 other symbol types result in 'n'.
213 (2) If the values of both symbols are equal, it returns 'y',
215 (3) If the values of both symbols are equal, it returns 'n',
217 (4) Returns the value of the expression. Used to override precedence.
218 (5) Returns the result of (2-/expr/).
219 (6) Returns the result of min(/expr/, /expr/).
220 (7) Returns the result of max(/expr/, /expr/).
222 An expression can have a value of 'n', 'm' or 'y' (or 0, 1, 2
223 respectively for calculations). A menu entry becomes visible when its
224 expression evaluates to 'm' or 'y'.
226 There are two types of symbols: constant and non-constant symbols.
227 Non-constant symbols are the most common ones and are defined with the
228 'config' statement. Non-constant symbols consist entirely of alphanumeric
229 characters or underscores.
230 Constant symbols are only part of expressions. Constant symbols are
231 always surrounded by single or double quotes. Within the quote, any
232 other character is allowed and the quotes can be escaped using '\'.
237 The position of a menu entry in the tree is determined in two ways. First
238 it can be specified explicitly:
240 menu "Network device support"
248 All entries within the "menu" ... "endmenu" block become a submenu of
249 "Network device support". All subentries inherit the dependencies from
250 the menu entry, e.g. this means the dependency "NET" is added to the
251 dependency list of the config option NETDEVICES.
253 The other way to generate the menu structure is done by analyzing the
254 dependencies. If a menu entry somehow depends on the previous entry, it
255 can be made a submenu of it. First, the previous (parent) symbol must
256 be part of the dependency list and then one of these two conditions
258 - the child entry must become invisible, if the parent is set to 'n'
259 - the child entry must only be visible, if the parent is visible
262 bool "Enable loadable module support"
265 bool "Set version information on all module symbols"
268 comment "module support disabled"
271 MODVERSIONS directly depends on MODULES, this means it's only visible if
272 MODULES is different from 'n'. The comment on the other hand is only
273 visible when MODULES is set to 'n'.
279 The configuration file describes a series of menu entries, where every
280 line starts with a keyword (except help texts). The following keywords
289 The first five also start the definition of a menu entry.
296 This defines a config symbol <symbol> and accepts any of above
297 attributes as options.
300 "menuconfig" <symbol>
303 This is similar to the simple config entry above, but it also gives a
304 hint to front ends, that all suboptions should be displayed as a
305 separate list of options. To make sure all the suboptions will really
306 show up under the menuconfig entry and not outside of it, every item
307 from the <config options> list must depend on the menuconfig symbol.
308 In practice, this is achieved by using one of the next two constructs:
324 In the following examples (3) and (4), C1 and C2 still have the M
325 dependency, but will not appear under menuconfig M anymore, because
326 of C0, which doesn't depend on M:
351 This defines a choice group and accepts any of the above attributes as
352 options. A choice can only be of type bool or tristate. If no type is
353 specified for a choice, it's type will be determined by the type of
354 the first choice element in the group or remain unknown if none of the
355 choice elements have a type specified, as well.
357 While a boolean choice only allows a single config entry to be
358 selected, a tristate choice also allows any number of config entries
359 to be set to 'm'. This can be used if multiple drivers for a single
360 hardware exists and only a single driver can be compiled/loaded into
361 the kernel, but all drivers can be compiled as modules.
363 A choice accepts another option "optional", which allows to set the
364 choice to 'n' and no entry needs to be selected.
365 If no [symbol] is associated with a choice, then you can not have multiple
366 definitions of that choice. If a [symbol] is associated to the choice,
367 then you may define the same choice (ie. with the same entries) in another
375 This defines a comment which is displayed to the user during the
376 configuration process and is also echoed to the output files. The only
377 possible options are dependencies.
386 This defines a menu block, see "Menu structure" above for more
387 information. The only possible options are dependencies and "visible"
396 This defines an if block. The dependency expression <expr> is appended
397 to all enclosed menu entries.
403 This reads the specified configuration file. This file is always parsed.
409 This sets the config program's title bar if the config program chooses
410 to use it. It should be placed at the top of the configuration, before any
416 This is a collection of Kconfig tips, most of which aren't obvious at
417 first glance and most of which have become idioms in several Kconfig
420 Adding common features and make the usage configurable
421 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
422 It is a common idiom to implement a feature/functionality that are
423 relevant for some architectures but not all.
