1GPIO Interfaces
   3This provides an overview of GPIO access conventions on Linux.
   5These calls use the gpio_* naming prefix.  No other calls should use that
   6prefix, or the related __gpio_* prefix.
   9What is a GPIO?
  11A "General Purpose Input/Output" (GPIO) is a flexible software-controlled
  12digital signal.  They are provided from many kinds of chip, and are familiar
  13to Linux developers working with embedded and custom hardware.  Each GPIO
  14represents a bit connected to a particular pin, or "ball" on Ball Grid Array
  15(BGA) packages.  Board schematics show which external hardware connects to
  16which GPIOs.  Drivers can be written generically, so that board setup code
  17passes such pin configuration data to drivers.
  19System-on-Chip (SOC) processors heavily rely on GPIOs.  In some cases, every
  20non-dedicated pin can be configured as a GPIO; and most chips have at least
  21several dozen of them.  Programmable logic devices (like FPGAs) can easily
  22provide GPIOs; multifunction chips like power managers, and audio codecs
  23often have a few such pins to help with pin scarcity on SOCs; and there are
  24also "GPIO Expander" chips that connect using the I2C or SPI serial busses.
  25Most PC southbridges have a few dozen GPIO-capable pins (with only the BIOS
  26firmware knowing how they're used).
  28The exact capabilities of GPIOs vary between systems.  Common options:
  30  - Output values are writable (high=1, low=0).  Some chips also have
  31    options about how that value is driven, so that for example only one
  32    value might be driven ... supporting "wire-OR" and similar schemes
  33    for the other value (notably, "open drain" signaling).
  35  - Input values are likewise readable (1, 0).  Some chips support readback
  36    of pins configured as "output", which is very useful in such "wire-OR"
  37    cases (to support bidirectional signaling).  GPIO controllers may have
  38    input de-glitch/debounce logic, sometimes with software controls.
  40  - Inputs can often be used as IRQ signals, often edge triggered but
  41    sometimes level triggered.  Such IRQs may be configurable as system
  42    wakeup events, to wake the system from a low power state.
  44  - Usually a GPIO will be configurable as either input or output, as needed
  45    by different product boards; single direction ones exist too.
  47  - Most GPIOs can be accessed while holding spinlocks, but those accessed
  48    through a serial bus normally can't.  Some systems support both types.
  50On a given board each GPIO is used for one specific purpose like monitoring
  51MMC/SD card insertion/removal, detecting card writeprotect status, driving
  52a LED, configuring a transceiver, bitbanging a serial bus, poking a hardware
  53watchdog, sensing a switch, and so on.
  56GPIO conventions
  58Note that this is called a "convention" because you don't need to do it this
  59way, and it's no crime if you don't.  There **are** cases where portability
  60is not the main issue; GPIOs are often used for the kind of board-specific
  61glue logic that may even change between board revisions, and can't ever be
  62used on a board that's wired differently.  Only least-common-denominator
  63functionality can be very portable.  Other features are platform-specific,
  64and that can be critical for glue logic.
  66Plus, this doesn't require any implementation framework, just an interface.
  67One platform might implement it as simple inline functions accessing chip
  68registers; another might implement it by delegating through abstractions
  69used for several very different kinds of GPIO controller.  (There is some
  70optional code supporting such an implementation strategy, described later
  71in this document, but drivers acting as clients to the GPIO interface must
  72not care how it's implemented.)
  74That said, if the convention is supported on their platform, drivers should
  75use it when possible.  Platforms must declare GENERIC_GPIO support in their
  76Kconfig (boolean true), and provide an <asm/gpio.h> file.  Drivers that can't
  77work without standard GPIO calls should have Kconfig entries which depend
  78on GENERIC_GPIO.  The GPIO calls are available, either as "real code" or as
  79optimized-away stubs, when drivers use the include file:
  81        #include <linux/gpio.h>
  83If you stick to this convention then it'll be easier for other developers to
  84see what your code is doing, and help maintain it.
  86Note that these operations include I/O barriers on platforms which need to
  87use them; drivers don't need to add them explicitly.
