1 Booting the Linux/ppc kernel without Open Firmware
2 --------------------------------------------------
5 (c) 2005 Benjamin Herrenschmidt <benh at kernel.crashing.org>,
7 (c) 2005 Becky Bruce <becky.bruce at freescale.com>,
8 Freescale Semiconductor, FSL SOC and 32-bit additions
9 (c) 2006 MontaVista Software, Inc.
10 Flash chip node definition
12 May 18, 2005: Rev 0.1 - Initial draft, no chapter III yet.
14 May 19, 2005: Rev 0.2 - Add chapter III and bits & pieces here or
15 clarifies the fact that a lot of things are
16 optional, the kernel only requires a very
17 small device tree, though it is encouraged
18 to provide an as complete one as possible.
20 May 24, 2005: Rev 0.3 - Precise that DT block has to be in RAM
22 - Define version 3 and new format version 16
23 for the DT block (version 16 needs kernel
24 patches, will be fwd separately).
25 String block now has a size, and full path
26 is replaced by unit name for more
28 linux,phandle is made optional, only nodes
29 that are referenced by other nodes need it.
30 "name" property is now automatically
31 deduced from the unit name
33 June 1, 2005: Rev 0.4 - Correct confusion between OF_DT_END and
34 OF_DT_END_NODE in structure definition.
35 - Change version 16 format to always align
36 property data to 4 bytes. Since tokens are
37 already aligned, that means no specific
38 required alignment between property size
39 and property data. The old style variable
40 alignment would make it impossible to do
41 "simple" insertion of properties using
42 memove (thanks Milton for
43 noticing). Updated kernel patch as well
44 - Correct a few more alignment constraints
45 - Add a chapter about the device-tree
46 compiler and the textural representation of
47 the tree that can be "compiled" by dtc.
49 November 21, 2005: Rev 0.5
50 - Additions/generalizations for 32-bit
51 - Changed to reflect the new arch/powerpc
57 - Add some definitions of interrupt tree (simple/complex)
58 - Add some definitions for pci host bridges
59 - Add some common address format examples
60 - Add definitions for standard properties and "compatible"
61 names for cells that are not already defined by the existing
63 - Compare FSL SOC use of PCI to standard and make sure no new
64 node definition required.
65 - Add more information about node definitions for SOC devices
66 that currently have no standard, like the FSL CPM.
72 During the recent development of the Linux/ppc64 kernel, and more
73 specifically, the addition of new platform types outside of the old
74 IBM pSeries/iSeries pair, it was decided to enforce some strict rules
75 regarding the kernel entry and bootloader <-> kernel interfaces, in
76 order to avoid the degeneration that had become the ppc32 kernel entry
77 point and the way a new platform should be added to the kernel. The
78 legacy iSeries platform breaks those rules as it predates this scheme,
79 but no new board support will be accepted in the main tree that
80 doesn't follows them properly. In addition, since the advent of the
81 arch/powerpc merged architecture for ppc32 and ppc64, new 32-bit
82 platforms and 32-bit platforms which move into arch/powerpc will be
83 required to use these rules as well.
85 The main requirement that will be defined in more detail below is
86 the presence of a device-tree whose format is defined after Open
87 Firmware specification. However, in order to make life easier
88 to embedded board vendors, the kernel doesn't require the device-tree
89 to represent every device in the system and only requires some nodes
90 and properties to be present. This will be described in detail in
91 section III, but, for example, the kernel does not require you to
92 create a node for every PCI device in the system. It is a requirement
93 to have a node for PCI host bridges in order to provide interrupt
94 routing informations and memory/IO ranges, among others. It is also
95 recommended to define nodes for on chip devices and other busses that
96 don't specifically fit in an existing OF specification. This creates a
97 great flexibility in the way the kernel can then probe those and match
98 drivers to device, without having to hard code all sorts of tables. It
99 also makes it more flexible for board vendors to do minor hardware
100 upgrades without significantly impacting the kernel code or cluttering
101 it with special cases.
104 1) Entry point for arch/powerpc
105 -------------------------------
107 There is one and one single entry point to the kernel, at the start
108 of the kernel image. That entry point supports two calling
111 a) Boot from Open Firmware. If your firmware is compatible
112 with Open Firmware (IEEE 1275) or provides an OF compatible
113 client interface API (support for "interpret" callback of
114 forth words isn't required), you can enter the kernel with:
116 r5 : OF callback pointer as defined by IEEE 1275
117 bindings to powerpc. Only the 32 bit client interface
118 is currently supported
120 r3, r4 : address & length of an initrd if any or 0
122 The MMU is either on or off; the kernel will run the
123 trampoline located in arch/powerpc/kernel/prom_init.c to
124 extract the device-tree and other information from open
125 firmware and build a flattened device-tree as described
126 in b). prom_init() will then re-enter the kernel using
127 the second method. This trampoline code runs in the
128 context of the firmware, which is supposed to handle all
129 exceptions during that time.
131 b) Direct entry with a flattened device-tree block. This entry
132 point is called by a) after the OF trampoline and can also be
133 called directly by a bootloader that does not support the Open
134 Firmware client interface. It is also used by "kexec" to
135 implement "hot" booting of a new kernel from a previous
136 running one. This method is what I will describe in more
137 details in this document, as method a) is simply standard Open
138 Firmware, and thus should be implemented according to the
139 various standard documents defining it and its binding to the
140 PowerPC platform. The entry point definition then becomes:
142 r3 : physical pointer to the device-tree block
143 (defined in chapter II) in RAM
145 r4 : physical pointer to the kernel itself. This is
146 used by the assembly code to properly disable the MMU
147 in case you are entering the kernel with MMU enabled
148 and a non-1:1 mapping.
150 r5 : NULL (as to differentiate with method a)
152 Note about SMP entry: Either your firmware puts your other
153 CPUs in some sleep loop or spin loop in ROM where you can get
154 them out via a soft reset or some other means, in which case
155 you don't need to care, or you'll have to enter the kernel
156 with all CPUs. The way to do that with method b) will be
157 described in a later revision of this document.
165 Board supports (platforms) are not exclusive config options. An
166 arbitrary set of board supports can be built in a single kernel
167 image. The kernel will "know" what set of functions to use for a
168 given platform based on the content of the device-tree. Thus, you
171 a) add your platform support as a _boolean_ option in
172 arch/powerpc/Kconfig, following the example of PPC_PSERIES,
173 PPC_PMAC and PPC_MAPLE. The later is probably a good
174 example of a board support to start from.
176 b) create your main platform file as
177 "arch/powerpc/platforms/myplatform/myboard_setup.c" and add it
178 to the Makefile under the condition of your CONFIG_
179 option. This file will define a structure of type "ppc_md"
180 containing the various callbacks that the generic code will
181 use to get to your platform specific code
183 c) Add a reference to your "ppc_md" structure in the
184 "machines" table in arch/powerpc/kernel/setup_64.c if you are
187 d) request and get assigned a platform number (see PLATFORM_*
188 constants in include/asm-powerpc/processor.h
190 32-bit embedded kernels:
192 Currently, board support is essentially an exclusive config option.
193 The kernel is configured for a single platform. Part of the reason
194 for this is to keep kernels on embedded systems small and efficient;
195 part of this is due to the fact the code is already that way. In the
196 future, a kernel may support multiple platforms, but only if the
197 platforms feature the same core architectire. A single kernel build
198 cannot support both configurations with Book E and configurations
199 with classic Powerpc architectures.
201 32-bit embedded platforms that are moved into arch/powerpc using a
202 flattened device tree should adopt the merged tree practice of
203 setting ppc_md up dynamically, even though the kernel is currently
204 built with support for only a single platform at a time. This allows
205 unification of the setup code, and will make it easier to go to a
206 multiple-platform-support model in the future.
208 NOTE: I believe the above will be true once Ben's done with the merge
209 of the boot sequences.... someone speak up if this is wrong!
