요약·해설과 원문, 전문 번역을 서로 분리했습니다. API 이름, symbol, source path는 원문 표기를 사용합니다.
1. 요약·해설
원문의 핵심 논리와 kernel programming 관점의 보충 설명입니다. 아래의 전문 번역과는 별도로 작성했습니다.
2. 영어 원문 전체
번역 기준이 된 Linux v6.18.37 원문입니다. 줄 번호는 이 버전의 파일 좌표입니다.
원문 전체 펼치기
==================================
GPIO Descriptor Consumer Interface
==================================
This document describes the consumer interface of the GPIO framework.
Guidelines for GPIOs consumers
==============================
Drivers that can't work without standard GPIO calls should have Kconfig entries
that depend on GPIOLIB or select GPIOLIB. The functions that allow a driver to
obtain and use GPIOs are available by including the following file::
#include <linux/gpio/consumer.h>
There are static inline stubs for all functions in the header file in the case
where GPIOLIB is disabled. When these stubs are called they will emit
warnings. These stubs are used for two use cases:
- Simple compile coverage with e.g. COMPILE_TEST - it does not matter that
the current platform does not enable or select GPIOLIB because we are not
going to execute the system anyway.
- Truly optional GPIOLIB support - where the driver does not really make use
of the GPIOs on certain compile-time configurations for certain systems, but
will use it under other compile-time configurations. In this case the
consumer must make sure not to call into these functions, or the user will
be met with console warnings that may be perceived as intimidating.
Combining truly optional GPIOLIB usage with calls to
``[devm_]gpiod_get_optional()`` is a *bad idea*, and will result in weird
error messages. Use the ordinary getter functions with optional GPIOLIB:
some open coding of error handling should be expected when you do this.
All the functions that work with the descriptor-based GPIO interface are
prefixed with ``gpiod_``. The ``gpio_`` prefix is used for the legacy
interface. No other function in the kernel should use these prefixes. The use
of the legacy functions is strongly discouraged, new code should use
<linux/gpio/consumer.h> and descriptors exclusively.
Obtaining and Disposing GPIOs
=============================
With the descriptor-based interface, GPIOs are identified with an opaque,
non-forgeable handler that must be obtained through a call to one of the
gpiod_get() functions. Like many other kernel subsystems, gpiod_get() takes the
device that will use the GPIO and the function the requested GPIO is supposed to
fulfill::
struct gpio_desc *gpiod_get(struct device *dev, const char *con_id,
enum gpiod_flags flags)
If a function is implemented by using several GPIOs together (e.g. a simple LED
device that displays digits), an additional index argument can be specified::
struct gpio_desc *gpiod_get_index(struct device *dev,
const char *con_id, unsigned int idx,
enum gpiod_flags flags)
For a more detailed description of the con_id parameter in the DeviceTree case
see Documentation/driver-api/gpio/board.rst
The flags parameter is used to optionally specify a direction and initial value
for the GPIO. Values can be:
* GPIOD_ASIS or 0 to not initialize the GPIO at all. The direction must be set
later with one of the dedicated functions.
* GPIOD_IN to initialize the GPIO as input.
* GPIOD_OUT_LOW to initialize the GPIO as output with a value of 0.
* GPIOD_OUT_HIGH to initialize the GPIO as output with a value of 1.
* GPIOD_OUT_LOW_OPEN_DRAIN same as GPIOD_OUT_LOW but also enforce the line
to be electrically used with open drain.
* GPIOD_OUT_HIGH_OPEN_DRAIN same as GPIOD_OUT_HIGH but also enforce the line
to be electrically used with open drain.
Note that the initial value is *logical* and the physical line level depends on
whether the line is configured active high or active low (see
:ref:`active_low_semantics`).
The two last flags are used for use cases where open drain is mandatory, such
as I2C: if the line is not already configured as open drain in the mappings
(see board.rst), then open drain will be enforced anyway and a warning will be
printed that the board configuration needs to be updated to match the use case.
Both functions return either a valid GPIO descriptor, or an error code checkable
with IS_ERR() (they will never return a NULL pointer). -ENOENT will be returned
if and only if no GPIO has been assigned to the device/function/index triplet,
other error codes are used for cases where a GPIO has been assigned but an error
occurred while trying to acquire it. This is useful to discriminate between mere
errors and an absence of GPIO for optional GPIO parameters. For the common
pattern where a GPIO is optional, the gpiod_get_optional() and
gpiod_get_index_optional() functions can be used. These functions return NULL
instead of -ENOENT if no GPIO has been assigned to the requested function::
struct gpio_desc *gpiod_get_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *gpiod_get_index_optional(struct device *dev,
const char *con_id,
unsigned int index,
enum gpiod_flags flags)
Note that gpio_get*_optional() functions (and their managed variants), unlike
the rest of gpiolib API, also return NULL when gpiolib support is disabled.
This is helpful to driver authors, since they do not need to special case
-ENOSYS return codes. System integrators should however be careful to enable
gpiolib on systems that need it.
For a function using multiple GPIOs all of those can be obtained with one call::
struct gpio_descs *gpiod_get_array(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
This function returns a struct gpio_descs which contains an array of
descriptors. It also contains a pointer to a gpiolib private structure which,
if passed back to get/set array functions, may speed up I/O processing::
struct gpio_descs {
struct gpio_array *info;
unsigned int ndescs;
struct gpio_desc *desc[];
}
The following function returns NULL instead of -ENOENT if no GPIOs have been
assigned to the requested function::
struct gpio_descs *gpiod_get_array_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
Device-managed variants of these functions are also defined::
struct gpio_desc *devm_gpiod_get(struct device *dev, const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_index(struct device *dev,
const char *con_id,
unsigned int idx,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_index_optional(struct device *dev,
const char *con_id,
unsigned int index,
enum gpiod_flags flags)
struct gpio_descs *devm_gpiod_get_array(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_descs *devm_gpiod_get_array_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
A GPIO descriptor can be disposed of using the gpiod_put() function::
void gpiod_put(struct gpio_desc *desc)
For an array of GPIOs this function can be used::
void gpiod_put_array(struct gpio_descs *descs)
It is strictly forbidden to use a descriptor after calling these functions.
It is also not allowed to individually release descriptors (using gpiod_put())
from an array acquired with gpiod_get_array().
The device-managed variants are, unsurprisingly::
void devm_gpiod_put(struct device *dev, struct gpio_desc *desc)
void devm_gpiod_put_array(struct device *dev, struct gpio_descs *descs)
Using GPIOs
===========
Setting Direction
-----------------
The first thing a driver must do with a GPIO is setting its direction. If no
direction-setting flags have been given to gpiod_get*(), this is done by
invoking one of the gpiod_direction_*() functions::
int gpiod_direction_input(struct gpio_desc *desc)
int gpiod_direction_output(struct gpio_desc *desc, int value)
The return value is zero for success, else a negative errno. It should be
checked, since the get/set calls don't return errors and since misconfiguration
is possible. You should normally issue these calls from a task context. However,
for spinlock-safe GPIOs it is OK to use them before tasking is enabled, as part
of early board setup.
For output GPIOs, the value provided becomes the initial output value. This
helps avoid signal glitching during system startup.
A driver can also query the current direction of a GPIO::
int gpiod_get_direction(const struct gpio_desc *desc)
This function returns 0 for output, 1 for input, or an error code in case of error.
Be aware that there is no default direction for GPIOs. Therefore, **using a GPIO
without setting its direction first is illegal and will result in undefined
behavior!**
Spinlock-Safe GPIO Access
-------------------------
Most GPIO controllers can be accessed with memory read/write instructions. Those
don't need to sleep, and can safely be done from inside hard (non-threaded) IRQ
handlers and similar contexts.
Use the following calls to access GPIOs from an atomic context::
int gpiod_get_value(const struct gpio_desc *desc);
void gpiod_set_value(struct gpio_desc *desc, int value);
The values are boolean, zero for inactive, nonzero for active. When reading the
value of an output pin, the value returned should be what's seen on the pin.
That won't always match the specified output value, because of issues including
open-drain signaling and output latencies.
The get/set calls do not return errors because "invalid GPIO" should have been
reported earlier from gpiod_direction_*(). However, note that not all platforms
can read the value of output pins; those that can't should always return zero.
Also, using these calls for GPIOs that can't safely be accessed without sleeping
(see below) is an error.
GPIO Access That May Sleep
--------------------------
Some GPIO controllers must be accessed using message based buses like I2C or
SPI. Commands to read or write those GPIO values require waiting to get to the
head of a queue to transmit a command and get its response. This requires
sleeping, which can't be done from inside IRQ handlers.
Platforms that support this type of GPIO distinguish them from other GPIOs by
returning nonzero from this call::
int gpiod_cansleep(const struct gpio_desc *desc)
To access such GPIOs, a different set of accessors is defined::
int gpiod_get_value_cansleep(const struct gpio_desc *desc)
void gpiod_set_value_cansleep(struct gpio_desc *desc, int value)
Accessing such GPIOs requires a context which may sleep, for example a threaded
IRQ handler, and those accessors must be used instead of spinlock-safe
accessors without the cansleep() name suffix.
Other than the fact that these accessors might sleep, and will work on GPIOs
that can't be accessed from hardIRQ handlers, these calls act the same as the
spinlock-safe calls.
.. _active_low_semantics:
The active low and open drain semantics
---------------------------------------
As a consumer should not have to care about the physical line level, all of the
gpiod_set_value_xxx() or gpiod_set_array_value_xxx() functions operate with
the *logical* value. With this they take the active low property into account.
This means that they check whether the GPIO is configured to be active low,
and if so, they manipulate the passed value before the physical line level is
driven.
The same is applicable for open drain or open source output lines: those do not
actively drive their output high (open drain) or low (open source), they just
switch their output to a high impedance value. The consumer should not need to
care. (For details read about open drain in driver.rst.)
With this, all the gpiod_set_(array)_value_xxx() functions interpret the
parameter "value" as "active" ("1") or "inactive" ("0"). The physical line
level will be driven accordingly.
As an example, if the active low property for a dedicated GPIO is set, and the
gpiod_set_(array)_value_xxx() passes "active" ("1"), the physical line level
will be driven low.
To summarize::
Function (example) line property physical line
gpiod_set_raw_value(desc, 0); don't care low
gpiod_set_raw_value(desc, 1); don't care high
gpiod_set_value(desc, 0); default (active high) low
gpiod_set_value(desc, 1); default (active high) high
gpiod_set_value(desc, 0); active low high
gpiod_set_value(desc, 1); active low low
gpiod_set_value(desc, 0); open drain low
gpiod_set_value(desc, 1); open drain high impedance
gpiod_set_value(desc, 0); open source high impedance
gpiod_set_value(desc, 1); open source high
It is possible to override these semantics using the set_raw/get_raw functions
but it should be avoided as much as possible, especially by system-agnostic drivers
which should not need to care about the actual physical line level and worry about
the logical value instead.
Accessing raw GPIO values
-------------------------
Consumers exist that need to manage the logical state of a GPIO line, i.e. the value
their device will actually receive, no matter what lies between it and the GPIO
line.
The following set of calls ignore the active-low or open drain property of a GPIO and
work on the raw line value::
int gpiod_get_raw_value(const struct gpio_desc *desc)
void gpiod_set_raw_value(struct gpio_desc *desc, int value)
int gpiod_get_raw_value_cansleep(const struct gpio_desc *desc)
void gpiod_set_raw_value_cansleep(struct gpio_desc *desc, int value)
int gpiod_direction_output_raw(struct gpio_desc *desc, int value)
The active low state of a GPIO can also be queried and toggled using the
following calls::
int gpiod_is_active_low(const struct gpio_desc *desc)
void gpiod_toggle_active_low(struct gpio_desc *desc)
Note that these functions should only be used with great moderation; a driver
should not have to care about the physical line level or open drain semantics.
Access multiple GPIOs with a single function call
-------------------------------------------------
The following functions get or set the values of an array of GPIOs::
int gpiod_get_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_raw_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_raw_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_set_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_raw_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_raw_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
The array can be an arbitrary set of GPIOs. The functions will try to access
GPIOs belonging to the same bank or chip simultaneously if supported by the
corresponding chip driver. In that case a significantly improved performance
can be expected. If simultaneous access is not possible the GPIOs will be
accessed sequentially.
The functions take four arguments:
* array_size - the number of array elements
* desc_array - an array of GPIO descriptors
* array_info - optional information obtained from gpiod_get_array()
* value_bitmap - a bitmap to store the GPIOs' values (get) or
a bitmap of values to assign to the GPIOs (set)
The descriptor array can be obtained using the gpiod_get_array() function
or one of its variants. If the group of descriptors returned by that function
matches the desired group of GPIOs, those GPIOs can be accessed by simply using
the struct gpio_descs returned by gpiod_get_array()::
struct gpio_descs *my_gpio_descs = gpiod_get_array(...);
gpiod_set_array_value(my_gpio_descs->ndescs, my_gpio_descs->desc,
my_gpio_descs->info, my_gpio_value_bitmap);
It is also possible to access a completely arbitrary array of descriptors. The
descriptors may be obtained using any combination of gpiod_get() and
gpiod_get_array(). Afterwards the array of descriptors has to be setup
manually before it can be passed to one of the above functions. In that case,
array_info should be set to NULL.
Note that for optimal performance GPIOs belonging to the same chip should be
contiguous within the array of descriptors.
Still better performance may be achieved if array indexes of the descriptors
match hardware pin numbers of a single chip. If an array passed to a get/set
array function matches the one obtained from gpiod_get_array() and array_info
associated with the array is also passed, the function may take a fast bitmap
processing path, passing the value_bitmap argument directly to the respective
.get/set_multiple() callback of the chip. That allows for utilization of GPIO
banks as data I/O ports without much loss of performance.
The return value of gpiod_get_array_value() and its variants is 0 on success
or negative on error. Note the difference to gpiod_get_value(), which returns
0 or 1 on success to convey the GPIO value. With the array functions, the GPIO
values are stored in value_array rather than passed back as return value.
GPIOs mapped to IRQs
--------------------
GPIO lines can quite often be used as IRQs. You can get the IRQ number
corresponding to a given GPIO using the following call::
int gpiod_to_irq(const struct gpio_desc *desc)
It will return an IRQ number, or a negative errno code if the mapping can't be
done (most likely because that particular GPIO cannot be used as IRQ). It is an
unchecked error to use a GPIO that wasn't set up as an input using
gpiod_direction_input(), or to use an IRQ number that didn't originally come
from gpiod_to_irq(). gpiod_to_irq() is not allowed to sleep.
Non-error values returned from gpiod_to_irq() can be passed to request_irq() or
free_irq(). They will often be stored into IRQ resources for platform devices,
by the board-specific initialization code. Note that IRQ trigger options are
part of the IRQ interface, e.g. IRQF_TRIGGER_FALLING, as are system wakeup
capabilities.
GPIOs and ACPI
==============
On ACPI systems, GPIOs are described by GpioIo()/GpioInt() resources listed by
the _CRS configuration objects of devices. Those resources do not provide
connection IDs (names) for GPIOs, so it is necessary to use an additional
mechanism for this purpose.
Systems compliant with ACPI 5.1 or newer may provide a _DSD configuration object
which, among other things, may be used to provide connection IDs for specific
GPIOs described by the GpioIo()/GpioInt() resources in _CRS. If that is the
case, it will be handled by the GPIO subsystem automatically. However, if the
_DSD is not present, the mappings between GpioIo()/GpioInt() resources and GPIO
connection IDs need to be provided by device drivers.
For details refer to Documentation/firmware-guide/acpi/gpio-properties.rst
Interacting With the Legacy GPIO Subsystem
==========================================
Many kernel subsystems and drivers still handle GPIOs using the legacy
integer-based interface. It is strongly recommended to update these to the new
gpiod interface. For cases where both interfaces need to be used, the following
two functions allow to convert a GPIO descriptor into the GPIO integer namespace
and vice-versa::
int desc_to_gpio(const struct gpio_desc *desc)
struct gpio_desc *gpio_to_desc(unsigned gpio)
The GPIO number returned by desc_to_gpio() can safely be used as a parameter of
the gpio\_*() functions for as long as the GPIO descriptor `desc` is not freed.
All the same, a GPIO number passed to gpio_to_desc() must first be properly
acquired using e.g. gpio_request_one(), and the returned GPIO descriptor is only
considered valid until that GPIO number is released using gpio_free().
Freeing a GPIO obtained by one API with the other API is forbidden and an
unchecked error.
3. 한국어 전문 번역
영어 원문의 문단 순서와 의미를 유지한 전체 번역입니다. 코드, 함수명, symbol과 URL은 원문 표기를 유지합니다.
GPIO consumer 지침
1-40문서 제목은 `GPIO Descriptor Consumer Interface`이며 GPIO framework의 consumer interface를 설명합니다.
표준 GPIO call 없이는 동작할 수 없는 driver의 Kconfig entry는 `GPIOLIB`에 depend하거나 이를 select해야 합니다. Driver가 GPIO를 얻고 사용하는 function은 source path `include/linux/gpio/consumer.h`를 include하면 사용할 수 있습니다.
#include <linux/gpio/consumer.h>
`GPIOLIB`이 disable된 경우 header의 모든 function에는 static inline stub가 있고 호출 시 warning을 냅니다. Stub는 두 경우를 지원합니다.
첫째, `COMPILE_TEST` 같은 단순 compile coverage에서는 실제 system을 실행하지 않으므로 현재 platform이 `GPIOLIB`을 enable·select하지 않아도 상관없습니다.
둘째, compile-time configuration에 따라 일부 system에서는 GPIO를 쓰지 않고 다른 system에서는 사용하는 진정한 optional `GPIOLIB` 지원입니다. 이 경우 consumer는 disabled configuration에서 stub를 호출하지 않도록 해야 하며, 그렇지 않으면 console warning이 발생합니다.
진정한 optional `GPIOLIB` 사용과 `[devm_]gpiod_get_optional()` 호출을 결합하면 이상한 error message가 생기므로 피해야 합니다. Optional GPIOLIB에서는 일반 getter와 명시적 error handling을 사용합니다.
Descriptor interface function은 `gpiod_` prefix를, legacy interface는 `gpio_` prefix를 사용합니다. Kernel의 다른 function은 이 prefix를 쓰지 않아야 합니다. 새 code는 legacy function 대신 `<linux/gpio/consumer.h>`와 descriptor만 사용해야 합니다.
필수·compile-test·진정한 optional 사용의 처리 차이입니다.
GPIO descriptor 획득과 초기 flag
41-84Descriptor interface에서 GPIO는 위조할 수 없는 opaque handler로 식별하며 `gpiod_get()` 계열을 통해 얻어야 합니다. `gpiod_get()`은 GPIO를 사용할 device와 요청한 GPIO의 function을 받습니다.
struct gpio_desc *gpiod_get(struct device *dev, const char *con_id,
enum gpiod_flags flags)
여러 GPIO가 하나의 function을 구현한다면 추가 index를 받는 `gpiod_get_index()`를 사용합니다.
struct gpio_desc *gpiod_get_index(struct device *dev,
const char *con_id, unsigned int idx,
enum gpiod_flags flags)
Device Tree에서 `con_id` parameter가 mapping되는 자세한 방식은 `Documentation/driver-api/gpio/board.rst`를 참조합니다.
`flags`는 GPIO direction과 초기 logical value를 선택적으로 지정합니다.
- `GPIOD_ASIS` 또는 0: 초기화하지 않으며 나중에 전용 function으로 direction 설정
- `GPIOD_IN`: input으로 초기화
- `GPIOD_OUT_LOW`: logical 0 output으로 초기화
- `GPIOD_OUT_HIGH`: logical 1 output으로 초기화
- `GPIOD_OUT_LOW_OPEN_DRAIN`: logical 0 output과 open-drain electrical mode 강제
- `GPIOD_OUT_HIGH_OPEN_DRAIN`: logical 1 output과 open-drain electrical mode 강제
초기값은 logical value이며 실제 line level은 active-high 또는 active-low 설정에 따라 달라집니다.
마지막 두 flag는 I2C처럼 open drain이 필수인 경우에 사용합니다. Mapping에서 open drain으로 구성하지 않았더라도 mode를 강제하고 board configuration을 고쳐야 한다는 warning을 출력합니다.
초기 direction·logical value·electrical mode를 정리했습니다.
Optional GPIO와 return semantics
85-110`gpiod_get()`과 `gpiod_get_index()`는 valid descriptor 또는 `IS_ERR()`로 검사할 error code를 반환하며 `NULL`은 반환하지 않습니다.
Device/function/index 조합에 GPIO가 전혀 할당되지 않았을 때만 `-ENOENT`를 반환합니다. GPIO가 할당됐지만 획득 중 실패한 경우에는 다른 error code를 사용하므로 optional parameter의 부재와 실제 error를 구분할 수 있습니다.
Optional GPIO에는 `gpiod_get_optional()`과 `gpiod_get_index_optional()`을 사용할 수 있으며 mapping이 없으면 `-ENOENT` 대신 `NULL`을 반환합니다.
struct gpio_desc *gpiod_get_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *gpiod_get_index_optional(struct device *dev,
const char *con_id,
unsigned int index,
enum gpiod_flags flags)
`gpio_get*_optional()` function과 managed variant는 나머지 gpiolib API와 달리 gpiolib support가 disable된 경우에도 `NULL`을 반환합니다. Driver는 `-ENOSYS`를 따로 처리할 필요가 없지만 system integrator는 필요한 system에서 gpiolib을 enable해야 합니다.
일반·optional getter의 mapping 부재와 실제 error 반환입니다.
GPIO array 획득·해제와 devm
111-179하나의 function이 여러 GPIO를 사용하면 `gpiod_get_array()` 한 번으로 모두 얻을 수 있습니다.
struct gpio_descs *gpiod_get_array(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
반환되는 `struct gpio_descs`에는 descriptor array와 I/O를 빠르게 할 수 있는 gpiolib private `gpio_array` pointer가 들어 있습니다.
struct gpio_descs {
struct gpio_array *info;
unsigned int ndescs;
struct gpio_desc *desc[];
}
Mapping이 없을 때 `NULL`을 반환하는 array optional variant도 있습니다.
struct gpio_descs *gpiod_get_array_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
Single·indexed·optional·array getter에는 device-managed variant가 모두 정의됩니다.
struct gpio_desc *devm_gpiod_get(struct device *dev, const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_index(struct device *dev,
const char *con_id,
unsigned int idx,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_desc *devm_gpiod_get_index_optional(struct device *dev,
const char *con_id,
unsigned int index,
enum gpiod_flags flags)
struct gpio_descs *devm_gpiod_get_array(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
struct gpio_descs *devm_gpiod_get_array_optional(struct device *dev,
const char *con_id,
enum gpiod_flags flags)
Single descriptor는 `gpiod_put()`으로 해제합니다.
void gpiod_put(struct gpio_desc *desc)
Array는 `gpiod_put_array()`로 전체를 해제합니다.
void gpiod_put_array(struct gpio_descs *descs)
해제한 descriptor는 절대 다시 사용하면 안 됩니다. `gpiod_get_array()`로 얻은 array의 개별 descriptor를 `gpiod_put()`으로 따로 해제하는 것도 허용되지 않습니다.
Device-managed 해제 variant는 다음과 같습니다.
void devm_gpiod_put(struct device *dev, struct gpio_desc *desc)
void devm_gpiod_put_array(struct device *dev, struct gpio_descs *descs)
일반·array·devm descriptor의 획득과 해제 규칙입니다.
GPIO direction 설정
180-211Driver가 GPIO에서 가장 먼저 해야 할 일은 direction 설정입니다. `gpiod_get*()`에 direction flag를 주지 않았다면 `gpiod_direction_*()` function을 호출합니다.
int gpiod_direction_input(struct gpio_desc *desc)
int gpiod_direction_output(struct gpio_desc *desc, int value)
성공하면 0, 실패하면 negative errno를 반환합니다. Get/set value call은 error를 반환하지 않고 misconfiguration 가능성이 있으므로 direction call의 return을 반드시 검사해야 합니다.
보통 task context에서 호출합니다. 다만 spinlock-safe GPIO라면 early board setup에서 tasking enable 전에도 사용할 수 있습니다.
Output GPIO에 전달한 value는 초기 output value가 되어 system startup 중 signal glitch를 방지합니다.
현재 direction은 다음 function으로 조회합니다.
int gpiod_get_direction(const struct gpio_desc *desc)
`gpiod_get_direction()`은 output이면 0, input이면 1, 실패하면 error code를 반환합니다. GPIO에는 default direction이 없으므로 먼저 direction을 설정하지 않고 사용하는 것은 illegal이며 undefined behavior입니다.
Direction 설정·조회와 return semantics입니다.
Spinlock-safe GPIO access
212-234대부분의 GPIO controller는 sleep이 필요 없는 memory read/write instruction으로 접근할 수 있어 hard non-threaded IRQ handler 같은 context에서도 안전합니다.
Atomic context에서는 다음 call을 사용합니다.
int gpiod_get_value(const struct gpio_desc *desc);
void gpiod_set_value(struct gpio_desc *desc, int value);
Value는 boolean이며 0은 inactive, nonzero는 active입니다. Output pin을 읽으면 pin에서 실제로 보이는 값을 반환해야 하므로 open-drain signaling이나 output latency 때문에 지정한 output value와 다를 수 있습니다.
Invalid GPIO는 앞선 `gpiod_direction_*()`에서 보고되어야 하므로 get/set call은 error를 반환하지 않습니다. Output read를 지원하지 않는 platform은 항상 0을 반환해야 합니다. Sleep 없이 안전하게 접근할 수 없는 GPIO에 이 call을 사용하는 것은 error입니다.
Hard IRQ에서도 가능한 accessor와 제한입니다.
Sleep 가능한 GPIO access
235-260일부 GPIO controller는 I2C나 SPI 같은 message bus를 사용합니다. Command 전송과 response를 기다리려면 sleep해야 하므로 IRQ handler 안에서는 접근할 수 없습니다.
이런 GPIO는 다음 function이 nonzero를 반환해 구분합니다.
int gpiod_cansleep(const struct gpio_desc *desc)
Sleep이 필요한 GPIO에는 다음 accessor를 사용합니다.
int gpiod_get_value_cansleep(const struct gpio_desc *desc)
void gpiod_set_value_cansleep(struct gpio_desc *desc, int value)
Threaded IRQ handler처럼 sleep할 수 있는 context가 필요하며 `cansleep()` suffix가 없는 spinlock-safe accessor 대신 이 function을 사용해야 합니다. Sleep 가능성과 hardIRQ 사용 불가를 제외하면 동작은 spinlock-safe call과 같습니다.
Controller 접근 방식과 실행 context에 따른 API입니다.
Active-low·open-drain logical semantics
261-304Consumer가 실제 physical line level을 신경 쓰지 않도록 모든 `gpiod_set_value_xxx()`와 `gpiod_set_array_value_xxx()`는 logical value를 사용합니다. GPIO가 active low이면 전달된 value를 변환한 뒤 physical line을 구동합니다.
Open drain output은 high를 적극 구동하지 않고, open source output은 low를 적극 구동하지 않으며 대신 high-impedance 상태로 전환합니다. 이 electrical detail 역시 consumer가 처리할 필요가 없습니다. 자세한 내용은 `driver.rst`의 open-drain 설명을 참조합니다.
따라서 `gpiod_set_(array)_value_xxx()`의 `value`는 1이면 active, 0이면 inactive를 뜻합니다. 예를 들어 active-low GPIO에 active 1을 전달하면 실제 line level은 low가 됩니다.
원문의 active-low·open-drain/source ASCII 표를 구조화했습니다.
`set_raw/get_raw` function으로 이 semantics를 override할 수 있지만 가능한 한 피해야 합니다. 특히 system-agnostic driver는 실제 physical level이 아니라 logical value에 집중해야 합니다.
Raw GPIO value access
305-329일부 consumer는 GPIO line과 device 사이의 요소와 무관하게 device가 실제로 받을 logical state를 직접 관리해야 합니다.
다음 call은 active-low나 open-drain property를 무시하고 raw line value에 작동합니다.
int gpiod_get_raw_value(const struct gpio_desc *desc)
void gpiod_set_raw_value(struct gpio_desc *desc, int value)
int gpiod_get_raw_value_cansleep(const struct gpio_desc *desc)
void gpiod_set_raw_value_cansleep(struct gpio_desc *desc, int value)
int gpiod_direction_output_raw(struct gpio_desc *desc, int value)
GPIO의 active-low state도 다음 call로 조회하고 toggle할 수 있습니다.
int gpiod_is_active_low(const struct gpio_desc *desc)
void gpiod_toggle_active_low(struct gpio_desc *desc)
이 function은 매우 제한적으로 사용해야 합니다. 일반 driver는 physical line level이나 open-drain semantics를 직접 다룰 필요가 없어야 합니다.
Mapping semantics 적용 여부를 비교합니다.
여러 GPIO 동시 access
330-413다음 function은 GPIO array의 value를 get 또는 set합니다.
int gpiod_get_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_raw_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_get_raw_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap);
int gpiod_set_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_raw_array_value(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
int gpiod_set_raw_array_value_cansleep(unsigned int array_size,
struct gpio_desc **desc_array,
struct gpio_array *array_info,
unsigned long *value_bitmap)
Array는 임의의 GPIO 집합일 수 있습니다. Chip driver가 지원하면 같은 bank나 chip의 GPIO를 동시에 접근해 성능을 크게 높이고, 지원하지 않으면 순차 접근합니다.
Function은 네 argument를 받습니다.
- `array_size`: array element 수
- `desc_array`: GPIO descriptor array
- `array_info`: `gpiod_get_array()`에서 얻은 optional 정보
- `value_bitmap`: get 결과를 저장하거나 set 값을 제공하는 bitmap
`gpiod_get_array()`가 반환한 descriptor group이 원하는 GPIO 집합과 같다면 `struct gpio_descs`를 그대로 사용할 수 있습니다.
struct gpio_descs *my_gpio_descs = gpiod_get_array(...);
gpiod_set_array_value(my_gpio_descs->ndescs, my_gpio_descs->desc,
my_gpio_descs->info, my_gpio_value_bitmap);
`gpiod_get()`과 `gpiod_get_array()`를 조합해 완전히 임의의 descriptor array도 만들 수 있습니다. 이 경우 array를 수동으로 구성하고 `array_info`는 `NULL`로 둡니다.
최적 성능을 위해 같은 chip의 GPIO는 descriptor array 안에서 연속해야 합니다. Descriptor index가 한 chip의 hardware pin number와 일치하고 array와 `array_info`가 `gpiod_get_array()` 결과와 같으면 `value_bitmap`을 chip의 `.get/set_multiple()` callback에 직접 전달하는 fast bitmap path를 사용할 수 있습니다.
`gpiod_get_array_value()` 계열은 성공 시 0, 실패 시 negative error를 반환합니다. 성공 시 GPIO value 0 또는 1을 return하는 `gpiod_get_value()`와 달리 array value는 `value_bitmap`에 저장됩니다.
Array get/set에 공통인 네 argument입니다.
Sequential access에서 fast bitmap processing까지의 선택입니다.
GPIO를 IRQ로 mapping
414-433GPIO line은 흔히 IRQ로 사용할 수 있습니다. 해당 GPIO의 IRQ number는 다음 call로 얻습니다.
int gpiod_to_irq(const struct gpio_desc *desc)
성공하면 IRQ number, mapping할 수 없으면 negative errno를 반환합니다. `gpiod_direction_input()`으로 input 설정하지 않은 GPIO를 사용하거나 `gpiod_to_irq()`에서 얻지 않은 IRQ number를 이 경로에서 사용하는 것은 unchecked error입니다. `gpiod_to_irq()`는 sleep할 수 없습니다.
성공한 IRQ number는 `request_irq()`와 `free_irq()`에 전달할 수 있고 board-specific initialization code가 platform device IRQ resource에 저장하기도 합니다. `IRQF_TRIGGER_FALLING` 같은 trigger option과 system wakeup 기능은 GPIO가 아니라 IRQ interface의 일부입니다.
Descriptor를 input으로 설정한 뒤 IRQ interface로 넘기는 과정입니다.
GPIO와 ACPI
434-451ACPI system에서 GPIO는 device의 `_CRS` configuration object에 나열된 `GpioIo()`와 `GpioInt()` resource로 기술됩니다. 이 resource에는 GPIO connection ID 또는 name이 없으므로 추가 mechanism이 필요합니다.
ACPI 5.1 이상 system은 `_DSD` configuration object로 `_CRS`의 특정 GPIO resource에 connection ID를 제공할 수 있으며 GPIO subsystem이 자동 처리합니다. `_DSD`가 없으면 device driver가 `GpioIo()`·`GpioInt()` resource와 GPIO connection ID의 mapping을 제공해야 합니다.
자세한 내용은 `Documentation/firmware-guide/acpi/gpio-properties.rst`를 참조합니다.
_CRS resource와 _DSD name mapping의 관계입니다.
Legacy GPIO subsystem과 상호 운용
452-470많은 kernel subsystem과 driver가 아직 legacy integer-based GPIO interface를 사용하지만 새 `gpiod` interface로 갱신하는 것을 강하게 권장합니다.
두 interface를 함께 써야 할 때 descriptor와 integer namespace를 다음 function으로 변환합니다.
int desc_to_gpio(const struct gpio_desc *desc)
struct gpio_desc *gpio_to_desc(unsigned gpio)
`desc_to_gpio()`가 반환한 GPIO number는 descriptor `desc`를 해제하기 전까지 `gpio_*()` function parameter로 안전하게 사용할 수 있습니다.
`gpio_to_desc()`에 전달할 GPIO number는 먼저 `gpio_request_one()` 같은 function으로 적절히 획득해야 하며, 반환 descriptor는 `gpio_free()`로 그 number를 해제할 때까지만 valid합니다.
한 API로 얻은 GPIO를 다른 API로 해제하는 것은 금지되며 unchecked error입니다.
변환 후 validity와 해제 API의 소유권 규칙입니다.
요약과 해설
consumer.rst:1-470GPIO consumer는 opaque descriptor를 획득한 뒤 direction을 먼저 설정하고 실행 context에 맞는 logical-value accessor를 사용해야 합니다. Mapping layer가 active-low와 open-drain/source semantics를 처리하며, array API는 chip topology가 맞으면 bitmap fast path를 사용합니다.
Legacy integer API와 descriptor API의 소유권은 섞어 해제할 수 없습니다. ACPI에서는 _CRS resource와 _DSD connection ID를 GPIO subsystem이 연결합니다.