blob: 06141cb200fa0734df549ebed4f6536aeba4409f [file] [log] [blame]
David Gibsonc125a182006-02-01 03:05:22 -08001 Booting the Linux/ppc kernel without Open Firmware
2 --------------------------------------------------
3
David Gibsonc125a182006-02-01 03:05:22 -08004(c) 2005 Benjamin Herrenschmidt <benh at kernel.crashing.org>,
5 IBM Corp.
6(c) 2005 Becky Bruce <becky.bruce at freescale.com>,
7 Freescale Semiconductor, FSL SOC and 32-bit additions
Vitaly Wool28f9ec32006-11-20 16:32:39 +03008(c) 2006 MontaVista Software, Inc.
9 Flash chip node definition
David Gibsonc125a182006-02-01 03:05:22 -080010
Stuart Yoder5e1e9ba2007-06-06 04:29:14 +100011Table of Contents
12=================
13
14 I - Introduction
15 1) Entry point for arch/powerpc
16 2) Board support
17
18 II - The DT block format
19 1) Header
20 2) Device tree generalities
21 3) Device tree "structure" block
22 4) Device tree "strings" block
23
24 III - Required content of the device tree
25 1) Note about cells and address representation
26 2) Note about "compatible" properties
27 3) Note about "name" properties
28 4) Note about node and property names and character set
29 5) Required nodes and properties
30 a) The root node
31 b) The /cpus node
32 c) The /cpus/* nodes
33 d) the /memory node(s)
34 e) The /chosen node
35 f) the /soc<SOCname> node
36
37 IV - "dtc", the device tree compiler
38
39 V - Recommendations for a bootloader
40
41 VI - System-on-a-chip devices and nodes
42 1) Defining child nodes of an SOC
43 2) Representing devices without a current OF specification
44 a) MDIO IO device
45 c) PHY nodes
46 b) Gianfar-compatible ethernet nodes
47 d) Interrupt controllers
48 e) I2C
49 f) Freescale SOC USB controllers
50 g) Freescale SOC SEC Security Engines
51 h) Board Control and Status (BCSR)
52 i) Freescale QUICC Engine module (QE)
53 g) Flash chip nodes
Roy Zang3b824f82007-06-19 15:19:18 +080054 k) Global Utilities Block
Stuart Yoder5e1e9ba2007-06-06 04:29:14 +100055
56 VII - Specifying interrupt information for devices
57 1) interrupts property
58 2) interrupt-parent property
59 3) OpenPIC Interrupt Controllers
60 4) ISA Interrupt Controllers
61
62 Appendix A - Sample SOC node for MPC8540
63
64
65Revision Information
66====================
67
David Gibsonc125a182006-02-01 03:05:22 -080068 May 18, 2005: Rev 0.1 - Initial draft, no chapter III yet.
69
70 May 19, 2005: Rev 0.2 - Add chapter III and bits & pieces here or
71 clarifies the fact that a lot of things are
72 optional, the kernel only requires a very
73 small device tree, though it is encouraged
74 to provide an as complete one as possible.
75
76 May 24, 2005: Rev 0.3 - Precise that DT block has to be in RAM
77 - Misc fixes
78 - Define version 3 and new format version 16
79 for the DT block (version 16 needs kernel
80 patches, will be fwd separately).
81 String block now has a size, and full path
82 is replaced by unit name for more
83 compactness.
84 linux,phandle is made optional, only nodes
85 that are referenced by other nodes need it.
86 "name" property is now automatically
87 deduced from the unit name
88
89 June 1, 2005: Rev 0.4 - Correct confusion between OF_DT_END and
90 OF_DT_END_NODE in structure definition.
91 - Change version 16 format to always align
92 property data to 4 bytes. Since tokens are
93 already aligned, that means no specific
Matt LaPlante5d3f0832006-11-30 05:21:10 +010094 required alignment between property size
David Gibsonc125a182006-02-01 03:05:22 -080095 and property data. The old style variable
96 alignment would make it impossible to do
97 "simple" insertion of properties using
Domen Puncer5dd60162007-03-02 21:44:45 +110098 memmove (thanks Milton for
David Gibsonc125a182006-02-01 03:05:22 -080099 noticing). Updated kernel patch as well
Matt LaPlante5d3f0832006-11-30 05:21:10 +0100100 - Correct a few more alignment constraints
David Gibsonc125a182006-02-01 03:05:22 -0800101 - Add a chapter about the device-tree
102 compiler and the textural representation of
103 the tree that can be "compiled" by dtc.
104
David Gibsonc125a182006-02-01 03:05:22 -0800105 November 21, 2005: Rev 0.5
106 - Additions/generalizations for 32-bit
107 - Changed to reflect the new arch/powerpc
108 structure
109 - Added chapter VI
110
111
112 ToDo:
113 - Add some definitions of interrupt tree (simple/complex)
Domen Puncer5dd60162007-03-02 21:44:45 +1100114 - Add some definitions for PCI host bridges
David Gibsonc125a182006-02-01 03:05:22 -0800115 - Add some common address format examples
116 - Add definitions for standard properties and "compatible"
117 names for cells that are not already defined by the existing
118 OF spec.
119 - Compare FSL SOC use of PCI to standard and make sure no new
120 node definition required.
121 - Add more information about node definitions for SOC devices
122 that currently have no standard, like the FSL CPM.
123
124
125I - Introduction
126================
127
128During the recent development of the Linux/ppc64 kernel, and more
129specifically, the addition of new platform types outside of the old
130IBM pSeries/iSeries pair, it was decided to enforce some strict rules
131regarding the kernel entry and bootloader <-> kernel interfaces, in
132order to avoid the degeneration that had become the ppc32 kernel entry
133point and the way a new platform should be added to the kernel. The
134legacy iSeries platform breaks those rules as it predates this scheme,
135but no new board support will be accepted in the main tree that
136doesn't follows them properly. In addition, since the advent of the
137arch/powerpc merged architecture for ppc32 and ppc64, new 32-bit
138platforms and 32-bit platforms which move into arch/powerpc will be
139required to use these rules as well.
140
141The main requirement that will be defined in more detail below is
142the presence of a device-tree whose format is defined after Open
143Firmware specification. However, in order to make life easier
144to embedded board vendors, the kernel doesn't require the device-tree
145to represent every device in the system and only requires some nodes
146and properties to be present. This will be described in detail in
147section III, but, for example, the kernel does not require you to
148create a node for every PCI device in the system. It is a requirement
149to have a node for PCI host bridges in order to provide interrupt
150routing informations and memory/IO ranges, among others. It is also
151recommended to define nodes for on chip devices and other busses that
152don't specifically fit in an existing OF specification. This creates a
153great flexibility in the way the kernel can then probe those and match
154drivers to device, without having to hard code all sorts of tables. It
155also makes it more flexible for board vendors to do minor hardware
156upgrades without significantly impacting the kernel code or cluttering
157it with special cases.
158
159
1601) Entry point for arch/powerpc
161-------------------------------
162
163 There is one and one single entry point to the kernel, at the start
164 of the kernel image. That entry point supports two calling
165 conventions:
166
167 a) Boot from Open Firmware. If your firmware is compatible
168 with Open Firmware (IEEE 1275) or provides an OF compatible
169 client interface API (support for "interpret" callback of
170 forth words isn't required), you can enter the kernel with:
171
172 r5 : OF callback pointer as defined by IEEE 1275
Domen Puncer5dd60162007-03-02 21:44:45 +1100173 bindings to powerpc. Only the 32-bit client interface
David Gibsonc125a182006-02-01 03:05:22 -0800174 is currently supported
175
176 r3, r4 : address & length of an initrd if any or 0
177
178 The MMU is either on or off; the kernel will run the
179 trampoline located in arch/powerpc/kernel/prom_init.c to
180 extract the device-tree and other information from open
181 firmware and build a flattened device-tree as described
182 in b). prom_init() will then re-enter the kernel using
183 the second method. This trampoline code runs in the
184 context of the firmware, which is supposed to handle all
185 exceptions during that time.
186
187 b) Direct entry with a flattened device-tree block. This entry
188 point is called by a) after the OF trampoline and can also be
189 called directly by a bootloader that does not support the Open
190 Firmware client interface. It is also used by "kexec" to
191 implement "hot" booting of a new kernel from a previous
192 running one. This method is what I will describe in more
193 details in this document, as method a) is simply standard Open
194 Firmware, and thus should be implemented according to the
195 various standard documents defining it and its binding to the
196 PowerPC platform. The entry point definition then becomes:
197
198 r3 : physical pointer to the device-tree block
199 (defined in chapter II) in RAM
200
201 r4 : physical pointer to the kernel itself. This is
202 used by the assembly code to properly disable the MMU
203 in case you are entering the kernel with MMU enabled
204 and a non-1:1 mapping.
205
Matt LaPlante2fe0ae72006-10-03 22:50:39 +0200206 r5 : NULL (as to differentiate with method a)
David Gibsonc125a182006-02-01 03:05:22 -0800207
208 Note about SMP entry: Either your firmware puts your other
209 CPUs in some sleep loop or spin loop in ROM where you can get
210 them out via a soft reset or some other means, in which case
211 you don't need to care, or you'll have to enter the kernel
212 with all CPUs. The way to do that with method b) will be
213 described in a later revision of this document.
214
215
2162) Board support
217----------------
218
21964-bit kernels:
220
221 Board supports (platforms) are not exclusive config options. An
222 arbitrary set of board supports can be built in a single kernel
223 image. The kernel will "know" what set of functions to use for a
224 given platform based on the content of the device-tree. Thus, you
225 should:
226
227 a) add your platform support as a _boolean_ option in
228 arch/powerpc/Kconfig, following the example of PPC_PSERIES,
229 PPC_PMAC and PPC_MAPLE. The later is probably a good
230 example of a board support to start from.
231
232 b) create your main platform file as
233 "arch/powerpc/platforms/myplatform/myboard_setup.c" and add it
234 to the Makefile under the condition of your CONFIG_
235 option. This file will define a structure of type "ppc_md"
236 containing the various callbacks that the generic code will
237 use to get to your platform specific code
238
239 c) Add a reference to your "ppc_md" structure in the
240 "machines" table in arch/powerpc/kernel/setup_64.c if you are
241 a 64-bit platform.
242
243 d) request and get assigned a platform number (see PLATFORM_*
244 constants in include/asm-powerpc/processor.h
245
24632-bit embedded kernels:
247
248 Currently, board support is essentially an exclusive config option.
249 The kernel is configured for a single platform. Part of the reason
250 for this is to keep kernels on embedded systems small and efficient;
251 part of this is due to the fact the code is already that way. In the
252 future, a kernel may support multiple platforms, but only if the
Domen Puncer5dd60162007-03-02 21:44:45 +1100253 platforms feature the same core architecture. A single kernel build
David Gibsonc125a182006-02-01 03:05:22 -0800254 cannot support both configurations with Book E and configurations
255 with classic Powerpc architectures.
