legacy.rst 34 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769
  1. ======================
  2. Legacy GPIO Interfaces
  3. ======================
  4. This provides an overview of GPIO access conventions on Linux.
  5. These calls use the gpio_* naming prefix. No other calls should use that
  6. prefix, or the related __gpio_* prefix.
  7. What is a GPIO?
  8. ===============
  9. A "General Purpose Input/Output" (GPIO) is a flexible software-controlled
  10. digital signal. They are provided from many kinds of chip, and are familiar
  11. to Linux developers working with embedded and custom hardware. Each GPIO
  12. represents a bit connected to a particular pin, or "ball" on Ball Grid Array
  13. (BGA) packages. Board schematics show which external hardware connects to
  14. which GPIOs. Drivers can be written generically, so that board setup code
  15. passes such pin configuration data to drivers.
  16. System-on-Chip (SOC) processors heavily rely on GPIOs. In some cases, every
  17. non-dedicated pin can be configured as a GPIO; and most chips have at least
  18. several dozen of them. Programmable logic devices (like FPGAs) can easily
  19. provide GPIOs; multifunction chips like power managers, and audio codecs
  20. often have a few such pins to help with pin scarcity on SOCs; and there are
  21. also "GPIO Expander" chips that connect using the I2C or SPI serial busses.
  22. Most PC southbridges have a few dozen GPIO-capable pins (with only the BIOS
  23. firmware knowing how they're used).
  24. The exact capabilities of GPIOs vary between systems. Common options:
  25. - Output values are writable (high=1, low=0). Some chips also have
  26. options about how that value is driven, so that for example only one
  27. value might be driven ... supporting "wire-OR" and similar schemes
  28. for the other value (notably, "open drain" signaling).
  29. - Input values are likewise readable (1, 0). Some chips support readback
  30. of pins configured as "output", which is very useful in such "wire-OR"
  31. cases (to support bidirectional signaling). GPIO controllers may have
  32. input de-glitch/debounce logic, sometimes with software controls.
  33. - Inputs can often be used as IRQ signals, often edge triggered but
  34. sometimes level triggered. Such IRQs may be configurable as system
  35. wakeup events, to wake the system from a low power state.
  36. - Usually a GPIO will be configurable as either input or output, as needed
  37. by different product boards; single direction ones exist too.
  38. - Most GPIOs can be accessed while holding spinlocks, but those accessed
  39. through a serial bus normally can't. Some systems support both types.
  40. On a given board each GPIO is used for one specific purpose like monitoring
  41. MMC/SD card insertion/removal, detecting card writeprotect status, driving
  42. a LED, configuring a transceiver, bitbanging a serial bus, poking a hardware
  43. watchdog, sensing a switch, and so on.
  44. GPIO conventions
  45. ================
  46. Note that this is called a "convention" because you don't need to do it this
  47. way, and it's no crime if you don't. There **are** cases where portability
  48. is not the main issue; GPIOs are often used for the kind of board-specific
  49. glue logic that may even change between board revisions, and can't ever be
  50. used on a board that's wired differently. Only least-common-denominator
  51. functionality can be very portable. Other features are platform-specific,
  52. and that can be critical for glue logic.
  53. Plus, this doesn't require any implementation framework, just an interface.
  54. One platform might implement it as simple inline functions accessing chip
  55. registers; another might implement it by delegating through abstractions
  56. used for several very different kinds of GPIO controller. (There is some
  57. optional code supporting such an implementation strategy, described later
  58. in this document, but drivers acting as clients to the GPIO interface must
  59. not care how it's implemented.)
  60. That said, if the convention is supported on their platform, drivers should
  61. use it when possible. Platforms must select GPIOLIB if GPIO functionality
  62. is strictly required. Drivers that can't work without
  63. standard GPIO calls should have Kconfig entries which depend on GPIOLIB. The
  64. GPIO calls are available, either as "real code" or as optimized-away stubs,
  65. when drivers use the include file:
  66. #include <linux/gpio.h>
  67. If you stick to this convention then it'll be easier for other developers to
  68. see what your code is doing, and help maintain it.
  69. Note that these operations include I/O barriers on platforms which need to
  70. use them; drivers don't need to add them explicitly.