  90Identifying GPIOs
  92GPIOs are identified by unsigned integers in the range 0..MAX_INT.  That
  93reserves "negative" numbers for other purposes like marking signals as
  94"not available on this board", or indicating faults.  Code that doesn't
  95touch the underlying hardware treats these integers as opaque cookies.
  97Platforms define how they use those integers, and usually #define symbols
  98for the GPIO lines so that board-specific setup code directly corresponds
  99to the relevant schematics.  In contrast, drivers should only use GPIO
 100numbers passed to them from that setup code, using platform_data to hold
 101board-specific pin configuration data (along with other board specific
 102data they need).  That avoids portability problems.
 104So for example one platform uses numbers 32-159 for GPIOs; while another
 105uses numbers 0..63 with one set of GPIO controllers, 64-79 with another
 106type of GPIO controller, and on one particular board 80-95 with an FPGA.
 107The numbers need not be contiguous; either of those platforms could also
 108use numbers 2000-2063 to identify GPIOs in a bank of I2C GPIO expanders.
 110If you want to initialize a structure with an invalid GPIO number, use
 111some negative number (perhaps "-EINVAL"); that will never be valid.  To
 112test if a number could reference a GPIO, you may use this predicate:
 114        int gpio_is_valid(int number);
 116A number that's not valid will be rejected by calls which may request
 117or free GPIOs (see below).  Other numbers may also be rejected; for
 118example, a number might be valid but unused on a given board.
 120Whether a platform supports multiple GPIO controllers is currently a
 121platform-specific implementation issue.
 124Using GPIOs
 126One of the first things to do with a GPIO, often in board setup code when
 127setting up a platform_device using the GPIO, is mark its direction:
 129        /* set as input or output, returning 0 or negative errno */
 130        int gpio_direction_input(unsigned gpio);
 131        int gpio_direction_output(unsigned gpio, int value);
 133The return value is zero for success, else a negative errno.  It should
 134be checked, since the get/set calls don't have error returns and since
 135misconfiguration is possible.  You should normally issue these calls from
 136a task context.  However, for spinlock-safe GPIOs it's OK to use them
 137before tasking is enabled, as part of early board setup.
 139For output GPIOs, the value provided becomes the initial output value.
 140This helps avoid signal glitching during system startup.
 142For compatibility with legacy interfaces to GPIOs, setting the direction
 143of a GPIO implicitly requests that GPIO (see below) if it has not been
 144requested already.  That compatibility may be removed in the future;
 145explicitly requesting GPIOs is strongly preferred.
 147Setting the direction can fail if the GPIO number is invalid, or when
 148that particular GPIO can't be used in that mode.  It's generally a bad
 149idea to rely on boot firmware to have set the direction correctly, since
 150it probably wasn't validated to do more than boot Linux.  (Similarly,
 151that board setup code probably needs to multiplex that pin as a GPIO,
 152and configure pullups/pulldowns appropriately.)
 155Spinlock-Safe GPIO access
 157Most GPIO controllers can be accessed with memory read/write instructions.
 158That doesn't need to sleep, and can safely be done from inside IRQ handlers.
 159(That includes hardirq contexts on RT kernels.)
 161Use these calls to access such GPIOs:
 163        /* GPIO INPUT:  return zero or nonzero */
 164        int gpio_get_value(unsigned gpio);
 166        /* GPIO OUTPUT */
 167        void gpio_set_value(unsigned gpio, int value);
 169The values are boolean, zero for low, nonzero for high.  When reading the
 170value of an output pin, the value returned should be what's seen on the
 171pin ... that won't always match the specified output value, because of
 172issues including open-drain signaling and output latencies.
 174The get/set calls have no error returns because "invalid GPIO" should have
 175been reported earlier from gpio_direction_*().  However, note that not all
 176platforms can read the value of output pins; those that can't should always
 177return zero.  Also, using these calls for GPIOs that can't safely be accessed
 178without sleeping (see below) is an error.
 180Platform-specific implementations are encouraged to optimize the two
 181calls to access the GPIO value in cases where the GPIO number (and for
 182output, value) are constant.  It's normal for them to need only a couple
 183of instructions in such cases (reading or writing a hardware register),
 184and not to need spinlocks.  Such optimized calls can make bitbanging
 185applications a lot more efficient (in both space and time) than spending
 186dozens of instructions on subroutine calls.