211 To add a 32-bit embedded platform support, follow the instructions
212 for 64-bit platforms above, with the exception that the Kconfig
213 option should be set up such that the kernel builds exclusively for
214 the platform selected. The processor type for the platform should
215 enable another config option to select the specific board
218 NOTE: If ben doesn't merge the setup files, may need to change this to
222 I will describe later the boot process and various callbacks that
223 your platform should implement.
226 II - The DT block format
227 ========================
230 This chapter defines the actual format of the flattened device-tree
231 passed to the kernel. The actual content of it and kernel requirements
232 are described later. You can find example of code manipulating that
233 format in various places, including arch/powerpc/kernel/prom_init.c
234 which will generate a flattened device-tree from the Open Firmware
235 representation, or the fs2dt utility which is part of the kexec tools
236 which will generate one from a filesystem representation. It is
237 expected that a bootloader like uboot provides a bit more support,
238 that will be discussed later as well.
240 Note: The block has to be in main memory. It has to be accessible in
241 both real mode and virtual mode with no mapping other than main
242 memory. If you are writing a simple flash bootloader, it should copy
243 the block to RAM before passing it to the kernel.
249 The kernel is entered with r3 pointing to an area of memory that is
250 roughly described in include/asm-powerpc/prom.h by the structure
253 struct boot_param_header {
254 u32 magic; /* magic word OF_DT_HEADER */
255 u32 totalsize; /* total size of DT block */
256 u32 off_dt_struct; /* offset to structure */
257 u32 off_dt_strings; /* offset to strings */
258 u32 off_mem_rsvmap; /* offset to memory reserve map
260 u32 version; /* format version */
261 u32 last_comp_version; /* last compatible version */
263 /* version 2 fields below */
264 u32 boot_cpuid_phys; /* Which physical CPU id we're
266 /* version 3 fields below */
267 u32 size_dt_strings; /* size of the strings block */
270 Along with the constants:
272 /* Definitions used by the flattened device tree */
273 #define OF_DT_HEADER 0xd00dfeed /* 4: version,
275 #define OF_DT_BEGIN_NODE 0x1 /* Start node: full name
277 #define OF_DT_END_NODE 0x2 /* End node */
278 #define OF_DT_PROP 0x3 /* Property: name off,
280 #define OF_DT_END 0x9
282 All values in this header are in big endian format, the various
283 fields in this header are defined more precisely below. All
284 "offset" values are in bytes from the start of the header; that is
285 from the value of r3.
289 This is a magic value that "marks" the beginning of the
290 device-tree block header. It contains the value 0xd00dfeed and is
291 defined by the constant OF_DT_HEADER
295 This is the total size of the DT block including the header. The
296 "DT" block should enclose all data structures defined in this
297 chapter (who are pointed to by offsets in this header). That is,
298 the device-tree structure, strings, and the memory reserve map.
302 This is an offset from the beginning of the header to the start
303 of the "structure" part the device tree. (see 2) device tree)
307 This is an offset from the beginning of the header to the start
308 of the "strings" part of the device-tree
312 This is an offset from the beginning of the header to the start
313 of the reserved memory map. This map is a list of pairs of 64
314 bit integers. Each pair is a physical address and a size. The
316 list is terminated by an entry of size 0. This map provides the
317 kernel with a list of physical memory areas that are "reserved"
318 and thus not to be used for memory allocations, especially during
319 early initialization. The kernel needs to allocate memory during
320 boot for things like un-flattening the device-tree, allocating an
321 MMU hash table, etc... Those allocations must be done in such a
322 way to avoid overriding critical things like, on Open Firmware
323 capable machines, the RTAS instance, or on some pSeries, the TCE
324 tables used for the iommu. Typically, the reserve map should
325 contain _at least_ this DT block itself (header,total_size). If
326 you are passing an initrd to the kernel, you should reserve it as
327 well. You do not need to reserve the kernel image itself. The map
328 should be 64 bit aligned.
332 This is the version of this structure. Version 1 stops
333 here. Version 2 adds an additional field boot_cpuid_phys.
334 Version 3 adds the size of the strings block, allowing the kernel
335 to reallocate it easily at boot and free up the unused flattened
336 structure after expansion. Version 16 introduces a new more
337 "compact" format for the tree itself that is however not backward
338 compatible. You should always generate a structure of the highest
339 version defined at the time of your implementation. Currently
340 that is version 16, unless you explicitly aim at being backward
345 Last compatible version. This indicates down to what version of
346 the DT block you are backward compatible. For example, version 2
347 is backward compatible with version 1 (that is, a kernel build
348 for version 1 will be able to boot with a version 2 format). You
349 should put a 1 in this field if you generate a device tree of
350 version 1 to 3, or 0x10 if you generate a tree of version 0x10
351 using the new unit name format.
355 This field only exist on version 2 headers. It indicate which
356 physical CPU ID is calling the kernel entry point. This is used,
357 among others, by kexec. If you are on an SMP system, this value
358 should match the content of the "reg" property of the CPU node in
359 the device-tree corresponding to the CPU calling the kernel entry
360 point (see further chapters for more informations on the required
361 device-tree contents)
364 So the typical layout of a DT block (though the various parts don't
365 need to be in that order) looks like this (addresses go from top to
369 ------------------------------
370 r3 -> | struct boot_param_header |
371 ------------------------------
372 | (alignment gap) (*) |
373 ------------------------------
374 | memory reserve map |
375 ------------------------------
377 ------------------------------
379 | device-tree structure |
381 ------------------------------
383 ------------------------------
385 | device-tree strings |
387 -----> ------------------------------
392 (*) The alignment gaps are not necessarily present; their presence
393 and size are dependent on the various alignment requirements of
394 the individual data blocks.
397 2) Device tree generalities
398 ---------------------------
400 This device-tree itself is separated in two different blocks, a
401 structure block and a strings block. Both need to be aligned to a 4
404 First, let's quickly describe the device-tree concept before detailing
405 the storage format. This chapter does _not_ describe the detail of the
406 required types of nodes & properties for the kernel, this is done
407 later in chapter III.
409 The device-tree layout is strongly inherited from the definition of
410 the Open Firmware IEEE 1275 device-tree. It's basically a tree of
411 nodes, each node having two or more named properties. A property can
414 It is a tree, so each node has one and only one parent except for the
415 root node who has no parent.
417 A node has 2 names. The actual node name is generally contained in a
418 property of type "name" in the node property list whose value is a
419 zero terminated string and is mandatory for version 1 to 3 of the
420 format definition (as it is in Open Firmware). Version 0x10 makes it
421 optional as it can generate it from the unit name defined below.
423 There is also a "unit name" that is used to differentiate nodes with
424 the same name at the same level, it is usually made of the node
425 names, the "@" sign, and a "unit address", which definition is
426 specific to the bus type the node sits on.
428 The unit name doesn't exist as a property per-se but is included in
429 the device-tree structure. It is typically used to represent "path" in
430 the device-tree. More details about the actual format of these will be
433 The kernel powerpc generic code does not make any formal use of the
434 unit address (though some board support code may do) so the only real
435 requirement here for the unit address is to ensure uniqueness of
436 the node unit name at a given level of the tree. Nodes with no notion
437 of address and no possible sibling of the same name (like /memory or
438 /cpus) may omit the unit address in the context of this specification,
439 or use the "@0" default unit address. The unit name is used to define
440 a node "full path", which is the concatenation of all parent node
441 unit names separated with "/".
443 The root node doesn't have a defined name, and isn't required to have
444 a name property either if you are using version 3 or earlier of the
445 format. It also has no unit address (no @ symbol followed by a unit
446 address). The root node unit name is thus an empty string. The full
447 path to the root node is "/".