  71. Identifying GPIOs
  72. -----------------
  73. GPIOs are identified by unsigned integers in the range 0..MAX_INT. That
  74. reserves "negative" numbers for other purposes like marking signals as
  75. "not available on this board", or indicating faults. Code that doesn't
  76. touch the underlying hardware treats these integers as opaque cookies.
  77. Platforms define how they use those integers, and usually #define symbols
  78. for the GPIO lines so that board-specific setup code directly corresponds
  79. to the relevant schematics. In contrast, drivers should only use GPIO
  80. numbers passed to them from that setup code, using platform_data to hold
  81. board-specific pin configuration data (along with other board specific
  82. data they need). That avoids portability problems.
  83. So for example one platform uses numbers 32-159 for GPIOs; while another
  84. uses numbers 0..63 with one set of GPIO controllers, 64-79 with another
  85. type of GPIO controller, and on one particular board 80-95 with an FPGA.
  86. The numbers need not be contiguous; either of those platforms could also
  87. use numbers 2000-2063 to identify GPIOs in a bank of I2C GPIO expanders.
  88. If you want to initialize a structure with an invalid GPIO number, use
  89. some negative number (perhaps "-EINVAL"); that will never be valid. To
  90. test if such number from such a structure could reference a GPIO, you
  91. may use this predicate:
  92. int gpio_is_valid(int number);
  93. A number that's not valid will be rejected by calls which may request
  94. or free GPIOs (see below). Other numbers may also be rejected; for
  95. example, a number might be valid but temporarily unused on a given board.
  96. Whether a platform supports multiple GPIO controllers is a platform-specific
  97. implementation issue, as are whether that support can leave "holes" in the space
  98. of GPIO numbers, and whether new controllers can be added at runtime. Such issues
  99. can affect things including whether adjacent GPIO numbers are both valid.
  100. Using GPIOs
  101. -----------
  102. The first thing a system should do with a GPIO is allocate it, using
  103. the gpio_request() call; see later.
  104. One of the next things to do with a GPIO, often in board setup code when
  105. setting up a platform_device using the GPIO, is mark its direction::
  106. /* set as input or output, returning 0 or negative errno */
  107. int gpio_direction_input(unsigned gpio);
  108. int gpio_direction_output(unsigned gpio, int value);
  109. The return value is zero for success, else a negative errno. It should
  110. be checked, since the get/set calls don't have error returns and since
  111. misconfiguration is possible. You should normally issue these calls from
  112. a task context. However, for spinlock-safe GPIOs it's OK to use them
  113. before tasking is enabled, as part of early board setup.
  114. For output GPIOs, the value provided becomes the initial output value.
  115. This helps avoid signal glitching during system startup.
  116. For compatibility with legacy interfaces to GPIOs, setting the direction
  117. of a GPIO implicitly requests that GPIO (see below) if it has not been
  118. requested already. That compatibility is being removed from the optional
  119. gpiolib framework.
  120. Setting the direction can fail if the GPIO number is invalid, or when
  121. that particular GPIO can't be used in that mode. It's generally a bad
  122. idea to rely on boot firmware to have set the direction correctly, since
  123. it probably wasn't validated to do more than boot Linux. (Similarly,
  124. that board setup code probably needs to multiplex that pin as a GPIO,
  125. and configure pullups/pulldowns appropriately.)
  126. Spinlock-Safe GPIO access
  127. -------------------------
  128. Most GPIO controllers can be accessed with memory read/write instructions.
  129. Those don't need to sleep, and can safely be done from inside hard
  130. (nonthreaded) IRQ handlers and similar contexts.
  131. Use the following calls to access such GPIOs,
  132. for which gpio_cansleep() will always return false (see below)::
  133. /* GPIO INPUT: return zero or nonzero */
  134. int gpio_get_value(unsigned gpio);
  135. /* GPIO OUTPUT */
  136. void gpio_set_value(unsigned gpio, int value);
  137. The values are boolean, zero for low, nonzero for high. When reading the
  138. value of an output pin, the value returned should be what's seen on the
  139. pin ... that won't always match the specified output value, because of
  140. issues including open-drain signaling and output latencies.
  141. The get/set calls have no error returns because "invalid GPIO" should have
  142. been reported earlier from gpio_direction_*(). However, note that not all
  143. platforms can read the value of output pins; those that can't should always
  144. return zero. Also, using these calls for GPIOs that can't safely be accessed
  145. without sleeping (see below) is an error.