 189GPIO access that may sleep
 191Some GPIO controllers must be accessed using message based busses like I2C
 192or SPI.  Commands to read or write those GPIO values require waiting to
 193get to the head of a queue to transmit a command and get its response.
 194This requires sleeping, which can't be done from inside IRQ handlers.
 196Platforms that support this type of GPIO distinguish them from other GPIOs
 197by returning nonzero from this call (which requires a valid GPIO number,
 198either explicitly or implicitly requested):
 200        int gpio_cansleep(unsigned gpio);
 202To access such GPIOs, a different set of accessors is defined:
 204        /* GPIO INPUT:  return zero or nonzero, might sleep */
 205        int gpio_get_value_cansleep(unsigned gpio);
 207        /* GPIO OUTPUT, might sleep */
 208        void gpio_set_value_cansleep(unsigned gpio, int value);
 210Other than the fact that these calls might sleep, and will not be ignored
 211for GPIOs that can't be accessed from IRQ handlers, these calls act the
 212same as the spinlock-safe calls.
 215Claiming and Releasing GPIOs (OPTIONAL)
 217To help catch system configuration errors, two calls are defined.
 218However, many platforms don't currently support this mechanism.
 220        /* request GPIO, returning 0 or negative errno.
 221         * non-null labels may be useful for diagnostics.
 222         */
 223        int gpio_request(unsigned gpio, const char *label);
 225        /* release previously-claimed GPIO */
 226        void gpio_free(unsigned gpio);
 228Passing invalid GPIO numbers to gpio_request() will fail, as will requesting
 229GPIOs that have already been claimed with that call.  The return value of
 230gpio_request() must be checked.  You should normally issue these calls from
 231a task context.  However, for spinlock-safe GPIOs it's OK to request GPIOs
 232before tasking is enabled, as part of early board setup.
 234These calls serve two basic purposes.  One is marking the signals which
 235are actually in use as GPIOs, for better diagnostics; systems may have
 236several hundred potential GPIOs, but often only a dozen are used on any
 237given board.  Another is to catch conflicts, identifying errors when
 238(a) two or more drivers wrongly think they have exclusive use of that
 239signal, or (b) something wrongly believes it's safe to remove drivers
 240needed to manage a signal that's in active use.  That is, requesting a
 241GPIO can serve as a kind of lock.
 243Some platforms may also use knowledge about what GPIOs are active for
 244power management, such as by powering down unused chip sectors and, more
 245easily, gating off unused clocks.
 247These two calls are optional because not not all current Linux platforms
 248offer such functionality in their GPIO support; a valid implementation
 249could return success for all gpio_request() calls.  Unlike the other calls,
 250the state they represent doesn't normally match anything from a hardware
 251register; it's just a software bitmap which clearly is not necessary for
 252correct operation of hardware or (bug free) drivers.
 254Note that requesting a GPIO does NOT cause it to be configured in any
 255way; it just marks that GPIO as in use.  Separate code must handle any
 256pin setup (e.g. controlling which pin the GPIO uses, pullup/pulldown).
 258Also note that it's your responsibility to have stopped using a GPIO
 259before you free it.
 262GPIOs mapped to IRQs
 264GPIO numbers are unsigned integers; so are IRQ numbers.  These make up
 265two logically distinct namespaces (GPIO 0 need not use IRQ 0).  You can
 266map between them using calls like:
 268        /* map GPIO numbers to IRQ numbers */
 269        int gpio_to_irq(unsigned gpio);
 271        /* map IRQ numbers to GPIO numbers (avoid using this) */
 272        int irq_to_gpio(unsigned irq);
 274Those return either the corresponding number in the other namespace, or
 275else a negative errno code if the mapping can't be done.  (For example,
 276some GPIOs can't be used as IRQs.)  It is an unchecked error to use a GPIO
 277number that wasn't set up as an input using gpio_direction_input(), or
 278to use an IRQ number that didn't originally come from gpio_to_irq().
 280These two mapping calls are expected to cost on the order of a single
 281addition or subtraction.  They're not allowed to sleep.