256
257 32-bit embedded platforms that are moved into arch/powerpc using a
258 flattened device tree should adopt the merged tree practice of
259 setting ppc_md up dynamically, even though the kernel is currently
260 built with support for only a single platform at a time. This allows
261 unification of the setup code, and will make it easier to go to a
262 multiple-platform-support model in the future.
263
264NOTE: I believe the above will be true once Ben's done with the merge
265of the boot sequences.... someone speak up if this is wrong!
266
267 To add a 32-bit embedded platform support, follow the instructions
268 for 64-bit platforms above, with the exception that the Kconfig
269 option should be set up such that the kernel builds exclusively for
270 the platform selected. The processor type for the platform should
271 enable another config option to select the specific board
272 supported.
273
Domen Puncer5dd60162007-03-02 21:44:45 +1100274NOTE: If Ben doesn't merge the setup files, may need to change this to
David Gibsonc125a182006-02-01 03:05:22 -0800275point to setup_32.c
276
277
278 I will describe later the boot process and various callbacks that
279 your platform should implement.
280
281
282II - The DT block format
283========================
284
285
286This chapter defines the actual format of the flattened device-tree
287passed to the kernel. The actual content of it and kernel requirements
288are described later. You can find example of code manipulating that
289format in various places, including arch/powerpc/kernel/prom_init.c
290which will generate a flattened device-tree from the Open Firmware
291representation, or the fs2dt utility which is part of the kexec tools
292which will generate one from a filesystem representation. It is
293expected that a bootloader like uboot provides a bit more support,
294that will be discussed later as well.
295
296Note: The block has to be in main memory. It has to be accessible in
297both real mode and virtual mode with no mapping other than main
298memory. If you are writing a simple flash bootloader, it should copy
299the block to RAM before passing it to the kernel.
300
301
3021) Header
303---------
304
305 The kernel is entered with r3 pointing to an area of memory that is
Matt LaPlanted6bc8ac2006-10-03 22:54:15 +0200306 roughly described in include/asm-powerpc/prom.h by the structure
David Gibsonc125a182006-02-01 03:05:22 -0800307 boot_param_header:
308
309struct boot_param_header {
310 u32 magic; /* magic word OF_DT_HEADER */
311 u32 totalsize; /* total size of DT block */
312 u32 off_dt_struct; /* offset to structure */
313 u32 off_dt_strings; /* offset to strings */
314 u32 off_mem_rsvmap; /* offset to memory reserve map
Domen Puncer5dd60162007-03-02 21:44:45 +1100315 */
David Gibsonc125a182006-02-01 03:05:22 -0800316 u32 version; /* format version */
317 u32 last_comp_version; /* last compatible version */
318
319 /* version 2 fields below */
320 u32 boot_cpuid_phys; /* Which physical CPU id we're
321 booting on */
322 /* version 3 fields below */
323 u32 size_dt_strings; /* size of the strings block */
David Gibson0e0293c2007-03-14 11:50:40 +1100324
325 /* version 17 fields below */
326 u32 size_dt_struct; /* size of the DT structure block */
David Gibsonc125a182006-02-01 03:05:22 -0800327};
328
329 Along with the constants:
330
331/* Definitions used by the flattened device tree */
332#define OF_DT_HEADER 0xd00dfeed /* 4: version,
333 4: total size */
334#define OF_DT_BEGIN_NODE 0x1 /* Start node: full name
Domen Puncer5dd60162007-03-02 21:44:45 +1100335 */
David Gibsonc125a182006-02-01 03:05:22 -0800336#define OF_DT_END_NODE 0x2 /* End node */
337#define OF_DT_PROP 0x3 /* Property: name off,
338 size, content */
339#define OF_DT_END 0x9
340
341 All values in this header are in big endian format, the various
342 fields in this header are defined more precisely below. All
343 "offset" values are in bytes from the start of the header; that is
344 from the value of r3.
345
346 - magic
347
348 This is a magic value that "marks" the beginning of the
349 device-tree block header. It contains the value 0xd00dfeed and is
350 defined by the constant OF_DT_HEADER
351
352 - totalsize
353
354 This is the total size of the DT block including the header. The
355 "DT" block should enclose all data structures defined in this
356 chapter (who are pointed to by offsets in this header). That is,
357 the device-tree structure, strings, and the memory reserve map.
358
359 - off_dt_struct
360
361 This is an offset from the beginning of the header to the start
362 of the "structure" part the device tree. (see 2) device tree)
363
364 - off_dt_strings
365
366 This is an offset from the beginning of the header to the start
367 of the "strings" part of the device-tree
368
369 - off_mem_rsvmap
370
371 This is an offset from the beginning of the header to the start
Domen Puncer5dd60162007-03-02 21:44:45 +1100372 of the reserved memory map. This map is a list of pairs of 64-
David Gibsonc125a182006-02-01 03:05:22 -0800373 bit integers. Each pair is a physical address and a size. The
David Gibsonc125a182006-02-01 03:05:22 -0800374 list is terminated by an entry of size 0. This map provides the
375 kernel with a list of physical memory areas that are "reserved"
376 and thus not to be used for memory allocations, especially during
377 early initialization. The kernel needs to allocate memory during
378 boot for things like un-flattening the device-tree, allocating an
379 MMU hash table, etc... Those allocations must be done in such a
380 way to avoid overriding critical things like, on Open Firmware
381 capable machines, the RTAS instance, or on some pSeries, the TCE
382 tables used for the iommu. Typically, the reserve map should
383 contain _at least_ this DT block itself (header,total_size). If
384 you are passing an initrd to the kernel, you should reserve it as
385 well. You do not need to reserve the kernel image itself. The map
Domen Puncer5dd60162007-03-02 21:44:45 +1100386 should be 64-bit aligned.
David Gibsonc125a182006-02-01 03:05:22 -0800387
388 - version
389
390 This is the version of this structure. Version 1 stops
391 here. Version 2 adds an additional field boot_cpuid_phys.
392 Version 3 adds the size of the strings block, allowing the kernel
393 to reallocate it easily at boot and free up the unused flattened
394 structure after expansion. Version 16 introduces a new more
395 "compact" format for the tree itself that is however not backward
David Gibson0e0293c2007-03-14 11:50:40 +1100396 compatible. Version 17 adds an additional field, size_dt_struct,
397 allowing it to be reallocated or moved more easily (this is
398 particularly useful for bootloaders which need to make
399 adjustments to a device tree based on probed information). You
400 should always generate a structure of the highest version defined
401 at the time of your implementation. Currently that is version 17,
402 unless you explicitly aim at being backward compatible.
David Gibsonc125a182006-02-01 03:05:22 -0800403
404 - last_comp_version
405
406 Last compatible version. This indicates down to what version of
407 the DT block you are backward compatible. For example, version 2
408 is backward compatible with version 1 (that is, a kernel build
409 for version 1 will be able to boot with a version 2 format). You
410 should put a 1 in this field if you generate a device tree of
David Gibson0e0293c2007-03-14 11:50:40 +1100411 version 1 to 3, or 16 if you generate a tree of version 16 or 17
David Gibsonc125a182006-02-01 03:05:22 -0800412 using the new unit name format.
413
414 - boot_cpuid_phys
415
416 This field only exist on version 2 headers. It indicate which
417 physical CPU ID is calling the kernel entry point. This is used,
418 among others, by kexec. If you are on an SMP system, this value
419 should match the content of the "reg" property of the CPU node in
420 the device-tree corresponding to the CPU calling the kernel entry
421 point (see further chapters for more informations on the required
422 device-tree contents)
423
David Gibson0e0293c2007-03-14 11:50:40 +1100424 - size_dt_strings
425
426 This field only exists on version 3 and later headers. It
427 gives the size of the "strings" section of the device tree (which
428 starts at the offset given by off_dt_strings).
429
430 - size_dt_struct
431
432 This field only exists on version 17 and later headers. It gives
433 the size of the "structure" section of the device tree (which
434 starts at the offset given by off_dt_struct).
David Gibsonc125a182006-02-01 03:05:22 -0800435
436 So the typical layout of a DT block (though the various parts don't
437 need to be in that order) looks like this (addresses go from top to
438 bottom):
439
440
441 ------------------------------
442 r3 -> | struct boot_param_header |
443 ------------------------------
444 | (alignment gap) (*) |
445 ------------------------------
446 | memory reserve map |
447 ------------------------------
448 | (alignment gap) |
449 ------------------------------
450 | |
451 | device-tree structure |
452 | |
453 ------------------------------
454 | (alignment gap) |
455 ------------------------------
456 | |
457 | device-tree strings |
458 | |
459 -----> ------------------------------
460 |
461 |
462 --- (r3 + totalsize)
463
464 (*) The alignment gaps are not necessarily present; their presence
465 and size are dependent on the various alignment requirements of
466 the individual data blocks.
467
468
4692) Device tree generalities
470---------------------------
471
472This device-tree itself is separated in two different blocks, a
473structure block and a strings block. Both need to be aligned to a 4
474byte boundary.
475
476First, let's quickly describe the device-tree concept before detailing
477the storage format. This chapter does _not_ describe the detail of the
478required types of nodes & properties for the kernel, this is done
479later in chapter III.
480
481The device-tree layout is strongly inherited from the definition of
482the Open Firmware IEEE 1275 device-tree. It's basically a tree of
483nodes, each node having two or more named properties. A property can
484have a value or not.
485
486It is a tree, so each node has one and only one parent except for the
487root node who has no parent.
488
489A node has 2 names. The actual node name is generally contained in a
490property of type "name" in the node property list whose value is a
491zero terminated string and is mandatory for version 1 to 3 of the
David Gibson0e0293c2007-03-14 11:50:40 +1100492format definition (as it is in Open Firmware). Version 16 makes it
David Gibsonc125a182006-02-01 03:05:22 -0800493optional as it can generate it from the unit name defined below.
494
Matt LaPlante2fe0ae72006-10-03 22:50:39 +0200495There is also a "unit name" that is used to differentiate nodes with
David Gibsonc125a182006-02-01 03:05:22 -0800496the same name at the same level, it is usually made of the node
Matt LaPlante2fe0ae72006-10-03 22:50:39 +0200497names, the "@" sign, and a "unit address", which definition is
David Gibsonc125a182006-02-01 03:05:22 -0800498specific to the bus type the node sits on.
499
500The unit name doesn't exist as a property per-se but is included in
501the device-tree structure. It is typically used to represent "path" in
502the device-tree. More details about the actual format of these will be
503below.
504
505The kernel powerpc generic code does not make any formal use of the
506unit address (though some board support code may do) so the only real
507requirement here for the unit address is to ensure uniqueness of
508the node unit name at a given level of the tree. Nodes with no notion
509of address and no possible sibling of the same name (like /memory or
510/cpus) may omit the unit address in the context of this specification,
511or use the "@0" default unit address. The unit name is used to define
512a node "full path", which is the concatenation of all parent node
513unit names separated with "/".