  146. Platform-specific implementations are encouraged to optimize the two
  147. calls to access the GPIO value in cases where the GPIO number (and for
  148. output, value) are constant. It's normal for them to need only a couple
  149. of instructions in such cases (reading or writing a hardware register),
  150. and not to need spinlocks. Such optimized calls can make bitbanging
  151. applications a lot more efficient (in both space and time) than spending
  152. dozens of instructions on subroutine calls.
  153. GPIO access that may sleep
  154. --------------------------
  155. Some GPIO controllers must be accessed using message based busses like I2C
  156. or SPI. Commands to read or write those GPIO values require waiting to
  157. get to the head of a queue to transmit a command and get its response.
  158. This requires sleeping, which can't be done from inside IRQ handlers.
  159. Platforms that support this type of GPIO distinguish them from other GPIOs
  160. by returning nonzero from this call (which requires a valid GPIO number,
  161. which should have been previously allocated with gpio_request)::
  162. int gpio_cansleep(unsigned gpio);
  163. To access such GPIOs, a different set of accessors is defined::
  164. /* GPIO INPUT: return zero or nonzero, might sleep */
  165. int gpio_get_value_cansleep(unsigned gpio);
  166. /* GPIO OUTPUT, might sleep */
  167. void gpio_set_value_cansleep(unsigned gpio, int value);
  168. Accessing such GPIOs requires a context which may sleep, for example
  169. a threaded IRQ handler, and those accessors must be used instead of
  170. spinlock-safe accessors without the cansleep() name suffix.
  171. Other than the fact that these accessors might sleep, and will work
  172. on GPIOs that can't be accessed from hardIRQ handlers, these calls act
  173. the same as the spinlock-safe calls.
  174. **IN ADDITION** calls to setup and configure such GPIOs must be made
  175. from contexts which may sleep, since they may need to access the GPIO
  176. controller chip too (These setup calls are usually made from board
  177. setup or driver probe/teardown code, so this is an easy constraint.)::
  178. gpio_direction_input()
  179. gpio_direction_output()
  180. gpio_request()
  181. ## gpio_request_one()
  182. ## gpio_request_array()
  183. ## gpio_free_array()
  184. gpio_free()
  185. gpio_set_debounce()
  186. Claiming and Releasing GPIOs
  187. ----------------------------
  188. To help catch system configuration errors, two calls are defined::
  189. /* request GPIO, returning 0 or negative errno.
  190. * non-null labels may be useful for diagnostics.
  191. */
  192. int gpio_request(unsigned gpio, const char *label);
  193. /* release previously-claimed GPIO */
  194. void gpio_free(unsigned gpio);
  195. Passing invalid GPIO numbers to gpio_request() will fail, as will requesting
  196. GPIOs that have already been claimed with that call. The return value of
  197. gpio_request() must be checked. You should normally issue these calls from
  198. a task context. However, for spinlock-safe GPIOs it's OK to request GPIOs
  199. before tasking is enabled, as part of early board setup.
  200. These calls serve two basic purposes. One is marking the signals which
  201. are actually in use as GPIOs, for better diagnostics; systems may have
  202. several hundred potential GPIOs, but often only a dozen are used on any
  203. given board. Another is to catch conflicts, identifying errors when
  204. (a) two or more drivers wrongly think they have exclusive use of that
  205. signal, or (b) something wrongly believes it's safe to remove drivers
  206. needed to manage a signal that's in active use. That is, requesting a
  207. GPIO can serve as a kind of lock.
  208. Some platforms may also use knowledge about what GPIOs are active for
  209. power management, such as by powering down unused chip sectors and, more
  210. easily, gating off unused clocks.
  211. For GPIOs that use pins known to the pinctrl subsystem, that subsystem should
  212. be informed of their use; a gpiolib driver's .request() operation may call
  213. pinctrl_gpio_request(), and a gpiolib driver's .free() operation may call
  214. pinctrl_gpio_free(). The pinctrl subsystem allows a pinctrl_gpio_request()
  215. to succeed concurrently with a pin or pingroup being "owned" by a device for
  216. pin multiplexing.
