1 Booting the Linux/ppc kernel without Open Firmware
2 --------------------------------------------------
4 (c) 2005 Benjamin Herrenschmidt <benh at kernel.crashing.org>,
6 (c) 2005 Becky Bruce <becky.bruce at freescale.com>,
7 Freescale Semiconductor, FSL SOC and 32-bit additions
8 (c) 2006 MontaVista Software, Inc.
9 Flash chip node definition
15 1) Entry point for arch/powerpc
18 II - The DT block format
20 2) Device tree generalities
21 3) Device tree "structure" block
22 4) Device tree "strings" block
24 III - Required content of the device tree
25 1) Note about cells and address representation
26 2) Note about "compatible" properties
27 3) Note about "name" properties
28 4) Note about node and property names and character set
29 5) Required nodes and properties
33 d) the /memory node(s)
35 f) the /soc<SOCname> node
37 IV - "dtc", the device tree compiler
39 V - Recommendations for a bootloader
41 VI - System-on-a-chip devices and nodes
42 1) Defining child nodes of an SOC
43 2) Representing devices without a current OF specification
45 b) Gianfar-compatible ethernet nodes
47 d) Interrupt controllers
49 f) Freescale SOC USB controllers
50 g) Freescale SOC SEC Security Engines
51 h) Board Control and Status (BCSR)
52 i) Freescale QUICC Engine module (QE)
53 j) CFI or JEDEC memory-mapped NOR flash
54 k) Global Utilities Block
57 VII - Specifying interrupt information for devices
58 1) interrupts property
59 2) interrupt-parent property
60 3) OpenPIC Interrupt Controllers
61 4) ISA Interrupt Controllers
63 Appendix A - Sample SOC node for MPC8540
69 May 18, 2005: Rev 0.1 - Initial draft, no chapter III yet.
71 May 19, 2005: Rev 0.2 - Add chapter III and bits & pieces here or
72 clarifies the fact that a lot of things are
73 optional, the kernel only requires a very
74 small device tree, though it is encouraged
75 to provide an as complete one as possible.
77 May 24, 2005: Rev 0.3 - Precise that DT block has to be in RAM
79 - Define version 3 and new format version 16
80 for the DT block (version 16 needs kernel
81 patches, will be fwd separately).
82 String block now has a size, and full path
83 is replaced by unit name for more
85 linux,phandle is made optional, only nodes
86 that are referenced by other nodes need it.
87 "name" property is now automatically
88 deduced from the unit name
90 June 1, 2005: Rev 0.4 - Correct confusion between OF_DT_END and
91 OF_DT_END_NODE in structure definition.
92 - Change version 16 format to always align
93 property data to 4 bytes. Since tokens are
94 already aligned, that means no specific
95 required alignment between property size
96 and property data. The old style variable
97 alignment would make it impossible to do
98 "simple" insertion of properties using
99 memmove (thanks Milton for
100 noticing). Updated kernel patch as well
101 - Correct a few more alignment constraints
102 - Add a chapter about the device-tree
103 compiler and the textural representation of
104 the tree that can be "compiled" by dtc.
106 November 21, 2005: Rev 0.5
107 - Additions/generalizations for 32-bit
108 - Changed to reflect the new arch/powerpc
114 - Add some definitions of interrupt tree (simple/complex)
115 - Add some definitions for PCI host bridges
116 - Add some common address format examples
117 - Add definitions for standard properties and "compatible"
118 names for cells that are not already defined by the existing
120 - Compare FSL SOC use of PCI to standard and make sure no new
121 node definition required.
122 - Add more information about node definitions for SOC devices
123 that currently have no standard, like the FSL CPM.
129 During the recent development of the Linux/ppc64 kernel, and more
130 specifically, the addition of new platform types outside of the old
131 IBM pSeries/iSeries pair, it was decided to enforce some strict rules
132 regarding the kernel entry and bootloader <-> kernel interfaces, in
133 order to avoid the degeneration that had become the ppc32 kernel entry
134 point and the way a new platform should be added to the kernel. The
135 legacy iSeries platform breaks those rules as it predates this scheme,
136 but no new board support will be accepted in the main tree that
137 doesn't follows them properly. In addition, since the advent of the
138 arch/powerpc merged architecture for ppc32 and ppc64, new 32-bit
139 platforms and 32-bit platforms which move into arch/powerpc will be
140 required to use these rules as well.
142 The main requirement that will be defined in more detail below is
143 the presence of a device-tree whose format is defined after Open
144 Firmware specification. However, in order to make life easier
145 to embedded board vendors, the kernel doesn't require the device-tree
146 to represent every device in the system and only requires some nodes
147 and properties to be present. This will be described in detail in
148 section III, but, for example, the kernel does not require you to
149 create a node for every PCI device in the system. It is a requirement
150 to have a node for PCI host bridges in order to provide interrupt
151 routing informations and memory/IO ranges, among others. It is also
152 recommended to define nodes for on chip devices and other busses that
153 don't specifically fit in an existing OF specification. This creates a
154 great flexibility in the way the kernel can then probe those and match
155 drivers to device, without having to hard code all sorts of tables. It
156 also makes it more flexible for board vendors to do minor hardware
157 upgrades without significantly impacting the kernel code or cluttering
158 it with special cases.
161 1) Entry point for arch/powerpc
162 -------------------------------
164 There is one and one single entry point to the kernel, at the start
165 of the kernel image. That entry point supports two calling
168 a) Boot from Open Firmware. If your firmware is compatible
169 with Open Firmware (IEEE 1275) or provides an OF compatible
170 client interface API (support for "interpret" callback of
171 forth words isn't required), you can enter the kernel with:
173 r5 : OF callback pointer as defined by IEEE 1275
174 bindings to powerpc. Only the 32-bit client interface
175 is currently supported
177 r3, r4 : address & length of an initrd if any or 0
179 The MMU is either on or off; the kernel will run the
180 trampoline located in arch/powerpc/kernel/prom_init.c to
181 extract the device-tree and other information from open
182 firmware and build a flattened device-tree as described
183 in b). prom_init() will then re-enter the kernel using
184 the second method. This trampoline code runs in the
185 context of the firmware, which is supposed to handle all
186 exceptions during that time.
188 b) Direct entry with a flattened device-tree block. This entry
189 point is called by a) after the OF trampoline and can also be
190 called directly by a bootloader that does not support the Open
191 Firmware client interface. It is also used by "kexec" to
192 implement "hot" booting of a new kernel from a previous
193 running one. This method is what I will describe in more
194 details in this document, as method a) is simply standard Open
195 Firmware, and thus should be implemented according to the
196 various standard documents defining it and its binding to the
197 PowerPC platform. The entry point definition then becomes:
199 r3 : physical pointer to the device-tree block
200 (defined in chapter II) in RAM
202 r4 : physical pointer to the kernel itself. This is
203 used by the assembly code to properly disable the MMU
204 in case you are entering the kernel with MMU enabled
205 and a non-1:1 mapping.
207 r5 : NULL (as to differentiate with method a)
209 Note about SMP entry: Either your firmware puts your other
210 CPUs in some sleep loop or spin loop in ROM where you can get
211 them out via a soft reset or some other means, in which case
212 you don't need to care, or you'll have to enter the kernel
213 with all CPUs. The way to do that with method b) will be
214 described in a later revision of this document.
222 Board supports (platforms) are not exclusive config options. An
223 arbitrary set of board supports can be built in a single kernel
224 image. The kernel will "know" what set of functions to use for a
225 given platform based on the content of the device-tree. Thus, you
228 a) add your platform support as a _boolean_ option in
229 arch/powerpc/Kconfig, following the example of PPC_PSERIES,
230 PPC_PMAC and PPC_MAPLE. The later is probably a good
231 example of a board support to start from.
