blob: b57e7da7897667b2e2702944cae029677219e7e6 [file] [log] [blame]
David Gibsonc125a182006-02-01 03:05:22 -08001 Booting the Linux/ppc kernel without Open Firmware
2 --------------------------------------------------
3
4
5(c) 2005 Benjamin Herrenschmidt <benh at kernel.crashing.org>,
6 IBM Corp.
7(c) 2005 Becky Bruce <becky.bruce at freescale.com>,
8 Freescale Semiconductor, FSL SOC and 32-bit additions
9
10 May 18, 2005: Rev 0.1 - Initial draft, no chapter III yet.
11
12 May 19, 2005: Rev 0.2 - Add chapter III and bits & pieces here or
13 clarifies the fact that a lot of things are
14 optional, the kernel only requires a very
15 small device tree, though it is encouraged
16 to provide an as complete one as possible.
17
18 May 24, 2005: Rev 0.3 - Precise that DT block has to be in RAM
19 - Misc fixes
20 - Define version 3 and new format version 16
21 for the DT block (version 16 needs kernel
22 patches, will be fwd separately).
23 String block now has a size, and full path
24 is replaced by unit name for more
25 compactness.
26 linux,phandle is made optional, only nodes
27 that are referenced by other nodes need it.
28 "name" property is now automatically
29 deduced from the unit name
30
31 June 1, 2005: Rev 0.4 - Correct confusion between OF_DT_END and
32 OF_DT_END_NODE in structure definition.
33 - Change version 16 format to always align
34 property data to 4 bytes. Since tokens are
35 already aligned, that means no specific
36 required alignement between property size
37 and property data. The old style variable
38 alignment would make it impossible to do
39 "simple" insertion of properties using
40 memove (thanks Milton for
41 noticing). Updated kernel patch as well
42 - Correct a few more alignement constraints
43 - Add a chapter about the device-tree
44 compiler and the textural representation of
45 the tree that can be "compiled" by dtc.
46
David Gibsonc125a182006-02-01 03:05:22 -080047 November 21, 2005: Rev 0.5
48 - Additions/generalizations for 32-bit
49 - Changed to reflect the new arch/powerpc
50 structure
51 - Added chapter VI
52
53
54 ToDo:
55 - Add some definitions of interrupt tree (simple/complex)
56 - Add some definitions for pci host bridges
57 - Add some common address format examples
58 - Add definitions for standard properties and "compatible"
59 names for cells that are not already defined by the existing
60 OF spec.
61 - Compare FSL SOC use of PCI to standard and make sure no new
62 node definition required.
63 - Add more information about node definitions for SOC devices
64 that currently have no standard, like the FSL CPM.
65
66
67I - Introduction
68================
69
70During the recent development of the Linux/ppc64 kernel, and more
71specifically, the addition of new platform types outside of the old
72IBM pSeries/iSeries pair, it was decided to enforce some strict rules
73regarding the kernel entry and bootloader <-> kernel interfaces, in
74order to avoid the degeneration that had become the ppc32 kernel entry
75point and the way a new platform should be added to the kernel. The
76legacy iSeries platform breaks those rules as it predates this scheme,
77but no new board support will be accepted in the main tree that
78doesn't follows them properly. In addition, since the advent of the
79arch/powerpc merged architecture for ppc32 and ppc64, new 32-bit
80platforms and 32-bit platforms which move into arch/powerpc will be
81required to use these rules as well.
82
83The main requirement that will be defined in more detail below is
84the presence of a device-tree whose format is defined after Open
85Firmware specification. However, in order to make life easier
86to embedded board vendors, the kernel doesn't require the device-tree
87to represent every device in the system and only requires some nodes
88and properties to be present. This will be described in detail in
89section III, but, for example, the kernel does not require you to
90create a node for every PCI device in the system. It is a requirement
91to have a node for PCI host bridges in order to provide interrupt
92routing informations and memory/IO ranges, among others. It is also
93recommended to define nodes for on chip devices and other busses that
94don't specifically fit in an existing OF specification. This creates a
95great flexibility in the way the kernel can then probe those and match
96drivers to device, without having to hard code all sorts of tables. It
97also makes it more flexible for board vendors to do minor hardware
98upgrades without significantly impacting the kernel code or cluttering
99it with special cases.
100
101
1021) Entry point for arch/powerpc
103-------------------------------
104
105 There is one and one single entry point to the kernel, at the start
106 of the kernel image. That entry point supports two calling
107 conventions:
108
109 a) Boot from Open Firmware. If your firmware is compatible
110 with Open Firmware (IEEE 1275) or provides an OF compatible
111 client interface API (support for "interpret" callback of
112 forth words isn't required), you can enter the kernel with:
113
114 r5 : OF callback pointer as defined by IEEE 1275
115 bindings to powerpc. Only the 32 bit client interface
116 is currently supported
117
118 r3, r4 : address & length of an initrd if any or 0
119
120 The MMU is either on or off; the kernel will run the
121 trampoline located in arch/powerpc/kernel/prom_init.c to
122 extract the device-tree and other information from open
123 firmware and build a flattened device-tree as described
124 in b). prom_init() will then re-enter the kernel using
125 the second method. This trampoline code runs in the
126 context of the firmware, which is supposed to handle all
127 exceptions during that time.
128
129 b) Direct entry with a flattened device-tree block. This entry
130 point is called by a) after the OF trampoline and can also be
131 called directly by a bootloader that does not support the Open
132 Firmware client interface. It is also used by "kexec" to
133 implement "hot" booting of a new kernel from a previous
134 running one. This method is what I will describe in more
135 details in this document, as method a) is simply standard Open
136 Firmware, and thus should be implemented according to the
137 various standard documents defining it and its binding to the
138 PowerPC platform. The entry point definition then becomes:
139
140 r3 : physical pointer to the device-tree block
141 (defined in chapter II) in RAM
142
143 r4 : physical pointer to the kernel itself. This is
144 used by the assembly code to properly disable the MMU
145 in case you are entering the kernel with MMU enabled
146 and a non-1:1 mapping.
147
148 r5 : NULL (as to differenciate with method a)
149
150 Note about SMP entry: Either your firmware puts your other
151 CPUs in some sleep loop or spin loop in ROM where you can get
152 them out via a soft reset or some other means, in which case
153 you don't need to care, or you'll have to enter the kernel
154 with all CPUs. The way to do that with method b) will be
155 described in a later revision of this document.
156
157
1582) Board support
159----------------
160
16164-bit kernels:
162
163 Board supports (platforms) are not exclusive config options. An
164 arbitrary set of board supports can be built in a single kernel
165 image. The kernel will "know" what set of functions to use for a
166 given platform based on the content of the device-tree. Thus, you
167 should:
168
169 a) add your platform support as a _boolean_ option in
170 arch/powerpc/Kconfig, following the example of PPC_PSERIES,
171 PPC_PMAC and PPC_MAPLE. The later is probably a good
172 example of a board support to start from.
173
174 b) create your main platform file as
175 "arch/powerpc/platforms/myplatform/myboard_setup.c" and add it
176 to the Makefile under the condition of your CONFIG_
177 option. This file will define a structure of type "ppc_md"
178 containing the various callbacks that the generic code will
179 use to get to your platform specific code
180
181 c) Add a reference to your "ppc_md" structure in the
182 "machines" table in arch/powerpc/kernel/setup_64.c if you are
183 a 64-bit platform.
184
185 d) request and get assigned a platform number (see PLATFORM_*
186 constants in include/asm-powerpc/processor.h
187
18832-bit embedded kernels:
189
190 Currently, board support is essentially an exclusive config option.
191 The kernel is configured for a single platform. Part of the reason
192 for this is to keep kernels on embedded systems small and efficient;
193 part of this is due to the fact the code is already that way. In the
194 future, a kernel may support multiple platforms, but only if the
195 platforms feature the same core architectire. A single kernel build
196 cannot support both configurations with Book E and configurations
197 with classic Powerpc architectures.
198
199 32-bit embedded platforms that are moved into arch/powerpc using a
200 flattened device tree should adopt the merged tree practice of
201 setting ppc_md up dynamically, even though the kernel is currently
202 built with support for only a single platform at a time. This allows
203 unification of the setup code, and will make it easier to go to a
204 multiple-platform-support model in the future.
205
206NOTE: I believe the above will be true once Ben's done with the merge
207of the boot sequences.... someone speak up if this is wrong!
208
209 To add a 32-bit embedded platform support, follow the instructions
210 for 64-bit platforms above, with the exception that the Kconfig
211 option should be set up such that the kernel builds exclusively for
212 the platform selected. The processor type for the platform should
213 enable another config option to select the specific board
214 supported.