  217. Any programming of pin multiplexing hardware that is needed to route the
  218. GPIO signal to the appropriate pin should occur within a GPIO driver's
  219. .direction_input() or .direction_output() operations, and occur after any
  220. setup of an output GPIO's value. This allows a glitch-free migration from a
  221. pin's special function to GPIO. This is sometimes required when using a GPIO
  222. to implement a workaround on signals typically driven by a non-GPIO HW block.
  223. Some platforms allow some or all GPIO signals to be routed to different pins.
  224. Similarly, other aspects of the GPIO or pin may need to be configured, such as
  225. pullup/pulldown. Platform software should arrange that any such details are
  226. configured prior to gpio_request() being called for those GPIOs, e.g. using
  227. the pinctrl subsystem's mapping table, so that GPIO users need not be aware
  228. of these details.
  229. Also note that it's your responsibility to have stopped using a GPIO
  230. before you free it.
  231. Considering in most cases GPIOs are actually configured right after they
  232. are claimed, three additional calls are defined::
  233. /* request a single GPIO, with initial configuration specified by
  234. * 'flags', identical to gpio_request() wrt other arguments and
  235. * return value
  236. */
  237. int gpio_request_one(unsigned gpio, unsigned long flags, const char *label);
  238. /* request multiple GPIOs in a single call
  239. */
  240. int gpio_request_array(struct gpio *array, size_t num);
  241. /* release multiple GPIOs in a single call
  242. */
  243. void gpio_free_array(struct gpio *array, size_t num);
  244. where 'flags' is currently defined to specify the following properties:
  245. * GPIOF_DIR_IN - to configure direction as input
  246. * GPIOF_DIR_OUT - to configure direction as output
  247. * GPIOF_INIT_LOW - as output, set initial level to LOW
  248. * GPIOF_INIT_HIGH - as output, set initial level to HIGH
  249. * GPIOF_OPEN_DRAIN - gpio pin is open drain type.
  250. * GPIOF_OPEN_SOURCE - gpio pin is open source type.
  251. * GPIOF_EXPORT_DIR_FIXED - export gpio to sysfs, keep direction
  252. * GPIOF_EXPORT_DIR_CHANGEABLE - also export, allow changing direction
  253. since GPIOF_INIT_* are only valid when configured as output, so group valid
  254. combinations as:
  255. * GPIOF_IN - configure as input
  256. * GPIOF_OUT_INIT_LOW - configured as output, initial level LOW
  257. * GPIOF_OUT_INIT_HIGH - configured as output, initial level HIGH
  258. When setting the flag as GPIOF_OPEN_DRAIN then it will assume that pins is
  259. open drain type. Such pins will not be driven to 1 in output mode. It is
  260. require to connect pull-up on such pins. By enabling this flag, gpio lib will
  261. make the direction to input when it is asked to set value of 1 in output mode
  262. to make the pin HIGH. The pin is make to LOW by driving value 0 in output mode.
  263. When setting the flag as GPIOF_OPEN_SOURCE then it will assume that pins is
  264. open source type. Such pins will not be driven to 0 in output mode. It is
  265. require to connect pull-down on such pin. By enabling this flag, gpio lib will
  266. make the direction to input when it is asked to set value of 0 in output mode
  267. to make the pin LOW. The pin is make to HIGH by driving value 1 in output mode.
  268. In the future, these flags can be extended to support more properties.
  269. Further more, to ease the claim/release of multiple GPIOs, 'struct gpio' is
  270. introduced to encapsulate all three fields as::
  271. struct gpio {
  272. unsigned gpio;
  273. unsigned long flags;
  274. const char *label;
  275. };
  276. A typical example of usage::
  277. static struct gpio leds_gpios[] = {
  278. { 32, GPIOF_OUT_INIT_HIGH, "Power LED" }, /* default to ON */
  279. { 33, GPIOF_OUT_INIT_LOW, "Green LED" }, /* default to OFF */
  280. { 34, GPIOF_OUT_INIT_LOW, "Red LED" }, /* default to OFF */
  281. { 35, GPIOF_OUT_INIT_LOW, "Blue LED" }, /* default to OFF */
  282. { ... },
  283. };
  284. err = gpio_request_one(31, GPIOF_IN, "Reset Button");
  285. if (err)
  286. ...
  287. err = gpio_request_array(leds_gpios, ARRAY_SIZE(leds_gpios));
  288. if (err)
  289. ...