233 b) create your main platform file as
234 "arch/powerpc/platforms/myplatform/myboard_setup.c" and add it
235 to the Makefile under the condition of your CONFIG_
236 option. This file will define a structure of type "ppc_md"
237 containing the various callbacks that the generic code will
238 use to get to your platform specific code
240 c) Add a reference to your "ppc_md" structure in the
241 "machines" table in arch/powerpc/kernel/setup_64.c if you are
244 d) request and get assigned a platform number (see PLATFORM_*
245 constants in include/asm-powerpc/processor.h
247 32-bit embedded kernels:
249 Currently, board support is essentially an exclusive config option.
250 The kernel is configured for a single platform. Part of the reason
251 for this is to keep kernels on embedded systems small and efficient;
252 part of this is due to the fact the code is already that way. In the
253 future, a kernel may support multiple platforms, but only if the
254 platforms feature the same core architecture. A single kernel build
255 cannot support both configurations with Book E and configurations
256 with classic Powerpc architectures.
258 32-bit embedded platforms that are moved into arch/powerpc using a
259 flattened device tree should adopt the merged tree practice of
260 setting ppc_md up dynamically, even though the kernel is currently
261 built with support for only a single platform at a time. This allows
262 unification of the setup code, and will make it easier to go to a
263 multiple-platform-support model in the future.
265 NOTE: I believe the above will be true once Ben's done with the merge
266 of the boot sequences.... someone speak up if this is wrong!
268 To add a 32-bit embedded platform support, follow the instructions
269 for 64-bit platforms above, with the exception that the Kconfig
270 option should be set up such that the kernel builds exclusively for
271 the platform selected. The processor type for the platform should
272 enable another config option to select the specific board
275 NOTE: If Ben doesn't merge the setup files, may need to change this to
279 I will describe later the boot process and various callbacks that
280 your platform should implement.
283 II - The DT block format
284 ========================
287 This chapter defines the actual format of the flattened device-tree
288 passed to the kernel. The actual content of it and kernel requirements
289 are described later. You can find example of code manipulating that
290 format in various places, including arch/powerpc/kernel/prom_init.c
291 which will generate a flattened device-tree from the Open Firmware
292 representation, or the fs2dt utility which is part of the kexec tools
293 which will generate one from a filesystem representation. It is
294 expected that a bootloader like uboot provides a bit more support,
295 that will be discussed later as well.
297 Note: The block has to be in main memory. It has to be accessible in
298 both real mode and virtual mode with no mapping other than main
299 memory. If you are writing a simple flash bootloader, it should copy
300 the block to RAM before passing it to the kernel.
306 The kernel is entered with r3 pointing to an area of memory that is
307 roughly described in include/asm-powerpc/prom.h by the structure
310 struct boot_param_header {
311 u32 magic; /* magic word OF_DT_HEADER */
312 u32 totalsize; /* total size of DT block */
313 u32 off_dt_struct; /* offset to structure */
314 u32 off_dt_strings; /* offset to strings */
315 u32 off_mem_rsvmap; /* offset to memory reserve map
317 u32 version; /* format version */
318 u32 last_comp_version; /* last compatible version */
320 /* version 2 fields below */
321 u32 boot_cpuid_phys; /* Which physical CPU id we're
323 /* version 3 fields below */
324 u32 size_dt_strings; /* size of the strings block */
326 /* version 17 fields below */
327 u32 size_dt_struct; /* size of the DT structure block */
330 Along with the constants:
332 /* Definitions used by the flattened device tree */
333 #define OF_DT_HEADER 0xd00dfeed /* 4: version,
335 #define OF_DT_BEGIN_NODE 0x1 /* Start node: full name
337 #define OF_DT_END_NODE 0x2 /* End node */
338 #define OF_DT_PROP 0x3 /* Property: name off,
340 #define OF_DT_END 0x9
342 All values in this header are in big endian format, the various
343 fields in this header are defined more precisely below. All
344 "offset" values are in bytes from the start of the header; that is
345 from the value of r3.
349 This is a magic value that "marks" the beginning of the
350 device-tree block header. It contains the value 0xd00dfeed and is
351 defined by the constant OF_DT_HEADER
355 This is the total size of the DT block including the header. The
356 "DT" block should enclose all data structures defined in this
357 chapter (who are pointed to by offsets in this header). That is,
358 the device-tree structure, strings, and the memory reserve map.
362 This is an offset from the beginning of the header to the start
363 of the "structure" part the device tree. (see 2) device tree)
367 This is an offset from the beginning of the header to the start
368 of the "strings" part of the device-tree
372 This is an offset from the beginning of the header to the start
373 of the reserved memory map. This map is a list of pairs of 64-
374 bit integers. Each pair is a physical address and a size. The
375 list is terminated by an entry of size 0. This map provides the
376 kernel with a list of physical memory areas that are "reserved"
377 and thus not to be used for memory allocations, especially during
378 early initialization. The kernel needs to allocate memory during
379 boot for things like un-flattening the device-tree, allocating an
380 MMU hash table, etc... Those allocations must be done in such a
381 way to avoid overriding critical things like, on Open Firmware
382 capable machines, the RTAS instance, or on some pSeries, the TCE
383 tables used for the iommu. Typically, the reserve map should
384 contain _at least_ this DT block itself (header,total_size). If
385 you are passing an initrd to the kernel, you should reserve it as
386 well. You do not need to reserve the kernel image itself. The map
387 should be 64-bit aligned.
391 This is the version of this structure. Version 1 stops
392 here. Version 2 adds an additional field boot_cpuid_phys.
393 Version 3 adds the size of the strings block, allowing the kernel
394 to reallocate it easily at boot and free up the unused flattened
395 structure after expansion. Version 16 introduces a new more
396 "compact" format for the tree itself that is however not backward
397 compatible. Version 17 adds an additional field, size_dt_struct,
398 allowing it to be reallocated or moved more easily (this is
399 particularly useful for bootloaders which need to make
400 adjustments to a device tree based on probed information). You
401 should always generate a structure of the highest version defined
402 at the time of your implementation. Currently that is version 17,
403 unless you explicitly aim at being backward compatible.
407 Last compatible version. This indicates down to what version of
408 the DT block you are backward compatible. For example, version 2
409 is backward compatible with version 1 (that is, a kernel build
410 for version 1 will be able to boot with a version 2 format). You
411 should put a 1 in this field if you generate a device tree of
412 version 1 to 3, or 16 if you generate a tree of version 16 or 17
413 using the new unit name format.
417 This field only exist on version 2 headers. It indicate which
418 physical CPU ID is calling the kernel entry point. This is used,
419 among others, by kexec. If you are on an SMP system, this value
420 should match the content of the "reg" property of the CPU node in
421 the device-tree corresponding to the CPU calling the kernel entry
422 point (see further chapters for more informations on the required
423 device-tree contents)
427 This field only exists on version 3 and later headers. It
428 gives the size of the "strings" section of the device tree (which
429 starts at the offset given by off_dt_strings).
433 This field only exists on version 17 and later headers. It gives
434 the size of the "structure" section of the device tree (which
435 starts at the offset given by off_dt_struct).
437 So the typical layout of a DT block (though the various parts don't
438 need to be in that order) looks like this (addresses go from top to
442 ------------------------------
443 r3 -> | struct boot_param_header |
444 ------------------------------
445 | (alignment gap) (*) |
446 ------------------------------
447 | memory reserve map |
448 ------------------------------
450 ------------------------------
452 | device-tree structure |
454 ------------------------------
456 ------------------------------
458 | device-tree strings |
460 -----> ------------------------------
465 (*) The alignment gaps are not necessarily present; their presence
466 and size are dependent on the various alignment requirements of
467 the individual data blocks.
