1 ACPI based device enumeration
2 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
3 ACPI 5 introduced a set of new resources (UartTSerialBus, I2cSerialBus,
4 SpiSerialBus, GpioIo and GpioInt) which can be used in enumerating slave
5 devices behind serial bus controllers.
7 In addition we are starting to see peripherals integrated in the
8 SoC/Chipset to appear only in ACPI namespace. These are typically devices
9 that are accessed through memory-mapped registers.
11 In order to support this and re-use the existing drivers as much as
12 possible we decided to do following:
14 o Devices that have no bus connector resource are represented as
17 o Devices behind real busses where there is a connector resource
18 are represented as struct spi_device or struct i2c_device
19 (standard UARTs are not busses so there is no struct uart_device).
21 As both ACPI and Device Tree represent a tree of devices (and their
22 resources) this implementation follows the Device Tree way as much as
25 The ACPI implementation enumerates devices behind busses (platform, SPI and
26 I2C), creates the physical devices and binds them to their ACPI handle in
29 This means that when ACPI_HANDLE(dev) returns non-NULL the device was
30 enumerated from ACPI namespace. This handle can be used to extract other
31 device-specific configuration. There is an example of this below.
35 Since we are using platform devices to represent devices that are not
36 connected to any physical bus we only need to implement a platform driver
37 for the device and add supported ACPI IDs. If this same IP-block is used on
38 some other non-ACPI platform, the driver might work out of the box or needs
41 Adding ACPI support for an existing driver should be pretty
42 straightforward. Here is the simplest example:
45 static struct acpi_device_id mydrv_acpi_match[] = {
49 MODULE_DEVICE_TABLE(acpi, mydrv_acpi_match);
52 static struct platform_driver my_driver = {
55 .acpi_match_table = ACPI_PTR(mydrv_acpi_match),
59 If the driver needs to perform more complex initialization like getting and
60 configuring GPIOs it can get its ACPI handle and extract this information
63 Currently the kernel is not able to automatically determine from which ACPI
64 device it should make the corresponding platform device so we need to add
65 the ACPI device explicitly to acpi_platform_device_ids list defined in
66 drivers/acpi/acpi_platform.c. This limitation is only for the platform
67 devices, SPI and I2C devices are created automatically as described below.
71 DMA controllers enumerated via ACPI should be registered in the system to
72 provide generic access to their resources. For example, a driver that would
73 like to be accessible to slave devices via generic API call
74 dma_request_slave_channel() must register itself at the end of the probe
77 err = devm_acpi_dma_controller_register(dev, xlate_func, dw);
78 /* Handle the error if it's not a case of !CONFIG_ACPI */
80 and implement custom xlate function if needed (usually acpi_dma_simple_xlate()
81 is enough) which converts the FixedDMA resource provided by struct
82 acpi_dma_spec into the corresponding DMA channel. A piece of code for that case
87 /* Provide necessary information for the filter_func */
91 static bool filter_func(struct dma_chan *chan, void *param)
93 /* Choose the proper channel */
97 static struct dma_chan *xlate_func(struct acpi_dma_spec *dma_spec,
98 struct acpi_dma *adma)
101 struct filter_args args;
103 /* Prepare arguments for filter_func */
105 return dma_request_channel(cap, filter_func, &args);
108 static struct dma_chan *xlate_func(struct acpi_dma_spec *dma_spec,
109 struct acpi_dma *adma)
115 dma_request_slave_channel() will call xlate_func() for each registered DMA
116 controller. In the xlate function the proper channel must be chosen based on
117 information in struct acpi_dma_spec and the properties of the controller
118 provided by struct acpi_dma.
120 Clients must call dma_request_slave_channel() with the string parameter that
121 corresponds to a specific FixedDMA resource. By default "tx" means the first
122 entry of the FixedDMA resource array, "rx" means the second entry. The table
123 below shows a layout:
128 Method (_CRS, 0, NotSerialized)
130 Name (DBUF, ResourceTemplate ()
132 FixedDMA (0x0018, 0x0004, Width32bit, _Y48)
133 FixedDMA (0x0019, 0x0005, Width32bit, )
139 So, the FixedDMA with request line 0x0018 is "tx" and next one is "rx" in
142 In robust cases the client unfortunately needs to call
143 acpi_dma_request_slave_chan_by_index() directly and therefore choose the
144 specific FixedDMA resource by its index.