449 Every node which actually represents an actual device (that is, a node
450 which isn't only a virtual "container" for more nodes, like "/cpus"
451 is) is also required to have a "device_type" property indicating the
454 Finally, every node that can be referenced from a property in another
455 node is required to have a "linux,phandle" property. Real open
456 firmware implementations provide a unique "phandle" value for every
457 node that the "prom_init()" trampoline code turns into
458 "linux,phandle" properties. However, this is made optional if the
459 flattened device tree is used directly. An example of a node
460 referencing another node via "phandle" is when laying out the
461 interrupt tree which will be described in a further version of this
464 This "linux, phandle" property is a 32 bit value that uniquely
465 identifies a node. You are free to use whatever values or system of
466 values, internal pointers, or whatever to generate these, the only
467 requirement is that every node for which you provide that property has
468 a unique value for it.
470 Here is an example of a simple device-tree. In this example, an "o"
471 designates a node followed by the node unit name. Properties are
472 presented with their name followed by their content. "content"
473 represents an ASCII string (zero terminated) value, while <content>
474 represents a 32 bit hexadecimal value. The various nodes in this
475 example will be discussed in a later chapter. At this point, it is
476 only meant to give you a idea of what a device-tree looks like. I have
477 purposefully kept the "name" and "linux,phandle" properties which
478 aren't necessary in order to give you a better idea of what the tree
479 looks like in practice.
482 |- name = "device-tree"
483 |- model = "MyBoardName"
484 |- compatible = "MyBoardFamilyName"
485 |- #address-cells = <2>
487 |- linux,phandle = <0>
491 | | - linux,phandle = <1>
492 | | - #address-cells = <1>
493 | | - #size-cells = <0>
496 | |- name = "PowerPC,970"
497 | |- device_type = "cpu"
499 | |- clock-frequency = <5f5e1000>
501 | |- linux,phandle = <2>
505 | |- device_type = "memory"
506 | |- reg = <00000000 00000000 00000000 20000000>
507 | |- linux,phandle = <3>
511 |- bootargs = "root=/dev/sda2"
512 |- linux,phandle = <4>
514 This tree is almost a minimal tree. It pretty much contains the
515 minimal set of required nodes and properties to boot a linux kernel;
516 that is, some basic model informations at the root, the CPUs, and the
517 physical memory layout. It also includes misc information passed
518 through /chosen, like in this example, the platform type (mandatory)
519 and the kernel command line arguments (optional).
521 The /cpus/PowerPC,970@0/64-bit property is an example of a
522 property without a value. All other properties have a value. The
523 significance of the #address-cells and #size-cells properties will be
524 explained in chapter IV which defines precisely the required nodes and
525 properties and their content.
528 3) Device tree "structure" block
530 The structure of the device tree is a linearized tree structure. The
531 "OF_DT_BEGIN_NODE" token starts a new node, and the "OF_DT_END_NODE"
532 ends that node definition. Child nodes are simply defined before
533 "OF_DT_END_NODE" (that is nodes within the node). A 'token' is a 32
534 bit value. The tree has to be "finished" with a OF_DT_END token
536 Here's the basic structure of a single node:
538 * token OF_DT_BEGIN_NODE (that is 0x00000001)
539 * for version 1 to 3, this is the node full path as a zero
540 terminated string, starting with "/". For version 16 and later,
541 this is the node unit name only (or an empty string for the
543 * [align gap to next 4 bytes boundary]
545 * token OF_DT_PROP (that is 0x00000003)
546 * 32 bit value of property value size in bytes (or 0 of no
548 * 32 bit value of offset in string block of property name
549 * property value data if any
550 * [align gap to next 4 bytes boundary]
551 * [child nodes if any]
552 * token OF_DT_END_NODE (that is 0x00000002)
554 So the node content can be summarised as a start token, a full path,
555 a list of properties, a list of child nodes, and an end token. Every
556 child node is a full node structure itself as defined above.
558 4) Device tree "strings" block
560 In order to save space, property names, which are generally redundant,
561 are stored separately in the "strings" block. This block is simply the
562 whole bunch of zero terminated strings for all property names
563 concatenated together. The device-tree property definitions in the
564 structure block will contain offset values from the beginning of the
568 III - Required content of the device tree
569 =========================================
571 WARNING: All "linux,*" properties defined in this document apply only
572 to a flattened device-tree. If your platform uses a real
573 implementation of Open Firmware or an implementation compatible with
574 the Open Firmware client interface, those properties will be created
575 by the trampoline code in the kernel's prom_init() file. For example,
576 that's where you'll have to add code to detect your board model and
577 set the platform number. However, when using the flattened device-tree
578 entry point, there is no prom_init() pass, and thus you have to
579 provide those properties yourself.
582 1) Note about cells and address representation
583 ----------------------------------------------
585 The general rule is documented in the various Open Firmware
586 documentations. If you chose to describe a bus with the device-tree
587 and there exist an OF bus binding, then you should follow the
588 specification. However, the kernel does not require every single
589 device or bus to be described by the device tree.
591 In general, the format of an address for a device is defined by the
592 parent bus type, based on the #address-cells and #size-cells
593 property. In the absence of such a property, the parent's parent
594 values are used, etc... The kernel requires the root node to have
595 those properties defining addresses format for devices directly mapped
596 on the processor bus.
598 Those 2 properties define 'cells' for representing an address and a
599 size. A "cell" is a 32 bit number. For example, if both contain 2
600 like the example tree given above, then an address and a size are both
601 composed of 2 cells, and each is a 64 bit number (cells are
602 concatenated and expected to be in big endian format). Another example
603 is the way Apple firmware defines them, with 2 cells for an address
604 and one cell for a size. Most 32-bit implementations should define
605 #address-cells and #size-cells to 1, which represents a 32-bit value.
606 Some 32-bit processors allow for physical addresses greater than 32
607 bits; these processors should define #address-cells as 2.
609 "reg" properties are always a tuple of the type "address size" where
610 the number of cells of address and size is specified by the bus
611 #address-cells and #size-cells. When a bus supports various address
612 spaces and other flags relative to a given address allocation (like
613 prefetchable, etc...) those flags are usually added to the top level
614 bits of the physical address. For example, a PCI physical address is
615 made of 3 cells, the bottom two containing the actual address itself
616 while the top cell contains address space indication, flags, and pci
617 bus & device numbers.
619 For busses that support dynamic allocation, it's the accepted practice
620 to then not provide the address in "reg" (keep it 0) though while
621 providing a flag indicating the address is dynamically allocated, and
622 then, to provide a separate "assigned-addresses" property that
623 contains the fully allocated addresses. See the PCI OF bindings for
626 In general, a simple bus with no address space bits and no dynamic
627 allocation is preferred if it reflects your hardware, as the existing
628 kernel address parsing functions will work out of the box. If you
629 define a bus type with a more complex address format, including things
630 like address space bits, you'll have to add a bus translator to the
631 prom_parse.c file of the recent kernels for your bus type.
633 The "reg" property only defines addresses and sizes (if #size-cells
634 is non-0) within a given bus. In order to translate addresses upward
635 (that is into parent bus addresses, and possibly into cpu physical
636 addresses), all busses must contain a "ranges" property. If the
637 "ranges" property is missing at a given level, it's assumed that
638 translation isn't possible. The format of the "ranges" property for a
641 bus address, parent bus address, size
643 "bus address" is in the format of the bus this bus node is defining,
644 that is, for a PCI bridge, it would be a PCI address. Thus, (bus
645 address, size) defines a range of addresses for child devices. "parent
646 bus address" is in the format of the parent bus of this bus. For
647 example, for a PCI host controller, that would be a CPU address. For a
648 PCI<->ISA bridge, that would be a PCI address. It defines the base
649 address in the parent bus where the beginning of that range is mapped.
651 For a new 64 bit powerpc board, I recommend either the 2/2 format or
652 Apple's 2/1 format which is slightly more compact since sizes usually
653 fit in a single 32 bit word. New 32 bit powerpc boards should use a
654 1/1 format, unless the processor supports physical addresses greater
655 than 32-bits, in which case a 2/1 format is recommended.
658 2) Note about "compatible" properties
659 -------------------------------------
661 These properties are optional, but recommended in devices and the root
662 node. The format of a "compatible" property is a list of concatenated
663 zero terminated strings. They allow a device to express its
664 compatibility with a family of similar devices, in some cases,
665 allowing a single driver to match against several devices regardless
666 of their actual names.