470 2) Device tree generalities
471 ---------------------------
473 This device-tree itself is separated in two different blocks, a
474 structure block and a strings block. Both need to be aligned to a 4
477 First, let's quickly describe the device-tree concept before detailing
478 the storage format. This chapter does _not_ describe the detail of the
479 required types of nodes & properties for the kernel, this is done
480 later in chapter III.
482 The device-tree layout is strongly inherited from the definition of
483 the Open Firmware IEEE 1275 device-tree. It's basically a tree of
484 nodes, each node having two or more named properties. A property can
487 It is a tree, so each node has one and only one parent except for the
488 root node who has no parent.
490 A node has 2 names. The actual node name is generally contained in a
491 property of type "name" in the node property list whose value is a
492 zero terminated string and is mandatory for version 1 to 3 of the
493 format definition (as it is in Open Firmware). Version 16 makes it
494 optional as it can generate it from the unit name defined below.
496 There is also a "unit name" that is used to differentiate nodes with
497 the same name at the same level, it is usually made of the node
498 names, the "@" sign, and a "unit address", which definition is
499 specific to the bus type the node sits on.
501 The unit name doesn't exist as a property per-se but is included in
502 the device-tree structure. It is typically used to represent "path" in
503 the device-tree. More details about the actual format of these will be
506 The kernel powerpc generic code does not make any formal use of the
507 unit address (though some board support code may do) so the only real
508 requirement here for the unit address is to ensure uniqueness of
509 the node unit name at a given level of the tree. Nodes with no notion
510 of address and no possible sibling of the same name (like /memory or
511 /cpus) may omit the unit address in the context of this specification,
512 or use the "@0" default unit address. The unit name is used to define
513 a node "full path", which is the concatenation of all parent node
514 unit names separated with "/".
516 The root node doesn't have a defined name, and isn't required to have
517 a name property either if you are using version 3 or earlier of the
518 format. It also has no unit address (no @ symbol followed by a unit
519 address). The root node unit name is thus an empty string. The full
520 path to the root node is "/".
522 Every node which actually represents an actual device (that is, a node
523 which isn't only a virtual "container" for more nodes, like "/cpus"
524 is) is also required to have a "device_type" property indicating the
527 Finally, every node that can be referenced from a property in another
528 node is required to have a "linux,phandle" property. Real open
529 firmware implementations provide a unique "phandle" value for every
530 node that the "prom_init()" trampoline code turns into
531 "linux,phandle" properties. However, this is made optional if the
532 flattened device tree is used directly. An example of a node
533 referencing another node via "phandle" is when laying out the
534 interrupt tree which will be described in a further version of this
537 This "linux, phandle" property is a 32-bit value that uniquely
538 identifies a node. You are free to use whatever values or system of
539 values, internal pointers, or whatever to generate these, the only
540 requirement is that every node for which you provide that property has
541 a unique value for it.
543 Here is an example of a simple device-tree. In this example, an "o"
544 designates a node followed by the node unit name. Properties are
545 presented with their name followed by their content. "content"
546 represents an ASCII string (zero terminated) value, while <content>
547 represents a 32-bit hexadecimal value. The various nodes in this
548 example will be discussed in a later chapter. At this point, it is
549 only meant to give you a idea of what a device-tree looks like. I have
550 purposefully kept the "name" and "linux,phandle" properties which
551 aren't necessary in order to give you a better idea of what the tree
552 looks like in practice.
555 |- name = "device-tree"
556 |- model = "MyBoardName"
557 |- compatible = "MyBoardFamilyName"
558 |- #address-cells = <2>
560 |- linux,phandle = <0>
564 | | - linux,phandle = <1>
565 | | - #address-cells = <1>
566 | | - #size-cells = <0>
569 | |- name = "PowerPC,970"
570 | |- device_type = "cpu"
572 | |- clock-frequency = <5f5e1000>
574 | |- linux,phandle = <2>
578 | |- device_type = "memory"
579 | |- reg = <00000000 00000000 00000000 20000000>
580 | |- linux,phandle = <3>
584 |- bootargs = "root=/dev/sda2"
585 |- linux,phandle = <4>
587 This tree is almost a minimal tree. It pretty much contains the
588 minimal set of required nodes and properties to boot a linux kernel;
589 that is, some basic model informations at the root, the CPUs, and the
590 physical memory layout. It also includes misc information passed
591 through /chosen, like in this example, the platform type (mandatory)
592 and the kernel command line arguments (optional).
594 The /cpus/PowerPC,970@0/64-bit property is an example of a
595 property without a value. All other properties have a value. The
596 significance of the #address-cells and #size-cells properties will be
597 explained in chapter IV which defines precisely the required nodes and
598 properties and their content.
601 3) Device tree "structure" block
603 The structure of the device tree is a linearized tree structure. The
604 "OF_DT_BEGIN_NODE" token starts a new node, and the "OF_DT_END_NODE"
605 ends that node definition. Child nodes are simply defined before
606 "OF_DT_END_NODE" (that is nodes within the node). A 'token' is a 32
607 bit value. The tree has to be "finished" with a OF_DT_END token
609 Here's the basic structure of a single node:
611 * token OF_DT_BEGIN_NODE (that is 0x00000001)
612 * for version 1 to 3, this is the node full path as a zero
613 terminated string, starting with "/". For version 16 and later,
614 this is the node unit name only (or an empty string for the
616 * [align gap to next 4 bytes boundary]
618 * token OF_DT_PROP (that is 0x00000003)
619 * 32-bit value of property value size in bytes (or 0 if no
621 * 32-bit value of offset in string block of property name
622 * property value data if any
623 * [align gap to next 4 bytes boundary]
624 * [child nodes if any]
625 * token OF_DT_END_NODE (that is 0x00000002)
627 So the node content can be summarized as a start token, a full path,
628 a list of properties, a list of child nodes, and an end token. Every
629 child node is a full node structure itself as defined above.
631 NOTE: The above definition requires that all property definitions for
632 a particular node MUST precede any subnode definitions for that node.
633 Although the structure would not be ambiguous if properties and
634 subnodes were intermingled, the kernel parser requires that the
635 properties come first (up until at least 2.6.22). Any tools
636 manipulating a flattened tree must take care to preserve this
639 4) Device tree "strings" block
641 In order to save space, property names, which are generally redundant,
642 are stored separately in the "strings" block. This block is simply the
643 whole bunch of zero terminated strings for all property names
644 concatenated together. The device-tree property definitions in the
645 structure block will contain offset values from the beginning of the
649 III - Required content of the device tree
650 =========================================
652 WARNING: All "linux,*" properties defined in this document apply only
653 to a flattened device-tree. If your platform uses a real
654 implementation of Open Firmware or an implementation compatible with
655 the Open Firmware client interface, those properties will be created
656 by the trampoline code in the kernel's prom_init() file. For example,
657 that's where you'll have to add code to detect your board model and
658 set the platform number. However, when using the flattened device-tree
659 entry point, there is no prom_init() pass, and thus you have to
660 provide those properties yourself.
663 1) Note about cells and address representation
664 ----------------------------------------------
666 The general rule is documented in the various Open Firmware
667 documentations. If you choose to describe a bus with the device-tree
668 and there exist an OF bus binding, then you should follow the
669 specification. However, the kernel does not require every single
670 device or bus to be described by the device tree.
672 In general, the format of an address for a device is defined by the
673 parent bus type, based on the #address-cells and #size-cells
674 property. In the absence of such a property, the parent's parent
675 values are used, etc... The kernel requires the root node to have
676 those properties defining addresses format for devices directly mapped
677 on the processor bus.