424 The recommended way to do so is to use a config variable named HAVE_*
425 that is defined in a common Kconfig file and selected by the relevant
427 An example is the generic IOMAP functionality.
429 We would in lib/Kconfig see:
431 # Generic IOMAP is used to ...
432 config HAVE_GENERIC_IOMAP
435 depends on HAVE_GENERIC_IOMAP && FOO
437 And in lib/Makefile we would see:
438 obj-$(CONFIG_GENERIC_IOMAP) += iomap.o
440 For each architecture using the generic IOMAP functionality we would see:
444 select HAVE_GENERIC_IOMAP
447 Note: we use the existing config option and avoid creating a new
448 config variable to select HAVE_GENERIC_IOMAP.
450 Note: the use of the internal config variable HAVE_GENERIC_IOMAP, it is
451 introduced to overcome the limitation of select which will force a
452 config option to 'y' no matter the dependencies.
453 The dependencies are moved to the symbol GENERIC_IOMAP and we avoid the
454 situation where select forces a symbol equals to 'y'.
458 To restrict a component build to module-only, qualify its config symbol
459 with "depends on m". E.g.:
464 limits FOO to module (=m) or disabled (=n).
466 Kconfig recursive dependency limitations
467 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
469 If you've hit the Kconfig error: "recursive dependency detected" you've run
470 into a recursive dependency issue with Kconfig, a recursive dependency can be
471 summarized as a circular dependency. The kconfig tools need to ensure that
472 Kconfig files comply with specified configuration requirements. In order to do
473 that kconfig must determine the values that are possible for all Kconfig
474 symbols, this is currently not possible if there is a circular relation
475 between two or more Kconfig symbols. For more details refer to the "Simple
476 Kconfig recursive issue" subsection below. Kconfig does not do recursive
477 dependency resolution; this has a few implications for Kconfig file writers.
478 We'll first explain why this issues exists and then provide an example
479 technical limitation which this brings upon Kconfig developers. Eager
480 developers wishing to try to address this limitation should read the next
483 Simple Kconfig recursive issue
484 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
486 Read: Documentation/kbuild/Kconfig.recursion-issue-01
490 make KBUILD_KCONFIG=Documentation/kbuild/Kconfig.recursion-issue-01 allnoconfig
492 Cumulative Kconfig recursive issue
493 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
495 Read: Documentation/kbuild/Kconfig.recursion-issue-02
499 make KBUILD_KCONFIG=Documentation/kbuild/Kconfig.recursion-issue-02 allnoconfig
501 Practical solutions to kconfig recursive issue
502 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
504 Developers who run into the recursive Kconfig issue have three options
505 at their disposal. We document them below and also provide a list of
506 historical issues resolved through these different solutions.
508 a) Remove any superfluous "select FOO" or "depends on FOO"
509 b) Match dependency semantics:
510 b1) Swap all "select FOO" to "depends on FOO" or,
511 b2) Swap all "depends on FOO" to "select FOO"
512 c) Consider the use of "imply" instead of "select"
514 The resolution to a) can be tested with the sample Kconfig file
515 Documentation/kbuild/Kconfig.recursion-issue-01 through the removal
516 of the "select CORE" from CORE_BELL_A_ADVANCED as that is implicit already
517 since CORE_BELL_A depends on CORE. At times it may not be possible to remove
518 some dependency criteria, for such cases you can work with solution b).
520 The two different resolutions for b) can be tested in the sample Kconfig file
521 Documentation/kbuild/Kconfig.recursion-issue-02.
523 Below is a list of examples of prior fixes for these types of recursive issues;
524 all errors appear to involve one or more select's and one or more "depends on".
528 06b718c01208 select A -> depends on A
529 c22eacfe82f9 depends on A -> depends on B
530 6a91e854442c select A -> depends on A
531 118c565a8f2e select A -> select B
532 f004e5594705 select A -> depends on A
533 c7861f37b4c6 depends on A -> (null)
534 80c69915e5fb select A -> (null) (1)
535 c2218e26c0d0 select A -> depends on A (1)
536 d6ae99d04e1c select A -> depends on A
537 95ca19cf8cbf select A -> depends on A
538 8f057d7bca54 depends on A -> (null)
539 8f057d7bca54 depends on A -> select A
540 a0701f04846e select A -> depends on A
541 0c8b92f7f259 depends on A -> (null)
542 e4e9e0540928 select A -> depends on A (2)
543 7453ea886e87 depends on A > (null) (1)
544 7b1fff7e4fdf select A -> depends on A
545 86c747d2a4f0 select A -> depends on A
546 d9f9ab51e55e select A -> depends on A
547 0c51a4d8abd6 depends on A -> select A (3)
548 e98062ed6dc4 select A -> depends on A (3)
549 91e5d284a7f1 select A -> (null)
551 (1) Partial (or no) quote of error.
