About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Documentation / spi / spi-summary

Custom Search

Based on kernel version 4.9. Page generated on 2016-12-21 14:37 EST.

1	Overview of Linux kernel SPI support
2	====================================
4	02-Feb-2012
6	What is SPI?
7	------------
8	The "Serial Peripheral Interface" (SPI) is a synchronous four wire serial
9	link used to connect microcontrollers to sensors, memory, and peripherals.
10	It's a simple "de facto" standard, not complicated enough to acquire a
11	standardization body.  SPI uses a master/slave configuration.
13	The three signal wires hold a clock (SCK, often on the order of 10 MHz),
14	and parallel data lines with "Master Out, Slave In" (MOSI) or "Master In,
15	Slave Out" (MISO) signals.  (Other names are also used.)  There are four
16	clocking modes through which data is exchanged; mode-0 and mode-3 are most
17	commonly used.  Each clock cycle shifts data out and data in; the clock
18	doesn't cycle except when there is a data bit to shift.  Not all data bits
19	are used though; not every protocol uses those full duplex capabilities.
21	SPI masters use a fourth "chip select" line to activate a given SPI slave
22	device, so those three signal wires may be connected to several chips
23	in parallel.  All SPI slaves support chipselects; they are usually active
24	low signals, labeled nCSx for slave 'x' (e.g. nCS0).  Some devices have
25	other signals, often including an interrupt to the master.
27	Unlike serial busses like USB or SMBus, even low level protocols for
28	SPI slave functions are usually not interoperable between vendors
29	(except for commodities like SPI memory chips).
31	  - SPI may be used for request/response style device protocols, as with
32	    touchscreen sensors and memory chips.
34	  - It may also be used to stream data in either direction (half duplex),
35	    or both of them at the same time (full duplex).
37	  - Some devices may use eight bit words.  Others may use different word
38	    lengths, such as streams of 12-bit or 20-bit digital samples.
40	  - Words are usually sent with their most significant bit (MSB) first,
41	    but sometimes the least significant bit (LSB) goes first instead.
43	  - Sometimes SPI is used to daisy-chain devices, like shift registers.
45	In the same way, SPI slaves will only rarely support any kind of automatic
46	discovery/enumeration protocol.  The tree of slave devices accessible from
47	a given SPI master will normally be set up manually, with configuration
48	tables.
50	SPI is only one of the names used by such four-wire protocols, and
51	most controllers have no problem handling "MicroWire" (think of it as
52	half-duplex SPI, for request/response protocols), SSP ("Synchronous
53	Serial Protocol"), PSP ("Programmable Serial Protocol"), and other
54	related protocols.
56	Some chips eliminate a signal line by combining MOSI and MISO, and
57	limiting themselves to half-duplex at the hardware level.  In fact
58	some SPI chips have this signal mode as a strapping option.  These
59	can be accessed using the same programming interface as SPI, but of
60	course they won't handle full duplex transfers.  You may find such
61	chips described as using "three wire" signaling: SCK, data, nCSx.
62	(That data line is sometimes called MOMI or SISO.)
64	Microcontrollers often support both master and slave sides of the SPI
65	protocol.  This document (and Linux) currently only supports the master
66	side of SPI interactions.
69	Who uses it?  On what kinds of systems?
70	---------------------------------------
71	Linux developers using SPI are probably writing device drivers for embedded
72	systems boards.  SPI is used to control external chips, and it is also a
73	protocol supported by every MMC or SD memory card.  (The older "DataFlash"
74	cards, predating MMC cards but using the same connectors and card shape,
75	support only SPI.)  Some PC hardware uses SPI flash for BIOS code.
77	SPI slave chips range from digital/analog converters used for analog
78	sensors and codecs, to memory, to peripherals like USB controllers
79	or Ethernet adapters; and more.
81	Most systems using SPI will integrate a few devices on a mainboard.
82	Some provide SPI links on expansion connectors; in cases where no
83	dedicated SPI controller exists, GPIO pins can be used to create a
84	low speed "bitbanging" adapter.  Very few systems will "hotplug" an SPI
85	controller; the reasons to use SPI focus on low cost and simple operation,
86	and if dynamic reconfiguration is important, USB will often be a more
87	appropriate low-pincount peripheral bus.