679 Those 2 properties define 'cells' for representing an address and a
680 size. A "cell" is a 32-bit number. For example, if both contain 2
681 like the example tree given above, then an address and a size are both
682 composed of 2 cells, and each is a 64-bit number (cells are
683 concatenated and expected to be in big endian format). Another example
684 is the way Apple firmware defines them, with 2 cells for an address
685 and one cell for a size. Most 32-bit implementations should define
686 #address-cells and #size-cells to 1, which represents a 32-bit value.
687 Some 32-bit processors allow for physical addresses greater than 32
688 bits; these processors should define #address-cells as 2.
690 "reg" properties are always a tuple of the type "address size" where
691 the number of cells of address and size is specified by the bus
692 #address-cells and #size-cells. When a bus supports various address
693 spaces and other flags relative to a given address allocation (like
694 prefetchable, etc...) those flags are usually added to the top level
695 bits of the physical address. For example, a PCI physical address is
696 made of 3 cells, the bottom two containing the actual address itself
697 while the top cell contains address space indication, flags, and pci
698 bus & device numbers.
700 For busses that support dynamic allocation, it's the accepted practice
701 to then not provide the address in "reg" (keep it 0) though while
702 providing a flag indicating the address is dynamically allocated, and
703 then, to provide a separate "assigned-addresses" property that
704 contains the fully allocated addresses. See the PCI OF bindings for
707 In general, a simple bus with no address space bits and no dynamic
708 allocation is preferred if it reflects your hardware, as the existing
709 kernel address parsing functions will work out of the box. If you
710 define a bus type with a more complex address format, including things
711 like address space bits, you'll have to add a bus translator to the
712 prom_parse.c file of the recent kernels for your bus type.
714 The "reg" property only defines addresses and sizes (if #size-cells
715 is non-0) within a given bus. In order to translate addresses upward
716 (that is into parent bus addresses, and possibly into CPU physical
717 addresses), all busses must contain a "ranges" property. If the
718 "ranges" property is missing at a given level, it's assumed that
719 translation isn't possible. The format of the "ranges" property for a
722 bus address, parent bus address, size
724 "bus address" is in the format of the bus this bus node is defining,
725 that is, for a PCI bridge, it would be a PCI address. Thus, (bus
726 address, size) defines a range of addresses for child devices. "parent
727 bus address" is in the format of the parent bus of this bus. For
728 example, for a PCI host controller, that would be a CPU address. For a
729 PCI<->ISA bridge, that would be a PCI address. It defines the base
730 address in the parent bus where the beginning of that range is mapped.
732 For a new 64-bit powerpc board, I recommend either the 2/2 format or
733 Apple's 2/1 format which is slightly more compact since sizes usually
734 fit in a single 32-bit word. New 32-bit powerpc boards should use a
735 1/1 format, unless the processor supports physical addresses greater
736 than 32-bits, in which case a 2/1 format is recommended.
739 2) Note about "compatible" properties
740 -------------------------------------
742 These properties are optional, but recommended in devices and the root
743 node. The format of a "compatible" property is a list of concatenated
744 zero terminated strings. They allow a device to express its
745 compatibility with a family of similar devices, in some cases,
746 allowing a single driver to match against several devices regardless
747 of their actual names.
749 3) Note about "name" properties
750 -------------------------------
752 While earlier users of Open Firmware like OldWorld macintoshes tended
753 to use the actual device name for the "name" property, it's nowadays
754 considered a good practice to use a name that is closer to the device
755 class (often equal to device_type). For example, nowadays, ethernet
756 controllers are named "ethernet", an additional "model" property
757 defining precisely the chip type/model, and "compatible" property
758 defining the family in case a single driver can driver more than one
759 of these chips. However, the kernel doesn't generally put any
760 restriction on the "name" property; it is simply considered good
761 practice to follow the standard and its evolutions as closely as
764 Note also that the new format version 16 makes the "name" property
765 optional. If it's absent for a node, then the node's unit name is then
766 used to reconstruct the name. That is, the part of the unit name
767 before the "@" sign is used (or the entire unit name if no "@" sign
770 4) Note about node and property names and character set
771 -------------------------------------------------------
773 While open firmware provides more flexible usage of 8859-1, this
774 specification enforces more strict rules. Nodes and properties should
775 be comprised only of ASCII characters 'a' to 'z', '0' to
776 '9', ',', '.', '_', '+', '#', '?', and '-'. Node names additionally
777 allow uppercase characters 'A' to 'Z' (property names should be
778 lowercase. The fact that vendors like Apple don't respect this rule is
779 irrelevant here). Additionally, node and property names should always
780 begin with a character in the range 'a' to 'z' (or 'A' to 'Z' for node
783 The maximum number of characters for both nodes and property names
784 is 31. In the case of node names, this is only the leftmost part of
785 a unit name (the pure "name" property), it doesn't include the unit
786 address which can extend beyond that limit.
789 5) Required nodes and properties
790 --------------------------------
791 These are all that are currently required. However, it is strongly
792 recommended that you expose PCI host bridges as documented in the
793 PCI binding to open firmware, and your interrupt tree as documented
794 in OF interrupt tree specification.
798 The root node requires some properties to be present:
800 - model : this is your board name/model
801 - #address-cells : address representation for "root" devices
802 - #size-cells: the size representation for "root" devices
803 - device_type : This property shouldn't be necessary. However, if
804 you decide to create a device_type for your root node, make sure it
805 is _not_ "chrp" unless your platform is a pSeries or PAPR compliant
806 one for 64-bit, or a CHRP-type machine for 32-bit as this will
807 matched by the kernel this way.
809 Additionally, some recommended properties are:
811 - compatible : the board "family" generally finds its way here,
812 for example, if you have 2 board models with a similar layout,
813 that typically get driven by the same platform code in the
814 kernel, you would use a different "model" property but put a
815 value in "compatible". The kernel doesn't directly use that
816 value but it is generally useful.
818 The root node is also generally where you add additional properties
819 specific to your board like the serial number if any, that sort of
820 thing. It is recommended that if you add any "custom" property whose
821 name may clash with standard defined ones, you prefix them with your
822 vendor name and a comma.
826 This node is the parent of all individual CPU nodes. It doesn't
827 have any specific requirements, though it's generally good practice
830 #address-cells = <00000001>
831 #size-cells = <00000000>
833 This defines that the "address" for a CPU is a single cell, and has
834 no meaningful size. This is not necessary but the kernel will assume
835 that format when reading the "reg" properties of a CPU node, see
840 So under /cpus, you are supposed to create a node for every CPU on
841 the machine. There is no specific restriction on the name of the
842 CPU, though It's common practice to call it PowerPC,<name>. For
843 example, Apple uses PowerPC,G5 while IBM uses PowerPC,970FX.
847 - device_type : has to be "cpu"
848 - reg : This is the physical CPU number, it's a single 32-bit cell
849 and is also used as-is as the unit number for constructing the
850 unit name in the full path. For example, with 2 CPUs, you would
852 /cpus/PowerPC,970FX@0
853 /cpus/PowerPC,970FX@1
854 (unit addresses do not require leading zeroes)
855 - d-cache-block-size : one cell, L1 data cache block size in bytes (*)
856 - i-cache-block-size : one cell, L1 instruction cache block size in
858 - d-cache-size : one cell, size of L1 data cache in bytes
859 - i-cache-size : one cell, size of L1 instruction cache in bytes
861 (*) The cache "block" size is the size on which the cache management
862 instructions operate. Historically, this document used the cache
863 "line" size here which is incorrect. The kernel will prefer the cache
864 block size and will fallback to cache line size for backward
867 Recommended properties:
869 - timebase-frequency : a cell indicating the frequency of the
870 timebase in Hz. This is not directly used by the generic code,
871 but you are welcome to copy/paste the pSeries code for setting
872 the kernel timebase/decrementer calibration based on this
874 - clock-frequency : a cell indicating the CPU core clock frequency
875 in Hz. A new property will be defined for 64-bit values, but if
876 your frequency is < 4Ghz, one cell is enough. Here as well as
877 for the above, the common code doesn't use that property, but
878 you are welcome to re-use the pSeries or Maple one. A future
879 kernel version might provide a common function for this.