552 (2) That seems to be the gist of that fix.
558 Work on kconfig is welcomed on both areas of clarifying semantics and on
559 evaluating the use of a full SAT solver for it. A full SAT solver can be
560 desirable to enable more complex dependency mappings and / or queries,
561 for instance on possible use case for a SAT solver could be that of handling
562 the current known recursive dependency issues. It is not known if this would
563 address such issues but such evaluation is desirable. If support for a full SAT
564 solver proves too complex or that it cannot address recursive dependency issues
565 Kconfig should have at least clear and well defined semantics which also
566 addresses and documents limitations or requirements such as the ones dealing
567 with recursive dependencies.
569 Further work on both of these areas is welcomed on Kconfig. We elaborate
570 on both of these in the next two subsections.
575 The use of Kconfig is broad, Linux is now only one of Kconfig's users:
576 one study has completed a broad analysis of Kconfig use in 12 projects [0].
577 Despite its widespread use, and although this document does a reasonable job
578 in documenting basic Kconfig syntax a more precise definition of Kconfig
579 semantics is welcomed. One project deduced Kconfig semantics through
580 the use of the xconfig configurator [1]. Work should be done to confirm if
581 the deduced semantics matches our intended Kconfig design goals.
583 Having well defined semantics can be useful for tools for practical
584 evaluation of depenencies, for instance one such use known case was work to
585 express in boolean abstraction of the inferred semantics of Kconfig to
586 translate Kconfig logic into boolean formulas and run a SAT solver on this to
587 find dead code / features (always inactive), 114 dead features were found in
588 Linux using this methodology [1] (Section 8: Threats to validity).
590 Confirming this could prove useful as Kconfig stands as one of the the leading
591 industrial variability modeling languages [1] [2]. Its study would help
592 evaluate practical uses of such languages, their use was only theoretical
593 and real world requirements were not well understood. As it stands though
594 only reverse engineering techniques have been used to deduce semantics from
595 variability modeling languages such as Kconfig [3].
597 [0] http://www.eng.uwaterloo.ca/~shshe/kconfig_semantics.pdf
598 [1] http://gsd.uwaterloo.ca/sites/default/files/vm-2013-berger.pdf
599 [2] http://gsd.uwaterloo.ca/sites/default/files/ase241-berger_0.pdf
600 [3] http://gsd.uwaterloo.ca/sites/default/files/icse2011.pdf
602 Full SAT solver for Kconfig
603 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
605 Although SAT solvers [0] haven't yet been used by Kconfig directly, as noted in
606 the previous subsection, work has been done however to express in boolean
607 abstraction the inferred semantics of Kconfig to translate Kconfig logic into
608 boolean formulas and run a SAT solver on it [1]. Another known related project
609 is CADOS [2] (former VAMOS [3]) and the tools, mainly undertaker [4], which has
610 been introduced first with [5]. The basic concept of undertaker is to exract
611 variability models from Kconfig, and put them together with a propositional
612 formula extracted from CPP #ifdefs and build-rules into a SAT solver in order
613 to find dead code, dead files, and dead symbols. If using a SAT solver is
614 desirable on Kconfig one approach would be to evaluate repurposing such efforts
615 somehow on Kconfig. There is enough interest from mentors of existing projects
616 to not only help advise how to integrate this work upstream but also help
617 maintain it long term. Interested developers should visit:
619 http://kernelnewbies.org/KernelProjects/kconfig-sat
621 [0] http://www.cs.cornell.edu/~sabhar/chapters/SATSolvers-KR-Handbook.pdf
622 [1] http://gsd.uwaterloo.ca/sites/default/files/vm-2013-berger.pdf
623 [2] https://cados.cs.fau.de
624 [3] https://vamos.cs.fau.de
625 [4] https://undertaker.cs.fau.de
626 [5] https://www4.cs.fau.de/Publications/2011/tartler_11_eurosys.pdf
git.samba.org / sfrench / cifs-2.6.git / blob