215
216NOTE: If ben doesn't merge the setup files, may need to change this to
217point to setup_32.c
218
219
220 I will describe later the boot process and various callbacks that
221 your platform should implement.
222
223
224II - The DT block format
225========================
226
227
228This chapter defines the actual format of the flattened device-tree
229passed to the kernel. The actual content of it and kernel requirements
230are described later. You can find example of code manipulating that
231format in various places, including arch/powerpc/kernel/prom_init.c
232which will generate a flattened device-tree from the Open Firmware
233representation, or the fs2dt utility which is part of the kexec tools
234which will generate one from a filesystem representation. It is
235expected that a bootloader like uboot provides a bit more support,
236that will be discussed later as well.
237
238Note: The block has to be in main memory. It has to be accessible in
239both real mode and virtual mode with no mapping other than main
240memory. If you are writing a simple flash bootloader, it should copy
241the block to RAM before passing it to the kernel.
242
243
2441) Header
245---------
246
247 The kernel is entered with r3 pointing to an area of memory that is
248 roughtly described in include/asm-powerpc/prom.h by the structure
249 boot_param_header:
250
251struct boot_param_header {
252 u32 magic; /* magic word OF_DT_HEADER */
253 u32 totalsize; /* total size of DT block */
254 u32 off_dt_struct; /* offset to structure */
255 u32 off_dt_strings; /* offset to strings */
256 u32 off_mem_rsvmap; /* offset to memory reserve map
257*/
258 u32 version; /* format version */
259 u32 last_comp_version; /* last compatible version */
260
261 /* version 2 fields below */
262 u32 boot_cpuid_phys; /* Which physical CPU id we're
263 booting on */
264 /* version 3 fields below */
265 u32 size_dt_strings; /* size of the strings block */
266};
267
268 Along with the constants:
269
270/* Definitions used by the flattened device tree */
271#define OF_DT_HEADER 0xd00dfeed /* 4: version,
272 4: total size */
273#define OF_DT_BEGIN_NODE 0x1 /* Start node: full name
274*/
275#define OF_DT_END_NODE 0x2 /* End node */
276#define OF_DT_PROP 0x3 /* Property: name off,
277 size, content */
278#define OF_DT_END 0x9
279
280 All values in this header are in big endian format, the various
281 fields in this header are defined more precisely below. All
282 "offset" values are in bytes from the start of the header; that is
283 from the value of r3.
284
285 - magic
286
287 This is a magic value that "marks" the beginning of the
288 device-tree block header. It contains the value 0xd00dfeed and is
289 defined by the constant OF_DT_HEADER
290
291 - totalsize
292
293 This is the total size of the DT block including the header. The
294 "DT" block should enclose all data structures defined in this
295 chapter (who are pointed to by offsets in this header). That is,
296 the device-tree structure, strings, and the memory reserve map.
297
298 - off_dt_struct
299
300 This is an offset from the beginning of the header to the start
301 of the "structure" part the device tree. (see 2) device tree)
302
303 - off_dt_strings
304
305 This is an offset from the beginning of the header to the start
306 of the "strings" part of the device-tree
307
308 - off_mem_rsvmap
309
310 This is an offset from the beginning of the header to the start
311 of the reserved memory map. This map is a list of pairs of 64
312 bit integers. Each pair is a physical address and a size. The
313
314 list is terminated by an entry of size 0. This map provides the
315 kernel with a list of physical memory areas that are "reserved"
316 and thus not to be used for memory allocations, especially during
317 early initialization. The kernel needs to allocate memory during
318 boot for things like un-flattening the device-tree, allocating an
319 MMU hash table, etc... Those allocations must be done in such a
320 way to avoid overriding critical things like, on Open Firmware
321 capable machines, the RTAS instance, or on some pSeries, the TCE
322 tables used for the iommu. Typically, the reserve map should
323 contain _at least_ this DT block itself (header,total_size). If
324 you are passing an initrd to the kernel, you should reserve it as
325 well. You do not need to reserve the kernel image itself. The map
326 should be 64 bit aligned.
327
328 - version
329
330 This is the version of this structure. Version 1 stops
331 here. Version 2 adds an additional field boot_cpuid_phys.
332 Version 3 adds the size of the strings block, allowing the kernel
333 to reallocate it easily at boot and free up the unused flattened
334 structure after expansion. Version 16 introduces a new more
335 "compact" format for the tree itself that is however not backward
336 compatible. You should always generate a structure of the highest
337 version defined at the time of your implementation. Currently
338 that is version 16, unless you explicitely aim at being backward
339 compatible.
340
341 - last_comp_version
342
343 Last compatible version. This indicates down to what version of
344 the DT block you are backward compatible. For example, version 2
345 is backward compatible with version 1 (that is, a kernel build
346 for version 1 will be able to boot with a version 2 format). You
347 should put a 1 in this field if you generate a device tree of
348 version 1 to 3, or 0x10 if you generate a tree of version 0x10
349 using the new unit name format.
350
351 - boot_cpuid_phys
352
353 This field only exist on version 2 headers. It indicate which
354 physical CPU ID is calling the kernel entry point. This is used,
355 among others, by kexec. If you are on an SMP system, this value
356 should match the content of the "reg" property of the CPU node in
357 the device-tree corresponding to the CPU calling the kernel entry
358 point (see further chapters for more informations on the required
359 device-tree contents)
360
361
362 So the typical layout of a DT block (though the various parts don't
363 need to be in that order) looks like this (addresses go from top to
364 bottom):
365
366
367 ------------------------------
368 r3 -> | struct boot_param_header |
369 ------------------------------
370 | (alignment gap) (*) |
371 ------------------------------
372 | memory reserve map |
373 ------------------------------
374 | (alignment gap) |
375 ------------------------------
376 | |
377 | device-tree structure |
378 | |
379 ------------------------------
380 | (alignment gap) |
381 ------------------------------
382 | |
383 | device-tree strings |
384 | |
385 -----> ------------------------------
386 |
387 |
388 --- (r3 + totalsize)
389
390 (*) The alignment gaps are not necessarily present; their presence
391 and size are dependent on the various alignment requirements of
392 the individual data blocks.
393
394
3952) Device tree generalities
396---------------------------
397
398This device-tree itself is separated in two different blocks, a
399structure block and a strings block. Both need to be aligned to a 4
400byte boundary.
401
402First, let's quickly describe the device-tree concept before detailing
403the storage format. This chapter does _not_ describe the detail of the
404required types of nodes & properties for the kernel, this is done
405later in chapter III.
406
407The device-tree layout is strongly inherited from the definition of
408the Open Firmware IEEE 1275 device-tree. It's basically a tree of
409nodes, each node having two or more named properties. A property can
410have a value or not.
411
412It is a tree, so each node has one and only one parent except for the
413root node who has no parent.
414
415A node has 2 names. The actual node name is generally contained in a
416property of type "name" in the node property list whose value is a
417zero terminated string and is mandatory for version 1 to 3 of the
418format definition (as it is in Open Firmware). Version 0x10 makes it
419optional as it can generate it from the unit name defined below.
420
421There is also a "unit name" that is used to differenciate nodes with
422the same name at the same level, it is usually made of the node
423name's, the "@" sign, and a "unit address", which definition is
424specific to the bus type the node sits on.
425
426The unit name doesn't exist as a property per-se but is included in
427the device-tree structure. It is typically used to represent "path" in
428the device-tree. More details about the actual format of these will be
429below.
430
431The kernel powerpc generic code does not make any formal use of the
432unit address (though some board support code may do) so the only real
433requirement here for the unit address is to ensure uniqueness of
434the node unit name at a given level of the tree. Nodes with no notion
435of address and no possible sibling of the same name (like /memory or
436/cpus) may omit the unit address in the context of this specification,
437or use the "@0" default unit address. The unit name is used to define
438a node "full path", which is the concatenation of all parent node
439unit names separated with "/".
440
441The root node doesn't have a defined name, and isn't required to have
442a name property either if you are using version 3 or earlier of the
443format. It also has no unit address (no @ symbol followed by a unit
444address). The root node unit name is thus an empty string. The full
445path to the root node is "/".
446
447Every node which actually represents an actual device (that is, a node
448which isn't only a virtual "container" for more nodes, like "/cpus"
449is) is also required to have a "device_type" property indicating the
450type of node .