880 - d-cache-line-size : one cell, L1 data cache line size in bytes
881 if different from the block size
882 - i-cache-line-size : one cell, L1 instruction cache line size in
883 bytes if different from the block size
885 You are welcome to add any property you find relevant to your board,
886 like some information about the mechanism used to soft-reset the
887 CPUs. For example, Apple puts the GPIO number for CPU soft reset
888 lines in there as a "soft-reset" property since they start secondary
889 CPUs by soft-resetting them.
892 d) the /memory node(s)
894 To define the physical memory layout of your board, you should
895 create one or more memory node(s). You can either create a single
896 node with all memory ranges in its reg property, or you can create
897 several nodes, as you wish. The unit address (@ part) used for the
898 full path is the address of the first range of memory defined by a
899 given node. If you use a single memory node, this will typically be
904 - device_type : has to be "memory"
905 - reg : This property contains all the physical memory ranges of
906 your board. It's a list of addresses/sizes concatenated
907 together, with the number of cells of each defined by the
908 #address-cells and #size-cells of the root node. For example,
909 with both of these properties being 2 like in the example given
910 earlier, a 970 based machine with 6Gb of RAM could typically
911 have a "reg" property here that looks like:
913 00000000 00000000 00000000 80000000
914 00000001 00000000 00000001 00000000
916 That is a range starting at 0 of 0x80000000 bytes and a range
917 starting at 0x100000000 and of 0x100000000 bytes. You can see
918 that there is no memory covering the IO hole between 2Gb and
919 4Gb. Some vendors prefer splitting those ranges into smaller
920 segments, but the kernel doesn't care.
924 This node is a bit "special". Normally, that's where open firmware
925 puts some variable environment information, like the arguments, or
926 the default input/output devices.
928 This specification makes a few of these mandatory, but also defines
929 some linux-specific properties that would be normally constructed by
930 the prom_init() trampoline when booting with an OF client interface,
931 but that you have to provide yourself when using the flattened format.
933 Recommended properties:
935 - bootargs : This zero-terminated string is passed as the kernel
937 - linux,stdout-path : This is the full path to your standard
938 console device if any. Typically, if you have serial devices on
939 your board, you may want to put the full path to the one set as
940 the default console in the firmware here, for the kernel to pick
941 it up as its own default console. If you look at the function
942 set_preferred_console() in arch/ppc64/kernel/setup.c, you'll see
943 that the kernel tries to find out the default console and has
944 knowledge of various types like 8250 serial ports. You may want
945 to extend this function to add your own.
947 Note that u-boot creates and fills in the chosen node for platforms
950 (Note: a practice that is now obsolete was to include a property
951 under /chosen called interrupt-controller which had a phandle value
952 that pointed to the main interrupt controller)
954 f) the /soc<SOCname> node
956 This node is used to represent a system-on-a-chip (SOC) and must be
957 present if the processor is a SOC. The top-level soc node contains
958 information that is global to all devices on the SOC. The node name
959 should contain a unit address for the SOC, which is the base address
960 of the memory-mapped register set for the SOC. The name of an soc
961 node should start with "soc", and the remainder of the name should
962 represent the part number for the soc. For example, the MPC8540's
963 soc node would be called "soc8540".
967 - device_type : Should be "soc"
968 - ranges : Should be defined as specified in 1) to describe the
969 translation of SOC addresses for memory mapped SOC registers.
970 - bus-frequency: Contains the bus frequency for the SOC node.
971 Typically, the value of this field is filled in by the boot
975 Recommended properties:
977 - reg : This property defines the address and size of the
978 memory-mapped registers that are used for the SOC node itself.
979 It does not include the child device registers - these will be
980 defined inside each child node. The address specified in the
981 "reg" property should match the unit address of the SOC node.
982 - #address-cells : Address representation for "soc" devices. The
983 format of this field may vary depending on whether or not the
984 device registers are memory mapped. For memory mapped
985 registers, this field represents the number of cells needed to
986 represent the address of the registers. For SOCs that do not
987 use MMIO, a special address format should be defined that
988 contains enough cells to represent the required information.
989 See 1) above for more details on defining #address-cells.
990 - #size-cells : Size representation for "soc" devices
991 - #interrupt-cells : Defines the width of cells used to represent
992 interrupts. Typically this value is <2>, which includes a
993 32-bit number that represents the interrupt number, and a
994 32-bit number that represents the interrupt sense and level.
995 This field is only needed if the SOC contains an interrupt
998 The SOC node may contain child nodes for each SOC device that the
999 platform uses. Nodes should not be created for devices which exist
1000 on the SOC but are not used by a particular platform. See chapter VI
1001 for more information on how to specify devices that are part of a SOC.
1003 Example SOC node for the MPC8540:
1006 #address-cells = <1>;
1008 #interrupt-cells = <2>;
1009 device_type = "soc";
1010 ranges = <00000000 e0000000 00100000>
1011 reg = <e0000000 00003000>;
1012 bus-frequency = <0>;
1017 IV - "dtc", the device tree compiler
1018 ====================================
1021 dtc source code can be found at
1022 <http://ozlabs.org/~dgibson/dtc/dtc.tar.gz>
1024 WARNING: This version is still in early development stage; the
1025 resulting device-tree "blobs" have not yet been validated with the
1026 kernel. The current generated bloc lacks a useful reserve map (it will
1027 be fixed to generate an empty one, it's up to the bootloader to fill
1028 it up) among others. The error handling needs work, bugs are lurking,
1031 dtc basically takes a device-tree in a given format and outputs a
1032 device-tree in another format. The currently supported formats are:
1037 - "dtb": "blob" format, that is a flattened device-tree block
1039 header all in a binary blob.
1040 - "dts": "source" format. This is a text file containing a
1041 "source" for a device-tree. The format is defined later in this
1043 - "fs" format. This is a representation equivalent to the
1044 output of /proc/device-tree, that is nodes are directories and
1045 properties are files
1050 - "dtb": "blob" format
1051 - "dts": "source" format
1052 - "asm": assembly language file. This is a file that can be
1053 sourced by gas to generate a device-tree "blob". That file can
1054 then simply be added to your Makefile. Additionally, the
1055 assembly file exports some symbols that can be used.
1058 The syntax of the dtc tool is
1060 dtc [-I <input-format>] [-O <output-format>]
1061 [-o output-filename] [-V output_version] input_filename
1064 The "output_version" defines what version of the "blob" format will be
1065 generated. Supported versions are 1,2,3 and 16. The default is
1066 currently version 3 but that may change in the future to version 16.
1068 Additionally, dtc performs various sanity checks on the tree, like the
1069 uniqueness of linux, phandle properties, validity of strings, etc...
1071 The format of the .dts "source" file is "C" like, supports C and C++
1077 The above is the "device-tree" definition. It's the only statement
1078 supported currently at the toplevel.