89	Many microcontrollers that can run Linux integrate one or more I/O
90	interfaces with SPI modes.  Given SPI support, they could use MMC or SD
91	cards without needing a special purpose MMC/SD/SDIO controller.
94	I'm confused.  What are these four SPI "clock modes"?
95	-----------------------------------------------------
96	It's easy to be confused here, and the vendor documentation you'll
97	find isn't necessarily helpful.  The four modes combine two mode bits:
99	 - CPOL indicates the initial clock polarity.  CPOL=0 means the
100	   clock starts low, so the first (leading) edge is rising, and
101	   the second (trailing) edge is falling.  CPOL=1 means the clock
102	   starts high, so the first (leading) edge is falling.
104	 - CPHA indicates the clock phase used to sample data; CPHA=0 says
105	   sample on the leading edge, CPHA=1 means the trailing edge.
107	   Since the signal needs to stablize before it's sampled, CPHA=0
108	   implies that its data is written half a clock before the first
109	   clock edge.  The chipselect may have made it become available.
111	Chip specs won't always say "uses SPI mode X" in as many words,
112	but their timing diagrams will make the CPOL and CPHA modes clear.
114	In the SPI mode number, CPOL is the high order bit and CPHA is the
115	low order bit.  So when a chip's timing diagram shows the clock
116	starting low (CPOL=0) and data stabilized for sampling during the
117	trailing clock edge (CPHA=1), that's SPI mode 1.
119	Note that the clock mode is relevant as soon as the chipselect goes
120	active.  So the master must set the clock to inactive before selecting
121	a slave, and the slave can tell the chosen polarity by sampling the
122	clock level when its select line goes active.  That's why many devices
123	support for example both modes 0 and 3:  they don't care about polarity,
124	and always clock data in/out on rising clock edges.
127	How do these driver programming interfaces work?
128	------------------------------------------------
129	The <linux/spi/spi.h> header file includes kerneldoc, as does the
130	main source code, and you should certainly read that chapter of the
131	kernel API document.  This is just an overview, so you get the big
132	picture before those details.
134	SPI requests always go into I/O queues.  Requests for a given SPI device
135	are always executed in FIFO order, and complete asynchronously through
136	completion callbacks.  There are also some simple synchronous wrappers
137	for those calls, including ones for common transaction types like writing
138	a command and then reading its response.
140	There are two types of SPI driver, here called:
142	  Controller drivers ... controllers may be built into System-On-Chip
143		processors, and often support both Master and Slave roles.
144		These drivers touch hardware registers and may use DMA.
145		Or they can be PIO bitbangers, needing just GPIO pins.
147	  Protocol drivers ... these pass messages through the controller
148		driver to communicate with a Slave or Master device on the
149		other side of an SPI link.
151	So for example one protocol driver might talk to the MTD layer to export
152	data to filesystems stored on SPI flash like DataFlash; and others might
153	control audio interfaces, present touchscreen sensors as input interfaces,
154	or monitor temperature and voltage levels during industrial processing.
155	And those might all be sharing the same controller driver.
157	A "struct spi_device" encapsulates the master-side interface between
158	those two types of driver.  At this writing, Linux has no slave side
159	programming interface.
161	There is a minimal core of SPI programming interfaces, focussing on
162	using the driver model to connect controller and protocol drivers using
163	device tables provided by board specific initialization code.  SPI
164	shows up in sysfs in several locations:
166	   /sys/devices/.../CTLR ... physical node for a given SPI controller
168	   /sys/devices/.../CTLR/spiB.C ... spi_device on bus "B",
169		chipselect C, accessed through CTLR.
171	   /sys/bus/spi/devices/spiB.C ... symlink to that physical
172	   	.../CTLR/spiB.C device
174	   /sys/devices/.../CTLR/spiB.C/modalias ... identifies the driver
175		that should be used with this device (for hotplug/coldplug)
177	   /sys/bus/spi/drivers/D ... driver for one or more spi*.* devices
179	   /sys/class/spi_master/spiB ... symlink (or actual device node) to
180		a logical node which could hold class related state for the
181		controller managing bus "B".  All spiB.* devices share one
182		physical SPI bus segment, with SCLK, MOSI, and MISO.
184	Note that the actual location of the controller's class state depends
185	on whether you enabled CONFIG_SYSFS_DEPRECATED or not.  At this time,
186	the only class-specific state is the bus number ("B" in "spiB"), so
187	those /sys/class entries are only useful to quickly identify busses.
190	How does board-specific init code declare SPI devices?
191	------------------------------------------------------
192	Linux needs several kinds of information to properly configure SPI devices.
193	That information is normally provided by board-specific code, even for
194	chips that do support some of automated discovery/enumeration.
198	The first kind of information is a list of what SPI controllers exist.
199	For System-on-Chip (SOC) based boards, these will usually be platform
200	devices, and the controller may need some platform_data in order to
201	operate properly.  The "struct platform_device" will include resources
202	like the physical address of the controller's first register and its IRQ.
204	Platforms will often abstract the "register SPI controller" operation,
205	maybe coupling it with code to initialize pin configurations, so that
206	the arch/.../mach-*/board-*.c files for several boards can all share the
207	same basic controller setup code.  This is because most SOCs have several
208	SPI-capable controllers, and only the ones actually usable on a given
209	board should normally be set up and registered.
211	So for example arch/.../mach-*/board-*.c files might have code like:
213		#include <mach/spi.h>	/* for mysoc_spi_data */
215		/* if your mach-* infrastructure doesn't support kernels that can
216		 * run on multiple boards, pdata wouldn't benefit from "__init".
217		 */
218		static struct mysoc_spi_data pdata __initdata = { ... };
220		static __init board_init(void)
221		{
222			...
223			/* this board only uses SPI controller #2 */
224			mysoc_register_spi(2, &pdata);
225			...
226		}
228	And SOC-specific utility code might look something like:
230		#include <mach/spi.h>
232		static struct platform_device spi2 = { ... };
234		void mysoc_register_spi(unsigned n, struct mysoc_spi_data *pdata)
235		{
236			struct mysoc_spi_data *pdata2;
238			pdata2 = kmalloc(sizeof *pdata2, GFP_KERNEL);
239			*pdata2 = pdata;
240			...
241			if (n == 2) {
242				spi2->dev.platform_data = pdata2;
243				register_platform_device(&spi2);
245				/* also: set up pin modes so the spi2 signals are
246				 * visible on the relevant pins ... bootloaders on
247				 * production boards may already have done this, but
248				 * developer boards will often need Linux to do it.
249				 */
250			}
251			...
252		}
254	Notice how the platform_data for boards may be different, even if the
255	same SOC controller is used.  For example, on one board SPI might use
256	an external clock, where another derives the SPI clock from current
257	settings of some master clock.
262	The second kind of information is a list of what SPI slave devices exist
263	on the target board, often with some board-specific data needed for the
264	driver to work correctly.
266	Normally your arch/.../mach-*/board-*.c files would provide a small table
267	listing the SPI devices on each board.  (This would typically be only a
268	small handful.)  That might look like:
270		static struct ads7846_platform_data ads_info = {
271			.vref_delay_usecs	= 100,
272			.x_plate_ohms		= 580,
273			.y_plate_ohms		= 410,
274		};
276		static struct spi_board_info spi_board_info[] __initdata = {
277		{
278			.modalias	= "ads7846",
279			.platform_data	= &ads_info,
280			.mode		= SPI_MODE_0,
281			.irq		= GPIO_IRQ(31),
282			.max_speed_hz	= 120000 /* max sample rate at 3V */ * 16,
283			.bus_num	= 1,
284			.chip_select	= 0,
285		},
286		};
288	Again, notice how board-specific information is provided; each chip may need
289	several types.  This example shows generic constraints like the fastest SPI
290	clock to allow (a function of board voltage in this case) or how an IRQ pin
291	is wired, plus chip-specific constraints like an important delay that's
292	changed by the capacitance at one pin.
294	(There's also "controller_data", information that may be useful to the
295	controller driver.  An example would be peripheral-specific DMA tuning
296	data or chipselect callbacks.  This is stored in spi_device later.)
298	The board_info should provide enough information to let the system work
299	without the chip's driver being loaded.  The most troublesome aspect of
300	that is likely the SPI_CS_HIGH bit in the spi_device.mode field, since
301	sharing a bus with a device that interprets chipselect "backwards" is
302	not possible until the infrastructure knows how to deselect it.
304	Then your board initialization code would register that table with the SPI
305	infrastructure, so that it's available later when the SPI master controller
306	driver is registered:
308		spi_register_board_info(spi_board_info, ARRAY_SIZE(spi_board_info));
310	Like with other static board-specific setup, you won't unregister those.
312	The widely used "card" style computers bundle memory, cpu, and little else
313	onto a card that's maybe just thirty square centimeters.  On such systems,
314	your arch/.../mach-.../board-*.c file would primarily provide information
315	about the devices on the mainboard into which such a card is plugged.  That
316	certainly includes SPI devices hooked up through the card connectors!
321	Developer boards often play by different rules than product boards, and one
322	example is the potential need to hotplug SPI devices and/or controllers.
324	For those cases you might need to use spi_busnum_to_master() to look
325	up the spi bus master, and will likely need spi_new_device() to provide the
326	board info based on the board that was hotplugged.  Of course, you'd later
327	call at least spi_unregister_device() when that board is removed.
329	When Linux includes support for MMC/SD/SDIO/DataFlash cards through SPI, those
330	configurations will also be dynamic.  Fortunately, such devices all support
331	basic device identification probes, so they should hotplug normally.
334	How do I write an "SPI Protocol Driver"?
335	----------------------------------------
336	Most SPI drivers are currently kernel drivers, but there's also support
337	for userspace drivers.  Here we talk only about kernel drivers.
339	SPI protocol drivers somewhat resemble platform device drivers:
341		static struct spi_driver CHIP_driver = {
342			.driver = {
343				.name		= "CHIP",
344				.owner		= THIS_MODULE,
345				.pm		= &CHIP_pm_ops,
346			},
348			.probe		= CHIP_probe,
349			.remove		= CHIP_remove,
350		};
352	The driver core will automatically attempt to bind this driver to any SPI
353	device whose board_info gave a modalias of "CHIP".  Your probe() code
354	might look like this unless you're creating a device which is managing
355	a bus (appearing under /sys/class/spi_master).
357		static int CHIP_probe(struct spi_device *spi)
358		{
359			struct CHIP			*chip;
360			struct CHIP_platform_data	*pdata;
362			/* assuming the driver requires board-specific data: */
363			pdata = &spi->dev.platform_data;
364			if (!pdata)
365				return -ENODEV;
367			/* get memory for driver's per-chip state */
368			chip = kzalloc(sizeof *chip, GFP_KERNEL);
369			if (!chip)
370				return -ENOMEM;
371			spi_set_drvdata(spi, chip);
373			... etc
374			return 0;
375		}
377	As soon as it enters probe(), the driver may issue I/O requests to
378	the SPI device using "struct spi_message".  When remove() returns,
379	or after probe() fails, the driver guarantees that it won't submit
380	any more such messages.
382	  - An spi_message is a sequence of protocol operations, executed
383	    as one atomic sequence.  SPI driver controls include:
385	      + when bidirectional reads and writes start ... by how its
386	        sequence of spi_transfer requests is arranged;
388	      + which I/O buffers are used ... each spi_transfer wraps a
389	        buffer for each transfer direction, supporting full duplex
390	        (two pointers, maybe the same one in both cases) and half
391	        duplex (one pointer is NULL) transfers;
393	      + optionally defining short delays after transfers ... using
394	        the spi_transfer.delay_usecs setting (this delay can be the
395	        only protocol effect, if the buffer length is zero);
397	      + whether the chipselect becomes inactive after a transfer and
398	        any delay ... by using the spi_transfer.cs_change flag;
400	      + hinting whether the next message is likely to go to this same
401	        device ... using the spi_transfer.cs_change flag on the last
402		transfer in that atomic group, and potentially saving costs
403		for chip deselect and select operations.
405	  - Follow standard kernel rules, and provide DMA-safe buffers in
406	    your messages.  That way controller drivers using DMA aren't forced
407	    to make extra copies unless the hardware requires it (e.g. working
408	    around hardware errata that force the use of bounce buffering).
410	    If standard dma_map_single() handling of these buffers is inappropriate,
411	    you can use spi_message.is_dma_mapped to tell the controller driver
412	    that you've already provided the relevant DMA addresses.
414	  - The basic I/O primitive is spi_async().  Async requests may be
415	    issued in any context (irq handler, task, etc) and completion
416	    is reported using a callback provided with the message.
417	    After any detected error, the chip is deselected and processing
418	    of that spi_message is aborted.
420	  - There are also synchronous wrappers like spi_sync(), and wrappers
421	    like spi_read(), spi_write(), and spi_write_then_read().  These
422	    may be issued only in contexts that may sleep, and they're all
423	    clean (and small, and "optional") layers over spi_async().
425	  - The spi_write_then_read() call, and convenience wrappers around
426	    it, should only be used with small amounts of data where the
427	    cost of an extra copy may be ignored.  It's designed to support
428	    common RPC-style requests, such as writing an eight bit command
429	    and reading a sixteen bit response -- spi_w8r16() being one its
430	    wrappers, doing exactly that.
432	Some drivers may need to modify spi_device characteristics like the
433	transfer mode, wordsize, or clock rate.  This is done with spi_setup(),
434	which would normally be called from probe() before the first I/O is
435	done to the device.  However, that can also be called at any time
436	that no message is pending for that device.
438	While "spi_device" would be the bottom boundary of the driver, the
439	upper boundaries might include sysfs (especially for sensor readings),
440	the input layer, ALSA, networking, MTD, the character device framework,
441	or other Linux subsystems.
443	Note that there are two types of memory your driver must manage as part
444	of interacting with SPI devices.
446	  - I/O buffers use the usual Linux rules, and must be DMA-safe.
447	    You'd normally allocate them from the heap or free page pool.
448	    Don't use the stack, or anything that's declared "static".
450	  - The spi_message and spi_transfer metadata used to glue those
451	    I/O buffers into a group of protocol transactions.  These can
452	    be allocated anywhere it's convenient, including as part of
453	    other allocate-once driver data structures.  Zero-init these.
455	If you like, spi_message_alloc() and spi_message_free() convenience
456	routines are available to allocate and zero-initialize an spi_message
457	with several transfers.
460	How do I write an "SPI Master Controller Driver"?
461	-------------------------------------------------
462	An SPI controller will probably be registered on the platform_bus; write
463	a driver to bind to the device, whichever bus is involved.
465	The main task of this type of driver is to provide an "spi_master".
466	Use spi_alloc_master() to allocate the master, and spi_master_get_devdata()
467	to get the driver-private data allocated for that device.
469		struct spi_master	*master;
470		struct CONTROLLER	*c;
472		master = spi_alloc_master(dev, sizeof *c);
473		if (!master)
474			return -ENODEV;
476		c = spi_master_get_devdata(master);
478	The driver will initialize the fields of that spi_master, including the
479	bus number (maybe the same as the platform device ID) and three methods
480	used to interact with the SPI core and SPI protocol drivers.  It will
481	also initialize its own internal state.  (See below about bus numbering
482	and those methods.)
484	After you initialize the spi_master, then use spi_register_master() to
485	publish it to the rest of the system. At that time, device nodes for the
486	controller and any predeclared spi devices will be made available, and
487	the driver model core will take care of binding them to drivers.
489	If you need to remove your SPI controller driver, spi_unregister_master()
490	will reverse the effect of spi_register_master().
495	Bus numbering is important, since that's how Linux identifies a given
496	SPI bus (shared SCK, MOSI, MISO).  Valid bus numbers start at zero.  On
497	SOC systems, the bus numbers should match the numbers defined by the chip
498	manufacturer.  For example, hardware controller SPI2 would be bus number 2,
499	and spi_board_info for devices connected to it would use that number.
501	If you don't have such hardware-assigned bus number, and for some reason
502	you can't just assign them, then provide a negative bus number.  That will
503	then be replaced by a dynamically assigned number. You'd then need to treat
504	this as a non-static configuration (see above).
509	    master->setup(struct spi_device *spi)
510		This sets up the device clock rate, SPI mode, and word sizes.
511		Drivers may change the defaults provided by board_info, and then
512		call spi_setup(spi) to invoke this routine.  It may sleep.
514		Unless each SPI slave has its own configuration registers, don't
515		change them right away ... otherwise drivers could corrupt I/O
516		that's in progress for other SPI devices.
518			** BUG ALERT:  for some reason the first version of
519			** many spi_master drivers seems to get this wrong.
520			** When you code setup(), ASSUME that the controller
521			** is actively processing transfers for another device.
523	    master->cleanup(struct spi_device *spi)
524		Your controller driver may use spi_device.controller_state to hold
525		state it dynamically associates with that device.  If you do that,
526		be sure to provide the cleanup() method to free that state.
528	    master->prepare_transfer_hardware(struct spi_master *master)
529		This will be called by the queue mechanism to signal to the driver
530		that a message is coming in soon, so the subsystem requests the
531		driver to prepare the transfer hardware by issuing this call.
532		This may sleep.
534	    master->unprepare_transfer_hardware(struct spi_master *master)
535		This will be called by the queue mechanism to signal to the driver
536		that there are no more messages pending in the queue and it may
537		relax the hardware (e.g. by power management calls). This may sleep.
539	    master->transfer_one_message(struct spi_master *master,
540					 struct spi_message *mesg)
541		The subsystem calls the driver to transfer a single message while
542		queuing transfers that arrive in the meantime. When the driver is
543		finished with this message, it must call
544		spi_finalize_current_message() so the subsystem can issue the next
545		message. This may sleep.
547	    master->transfer_one(struct spi_master *master, struct spi_device *spi,
548				 struct spi_transfer *transfer)
549		The subsystem calls the driver to transfer a single transfer while
550		queuing transfers that arrive in the meantime. When the driver is
551		finished with this transfer, it must call
552		spi_finalize_current_transfer() so the subsystem can issue the next
553		transfer. This may sleep. Note: transfer_one and transfer_one_message
554		are mutually exclusive; when both are set, the generic subsystem does
555		not call your transfer_one callback.
557		Return values:
558		negative errno: error
559		0: transfer is finished
560		1: transfer is still in progress
564	    master->transfer(struct spi_device *spi, struct spi_message *message)
565		This must not sleep. Its responsibility is to arrange that the
566		transfer happens and its complete() callback is issued. The two
567		will normally happen later, after other transfers complete, and
568		if the controller is idle it will need to be kickstarted. This
569		method is not used on queued controllers and must be NULL if
570		transfer_one_message() and (un)prepare_transfer_hardware() are
571		implemented.
576	If you are happy with the standard queueing mechanism provided by the
577	SPI subsystem, just implement the queued methods specified above. Using
578	the message queue has the upside of centralizing a lot of code and
579	providing pure process-context execution of methods. The message queue
580	can also be elevated to realtime priority on high-priority SPI traffic.
582	Unless the queueing mechanism in the SPI subsystem is selected, the bulk
583	of the driver will be managing the I/O queue fed by the now deprecated
584	function transfer().
586	That queue could be purely conceptual.  For example, a driver used only
587	for low-frequency sensor access might be fine using synchronous PIO.
589	But the queue will probably be very real, using message->queue, PIO,
590	often DMA (especially if the root filesystem is in SPI flash), and
591	execution contexts like IRQ handlers, tasklets, or workqueues (such
592	as keventd).  Your driver can be as fancy, or as simple, as you need.
593	Such a transfer() method would normally just add the message to a
594	queue, and then start some asynchronous transfer engine (unless it's
595	already running).
599	---------
600	Contributors to Linux-SPI discussions include (in alphabetical order,
601	by last name):
603	Mark Brown
604	David Brownell
605	Russell King
606	Grant Likely
607	Dmitry Pervushin
608	Stephen Street
609	Mark Underwood
610	Andrew Victor
611	Linus Walleij
612	Vitaly Wool
Hide Line Numbers
About Kernel Documentation Linux Kernel Contact Linux Resources Linux Blog

Information is copyright its respective author. All material is available from the Linux Kernel Source distributed under a GPL License. This page is provided as a free service by mjmwired.net.