  290. gpio_free_array(leds_gpios, ARRAY_SIZE(leds_gpios));
  291. GPIOs mapped to IRQs
  292. --------------------
  293. GPIO numbers are unsigned integers; so are IRQ numbers. These make up
  294. two logically distinct namespaces (GPIO 0 need not use IRQ 0). You can
  295. map between them using calls like::
  296. /* map GPIO numbers to IRQ numbers */
  297. int gpio_to_irq(unsigned gpio);
  298. /* map IRQ numbers to GPIO numbers (avoid using this) */
  299. int irq_to_gpio(unsigned irq);
  300. Those return either the corresponding number in the other namespace, or
  301. else a negative errno code if the mapping can't be done. (For example,
  302. some GPIOs can't be used as IRQs.) It is an unchecked error to use a GPIO
  303. number that wasn't set up as an input using gpio_direction_input(), or
  304. to use an IRQ number that didn't originally come from gpio_to_irq().
  305. These two mapping calls are expected to cost on the order of a single
  306. addition or subtraction. They're not allowed to sleep.
  307. Non-error values returned from gpio_to_irq() can be passed to request_irq()
  308. or free_irq(). They will often be stored into IRQ resources for platform
  309. devices, by the board-specific initialization code. Note that IRQ trigger
  310. options are part of the IRQ interface, e.g. IRQF_TRIGGER_FALLING, as are
  311. system wakeup capabilities.
  312. Non-error values returned from irq_to_gpio() would most commonly be used
  313. with gpio_get_value(), for example to initialize or update driver state
  314. when the IRQ is edge-triggered. Note that some platforms don't support
  315. this reverse mapping, so you should avoid using it.
  316. Emulating Open Drain Signals
  317. ----------------------------
  318. Sometimes shared signals need to use "open drain" signaling, where only the
  319. low signal level is actually driven. (That term applies to CMOS transistors;
  320. "open collector" is used for TTL.) A pullup resistor causes the high signal
  321. level. This is sometimes called a "wire-AND"; or more practically, from the
  322. negative logic (low=true) perspective this is a "wire-OR".
  323. One common example of an open drain signal is a shared active-low IRQ line.
  324. Also, bidirectional data bus signals sometimes use open drain signals.
  325. Some GPIO controllers directly support open drain outputs; many don't. When
  326. you need open drain signaling but your hardware doesn't directly support it,
  327. there's a common idiom you can use to emulate it with any GPIO pin that can
  328. be used as either an input or an output:
  329. LOW: gpio_direction_output(gpio, 0) ... this drives the signal
  330. and overrides the pullup.
  331. HIGH: gpio_direction_input(gpio) ... this turns off the output,
  332. so the pullup (or some other device) controls the signal.
  333. If you are "driving" the signal high but gpio_get_value(gpio) reports a low
  334. value (after the appropriate rise time passes), you know some other component
  335. is driving the shared signal low. That's not necessarily an error. As one
  336. common example, that's how I2C clocks are stretched: a slave that needs a
  337. slower clock delays the rising edge of SCK, and the I2C master adjusts its
  338. signaling rate accordingly.
  339. GPIO controllers and the pinctrl subsystem
  340. ------------------------------------------
  341. A GPIO controller on a SOC might be tightly coupled with the pinctrl
  342. subsystem, in the sense that the pins can be used by other functions
  343. together with an optional gpio feature. We have already covered the
  344. case where e.g. a GPIO controller need to reserve a pin or set the
  345. direction of a pin by calling any of::
  346. pinctrl_gpio_request()
  347. pinctrl_gpio_free()
  348. pinctrl_gpio_direction_input()
  349. pinctrl_gpio_direction_output()
  350. But how does the pin control subsystem cross-correlate the GPIO
  351. numbers (which are a global business) to a certain pin on a certain
  352. pin controller?
  353. This is done by registering "ranges" of pins, which are essentially
  354. cross-reference tables. These are described in
  355. Documentation/driver-api/pinctl.rst
  356. While the pin allocation is totally managed by the pinctrl subsystem,
  357. gpio (under gpiolib) is still maintained by gpio drivers. It may happen
  358. that different pin ranges in a SoC is managed by different gpio drivers.
  359. This makes it logical to let gpio drivers announce their pin ranges to
  360. the pin ctrl subsystem before it will call 'pinctrl_gpio_request' in order
  361. to request the corresponding pin to be prepared by the pinctrl subsystem
  362. before any gpio usage.