1081 property1 = "string_value"; /* define a property containing a 0
1085 property2 = <1234abcd>; /* define a property containing a
1086 * numerical 32-bit value (hexadecimal)
1089 property3 = <12345678 12345678 deadbeef>;
1090 /* define a property containing 3
1091 * numerical 32-bit values (cells) in
1094 property4 = [0a 0b 0c 0d de ea ad be ef];
1095 /* define a property whose content is
1096 * an arbitrary array of bytes
1099 childnode@addresss { /* define a child node named "childnode"
1100 * whose unit name is "childnode at
1104 childprop = "hello\n"; /* define a property "childprop" of
1105 * childnode (in this case, a string)
1110 Nodes can contain other nodes etc... thus defining the hierarchical
1111 structure of the tree.
1113 Strings support common escape sequences from C: "\n", "\t", "\r",
1114 "\(octal value)", "\x(hex value)".
1116 It is also suggested that you pipe your source file through cpp (gcc
1117 preprocessor) so you can use #include's, #define for constants, etc...
1119 Finally, various options are planned but not yet implemented, like
1120 automatic generation of phandles, labels (exported to the asm file so
1121 you can point to a property content and change it easily from whatever
1122 you link the device-tree with), label or path instead of numeric value
1123 in some cells to "point" to a node (replaced by a phandle at compile
1124 time), export of reserve map address to the asm file, ability to
1125 specify reserve map content at compile time, etc...
1127 We may provide a .h include file with common definitions of that
1128 proves useful for some properties (like building PCI properties or
1129 interrupt maps) though it may be better to add a notion of struct
1130 definitions to the compiler...
1133 V - Recommendations for a bootloader
1134 ====================================
1137 Here are some various ideas/recommendations that have been proposed
1138 while all this has been defined and implemented.
1140 - The bootloader may want to be able to use the device-tree itself
1141 and may want to manipulate it (to add/edit some properties,
1142 like physical memory size or kernel arguments). At this point, 2
1143 choices can be made. Either the bootloader works directly on the
1144 flattened format, or the bootloader has its own internal tree
1145 representation with pointers (similar to the kernel one) and
1146 re-flattens the tree when booting the kernel. The former is a bit
1147 more difficult to edit/modify, the later requires probably a bit
1148 more code to handle the tree structure. Note that the structure
1149 format has been designed so it's relatively easy to "insert"
1150 properties or nodes or delete them by just memmoving things
1151 around. It contains no internal offsets or pointers for this
1154 - An example of code for iterating nodes & retrieving properties
1155 directly from the flattened tree format can be found in the kernel
1156 file arch/ppc64/kernel/prom.c, look at scan_flat_dt() function,
1157 its usage in early_init_devtree(), and the corresponding various
1158 early_init_dt_scan_*() callbacks. That code can be re-used in a
1159 GPL bootloader, and as the author of that code, I would be happy
1160 to discuss possible free licensing to any vendor who wishes to
1161 integrate all or part of this code into a non-GPL bootloader.
1165 VI - System-on-a-chip devices and nodes
1166 =======================================
1168 Many companies are now starting to develop system-on-a-chip
1169 processors, where the processor core (CPU) and many peripheral devices
1170 exist on a single piece of silicon. For these SOCs, an SOC node
1171 should be used that defines child nodes for the devices that make
1172 up the SOC. While platforms are not required to use this model in
1173 order to boot the kernel, it is highly encouraged that all SOC
1174 implementations define as complete a flat-device-tree as possible to
1175 describe the devices on the SOC. This will allow for the
1176 genericization of much of the kernel code.
1179 1) Defining child nodes of an SOC
1180 ---------------------------------
1182 Each device that is part of an SOC may have its own node entry inside
1183 the SOC node. For each device that is included in the SOC, the unit
1184 address property represents the address offset for this device's
1185 memory-mapped registers in the parent's address space. The parent's
1186 address space is defined by the "ranges" property in the top-level soc
1187 node. The "reg" property for each node that exists directly under the
1188 SOC node should contain the address mapping from the child address space
1189 to the parent SOC address space and the size of the device's
1190 memory-mapped register file.
1192 For many devices that may exist inside an SOC, there are predefined
1193 specifications for the format of the device tree node. All SOC child
1194 nodes should follow these specifications, except where noted in this
1197 See appendix A for an example partial SOC node definition for the
1201 2) Representing devices without a current OF specification
1202 ----------------------------------------------------------
1204 Currently, there are many devices on SOCs that do not have a standard
1205 representation pre-defined as part of the open firmware
1206 specifications, mainly because the boards that contain these SOCs are
1207 not currently booted using open firmware. This section contains
1208 descriptions for the SOC devices for which new nodes have been
1209 defined; this list will expand as more and more SOC-containing
1210 platforms are moved over to use the flattened-device-tree model.
1214 The MDIO is a bus to which the PHY devices are connected. For each
1215 device that exists on this bus, a child node should be created. See
1216 the definition of the PHY node below for an example of how to define
1219 Required properties:
1220 - reg : Offset and length of the register set for the device
1221 - device_type : Should be "mdio"
1222 - compatible : Should define the compatible device type for the
1223 mdio. Currently, this is most likely to be "gianfar"
1229 device_type = "mdio";
1230 compatible = "gianfar";
1238 b) Gianfar-compatible ethernet nodes
1240 Required properties:
1242 - device_type : Should be "network"
1243 - model : Model of the device. Can be "TSEC", "eTSEC", or "FEC"
1244 - compatible : Should be "gianfar"
1245 - reg : Offset and length of the register set for the device
1246 - mac-address : List of bytes representing the ethernet address of
1248 - interrupts : <a b> where a is the interrupt number and b is a
1249 field that represents an encoding of the sense and level
1250 information for the interrupt. This should be encoded based on
1251 the information in section 2) depending on the type of interrupt
1252 controller you have.
1253 - interrupt-parent : the phandle for the interrupt controller that
1254 services interrupts for this device.
1255 - phy-handle : The phandle for the PHY connected to this ethernet
1258 Recommended properties:
1260 - linux,network-index : This is the intended "index" of this
1261 network device. This is used by the bootwrapper to interpret
1262 MAC addresses passed by the firmware when no information other
1263 than indices is available to associate an address with a device.
1264 - phy-connection-type : a string naming the controller/PHY interface type,
1265 i.e., "mii" (default), "rmii", "gmii", "rgmii", "rgmii-id", "sgmii",
1266 "tbi", or "rtbi". This property is only really needed if the connection
1267 is of type "rgmii-id", as all other connection types are detected by
1275 device_type = "network";
1277 compatible = "gianfar";
1279 mac-address = [ 00 E0 0C 00 73 00 ];
1280 interrupts = <d 3 e 3 12 3>;
1281 interrupt-parent = <40000>;
1282 phy-handle = <2452000>
1289 Required properties:
1291 - device_type : Should be "ethernet-phy"
1292 - interrupts : <a b> where a is the interrupt number and b is a
1293 field that represents an encoding of the sense and level
1294 information for the interrupt. This should be encoded based on
1295 the information in section 2) depending on the type of interrupt
1296 controller you have.
1297 - interrupt-parent : the phandle for the interrupt controller that
1298 services interrupts for this device.
1299 - reg : The ID number for the phy, usually a small integer
1300 - linux,phandle : phandle for this node; likely referenced by an
1301 ethernet controller node.
1307 linux,phandle = <2452000>
1308 interrupt-parent = <40000>;
1309 interrupts = <35 1>;
1311 device_type = "ethernet-phy";
1315 d) Interrupt controllers
1317 Some SOC devices contain interrupt controllers that are different
1318 from the standard Open PIC specification. The SOC device nodes for
1319 these types of controllers should be specified just like a standard
1320 OpenPIC controller. Sense and level information should be encoded
1321 as specified in section 2) of this chapter for each device that
1322 specifies an interrupt.
1327 linux,phandle = <40000>;
1328 clock-frequency = <0>;
1329 interrupt-controller;
1330 #address-cells = <0>;
1331 reg = <40000 40000>;
1333 compatible = "chrp,open-pic";
1334 device_type = "open-pic";
1341 Required properties :
1343 - device_type : Should be "i2c"
1344 - reg : Offset and length of the register set for the device
1346 Recommended properties :
1348 - compatible : Should be "fsl-i2c" for parts compatible with
1349 Freescale I2C specifications.
1350 - interrupts : <a b> where a is the interrupt number and b is a
1351 field that represents an encoding of the sense and level
1352 information for the interrupt. This should be encoded based on
1353 the information in section 2) depending on the type of interrupt
1354 controller you have.
1355 - interrupt-parent : the phandle for the interrupt controller that
1356 services interrupts for this device.
1357 - dfsrr : boolean; if defined, indicates that this I2C device has
1358 a digital filter sampling rate register
1359 - fsl5200-clocking : boolean; if defined, indicated that this device
1360 uses the FSL 5200 clocking mechanism.
1365 interrupt-parent = <40000>;
1366 interrupts = <1b 3>;
1368 device_type = "i2c";
1369 compatible = "fsl-i2c";
1374 f) Freescale SOC USB controllers
1376 The device node for a USB controller that is part of a Freescale
1377 SOC is as described in the document "Open Firmware Recommended
1378 Practice : Universal Serial Bus" with the following modifications
1381 Required properties :
1382 - compatible : Should be "fsl-usb2-mph" for multi port host USB
1383 controllers, or "fsl-usb2-dr" for dual role USB controllers
1384 - phy_type : For multi port host USB controllers, should be one of
1385 "ulpi", or "serial". For dual role USB controllers, should be
1386 one of "ulpi", "utmi", "utmi_wide", or "serial".
1387 - reg : Offset and length of the register set for the device
1388 - port0 : boolean; if defined, indicates port0 is connected for
1389 fsl-usb2-mph compatible controllers. Either this property or
1390 "port1" (or both) must be defined for "fsl-usb2-mph" compatible
1392 - port1 : boolean; if defined, indicates port1 is connected for
1393 fsl-usb2-mph compatible controllers. Either this property or
1394 "port0" (or both) must be defined for "fsl-usb2-mph" compatible
1396 - dr_mode : indicates the working mode for "fsl-usb2-dr" compatible
1397 controllers. Can be "host", "peripheral", or "otg". Default to
1398 "host" if not defined for backward compatibility.
1400 Recommended properties :
1401 - interrupts : <a b> where a is the interrupt number and b is a
1402 field that represents an encoding of the sense and level
1403 information for the interrupt. This should be encoded based on
1404 the information in section 2) depending on the type of interrupt
1405 controller you have.
1406 - interrupt-parent : the phandle for the interrupt controller that
1407 services interrupts for this device.
1409 Example multi port host USB controller device node :
1411 device_type = "usb";
1412 compatible = "fsl-usb2-mph";
1414 #address-cells = <1>;
1416 interrupt-parent = <700>;
1417 interrupts = <27 1>;
1423 Example dual role USB controller device node :
1425 device_type = "usb";
1426 compatible = "fsl-usb2-dr";
1428 #address-cells = <1>;
1430 interrupt-parent = <700>;
1431 interrupts = <26 1>;
1437 g) Freescale SOC SEC Security Engines
1439 Required properties:
1441 - device_type : Should be "crypto"
1442 - model : Model of the device. Should be "SEC1" or "SEC2"
1443 - compatible : Should be "talitos"
1444 - reg : Offset and length of the register set for the device
1445 - interrupts : <a b> where a is the interrupt number and b is a
1446 field that represents an encoding of the sense and level
1447 information for the interrupt. This should be encoded based on
1448 the information in section 2) depending on the type of interrupt
1449 controller you have.
1450 - interrupt-parent : the phandle for the interrupt controller that
1451 services interrupts for this device.
1452 - num-channels : An integer representing the number of channels
1454 - channel-fifo-len : An integer representing the number of
1455 descriptor pointers each channel fetch fifo can hold.
1456 - exec-units-mask : The bitmask representing what execution units
1457 (EUs) are available. It's a single 32-bit cell. EU information
1458 should be encoded following the SEC's Descriptor Header Dword
1459 EU_SEL0 field documentation, i.e. as follows:
1461 bit 0 = reserved - should be 0
1462 bit 1 = set if SEC has the ARC4 EU (AFEU)
1463 bit 2 = set if SEC has the DES/3DES EU (DEU)
1464 bit 3 = set if SEC has the message digest EU (MDEU)
1465 bit 4 = set if SEC has the random number generator EU (RNG)
1466 bit 5 = set if SEC has the public key EU (PKEU)
1467 bit 6 = set if SEC has the AES EU (AESU)
1468 bit 7 = set if SEC has the Kasumi EU (KEU)
1470 bits 8 through 31 are reserved for future SEC EUs.
1472 - descriptor-types-mask : The bitmask representing what descriptors
1473 are available. It's a single 32-bit cell. Descriptor type
1474 information should be encoded following the SEC's Descriptor
1475 Header Dword DESC_TYPE field documentation, i.e. as follows:
1477 bit 0 = set if SEC supports the aesu_ctr_nonsnoop desc. type
1478 bit 1 = set if SEC supports the ipsec_esp descriptor type
1479 bit 2 = set if SEC supports the common_nonsnoop desc. type
1480 bit 3 = set if SEC supports the 802.11i AES ccmp desc. type
1481 bit 4 = set if SEC supports the hmac_snoop_no_afeu desc. type
1482 bit 5 = set if SEC supports the srtp descriptor type
1483 bit 6 = set if SEC supports the non_hmac_snoop_no_afeu desc.type
1484 bit 7 = set if SEC supports the pkeu_assemble descriptor type
1485 bit 8 = set if SEC supports the aesu_key_expand_output desc.type
1486 bit 9 = set if SEC supports the pkeu_ptmul descriptor type
1487 bit 10 = set if SEC supports the common_nonsnoop_afeu desc. type
1488 bit 11 = set if SEC supports the pkeu_ptadd_dbl descriptor type
1490 ..and so on and so forth.
1496 device_type = "crypto";
1498 compatible = "talitos";
1499 reg = <30000 10000>;
1500 interrupts = <1d 3>;
1501 interrupt-parent = <40000>;
1503 channel-fifo-len = <18>;
1504 exec-units-mask = <000000fe>;
1505 descriptor-types-mask = <012b0ebf>;
1508 h) Board Control and Status (BCSR)
1510 Required properties:
1512 - device_type : Should be "board-control"
1513 - reg : Offset and length of the register set for the device
1518 device_type = "board-control";
1519 reg = <f8000000 8000>;
1522 i) Freescale QUICC Engine module (QE)
1523 This represents qe module that is installed on PowerQUICC II Pro.
1525 NOTE: This is an interim binding; it should be updated to fit
1526 in with the CPM binding later in this document.
1528 Basically, it is a bus of devices, that could act more or less
1529 as a complete entity (UCC, USB etc ). All of them should be siblings on
1530 the "root" qe node, using the common properties from there.
1531 The description below applies to the qe of MPC8360 and
1532 more nodes and properties would be extended in the future.
1536 Required properties:
1537 - device_type : should be "qe";
1538 - model : precise model of the QE, Can be "QE", "CPM", or "CPM2"
1539 - reg : offset and length of the device registers.
1540 - bus-frequency : the clock frequency for QUICC Engine.