146 SPI serial bus support
147 ~~~~~~~~~~~~~~~~~~~~~~
148 Slave devices behind SPI bus have SpiSerialBus resource attached to them.
149 This is extracted automatically by the SPI core and the slave devices are
150 enumerated once spi_register_master() is called by the bus driver.
152 Here is what the ACPI namespace for a SPI slave might look like:
157 Name (_CID, Package() {
162 Method (_CRS, 0, NotSerialized)
164 SPISerialBus(1, PolarityLow, FourWireMode, 8,
165 ControllerInitiated, 1000000, ClockPolarityLow,
166 ClockPhaseFirst, "\\_SB.PCI0.SPI1",)
170 The SPI device drivers only need to add ACPI IDs in a similar way than with
171 the platform device drivers. Below is an example where we add ACPI support
172 to at25 SPI eeprom driver (this is meant for the above ACPI snippet):
175 static struct acpi_device_id at25_acpi_match[] = {
179 MODULE_DEVICE_TABLE(acpi, at25_acpi_match);
182 static struct spi_driver at25_driver = {
185 .acpi_match_table = ACPI_PTR(at25_acpi_match),
189 Note that this driver actually needs more information like page size of the
190 eeprom etc. but at the time writing this there is no standard way of
191 passing those. One idea is to return this in _DSM method like:
196 Method (_DSM, 4, NotSerialized)
210 Then the at25 SPI driver can get this configuration by calling _DSM on its
213 struct acpi_buffer output = { ACPI_ALLOCATE_BUFFER, NULL };
214 struct acpi_object_list input;
217 /* Fill in the input buffer */
219 status = acpi_evaluate_object(ACPI_HANDLE(&spi->dev), "_DSM",
221 if (ACPI_FAILURE(status))
222 /* Handle the error */
224 /* Extract the data here */
226 kfree(output.pointer);
228 I2C serial bus support
229 ~~~~~~~~~~~~~~~~~~~~~~
230 The slaves behind I2C bus controller only need to add the ACPI IDs like
231 with the platform and SPI drivers. The I2C core automatically enumerates
232 any slave devices behind the controller device once the adapter is
235 Below is an example of how to add ACPI support to the existing mpu3050
239 static struct acpi_device_id mpu3050_acpi_match[] = {
243 MODULE_DEVICE_TABLE(acpi, mpu3050_acpi_match);
246 static struct i2c_driver mpu3050_i2c_driver = {
249 .owner = THIS_MODULE,
251 .of_match_table = mpu3050_of_match,
252 .acpi_match_table ACPI_PTR(mpu3050_acpi_match),
254 .probe = mpu3050_probe,
255 .remove = mpu3050_remove,
256 .id_table = mpu3050_ids,
261 ACPI 5 introduced two new resources to describe GPIO connections: GpioIo
262 and GpioInt. These resources are used be used to pass GPIO numbers used by
263 the device to the driver. For example:
265 Method (_CRS, 0, NotSerialized)
267 Name (SBUF, ResourceTemplate()
270 // Used to power on/off the device
271 GpioIo (Exclusive, PullDefault, 0x0000, 0x0000,
272 IoRestrictionOutputOnly, "\\_SB.PCI0.GPI0",
273 0x00, ResourceConsumer,,)
279 // Interrupt for the device
280 GpioInt (Edge, ActiveHigh, ExclusiveAndWake, PullNone,
281 0x0000, "\\_SB.PCI0.GPI0", 0x00, ResourceConsumer,,)
294 These GPIO numbers are controller relative and path "\\_SB.PCI0.GPI0"
295 specifies the path to the controller. In order to use these GPIOs in Linux
296 we need to translate them to the corresponding Linux GPIO descriptors.
298 There is a standard GPIO API for that and is documented in
301 In the above example we can get the corresponding two GPIO descriptors with
304 #include <linux/gpio/consumer.h>
307 struct gpio_desc *irq_desc, *power_desc;
309 irq_desc = gpiod_get_index(dev, NULL, 1);
310 if (IS_ERR(irq_desc))
313 power_desc = gpiod_get_index(dev, NULL, 0);
314 if (IS_ERR(power_desc))
317 /* Now we can use the GPIO descriptors */
319 There are also devm_* versions of these functions which release the
320 descriptors once the device is released.