514
515The root node doesn't have a defined name, and isn't required to have
516a name property either if you are using version 3 or earlier of the
517format. It also has no unit address (no @ symbol followed by a unit
518address). The root node unit name is thus an empty string. The full
519path to the root node is "/".
520
521Every node which actually represents an actual device (that is, a node
522which isn't only a virtual "container" for more nodes, like "/cpus"
523is) is also required to have a "device_type" property indicating the
524type of node .
525
526Finally, every node that can be referenced from a property in another
527node is required to have a "linux,phandle" property. Real open
528firmware implementations provide a unique "phandle" value for every
529node that the "prom_init()" trampoline code turns into
530"linux,phandle" properties. However, this is made optional if the
531flattened device tree is used directly. An example of a node
532referencing another node via "phandle" is when laying out the
533interrupt tree which will be described in a further version of this
534document.
535
Domen Puncer5dd60162007-03-02 21:44:45 +1100536This "linux, phandle" property is a 32-bit value that uniquely
David Gibsonc125a182006-02-01 03:05:22 -0800537identifies a node. You are free to use whatever values or system of
538values, internal pointers, or whatever to generate these, the only
539requirement is that every node for which you provide that property has
540a unique value for it.
541
542Here is an example of a simple device-tree. In this example, an "o"
543designates a node followed by the node unit name. Properties are
544presented with their name followed by their content. "content"
545represents an ASCII string (zero terminated) value, while <content>
Domen Puncer5dd60162007-03-02 21:44:45 +1100546represents a 32-bit hexadecimal value. The various nodes in this
David Gibsonc125a182006-02-01 03:05:22 -0800547example will be discussed in a later chapter. At this point, it is
548only meant to give you a idea of what a device-tree looks like. I have
549purposefully kept the "name" and "linux,phandle" properties which
550aren't necessary in order to give you a better idea of what the tree
551looks like in practice.
552
553 / o device-tree
554 |- name = "device-tree"
555 |- model = "MyBoardName"
556 |- compatible = "MyBoardFamilyName"
557 |- #address-cells = <2>
558 |- #size-cells = <2>
559 |- linux,phandle = <0>
560 |
561 o cpus
562 | | - name = "cpus"
563 | | - linux,phandle = <1>
564 | | - #address-cells = <1>
565 | | - #size-cells = <0>
566 | |
567 | o PowerPC,970@0
568 | |- name = "PowerPC,970"
569 | |- device_type = "cpu"
570 | |- reg = <0>
571 | |- clock-frequency = <5f5e1000>
Timur Tabi32aed2a2007-02-14 15:29:07 -0600572 | |- 64-bit
David Gibsonc125a182006-02-01 03:05:22 -0800573 | |- linux,phandle = <2>
574 |
575 o memory@0
576 | |- name = "memory"
577 | |- device_type = "memory"
578 | |- reg = <00000000 00000000 00000000 20000000>
579 | |- linux,phandle = <3>
580 |
581 o chosen
582 |- name = "chosen"
583 |- bootargs = "root=/dev/sda2"
David Gibsonc125a182006-02-01 03:05:22 -0800584 |- linux,phandle = <4>
585
586This tree is almost a minimal tree. It pretty much contains the
587minimal set of required nodes and properties to boot a linux kernel;
588that is, some basic model informations at the root, the CPUs, and the
589physical memory layout. It also includes misc information passed
590through /chosen, like in this example, the platform type (mandatory)
591and the kernel command line arguments (optional).
592
Timur Tabi32aed2a2007-02-14 15:29:07 -0600593The /cpus/PowerPC,970@0/64-bit property is an example of a
David Gibsonc125a182006-02-01 03:05:22 -0800594property without a value. All other properties have a value. The
595significance of the #address-cells and #size-cells properties will be
596explained in chapter IV which defines precisely the required nodes and
597properties and their content.
598
599
6003) Device tree "structure" block
601
602The structure of the device tree is a linearized tree structure. The
603"OF_DT_BEGIN_NODE" token starts a new node, and the "OF_DT_END_NODE"
604ends that node definition. Child nodes are simply defined before
605"OF_DT_END_NODE" (that is nodes within the node). A 'token' is a 32
606bit value. The tree has to be "finished" with a OF_DT_END token
607
608Here's the basic structure of a single node:
609
610 * token OF_DT_BEGIN_NODE (that is 0x00000001)
611 * for version 1 to 3, this is the node full path as a zero
612 terminated string, starting with "/". For version 16 and later,
613 this is the node unit name only (or an empty string for the
614 root node)
615 * [align gap to next 4 bytes boundary]
616 * for each property:
617 * token OF_DT_PROP (that is 0x00000003)
Domen Puncer5dd60162007-03-02 21:44:45 +1100618 * 32-bit value of property value size in bytes (or 0 if no
619 value)
620 * 32-bit value of offset in string block of property name
David Gibsonc125a182006-02-01 03:05:22 -0800621 * property value data if any
622 * [align gap to next 4 bytes boundary]
623 * [child nodes if any]
624 * token OF_DT_END_NODE (that is 0x00000002)
625
Domen Puncer5dd60162007-03-02 21:44:45 +1100626So the node content can be summarized as a start token, a full path,
Matt LaPlante53cb4722006-10-03 22:55:17 +0200627a list of properties, a list of child nodes, and an end token. Every
David Gibsonc125a182006-02-01 03:05:22 -0800628child node is a full node structure itself as defined above.
629
David Gibsoneff2ebd2007-06-28 15:56:26 +1000630NOTE: The above definition requires that all property definitions for
631a particular node MUST precede any subnode definitions for that node.
632Although the structure would not be ambiguous if properties and
633subnodes were intermingled, the kernel parser requires that the
634properties come first (up until at least 2.6.22). Any tools
635manipulating a flattened tree must take care to preserve this
636constraint.
637
Matt LaPlante53cb4722006-10-03 22:55:17 +02006384) Device tree "strings" block
David Gibsonc125a182006-02-01 03:05:22 -0800639
640In order to save space, property names, which are generally redundant,
641are stored separately in the "strings" block. This block is simply the
642whole bunch of zero terminated strings for all property names
643concatenated together. The device-tree property definitions in the
644structure block will contain offset values from the beginning of the
645strings block.
646
647
648III - Required content of the device tree
649=========================================
650
651WARNING: All "linux,*" properties defined in this document apply only
652to a flattened device-tree. If your platform uses a real
653implementation of Open Firmware or an implementation compatible with
654the Open Firmware client interface, those properties will be created
655by the trampoline code in the kernel's prom_init() file. For example,
656that's where you'll have to add code to detect your board model and
Matt LaPlantea2ffd272006-10-03 22:49:15 +0200657set the platform number. However, when using the flattened device-tree
David Gibsonc125a182006-02-01 03:05:22 -0800658entry point, there is no prom_init() pass, and thus you have to
659provide those properties yourself.
660
661
6621) Note about cells and address representation
663----------------------------------------------
664
665The general rule is documented in the various Open Firmware
Domen Puncer5dd60162007-03-02 21:44:45 +1100666documentations. If you choose to describe a bus with the device-tree
David Gibsonc125a182006-02-01 03:05:22 -0800667and there exist an OF bus binding, then you should follow the
668specification. However, the kernel does not require every single
669device or bus to be described by the device tree.
670
671In general, the format of an address for a device is defined by the
672parent bus type, based on the #address-cells and #size-cells
673property. In the absence of such a property, the parent's parent
674values are used, etc... The kernel requires the root node to have
675those properties defining addresses format for devices directly mapped
676on the processor bus.
677
678Those 2 properties define 'cells' for representing an address and a
Domen Puncer5dd60162007-03-02 21:44:45 +1100679size. A "cell" is a 32-bit number. For example, if both contain 2
David Gibsonc125a182006-02-01 03:05:22 -0800680like the example tree given above, then an address and a size are both
Domen Puncer5dd60162007-03-02 21:44:45 +1100681composed of 2 cells, and each is a 64-bit number (cells are
David Gibsonc125a182006-02-01 03:05:22 -0800682concatenated and expected to be in big endian format). Another example
683is the way Apple firmware defines them, with 2 cells for an address
684and one cell for a size. Most 32-bit implementations should define
685#address-cells and #size-cells to 1, which represents a 32-bit value.
686Some 32-bit processors allow for physical addresses greater than 32
687bits; these processors should define #address-cells as 2.
688
689"reg" properties are always a tuple of the type "address size" where
690the number of cells of address and size is specified by the bus
691#address-cells and #size-cells. When a bus supports various address
692spaces and other flags relative to a given address allocation (like
693prefetchable, etc...) those flags are usually added to the top level
694bits of the physical address. For example, a PCI physical address is
695made of 3 cells, the bottom two containing the actual address itself
696while the top cell contains address space indication, flags, and pci
697bus & device numbers.
698
699For busses that support dynamic allocation, it's the accepted practice
700to then not provide the address in "reg" (keep it 0) though while
701providing a flag indicating the address is dynamically allocated, and
702then, to provide a separate "assigned-addresses" property that
703contains the fully allocated addresses. See the PCI OF bindings for
704details.
705
706In general, a simple bus with no address space bits and no dynamic
707allocation is preferred if it reflects your hardware, as the existing
708kernel address parsing functions will work out of the box. If you
709define a bus type with a more complex address format, including things
710like address space bits, you'll have to add a bus translator to the
711prom_parse.c file of the recent kernels for your bus type.
712
713The "reg" property only defines addresses and sizes (if #size-cells
Matt LaPlante992caac2006-10-03 22:52:05 +0200714is non-0) within a given bus. In order to translate addresses upward
Domen Puncer5dd60162007-03-02 21:44:45 +1100715(that is into parent bus addresses, and possibly into CPU physical
David Gibsonc125a182006-02-01 03:05:22 -0800716addresses), all busses must contain a "ranges" property. If the
717"ranges" property is missing at a given level, it's assumed that
Matt LaPlante992caac2006-10-03 22:52:05 +0200718translation isn't possible. The format of the "ranges" property for a
David Gibsonc125a182006-02-01 03:05:22 -0800719bus is a list of:
720
721 bus address, parent bus address, size
722
723"bus address" is in the format of the bus this bus node is defining,
724that is, for a PCI bridge, it would be a PCI address. Thus, (bus
725address, size) defines a range of addresses for child devices. "parent
726bus address" is in the format of the parent bus of this bus. For
727example, for a PCI host controller, that would be a CPU address. For a
728PCI<->ISA bridge, that would be a PCI address. It defines the base
729address in the parent bus where the beginning of that range is mapped.
730
Domen Puncer5dd60162007-03-02 21:44:45 +1100731For a new 64-bit powerpc board, I recommend either the 2/2 format or
David Gibsonc125a182006-02-01 03:05:22 -0800732Apple's 2/1 format which is slightly more compact since sizes usually
Domen Puncer5dd60162007-03-02 21:44:45 +1100733fit in a single 32-bit word. New 32-bit powerpc boards should use a
David Gibsonc125a182006-02-01 03:05:22 -08007341/1 format, unless the processor supports physical addresses greater
735than 32-bits, in which case a 2/1 format is recommended.
736
737
7382) Note about "compatible" properties
739-------------------------------------
740
741These properties are optional, but recommended in devices and the root
742node. The format of a "compatible" property is a list of concatenated
743zero terminated strings. They allow a device to express its
744compatibility with a family of similar devices, in some cases,
745allowing a single driver to match against several devices regardless
746of their actual names.
747
7483) Note about "name" properties
749-------------------------------
750
751While earlier users of Open Firmware like OldWorld macintoshes tended
752to use the actual device name for the "name" property, it's nowadays
753considered a good practice to use a name that is closer to the device
754class (often equal to device_type). For example, nowadays, ethernet
755controllers are named "ethernet", an additional "model" property
756defining precisely the chip type/model, and "compatible" property
757defining the family in case a single driver can driver more than one
758of these chips. However, the kernel doesn't generally put any
759restriction on the "name" property; it is simply considered good
760practice to follow the standard and its evolutions as closely as
761possible.
762
763Note also that the new format version 16 makes the "name" property
764optional. If it's absent for a node, then the node's unit name is then
765used to reconstruct the name. That is, the part of the unit name
766before the "@" sign is used (or the entire unit name if no "@" sign
767is present).
768
7694) Note about node and property names and character set
770-------------------------------------------------------
771
Matt LaPlantea2ffd272006-10-03 22:49:15 +0200772While open firmware provides more flexible usage of 8859-1, this
David Gibsonc125a182006-02-01 03:05:22 -0800773specification enforces more strict rules. Nodes and properties should
774be comprised only of ASCII characters 'a' to 'z', '0' to
775'9', ',', '.', '_', '+', '#', '?', and '-'. Node names additionally
776allow uppercase characters 'A' to 'Z' (property names should be
777lowercase. The fact that vendors like Apple don't respect this rule is
778irrelevant here). Additionally, node and property names should always
779begin with a character in the range 'a' to 'z' (or 'A' to 'Z' for node
780names).
781
782The maximum number of characters for both nodes and property names
783is 31. In the case of node names, this is only the leftmost part of
784a unit name (the pure "name" property), it doesn't include the unit
785address which can extend beyond that limit.
786
787
7885) Required nodes and properties
789--------------------------------
790 These are all that are currently required. However, it is strongly
791 recommended that you expose PCI host bridges as documented in the
792 PCI binding to open firmware, and your interrupt tree as documented
793 in OF interrupt tree specification.
794
795 a) The root node
796
797 The root node requires some properties to be present:
798
799 - model : this is your board name/model
800 - #address-cells : address representation for "root" devices
801 - #size-cells: the size representation for "root" devices
Benjamin Herrenschmidte8222502006-03-28 23:15:54 +1100802 - device_type : This property shouldn't be necessary. However, if
803 you decide to create a device_type for your root node, make sure it
804 is _not_ "chrp" unless your platform is a pSeries or PAPR compliant
805 one for 64-bit, or a CHRP-type machine for 32-bit as this will
806 matched by the kernel this way.
David Gibsonc125a182006-02-01 03:05:22 -0800807
808 Additionally, some recommended properties are:
809
810 - compatible : the board "family" generally finds its way here,
811 for example, if you have 2 board models with a similar layout,
812 that typically get driven by the same platform code in the
813 kernel, you would use a different "model" property but put a
814 value in "compatible". The kernel doesn't directly use that
Stuart Yoder143a42d2007-02-16 11:30:29 -0600815 value but it is generally useful.
David Gibsonc125a182006-02-01 03:05:22 -0800816
817 The root node is also generally where you add additional properties
818 specific to your board like the serial number if any, that sort of
Matt LaPlante6c28f2c2006-10-03 22:46:31 +0200819 thing. It is recommended that if you add any "custom" property whose
David Gibsonc125a182006-02-01 03:05:22 -0800820 name may clash with standard defined ones, you prefix them with your
821 vendor name and a comma.
822
823 b) The /cpus node
824
825 This node is the parent of all individual CPU nodes. It doesn't
826 have any specific requirements, though it's generally good practice
827 to have at least:
828
829 #address-cells = <00000001>
830 #size-cells = <00000000>
831
832 This defines that the "address" for a CPU is a single cell, and has
833 no meaningful size. This is not necessary but the kernel will assume
834 that format when reading the "reg" properties of a CPU node, see
835 below
836
837 c) The /cpus/* nodes
838
839 So under /cpus, you are supposed to create a node for every CPU on
840 the machine. There is no specific restriction on the name of the
841 CPU, though It's common practice to call it PowerPC,<name>. For
842 example, Apple uses PowerPC,G5 while IBM uses PowerPC,970FX.
843
844 Required properties:
845
846 - device_type : has to be "cpu"
Domen Puncer5dd60162007-03-02 21:44:45 +1100847 - reg : This is the physical CPU number, it's a single 32-bit cell
David Gibsonc125a182006-02-01 03:05:22 -0800848 and is also used as-is as the unit number for constructing the
849 unit name in the full path. For example, with 2 CPUs, you would
850 have the full path:
851 /cpus/PowerPC,970FX@0
852 /cpus/PowerPC,970FX@1
853 (unit addresses do not require leading zeroes)
854 - d-cache-line-size : one cell, L1 data cache line size in bytes
855 - i-cache-line-size : one cell, L1 instruction cache line size in
856 bytes
857 - d-cache-size : one cell, size of L1 data cache in bytes
858 - i-cache-size : one cell, size of L1 instruction cache in bytes
David Gibsonc125a182006-02-01 03:05:22 -0800859
860 Recommended properties:
861
862 - timebase-frequency : a cell indicating the frequency of the
863 timebase in Hz. This is not directly used by the generic code,
864 but you are welcome to copy/paste the pSeries code for setting
865 the kernel timebase/decrementer calibration based on this
866 value.
867 - clock-frequency : a cell indicating the CPU core clock frequency
Domen Puncer5dd60162007-03-02 21:44:45 +1100868 in Hz. A new property will be defined for 64-bit values, but if
David Gibsonc125a182006-02-01 03:05:22 -0800869 your frequency is < 4Ghz, one cell is enough. Here as well as
870 for the above, the common code doesn't use that property, but
871 you are welcome to re-use the pSeries or Maple one. A future
872 kernel version might provide a common function for this.
873
874 You are welcome to add any property you find relevant to your board,
875 like some information about the mechanism used to soft-reset the
876 CPUs. For example, Apple puts the GPIO number for CPU soft reset
877 lines in there as a "soft-reset" property since they start secondary
878 CPUs by soft-resetting them.
879
880
881 d) the /memory node(s)
882
883 To define the physical memory layout of your board, you should
884 create one or more memory node(s). You can either create a single
885 node with all memory ranges in its reg property, or you can create
886 several nodes, as you wish. The unit address (@ part) used for the
887 full path is the address of the first range of memory defined by a
888 given node. If you use a single memory node, this will typically be
889 @0.
890
891 Required properties:
892
893 - device_type : has to be "memory"
894 - reg : This property contains all the physical memory ranges of
895 your board. It's a list of addresses/sizes concatenated
896 together, with the number of cells of each defined by the
897 #address-cells and #size-cells of the root node. For example,
Matt LaPlante6c28f2c2006-10-03 22:46:31 +0200898 with both of these properties being 2 like in the example given
David Gibsonc125a182006-02-01 03:05:22 -0800899 earlier, a 970 based machine with 6Gb of RAM could typically
900 have a "reg" property here that looks like:
901
902 00000000 00000000 00000000 80000000
903 00000001 00000000 00000001 00000000
904
905 That is a range starting at 0 of 0x80000000 bytes and a range
906 starting at 0x100000000 and of 0x100000000 bytes. You can see
907 that there is no memory covering the IO hole between 2Gb and
908 4Gb. Some vendors prefer splitting those ranges into smaller
909 segments, but the kernel doesn't care.
910
911 e) The /chosen node
912
913 This node is a bit "special". Normally, that's where open firmware
914 puts some variable environment information, like the arguments, or
Stuart Yoderd1bff9e2007-02-19 11:25:05 -0600915 the default input/output devices.
David Gibsonc125a182006-02-01 03:05:22 -0800916
917 This specification makes a few of these mandatory, but also defines
918 some linux-specific properties that would be normally constructed by
919 the prom_init() trampoline when booting with an OF client interface,
920 but that you have to provide yourself when using the flattened format.
921
David Gibsonc125a182006-02-01 03:05:22 -0800922 Recommended properties:
923
924 - bootargs : This zero-terminated string is passed as the kernel
925 command line
926 - linux,stdout-path : This is the full path to your standard
927 console device if any. Typically, if you have serial devices on
928 your board, you may want to put the full path to the one set as
929 the default console in the firmware here, for the kernel to pick
Matt LaPlante5d3f0832006-11-30 05:21:10 +0100930 it up as its own default console. If you look at the function
David Gibsonc125a182006-02-01 03:05:22 -0800931 set_preferred_console() in arch/ppc64/kernel/setup.c, you'll see
932 that the kernel tries to find out the default console and has
933 knowledge of various types like 8250 serial ports. You may want
934 to extend this function to add your own.
David Gibsonc125a182006-02-01 03:05:22 -0800935
936 Note that u-boot creates and fills in the chosen node for platforms
937 that use it.
938
Stuart Yoderd1bff9e2007-02-19 11:25:05 -0600939 (Note: a practice that is now obsolete was to include a property
940 under /chosen called interrupt-controller which had a phandle value
941 that pointed to the main interrupt controller)
942
David Gibsonc125a182006-02-01 03:05:22 -0800943 f) the /soc<SOCname> node
944
945 This node is used to represent a system-on-a-chip (SOC) and must be
946 present if the processor is a SOC. The top-level soc node contains
947 information that is global to all devices on the SOC. The node name
948 should contain a unit address for the SOC, which is the base address
949 of the memory-mapped register set for the SOC. The name of an soc
950 node should start with "soc", and the remainder of the name should
951 represent the part number for the soc. For example, the MPC8540's
952 soc node would be called "soc8540".
953
954 Required properties:
955
956 - device_type : Should be "soc"
957 - ranges : Should be defined as specified in 1) to describe the
958 translation of SOC addresses for memory mapped SOC registers.
Becky Bruce7d4b95a2006-02-06 14:26:31 -0600959 - bus-frequency: Contains the bus frequency for the SOC node.
960 Typically, the value of this field is filled in by the boot
961 loader.
962
David Gibsonc125a182006-02-01 03:05:22 -0800963
964 Recommended properties:
965
966 - reg : This property defines the address and size of the
967 memory-mapped registers that are used for the SOC node itself.
968 It does not include the child device registers - these will be
969 defined inside each child node. The address specified in the
970 "reg" property should match the unit address of the SOC node.
971 - #address-cells : Address representation for "soc" devices. The
972 format of this field may vary depending on whether or not the
973 device registers are memory mapped. For memory mapped
974 registers, this field represents the number of cells needed to
975 represent the address of the registers. For SOCs that do not
976 use MMIO, a special address format should be defined that
977 contains enough cells to represent the required information.
978 See 1) above for more details on defining #address-cells.
979 - #size-cells : Size representation for "soc" devices
980 - #interrupt-cells : Defines the width of cells used to represent
981 interrupts. Typically this value is <2>, which includes a
982 32-bit number that represents the interrupt number, and a
983 32-bit number that represents the interrupt sense and level.
984 This field is only needed if the SOC contains an interrupt
985 controller.
986
987 The SOC node may contain child nodes for each SOC device that the
988 platform uses. Nodes should not be created for devices which exist
989 on the SOC but are not used by a particular platform. See chapter VI
Domen Puncer5dd60162007-03-02 21:44:45 +1100990 for more information on how to specify devices that are part of a SOC.
David Gibsonc125a182006-02-01 03:05:22 -0800991
992 Example SOC node for the MPC8540:
993
994 soc8540@e0000000 {
995 #address-cells = <1>;
996 #size-cells = <1>;
997 #interrupt-cells = <2>;
998 device_type = "soc";
999 ranges = <00000000 e0000000 00100000>
1000 reg = <e0000000 00003000>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -06001001 bus-frequency = <0>;
David Gibsonc125a182006-02-01 03:05:22 -08001002 }
1003
1004
1005
1006IV - "dtc", the device tree compiler
1007====================================
1008
1009
1010dtc source code can be found at
1011<http://ozlabs.org/~dgibson/dtc/dtc.tar.gz>
1012
1013WARNING: This version is still in early development stage; the
1014resulting device-tree "blobs" have not yet been validated with the
1015kernel. The current generated bloc lacks a useful reserve map (it will
1016be fixed to generate an empty one, it's up to the bootloader to fill
1017it up) among others. The error handling needs work, bugs are lurking,
1018etc...
1019
1020dtc basically takes a device-tree in a given format and outputs a
1021device-tree in another format. The currently supported formats are:
1022
1023 Input formats:
1024 -------------
1025
1026 - "dtb": "blob" format, that is a flattened device-tree block
1027 with
1028 header all in a binary blob.
1029 - "dts": "source" format. This is a text file containing a
1030 "source" for a device-tree. The format is defined later in this
1031 chapter.
1032 - "fs" format. This is a representation equivalent to the
1033 output of /proc/device-tree, that is nodes are directories and
1034 properties are files
1035
1036 Output formats:
1037 ---------------
1038
1039 - "dtb": "blob" format
1040 - "dts": "source" format
1041 - "asm": assembly language file. This is a file that can be
1042 sourced by gas to generate a device-tree "blob". That file can
1043 then simply be added to your Makefile. Additionally, the
Matt LaPlante6c28f2c2006-10-03 22:46:31 +02001044 assembly file exports some symbols that can be used.
David Gibsonc125a182006-02-01 03:05:22 -08001045
1046
1047The syntax of the dtc tool is
1048
1049 dtc [-I <input-format>] [-O <output-format>]
1050 [-o output-filename] [-V output_version] input_filename
1051
1052
Domen Puncer5dd60162007-03-02 21:44:45 +11001053The "output_version" defines what version of the "blob" format will be
David Gibsonc125a182006-02-01 03:05:22 -08001054generated. Supported versions are 1,2,3 and 16. The default is
1055currently version 3 but that may change in the future to version 16.
1056
1057Additionally, dtc performs various sanity checks on the tree, like the
Matt LaPlante6c28f2c2006-10-03 22:46:31 +02001058uniqueness of linux, phandle properties, validity of strings, etc...
David Gibsonc125a182006-02-01 03:05:22 -08001059
1060The format of the .dts "source" file is "C" like, supports C and C++
Matt LaPlante6c28f2c2006-10-03 22:46:31 +02001061style comments.
David Gibsonc125a182006-02-01 03:05:22 -08001062
1063/ {
1064}
1065
1066The above is the "device-tree" definition. It's the only statement
1067supported currently at the toplevel.
1068
1069/ {
1070 property1 = "string_value"; /* define a property containing a 0
1071 * terminated string
1072 */
1073
1074 property2 = <1234abcd>; /* define a property containing a
Domen Puncer5dd60162007-03-02 21:44:45 +11001075 * numerical 32-bit value (hexadecimal)
David Gibsonc125a182006-02-01 03:05:22 -08001076 */
1077
1078 property3 = <12345678 12345678 deadbeef>;
1079 /* define a property containing 3
Domen Puncer5dd60162007-03-02 21:44:45 +11001080 * numerical 32-bit values (cells) in
David Gibsonc125a182006-02-01 03:05:22 -08001081 * hexadecimal
1082 */
1083 property4 = [0a 0b 0c 0d de ea ad be ef];
1084 /* define a property whose content is
1085 * an arbitrary array of bytes
1086 */
1087
1088 childnode@addresss { /* define a child node named "childnode"
1089 * whose unit name is "childnode at
1090 * address"
1091 */
1092
1093 childprop = "hello\n"; /* define a property "childprop" of
1094 * childnode (in this case, a string)
1095 */
1096 };
1097};
1098
1099Nodes can contain other nodes etc... thus defining the hierarchical
1100structure of the tree.
1101
1102Strings support common escape sequences from C: "\n", "\t", "\r",
1103"\(octal value)", "\x(hex value)".
1104
1105It is also suggested that you pipe your source file through cpp (gcc
1106preprocessor) so you can use #include's, #define for constants, etc...
1107
1108Finally, various options are planned but not yet implemented, like
1109automatic generation of phandles, labels (exported to the asm file so
1110you can point to a property content and change it easily from whatever
1111you link the device-tree with), label or path instead of numeric value
1112in some cells to "point" to a node (replaced by a phandle at compile
1113time), export of reserve map address to the asm file, ability to
1114specify reserve map content at compile time, etc...
1115
1116We may provide a .h include file with common definitions of that
1117proves useful for some properties (like building PCI properties or
1118interrupt maps) though it may be better to add a notion of struct
1119definitions to the compiler...
1120
1121
1122V - Recommendations for a bootloader
1123====================================
1124
1125
1126Here are some various ideas/recommendations that have been proposed
1127while all this has been defined and implemented.
1128
1129 - The bootloader may want to be able to use the device-tree itself
1130 and may want to manipulate it (to add/edit some properties,
1131 like physical memory size or kernel arguments). At this point, 2
1132 choices can be made. Either the bootloader works directly on the
1133 flattened format, or the bootloader has its own internal tree
1134 representation with pointers (similar to the kernel one) and
1135 re-flattens the tree when booting the kernel. The former is a bit
1136 more difficult to edit/modify, the later requires probably a bit
1137 more code to handle the tree structure. Note that the structure
1138 format has been designed so it's relatively easy to "insert"
1139 properties or nodes or delete them by just memmoving things
1140 around. It contains no internal offsets or pointers for this
1141 purpose.
1142
Matt LaPlanted6bc8ac2006-10-03 22:54:15 +02001143 - An example of code for iterating nodes & retrieving properties
David Gibsonc125a182006-02-01 03:05:22 -08001144 directly from the flattened tree format can be found in the kernel
1145 file arch/ppc64/kernel/prom.c, look at scan_flat_dt() function,
Matt LaPlanted6bc8ac2006-10-03 22:54:15 +02001146 its usage in early_init_devtree(), and the corresponding various
David Gibsonc125a182006-02-01 03:05:22 -08001147 early_init_dt_scan_*() callbacks. That code can be re-used in a
1148 GPL bootloader, and as the author of that code, I would be happy
Domen Puncer5dd60162007-03-02 21:44:45 +11001149 to discuss possible free licensing to any vendor who wishes to
David Gibsonc125a182006-02-01 03:05:22 -08001150 integrate all or part of this code into a non-GPL bootloader.
1151
1152
1153
1154VI - System-on-a-chip devices and nodes
1155=======================================
1156
1157Many companies are now starting to develop system-on-a-chip
Domen Puncer5dd60162007-03-02 21:44:45 +11001158processors, where the processor core (CPU) and many peripheral devices
David Gibsonc125a182006-02-01 03:05:22 -08001159exist on a single piece of silicon. For these SOCs, an SOC node
1160should be used that defines child nodes for the devices that make
1161up the SOC. While platforms are not required to use this model in
1162order to boot the kernel, it is highly encouraged that all SOC
1163implementations define as complete a flat-device-tree as possible to
1164describe the devices on the SOC. This will allow for the
1165genericization of much of the kernel code.
1166
1167
11681) Defining child nodes of an SOC
1169---------------------------------
1170
1171Each device that is part of an SOC may have its own node entry inside
1172the SOC node. For each device that is included in the SOC, the unit
1173address property represents the address offset for this device's
1174memory-mapped registers in the parent's address space. The parent's
1175address space is defined by the "ranges" property in the top-level soc
1176node. The "reg" property for each node that exists directly under the
1177SOC node should contain the address mapping from the child address space
1178to the parent SOC address space and the size of the device's
1179memory-mapped register file.
1180
1181For many devices that may exist inside an SOC, there are predefined
1182specifications for the format of the device tree node. All SOC child
1183nodes should follow these specifications, except where noted in this
1184document.
1185
1186See appendix A for an example partial SOC node definition for the
1187MPC8540.
1188
1189
Stuart Yoder27565902007-03-02 13:42:33 -060011902) Representing devices without a current OF specification
David Gibsonc125a182006-02-01 03:05:22 -08001191----------------------------------------------------------
1192
1193Currently, there are many devices on SOCs that do not have a standard
1194representation pre-defined as part of the open firmware
1195specifications, mainly because the boards that contain these SOCs are
1196not currently booted using open firmware. This section contains
1197descriptions for the SOC devices for which new nodes have been
1198defined; this list will expand as more and more SOC-containing
1199platforms are moved over to use the flattened-device-tree model.
1200
1201 a) MDIO IO device
1202
1203 The MDIO is a bus to which the PHY devices are connected. For each
1204 device that exists on this bus, a child node should be created. See
1205 the definition of the PHY node below for an example of how to define
1206 a PHY.
1207
1208 Required properties:
1209 - reg : Offset and length of the register set for the device
1210 - device_type : Should be "mdio"
1211 - compatible : Should define the compatible device type for the
1212 mdio. Currently, this is most likely to be "gianfar"
1213
1214 Example:
1215
1216 mdio@24520 {
1217 reg = <24520 20>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -06001218 device_type = "mdio";
1219 compatible = "gianfar";
David Gibsonc125a182006-02-01 03:05:22 -08001220
1221 ethernet-phy@0 {
1222 ......
1223 };
1224 };
1225
1226
1227 b) Gianfar-compatible ethernet nodes
1228
1229 Required properties:
1230
1231 - device_type : Should be "network"
1232 - model : Model of the device. Can be "TSEC", "eTSEC", or "FEC"
1233 - compatible : Should be "gianfar"
1234 - reg : Offset and length of the register set for the device
Jon Loeligerf5831652006-08-17 08:42:35 -05001235 - mac-address : List of bytes representing the ethernet address of
David Gibsonc125a182006-02-01 03:05:22 -08001236 this controller
1237 - interrupts : <a b> where a is the interrupt number and b is a
1238 field that represents an encoding of the sense and level
1239 information for the interrupt. This should be encoded based on
1240 the information in section 2) depending on the type of interrupt
1241 controller you have.
1242 - interrupt-parent : the phandle for the interrupt controller that
1243 services interrupts for this device.
1244 - phy-handle : The phandle for the PHY connected to this ethernet
1245 controller.
1246
Scott Woode0a2f282007-03-16 12:28:46 -05001247 Recommended properties:
1248
1249 - linux,network-index : This is the intended "index" of this
1250 network device. This is used by the bootwrapper to interpret
1251 MAC addresses passed by the firmware when no information other
1252 than indices is available to associate an address with a device.
1253
David Gibsonc125a182006-02-01 03:05:22 -08001254 Example:
1255
1256 ethernet@24000 {
1257 #size-cells = <0>;
1258 device_type = "network";
1259 model = "TSEC";
1260 compatible = "gianfar";
1261 reg = <24000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001262 mac-address = [ 00 E0 0C 00 73 00 ];
David Gibsonc125a182006-02-01 03:05:22 -08001263 interrupts = <d 3 e 3 12 3>;
1264 interrupt-parent = <40000>;
1265 phy-handle = <2452000>
1266 };
1267
1268
1269
1270 c) PHY nodes
1271
1272 Required properties:
1273
1274 - device_type : Should be "ethernet-phy"
1275 - interrupts : <a b> where a is the interrupt number and b is a
1276 field that represents an encoding of the sense and level
1277 information for the interrupt. This should be encoded based on
1278 the information in section 2) depending on the type of interrupt
1279 controller you have.
1280 - interrupt-parent : the phandle for the interrupt controller that
1281 services interrupts for this device.
1282 - reg : The ID number for the phy, usually a small integer
1283 - linux,phandle : phandle for this node; likely referenced by an
1284 ethernet controller node.
1285
1286
1287 Example:
1288
1289 ethernet-phy@0 {
1290 linux,phandle = <2452000>
1291 interrupt-parent = <40000>;
1292 interrupts = <35 1>;
1293 reg = <0>;
1294 device_type = "ethernet-phy";
1295 };
1296
1297
1298 d) Interrupt controllers
1299
1300 Some SOC devices contain interrupt controllers that are different
1301 from the standard Open PIC specification. The SOC device nodes for
1302 these types of controllers should be specified just like a standard
1303 OpenPIC controller. Sense and level information should be encoded
1304 as specified in section 2) of this chapter for each device that
1305 specifies an interrupt.
1306
1307 Example :
1308
1309 pic@40000 {
1310 linux,phandle = <40000>;
1311 clock-frequency = <0>;
1312 interrupt-controller;
1313 #address-cells = <0>;
1314 reg = <40000 40000>;
1315 built-in;
1316 compatible = "chrp,open-pic";
1317 device_type = "open-pic";
1318 big-endian;
1319 };
1320
1321
1322 e) I2C
1323
1324 Required properties :
1325
1326 - device_type : Should be "i2c"
1327 - reg : Offset and length of the register set for the device
1328
1329 Recommended properties :
1330
1331 - compatible : Should be "fsl-i2c" for parts compatible with
1332 Freescale I2C specifications.
1333 - interrupts : <a b> where a is the interrupt number and b is a
1334 field that represents an encoding of the sense and level
1335 information for the interrupt. This should be encoded based on
1336 the information in section 2) depending on the type of interrupt
1337 controller you have.
1338 - interrupt-parent : the phandle for the interrupt controller that
1339 services interrupts for this device.
1340 - dfsrr : boolean; if defined, indicates that this I2C device has
1341 a digital filter sampling rate register
1342 - fsl5200-clocking : boolean; if defined, indicated that this device
1343 uses the FSL 5200 clocking mechanism.
1344
1345 Example :
1346
1347 i2c@3000 {
1348 interrupt-parent = <40000>;
1349 interrupts = <1b 3>;
1350 reg = <3000 18>;
1351 device_type = "i2c";
1352 compatible = "fsl-i2c";
1353 dfsrr;
1354 };
1355
1356
Becky Brucead71f122006-02-07 13:44:08 -06001357 f) Freescale SOC USB controllers
1358
1359 The device node for a USB controller that is part of a Freescale
1360 SOC is as described in the document "Open Firmware Recommended
1361 Practice : Universal Serial Bus" with the following modifications
1362 and additions :
1363
1364 Required properties :
Domen Puncer5dd60162007-03-02 21:44:45 +11001365 - compatible : Should be "fsl-usb2-mph" for multi port host USB
1366 controllers, or "fsl-usb2-dr" for dual role USB controllers
1367 - phy_type : For multi port host USB controllers, should be one of
1368 "ulpi", or "serial". For dual role USB controllers, should be
Becky Brucead71f122006-02-07 13:44:08 -06001369 one of "ulpi", "utmi", "utmi_wide", or "serial".
1370 - reg : Offset and length of the register set for the device
1371 - port0 : boolean; if defined, indicates port0 is connected for
1372 fsl-usb2-mph compatible controllers. Either this property or
1373 "port1" (or both) must be defined for "fsl-usb2-mph" compatible
1374 controllers.
1375 - port1 : boolean; if defined, indicates port1 is connected for
1376 fsl-usb2-mph compatible controllers. Either this property or
1377 "port0" (or both) must be defined for "fsl-usb2-mph" compatible
1378 controllers.
Li Yangea5b7a62007-02-07 13:51:09 +08001379 - dr_mode : indicates the working mode for "fsl-usb2-dr" compatible
1380 controllers. Can be "host", "peripheral", or "otg". Default to
1381 "host" if not defined for backward compatibility.
Becky Brucead71f122006-02-07 13:44:08 -06001382
1383 Recommended properties :
1384 - interrupts : <a b> where a is the interrupt number and b is a
1385 field that represents an encoding of the sense and level
1386 information for the interrupt. This should be encoded based on
1387 the information in section 2) depending on the type of interrupt
1388 controller you have.
1389 - interrupt-parent : the phandle for the interrupt controller that
1390 services interrupts for this device.
1391
Domen Puncer5dd60162007-03-02 21:44:45 +11001392 Example multi port host USB controller device node :
Becky Brucead71f122006-02-07 13:44:08 -06001393 usb@22000 {
1394 device_type = "usb";
1395 compatible = "fsl-usb2-mph";
1396 reg = <22000 1000>;
1397 #address-cells = <1>;
1398 #size-cells = <0>;
1399 interrupt-parent = <700>;
1400 interrupts = <27 1>;
1401 phy_type = "ulpi";
1402 port0;
1403 port1;
1404 };
1405
Domen Puncer5dd60162007-03-02 21:44:45 +11001406 Example dual role USB controller device node :
Becky Brucead71f122006-02-07 13:44:08 -06001407 usb@23000 {
1408 device_type = "usb";
1409 compatible = "fsl-usb2-dr";
1410 reg = <23000 1000>;
1411 #address-cells = <1>;
1412 #size-cells = <0>;
1413 interrupt-parent = <700>;
1414 interrupts = <26 1>;
Li Yangea5b7a62007-02-07 13:51:09 +08001415 dr_mode = "otg";
Becky Brucead71f122006-02-07 13:44:08 -06001416 phy = "ulpi";
1417 };
1418
1419
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001420 g) Freescale SOC SEC Security Engines
1421
1422 Required properties:
1423
1424 - device_type : Should be "crypto"
1425 - model : Model of the device. Should be "SEC1" or "SEC2"
1426 - compatible : Should be "talitos"
1427 - reg : Offset and length of the register set for the device
1428 - interrupts : <a b> where a is the interrupt number and b is a
1429 field that represents an encoding of the sense and level
1430 information for the interrupt. This should be encoded based on
1431 the information in section 2) depending on the type of interrupt
1432 controller you have.
1433 - interrupt-parent : the phandle for the interrupt controller that
1434 services interrupts for this device.
1435 - num-channels : An integer representing the number of channels
1436 available.
1437 - channel-fifo-len : An integer representing the number of
1438 descriptor pointers each channel fetch fifo can hold.
1439 - exec-units-mask : The bitmask representing what execution units
Domen Puncer5dd60162007-03-02 21:44:45 +11001440 (EUs) are available. It's a single 32-bit cell. EU information
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001441 should be encoded following the SEC's Descriptor Header Dword
1442 EU_SEL0 field documentation, i.e. as follows:
1443
1444 bit 0 = reserved - should be 0
1445 bit 1 = set if SEC has the ARC4 EU (AFEU)
1446 bit 2 = set if SEC has the DES/3DES EU (DEU)
1447 bit 3 = set if SEC has the message digest EU (MDEU)
1448 bit 4 = set if SEC has the random number generator EU (RNG)
1449 bit 5 = set if SEC has the public key EU (PKEU)
1450 bit 6 = set if SEC has the AES EU (AESU)
1451 bit 7 = set if SEC has the Kasumi EU (KEU)
1452
1453 bits 8 through 31 are reserved for future SEC EUs.
1454
1455 - descriptor-types-mask : The bitmask representing what descriptors
Domen Puncer5dd60162007-03-02 21:44:45 +11001456 are available. It's a single 32-bit cell. Descriptor type
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001457 information should be encoded following the SEC's Descriptor
1458 Header Dword DESC_TYPE field documentation, i.e. as follows:
1459
1460 bit 0 = set if SEC supports the aesu_ctr_nonsnoop desc. type
1461 bit 1 = set if SEC supports the ipsec_esp descriptor type
1462 bit 2 = set if SEC supports the common_nonsnoop desc. type
1463 bit 3 = set if SEC supports the 802.11i AES ccmp desc. type
1464 bit 4 = set if SEC supports the hmac_snoop_no_afeu desc. type
1465 bit 5 = set if SEC supports the srtp descriptor type
1466 bit 6 = set if SEC supports the non_hmac_snoop_no_afeu desc.type
1467 bit 7 = set if SEC supports the pkeu_assemble descriptor type
1468 bit 8 = set if SEC supports the aesu_key_expand_output desc.type
1469 bit 9 = set if SEC supports the pkeu_ptmul descriptor type
1470 bit 10 = set if SEC supports the common_nonsnoop_afeu desc. type
1471 bit 11 = set if SEC supports the pkeu_ptadd_dbl descriptor type
1472
1473 ..and so on and so forth.
1474
1475 Example:
1476
1477 /* MPC8548E */
1478 crypto@30000 {
1479 device_type = "crypto";
1480 model = "SEC2";
1481 compatible = "talitos";
1482 reg = <30000 10000>;
1483 interrupts = <1d 3>;
1484 interrupt-parent = <40000>;
1485 num-channels = <4>;
Kim Phillipscbdb54d2006-07-03 15:10:14 -05001486 channel-fifo-len = <18>;
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001487 exec-units-mask = <000000fe>;
Kim Phillipscbdb54d2006-07-03 15:10:14 -05001488 descriptor-types-mask = <012b0ebf>;
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001489 };
1490
Li Yang9a1ab882006-10-02 20:08:59 -05001491 h) Board Control and Status (BCSR)
1492
1493 Required properties:
1494
1495 - device_type : Should be "board-control"
1496 - reg : Offset and length of the register set for the device
1497
1498 Example:
1499
1500 bcsr@f8000000 {
1501 device_type = "board-control";
1502 reg = <f8000000 8000>;
1503 };
1504
1505 i) Freescale QUICC Engine module (QE)
1506 This represents qe module that is installed on PowerQUICC II Pro.
1507 Hopefully it will merge backward compatibility with CPM/CPM2.
1508 Basically, it is a bus of devices, that could act more or less
1509 as a complete entity (UCC, USB etc ). All of them should be siblings on
1510 the "root" qe node, using the common properties from there.
Michael Opdenacker59c51592007-05-09 08:57:56 +02001511 The description below applies to the qe of MPC8360 and
Li Yang9a1ab882006-10-02 20:08:59 -05001512 more nodes and properties would be extended in the future.
1513
1514 i) Root QE device
1515
1516 Required properties:
1517 - device_type : should be "qe";
1518 - model : precise model of the QE, Can be "QE", "CPM", or "CPM2"
1519 - reg : offset and length of the device registers.
1520 - bus-frequency : the clock frequency for QUICC Engine.
1521
1522 Recommended properties
1523 - brg-frequency : the internal clock source frequency for baud-rate
1524 generators in Hz.
1525
1526 Example:
1527 qe@e0100000 {
1528 #address-cells = <1>;
1529 #size-cells = <1>;
1530 #interrupt-cells = <2>;
1531 device_type = "qe";
1532 model = "QE";
1533 ranges = <0 e0100000 00100000>;
1534 reg = <e0100000 480>;
1535 brg-frequency = <0>;
1536 bus-frequency = <179A7B00>;
1537 }
1538
1539
1540 ii) SPI (Serial Peripheral Interface)
1541
1542 Required properties:
1543 - device_type : should be "spi".
1544 - compatible : should be "fsl_spi".
Domen Puncer5dd60162007-03-02 21:44:45 +11001545 - mode : the SPI operation mode, it can be "cpu" or "qe".
Li Yang9a1ab882006-10-02 20:08:59 -05001546 - reg : Offset and length of the register set for the device
1547 - interrupts : <a b> where a is the interrupt number and b is a
1548 field that represents an encoding of the sense and level
1549 information for the interrupt. This should be encoded based on
1550 the information in section 2) depending on the type of interrupt
1551 controller you have.
1552 - interrupt-parent : the phandle for the interrupt controller that
1553 services interrupts for this device.
1554
1555 Example:
1556 spi@4c0 {
1557 device_type = "spi";
1558 compatible = "fsl_spi";
1559 reg = <4c0 40>;
1560 interrupts = <82 0>;
1561 interrupt-parent = <700>;
1562 mode = "cpu";
1563 };
1564
1565
1566 iii) USB (Universal Serial Bus Controller)
1567
1568 Required properties:
1569 - device_type : should be "usb".
1570 - compatible : could be "qe_udc" or "fhci-hcd".
1571 - mode : the could be "host" or "slave".
1572 - reg : Offset and length of the register set for the device
1573 - interrupts : <a b> where a is the interrupt number and b is a
1574 field that represents an encoding of the sense and level
1575 information for the interrupt. This should be encoded based on
1576 the information in section 2) depending on the type of interrupt
1577 controller you have.
1578 - interrupt-parent : the phandle for the interrupt controller that
1579 services interrupts for this device.
1580
1581 Example(slave):
1582 usb@6c0 {
1583 device_type = "usb";
1584 compatible = "qe_udc";
1585 reg = <6c0 40>;
1586 interrupts = <8b 0>;
1587 interrupt-parent = <700>;
1588 mode = "slave";
1589 };
1590
1591
1592 iv) UCC (Unified Communications Controllers)
1593
1594 Required properties:
1595 - device_type : should be "network", "hldc", "uart", "transparent"
1596 "bisync" or "atm".
1597 - compatible : could be "ucc_geth" or "fsl_atm" and so on.
1598 - model : should be "UCC".
1599 - device-id : the ucc number(1-8), corresponding to UCCx in UM.
1600 - reg : Offset and length of the register set for the device
1601 - interrupts : <a b> where a is the interrupt number and b is a
1602 field that represents an encoding of the sense and level
1603 information for the interrupt. This should be encoded based on
1604 the information in section 2) depending on the type of interrupt
1605 controller you have.
1606 - interrupt-parent : the phandle for the interrupt controller that
1607 services interrupts for this device.
1608 - pio-handle : The phandle for the Parallel I/O port configuration.
1609 - rx-clock : represents the UCC receive clock source.
1610 0x00 : clock source is disabled;
1611 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1612 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1613 - tx-clock: represents the UCC transmit clock source;
1614 0x00 : clock source is disabled;
1615 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1616 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1617
1618 Required properties for network device_type:
1619 - mac-address : list of bytes representing the ethernet address.
1620 - phy-handle : The phandle for the PHY connected to this controller.
1621
Scott Woode0a2f282007-03-16 12:28:46 -05001622 Recommended properties:
1623 - linux,network-index : This is the intended "index" of this
1624 network device. This is used by the bootwrapper to interpret
1625 MAC addresses passed by the firmware when no information other
1626 than indices is available to associate an address with a device.
Kim Phillips60c19222007-04-24 07:26:10 +10001627 - phy-connection-type : a string naming the controller/PHY interface type,
1628 i.e., "mii" (default), "rmii", "gmii", "rgmii", "rgmii-id", "tbi",
1629 or "rtbi".
Scott Woode0a2f282007-03-16 12:28:46 -05001630
Li Yang9a1ab882006-10-02 20:08:59 -05001631 Example:
1632 ucc@2000 {
1633 device_type = "network";
1634 compatible = "ucc_geth";
1635 model = "UCC";
1636 device-id = <1>;
1637 reg = <2000 200>;
1638 interrupts = <a0 0>;
1639 interrupt-parent = <700>;
1640 mac-address = [ 00 04 9f 00 23 23 ];
1641 rx-clock = "none";
1642 tx-clock = "clk9";
1643 phy-handle = <212000>;
Kim Phillips60c19222007-04-24 07:26:10 +10001644 phy-connection-type = "gmii";
Li Yang9a1ab882006-10-02 20:08:59 -05001645 pio-handle = <140001>;
1646 };
1647
1648
1649 v) Parallel I/O Ports
1650
1651 This node configures Parallel I/O ports for CPUs with QE support.
1652 The node should reside in the "soc" node of the tree. For each
1653 device that using parallel I/O ports, a child node should be created.
1654 See the definition of the Pin configuration nodes below for more
1655 information.
1656
1657 Required properties:
1658 - device_type : should be "par_io".
1659 - reg : offset to the register set and its length.
1660 - num-ports : number of Parallel I/O ports
1661
1662 Example:
1663 par_io@1400 {
1664 reg = <1400 100>;
1665 #address-cells = <1>;
1666 #size-cells = <0>;
1667 device_type = "par_io";
1668 num-ports = <7>;
1669 ucc_pin@01 {
1670 ......
1671 };
1672
1673
1674 vi) Pin configuration nodes
1675
1676 Required properties:
1677 - linux,phandle : phandle of this node; likely referenced by a QE
1678 device.
1679 - pio-map : array of pin configurations. Each pin is defined by 6
1680 integers. The six numbers are respectively: port, pin, dir,
1681 open_drain, assignment, has_irq.
1682 - port : port number of the pin; 0-6 represent port A-G in UM.
1683 - pin : pin number in the port.
1684 - dir : direction of the pin, should encode as follows:
1685
1686 0 = The pin is disabled
1687 1 = The pin is an output
1688 2 = The pin is an input
1689 3 = The pin is I/O
1690
1691 - open_drain : indicates the pin is normal or wired-OR:
1692
1693 0 = The pin is actively driven as an output
1694 1 = The pin is an open-drain driver. As an output, the pin is
1695 driven active-low, otherwise it is three-stated.
1696
1697 - assignment : function number of the pin according to the Pin Assignment
1698 tables in User Manual. Each pin can have up to 4 possible functions in
1699 QE and two options for CPM.
Matt LaPlantea982ac02007-05-09 07:35:06 +02001700 - has_irq : indicates if the pin is used as source of external
Li Yang9a1ab882006-10-02 20:08:59 -05001701 interrupts.
1702
1703 Example:
1704 ucc_pin@01 {
1705 linux,phandle = <140001>;
1706 pio-map = <
1707 /* port pin dir open_drain assignment has_irq */
1708 0 3 1 0 1 0 /* TxD0 */
1709 0 4 1 0 1 0 /* TxD1 */
1710 0 5 1 0 1 0 /* TxD2 */
1711 0 6 1 0 1 0 /* TxD3 */
1712 1 6 1 0 3 0 /* TxD4 */
1713 1 7 1 0 1 0 /* TxD5 */
1714 1 9 1 0 2 0 /* TxD6 */
1715 1 a 1 0 2 0 /* TxD7 */
1716 0 9 2 0 1 0 /* RxD0 */
1717 0 a 2 0 1 0 /* RxD1 */
1718 0 b 2 0 1 0 /* RxD2 */
1719 0 c 2 0 1 0 /* RxD3 */
1720 0 d 2 0 1 0 /* RxD4 */
1721 1 1 2 0 2 0 /* RxD5 */
1722 1 0 2 0 2 0 /* RxD6 */
1723 1 4 2 0 2 0 /* RxD7 */
1724 0 7 1 0 1 0 /* TX_EN */
1725 0 8 1 0 1 0 /* TX_ER */
1726 0 f 2 0 1 0 /* RX_DV */
1727 0 10 2 0 1 0 /* RX_ER */
1728 0 0 2 0 1 0 /* RX_CLK */
1729 2 9 1 0 3 0 /* GTX_CLK - CLK10 */
1730 2 8 2 0 1 0>; /* GTX125 - CLK9 */
1731 };
1732
1733 vii) Multi-User RAM (MURAM)
1734
1735 Required properties:
1736 - device_type : should be "muram".
1737 - mode : the could be "host" or "slave".
1738 - ranges : Should be defined as specified in 1) to describe the
1739 translation of MURAM addresses.
1740 - data-only : sub-node which defines the address area under MURAM
1741 bus that can be allocated as data/parameter
1742
1743 Example:
1744
1745 muram@10000 {
1746 device_type = "muram";
1747 ranges = <0 00010000 0000c000>;
1748
1749 data-only@0{
1750 reg = <0 c000>;
1751 };
1752 };
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001753
Stuart Yoderd30ac122007-06-05 07:16:15 +10001754 j) Flash chip nodes
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001755
1756 Flash chips (Memory Technology Devices) are often used for solid state
1757 file systems on embedded devices.
1758
1759 Required properties:
1760
1761 - device_type : has to be "rom"
Vitaly Wool173935f2006-12-19 18:44:25 +03001762 - compatible : Should specify what this flash device is compatible with.
1763 Currently, this is most likely to be "direct-mapped" (which
1764 corresponds to the MTD physmap mapping driver).
1765 - reg : Offset and length of the register set (or memory mapping) for
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001766 the device.
Vitaly Wool173935f2006-12-19 18:44:25 +03001767 - bank-width : Width of the flash data bus in bytes. Required
1768 for the NOR flashes (compatible == "direct-mapped" and others) ONLY.
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001769
1770 Recommended properties :
1771
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001772 - partitions : Several pairs of 32-bit values where the first value is
1773 partition's offset from the start of the device and the second one is
1774 partition size in bytes with LSB used to signify a read only
Domen Puncer5dd60162007-03-02 21:44:45 +11001775 partition (so, the partition size should always be an even number).
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001776 - partition-names : The list of concatenated zero terminated strings
1777 representing the partition names.
Vitaly Wool173935f2006-12-19 18:44:25 +03001778 - probe-type : The type of probe which should be done for the chip
1779 (JEDEC vs CFI actually). Valid ONLY for NOR flashes.
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001780
1781 Example:
1782
1783 flash@ff000000 {
1784 device_type = "rom";
1785 compatible = "direct-mapped";
Vitaly Wool173935f2006-12-19 18:44:25 +03001786 probe-type = "CFI";
1787 reg = <ff000000 01000000>;
Vitaly Wool28f9ec32006-11-20 16:32:39 +03001788 bank-width = <4>;
1789 partitions = <00000000 00f80000
1790 00f80000 00080001>;
1791 partition-names = "fs\0firmware";
1792 };
1793
Roy Zang3b824f82007-06-19 15:19:18 +08001794 k) Global Utilities Block
1795
1796 The global utilities block controls power management, I/O device
1797 enabling, power-on-reset configuration monitoring, general-purpose
1798 I/O signal configuration, alternate function selection for multiplexed
1799 signals, and clock control.
1800
1801 Required properties:
1802
1803 - compatible : Should define the compatible device type for
1804 global-utilities.
1805 - reg : Offset and length of the register set for the device.
1806
1807 Recommended properties:
1808
1809 - fsl,has-rstcr : Indicates that the global utilities register set
1810 contains a functioning "reset control register" (i.e. the board
1811 is wired to reset upon setting the HRESET_REQ bit in this register).
1812
1813 Example:
1814
1815 global-utilities@e0000 { /* global utilities block */
1816 compatible = "fsl,mpc8548-guts";
1817 reg = <e0000 1000>;
1818 fsl,has-rstcr;
1819 };
1820
David Gibsonc125a182006-02-01 03:05:22 -08001821 More devices will be defined as this spec matures.
1822
Stuart Yoder27565902007-03-02 13:42:33 -06001823VII - Specifying interrupt information for devices
1824===================================================
1825
1826The device tree represents the busses and devices of a hardware
1827system in a form similar to the physical bus topology of the
1828hardware.
1829
1830In addition, a logical 'interrupt tree' exists which represents the
1831hierarchy and routing of interrupts in the hardware.
1832
1833The interrupt tree model is fully described in the
1834document "Open Firmware Recommended Practice: Interrupt
1835Mapping Version 0.9". The document is available at:
1836<http://playground.sun.com/1275/practice>.
1837
18381) interrupts property
1839----------------------
1840
1841Devices that generate interrupts to a single interrupt controller
1842should use the conventional OF representation described in the
1843OF interrupt mapping documentation.
1844
1845Each device which generates interrupts must have an 'interrupt'
1846property. The interrupt property value is an arbitrary number of
1847of 'interrupt specifier' values which describe the interrupt or
1848interrupts for the device.
1849
1850The encoding of an interrupt specifier is determined by the
1851interrupt domain in which the device is located in the
1852interrupt tree. The root of an interrupt domain specifies in
1853its #interrupt-cells property the number of 32-bit cells
1854required to encode an interrupt specifier. See the OF interrupt
1855mapping documentation for a detailed description of domains.
1856
1857For example, the binding for the OpenPIC interrupt controller
1858specifies an #interrupt-cells value of 2 to encode the interrupt
1859number and level/sense information. All interrupt children in an
1860OpenPIC interrupt domain use 2 cells per interrupt in their interrupts
1861property.
1862
1863The PCI bus binding specifies a #interrupt-cell value of 1 to encode
1864which interrupt pin (INTA,INTB,INTC,INTD) is used.
1865
18662) interrupt-parent property
1867----------------------------
1868
1869The interrupt-parent property is specified to define an explicit
1870link between a device node and its interrupt parent in
1871the interrupt tree. The value of interrupt-parent is the
1872phandle of the parent node.
1873
1874If the interrupt-parent property is not defined for a node, it's
1875interrupt parent is assumed to be an ancestor in the node's
1876_device tree_ hierarchy.
1877
18783) OpenPIC Interrupt Controllers
1879--------------------------------
1880
1881OpenPIC interrupt controllers require 2 cells to encode
1882interrupt information. The first cell defines the interrupt
1883number. The second cell defines the sense and level
1884information.
1885
1886Sense and level information should be encoded as follows:
1887
1888 0 = low to high edge sensitive type enabled
1889 1 = active low level sensitive type enabled
1890 2 = active high level sensitive type enabled
1891 3 = high to low edge sensitive type enabled
1892
18934) ISA Interrupt Controllers
1894----------------------------
1895
1896ISA PIC interrupt controllers require 2 cells to encode
1897interrupt information. The first cell defines the interrupt
1898number. The second cell defines the sense and level
1899information.
1900
1901ISA PIC interrupt controllers should adhere to the ISA PIC
1902encodings listed below:
1903
1904 0 = active low level sensitive type enabled
1905 1 = active high level sensitive type enabled
1906 2 = high to low edge sensitive type enabled
1907 3 = low to high edge sensitive type enabled
1908
David Gibsonc125a182006-02-01 03:05:22 -08001909
1910Appendix A - Sample SOC node for MPC8540
1911========================================
1912
1913Note that the #address-cells and #size-cells for the SoC node
1914in this example have been explicitly listed; these are likely
1915not necessary as they are usually the same as the root node.
1916
1917 soc8540@e0000000 {
1918 #address-cells = <1>;
1919 #size-cells = <1>;
1920 #interrupt-cells = <2>;
1921 device_type = "soc";
1922 ranges = <00000000 e0000000 00100000>
1923 reg = <e0000000 00003000>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -06001924 bus-frequency = <0>;
David Gibsonc125a182006-02-01 03:05:22 -08001925
1926 mdio@24520 {
1927 reg = <24520 20>;
1928 device_type = "mdio";
1929 compatible = "gianfar";
1930
1931 ethernet-phy@0 {
1932 linux,phandle = <2452000>
1933 interrupt-parent = <40000>;
1934 interrupts = <35 1>;
1935 reg = <0>;
1936 device_type = "ethernet-phy";
1937 };
1938
1939 ethernet-phy@1 {
1940 linux,phandle = <2452001>
1941 interrupt-parent = <40000>;
1942 interrupts = <35 1>;
1943 reg = <1>;
1944 device_type = "ethernet-phy";
1945 };
1946
1947 ethernet-phy@3 {
1948 linux,phandle = <2452002>
1949 interrupt-parent = <40000>;
1950 interrupts = <35 1>;
1951 reg = <3>;
1952 device_type = "ethernet-phy";
1953 };
1954
1955 };
1956
1957 ethernet@24000 {
1958 #size-cells = <0>;
1959 device_type = "network";
1960 model = "TSEC";
1961 compatible = "gianfar";
1962 reg = <24000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001963 mac-address = [ 00 E0 0C 00 73 00 ];
David Gibsonc125a182006-02-01 03:05:22 -08001964 interrupts = <d 3 e 3 12 3>;
1965 interrupt-parent = <40000>;
1966 phy-handle = <2452000>;
1967 };
1968
1969 ethernet@25000 {
1970 #address-cells = <1>;
1971 #size-cells = <0>;
1972 device_type = "network";
1973 model = "TSEC";
1974 compatible = "gianfar";
1975 reg = <25000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001976 mac-address = [ 00 E0 0C 00 73 01 ];
David Gibsonc125a182006-02-01 03:05:22 -08001977 interrupts = <13 3 14 3 18 3>;
1978 interrupt-parent = <40000>;
1979 phy-handle = <2452001>;
1980 };
1981
1982 ethernet@26000 {
1983 #address-cells = <1>;
1984 #size-cells = <0>;
1985 device_type = "network";
1986 model = "FEC";
1987 compatible = "gianfar";
1988 reg = <26000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001989 mac-address = [ 00 E0 0C 00 73 02 ];
David Gibsonc125a182006-02-01 03:05:22 -08001990 interrupts = <19 3>;
1991 interrupt-parent = <40000>;
1992 phy-handle = <2452002>;
1993 };
1994
1995 serial@4500 {
1996 device_type = "serial";
1997 compatible = "ns16550";
1998 reg = <4500 100>;
1999 clock-frequency = <0>;
2000 interrupts = <1a 3>;
2001 interrupt-parent = <40000>;
2002 };
2003
2004 pic@40000 {
2005 linux,phandle = <40000>;
2006 clock-frequency = <0>;
2007 interrupt-controller;
2008 #address-cells = <0>;
2009 reg = <40000 40000>;
2010 built-in;
2011 compatible = "chrp,open-pic";
2012 device_type = "open-pic";
2013 big-endian;
2014 };
2015
2016 i2c@3000 {
2017 interrupt-parent = <40000>;
2018 interrupts = <1b 3>;
2019 reg = <3000 18>;
2020 device_type = "i2c";
2021 compatible = "fsl-i2c";
2022 dfsrr;
2023 };
2024
2025 };