  363. For this, the gpio controller can register its pin range with pinctrl
  364. subsystem. There are two ways of doing it currently: with or without DT.
  365. For with DT support refer to Documentation/devicetree/bindings/gpio/gpio.txt.
  366. For non-DT support, user can call gpiochip_add_pin_range() with appropriate
  367. parameters to register a range of gpio pins with a pinctrl driver. For this
  368. exact name string of pinctrl device has to be passed as one of the
  369. argument to this routine.
  370. What do these conventions omit?
  371. ===============================
  372. One of the biggest things these conventions omit is pin multiplexing, since
  373. this is highly chip-specific and nonportable. One platform might not need
  374. explicit multiplexing; another might have just two options for use of any
  375. given pin; another might have eight options per pin; another might be able
  376. to route a given GPIO to any one of several pins. (Yes, those examples all
  377. come from systems that run Linux today.)
  378. Related to multiplexing is configuration and enabling of the pullups or
  379. pulldowns integrated on some platforms. Not all platforms support them,
  380. or support them in the same way; and any given board might use external
  381. pullups (or pulldowns) so that the on-chip ones should not be used.
  382. (When a circuit needs 5 kOhm, on-chip 100 kOhm resistors won't do.)
  383. Likewise drive strength (2 mA vs 20 mA) and voltage (1.8V vs 3.3V) is a
  384. platform-specific issue, as are models like (not) having a one-to-one
  385. correspondence between configurable pins and GPIOs.
  386. There are other system-specific mechanisms that are not specified here,
  387. like the aforementioned options for input de-glitching and wire-OR output.
  388. Hardware may support reading or writing GPIOs in gangs, but that's usually
  389. configuration dependent: for GPIOs sharing the same bank. (GPIOs are
  390. commonly grouped in banks of 16 or 32, with a given SOC having several such
  391. banks.) Some systems can trigger IRQs from output GPIOs, or read values
  392. from pins not managed as GPIOs. Code relying on such mechanisms will
  393. necessarily be nonportable.
  394. Dynamic definition of GPIOs is not currently standard; for example, as
  395. a side effect of configuring an add-on board with some GPIO expanders.
  396. GPIO implementor's framework (OPTIONAL)
  397. =======================================
  398. As noted earlier, there is an optional implementation framework making it
  399. easier for platforms to support different kinds of GPIO controller using
  400. the same programming interface. This framework is called "gpiolib".
  401. As a debugging aid, if debugfs is available a /sys/kernel/debug/gpio file
  402. will be found there. That will list all the controllers registered through
  403. this framework, and the state of the GPIOs currently in use.
  404. Controller Drivers: gpio_chip
  405. -----------------------------
  406. In this framework each GPIO controller is packaged as a "struct gpio_chip"
  407. with information common to each controller of that type:
  408. - methods to establish GPIO direction
  409. - methods used to access GPIO values
  410. - flag saying whether calls to its methods may sleep
  411. - optional debugfs dump method (showing extra state like pullup config)
  412. - label for diagnostics
  413. There is also per-instance data, which may come from device.platform_data:
  414. the number of its first GPIO, and how many GPIOs it exposes.
  415. The code implementing a gpio_chip should support multiple instances of the
  416. controller, possibly using the driver model. That code will configure each
  417. gpio_chip and issue gpiochip_add(). Removing a GPIO controller should be
  418. rare; use gpiochip_remove() when it is unavoidable.
  419. Most often a gpio_chip is part of an instance-specific structure with state
  420. not exposed by the GPIO interfaces, such as addressing, power management,
  421. and more. Chips such as codecs will have complex non-GPIO state.
  422. Any debugfs dump method should normally ignore signals which haven't been
  423. requested as GPIOs. They can use gpiochip_is_requested(), which returns
  424. either NULL or the label associated with that GPIO when it was requested.
  425. Platform Support
  426. ----------------
  427. To force-enable this framework, a platform's Kconfig will "select" GPIOLIB,
  428. else it is up to the user to configure support for GPIO.
  429. It may also provide a custom value for ARCH_NR_GPIOS, so that it better
  430. reflects the number of GPIOs in actual use on that platform, without
  431. wasting static table space. (It should count both built-in/SoC GPIOs and
  432. also ones on GPIO expanders.