 283Non-error values returned from gpio_to_irq() can be passed to request_irq()
 284or free_irq().  They will often be stored into IRQ resources for platform
 285devices, by the board-specific initialization code.  Note that IRQ trigger
 286options are part of the IRQ interface, e.g. IRQF_TRIGGER_FALLING, as are
 287system wakeup capabilities.
 289Non-error values returned from irq_to_gpio() would most commonly be used
 290with gpio_get_value(), for example to initialize or update driver state
 291when the IRQ is edge-triggered.  Note that some platforms don't support
 292this reverse mapping, so you should avoid using it.
 295Emulating Open Drain Signals
 297Sometimes shared signals need to use "open drain" signaling, where only the
 298low signal level is actually driven.  (That term applies to CMOS transistors;
 299"open collector" is used for TTL.)  A pullup resistor causes the high signal
 300level.  This is sometimes called a "wire-AND"; or more practically, from the
 301negative logic (low=true) perspective this is a "wire-OR".
 303One common example of an open drain signal is a shared active-low IRQ line.
 304Also, bidirectional data bus signals sometimes use open drain signals.
 306Some GPIO controllers directly support open drain outputs; many don't.  When
 307you need open drain signaling but your hardware doesn't directly support it,
 308there's a common idiom you can use to emulate it with any GPIO pin that can
 309be used as either an input or an output:
 311 LOW:   gpio_direction_output(gpio, 0) ... this drives the signal
 312        and overrides the pullup.
 314 HIGH:  gpio_direction_input(gpio) ... this turns off the output,
 315        so the pullup (or some other device) controls the signal.
 317If you are "driving" the signal high but gpio_get_value(gpio) reports a low
 318value (after the appropriate rise time passes), you know some other component
 319is driving the shared signal low.  That's not necessarily an error.  As one
 320common example, that's how I2C clocks are stretched:  a slave that needs a
 321slower clock delays the rising edge of SCK, and the I2C master adjusts its
 322signaling rate accordingly.
 325What do these conventions omit?
 327One of the biggest things these conventions omit is pin multiplexing, since
 328this is highly chip-specific and nonportable.  One platform might not need
 329explicit multiplexing; another might have just two options for use of any
 330given pin; another might have eight options per pin; another might be able
 331to route a given GPIO to any one of several pins.  (Yes, those examples all
 332come from systems that run Linux today.)
 334Related to multiplexing is configuration and enabling of the pullups or
 335pulldowns integrated on some platforms.  Not all platforms support them,
 336or support them in the same way; and any given board might use external
 337pullups (or pulldowns) so that the on-chip ones should not be used.
 338(When a circuit needs 5 kOhm, on-chip 100 kOhm resistors won't do.)
 339Likewise drive strength (2 mA vs 20 mA) and voltage (1.8V vs 3.3V) is a
 340platform-specific issue, as are models like (not) having a one-to-one
 341correspondence between configurable pins and GPIOs.
 343There are other system-specific mechanisms that are not specified here,
 344like the aforementioned options for input de-glitching and wire-OR output.
 345Hardware may support reading or writing GPIOs in gangs, but that's usually
 346configuration dependent:  for GPIOs sharing the same bank.  (GPIOs are
 347commonly grouped in banks of 16 or 32, with a given SOC having several such
 348banks.)  Some systems can trigger IRQs from output GPIOs, or read values
 349from pins not managed as GPIOs.  Code relying on such mechanisms will
 350necessarily be nonportable.
 352Dynamic definition of GPIOs is not currently standard; for example, as
 353a side effect of configuring an add-on board with some GPIO expanders.
 356GPIO implementor's framework (OPTIONAL)
 358As noted earlier, there is an optional implementation framework making it
 359easier for platforms to support different kinds of GPIO controller using
 360the same programming interface.  This framework is called "gpiolib".
 362As a debugging aid, if debugfs is available a /sys/kernel/debug/gpio file
 363will be found there.  That will list all the controllers registered through
 364this framework, and the state of the GPIOs currently in use.
 367Controller Drivers: gpio_chip
 369In this framework each GPIO controller is packaged as a "struct gpio_chip"
 370with information common to each controller of that type:
 372 - methods to establish GPIO direction
 373 - methods used to access GPIO values
 374 - flag saying whether calls to its methods may sleep
 375 - optional debugfs dump method (showing extra state like pullup config)
 376 - label for diagnostics
 378There is also per-instance data, which may come from device.platform_data:
 379the number of its first GPIO, and how many GPIOs it exposes.
 381The code implementing a gpio_chip should support multiple instances of the
 382controller, possibly using the driver model.  That code will configure each
 383gpio_chip and issue gpiochip_add().  Removing a GPIO controller should be
 384rare; use gpiochip_remove() when it is unavoidable.
 386Most often a gpio_chip is part of an instance-specific structure with state
 387not exposed by the GPIO interfaces, such as addressing, power management,
 388and more.  Chips such as codecs will have complex non-GPIO state,
 390Any debugfs dump method should normally ignore signals which haven't been
 391requested as GPIOs.  They can use gpiochip_is_requested(), which returns
 392either NULL or the label associated with that GPIO when it was requested.
 395Platform Support
 397To support this framework, a platform's Kconfig will "select" either
 399and arrange that its <asm/gpio.h> includes <asm-generic/gpio.h> and defines
 400three functions: gpio_get_value(), gpio_set_value(), and gpio_cansleep().
 401They may also want to provide a custom value for ARCH_NR_GPIOS.
 403ARCH_REQUIRE_GPIOLIB means that the gpio-lib code will always get compiled
 404into the kernel on that architecture.
 406ARCH_WANT_OPTIONAL_GPIOLIB means the gpio-lib code defaults to off and the user
 407can enable it and build it into the kernel optionally.
 409If neither of these options are selected, the platform does not support
 410GPIOs through GPIO-lib and the code cannot be enabled by the user.
 412Trivial implementations of those functions can directly use framework
 413code, which always dispatches through the gpio_chip:
 415  #define gpio_get_value        __gpio_get_value
 416  #define gpio_set_value        __gpio_set_value
 417  #define gpio_cansleep         __gpio_cansleep
 419Fancier implementations could instead define those as inline functions with
 420logic optimizing access to specific SOC-based GPIOs.  For example, if the
 421referenced GPIO is the constant "12", getting or setting its value could
 422cost as little as two or three instructions, never sleeping.  When such an
 423optimization is not possible those calls must delegate to the framework
 424code, costing at least a few dozen instructions.  For bitbanged I/O, such
 425instruction savings can be significant.
 427For SOCs, platform-specific code defines and registers gpio_chip instances
 428for each bank of on-chip GPIOs.  Those GPIOs should be numbered/labeled to
 429match chip vendor documentation, and directly match board schematics.  They
 430may well start at zero and go up to a platform-specific limit.  Such GPIOs
 431are normally integrated into platform initialization to make them always be
 432available, from arch_initcall() or earlier; they can often serve as IRQs.
 435Board Support
 437For external GPIO controllers -- such as I2C or SPI expanders, ASICs, multi
 438function devices, FPGAs or CPLDs -- most often board-specific code handles
 439registering controller devices and ensures that their drivers know what GPIO
 440numbers to use with gpiochip_add().  Their numbers often start right after
 441platform-specific GPIOs.
 443For example, board setup code could create structures identifying the range
 444of GPIOs that chip will expose, and passes them to each GPIO expander chip
 445using platform_data.  Then the chip driver's probe() routine could pass that
 446data to gpiochip_add().
 448Initialization order can be important.  For example, when a device relies on
 449an I2C-based GPIO, its probe() routine should only be called after that GPIO
 450becomes available.  That may mean the device should not be registered until
 451calls for that GPIO can work.  One way to address such dependencies is for
 452such gpio_chip controllers to provide setup() and teardown() callbacks to
 453board specific code; those board specific callbacks would register devices
 454once all the necessary resources are available, and remove them later when
 455the GPIO controller device becomes unavailable.
 458Sysfs Interface for Userspace (OPTIONAL)
 460Platforms which use the "gpiolib" implementors framework may choose to
 461configure a sysfs user interface to GPIOs.  This is different from the
 462debugfs interface, since it provides control over GPIO direction and
 463value instead of just showing a gpio state summary.  Plus, it could be
 464present on production systems without debugging support.