451
452Finally, every node that can be referenced from a property in another
453node is required to have a "linux,phandle" property. Real open
454firmware implementations provide a unique "phandle" value for every
455node that the "prom_init()" trampoline code turns into
456"linux,phandle" properties. However, this is made optional if the
457flattened device tree is used directly. An example of a node
458referencing another node via "phandle" is when laying out the
459interrupt tree which will be described in a further version of this
460document.
461
462This "linux, phandle" property is a 32 bit value that uniquely
463identifies a node. You are free to use whatever values or system of
464values, internal pointers, or whatever to generate these, the only
465requirement is that every node for which you provide that property has
466a unique value for it.
467
468Here is an example of a simple device-tree. In this example, an "o"
469designates a node followed by the node unit name. Properties are
470presented with their name followed by their content. "content"
471represents an ASCII string (zero terminated) value, while <content>
472represents a 32 bit hexadecimal value. The various nodes in this
473example will be discussed in a later chapter. At this point, it is
474only meant to give you a idea of what a device-tree looks like. I have
475purposefully kept the "name" and "linux,phandle" properties which
476aren't necessary in order to give you a better idea of what the tree
477looks like in practice.
478
479 / o device-tree
480 |- name = "device-tree"
481 |- model = "MyBoardName"
482 |- compatible = "MyBoardFamilyName"
483 |- #address-cells = <2>
484 |- #size-cells = <2>
485 |- linux,phandle = <0>
486 |
487 o cpus
488 | | - name = "cpus"
489 | | - linux,phandle = <1>
490 | | - #address-cells = <1>
491 | | - #size-cells = <0>
492 | |
493 | o PowerPC,970@0
494 | |- name = "PowerPC,970"
495 | |- device_type = "cpu"
496 | |- reg = <0>
497 | |- clock-frequency = <5f5e1000>
498 | |- linux,boot-cpu
499 | |- linux,phandle = <2>
500 |
501 o memory@0
502 | |- name = "memory"
503 | |- device_type = "memory"
504 | |- reg = <00000000 00000000 00000000 20000000>
505 | |- linux,phandle = <3>
506 |
507 o chosen
508 |- name = "chosen"
509 |- bootargs = "root=/dev/sda2"
510 |- linux,platform = <00000600>
511 |- linux,phandle = <4>
512
513This tree is almost a minimal tree. It pretty much contains the
514minimal set of required nodes and properties to boot a linux kernel;
515that is, some basic model informations at the root, the CPUs, and the
516physical memory layout. It also includes misc information passed
517through /chosen, like in this example, the platform type (mandatory)
518and the kernel command line arguments (optional).
519
520The /cpus/PowerPC,970@0/linux,boot-cpu property is an example of a
521property without a value. All other properties have a value. The
522significance of the #address-cells and #size-cells properties will be
523explained in chapter IV which defines precisely the required nodes and
524properties and their content.
525
526
5273) Device tree "structure" block
528
529The structure of the device tree is a linearized tree structure. The
530"OF_DT_BEGIN_NODE" token starts a new node, and the "OF_DT_END_NODE"
531ends that node definition. Child nodes are simply defined before
532"OF_DT_END_NODE" (that is nodes within the node). A 'token' is a 32
533bit value. The tree has to be "finished" with a OF_DT_END token
534
535Here's the basic structure of a single node:
536
537 * token OF_DT_BEGIN_NODE (that is 0x00000001)
538 * for version 1 to 3, this is the node full path as a zero
539 terminated string, starting with "/". For version 16 and later,
540 this is the node unit name only (or an empty string for the
541 root node)
542 * [align gap to next 4 bytes boundary]
543 * for each property:
544 * token OF_DT_PROP (that is 0x00000003)
545 * 32 bit value of property value size in bytes (or 0 of no
546 * value)
547 * 32 bit value of offset in string block of property name
548 * property value data if any
549 * [align gap to next 4 bytes boundary]
550 * [child nodes if any]
551 * token OF_DT_END_NODE (that is 0x00000002)
552
553So the node content can be summmarised as a start token, a full path,
554a list of properties, a list of child node and an end token. Every
555child node is a full node structure itself as defined above.
556
5574) Device tree 'strings" block
558
559In order to save space, property names, which are generally redundant,
560are stored separately in the "strings" block. This block is simply the
561whole bunch of zero terminated strings for all property names
562concatenated together. The device-tree property definitions in the
563structure block will contain offset values from the beginning of the
564strings block.
565
566
567III - Required content of the device tree
568=========================================
569
570WARNING: All "linux,*" properties defined in this document apply only
571to a flattened device-tree. If your platform uses a real
572implementation of Open Firmware or an implementation compatible with
573the Open Firmware client interface, those properties will be created
574by the trampoline code in the kernel's prom_init() file. For example,
575that's where you'll have to add code to detect your board model and
576set the platform number. However, when using the flatenned device-tree
577entry point, there is no prom_init() pass, and thus you have to
578provide those properties yourself.
579
580
5811) Note about cells and address representation
582----------------------------------------------
583
584The general rule is documented in the various Open Firmware
585documentations. If you chose to describe a bus with the device-tree
586and there exist an OF bus binding, then you should follow the
587specification. However, the kernel does not require every single
588device or bus to be described by the device tree.
589
590In general, the format of an address for a device is defined by the
591parent bus type, based on the #address-cells and #size-cells
592property. In the absence of such a property, the parent's parent
593values are used, etc... The kernel requires the root node to have
594those properties defining addresses format for devices directly mapped
595on the processor bus.
596
597Those 2 properties define 'cells' for representing an address and a
598size. A "cell" is a 32 bit number. For example, if both contain 2
599like the example tree given above, then an address and a size are both
600composed of 2 cells, and each is a 64 bit number (cells are
601concatenated and expected to be in big endian format). Another example
602is the way Apple firmware defines them, with 2 cells for an address
603and one cell for a size. Most 32-bit implementations should define
604#address-cells and #size-cells to 1, which represents a 32-bit value.
605Some 32-bit processors allow for physical addresses greater than 32
606bits; these processors should define #address-cells as 2.
607
608"reg" properties are always a tuple of the type "address size" where
609the number of cells of address and size is specified by the bus
610#address-cells and #size-cells. When a bus supports various address
611spaces and other flags relative to a given address allocation (like
612prefetchable, etc...) those flags are usually added to the top level
613bits of the physical address. For example, a PCI physical address is
614made of 3 cells, the bottom two containing the actual address itself
615while the top cell contains address space indication, flags, and pci
616bus & device numbers.
617
618For busses that support dynamic allocation, it's the accepted practice
619to then not provide the address in "reg" (keep it 0) though while
620providing a flag indicating the address is dynamically allocated, and
621then, to provide a separate "assigned-addresses" property that
622contains the fully allocated addresses. See the PCI OF bindings for
623details.
624
625In general, a simple bus with no address space bits and no dynamic
626allocation is preferred if it reflects your hardware, as the existing
627kernel address parsing functions will work out of the box. If you
628define a bus type with a more complex address format, including things
629like address space bits, you'll have to add a bus translator to the
630prom_parse.c file of the recent kernels for your bus type.
631
632The "reg" property only defines addresses and sizes (if #size-cells
633is
634non-0) within a given bus. In order to translate addresses upward
635(that is into parent bus addresses, and possibly into cpu physical
636addresses), all busses must contain a "ranges" property. If the
637"ranges" property is missing at a given level, it's assumed that
638translation isn't possible. The format of the "ranges" proprety for a
639bus is a list of:
640
641 bus address, parent bus address, size
642
643"bus address" is in the format of the bus this bus node is defining,
644that is, for a PCI bridge, it would be a PCI address. Thus, (bus
645address, size) defines a range of addresses for child devices. "parent
646bus address" is in the format of the parent bus of this bus. For
647example, for a PCI host controller, that would be a CPU address. For a
648PCI<->ISA bridge, that would be a PCI address. It defines the base
649address in the parent bus where the beginning of that range is mapped.
650
651For a new 64 bit powerpc board, I recommend either the 2/2 format or
652Apple's 2/1 format which is slightly more compact since sizes usually
653fit in a single 32 bit word. New 32 bit powerpc boards should use a
6541/1 format, unless the processor supports physical addresses greater
655than 32-bits, in which case a 2/1 format is recommended.
656
657
6582) Note about "compatible" properties
659-------------------------------------
660
661These properties are optional, but recommended in devices and the root
662node. The format of a "compatible" property is a list of concatenated
663zero terminated strings. They allow a device to express its
664compatibility with a family of similar devices, in some cases,
665allowing a single driver to match against several devices regardless
666of their actual names.