  433. If neither of these options are selected, the platform does not support
  434. GPIOs through GPIO-lib and the code cannot be enabled by the user.
  435. Trivial implementations of those functions can directly use framework
  436. code, which always dispatches through the gpio_chip::
  437. #define gpio_get_value __gpio_get_value
  438. #define gpio_set_value __gpio_set_value
  439. #define gpio_cansleep __gpio_cansleep
  440. Fancier implementations could instead define those as inline functions with
  441. logic optimizing access to specific SOC-based GPIOs. For example, if the
  442. referenced GPIO is the constant "12", getting or setting its value could
  443. cost as little as two or three instructions, never sleeping. When such an
  444. optimization is not possible those calls must delegate to the framework
  445. code, costing at least a few dozen instructions. For bitbanged I/O, such
  446. instruction savings can be significant.
  447. For SOCs, platform-specific code defines and registers gpio_chip instances
  448. for each bank of on-chip GPIOs. Those GPIOs should be numbered/labeled to
  449. match chip vendor documentation, and directly match board schematics. They
  450. may well start at zero and go up to a platform-specific limit. Such GPIOs
  451. are normally integrated into platform initialization to make them always be
  452. available, from arch_initcall() or earlier; they can often serve as IRQs.
  453. Board Support
  454. -------------
  455. For external GPIO controllers -- such as I2C or SPI expanders, ASICs, multi
  456. function devices, FPGAs or CPLDs -- most often board-specific code handles
  457. registering controller devices and ensures that their drivers know what GPIO
  458. numbers to use with gpiochip_add(). Their numbers often start right after
  459. platform-specific GPIOs.
  460. For example, board setup code could create structures identifying the range
  461. of GPIOs that chip will expose, and passes them to each GPIO expander chip
  462. using platform_data. Then the chip driver's probe() routine could pass that
  463. data to gpiochip_add().
  464. Initialization order can be important. For example, when a device relies on
  465. an I2C-based GPIO, its probe() routine should only be called after that GPIO
  466. becomes available. That may mean the device should not be registered until
  467. calls for that GPIO can work. One way to address such dependencies is for
  468. such gpio_chip controllers to provide setup() and teardown() callbacks to
  469. board specific code; those board specific callbacks would register devices
  470. once all the necessary resources are available, and remove them later when
  471. the GPIO controller device becomes unavailable.
  472. Sysfs Interface for Userspace (OPTIONAL)
  473. ========================================
  474. Platforms which use the "gpiolib" implementors framework may choose to
  475. configure a sysfs user interface to GPIOs. This is different from the
  476. debugfs interface, since it provides control over GPIO direction and
  477. value instead of just showing a gpio state summary. Plus, it could be
  478. present on production systems without debugging support.
  479. Given appropriate hardware documentation for the system, userspace could
  480. know for example that GPIO #23 controls the write protect line used to
  481. protect boot loader segments in flash memory. System upgrade procedures
  482. may need to temporarily remove that protection, first importing a GPIO,
  483. then changing its output state, then updating the code before re-enabling
  484. the write protection. In normal use, GPIO #23 would never be touched,
  485. and the kernel would have no need to know about it.
  486. Again depending on appropriate hardware documentation, on some systems
  487. userspace GPIO can be used to determine system configuration data that
  488. standard kernels won't know about. And for some tasks, simple userspace
  489. GPIO drivers could be all that the system really needs.
  490. Note that standard kernel drivers exist for common "LEDs and Buttons"
  491. GPIO tasks: "leds-gpio" and "gpio_keys", respectively. Use those
  492. instead of talking directly to the GPIOs; they integrate with kernel
  493. frameworks better than your userspace code could.
  494. Paths in Sysfs
  495. --------------
  496. There are three kinds of entry in /sys/class/gpio:
  497. - Control interfaces used to get userspace control over GPIOs;
  498. - GPIOs themselves; and
  499. - GPIO controllers ("gpio_chip" instances).
  500. That's in addition to standard files including the "device" symlink.
  501. The control interfaces are write-only:
  502. /sys/class/gpio/
  503. "export" ... Userspace may ask the kernel to export control of
  504. a GPIO to userspace by writing its number to this file.