 466Given approprate hardware documentation for the system, userspace could
 467know for example that GPIO #23 controls the write protect line used to
 468protect boot loader segments in flash memory.  System upgrade procedures
 469may need to temporarily remove that protection, first importing a GPIO,
 470then changing its output state, then updating the code before re-enabling
 471the write protection.  In normal use, GPIO #23 would never be touched,
 472and the kernel would have no need to know about it.
 474Again depending on appropriate hardware documentation, on some systems
 475userspace GPIO can be used to determine system configuration data that
 476standard kernels won't know about.  And for some tasks, simple userspace
 477GPIO drivers could be all that the system really needs.
 479Note that standard kernel drivers exist for common "LEDs and Buttons"
 480GPIO tasks:  "leds-gpio" and "gpio_keys", respectively.  Use those
 481instead of talking directly to the GPIOs; they integrate with kernel
 482frameworks better than your userspace code could.
 485Paths in Sysfs
 487There are three kinds of entry in /sys/class/gpio:
 489   -    Control interfaces used to get userspace control over GPIOs;
 491   -    GPIOs themselves; and
 493   -    GPIO controllers ("gpio_chip" instances).
 495That's in addition to standard files including the "device" symlink.
 497The control interfaces are write-only:
 499    /sys/class/gpio/
 501        "export" ... Userspace may ask the kernel to export control of
 502                a GPIO to userspace by writing its number to this file.
 504                Example:  "echo 19 > export" will create a "gpio19" node
 505                for GPIO #19, if that's not requested by kernel code.
 507        "unexport" ... Reverses the effect of exporting to userspace.
 509                Example:  "echo 19 > unexport" will remove a "gpio19"
 510                node exported using the "export" file.
 512GPIO signals have paths like /sys/class/gpio/gpio42/ (for GPIO #42)
 513and have the following read/write attributes:
 515    /sys/class/gpio/gpioN/
 517        "direction" ... reads as either "in" or "out".  This value may
 518                normally be written.  Writing as "out" defaults to
 519                initializing the value as low.  To ensure glitch free
 520                operation, values "low" and "high" may be written to
 521                configure the GPIO as an output with that initial value.
 523                Note that this attribute *will not exist* if the kernel
 524                doesn't support changing the direction of a GPIO, or
 525                it was exported by kernel code that didn't explicitly
 526                allow userspace to reconfigure this GPIO's direction.
 528        "value" ... reads as either 0 (low) or 1 (high).  If the GPIO
 529                is configured as an output, this value may be written;
 530                any nonzero value is treated as high.
 532GPIO controllers have paths like /sys/class/gpio/chipchip42/ (for the
 533controller implementing GPIOs starting at #42) and have the following
 534read-only attributes:
 536    /sys/class/gpio/gpiochipN/
 538        "base" ... same as N, the first GPIO managed by this chip
 540        "label" ... provided for diagnostics (not always unique)
 542        "ngpio" ... how many GPIOs this manges (N to N + ngpio - 1)
 544Board documentation should in most cases cover what GPIOs are used for
 545what purposes.  However, those numbers are not always stable; GPIOs on
 546a daughtercard might be different depending on the base board being used,
 547or other cards in the stack.  In such cases, you may need to use the
 548gpiochip nodes (possibly in conjunction with schematics) to determine
 549the correct GPIO number to use for a given signal.
 552Exporting from Kernel code
 554Kernel code can explicitly manage exports of GPIOs which have already been
 555requested using gpio_request():
 557        /* export the GPIO to userspace */
 558        int gpio_export(unsigned gpio, bool direction_may_change);
 560        /* reverse gpio_export() */
 561        void gpio_unexport();
 563After a kernel driver requests a GPIO, it may only be made available in
 564the sysfs interface by gpio_export().  The driver can control whether the
 565signal direction may change.  This helps drivers prevent userspace code
 566from accidentally clobbering important system state.
 568This explicit exporting can help with debugging (by making some kinds
 569of experiments easier), or can provide an always-there interface that's
 570suitable for documenting as part of a board support package.