667
6683) Note about "name" properties
669-------------------------------
670
671While earlier users of Open Firmware like OldWorld macintoshes tended
672to use the actual device name for the "name" property, it's nowadays
673considered a good practice to use a name that is closer to the device
674class (often equal to device_type). For example, nowadays, ethernet
675controllers are named "ethernet", an additional "model" property
676defining precisely the chip type/model, and "compatible" property
677defining the family in case a single driver can driver more than one
678of these chips. However, the kernel doesn't generally put any
679restriction on the "name" property; it is simply considered good
680practice to follow the standard and its evolutions as closely as
681possible.
682
683Note also that the new format version 16 makes the "name" property
684optional. If it's absent for a node, then the node's unit name is then
685used to reconstruct the name. That is, the part of the unit name
686before the "@" sign is used (or the entire unit name if no "@" sign
687is present).
688
6894) Note about node and property names and character set
690-------------------------------------------------------
691
692While open firmware provides more flexibe usage of 8859-1, this
693specification enforces more strict rules. Nodes and properties should
694be comprised only of ASCII characters 'a' to 'z', '0' to
695'9', ',', '.', '_', '+', '#', '?', and '-'. Node names additionally
696allow uppercase characters 'A' to 'Z' (property names should be
697lowercase. The fact that vendors like Apple don't respect this rule is
698irrelevant here). Additionally, node and property names should always
699begin with a character in the range 'a' to 'z' (or 'A' to 'Z' for node
700names).
701
702The maximum number of characters for both nodes and property names
703is 31. In the case of node names, this is only the leftmost part of
704a unit name (the pure "name" property), it doesn't include the unit
705address which can extend beyond that limit.
706
707
7085) 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.
714
715 a) The root node
716
717 The root node requires some properties to be present:
718
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
Benjamin Herrenschmidte8222502006-03-28 23:15:54 +1100722 - 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.
David Gibsonc125a182006-02-01 03:05:22 -0800727
728 Additionally, some recommended properties are:
729
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 (see /chosen/linux,platform for how the kernel choses a
736 platform type) but it is generally useful.
737
738 The root node is also generally where you add additional properties
739 specific to your board like the serial number if any, that sort of
740 thing. it is recommended that if you add any "custom" property whose
741 name may clash with standard defined ones, you prefix them with your
742 vendor name and a comma.
743
744 b) The /cpus node
745
746 This node is the parent of all individual CPU nodes. It doesn't
747 have any specific requirements, though it's generally good practice
748 to have at least:
749
750 #address-cells = <00000001>
751 #size-cells = <00000000>
752
753 This defines that the "address" for a CPU is a single cell, and has
754 no meaningful size. This is not necessary but the kernel will assume
755 that format when reading the "reg" properties of a CPU node, see
756 below
757
758 c) The /cpus/* nodes
759
760 So under /cpus, you are supposed to create a node for every CPU on
761 the machine. There is no specific restriction on the name of the
762 CPU, though It's common practice to call it PowerPC,<name>. For
763 example, Apple uses PowerPC,G5 while IBM uses PowerPC,970FX.
764
765 Required properties:
766
767 - device_type : has to be "cpu"
768 - reg : This is the physical cpu number, it's a single 32 bit cell
769 and is also used as-is as the unit number for constructing the
770 unit name in the full path. For example, with 2 CPUs, you would
771 have the full path:
772 /cpus/PowerPC,970FX@0
773 /cpus/PowerPC,970FX@1
774 (unit addresses do not require leading zeroes)
775 - d-cache-line-size : one cell, L1 data cache line size in bytes
776 - i-cache-line-size : one cell, L1 instruction cache line size in
777 bytes
778 - d-cache-size : one cell, size of L1 data cache in bytes
779 - i-cache-size : one cell, size of L1 instruction cache in bytes
780 - linux, boot-cpu : Should be defined if this cpu is the boot cpu.
781
782 Recommended properties:
783
784 - timebase-frequency : a cell indicating the frequency of the
785 timebase in Hz. This is not directly used by the generic code,
786 but you are welcome to copy/paste the pSeries code for setting
787 the kernel timebase/decrementer calibration based on this
788 value.
789 - clock-frequency : a cell indicating the CPU core clock frequency
790 in Hz. A new property will be defined for 64 bit values, but if
791 your frequency is < 4Ghz, one cell is enough. Here as well as
792 for the above, the common code doesn't use that property, but
793 you are welcome to re-use the pSeries or Maple one. A future
794 kernel version might provide a common function for this.
795
796 You are welcome to add any property you find relevant to your board,
797 like some information about the mechanism used to soft-reset the
798 CPUs. For example, Apple puts the GPIO number for CPU soft reset
799 lines in there as a "soft-reset" property since they start secondary
800 CPUs by soft-resetting them.
801
802
803 d) the /memory node(s)
804
805 To define the physical memory layout of your board, you should
806 create one or more memory node(s). You can either create a single
807 node with all memory ranges in its reg property, or you can create
808 several nodes, as you wish. The unit address (@ part) used for the
809 full path is the address of the first range of memory defined by a
810 given node. If you use a single memory node, this will typically be
811 @0.
812
813 Required properties:
814
815 - device_type : has to be "memory"
816 - reg : This property contains all the physical memory ranges of
817 your board. It's a list of addresses/sizes concatenated
818 together, with the number of cells of each defined by the
819 #address-cells and #size-cells of the root node. For example,
820 with both of these properties beeing 2 like in the example given
821 earlier, a 970 based machine with 6Gb of RAM could typically
822 have a "reg" property here that looks like:
823
824 00000000 00000000 00000000 80000000
825 00000001 00000000 00000001 00000000
826
827 That is a range starting at 0 of 0x80000000 bytes and a range
828 starting at 0x100000000 and of 0x100000000 bytes. You can see
829 that there is no memory covering the IO hole between 2Gb and
830 4Gb. Some vendors prefer splitting those ranges into smaller
831 segments, but the kernel doesn't care.
832
833 e) The /chosen node
834
835 This node is a bit "special". Normally, that's where open firmware
836 puts some variable environment information, like the arguments, or
837 phandle pointers to nodes like the main interrupt controller, or the
838 default input/output devices.
839
840 This specification makes a few of these mandatory, but also defines
841 some linux-specific properties that would be normally constructed by
842 the prom_init() trampoline when booting with an OF client interface,
843 but that you have to provide yourself when using the flattened format.
844
845 Required properties:
846
847 - linux,platform : This is your platform number as assigned by the
848 architecture maintainers
849
850 Recommended properties:
851
852 - bootargs : This zero-terminated string is passed as the kernel
853 command line
854 - linux,stdout-path : This is the full path to your standard
855 console device if any. Typically, if you have serial devices on
856 your board, you may want to put the full path to the one set as
857 the default console in the firmware here, for the kernel to pick
858 it up as it's own default console. If you look at the funciton
859 set_preferred_console() in arch/ppc64/kernel/setup.c, you'll see
860 that the kernel tries to find out the default console and has
861 knowledge of various types like 8250 serial ports. You may want
862 to extend this function to add your own.
863 - interrupt-controller : This is one cell containing a phandle
864 value that matches the "linux,phandle" property of your main
865 interrupt controller node. May be used for interrupt routing.
866
867
868 Note that u-boot creates and fills in the chosen node for platforms
869 that use it.
870
871 f) the /soc<SOCname> node
872
873 This node is used to represent a system-on-a-chip (SOC) and must be
874 present if the processor is a SOC. The top-level soc node contains
875 information that is global to all devices on the SOC. The node name
876 should contain a unit address for the SOC, which is the base address
877 of the memory-mapped register set for the SOC. The name of an soc
878 node should start with "soc", and the remainder of the name should
879 represent the part number for the soc. For example, the MPC8540's
880 soc node would be called "soc8540".
881
882 Required properties:
883
884 - device_type : Should be "soc"
885 - ranges : Should be defined as specified in 1) to describe the
886 translation of SOC addresses for memory mapped SOC registers.
Becky Bruce7d4b95a2006-02-06 14:26:31 -0600887 - bus-frequency: Contains the bus frequency for the SOC node.
888 Typically, the value of this field is filled in by the boot
889 loader.
890
David Gibsonc125a182006-02-01 03:05:22 -0800891
892 Recommended properties:
893
894 - reg : This property defines the address and size of the
895 memory-mapped registers that are used for the SOC node itself.
896 It does not include the child device registers - these will be
897 defined inside each child node. The address specified in the
898 "reg" property should match the unit address of the SOC node.
899 - #address-cells : Address representation for "soc" devices. The
900 format of this field may vary depending on whether or not the
901 device registers are memory mapped. For memory mapped
902 registers, this field represents the number of cells needed to
903 represent the address of the registers. For SOCs that do not
904 use MMIO, a special address format should be defined that
905 contains enough cells to represent the required information.
906 See 1) above for more details on defining #address-cells.
907 - #size-cells : Size representation for "soc" devices
908 - #interrupt-cells : Defines the width of cells used to represent
909 interrupts. Typically this value is <2>, which includes a
910 32-bit number that represents the interrupt number, and a
911 32-bit number that represents the interrupt sense and level.
912 This field is only needed if the SOC contains an interrupt
913 controller.
914
915 The SOC node may contain child nodes for each SOC device that the
916 platform uses. Nodes should not be created for devices which exist
917 on the SOC but are not used by a particular platform. See chapter VI
918 for more information on how to specify devices that are part of an
919SOC.
920
921 Example SOC node for the MPC8540:
922
923 soc8540@e0000000 {
924 #address-cells = <1>;
925 #size-cells = <1>;
926 #interrupt-cells = <2>;
927 device_type = "soc";
928 ranges = <00000000 e0000000 00100000>
929 reg = <e0000000 00003000>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -0600930 bus-frequency = <0>;
David Gibsonc125a182006-02-01 03:05:22 -0800931 }
932
933
934
935IV - "dtc", the device tree compiler
936====================================
937
938
939dtc source code can be found at
940<http://ozlabs.org/~dgibson/dtc/dtc.tar.gz>
941
942WARNING: This version is still in early development stage; the
943resulting device-tree "blobs" have not yet been validated with the
944kernel. The current generated bloc lacks a useful reserve map (it will
945be fixed to generate an empty one, it's up to the bootloader to fill
946it up) among others. The error handling needs work, bugs are lurking,
947etc...
948
949dtc basically takes a device-tree in a given format and outputs a
950device-tree in another format. The currently supported formats are:
951
952 Input formats:
953 -------------
954
955 - "dtb": "blob" format, that is a flattened device-tree block
956 with
957 header all in a binary blob.
958 - "dts": "source" format. This is a text file containing a
959 "source" for a device-tree. The format is defined later in this
960 chapter.
961 - "fs" format. This is a representation equivalent to the
962 output of /proc/device-tree, that is nodes are directories and
963 properties are files
964
965 Output formats:
966 ---------------
967
968 - "dtb": "blob" format
969 - "dts": "source" format
970 - "asm": assembly language file. This is a file that can be
971 sourced by gas to generate a device-tree "blob". That file can
972 then simply be added to your Makefile. Additionally, the
973 assembly file exports some symbols that can be use
974
975
976The syntax of the dtc tool is
977
978 dtc [-I <input-format>] [-O <output-format>]
979 [-o output-filename] [-V output_version] input_filename
980
981
982The "output_version" defines what versio of the "blob" format will be
983generated. Supported versions are 1,2,3 and 16. The default is
984currently version 3 but that may change in the future to version 16.
985
986Additionally, dtc performs various sanity checks on the tree, like the
987uniqueness of linux,phandle properties, validity of strings, etc...
988
989The format of the .dts "source" file is "C" like, supports C and C++
990style commments.
991
992/ {
993}
994
995The above is the "device-tree" definition. It's the only statement
996supported currently at the toplevel.
997
998/ {
999 property1 = "string_value"; /* define a property containing a 0
1000 * terminated string
1001 */
1002
1003 property2 = <1234abcd>; /* define a property containing a
1004 * numerical 32 bits value (hexadecimal)
1005 */
1006
1007 property3 = <12345678 12345678 deadbeef>;
1008 /* define a property containing 3
1009 * numerical 32 bits values (cells) in
1010 * hexadecimal
1011 */
1012 property4 = [0a 0b 0c 0d de ea ad be ef];
1013 /* define a property whose content is
1014 * an arbitrary array of bytes
1015 */
1016
1017 childnode@addresss { /* define a child node named "childnode"
1018 * whose unit name is "childnode at
1019 * address"
1020 */
1021
1022 childprop = "hello\n"; /* define a property "childprop" of
1023 * childnode (in this case, a string)
1024 */
1025 };
1026};
1027
1028Nodes can contain other nodes etc... thus defining the hierarchical
1029structure of the tree.
1030
1031Strings support common escape sequences from C: "\n", "\t", "\r",
1032"\(octal value)", "\x(hex value)".
1033
1034It is also suggested that you pipe your source file through cpp (gcc
1035preprocessor) so you can use #include's, #define for constants, etc...
1036
1037Finally, various options are planned but not yet implemented, like
1038automatic generation of phandles, labels (exported to the asm file so
1039you can point to a property content and change it easily from whatever
1040you link the device-tree with), label or path instead of numeric value
1041in some cells to "point" to a node (replaced by a phandle at compile
1042time), export of reserve map address to the asm file, ability to
1043specify reserve map content at compile time, etc...
1044
1045We may provide a .h include file with common definitions of that
1046proves useful for some properties (like building PCI properties or
1047interrupt maps) though it may be better to add a notion of struct
1048definitions to the compiler...
1049
1050
1051V - Recommendations for a bootloader
1052====================================
1053
1054
1055Here are some various ideas/recommendations that have been proposed
1056while all this has been defined and implemented.
1057
1058 - The bootloader may want to be able to use the device-tree itself
1059 and may want to manipulate it (to add/edit some properties,
1060 like physical memory size or kernel arguments). At this point, 2
1061 choices can be made. Either the bootloader works directly on the
1062 flattened format, or the bootloader has its own internal tree
1063 representation with pointers (similar to the kernel one) and
1064 re-flattens the tree when booting the kernel. The former is a bit
1065 more difficult to edit/modify, the later requires probably a bit
1066 more code to handle the tree structure. Note that the structure
1067 format has been designed so it's relatively easy to "insert"
1068 properties or nodes or delete them by just memmoving things
1069 around. It contains no internal offsets or pointers for this
1070 purpose.
1071
1072 - An example of code for iterating nodes & retreiving properties
1073 directly from the flattened tree format can be found in the kernel
1074 file arch/ppc64/kernel/prom.c, look at scan_flat_dt() function,
1075 it's usage in early_init_devtree(), and the corresponding various
1076 early_init_dt_scan_*() callbacks. That code can be re-used in a
1077 GPL bootloader, and as the author of that code, I would be happy
1078 do discuss possible free licencing to any vendor who wishes to
1079 integrate all or part of this code into a non-GPL bootloader.
1080
1081
1082
1083VI - System-on-a-chip devices and nodes
1084=======================================
1085
1086Many companies are now starting to develop system-on-a-chip
1087processors, where the processor core (cpu) and many peripheral devices
1088exist on a single piece of silicon. For these SOCs, an SOC node
1089should be used that defines child nodes for the devices that make
1090up the SOC. While platforms are not required to use this model in
1091order to boot the kernel, it is highly encouraged that all SOC
1092implementations define as complete a flat-device-tree as possible to
1093describe the devices on the SOC. This will allow for the
1094genericization of much of the kernel code.
1095
1096
10971) Defining child nodes of an SOC
1098---------------------------------
1099
1100Each device that is part of an SOC may have its own node entry inside
1101the SOC node. For each device that is included in the SOC, the unit
1102address property represents the address offset for this device's
1103memory-mapped registers in the parent's address space. The parent's
1104address space is defined by the "ranges" property in the top-level soc
1105node. The "reg" property for each node that exists directly under the
1106SOC node should contain the address mapping from the child address space
1107to the parent SOC address space and the size of the device's
1108memory-mapped register file.
1109
1110For many devices that may exist inside an SOC, there are predefined
1111specifications for the format of the device tree node. All SOC child
1112nodes should follow these specifications, except where noted in this
1113document.
1114
1115See appendix A for an example partial SOC node definition for the
1116MPC8540.
1117
1118
11192) Specifying interrupt information for SOC devices
1120---------------------------------------------------
1121
1122Each device that is part of an SOC and which generates interrupts
1123should have the following properties:
1124
1125 - interrupt-parent : contains the phandle of the interrupt
1126 controller which handles interrupts for this device
1127 - interrupts : a list of tuples representing the interrupt
1128 number and the interrupt sense and level for each interupt
1129 for this device.
1130
1131This information is used by the kernel to build the interrupt table
1132for the interrupt controllers in the system.
1133
1134Sense and level information should be encoded as follows:
1135
1136 Devices connected to openPIC-compatible controllers should encode
1137 sense and polarity as follows:
1138
Benjamin Herrenschmidt3efbdd12006-08-30 08:58:00 +10001139 0 = low to high edge sensitive type enabled
David Gibsonc125a182006-02-01 03:05:22 -08001140 1 = active low level sensitive type enabled
Benjamin Herrenschmidt3efbdd12006-08-30 08:58:00 +10001141 2 = active high level sensitive type enabled
1142 3 = high to low edge sensitive type enabled
David Gibsonc125a182006-02-01 03:05:22 -08001143
1144 ISA PIC interrupt controllers should adhere to the ISA PIC
1145 encodings listed below:
1146
1147 0 = active low level sensitive type enabled
1148 1 = active high level sensitive type enabled
1149 2 = high to low edge sensitive type enabled
1150 3 = low to high edge sensitive type enabled
1151
1152
1153
11543) Representing devices without a current OF specification
1155----------------------------------------------------------
1156
1157Currently, there are many devices on SOCs that do not have a standard
1158representation pre-defined as part of the open firmware
1159specifications, mainly because the boards that contain these SOCs are
1160not currently booted using open firmware. This section contains
1161descriptions for the SOC devices for which new nodes have been
1162defined; this list will expand as more and more SOC-containing
1163platforms are moved over to use the flattened-device-tree model.
1164
1165 a) MDIO IO device
1166
1167 The MDIO is a bus to which the PHY devices are connected. For each
1168 device that exists on this bus, a child node should be created. See
1169 the definition of the PHY node below for an example of how to define
1170 a PHY.
1171
1172 Required properties:
1173 - reg : Offset and length of the register set for the device
1174 - device_type : Should be "mdio"
1175 - compatible : Should define the compatible device type for the
1176 mdio. Currently, this is most likely to be "gianfar"
1177
1178 Example:
1179
1180 mdio@24520 {
1181 reg = <24520 20>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -06001182 device_type = "mdio";
1183 compatible = "gianfar";
David Gibsonc125a182006-02-01 03:05:22 -08001184
1185 ethernet-phy@0 {
1186 ......
1187 };
1188 };
1189
1190
1191 b) Gianfar-compatible ethernet nodes
1192
1193 Required properties:
1194
1195 - device_type : Should be "network"
1196 - model : Model of the device. Can be "TSEC", "eTSEC", or "FEC"
1197 - compatible : Should be "gianfar"
1198 - reg : Offset and length of the register set for the device
Jon Loeligerf5831652006-08-17 08:42:35 -05001199 - mac-address : List of bytes representing the ethernet address of
David Gibsonc125a182006-02-01 03:05:22 -08001200 this controller
1201 - interrupts : <a b> where a is the interrupt number and b is a
1202 field that represents an encoding of the sense and level
1203 information for the interrupt. This should be encoded based on
1204 the information in section 2) depending on the type of interrupt
1205 controller you have.
1206 - interrupt-parent : the phandle for the interrupt controller that
1207 services interrupts for this device.
1208 - phy-handle : The phandle for the PHY connected to this ethernet
1209 controller.
1210
1211 Example:
1212
1213 ethernet@24000 {
1214 #size-cells = <0>;
1215 device_type = "network";
1216 model = "TSEC";
1217 compatible = "gianfar";
1218 reg = <24000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001219 mac-address = [ 00 E0 0C 00 73 00 ];
David Gibsonc125a182006-02-01 03:05:22 -08001220 interrupts = <d 3 e 3 12 3>;
1221 interrupt-parent = <40000>;
1222 phy-handle = <2452000>
1223 };
1224
1225
1226
1227 c) PHY nodes
1228
1229 Required properties:
1230
1231 - device_type : Should be "ethernet-phy"
1232 - interrupts : <a b> where a is the interrupt number and b is a
1233 field that represents an encoding of the sense and level
1234 information for the interrupt. This should be encoded based on
1235 the information in section 2) depending on the type of interrupt
1236 controller you have.
1237 - interrupt-parent : the phandle for the interrupt controller that
1238 services interrupts for this device.
1239 - reg : The ID number for the phy, usually a small integer
1240 - linux,phandle : phandle for this node; likely referenced by an
1241 ethernet controller node.
1242
1243
1244 Example:
1245
1246 ethernet-phy@0 {
1247 linux,phandle = <2452000>
1248 interrupt-parent = <40000>;
1249 interrupts = <35 1>;
1250 reg = <0>;
1251 device_type = "ethernet-phy";
1252 };
1253
1254
1255 d) Interrupt controllers
1256
1257 Some SOC devices contain interrupt controllers that are different
1258 from the standard Open PIC specification. The SOC device nodes for
1259 these types of controllers should be specified just like a standard
1260 OpenPIC controller. Sense and level information should be encoded
1261 as specified in section 2) of this chapter for each device that
1262 specifies an interrupt.
1263
1264 Example :
1265
1266 pic@40000 {
1267 linux,phandle = <40000>;
1268 clock-frequency = <0>;
1269 interrupt-controller;
1270 #address-cells = <0>;
1271 reg = <40000 40000>;
1272 built-in;
1273 compatible = "chrp,open-pic";
1274 device_type = "open-pic";
1275 big-endian;
1276 };
1277
1278
1279 e) I2C
1280
1281 Required properties :
1282
1283 - device_type : Should be "i2c"
1284 - reg : Offset and length of the register set for the device
1285
1286 Recommended properties :
1287
1288 - compatible : Should be "fsl-i2c" for parts compatible with
1289 Freescale I2C specifications.
1290 - interrupts : <a b> where a is the interrupt number and b is a
1291 field that represents an encoding of the sense and level
1292 information for the interrupt. This should be encoded based on
1293 the information in section 2) depending on the type of interrupt
1294 controller you have.
1295 - interrupt-parent : the phandle for the interrupt controller that
1296 services interrupts for this device.
1297 - dfsrr : boolean; if defined, indicates that this I2C device has
1298 a digital filter sampling rate register
1299 - fsl5200-clocking : boolean; if defined, indicated that this device
1300 uses the FSL 5200 clocking mechanism.
1301
1302 Example :
1303
1304 i2c@3000 {
1305 interrupt-parent = <40000>;
1306 interrupts = <1b 3>;
1307 reg = <3000 18>;
1308 device_type = "i2c";
1309 compatible = "fsl-i2c";
1310 dfsrr;
1311 };
1312
1313
Becky Brucead71f122006-02-07 13:44:08 -06001314 f) Freescale SOC USB controllers
1315
1316 The device node for a USB controller that is part of a Freescale
1317 SOC is as described in the document "Open Firmware Recommended
1318 Practice : Universal Serial Bus" with the following modifications
1319 and additions :
1320
1321 Required properties :
1322 - compatible : Should be "fsl-usb2-mph" for multi port host usb
1323 controllers, or "fsl-usb2-dr" for dual role usb controllers
1324 - phy_type : For multi port host usb controllers, should be one of
1325 "ulpi", or "serial". For dual role usb controllers, should be
1326 one of "ulpi", "utmi", "utmi_wide", or "serial".
1327 - reg : Offset and length of the register set for the device
1328 - port0 : boolean; if defined, indicates port0 is connected for
1329 fsl-usb2-mph compatible controllers. Either this property or
1330 "port1" (or both) must be defined for "fsl-usb2-mph" compatible
1331 controllers.
1332 - port1 : boolean; if defined, indicates port1 is connected for
1333 fsl-usb2-mph compatible controllers. Either this property or
1334 "port0" (or both) must be defined for "fsl-usb2-mph" compatible
1335 controllers.
1336
1337 Recommended properties :
1338 - interrupts : <a b> where a is the interrupt number and b is a
1339 field that represents an encoding of the sense and level
1340 information for the interrupt. This should be encoded based on
1341 the information in section 2) depending on the type of interrupt
1342 controller you have.
1343 - interrupt-parent : the phandle for the interrupt controller that
1344 services interrupts for this device.
1345
1346 Example multi port host usb controller device node :
1347 usb@22000 {
1348 device_type = "usb";
1349 compatible = "fsl-usb2-mph";
1350 reg = <22000 1000>;
1351 #address-cells = <1>;
1352 #size-cells = <0>;
1353 interrupt-parent = <700>;
1354 interrupts = <27 1>;
1355 phy_type = "ulpi";
1356 port0;
1357 port1;
1358 };
1359
1360 Example dual role usb controller device node :
1361 usb@23000 {
1362 device_type = "usb";
1363 compatible = "fsl-usb2-dr";
1364 reg = <23000 1000>;
1365 #address-cells = <1>;
1366 #size-cells = <0>;
1367 interrupt-parent = <700>;
1368 interrupts = <26 1>;
1369 phy = "ulpi";
1370 };
1371
1372
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001373 g) Freescale SOC SEC Security Engines
1374
1375 Required properties:
1376
1377 - device_type : Should be "crypto"
1378 - model : Model of the device. Should be "SEC1" or "SEC2"
1379 - compatible : Should be "talitos"
1380 - reg : Offset and length of the register set for the device
1381 - interrupts : <a b> where a is the interrupt number and b is a
1382 field that represents an encoding of the sense and level
1383 information for the interrupt. This should be encoded based on
1384 the information in section 2) depending on the type of interrupt
1385 controller you have.
1386 - interrupt-parent : the phandle for the interrupt controller that
1387 services interrupts for this device.
1388 - num-channels : An integer representing the number of channels
1389 available.
1390 - channel-fifo-len : An integer representing the number of
1391 descriptor pointers each channel fetch fifo can hold.
1392 - exec-units-mask : The bitmask representing what execution units
1393 (EUs) are available. It's a single 32 bit cell. EU information
1394 should be encoded following the SEC's Descriptor Header Dword
1395 EU_SEL0 field documentation, i.e. as follows:
1396
1397 bit 0 = reserved - should be 0
1398 bit 1 = set if SEC has the ARC4 EU (AFEU)
1399 bit 2 = set if SEC has the DES/3DES EU (DEU)
1400 bit 3 = set if SEC has the message digest EU (MDEU)
1401 bit 4 = set if SEC has the random number generator EU (RNG)
1402 bit 5 = set if SEC has the public key EU (PKEU)
1403 bit 6 = set if SEC has the AES EU (AESU)
1404 bit 7 = set if SEC has the Kasumi EU (KEU)
1405
1406 bits 8 through 31 are reserved for future SEC EUs.
1407
1408 - descriptor-types-mask : The bitmask representing what descriptors
1409 are available. It's a single 32 bit cell. Descriptor type
1410 information should be encoded following the SEC's Descriptor
1411 Header Dword DESC_TYPE field documentation, i.e. as follows:
1412
1413 bit 0 = set if SEC supports the aesu_ctr_nonsnoop desc. type
1414 bit 1 = set if SEC supports the ipsec_esp descriptor type
1415 bit 2 = set if SEC supports the common_nonsnoop desc. type
1416 bit 3 = set if SEC supports the 802.11i AES ccmp desc. type
1417 bit 4 = set if SEC supports the hmac_snoop_no_afeu desc. type
1418 bit 5 = set if SEC supports the srtp descriptor type
1419 bit 6 = set if SEC supports the non_hmac_snoop_no_afeu desc.type
1420 bit 7 = set if SEC supports the pkeu_assemble descriptor type
1421 bit 8 = set if SEC supports the aesu_key_expand_output desc.type
1422 bit 9 = set if SEC supports the pkeu_ptmul descriptor type
1423 bit 10 = set if SEC supports the common_nonsnoop_afeu desc. type
1424 bit 11 = set if SEC supports the pkeu_ptadd_dbl descriptor type
1425
1426 ..and so on and so forth.
1427
1428 Example:
1429
1430 /* MPC8548E */
1431 crypto@30000 {
1432 device_type = "crypto";
1433 model = "SEC2";
1434 compatible = "talitos";
1435 reg = <30000 10000>;
1436 interrupts = <1d 3>;
1437 interrupt-parent = <40000>;
1438 num-channels = <4>;
Kim Phillipscbdb54d2006-07-03 15:10:14 -05001439 channel-fifo-len = <18>;
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001440 exec-units-mask = <000000fe>;
Kim Phillipscbdb54d2006-07-03 15:10:14 -05001441 descriptor-types-mask = <012b0ebf>;
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001442 };
1443
Li Yang9a1ab882006-10-02 20:08:59 -05001444 h) Board Control and Status (BCSR)
1445
1446 Required properties:
1447
1448 - device_type : Should be "board-control"
1449 - reg : Offset and length of the register set for the device
1450
1451 Example:
1452
1453 bcsr@f8000000 {
1454 device_type = "board-control";
1455 reg = <f8000000 8000>;
1456 };
1457
1458 i) Freescale QUICC Engine module (QE)
1459 This represents qe module that is installed on PowerQUICC II Pro.
1460 Hopefully it will merge backward compatibility with CPM/CPM2.
1461 Basically, it is a bus of devices, that could act more or less
1462 as a complete entity (UCC, USB etc ). All of them should be siblings on
1463 the "root" qe node, using the common properties from there.
1464 The description below applies to the the qe of MPC8360 and
1465 more nodes and properties would be extended in the future.
1466
1467 i) Root QE device
1468
1469 Required properties:
1470 - device_type : should be "qe";
1471 - model : precise model of the QE, Can be "QE", "CPM", or "CPM2"
1472 - reg : offset and length of the device registers.
1473 - bus-frequency : the clock frequency for QUICC Engine.
1474
1475 Recommended properties
1476 - brg-frequency : the internal clock source frequency for baud-rate
1477 generators in Hz.
1478
1479 Example:
1480 qe@e0100000 {
1481 #address-cells = <1>;
1482 #size-cells = <1>;
1483 #interrupt-cells = <2>;
1484 device_type = "qe";
1485 model = "QE";
1486 ranges = <0 e0100000 00100000>;
1487 reg = <e0100000 480>;
1488 brg-frequency = <0>;
1489 bus-frequency = <179A7B00>;
1490 }
1491
1492
1493 ii) SPI (Serial Peripheral Interface)
1494
1495 Required properties:
1496 - device_type : should be "spi".
1497 - compatible : should be "fsl_spi".
1498 - mode : the spi operation mode, it can be "cpu" or "qe".
1499 - reg : Offset and length of the register set for the device
1500 - interrupts : <a b> where a is the interrupt number and b is a
1501 field that represents an encoding of the sense and level
1502 information for the interrupt. This should be encoded based on
1503 the information in section 2) depending on the type of interrupt
1504 controller you have.
1505 - interrupt-parent : the phandle for the interrupt controller that
1506 services interrupts for this device.
1507
1508 Example:
1509 spi@4c0 {
1510 device_type = "spi";
1511 compatible = "fsl_spi";
1512 reg = <4c0 40>;
1513 interrupts = <82 0>;
1514 interrupt-parent = <700>;
1515 mode = "cpu";
1516 };
1517
1518
1519 iii) USB (Universal Serial Bus Controller)
1520
1521 Required properties:
1522 - device_type : should be "usb".
1523 - compatible : could be "qe_udc" or "fhci-hcd".
1524 - mode : the could be "host" or "slave".
1525 - reg : Offset and length of the register set for the device
1526 - interrupts : <a b> where a is the interrupt number and b is a
1527 field that represents an encoding of the sense and level
1528 information for the interrupt. This should be encoded based on
1529 the information in section 2) depending on the type of interrupt
1530 controller you have.
1531 - interrupt-parent : the phandle for the interrupt controller that
1532 services interrupts for this device.
1533
1534 Example(slave):
1535 usb@6c0 {
1536 device_type = "usb";
1537 compatible = "qe_udc";
1538 reg = <6c0 40>;
1539 interrupts = <8b 0>;
1540 interrupt-parent = <700>;
1541 mode = "slave";
1542 };
1543
1544
1545 iv) UCC (Unified Communications Controllers)
1546
1547 Required properties:
1548 - device_type : should be "network", "hldc", "uart", "transparent"
1549 "bisync" or "atm".
1550 - compatible : could be "ucc_geth" or "fsl_atm" and so on.
1551 - model : should be "UCC".
1552 - device-id : the ucc number(1-8), corresponding to UCCx in UM.
1553 - reg : Offset and length of the register set for the device
1554 - interrupts : <a b> where a is the interrupt number and b is a
1555 field that represents an encoding of the sense and level
1556 information for the interrupt. This should be encoded based on
1557 the information in section 2) depending on the type of interrupt
1558 controller you have.
1559 - interrupt-parent : the phandle for the interrupt controller that
1560 services interrupts for this device.
1561 - pio-handle : The phandle for the Parallel I/O port configuration.
1562 - rx-clock : represents the UCC receive clock source.
1563 0x00 : clock source is disabled;
1564 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1565 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1566 - tx-clock: represents the UCC transmit clock source;
1567 0x00 : clock source is disabled;
1568 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1569 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1570
1571 Required properties for network device_type:
1572 - mac-address : list of bytes representing the ethernet address.
1573 - phy-handle : The phandle for the PHY connected to this controller.
1574
1575 Example:
1576 ucc@2000 {
1577 device_type = "network";
1578 compatible = "ucc_geth";
1579 model = "UCC";
1580 device-id = <1>;
1581 reg = <2000 200>;
1582 interrupts = <a0 0>;
1583 interrupt-parent = <700>;
1584 mac-address = [ 00 04 9f 00 23 23 ];
1585 rx-clock = "none";
1586 tx-clock = "clk9";
1587 phy-handle = <212000>;
1588 pio-handle = <140001>;
1589 };
1590
1591
1592 v) Parallel I/O Ports
1593
1594 This node configures Parallel I/O ports for CPUs with QE support.
1595 The node should reside in the "soc" node of the tree. For each
1596 device that using parallel I/O ports, a child node should be created.
1597 See the definition of the Pin configuration nodes below for more
1598 information.
1599
1600 Required properties:
1601 - device_type : should be "par_io".
1602 - reg : offset to the register set and its length.
1603 - num-ports : number of Parallel I/O ports
1604
1605 Example:
1606 par_io@1400 {
1607 reg = <1400 100>;
1608 #address-cells = <1>;
1609 #size-cells = <0>;
1610 device_type = "par_io";
1611 num-ports = <7>;
1612 ucc_pin@01 {
1613 ......
1614 };
1615
1616
1617 vi) Pin configuration nodes
1618
1619 Required properties:
1620 - linux,phandle : phandle of this node; likely referenced by a QE
1621 device.
1622 - pio-map : array of pin configurations. Each pin is defined by 6
1623 integers. The six numbers are respectively: port, pin, dir,
1624 open_drain, assignment, has_irq.
1625 - port : port number of the pin; 0-6 represent port A-G in UM.
1626 - pin : pin number in the port.
1627 - dir : direction of the pin, should encode as follows:
1628
1629 0 = The pin is disabled
1630 1 = The pin is an output
1631 2 = The pin is an input
1632 3 = The pin is I/O
1633
1634 - open_drain : indicates the pin is normal or wired-OR:
1635
1636 0 = The pin is actively driven as an output
1637 1 = The pin is an open-drain driver. As an output, the pin is
1638 driven active-low, otherwise it is three-stated.
1639
1640 - assignment : function number of the pin according to the Pin Assignment
1641 tables in User Manual. Each pin can have up to 4 possible functions in
1642 QE and two options for CPM.
1643 - has_irq : indicates if the pin is used as source of exteral
1644 interrupts.
1645
1646 Example:
1647 ucc_pin@01 {
1648 linux,phandle = <140001>;
1649 pio-map = <
1650 /* port pin dir open_drain assignment has_irq */
1651 0 3 1 0 1 0 /* TxD0 */
1652 0 4 1 0 1 0 /* TxD1 */
1653 0 5 1 0 1 0 /* TxD2 */
1654 0 6 1 0 1 0 /* TxD3 */
1655 1 6 1 0 3 0 /* TxD4 */
1656 1 7 1 0 1 0 /* TxD5 */
1657 1 9 1 0 2 0 /* TxD6 */
1658 1 a 1 0 2 0 /* TxD7 */
1659 0 9 2 0 1 0 /* RxD0 */
1660 0 a 2 0 1 0 /* RxD1 */
1661 0 b 2 0 1 0 /* RxD2 */
1662 0 c 2 0 1 0 /* RxD3 */
1663 0 d 2 0 1 0 /* RxD4 */
1664 1 1 2 0 2 0 /* RxD5 */
1665 1 0 2 0 2 0 /* RxD6 */
1666 1 4 2 0 2 0 /* RxD7 */
1667 0 7 1 0 1 0 /* TX_EN */
1668 0 8 1 0 1 0 /* TX_ER */
1669 0 f 2 0 1 0 /* RX_DV */
1670 0 10 2 0 1 0 /* RX_ER */
1671 0 0 2 0 1 0 /* RX_CLK */
1672 2 9 1 0 3 0 /* GTX_CLK - CLK10 */
1673 2 8 2 0 1 0>; /* GTX125 - CLK9 */
1674 };
1675
1676 vii) Multi-User RAM (MURAM)
1677
1678 Required properties:
1679 - device_type : should be "muram".
1680 - mode : the could be "host" or "slave".
1681 - ranges : Should be defined as specified in 1) to describe the
1682 translation of MURAM addresses.
1683 - data-only : sub-node which defines the address area under MURAM
1684 bus that can be allocated as data/parameter
1685
1686 Example:
1687
1688 muram@10000 {
1689 device_type = "muram";
1690 ranges = <0 00010000 0000c000>;
1691
1692 data-only@0{
1693 reg = <0 c000>;
1694 };
1695 };
Kim Phillipsb88a0b12006-03-22 14:39:03 -06001696
David Gibsonc125a182006-02-01 03:05:22 -08001697 More devices will be defined as this spec matures.
1698
1699
1700Appendix A - Sample SOC node for MPC8540
1701========================================
1702
1703Note that the #address-cells and #size-cells for the SoC node
1704in this example have been explicitly listed; these are likely
1705not necessary as they are usually the same as the root node.
1706
1707 soc8540@e0000000 {
1708 #address-cells = <1>;
1709 #size-cells = <1>;
1710 #interrupt-cells = <2>;
1711 device_type = "soc";
1712 ranges = <00000000 e0000000 00100000>
1713 reg = <e0000000 00003000>;
Becky Bruce7d4b95a2006-02-06 14:26:31 -06001714 bus-frequency = <0>;
David Gibsonc125a182006-02-01 03:05:22 -08001715
1716 mdio@24520 {
1717 reg = <24520 20>;
1718 device_type = "mdio";
1719 compatible = "gianfar";
1720
1721 ethernet-phy@0 {
1722 linux,phandle = <2452000>
1723 interrupt-parent = <40000>;
1724 interrupts = <35 1>;
1725 reg = <0>;
1726 device_type = "ethernet-phy";
1727 };
1728
1729 ethernet-phy@1 {
1730 linux,phandle = <2452001>
1731 interrupt-parent = <40000>;
1732 interrupts = <35 1>;
1733 reg = <1>;
1734 device_type = "ethernet-phy";
1735 };
1736
1737 ethernet-phy@3 {
1738 linux,phandle = <2452002>
1739 interrupt-parent = <40000>;
1740 interrupts = <35 1>;
1741 reg = <3>;
1742 device_type = "ethernet-phy";
1743 };
1744
1745 };
1746
1747 ethernet@24000 {
1748 #size-cells = <0>;
1749 device_type = "network";
1750 model = "TSEC";
1751 compatible = "gianfar";
1752 reg = <24000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001753 mac-address = [ 00 E0 0C 00 73 00 ];
David Gibsonc125a182006-02-01 03:05:22 -08001754 interrupts = <d 3 e 3 12 3>;
1755 interrupt-parent = <40000>;
1756 phy-handle = <2452000>;
1757 };
1758
1759 ethernet@25000 {
1760 #address-cells = <1>;
1761 #size-cells = <0>;
1762 device_type = "network";
1763 model = "TSEC";
1764 compatible = "gianfar";
1765 reg = <25000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001766 mac-address = [ 00 E0 0C 00 73 01 ];
David Gibsonc125a182006-02-01 03:05:22 -08001767 interrupts = <13 3 14 3 18 3>;
1768 interrupt-parent = <40000>;
1769 phy-handle = <2452001>;
1770 };
1771
1772 ethernet@26000 {
1773 #address-cells = <1>;
1774 #size-cells = <0>;
1775 device_type = "network";
1776 model = "FEC";
1777 compatible = "gianfar";
1778 reg = <26000 1000>;
Jon Loeligerf5831652006-08-17 08:42:35 -05001779 mac-address = [ 00 E0 0C 00 73 02 ];
David Gibsonc125a182006-02-01 03:05:22 -08001780 interrupts = <19 3>;
1781 interrupt-parent = <40000>;
1782 phy-handle = <2452002>;
1783 };
1784
1785 serial@4500 {
1786 device_type = "serial";
1787 compatible = "ns16550";
1788 reg = <4500 100>;
1789 clock-frequency = <0>;
1790 interrupts = <1a 3>;
1791 interrupt-parent = <40000>;
1792 };
1793
1794 pic@40000 {
1795 linux,phandle = <40000>;
1796 clock-frequency = <0>;
1797 interrupt-controller;
1798 #address-cells = <0>;
1799 reg = <40000 40000>;
1800 built-in;
1801 compatible = "chrp,open-pic";
1802 device_type = "open-pic";
1803 big-endian;
1804 };
1805
1806 i2c@3000 {
1807 interrupt-parent = <40000>;
1808 interrupts = <1b 3>;
1809 reg = <3000 18>;
1810 device_type = "i2c";
1811 compatible = "fsl-i2c";
1812 dfsrr;
1813 };
1814
1815 };