668 3) Note about "name" properties
669 -------------------------------
671 While earlier users of Open Firmware like OldWorld macintoshes tended
672 to use the actual device name for the "name" property, it's nowadays
673 considered a good practice to use a name that is closer to the device
674 class (often equal to device_type). For example, nowadays, ethernet
675 controllers are named "ethernet", an additional "model" property
676 defining precisely the chip type/model, and "compatible" property
677 defining the family in case a single driver can driver more than one
678 of these chips. However, the kernel doesn't generally put any
679 restriction on the "name" property; it is simply considered good
680 practice to follow the standard and its evolutions as closely as
683 Note also that the new format version 16 makes the "name" property
684 optional. If it's absent for a node, then the node's unit name is then
685 used to reconstruct the name. That is, the part of the unit name
686 before the "@" sign is used (or the entire unit name if no "@" sign
689 4) Note about node and property names and character set
690 -------------------------------------------------------
692 While open firmware provides more flexible usage of 8859-1, this
693 specification enforces more strict rules. Nodes and properties should
694 be comprised only of ASCII characters 'a' to 'z', '0' to
695 '9', ',', '.', '_', '+', '#', '?', and '-'. Node names additionally
696 allow uppercase characters 'A' to 'Z' (property names should be
697 lowercase. The fact that vendors like Apple don't respect this rule is
698 irrelevant here). Additionally, node and property names should always
699 begin with a character in the range 'a' to 'z' (or 'A' to 'Z' for node
702 The maximum number of characters for both nodes and property names
703 is 31. In the case of node names, this is only the leftmost part of
704 a unit name (the pure "name" property), it doesn't include the unit
705 address which can extend beyond that limit.
708 5) Required nodes and properties
709 --------------------------------
710 These are all that are currently required. However, it is strongly
711 recommended that you expose PCI host bridges as documented in the
712 PCI binding to open firmware, and your interrupt tree as documented
713 in OF interrupt tree specification.
717 The root node requires some properties to be present:
719 - model : this is your board name/model
720 - #address-cells : address representation for "root" devices
721 - #size-cells: the size representation for "root" devices
722 - device_type : This property shouldn't be necessary. However, if
723 you decide to create a device_type for your root node, make sure it
724 is _not_ "chrp" unless your platform is a pSeries or PAPR compliant
725 one for 64-bit, or a CHRP-type machine for 32-bit as this will
726 matched by the kernel this way.
728 Additionally, some recommended properties are:
730 - compatible : the board "family" generally finds its way here,
731 for example, if you have 2 board models with a similar layout,
732 that typically get driven by the same platform code in the
733 kernel, you would use a different "model" property but put a
734 value in "compatible". The kernel doesn't directly use that
735 value but it is generally useful.
737 The root node is also generally where you add additional properties
738 specific to your board like the serial number if any, that sort of
739 thing. It is recommended that if you add any "custom" property whose
740 name may clash with standard defined ones, you prefix them with your
741 vendor name and a comma.
745 This node is the parent of all individual CPU nodes. It doesn't
746 have any specific requirements, though it's generally good practice
749 #address-cells = <00000001>
750 #size-cells = <00000000>
752 This defines that the "address" for a CPU is a single cell, and has
753 no meaningful size. This is not necessary but the kernel will assume
754 that format when reading the "reg" properties of a CPU node, see
759 So under /cpus, you are supposed to create a node for every CPU on
760 the machine. There is no specific restriction on the name of the
761 CPU, though It's common practice to call it PowerPC,<name>. For
762 example, Apple uses PowerPC,G5 while IBM uses PowerPC,970FX.
766 - device_type : has to be "cpu"
767 - reg : This is the physical cpu number, it's a single 32 bit cell
768 and is also used as-is as the unit number for constructing the
769 unit name in the full path. For example, with 2 CPUs, you would
771 /cpus/PowerPC,970FX@0
772 /cpus/PowerPC,970FX@1
773 (unit addresses do not require leading zeroes)
774 - d-cache-line-size : one cell, L1 data cache line size in bytes
775 - i-cache-line-size : one cell, L1 instruction cache line size in
777 - d-cache-size : one cell, size of L1 data cache in bytes
778 - i-cache-size : one cell, size of L1 instruction cache in bytes
780 Recommended properties:
782 - timebase-frequency : a cell indicating the frequency of the
783 timebase in Hz. This is not directly used by the generic code,
784 but you are welcome to copy/paste the pSeries code for setting
785 the kernel timebase/decrementer calibration based on this
787 - clock-frequency : a cell indicating the CPU core clock frequency
788 in Hz. A new property will be defined for 64 bit values, but if
789 your frequency is < 4Ghz, one cell is enough. Here as well as
790 for the above, the common code doesn't use that property, but
791 you are welcome to re-use the pSeries or Maple one. A future
792 kernel version might provide a common function for this.
794 You are welcome to add any property you find relevant to your board,
795 like some information about the mechanism used to soft-reset the
796 CPUs. For example, Apple puts the GPIO number for CPU soft reset
797 lines in there as a "soft-reset" property since they start secondary
798 CPUs by soft-resetting them.
801 d) the /memory node(s)
803 To define the physical memory layout of your board, you should
804 create one or more memory node(s). You can either create a single
805 node with all memory ranges in its reg property, or you can create
806 several nodes, as you wish. The unit address (@ part) used for the
807 full path is the address of the first range of memory defined by a
808 given node. If you use a single memory node, this will typically be
813 - device_type : has to be "memory"
814 - reg : This property contains all the physical memory ranges of
815 your board. It's a list of addresses/sizes concatenated
816 together, with the number of cells of each defined by the
817 #address-cells and #size-cells of the root node. For example,
818 with both of these properties being 2 like in the example given
819 earlier, a 970 based machine with 6Gb of RAM could typically
820 have a "reg" property here that looks like:
822 00000000 00000000 00000000 80000000
823 00000001 00000000 00000001 00000000
825 That is a range starting at 0 of 0x80000000 bytes and a range
826 starting at 0x100000000 and of 0x100000000 bytes. You can see
827 that there is no memory covering the IO hole between 2Gb and
828 4Gb. Some vendors prefer splitting those ranges into smaller
829 segments, but the kernel doesn't care.
833 This node is a bit "special". Normally, that's where open firmware
834 puts some variable environment information, like the arguments, or
835 the default input/output devices.
837 This specification makes a few of these mandatory, but also defines
838 some linux-specific properties that would be normally constructed by
839 the prom_init() trampoline when booting with an OF client interface,
840 but that you have to provide yourself when using the flattened format.
842 Recommended properties:
844 - bootargs : This zero-terminated string is passed as the kernel
846 - linux,stdout-path : This is the full path to your standard
847 console device if any. Typically, if you have serial devices on
848 your board, you may want to put the full path to the one set as
849 the default console in the firmware here, for the kernel to pick
850 it up as its own default console. If you look at the function
851 set_preferred_console() in arch/ppc64/kernel/setup.c, you'll see
852 that the kernel tries to find out the default console and has
853 knowledge of various types like 8250 serial ports. You may want
854 to extend this function to add your own.
856 Note that u-boot creates and fills in the chosen node for platforms
859 (Note: a practice that is now obsolete was to include a property
860 under /chosen called interrupt-controller which had a phandle value
861 that pointed to the main interrupt controller)
863 f) the /soc<SOCname> node
865 This node is used to represent a system-on-a-chip (SOC) and must be
866 present if the processor is a SOC. The top-level soc node contains
867 information that is global to all devices on the SOC. The node name
868 should contain a unit address for the SOC, which is the base address
869 of the memory-mapped register set for the SOC. The name of an soc
870 node should start with "soc", and the remainder of the name should
871 represent the part number for the soc. For example, the MPC8540's
872 soc node would be called "soc8540".