  505. Example: "echo 19 > export" will create a "gpio19" node
  506. for GPIO #19, if that's not requested by kernel code.
  507. "unexport" ... Reverses the effect of exporting to userspace.
  508. Example: "echo 19 > unexport" will remove a "gpio19"
  509. node exported using the "export" file.
  510. GPIO signals have paths like /sys/class/gpio/gpio42/ (for GPIO #42)
  511. and have the following read/write attributes:
  512. /sys/class/gpio/gpioN/
  513. "direction" ... reads as either "in" or "out". This value may
  514. normally be written. Writing as "out" defaults to
  515. initializing the value as low. To ensure glitch free
  516. operation, values "low" and "high" may be written to
  517. configure the GPIO as an output with that initial value.
  518. Note that this attribute *will not exist* if the kernel
  519. doesn't support changing the direction of a GPIO, or
  520. it was exported by kernel code that didn't explicitly
  521. allow userspace to reconfigure this GPIO's direction.
  522. "value" ... reads as either 0 (low) or 1 (high). If the GPIO
  523. is configured as an output, this value may be written;
  524. any nonzero value is treated as high.
  525. If the pin can be configured as interrupt-generating interrupt
  526. and if it has been configured to generate interrupts (see the
  527. description of "edge"), you can poll(2) on that file and
  528. poll(2) will return whenever the interrupt was triggered. If
  529. you use poll(2), set the events POLLPRI. If you use select(2),
  530. set the file descriptor in exceptfds. After poll(2) returns,
  531. either lseek(2) to the beginning of the sysfs file and read the
  532. new value or close the file and re-open it to read the value.
  533. "edge" ... reads as either "none", "rising", "falling", or
  534. "both". Write these strings to select the signal edge(s)
  535. that will make poll(2) on the "value" file return.
  536. This file exists only if the pin can be configured as an
  537. interrupt generating input pin.
  538. "active_low" ... reads as either 0 (false) or 1 (true). Write
  539. any nonzero value to invert the value attribute both
  540. for reading and writing. Existing and subsequent
  541. poll(2) support configuration via the edge attribute
  542. for "rising" and "falling" edges will follow this
  543. setting.
  544. GPIO controllers have paths like /sys/class/gpio/gpiochip42/ (for the
  545. controller implementing GPIOs starting at #42) and have the following
  546. read-only attributes:
  547. /sys/class/gpio/gpiochipN/
  548. "base" ... same as N, the first GPIO managed by this chip
  549. "label" ... provided for diagnostics (not always unique)
  550. "ngpio" ... how many GPIOs this manges (N to N + ngpio - 1)
  551. Board documentation should in most cases cover what GPIOs are used for
  552. what purposes. However, those numbers are not always stable; GPIOs on
  553. a daughtercard might be different depending on the base board being used,
  554. or other cards in the stack. In such cases, you may need to use the
  555. gpiochip nodes (possibly in conjunction with schematics) to determine
  556. the correct GPIO number to use for a given signal.
  557. Exporting from Kernel code
  558. --------------------------
  559. Kernel code can explicitly manage exports of GPIOs which have already been
  560. requested using gpio_request()::
  561. /* export the GPIO to userspace */
  562. int gpio_export(unsigned gpio, bool direction_may_change);
  563. /* reverse gpio_export() */
  564. void gpio_unexport();
  565. /* create a sysfs link to an exported GPIO node */
  566. int gpio_export_link(struct device *dev, const char *name,
  567. unsigned gpio)
  568. After a kernel driver requests a GPIO, it may only be made available in
  569. the sysfs interface by gpio_export(). The driver can control whether the
  570. signal direction may change. This helps drivers prevent userspace code
  571. from accidentally clobbering important system state.
  572. This explicit exporting can help with debugging (by making some kinds
  573. of experiments easier), or can provide an always-there interface that's
  574. suitable for documenting as part of a board support package.
  575. After the GPIO has been exported, gpio_export_link() allows creating
  576. symlinks from elsewhere in sysfs to the GPIO sysfs node. Drivers can
  577. use this to provide the interface under their own device in sysfs with
  578. a descriptive name.
  579. API Reference
  580. =============
  581. The functions listed in this section are deprecated. The GPIO descriptor based
  582. API should be used in new code.
  583. .. kernel-doc:: drivers/gpio/gpiolib-legacy.c
  584. :export: