Linux Audio

Check our new training course

Loading...
   1===============================
   2PINCTRL (PIN CONTROL) subsystem
   3===============================
   4
   5This document outlines the pin control subsystem in Linux
   6
   7This subsystem deals with:
   8
   9- Enumerating and naming controllable pins
  10
  11- Multiplexing of pins, pads, fingers (etc) see below for details
  12
  13- Configuration of pins, pads, fingers (etc), such as software-controlled
  14  biasing and driving mode specific pins, such as pull-up/down, open drain,
  15  load capacitance etc.
  16
  17Top-level interface
  18===================
  19
  20Definition of PIN CONTROLLER:
  21
  22- A pin controller is a piece of hardware, usually a set of registers, that
  23  can control PINs. It may be able to multiplex, bias, set load capacitance,
  24  set drive strength, etc. for individual pins or groups of pins.
  25
  26Definition of PIN:
  27
  28- PINS are equal to pads, fingers, balls or whatever packaging input or
  29  output line you want to control and these are denoted by unsigned integers
  30  in the range 0..maxpin. This numberspace is local to each PIN CONTROLLER, so
  31  there may be several such number spaces in a system. This pin space may
  32  be sparse - i.e. there may be gaps in the space with numbers where no
  33  pin exists.
  34
  35When a PIN CONTROLLER is instantiated, it will register a descriptor to the
  36pin control framework, and this descriptor contains an array of pin descriptors
  37describing the pins handled by this specific pin controller.
  38
  39Here is an example of a PGA (Pin Grid Array) chip seen from underneath::
  40
  41        A   B   C   D   E   F   G   H
  42
  43   8    o   o   o   o   o   o   o   o
  44
  45   7    o   o   o   o   o   o   o   o
  46
  47   6    o   o   o   o   o   o   o   o
  48
  49   5    o   o   o   o   o   o   o   o
  50
  51   4    o   o   o   o   o   o   o   o
  52
  53   3    o   o   o   o   o   o   o   o
  54
  55   2    o   o   o   o   o   o   o   o
  56
  57   1    o   o   o   o   o   o   o   o
  58
  59To register a pin controller and name all the pins on this package we can do
  60this in our driver::
  61
  62	#include <linux/pinctrl/pinctrl.h>
  63
  64	const struct pinctrl_pin_desc foo_pins[] = {
  65		PINCTRL_PIN(0, "A8"),
  66		PINCTRL_PIN(1, "B8"),
  67		PINCTRL_PIN(2, "C8"),
  68		...
  69		PINCTRL_PIN(61, "F1"),
  70		PINCTRL_PIN(62, "G1"),
  71		PINCTRL_PIN(63, "H1"),
  72	};
  73
  74	static struct pinctrl_desc foo_desc = {
  75		.name = "foo",
  76		.pins = foo_pins,
  77		.npins = ARRAY_SIZE(foo_pins),
  78		.owner = THIS_MODULE,
  79	};
  80
  81	int __init foo_probe(void)
  82	{
  83		int error;
  84
  85		struct pinctrl_dev *pctl;
  86
  87		error = pinctrl_register_and_init(&foo_desc, <PARENT>,
  88						  NULL, &pctl);
  89		if (error)
  90			return error;
  91
  92		return pinctrl_enable(pctl);
  93	}
  94
  95To enable the pinctrl subsystem and the subgroups for PINMUX and PINCONF and
  96selected drivers, you need to select them from your machine's Kconfig entry,
  97since these are so tightly integrated with the machines they are used on.
  98See for example arch/arm/mach-u300/Kconfig for an example.
  99
 100Pins usually have fancier names than this. You can find these in the datasheet
 101for your chip. Notice that the core pinctrl.h file provides a fancy macro
 102called PINCTRL_PIN() to create the struct entries. As you can see I enumerated
 103the pins from 0 in the upper left corner to 63 in the lower right corner.
 104This enumeration was arbitrarily chosen, in practice you need to think
 105through your numbering system so that it matches the layout of registers
 106and such things in your driver, or the code may become complicated. You must
 107also consider matching of offsets to the GPIO ranges that may be handled by
 108the pin controller.
 109
 110For a padring with 467 pads, as opposed to actual pins, I used an enumeration
 111like this, walking around the edge of the chip, which seems to be industry
 112standard too (all these pads had names, too)::
 113
 114
 115     0 ..... 104
 116   466        105
 117     .        .
 118     .        .
 119   358        224
 120    357 .... 225
 121
 122
 123Pin groups
 124==========
 125
 126Many controllers need to deal with groups of pins, so the pin controller
 127subsystem has a mechanism for enumerating groups of pins and retrieving the
 128actual enumerated pins that are part of a certain group.
 129
 130For example, say that we have a group of pins dealing with an SPI interface
 131on { 0, 8, 16, 24 }, and a group of pins dealing with an I2C interface on pins
 132on { 24, 25 }.
 133
 134These two groups are presented to the pin control subsystem by implementing
 135some generic pinctrl_ops like this::
 136
 137	#include <linux/pinctrl/pinctrl.h>
 138
 139	struct foo_group {
 140		const char *name;
 141		const unsigned int *pins;
 142		const unsigned num_pins;
 143	};
 144
 145	static const unsigned int spi0_pins[] = { 0, 8, 16, 24 };
 146	static const unsigned int i2c0_pins[] = { 24, 25 };
 147
 148	static const struct foo_group foo_groups[] = {
 149		{
 150			.name = "spi0_grp",
 151			.pins = spi0_pins,
 152			.num_pins = ARRAY_SIZE(spi0_pins),
 153		},
 154		{
 155			.name = "i2c0_grp",
 156			.pins = i2c0_pins,
 157			.num_pins = ARRAY_SIZE(i2c0_pins),
 158		},
 159	};
 160
 161
 162	static int foo_get_groups_count(struct pinctrl_dev *pctldev)
 163	{
 164		return ARRAY_SIZE(foo_groups);
 165	}
 166
 167	static const char *foo_get_group_name(struct pinctrl_dev *pctldev,
 168					unsigned selector)
 169	{
 170		return foo_groups[selector].name;
 171	}
 172
 173	static int foo_get_group_pins(struct pinctrl_dev *pctldev, unsigned selector,
 174				const unsigned **pins,
 175				unsigned *num_pins)
 176	{
 177		*pins = (unsigned *) foo_groups[selector].pins;
 178		*num_pins = foo_groups[selector].num_pins;
 179		return 0;
 180	}
 181
 182	static struct pinctrl_ops foo_pctrl_ops = {
 183		.get_groups_count = foo_get_groups_count,
 184		.get_group_name = foo_get_group_name,
 185		.get_group_pins = foo_get_group_pins,
 186	};
 187
 188
 189	static struct pinctrl_desc foo_desc = {
 190	...
 191	.pctlops = &foo_pctrl_ops,
 192	};
 193
 194The pin control subsystem will call the .get_groups_count() function to
 195determine the total number of legal selectors, then it will call the other functions
 196to retrieve the name and pins of the group. Maintaining the data structure of
 197the groups is up to the driver, this is just a simple example - in practice you
 198may need more entries in your group structure, for example specific register
 199ranges associated with each group and so on.
 200
 201
 202Pin configuration
 203=================
 204
 205Pins can sometimes be software-configured in various ways, mostly related
 206to their electronic properties when used as inputs or outputs. For example you
 207may be able to make an output pin high impedance, or "tristate" meaning it is
 208effectively disconnected. You may be able to connect an input pin to VDD or GND
 209using a certain resistor value - pull up and pull down - so that the pin has a
 210stable value when nothing is driving the rail it is connected to, or when it's
 211unconnected.
 212
 213Pin configuration can be programmed by adding configuration entries into the
 214mapping table; see section "Board/machine configuration" below.
 215
 216The format and meaning of the configuration parameter, PLATFORM_X_PULL_UP
 217above, is entirely defined by the pin controller driver.
 218
 219The pin configuration driver implements callbacks for changing pin
 220configuration in the pin controller ops like this::
 221
 222	#include <linux/pinctrl/pinctrl.h>
 223	#include <linux/pinctrl/pinconf.h>
 224	#include "platform_x_pindefs.h"
 225
 226	static int foo_pin_config_get(struct pinctrl_dev *pctldev,
 227			unsigned offset,
 228			unsigned long *config)
 229	{
 230		struct my_conftype conf;
 231
 232		... Find setting for pin @ offset ...
 233
 234		*config = (unsigned long) conf;
 235	}
 236
 237	static int foo_pin_config_set(struct pinctrl_dev *pctldev,
 238			unsigned offset,
 239			unsigned long config)
 240	{
 241		struct my_conftype *conf = (struct my_conftype *) config;
 242
 243		switch (conf) {
 244			case PLATFORM_X_PULL_UP:
 245			...
 246			}
 247		}
 248	}
 249
 250	static int foo_pin_config_group_get (struct pinctrl_dev *pctldev,
 251			unsigned selector,
 252			unsigned long *config)
 253	{
 254		...
 255	}
 256
 257	static int foo_pin_config_group_set (struct pinctrl_dev *pctldev,
 258			unsigned selector,
 259			unsigned long config)
 260	{
 261		...
 262	}
 263
 264	static struct pinconf_ops foo_pconf_ops = {
 265		.pin_config_get = foo_pin_config_get,
 266		.pin_config_set = foo_pin_config_set,
 267		.pin_config_group_get = foo_pin_config_group_get,
 268		.pin_config_group_set = foo_pin_config_group_set,
 269	};
 270
 271	/* Pin config operations are handled by some pin controller */
 272	static struct pinctrl_desc foo_desc = {
 273		...
 274		.confops = &foo_pconf_ops,
 275	};
 276
 277Interaction with the GPIO subsystem
 278===================================
 279
 280The GPIO drivers may want to perform operations of various types on the same
 281physical pins that are also registered as pin controller pins.
 282
 283First and foremost, the two subsystems can be used as completely orthogonal,
 284see the section named "pin control requests from drivers" and
 285"drivers needing both pin control and GPIOs" below for details. But in some
 286situations a cross-subsystem mapping between pins and GPIOs is needed.
 287
 288Since the pin controller subsystem has its pinspace local to the pin controller
 289we need a mapping so that the pin control subsystem can figure out which pin
 290controller handles control of a certain GPIO pin. Since a single pin controller
 291may be muxing several GPIO ranges (typically SoCs that have one set of pins,
 292but internally several GPIO silicon blocks, each modelled as a struct
 293gpio_chip) any number of GPIO ranges can be added to a pin controller instance
 294like this::
 295
 296	struct gpio_chip chip_a;
 297	struct gpio_chip chip_b;
 298
 299	static struct pinctrl_gpio_range gpio_range_a = {
 300		.name = "chip a",
 301		.id = 0,
 302		.base = 32,
 303		.pin_base = 32,
 304		.npins = 16,
 305		.gc = &chip_a;
 306	};
 307
 308	static struct pinctrl_gpio_range gpio_range_b = {
 309		.name = "chip b",
 310		.id = 0,
 311		.base = 48,
 312		.pin_base = 64,
 313		.npins = 8,
 314		.gc = &chip_b;
 315	};
 316
 317	{
 318		struct pinctrl_dev *pctl;
 319		...
 320		pinctrl_add_gpio_range(pctl, &gpio_range_a);
 321		pinctrl_add_gpio_range(pctl, &gpio_range_b);
 322	}
 323
 324So this complex system has one pin controller handling two different
 325GPIO chips. "chip a" has 16 pins and "chip b" has 8 pins. The "chip a" and
 326"chip b" have different .pin_base, which means a start pin number of the
 327GPIO range.
 328
 329The GPIO range of "chip a" starts from the GPIO base of 32 and actual
 330pin range also starts from 32. However "chip b" has different starting
 331offset for the GPIO range and pin range. The GPIO range of "chip b" starts
 332from GPIO number 48, while the pin range of "chip b" starts from 64.
 333
 334We can convert a gpio number to actual pin number using this "pin_base".
 335They are mapped in the global GPIO pin space at:
 336
 337chip a:
 338 - GPIO range : [32 .. 47]
 339 - pin range  : [32 .. 47]
 340chip b:
 341 - GPIO range : [48 .. 55]
 342 - pin range  : [64 .. 71]
 343
 344The above examples assume the mapping between the GPIOs and pins is
 345linear. If the mapping is sparse or haphazard, an array of arbitrary pin
 346numbers can be encoded in the range like this::
 347
 348	static const unsigned range_pins[] = { 14, 1, 22, 17, 10, 8, 6, 2 };
 349
 350	static struct pinctrl_gpio_range gpio_range = {
 351		.name = "chip",
 352		.id = 0,
 353		.base = 32,
 354		.pins = &range_pins,
 355		.npins = ARRAY_SIZE(range_pins),
 356		.gc = &chip;
 357	};
 358
 359In this case the pin_base property will be ignored. If the name of a pin
 360group is known, the pins and npins elements of the above structure can be
 361initialised using the function pinctrl_get_group_pins(), e.g. for pin
 362group "foo"::
 363
 364	pinctrl_get_group_pins(pctl, "foo", &gpio_range.pins,
 365			       &gpio_range.npins);
 366
 367When GPIO-specific functions in the pin control subsystem are called, these
 368ranges will be used to look up the appropriate pin controller by inspecting
 369and matching the pin to the pin ranges across all controllers. When a
 370pin controller handling the matching range is found, GPIO-specific functions
 371will be called on that specific pin controller.
 372
 373For all functionalities dealing with pin biasing, pin muxing etc, the pin
 374controller subsystem will look up the corresponding pin number from the passed
 375in gpio number, and use the range's internals to retrieve a pin number. After
 376that, the subsystem passes it on to the pin control driver, so the driver
 377will get a pin number into its handled number range. Further it is also passed
 378the range ID value, so that the pin controller knows which range it should
 379deal with.
 380
 381Calling pinctrl_add_gpio_range from pinctrl driver is DEPRECATED. Please see
 382section 2.1 of Documentation/devicetree/bindings/gpio/gpio.txt on how to bind
 383pinctrl and gpio drivers.
 384
 385
 386PINMUX interfaces
 387=================
 388
 389These calls use the pinmux_* naming prefix.  No other calls should use that
 390prefix.
 391
 392
 393What is pinmuxing?
 394==================
 395
 396PINMUX, also known as padmux, ballmux, alternate functions or mission modes
 397is a way for chip vendors producing some kind of electrical packages to use
 398a certain physical pin (ball, pad, finger, etc) for multiple mutually exclusive
 399functions, depending on the application. By "application" in this context
 400we usually mean a way of soldering or wiring the package into an electronic
 401system, even though the framework makes it possible to also change the function
 402at runtime.
 403
 404Here is an example of a PGA (Pin Grid Array) chip seen from underneath::
 405
 406        A   B   C   D   E   F   G   H
 407      +---+
 408   8  | o | o   o   o   o   o   o   o
 409      |   |
 410   7  | o | o   o   o   o   o   o   o
 411      |   |
 412   6  | o | o   o   o   o   o   o   o
 413      +---+---+
 414   5  | o | o | o   o   o   o   o   o
 415      +---+---+               +---+
 416   4    o   o   o   o   o   o | o | o
 417                              |   |
 418   3    o   o   o   o   o   o | o | o
 419                              |   |
 420   2    o   o   o   o   o   o | o | o
 421      +-------+-------+-------+---+---+
 422   1  | o   o | o   o | o   o | o | o |
 423      +-------+-------+-------+---+---+
 424
 425This is not tetris. The game to think of is chess. Not all PGA/BGA packages
 426are chessboard-like, big ones have "holes" in some arrangement according to
 427different design patterns, but we're using this as a simple example. Of the
 428pins you see some will be taken by things like a few VCC and GND to feed power
 429to the chip, and quite a few will be taken by large ports like an external
 430memory interface. The remaining pins will often be subject to pin multiplexing.
 431
 432The example 8x8 PGA package above will have pin numbers 0 through 63 assigned
 433to its physical pins. It will name the pins { A1, A2, A3 ... H6, H7, H8 } using
 434pinctrl_register_pins() and a suitable data set as shown earlier.
 435
 436In this 8x8 BGA package the pins { A8, A7, A6, A5 } can be used as an SPI port
 437(these are four pins: CLK, RXD, TXD, FRM). In that case, pin B5 can be used as
 438some general-purpose GPIO pin. However, in another setting, pins { A5, B5 } can
 439be used as an I2C port (these are just two pins: SCL, SDA). Needless to say,
 440we cannot use the SPI port and I2C port at the same time. However in the inside
 441of the package the silicon performing the SPI logic can alternatively be routed
 442out on pins { G4, G3, G2, G1 }.
 443
 444On the bottom row at { A1, B1, C1, D1, E1, F1, G1, H1 } we have something
 445special - it's an external MMC bus that can be 2, 4 or 8 bits wide, and it will
 446consume 2, 4 or 8 pins respectively, so either { A1, B1 } are taken or
 447{ A1, B1, C1, D1 } or all of them. If we use all 8 bits, we cannot use the SPI
 448port on pins { G4, G3, G2, G1 } of course.
 449
 450This way the silicon blocks present inside the chip can be multiplexed "muxed"
 451out on different pin ranges. Often contemporary SoC (systems on chip) will
 452contain several I2C, SPI, SDIO/MMC, etc silicon blocks that can be routed to
 453different pins by pinmux settings.
 454
 455Since general-purpose I/O pins (GPIO) are typically always in shortage, it is
 456common to be able to use almost any pin as a GPIO pin if it is not currently
 457in use by some other I/O port.
 458
 459
 460Pinmux conventions
 461==================
 462
 463The purpose of the pinmux functionality in the pin controller subsystem is to
 464abstract and provide pinmux settings to the devices you choose to instantiate
 465in your machine configuration. It is inspired by the clk, GPIO and regulator
 466subsystems, so devices will request their mux setting, but it's also possible
 467to request a single pin for e.g. GPIO.
 468
 469Definitions:
 470
 471- FUNCTIONS can be switched in and out by a driver residing with the pin
 472  control subsystem in the drivers/pinctrl/* directory of the kernel. The
 473  pin control driver knows the possible functions. In the example above you can
 474  identify three pinmux functions, one for spi, one for i2c and one for mmc.
 475
 476- FUNCTIONS are assumed to be enumerable from zero in a one-dimensional array.
 477  In this case the array could be something like: { spi0, i2c0, mmc0 }
 478  for the three available functions.
 479
 480- FUNCTIONS have PIN GROUPS as defined on the generic level - so a certain
 481  function is *always* associated with a certain set of pin groups, could
 482  be just a single one, but could also be many. In the example above the
 483  function i2c is associated with the pins { A5, B5 }, enumerated as
 484  { 24, 25 } in the controller pin space.
 485
 486  The Function spi is associated with pin groups { A8, A7, A6, A5 }
 487  and { G4, G3, G2, G1 }, which are enumerated as { 0, 8, 16, 24 } and
 488  { 38, 46, 54, 62 } respectively.
 489
 490  Group names must be unique per pin controller, no two groups on the same
 491  controller may have the same name.
 492
 493- The combination of a FUNCTION and a PIN GROUP determine a certain function
 494  for a certain set of pins. The knowledge of the functions and pin groups
 495  and their machine-specific particulars are kept inside the pinmux driver,
 496  from the outside only the enumerators are known, and the driver core can
 497  request:
 498
 499  - The name of a function with a certain selector (>= 0)
 500  - A list of groups associated with a certain function
 501  - That a certain group in that list to be activated for a certain function
 502
 503  As already described above, pin groups are in turn self-descriptive, so
 504  the core will retrieve the actual pin range in a certain group from the
 505  driver.
 506
 507- FUNCTIONS and GROUPS on a certain PIN CONTROLLER are MAPPED to a certain
 508  device by the board file, device tree or similar machine setup configuration
 509  mechanism, similar to how regulators are connected to devices, usually by
 510  name. Defining a pin controller, function and group thus uniquely identify
 511  the set of pins to be used by a certain device. (If only one possible group
 512  of pins is available for the function, no group name need to be supplied -
 513  the core will simply select the first and only group available.)
 514
 515  In the example case we can define that this particular machine shall
 516  use device spi0 with pinmux function fspi0 group gspi0 and i2c0 on function
 517  fi2c0 group gi2c0, on the primary pin controller, we get mappings
 518  like these::
 519
 520	{
 521		{"map-spi0", spi0, pinctrl0, fspi0, gspi0},
 522		{"map-i2c0", i2c0, pinctrl0, fi2c0, gi2c0}
 523	}
 524
 525  Every map must be assigned a state name, pin controller, device and
 526  function. The group is not compulsory - if it is omitted the first group
 527  presented by the driver as applicable for the function will be selected,
 528  which is useful for simple cases.
 529
 530  It is possible to map several groups to the same combination of device,
 531  pin controller and function. This is for cases where a certain function on
 532  a certain pin controller may use different sets of pins in different
 533  configurations.
 534
 535- PINS for a certain FUNCTION using a certain PIN GROUP on a certain
 536  PIN CONTROLLER are provided on a first-come first-serve basis, so if some
 537  other device mux setting or GPIO pin request has already taken your physical
 538  pin, you will be denied the use of it. To get (activate) a new setting, the
 539  old one has to be put (deactivated) first.
 540
 541Sometimes the documentation and hardware registers will be oriented around
 542pads (or "fingers") rather than pins - these are the soldering surfaces on the
 543silicon inside the package, and may or may not match the actual number of
 544pins/balls underneath the capsule. Pick some enumeration that makes sense to
 545you. Define enumerators only for the pins you can control if that makes sense.
 546
 547Assumptions:
 548
 549We assume that the number of possible function maps to pin groups is limited by
 550the hardware. I.e. we assume that there is no system where any function can be
 551mapped to any pin, like in a phone exchange. So the available pin groups for
 552a certain function will be limited to a few choices (say up to eight or so),
 553not hundreds or any amount of choices. This is the characteristic we have found
 554by inspecting available pinmux hardware, and a necessary assumption since we
 555expect pinmux drivers to present *all* possible function vs pin group mappings
 556to the subsystem.
 557
 558
 559Pinmux drivers
 560==============
 561
 562The pinmux core takes care of preventing conflicts on pins and calling
 563the pin controller driver to execute different settings.
 564
 565It is the responsibility of the pinmux driver to impose further restrictions
 566(say for example infer electronic limitations due to load, etc.) to determine
 567whether or not the requested function can actually be allowed, and in case it
 568is possible to perform the requested mux setting, poke the hardware so that
 569this happens.
 570
 571Pinmux drivers are required to supply a few callback functions, some are
 572optional. Usually the set_mux() function is implemented, writing values into
 573some certain registers to activate a certain mux setting for a certain pin.
 574
 575A simple driver for the above example will work by setting bits 0, 1, 2, 3 or 4
 576into some register named MUX to select a certain function with a certain
 577group of pins would work something like this::
 578
 579	#include <linux/pinctrl/pinctrl.h>
 580	#include <linux/pinctrl/pinmux.h>
 581
 582	struct foo_group {
 583		const char *name;
 584		const unsigned int *pins;
 585		const unsigned num_pins;
 586	};
 587
 588	static const unsigned spi0_0_pins[] = { 0, 8, 16, 24 };
 589	static const unsigned spi0_1_pins[] = { 38, 46, 54, 62 };
 590	static const unsigned i2c0_pins[] = { 24, 25 };
 591	static const unsigned mmc0_1_pins[] = { 56, 57 };
 592	static const unsigned mmc0_2_pins[] = { 58, 59 };
 593	static const unsigned mmc0_3_pins[] = { 60, 61, 62, 63 };
 594
 595	static const struct foo_group foo_groups[] = {
 596		{
 597			.name = "spi0_0_grp",
 598			.pins = spi0_0_pins,
 599			.num_pins = ARRAY_SIZE(spi0_0_pins),
 600		},
 601		{
 602			.name = "spi0_1_grp",
 603			.pins = spi0_1_pins,
 604			.num_pins = ARRAY_SIZE(spi0_1_pins),
 605		},
 606		{
 607			.name = "i2c0_grp",
 608			.pins = i2c0_pins,
 609			.num_pins = ARRAY_SIZE(i2c0_pins),
 610		},
 611		{
 612			.name = "mmc0_1_grp",
 613			.pins = mmc0_1_pins,
 614			.num_pins = ARRAY_SIZE(mmc0_1_pins),
 615		},
 616		{
 617			.name = "mmc0_2_grp",
 618			.pins = mmc0_2_pins,
 619			.num_pins = ARRAY_SIZE(mmc0_2_pins),
 620		},
 621		{
 622			.name = "mmc0_3_grp",
 623			.pins = mmc0_3_pins,
 624			.num_pins = ARRAY_SIZE(mmc0_3_pins),
 625		},
 626	};
 627
 628
 629	static int foo_get_groups_count(struct pinctrl_dev *pctldev)
 630	{
 631		return ARRAY_SIZE(foo_groups);
 632	}
 633
 634	static const char *foo_get_group_name(struct pinctrl_dev *pctldev,
 635					unsigned selector)
 636	{
 637		return foo_groups[selector].name;
 638	}
 639
 640	static int foo_get_group_pins(struct pinctrl_dev *pctldev, unsigned selector,
 641				const unsigned ** pins,
 642				unsigned * num_pins)
 643	{
 644		*pins = (unsigned *) foo_groups[selector].pins;
 645		*num_pins = foo_groups[selector].num_pins;
 646		return 0;
 647	}
 648
 649	static struct pinctrl_ops foo_pctrl_ops = {
 650		.get_groups_count = foo_get_groups_count,
 651		.get_group_name = foo_get_group_name,
 652		.get_group_pins = foo_get_group_pins,
 653	};
 654
 655	struct foo_pmx_func {
 656		const char *name;
 657		const char * const *groups;
 658		const unsigned num_groups;
 659	};
 660
 661	static const char * const spi0_groups[] = { "spi0_0_grp", "spi0_1_grp" };
 662	static const char * const i2c0_groups[] = { "i2c0_grp" };
 663	static const char * const mmc0_groups[] = { "mmc0_1_grp", "mmc0_2_grp",
 664						"mmc0_3_grp" };
 665
 666	static const struct foo_pmx_func foo_functions[] = {
 667		{
 668			.name = "spi0",
 669			.groups = spi0_groups,
 670			.num_groups = ARRAY_SIZE(spi0_groups),
 671		},
 672		{
 673			.name = "i2c0",
 674			.groups = i2c0_groups,
 675			.num_groups = ARRAY_SIZE(i2c0_groups),
 676		},
 677		{
 678			.name = "mmc0",
 679			.groups = mmc0_groups,
 680			.num_groups = ARRAY_SIZE(mmc0_groups),
 681		},
 682	};
 683
 684	static int foo_get_functions_count(struct pinctrl_dev *pctldev)
 685	{
 686		return ARRAY_SIZE(foo_functions);
 687	}
 688
 689	static const char *foo_get_fname(struct pinctrl_dev *pctldev, unsigned selector)
 690	{
 691		return foo_functions[selector].name;
 692	}
 693
 694	static int foo_get_groups(struct pinctrl_dev *pctldev, unsigned selector,
 695				const char * const **groups,
 696				unsigned * const num_groups)
 697	{
 698		*groups = foo_functions[selector].groups;
 699		*num_groups = foo_functions[selector].num_groups;
 700		return 0;
 701	}
 702
 703	static int foo_set_mux(struct pinctrl_dev *pctldev, unsigned selector,
 704			unsigned group)
 705	{
 706		u8 regbit = (1 << selector + group);
 707
 708		writeb((readb(MUX)|regbit), MUX);
 709		return 0;
 710	}
 711
 712	static struct pinmux_ops foo_pmxops = {
 713		.get_functions_count = foo_get_functions_count,
 714		.get_function_name = foo_get_fname,
 715		.get_function_groups = foo_get_groups,
 716		.set_mux = foo_set_mux,
 717		.strict = true,
 718	};
 719
 720	/* Pinmux operations are handled by some pin controller */
 721	static struct pinctrl_desc foo_desc = {
 722		...
 723		.pctlops = &foo_pctrl_ops,
 724		.pmxops = &foo_pmxops,
 725	};
 726
 727In the example activating muxing 0 and 1 at the same time setting bits
 7280 and 1, uses one pin in common so they would collide.
 729
 730The beauty of the pinmux subsystem is that since it keeps track of all
 731pins and who is using them, it will already have denied an impossible
 732request like that, so the driver does not need to worry about such
 733things - when it gets a selector passed in, the pinmux subsystem makes
 734sure no other device or GPIO assignment is already using the selected
 735pins. Thus bits 0 and 1 in the control register will never be set at the
 736same time.
 737
 738All the above functions are mandatory to implement for a pinmux driver.
 739
 740
 741Pin control interaction with the GPIO subsystem
 742===============================================
 743
 744Note that the following implies that the use case is to use a certain pin
 745from the Linux kernel using the API in <linux/gpio.h> with gpio_request()
 746and similar functions. There are cases where you may be using something
 747that your datasheet calls "GPIO mode", but actually is just an electrical
 748configuration for a certain device. See the section below named
 749"GPIO mode pitfalls" for more details on this scenario.
 750
 751The public pinmux API contains two functions named pinctrl_gpio_request()
 752and pinctrl_gpio_free(). These two functions shall *ONLY* be called from
 753gpiolib-based drivers as part of their gpio_request() and
 754gpio_free() semantics. Likewise the pinctrl_gpio_direction_[input|output]
 755shall only be called from within respective gpio_direction_[input|output]
 756gpiolib implementation.
 757
 758NOTE that platforms and individual drivers shall *NOT* request GPIO pins to be
 759controlled e.g. muxed in. Instead, implement a proper gpiolib driver and have
 760that driver request proper muxing and other control for its pins.
 761
 762The function list could become long, especially if you can convert every
 763individual pin into a GPIO pin independent of any other pins, and then try
 764the approach to define every pin as a function.
 765
 766In this case, the function array would become 64 entries for each GPIO
 767setting and then the device functions.
 768
 769For this reason there are two functions a pin control driver can implement
 770to enable only GPIO on an individual pin: .gpio_request_enable() and
 771.gpio_disable_free().
 772
 773This function will pass in the affected GPIO range identified by the pin
 774controller core, so you know which GPIO pins are being affected by the request
 775operation.
 776
 777If your driver needs to have an indication from the framework of whether the
 778GPIO pin shall be used for input or output you can implement the
 779.gpio_set_direction() function. As described this shall be called from the
 780gpiolib driver and the affected GPIO range, pin offset and desired direction
 781will be passed along to this function.
 782
 783Alternatively to using these special functions, it is fully allowed to use
 784named functions for each GPIO pin, the pinctrl_gpio_request() will attempt to
 785obtain the function "gpioN" where "N" is the global GPIO pin number if no
 786special GPIO-handler is registered.
 787
 788
 789GPIO mode pitfalls
 790==================
 791
 792Due to the naming conventions used by hardware engineers, where "GPIO"
 793is taken to mean different things than what the kernel does, the developer
 794may be confused by a datasheet talking about a pin being possible to set
 795into "GPIO mode". It appears that what hardware engineers mean with
 796"GPIO mode" is not necessarily the use case that is implied in the kernel
 797interface <linux/gpio.h>: a pin that you grab from kernel code and then
 798either listen for input or drive high/low to assert/deassert some
 799external line.
 800
 801Rather hardware engineers think that "GPIO mode" means that you can
 802software-control a few electrical properties of the pin that you would
 803not be able to control if the pin was in some other mode, such as muxed in
 804for a device.
 805
 806The GPIO portions of a pin and its relation to a certain pin controller
 807configuration and muxing logic can be constructed in several ways. Here
 808are two examples::
 809
 810     (A)
 811                       pin config
 812                       logic regs
 813                       |               +- SPI
 814     Physical pins --- pad --- pinmux -+- I2C
 815                               |       +- mmc
 816                               |       +- GPIO
 817                               pin
 818                               multiplex
 819                               logic regs
 820
 821Here some electrical properties of the pin can be configured no matter
 822whether the pin is used for GPIO or not. If you multiplex a GPIO onto a
 823pin, you can also drive it high/low from "GPIO" registers.
 824Alternatively, the pin can be controlled by a certain peripheral, while
 825still applying desired pin config properties. GPIO functionality is thus
 826orthogonal to any other device using the pin.
 827
 828In this arrangement the registers for the GPIO portions of the pin controller,
 829or the registers for the GPIO hardware module are likely to reside in a
 830separate memory range only intended for GPIO driving, and the register
 831range dealing with pin config and pin multiplexing get placed into a
 832different memory range and a separate section of the data sheet.
 833
 834A flag "strict" in struct pinmux_ops is available to check and deny
 835simultaneous access to the same pin from GPIO and pin multiplexing
 836consumers on hardware of this type. The pinctrl driver should set this flag
 837accordingly.
 838
 839::
 840
 841     (B)
 842
 843                       pin config
 844                       logic regs
 845                       |               +- SPI
 846     Physical pins --- pad --- pinmux -+- I2C
 847                       |       |       +- mmc
 848                       |       |
 849                       GPIO    pin
 850                               multiplex
 851                               logic regs
 852
 853In this arrangement, the GPIO functionality can always be enabled, such that
 854e.g. a GPIO input can be used to "spy" on the SPI/I2C/MMC signal while it is
 855pulsed out. It is likely possible to disrupt the traffic on the pin by doing
 856wrong things on the GPIO block, as it is never really disconnected. It is
 857possible that the GPIO, pin config and pin multiplex registers are placed into
 858the same memory range and the same section of the data sheet, although that
 859need not be the case.
 860
 861In some pin controllers, although the physical pins are designed in the same
 862way as (B), the GPIO function still can't be enabled at the same time as the
 863peripheral functions. So again the "strict" flag should be set, denying
 864simultaneous activation by GPIO and other muxed in devices.
 865
 866From a kernel point of view, however, these are different aspects of the
 867hardware and shall be put into different subsystems:
 868
 869- Registers (or fields within registers) that control electrical
 870  properties of the pin such as biasing and drive strength should be
 871  exposed through the pinctrl subsystem, as "pin configuration" settings.
 872
 873- Registers (or fields within registers) that control muxing of signals
 874  from various other HW blocks (e.g. I2C, MMC, or GPIO) onto pins should
 875  be exposed through the pinctrl subsystem, as mux functions.
 876
 877- Registers (or fields within registers) that control GPIO functionality
 878  such as setting a GPIO's output value, reading a GPIO's input value, or
 879  setting GPIO pin direction should be exposed through the GPIO subsystem,
 880  and if they also support interrupt capabilities, through the irqchip
 881  abstraction.
 882
 883Depending on the exact HW register design, some functions exposed by the
 884GPIO subsystem may call into the pinctrl subsystem in order to
 885co-ordinate register settings across HW modules. In particular, this may
 886be needed for HW with separate GPIO and pin controller HW modules, where
 887e.g. GPIO direction is determined by a register in the pin controller HW
 888module rather than the GPIO HW module.
 889
 890Electrical properties of the pin such as biasing and drive strength
 891may be placed at some pin-specific register in all cases or as part
 892of the GPIO register in case (B) especially. This doesn't mean that such
 893properties necessarily pertain to what the Linux kernel calls "GPIO".
 894
 895Example: a pin is usually muxed in to be used as a UART TX line. But during
 896system sleep, we need to put this pin into "GPIO mode" and ground it.
 897
 898If you make a 1-to-1 map to the GPIO subsystem for this pin, you may start
 899to think that you need to come up with something really complex, that the
 900pin shall be used for UART TX and GPIO at the same time, that you will grab
 901a pin control handle and set it to a certain state to enable UART TX to be
 902muxed in, then twist it over to GPIO mode and use gpio_direction_output()
 903to drive it low during sleep, then mux it over to UART TX again when you
 904wake up and maybe even gpio_request/gpio_free as part of this cycle. This
 905all gets very complicated.
 906
 907The solution is to not think that what the datasheet calls "GPIO mode"
 908has to be handled by the <linux/gpio.h> interface. Instead view this as
 909a certain pin config setting. Look in e.g. <linux/pinctrl/pinconf-generic.h>
 910and you find this in the documentation:
 911
 912  PIN_CONFIG_OUTPUT:
 913     this will configure the pin in output, use argument
 914     1 to indicate high level, argument 0 to indicate low level.
 915
 916So it is perfectly possible to push a pin into "GPIO mode" and drive the
 917line low as part of the usual pin control map. So for example your UART
 918driver may look like this::
 919
 920	#include <linux/pinctrl/consumer.h>
 921
 922	struct pinctrl          *pinctrl;
 923	struct pinctrl_state    *pins_default;
 924	struct pinctrl_state    *pins_sleep;
 925
 926	pins_default = pinctrl_lookup_state(uap->pinctrl, PINCTRL_STATE_DEFAULT);
 927	pins_sleep = pinctrl_lookup_state(uap->pinctrl, PINCTRL_STATE_SLEEP);
 928
 929	/* Normal mode */
 930	retval = pinctrl_select_state(pinctrl, pins_default);
 931	/* Sleep mode */
 932	retval = pinctrl_select_state(pinctrl, pins_sleep);
 933
 934And your machine configuration may look like this:
 935--------------------------------------------------
 936
 937::
 938
 939	static unsigned long uart_default_mode[] = {
 940		PIN_CONF_PACKED(PIN_CONFIG_DRIVE_PUSH_PULL, 0),
 941	};
 942
 943	static unsigned long uart_sleep_mode[] = {
 944		PIN_CONF_PACKED(PIN_CONFIG_OUTPUT, 0),
 945	};
 946
 947	static struct pinctrl_map pinmap[] __initdata = {
 948		PIN_MAP_MUX_GROUP("uart", PINCTRL_STATE_DEFAULT, "pinctrl-foo",
 949			"u0_group", "u0"),
 950		PIN_MAP_CONFIGS_PIN("uart", PINCTRL_STATE_DEFAULT, "pinctrl-foo",
 951				"UART_TX_PIN", uart_default_mode),
 952		PIN_MAP_MUX_GROUP("uart", PINCTRL_STATE_SLEEP, "pinctrl-foo",
 953			"u0_group", "gpio-mode"),
 954		PIN_MAP_CONFIGS_PIN("uart", PINCTRL_STATE_SLEEP, "pinctrl-foo",
 955				"UART_TX_PIN", uart_sleep_mode),
 956	};
 957
 958	foo_init(void) {
 959		pinctrl_register_mappings(pinmap, ARRAY_SIZE(pinmap));
 960	}
 961
 962Here the pins we want to control are in the "u0_group" and there is some
 963function called "u0" that can be enabled on this group of pins, and then
 964everything is UART business as usual. But there is also some function
 965named "gpio-mode" that can be mapped onto the same pins to move them into
 966GPIO mode.
 967
 968This will give the desired effect without any bogus interaction with the
 969GPIO subsystem. It is just an electrical configuration used by that device
 970when going to sleep, it might imply that the pin is set into something the
 971datasheet calls "GPIO mode", but that is not the point: it is still used
 972by that UART device to control the pins that pertain to that very UART
 973driver, putting them into modes needed by the UART. GPIO in the Linux
 974kernel sense are just some 1-bit line, and is a different use case.
 975
 976How the registers are poked to attain the push or pull, and output low
 977configuration and the muxing of the "u0" or "gpio-mode" group onto these
 978pins is a question for the driver.
 979
 980Some datasheets will be more helpful and refer to the "GPIO mode" as
 981"low power mode" rather than anything to do with GPIO. This often means
 982the same thing electrically speaking, but in this latter case the
 983software engineers will usually quickly identify that this is some
 984specific muxing or configuration rather than anything related to the GPIO
 985API.
 986
 987
 988Board/machine configuration
 989===========================
 990
 991Boards and machines define how a certain complete running system is put
 992together, including how GPIOs and devices are muxed, how regulators are
 993constrained and how the clock tree looks. Of course pinmux settings are also
 994part of this.
 995
 996A pin controller configuration for a machine looks pretty much like a simple
 997regulator configuration, so for the example array above we want to enable i2c
 998and spi on the second function mapping::
 999
1000	#include <linux/pinctrl/machine.h>
1001
1002	static const struct pinctrl_map mapping[] __initconst = {
1003		{
1004			.dev_name = "foo-spi.0",
1005			.name = PINCTRL_STATE_DEFAULT,
1006			.type = PIN_MAP_TYPE_MUX_GROUP,
1007			.ctrl_dev_name = "pinctrl-foo",
1008			.data.mux.function = "spi0",
1009		},
1010		{
1011			.dev_name = "foo-i2c.0",
1012			.name = PINCTRL_STATE_DEFAULT,
1013			.type = PIN_MAP_TYPE_MUX_GROUP,
1014			.ctrl_dev_name = "pinctrl-foo",
1015			.data.mux.function = "i2c0",
1016		},
1017		{
1018			.dev_name = "foo-mmc.0",
1019			.name = PINCTRL_STATE_DEFAULT,
1020			.type = PIN_MAP_TYPE_MUX_GROUP,
1021			.ctrl_dev_name = "pinctrl-foo",
1022			.data.mux.function = "mmc0",
1023		},
1024	};
1025
1026The dev_name here matches to the unique device name that can be used to look
1027up the device struct (just like with clockdev or regulators). The function name
1028must match a function provided by the pinmux driver handling this pin range.
1029
1030As you can see we may have several pin controllers on the system and thus
1031we need to specify which one of them contains the functions we wish to map.
1032
1033You register this pinmux mapping to the pinmux subsystem by simply::
1034
1035       ret = pinctrl_register_mappings(mapping, ARRAY_SIZE(mapping));
1036
1037Since the above construct is pretty common there is a helper macro to make
1038it even more compact which assumes you want to use pinctrl-foo and position
10390 for mapping, for example::
1040
1041	static struct pinctrl_map mapping[] __initdata = {
1042		PIN_MAP_MUX_GROUP("foo-i2c.o", PINCTRL_STATE_DEFAULT,
1043				  "pinctrl-foo", NULL, "i2c0"),
1044	};
1045
1046The mapping table may also contain pin configuration entries. It's common for
1047each pin/group to have a number of configuration entries that affect it, so
1048the table entries for configuration reference an array of config parameters
1049and values. An example using the convenience macros is shown below::
1050
1051	static unsigned long i2c_grp_configs[] = {
1052		FOO_PIN_DRIVEN,
1053		FOO_PIN_PULLUP,
1054	};
1055
1056	static unsigned long i2c_pin_configs[] = {
1057		FOO_OPEN_COLLECTOR,
1058		FOO_SLEW_RATE_SLOW,
1059	};
1060
1061	static struct pinctrl_map mapping[] __initdata = {
1062		PIN_MAP_MUX_GROUP("foo-i2c.0", PINCTRL_STATE_DEFAULT,
1063				  "pinctrl-foo", "i2c0", "i2c0"),
1064		PIN_MAP_CONFIGS_GROUP("foo-i2c.0", PINCTRL_STATE_DEFAULT,
1065				      "pinctrl-foo", "i2c0", i2c_grp_configs),
1066		PIN_MAP_CONFIGS_PIN("foo-i2c.0", PINCTRL_STATE_DEFAULT,
1067				    "pinctrl-foo", "i2c0scl", i2c_pin_configs),
1068		PIN_MAP_CONFIGS_PIN("foo-i2c.0", PINCTRL_STATE_DEFAULT,
1069				    "pinctrl-foo", "i2c0sda", i2c_pin_configs),
1070	};
1071
1072Finally, some devices expect the mapping table to contain certain specific
1073named states. When running on hardware that doesn't need any pin controller
1074configuration, the mapping table must still contain those named states, in
1075order to explicitly indicate that the states were provided and intended to
1076be empty. Table entry macro PIN_MAP_DUMMY_STATE serves the purpose of defining
1077a named state without causing any pin controller to be programmed::
1078
1079	static struct pinctrl_map mapping[] __initdata = {
1080		PIN_MAP_DUMMY_STATE("foo-i2c.0", PINCTRL_STATE_DEFAULT),
1081	};
1082
1083
1084Complex mappings
1085================
1086
1087As it is possible to map a function to different groups of pins an optional
1088.group can be specified like this::
1089
1090	...
1091	{
1092		.dev_name = "foo-spi.0",
1093		.name = "spi0-pos-A",
1094		.type = PIN_MAP_TYPE_MUX_GROUP,
1095		.ctrl_dev_name = "pinctrl-foo",
1096		.function = "spi0",
1097		.group = "spi0_0_grp",
1098	},
1099	{
1100		.dev_name = "foo-spi.0",
1101		.name = "spi0-pos-B",
1102		.type = PIN_MAP_TYPE_MUX_GROUP,
1103		.ctrl_dev_name = "pinctrl-foo",
1104		.function = "spi0",
1105		.group = "spi0_1_grp",
1106	},
1107	...
1108
1109This example mapping is used to switch between two positions for spi0 at
1110runtime, as described further below under the heading "Runtime pinmuxing".
1111
1112Further it is possible for one named state to affect the muxing of several
1113groups of pins, say for example in the mmc0 example above, where you can
1114additively expand the mmc0 bus from 2 to 4 to 8 pins. If we want to use all
1115three groups for a total of 2+2+4 = 8 pins (for an 8-bit MMC bus as is the
1116case), we define a mapping like this::
1117
1118	...
1119	{
1120		.dev_name = "foo-mmc.0",
1121		.name = "2bit"
1122		.type = PIN_MAP_TYPE_MUX_GROUP,
1123		.ctrl_dev_name = "pinctrl-foo",
1124		.function = "mmc0",
1125		.group = "mmc0_1_grp",
1126	},
1127	{
1128		.dev_name = "foo-mmc.0",
1129		.name = "4bit"
1130		.type = PIN_MAP_TYPE_MUX_GROUP,
1131		.ctrl_dev_name = "pinctrl-foo",
1132		.function = "mmc0",
1133		.group = "mmc0_1_grp",
1134	},
1135	{
1136		.dev_name = "foo-mmc.0",
1137		.name = "4bit"
1138		.type = PIN_MAP_TYPE_MUX_GROUP,
1139		.ctrl_dev_name = "pinctrl-foo",
1140		.function = "mmc0",
1141		.group = "mmc0_2_grp",
1142	},
1143	{
1144		.dev_name = "foo-mmc.0",
1145		.name = "8bit"
1146		.type = PIN_MAP_TYPE_MUX_GROUP,
1147		.ctrl_dev_name = "pinctrl-foo",
1148		.function = "mmc0",
1149		.group = "mmc0_1_grp",
1150	},
1151	{
1152		.dev_name = "foo-mmc.0",
1153		.name = "8bit"
1154		.type = PIN_MAP_TYPE_MUX_GROUP,
1155		.ctrl_dev_name = "pinctrl-foo",
1156		.function = "mmc0",
1157		.group = "mmc0_2_grp",
1158	},
1159	{
1160		.dev_name = "foo-mmc.0",
1161		.name = "8bit"
1162		.type = PIN_MAP_TYPE_MUX_GROUP,
1163		.ctrl_dev_name = "pinctrl-foo",
1164		.function = "mmc0",
1165		.group = "mmc0_3_grp",
1166	},
1167	...
1168
1169The result of grabbing this mapping from the device with something like
1170this (see next paragraph)::
1171
1172	p = devm_pinctrl_get(dev);
1173	s = pinctrl_lookup_state(p, "8bit");
1174	ret = pinctrl_select_state(p, s);
1175
1176or more simply::
1177
1178	p = devm_pinctrl_get_select(dev, "8bit");
1179
1180Will be that you activate all the three bottom records in the mapping at
1181once. Since they share the same name, pin controller device, function and
1182device, and since we allow multiple groups to match to a single device, they
1183all get selected, and they all get enabled and disable simultaneously by the
1184pinmux core.
1185
1186
1187Pin control requests from drivers
1188=================================
1189
1190When a device driver is about to probe the device core will automatically
1191attempt to issue pinctrl_get_select_default() on these devices.
1192This way driver writers do not need to add any of the boilerplate code
1193of the type found below. However when doing fine-grained state selection
1194and not using the "default" state, you may have to do some device driver
1195handling of the pinctrl handles and states.
1196
1197So if you just want to put the pins for a certain device into the default
1198state and be done with it, there is nothing you need to do besides
1199providing the proper mapping table. The device core will take care of
1200the rest.
1201
1202Generally it is discouraged to let individual drivers get and enable pin
1203control. So if possible, handle the pin control in platform code or some other
1204place where you have access to all the affected struct device * pointers. In
1205some cases where a driver needs to e.g. switch between different mux mappings
1206at runtime this is not possible.
1207
1208A typical case is if a driver needs to switch bias of pins from normal
1209operation and going to sleep, moving from the PINCTRL_STATE_DEFAULT to
1210PINCTRL_STATE_SLEEP at runtime, re-biasing or even re-muxing pins to save
1211current in sleep mode.
1212
1213A driver may request a certain control state to be activated, usually just the
1214default state like this::
1215
1216	#include <linux/pinctrl/consumer.h>
1217
1218	struct foo_state {
1219	struct pinctrl *p;
1220	struct pinctrl_state *s;
1221	...
1222	};
1223
1224	foo_probe()
1225	{
1226		/* Allocate a state holder named "foo" etc */
1227		struct foo_state *foo = ...;
1228
1229		foo->p = devm_pinctrl_get(&device);
1230		if (IS_ERR(foo->p)) {
1231			/* FIXME: clean up "foo" here */
1232			return PTR_ERR(foo->p);
1233		}
1234
1235		foo->s = pinctrl_lookup_state(foo->p, PINCTRL_STATE_DEFAULT);
1236		if (IS_ERR(foo->s)) {
1237			/* FIXME: clean up "foo" here */
1238			return PTR_ERR(s);
1239		}
1240
1241		ret = pinctrl_select_state(foo->s);
1242		if (ret < 0) {
1243			/* FIXME: clean up "foo" here */
1244			return ret;
1245		}
1246	}
1247
1248This get/lookup/select/put sequence can just as well be handled by bus drivers
1249if you don't want each and every driver to handle it and you know the
1250arrangement on your bus.
1251
1252The semantics of the pinctrl APIs are:
1253
1254- pinctrl_get() is called in process context to obtain a handle to all pinctrl
1255  information for a given client device. It will allocate a struct from the
1256  kernel memory to hold the pinmux state. All mapping table parsing or similar
1257  slow operations take place within this API.
1258
1259- devm_pinctrl_get() is a variant of pinctrl_get() that causes pinctrl_put()
1260  to be called automatically on the retrieved pointer when the associated
1261  device is removed. It is recommended to use this function over plain
1262  pinctrl_get().
1263
1264- pinctrl_lookup_state() is called in process context to obtain a handle to a
1265  specific state for a client device. This operation may be slow, too.
1266
1267- pinctrl_select_state() programs pin controller hardware according to the
1268  definition of the state as given by the mapping table. In theory, this is a
1269  fast-path operation, since it only involved blasting some register settings
1270  into hardware. However, note that some pin controllers may have their
1271  registers on a slow/IRQ-based bus, so client devices should not assume they
1272  can call pinctrl_select_state() from non-blocking contexts.
1273
1274- pinctrl_put() frees all information associated with a pinctrl handle.
1275
1276- devm_pinctrl_put() is a variant of pinctrl_put() that may be used to
1277  explicitly destroy a pinctrl object returned by devm_pinctrl_get().
1278  However, use of this function will be rare, due to the automatic cleanup
1279  that will occur even without calling it.
1280
1281  pinctrl_get() must be paired with a plain pinctrl_put().
1282  pinctrl_get() may not be paired with devm_pinctrl_put().
1283  devm_pinctrl_get() can optionally be paired with devm_pinctrl_put().
1284  devm_pinctrl_get() may not be paired with plain pinctrl_put().
1285
1286Usually the pin control core handled the get/put pair and call out to the
1287device drivers bookkeeping operations, like checking available functions and
1288the associated pins, whereas select_state pass on to the pin controller
1289driver which takes care of activating and/or deactivating the mux setting by
1290quickly poking some registers.
1291
1292The pins are allocated for your device when you issue the devm_pinctrl_get()
1293call, after this you should be able to see this in the debugfs listing of all
1294pins.
1295
1296NOTE: the pinctrl system will return -EPROBE_DEFER if it cannot find the
1297requested pinctrl handles, for example if the pinctrl driver has not yet
1298registered. Thus make sure that the error path in your driver gracefully
1299cleans up and is ready to retry the probing later in the startup process.
1300
1301
1302Drivers needing both pin control and GPIOs
1303==========================================
1304
1305Again, it is discouraged to let drivers lookup and select pin control states
1306themselves, but again sometimes this is unavoidable.
1307
1308So say that your driver is fetching its resources like this::
1309
1310	#include <linux/pinctrl/consumer.h>
1311	#include <linux/gpio.h>
1312
1313	struct pinctrl *pinctrl;
1314	int gpio;
1315
1316	pinctrl = devm_pinctrl_get_select_default(&dev);
1317	gpio = devm_gpio_request(&dev, 14, "foo");
1318
1319Here we first request a certain pin state and then request GPIO 14 to be
1320used. If you're using the subsystems orthogonally like this, you should
1321nominally always get your pinctrl handle and select the desired pinctrl
1322state BEFORE requesting the GPIO. This is a semantic convention to avoid
1323situations that can be electrically unpleasant, you will certainly want to
1324mux in and bias pins in a certain way before the GPIO subsystems starts to
1325deal with them.
1326
1327The above can be hidden: using the device core, the pinctrl core may be
1328setting up the config and muxing for the pins right before the device is
1329probing, nevertheless orthogonal to the GPIO subsystem.
1330
1331But there are also situations where it makes sense for the GPIO subsystem
1332to communicate directly with the pinctrl subsystem, using the latter as a
1333back-end. This is when the GPIO driver may call out to the functions
1334described in the section "Pin control interaction with the GPIO subsystem"
1335above. This only involves per-pin multiplexing, and will be completely
1336hidden behind the gpio_*() function namespace. In this case, the driver
1337need not interact with the pin control subsystem at all.
1338
1339If a pin control driver and a GPIO driver is dealing with the same pins
1340and the use cases involve multiplexing, you MUST implement the pin controller
1341as a back-end for the GPIO driver like this, unless your hardware design
1342is such that the GPIO controller can override the pin controller's
1343multiplexing state through hardware without the need to interact with the
1344pin control system.
1345
1346
1347System pin control hogging
1348==========================
1349
1350Pin control map entries can be hogged by the core when the pin controller
1351is registered. This means that the core will attempt to call pinctrl_get(),
1352lookup_state() and select_state() on it immediately after the pin control
1353device has been registered.
1354
1355This occurs for mapping table entries where the client device name is equal
1356to the pin controller device name, and the state name is PINCTRL_STATE_DEFAULT::
1357
1358	{
1359		.dev_name = "pinctrl-foo",
1360		.name = PINCTRL_STATE_DEFAULT,
1361		.type = PIN_MAP_TYPE_MUX_GROUP,
1362		.ctrl_dev_name = "pinctrl-foo",
1363		.function = "power_func",
1364	},
1365
1366Since it may be common to request the core to hog a few always-applicable
1367mux settings on the primary pin controller, there is a convenience macro for
1368this::
1369
1370	PIN_MAP_MUX_GROUP_HOG_DEFAULT("pinctrl-foo", NULL /* group */,
1371				      "power_func")
1372
1373This gives the exact same result as the above construction.
1374
1375
1376Runtime pinmuxing
1377=================
1378
1379It is possible to mux a certain function in and out at runtime, say to move
1380an SPI port from one set of pins to another set of pins. Say for example for
1381spi0 in the example above, we expose two different groups of pins for the same
1382function, but with different named in the mapping as described under
1383"Advanced mapping" above. So that for an SPI device, we have two states named
1384"pos-A" and "pos-B".
1385
1386This snippet first initializes a state object for both groups (in foo_probe()),
1387then muxes the function in the pins defined by group A, and finally muxes it in
1388on the pins defined by group B::
1389
1390	#include <linux/pinctrl/consumer.h>
1391
1392	struct pinctrl *p;
1393	struct pinctrl_state *s1, *s2;
1394
1395	foo_probe()
1396	{
1397		/* Setup */
1398		p = devm_pinctrl_get(&device);
1399		if (IS_ERR(p))
1400			...
1401
1402		s1 = pinctrl_lookup_state(foo->p, "pos-A");
1403		if (IS_ERR(s1))
1404			...
1405
1406		s2 = pinctrl_lookup_state(foo->p, "pos-B");
1407		if (IS_ERR(s2))
1408			...
1409	}
1410
1411	foo_switch()
1412	{
1413		/* Enable on position A */
1414		ret = pinctrl_select_state(s1);
1415		if (ret < 0)
1416		...
1417
1418		...
1419
1420		/* Enable on position B */
1421		ret = pinctrl_select_state(s2);
1422		if (ret < 0)
1423		...
1424
1425		...
1426	}
1427
1428The above has to be done from process context. The reservation of the pins
1429will be done when the state is activated, so in effect one specific pin
1430can be used by different functions at different times on a running system.