876 - device_type : Should be "soc"
877 - ranges : Should be defined as specified in 1) to describe the
878 translation of SOC addresses for memory mapped SOC registers.
879 - bus-frequency: Contains the bus frequency for the SOC node.
880 Typically, the value of this field is filled in by the boot
884 Recommended properties:
886 - reg : This property defines the address and size of the
887 memory-mapped registers that are used for the SOC node itself.
888 It does not include the child device registers - these will be
889 defined inside each child node. The address specified in the
890 "reg" property should match the unit address of the SOC node.
891 - #address-cells : Address representation for "soc" devices. The
892 format of this field may vary depending on whether or not the
893 device registers are memory mapped. For memory mapped
894 registers, this field represents the number of cells needed to
895 represent the address of the registers. For SOCs that do not
896 use MMIO, a special address format should be defined that
897 contains enough cells to represent the required information.
898 See 1) above for more details on defining #address-cells.
899 - #size-cells : Size representation for "soc" devices
900 - #interrupt-cells : Defines the width of cells used to represent
901 interrupts. Typically this value is <2>, which includes a
902 32-bit number that represents the interrupt number, and a
903 32-bit number that represents the interrupt sense and level.
904 This field is only needed if the SOC contains an interrupt
907 The SOC node may contain child nodes for each SOC device that the
908 platform uses. Nodes should not be created for devices which exist
909 on the SOC but are not used by a particular platform. See chapter VI
910 for more information on how to specify devices that are part of an
913 Example SOC node for the MPC8540:
916 #address-cells = <1>;
918 #interrupt-cells = <2>;
920 ranges = <00000000 e0000000 00100000>
921 reg = <e0000000 00003000>;
927 IV - "dtc", the device tree compiler
928 ====================================
931 dtc source code can be found at
932 <http://ozlabs.org/~dgibson/dtc/dtc.tar.gz>
934 WARNING: This version is still in early development stage; the
935 resulting device-tree "blobs" have not yet been validated with the
936 kernel. The current generated bloc lacks a useful reserve map (it will
937 be fixed to generate an empty one, it's up to the bootloader to fill
938 it up) among others. The error handling needs work, bugs are lurking,
941 dtc basically takes a device-tree in a given format and outputs a
942 device-tree in another format. The currently supported formats are:
947 - "dtb": "blob" format, that is a flattened device-tree block
949 header all in a binary blob.
950 - "dts": "source" format. This is a text file containing a
951 "source" for a device-tree. The format is defined later in this
953 - "fs" format. This is a representation equivalent to the
954 output of /proc/device-tree, that is nodes are directories and
960 - "dtb": "blob" format
961 - "dts": "source" format
962 - "asm": assembly language file. This is a file that can be
963 sourced by gas to generate a device-tree "blob". That file can
964 then simply be added to your Makefile. Additionally, the
965 assembly file exports some symbols that can be used.
968 The syntax of the dtc tool is
970 dtc [-I <input-format>] [-O <output-format>]
971 [-o output-filename] [-V output_version] input_filename
974 The "output_version" defines what versio of the "blob" format will be
975 generated. Supported versions are 1,2,3 and 16. The default is
976 currently version 3 but that may change in the future to version 16.
978 Additionally, dtc performs various sanity checks on the tree, like the
979 uniqueness of linux, phandle properties, validity of strings, etc...
981 The format of the .dts "source" file is "C" like, supports C and C++
987 The above is the "device-tree" definition. It's the only statement
988 supported currently at the toplevel.
991 property1 = "string_value"; /* define a property containing a 0
995 property2 = <1234abcd>; /* define a property containing a
996 * numerical 32 bits value (hexadecimal)
999 property3 = <12345678 12345678 deadbeef>;
1000 /* define a property containing 3
1001 * numerical 32 bits values (cells) in
1004 property4 = [0a 0b 0c 0d de ea ad be ef];
1005 /* define a property whose content is
1006 * an arbitrary array of bytes
1009 childnode@addresss { /* define a child node named "childnode"
1010 * whose unit name is "childnode at
1014 childprop = "hello\n"; /* define a property "childprop" of
1015 * childnode (in this case, a string)
1020 Nodes can contain other nodes etc... thus defining the hierarchical
1021 structure of the tree.
1023 Strings support common escape sequences from C: "\n", "\t", "\r",
1024 "\(octal value)", "\x(hex value)".
1026 It is also suggested that you pipe your source file through cpp (gcc
1027 preprocessor) so you can use #include's, #define for constants, etc...
1029 Finally, various options are planned but not yet implemented, like
1030 automatic generation of phandles, labels (exported to the asm file so
1031 you can point to a property content and change it easily from whatever
1032 you link the device-tree with), label or path instead of numeric value
1033 in some cells to "point" to a node (replaced by a phandle at compile
1034 time), export of reserve map address to the asm file, ability to
1035 specify reserve map content at compile time, etc...
1037 We may provide a .h include file with common definitions of that
1038 proves useful for some properties (like building PCI properties or
1039 interrupt maps) though it may be better to add a notion of struct
1040 definitions to the compiler...
1043 V - Recommendations for a bootloader
1044 ====================================
1047 Here are some various ideas/recommendations that have been proposed
1048 while all this has been defined and implemented.
1050 - The bootloader may want to be able to use the device-tree itself
1051 and may want to manipulate it (to add/edit some properties,
1052 like physical memory size or kernel arguments). At this point, 2
1053 choices can be made. Either the bootloader works directly on the
1054 flattened format, or the bootloader has its own internal tree
1055 representation with pointers (similar to the kernel one) and
1056 re-flattens the tree when booting the kernel. The former is a bit
1057 more difficult to edit/modify, the later requires probably a bit
1058 more code to handle the tree structure. Note that the structure
1059 format has been designed so it's relatively easy to "insert"
1060 properties or nodes or delete them by just memmoving things
1061 around. It contains no internal offsets or pointers for this
1064 - An example of code for iterating nodes & retrieving properties
1065 directly from the flattened tree format can be found in the kernel
1066 file arch/ppc64/kernel/prom.c, look at scan_flat_dt() function,
1067 its usage in early_init_devtree(), and the corresponding various
1068 early_init_dt_scan_*() callbacks. That code can be re-used in a
1069 GPL bootloader, and as the author of that code, I would be happy
1070 to discuss possible free licencing to any vendor who wishes to
1071 integrate all or part of this code into a non-GPL bootloader.
1075 VI - System-on-a-chip devices and nodes
1076 =======================================
1078 Many companies are now starting to develop system-on-a-chip
1079 processors, where the processor core (cpu) and many peripheral devices
1080 exist on a single piece of silicon. For these SOCs, an SOC node
1081 should be used that defines child nodes for the devices that make
1082 up the SOC. While platforms are not required to use this model in
1083 order to boot the kernel, it is highly encouraged that all SOC
1084 implementations define as complete a flat-device-tree as possible to
1085 describe the devices on the SOC. This will allow for the
1086 genericization of much of the kernel code.
1089 1) Defining child nodes of an SOC
1090 ---------------------------------
1092 Each device that is part of an SOC may have its own node entry inside
1093 the SOC node. For each device that is included in the SOC, the unit
1094 address property represents the address offset for this device's
1095 memory-mapped registers in the parent's address space. The parent's
1096 address space is defined by the "ranges" property in the top-level soc
1097 node. The "reg" property for each node that exists directly under the
1098 SOC node should contain the address mapping from the child address space
1099 to the parent SOC address space and the size of the device's
1100 memory-mapped register file.
1102 For many devices that may exist inside an SOC, there are predefined
1103 specifications for the format of the device tree node. All SOC child
1104 nodes should follow these specifications, except where noted in this
1107 See appendix A for an example partial SOC node definition for the
1111 2) Representing devices without a current OF specification
1112 ----------------------------------------------------------
1114 Currently, there are many devices on SOCs that do not have a standard
1115 representation pre-defined as part of the open firmware
1116 specifications, mainly because the boards that contain these SOCs are
1117 not currently booted using open firmware. This section contains
1118 descriptions for the SOC devices for which new nodes have been
1119 defined; this list will expand as more and more SOC-containing
1120 platforms are moved over to use the flattened-device-tree model.
1124 The MDIO is a bus to which the PHY devices are connected. For each
1125 device that exists on this bus, a child node should be created. See
1126 the definition of the PHY node below for an example of how to define
1129 Required properties:
1130 - reg : Offset and length of the register set for the device
1131 - device_type : Should be "mdio"
1132 - compatible : Should define the compatible device type for the
1133 mdio. Currently, this is most likely to be "gianfar"
1139 device_type = "mdio";
1140 compatible = "gianfar";
1148 b) Gianfar-compatible ethernet nodes
1150 Required properties:
1152 - device_type : Should be "network"
1153 - model : Model of the device. Can be "TSEC", "eTSEC", or "FEC"
1154 - compatible : Should be "gianfar"
1155 - reg : Offset and length of the register set for the device
1156 - mac-address : List of bytes representing the ethernet address of
1158 - interrupts : <a b> where a is the interrupt number and b is a
1159 field that represents an encoding of the sense and level
1160 information for the interrupt. This should be encoded based on
1161 the information in section 2) depending on the type of interrupt
1162 controller you have.
1163 - interrupt-parent : the phandle for the interrupt controller that
1164 services interrupts for this device.
1165 - phy-handle : The phandle for the PHY connected to this ethernet
1172 device_type = "network";
1174 compatible = "gianfar";
1176 mac-address = [ 00 E0 0C 00 73 00 ];
1177 interrupts = <d 3 e 3 12 3>;
1178 interrupt-parent = <40000>;
1179 phy-handle = <2452000>
1186 Required properties:
1188 - device_type : Should be "ethernet-phy"
1189 - interrupts : <a b> where a is the interrupt number and b is a
1190 field that represents an encoding of the sense and level
1191 information for the interrupt. This should be encoded based on
1192 the information in section 2) depending on the type of interrupt
1193 controller you have.
1194 - interrupt-parent : the phandle for the interrupt controller that
1195 services interrupts for this device.
1196 - reg : The ID number for the phy, usually a small integer
1197 - linux,phandle : phandle for this node; likely referenced by an
1198 ethernet controller node.
1204 linux,phandle = <2452000>
1205 interrupt-parent = <40000>;
1206 interrupts = <35 1>;
1208 device_type = "ethernet-phy";
1212 d) Interrupt controllers
1214 Some SOC devices contain interrupt controllers that are different
1215 from the standard Open PIC specification. The SOC device nodes for
1216 these types of controllers should be specified just like a standard
1217 OpenPIC controller. Sense and level information should be encoded
1218 as specified in section 2) of this chapter for each device that
1219 specifies an interrupt.
1224 linux,phandle = <40000>;
1225 clock-frequency = <0>;
1226 interrupt-controller;
1227 #address-cells = <0>;
1228 reg = <40000 40000>;
1230 compatible = "chrp,open-pic";
1231 device_type = "open-pic";
1238 Required properties :
1240 - device_type : Should be "i2c"
1241 - reg : Offset and length of the register set for the device
1243 Recommended properties :
1245 - compatible : Should be "fsl-i2c" for parts compatible with
1246 Freescale I2C specifications.
1247 - interrupts : <a b> where a is the interrupt number and b is a
1248 field that represents an encoding of the sense and level
1249 information for the interrupt. This should be encoded based on
1250 the information in section 2) depending on the type of interrupt
1251 controller you have.
1252 - interrupt-parent : the phandle for the interrupt controller that
1253 services interrupts for this device.
1254 - dfsrr : boolean; if defined, indicates that this I2C device has
1255 a digital filter sampling rate register
1256 - fsl5200-clocking : boolean; if defined, indicated that this device
1257 uses the FSL 5200 clocking mechanism.
1262 interrupt-parent = <40000>;
1263 interrupts = <1b 3>;
1265 device_type = "i2c";
1266 compatible = "fsl-i2c";
1271 f) Freescale SOC USB controllers
1273 The device node for a USB controller that is part of a Freescale
1274 SOC is as described in the document "Open Firmware Recommended
1275 Practice : Universal Serial Bus" with the following modifications
1278 Required properties :
1279 - compatible : Should be "fsl-usb2-mph" for multi port host usb
1280 controllers, or "fsl-usb2-dr" for dual role usb controllers
1281 - phy_type : For multi port host usb controllers, should be one of
1282 "ulpi", or "serial". For dual role usb controllers, should be
1283 one of "ulpi", "utmi", "utmi_wide", or "serial".
1284 - reg : Offset and length of the register set for the device
1285 - port0 : boolean; if defined, indicates port0 is connected for
1286 fsl-usb2-mph compatible controllers. Either this property or
1287 "port1" (or both) must be defined for "fsl-usb2-mph" compatible
1289 - port1 : boolean; if defined, indicates port1 is connected for
1290 fsl-usb2-mph compatible controllers. Either this property or
1291 "port0" (or both) must be defined for "fsl-usb2-mph" compatible
1293 - dr_mode : indicates the working mode for "fsl-usb2-dr" compatible
1294 controllers. Can be "host", "peripheral", or "otg". Default to
1295 "host" if not defined for backward compatibility.
1297 Recommended properties :
1298 - interrupts : <a b> where a is the interrupt number and b is a
1299 field that represents an encoding of the sense and level
1300 information for the interrupt. This should be encoded based on
1301 the information in section 2) depending on the type of interrupt
1302 controller you have.
1303 - interrupt-parent : the phandle for the interrupt controller that
1304 services interrupts for this device.
1306 Example multi port host usb controller device node :
1308 device_type = "usb";
1309 compatible = "fsl-usb2-mph";
1311 #address-cells = <1>;
1313 interrupt-parent = <700>;
1314 interrupts = <27 1>;
1320 Example dual role usb controller device node :
1322 device_type = "usb";
1323 compatible = "fsl-usb2-dr";
1325 #address-cells = <1>;
1327 interrupt-parent = <700>;
1328 interrupts = <26 1>;
1334 g) Freescale SOC SEC Security Engines
1336 Required properties:
1338 - device_type : Should be "crypto"
1339 - model : Model of the device. Should be "SEC1" or "SEC2"
1340 - compatible : Should be "talitos"
1341 - reg : Offset and length of the register set for the device
1342 - interrupts : <a b> where a is the interrupt number and b is a
1343 field that represents an encoding of the sense and level
1344 information for the interrupt. This should be encoded based on
1345 the information in section 2) depending on the type of interrupt
1346 controller you have.
1347 - interrupt-parent : the phandle for the interrupt controller that
1348 services interrupts for this device.
1349 - num-channels : An integer representing the number of channels
1351 - channel-fifo-len : An integer representing the number of
1352 descriptor pointers each channel fetch fifo can hold.
1353 - exec-units-mask : The bitmask representing what execution units
1354 (EUs) are available. It's a single 32 bit cell. EU information
1355 should be encoded following the SEC's Descriptor Header Dword
1356 EU_SEL0 field documentation, i.e. as follows:
1358 bit 0 = reserved - should be 0
1359 bit 1 = set if SEC has the ARC4 EU (AFEU)
1360 bit 2 = set if SEC has the DES/3DES EU (DEU)
1361 bit 3 = set if SEC has the message digest EU (MDEU)
1362 bit 4 = set if SEC has the random number generator EU (RNG)
1363 bit 5 = set if SEC has the public key EU (PKEU)
1364 bit 6 = set if SEC has the AES EU (AESU)
1365 bit 7 = set if SEC has the Kasumi EU (KEU)
1367 bits 8 through 31 are reserved for future SEC EUs.
1369 - descriptor-types-mask : The bitmask representing what descriptors
1370 are available. It's a single 32 bit cell. Descriptor type
1371 information should be encoded following the SEC's Descriptor
1372 Header Dword DESC_TYPE field documentation, i.e. as follows:
1374 bit 0 = set if SEC supports the aesu_ctr_nonsnoop desc. type
1375 bit 1 = set if SEC supports the ipsec_esp descriptor type
1376 bit 2 = set if SEC supports the common_nonsnoop desc. type
1377 bit 3 = set if SEC supports the 802.11i AES ccmp desc. type
1378 bit 4 = set if SEC supports the hmac_snoop_no_afeu desc. type
1379 bit 5 = set if SEC supports the srtp descriptor type
1380 bit 6 = set if SEC supports the non_hmac_snoop_no_afeu desc.type
1381 bit 7 = set if SEC supports the pkeu_assemble descriptor type
1382 bit 8 = set if SEC supports the aesu_key_expand_output desc.type
1383 bit 9 = set if SEC supports the pkeu_ptmul descriptor type
1384 bit 10 = set if SEC supports the common_nonsnoop_afeu desc. type
1385 bit 11 = set if SEC supports the pkeu_ptadd_dbl descriptor type
1387 ..and so on and so forth.
1393 device_type = "crypto";
1395 compatible = "talitos";
1396 reg = <30000 10000>;
1397 interrupts = <1d 3>;
1398 interrupt-parent = <40000>;
1400 channel-fifo-len = <18>;
1401 exec-units-mask = <000000fe>;
1402 descriptor-types-mask = <012b0ebf>;
1405 h) Board Control and Status (BCSR)
1407 Required properties:
1409 - device_type : Should be "board-control"
1410 - reg : Offset and length of the register set for the device
1415 device_type = "board-control";
1416 reg = <f8000000 8000>;
1419 i) Freescale QUICC Engine module (QE)
1420 This represents qe module that is installed on PowerQUICC II Pro.
1421 Hopefully it will merge backward compatibility with CPM/CPM2.
1422 Basically, it is a bus of devices, that could act more or less
1423 as a complete entity (UCC, USB etc ). All of them should be siblings on
1424 the "root" qe node, using the common properties from there.
1425 The description below applies to the the qe of MPC8360 and
1426 more nodes and properties would be extended in the future.
1430 Required properties:
1431 - device_type : should be "qe";
1432 - model : precise model of the QE, Can be "QE", "CPM", or "CPM2"
1433 - reg : offset and length of the device registers.
1434 - bus-frequency : the clock frequency for QUICC Engine.
1436 Recommended properties
1437 - brg-frequency : the internal clock source frequency for baud-rate
1442 #address-cells = <1>;
1444 #interrupt-cells = <2>;
1447 ranges = <0 e0100000 00100000>;
1448 reg = <e0100000 480>;
1449 brg-frequency = <0>;
1450 bus-frequency = <179A7B00>;
1454 ii) SPI (Serial Peripheral Interface)
1456 Required properties:
1457 - device_type : should be "spi".
1458 - compatible : should be "fsl_spi".
1459 - mode : the spi operation mode, it can be "cpu" or "qe".
1460 - reg : Offset and length of the register set for the device
1461 - interrupts : <a b> where a is the interrupt number and b is a
1462 field that represents an encoding of the sense and level
1463 information for the interrupt. This should be encoded based on
1464 the information in section 2) depending on the type of interrupt
1465 controller you have.
1466 - interrupt-parent : the phandle for the interrupt controller that
1467 services interrupts for this device.
1471 device_type = "spi";
1472 compatible = "fsl_spi";
1474 interrupts = <82 0>;
1475 interrupt-parent = <700>;
1480 iii) USB (Universal Serial Bus Controller)
1482 Required properties:
1483 - device_type : should be "usb".
1484 - compatible : could be "qe_udc" or "fhci-hcd".
1485 - mode : the could be "host" or "slave".
1486 - reg : Offset and length of the register set for the device
1487 - interrupts : <a b> where a is the interrupt number and b is a
1488 field that represents an encoding of the sense and level
1489 information for the interrupt. This should be encoded based on
1490 the information in section 2) depending on the type of interrupt
1491 controller you have.
1492 - interrupt-parent : the phandle for the interrupt controller that
1493 services interrupts for this device.
1497 device_type = "usb";
1498 compatible = "qe_udc";
1500 interrupts = <8b 0>;
1501 interrupt-parent = <700>;
1506 iv) UCC (Unified Communications Controllers)
1508 Required properties:
1509 - device_type : should be "network", "hldc", "uart", "transparent"
1511 - compatible : could be "ucc_geth" or "fsl_atm" and so on.
1512 - model : should be "UCC".
1513 - device-id : the ucc number(1-8), corresponding to UCCx in UM.
1514 - reg : Offset and length of the register set for the device
1515 - interrupts : <a b> where a is the interrupt number and b is a
1516 field that represents an encoding of the sense and level
1517 information for the interrupt. This should be encoded based on
1518 the information in section 2) depending on the type of interrupt
1519 controller you have.
1520 - interrupt-parent : the phandle for the interrupt controller that
1521 services interrupts for this device.
1522 - pio-handle : The phandle for the Parallel I/O port configuration.
1523 - rx-clock : represents the UCC receive clock source.
1524 0x00 : clock source is disabled;
1525 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1526 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1527 - tx-clock: represents the UCC transmit clock source;
1528 0x00 : clock source is disabled;
1529 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1530 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1532 Required properties for network device_type:
1533 - mac-address : list of bytes representing the ethernet address.
1534 - phy-handle : The phandle for the PHY connected to this controller.
1538 device_type = "network";
1539 compatible = "ucc_geth";
1543 interrupts = <a0 0>;
1544 interrupt-parent = <700>;
1545 mac-address = [ 00 04 9f 00 23 23 ];
1548 phy-handle = <212000>;
1549 pio-handle = <140001>;
1553 v) Parallel I/O Ports
1555 This node configures Parallel I/O ports for CPUs with QE support.
1556 The node should reside in the "soc" node of the tree. For each
1557 device that using parallel I/O ports, a child node should be created.
1558 See the definition of the Pin configuration nodes below for more
1561 Required properties:
1562 - device_type : should be "par_io".
1563 - reg : offset to the register set and its length.
1564 - num-ports : number of Parallel I/O ports
1569 #address-cells = <1>;
1571 device_type = "par_io";
1578 vi) Pin configuration nodes
1580 Required properties:
1581 - linux,phandle : phandle of this node; likely referenced by a QE
1583 - pio-map : array of pin configurations. Each pin is defined by 6
1584 integers. The six numbers are respectively: port, pin, dir,
1585 open_drain, assignment, has_irq.
1586 - port : port number of the pin; 0-6 represent port A-G in UM.
1587 - pin : pin number in the port.
1588 - dir : direction of the pin, should encode as follows:
1590 0 = The pin is disabled
1591 1 = The pin is an output
1592 2 = The pin is an input
1595 - open_drain : indicates the pin is normal or wired-OR:
1597 0 = The pin is actively driven as an output
1598 1 = The pin is an open-drain driver. As an output, the pin is
1599 driven active-low, otherwise it is three-stated.
1601 - assignment : function number of the pin according to the Pin Assignment
1602 tables in User Manual. Each pin can have up to 4 possible functions in
1603 QE and two options for CPM.
1604 - has_irq : indicates if the pin is used as source of exteral
1609 linux,phandle = <140001>;
1611 /* port pin dir open_drain assignment has_irq */
1612 0 3 1 0 1 0 /* TxD0 */
1613 0 4 1 0 1 0 /* TxD1 */
1614 0 5 1 0 1 0 /* TxD2 */
1615 0 6 1 0 1 0 /* TxD3 */
1616 1 6 1 0 3 0 /* TxD4 */
1617 1 7 1 0 1 0 /* TxD5 */
1618 1 9 1 0 2 0 /* TxD6 */
1619 1 a 1 0 2 0 /* TxD7 */
1620 0 9 2 0 1 0 /* RxD0 */
1621 0 a 2 0 1 0 /* RxD1 */
1622 0 b 2 0 1 0 /* RxD2 */
1623 0 c 2 0 1 0 /* RxD3 */
1624 0 d 2 0 1 0 /* RxD4 */
1625 1 1 2 0 2 0 /* RxD5 */
1626 1 0 2 0 2 0 /* RxD6 */
1627 1 4 2 0 2 0 /* RxD7 */
1628 0 7 1 0 1 0 /* TX_EN */
1629 0 8 1 0 1 0 /* TX_ER */
1630 0 f 2 0 1 0 /* RX_DV */
1631 0 10 2 0 1 0 /* RX_ER */
1632 0 0 2 0 1 0 /* RX_CLK */
1633 2 9 1 0 3 0 /* GTX_CLK - CLK10 */
1634 2 8 2 0 1 0>; /* GTX125 - CLK9 */
1637 vii) Multi-User RAM (MURAM)
1639 Required properties:
1640 - device_type : should be "muram".
1641 - mode : the could be "host" or "slave".
1642 - ranges : Should be defined as specified in 1) to describe the
1643 translation of MURAM addresses.
1644 - data-only : sub-node which defines the address area under MURAM
1645 bus that can be allocated as data/parameter
1650 device_type = "muram";
1651 ranges = <0 00010000 0000c000>;
1660 Flash chips (Memory Technology Devices) are often used for solid state
1661 file systems on embedded devices.
1663 Required properties:
1665 - device_type : has to be "rom"
1666 - compatible : Should specify what this flash device is compatible with.
1667 Currently, this is most likely to be "direct-mapped" (which
1668 corresponds to the MTD physmap mapping driver).
1669 - reg : Offset and length of the register set (or memory mapping) for
1671 - bank-width : Width of the flash data bus in bytes. Required
1672 for the NOR flashes (compatible == "direct-mapped" and others) ONLY.
1674 Recommended properties :
1676 - partitions : Several pairs of 32-bit values where the first value is
1677 partition's offset from the start of the device and the second one is
1678 partition size in bytes with LSB used to signify a read only
1679 partition (so, the parition size should always be an even number).
1680 - partition-names : The list of concatenated zero terminated strings
1681 representing the partition names.
1682 - probe-type : The type of probe which should be done for the chip
1683 (JEDEC vs CFI actually). Valid ONLY for NOR flashes.
1688 device_type = "rom";
1689 compatible = "direct-mapped";
1691 reg = <ff000000 01000000>;
1693 partitions = <00000000 00f80000
1695 partition-names = "fs\0firmware";
1698 More devices will be defined as this spec matures.
1700 VII - Specifying interrupt information for devices
1701 ===================================================
1703 The device tree represents the busses and devices of a hardware
1704 system in a form similar to the physical bus topology of the
1707 In addition, a logical 'interrupt tree' exists which represents the
1708 hierarchy and routing of interrupts in the hardware.
1710 The interrupt tree model is fully described in the
1711 document "Open Firmware Recommended Practice: Interrupt
1712 Mapping Version 0.9". The document is available at:
1713 <http://playground.sun.com/1275/practice>.
1715 1) interrupts property
1716 ----------------------
1718 Devices that generate interrupts to a single interrupt controller
1719 should use the conventional OF representation described in the
1720 OF interrupt mapping documentation.
1722 Each device which generates interrupts must have an 'interrupt'
1723 property. The interrupt property value is an arbitrary number of
1724 of 'interrupt specifier' values which describe the interrupt or
1725 interrupts for the device.
1727 The encoding of an interrupt specifier is determined by the
1728 interrupt domain in which the device is located in the
1729 interrupt tree. The root of an interrupt domain specifies in
1730 its #interrupt-cells property the number of 32-bit cells
1731 required to encode an interrupt specifier. See the OF interrupt
1732 mapping documentation for a detailed description of domains.
1734 For example, the binding for the OpenPIC interrupt controller
1735 specifies an #interrupt-cells value of 2 to encode the interrupt
1736 number and level/sense information. All interrupt children in an
1737 OpenPIC interrupt domain use 2 cells per interrupt in their interrupts
1740 The PCI bus binding specifies a #interrupt-cell value of 1 to encode
1741 which interrupt pin (INTA,INTB,INTC,INTD) is used.
1743 2) interrupt-parent property
1744 ----------------------------
1746 The interrupt-parent property is specified to define an explicit
1747 link between a device node and its interrupt parent in
1748 the interrupt tree. The value of interrupt-parent is the
1749 phandle of the parent node.
1751 If the interrupt-parent property is not defined for a node, it's
1752 interrupt parent is assumed to be an ancestor in the node's
1753 _device tree_ hierarchy.
1755 3) OpenPIC Interrupt Controllers
1756 --------------------------------
1758 OpenPIC interrupt controllers require 2 cells to encode
1759 interrupt information. The first cell defines the interrupt
1760 number. The second cell defines the sense and level
1763 Sense and level information should be encoded as follows:
1765 0 = low to high edge sensitive type enabled
1766 1 = active low level sensitive type enabled
1767 2 = active high level sensitive type enabled
1768 3 = high to low edge sensitive type enabled
1770 4) ISA Interrupt Controllers
1771 ----------------------------
1773 ISA PIC interrupt controllers require 2 cells to encode
1774 interrupt information. The first cell defines the interrupt
1775 number. The second cell defines the sense and level
1778 ISA PIC interrupt controllers should adhere to the ISA PIC
1779 encodings listed below:
1781 0 = active low level sensitive type enabled
1782 1 = active high level sensitive type enabled
1783 2 = high to low edge sensitive type enabled
1784 3 = low to high edge sensitive type enabled
1787 Appendix A - Sample SOC node for MPC8540
1788 ========================================
1790 Note that the #address-cells and #size-cells for the SoC node
1791 in this example have been explicitly listed; these are likely
1792 not necessary as they are usually the same as the root node.
1795 #address-cells = <1>;
1797 #interrupt-cells = <2>;
1798 device_type = "soc";
1799 ranges = <00000000 e0000000 00100000>
1800 reg = <e0000000 00003000>;
1801 bus-frequency = <0>;
1805 device_type = "mdio";
1806 compatible = "gianfar";
1809 linux,phandle = <2452000>
1810 interrupt-parent = <40000>;
1811 interrupts = <35 1>;
1813 device_type = "ethernet-phy";
1817 linux,phandle = <2452001>
1818 interrupt-parent = <40000>;
1819 interrupts = <35 1>;
1821 device_type = "ethernet-phy";
1825 linux,phandle = <2452002>
1826 interrupt-parent = <40000>;
1827 interrupts = <35 1>;
1829 device_type = "ethernet-phy";
1836 device_type = "network";
1838 compatible = "gianfar";
1840 mac-address = [ 00 E0 0C 00 73 00 ];
1841 interrupts = <d 3 e 3 12 3>;
1842 interrupt-parent = <40000>;
1843 phy-handle = <2452000>;
1847 #address-cells = <1>;
1849 device_type = "network";
1851 compatible = "gianfar";
1853 mac-address = [ 00 E0 0C 00 73 01 ];
1854 interrupts = <13 3 14 3 18 3>;
1855 interrupt-parent = <40000>;
1856 phy-handle = <2452001>;
1860 #address-cells = <1>;
1862 device_type = "network";
1864 compatible = "gianfar";
1866 mac-address = [ 00 E0 0C 00 73 02 ];
1867 interrupts = <19 3>;
1868 interrupt-parent = <40000>;
1869 phy-handle = <2452002>;
1873 device_type = "serial";
1874 compatible = "ns16550";
1876 clock-frequency = <0>;
1877 interrupts = <1a 3>;
1878 interrupt-parent = <40000>;
1882 linux,phandle = <40000>;
1883 clock-frequency = <0>;
1884 interrupt-controller;
1885 #address-cells = <0>;
1886 reg = <40000 40000>;
1888 compatible = "chrp,open-pic";
1889 device_type = "open-pic";
1894 interrupt-parent = <40000>;
1895 interrupts = <1b 3>;
1897 device_type = "i2c";
1898 compatible = "fsl-i2c";