1542 Recommended properties
1543 - brg-frequency : the internal clock source frequency for baud-rate
1548 #address-cells = <1>;
1550 #interrupt-cells = <2>;
1553 ranges = <0 e0100000 00100000>;
1554 reg = <e0100000 480>;
1555 brg-frequency = <0>;
1556 bus-frequency = <179A7B00>;
1560 ii) SPI (Serial Peripheral Interface)
1562 Required properties:
1563 - device_type : should be "spi".
1564 - compatible : should be "fsl_spi".
1565 - mode : the SPI operation mode, it can be "cpu" or "cpu-qe".
1566 - reg : Offset and length of the register set for the device
1567 - interrupts : <a b> where a is the interrupt number and b is a
1568 field that represents an encoding of the sense and level
1569 information for the interrupt. This should be encoded based on
1570 the information in section 2) depending on the type of interrupt
1571 controller you have.
1572 - interrupt-parent : the phandle for the interrupt controller that
1573 services interrupts for this device.
1577 device_type = "spi";
1578 compatible = "fsl_spi";
1580 interrupts = <82 0>;
1581 interrupt-parent = <700>;
1586 iii) USB (Universal Serial Bus Controller)
1588 Required properties:
1589 - device_type : should be "usb".
1590 - compatible : could be "qe_udc" or "fhci-hcd".
1591 - mode : the could be "host" or "slave".
1592 - reg : Offset and length of the register set for the device
1593 - interrupts : <a b> where a is the interrupt number and b is a
1594 field that represents an encoding of the sense and level
1595 information for the interrupt. This should be encoded based on
1596 the information in section 2) depending on the type of interrupt
1597 controller you have.
1598 - interrupt-parent : the phandle for the interrupt controller that
1599 services interrupts for this device.
1603 device_type = "usb";
1604 compatible = "qe_udc";
1606 interrupts = <8b 0>;
1607 interrupt-parent = <700>;
1612 iv) UCC (Unified Communications Controllers)
1614 Required properties:
1615 - device_type : should be "network", "hldc", "uart", "transparent"
1617 - compatible : could be "ucc_geth" or "fsl_atm" and so on.
1618 - model : should be "UCC".
1619 - device-id : the ucc number(1-8), corresponding to UCCx in UM.
1620 - reg : Offset and length of the register set for the device
1621 - interrupts : <a b> where a is the interrupt number and b is a
1622 field that represents an encoding of the sense and level
1623 information for the interrupt. This should be encoded based on
1624 the information in section 2) depending on the type of interrupt
1625 controller you have.
1626 - interrupt-parent : the phandle for the interrupt controller that
1627 services interrupts for this device.
1628 - pio-handle : The phandle for the Parallel I/O port configuration.
1629 - rx-clock : represents the UCC receive clock source.
1630 0x00 : clock source is disabled;
1631 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1632 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1633 - tx-clock: represents the UCC transmit clock source;
1634 0x00 : clock source is disabled;
1635 0x1~0x10 : clock source is BRG1~BRG16 respectively;
1636 0x11~0x28: clock source is QE_CLK1~QE_CLK24 respectively.
1638 Required properties for network device_type:
1639 - mac-address : list of bytes representing the ethernet address.
1640 - phy-handle : The phandle for the PHY connected to this controller.
1642 Recommended properties:
1643 - linux,network-index : This is the intended "index" of this
1644 network device. This is used by the bootwrapper to interpret
1645 MAC addresses passed by the firmware when no information other
1646 than indices is available to associate an address with a device.
1647 - phy-connection-type : a string naming the controller/PHY interface type,
1648 i.e., "mii" (default), "rmii", "gmii", "rgmii", "rgmii-id", "tbi",
1653 device_type = "network";
1654 compatible = "ucc_geth";
1658 interrupts = <a0 0>;
1659 interrupt-parent = <700>;
1660 mac-address = [ 00 04 9f 00 23 23 ];
1663 phy-handle = <212000>;
1664 phy-connection-type = "gmii";
1665 pio-handle = <140001>;
1669 v) Parallel I/O Ports
1671 This node configures Parallel I/O ports for CPUs with QE support.
1672 The node should reside in the "soc" node of the tree. For each
1673 device that using parallel I/O ports, a child node should be created.
1674 See the definition of the Pin configuration nodes below for more
1677 Required properties:
1678 - device_type : should be "par_io".
1679 - reg : offset to the register set and its length.
1680 - num-ports : number of Parallel I/O ports
1685 #address-cells = <1>;
1687 device_type = "par_io";
1694 vi) Pin configuration nodes
1696 Required properties:
1697 - linux,phandle : phandle of this node; likely referenced by a QE
1699 - pio-map : array of pin configurations. Each pin is defined by 6
1700 integers. The six numbers are respectively: port, pin, dir,
1701 open_drain, assignment, has_irq.
1702 - port : port number of the pin; 0-6 represent port A-G in UM.
1703 - pin : pin number in the port.
1704 - dir : direction of the pin, should encode as follows:
1706 0 = The pin is disabled
1707 1 = The pin is an output
1708 2 = The pin is an input
1711 - open_drain : indicates the pin is normal or wired-OR:
1713 0 = The pin is actively driven as an output
1714 1 = The pin is an open-drain driver. As an output, the pin is
1715 driven active-low, otherwise it is three-stated.
1717 - assignment : function number of the pin according to the Pin Assignment
1718 tables in User Manual. Each pin can have up to 4 possible functions in
1719 QE and two options for CPM.
1720 - has_irq : indicates if the pin is used as source of external
1725 linux,phandle = <140001>;
1727 /* port pin dir open_drain assignment has_irq */
1728 0 3 1 0 1 0 /* TxD0 */
1729 0 4 1 0 1 0 /* TxD1 */
1730 0 5 1 0 1 0 /* TxD2 */
1731 0 6 1 0 1 0 /* TxD3 */
1732 1 6 1 0 3 0 /* TxD4 */
1733 1 7 1 0 1 0 /* TxD5 */
1734 1 9 1 0 2 0 /* TxD6 */
1735 1 a 1 0 2 0 /* TxD7 */
1736 0 9 2 0 1 0 /* RxD0 */
1737 0 a 2 0 1 0 /* RxD1 */
1738 0 b 2 0 1 0 /* RxD2 */
1739 0 c 2 0 1 0 /* RxD3 */
1740 0 d 2 0 1 0 /* RxD4 */
1741 1 1 2 0 2 0 /* RxD5 */
1742 1 0 2 0 2 0 /* RxD6 */
1743 1 4 2 0 2 0 /* RxD7 */
1744 0 7 1 0 1 0 /* TX_EN */
1745 0 8 1 0 1 0 /* TX_ER */
1746 0 f 2 0 1 0 /* RX_DV */
1747 0 10 2 0 1 0 /* RX_ER */
1748 0 0 2 0 1 0 /* RX_CLK */
1749 2 9 1 0 3 0 /* GTX_CLK - CLK10 */
1750 2 8 2 0 1 0>; /* GTX125 - CLK9 */
1753 vii) Multi-User RAM (MURAM)
1755 Required properties:
1756 - device_type : should be "muram".
1757 - mode : the could be "host" or "slave".
1758 - ranges : Should be defined as specified in 1) to describe the
1759 translation of MURAM addresses.
1760 - data-only : sub-node which defines the address area under MURAM
1761 bus that can be allocated as data/parameter
1766 device_type = "muram";
1767 ranges = <0 00010000 0000c000>;
1774 j) CFI or JEDEC memory-mapped NOR flash
1776 Flash chips (Memory Technology Devices) are often used for solid state
1777 file systems on embedded devices.
1779 - compatible : should contain the specific model of flash chip(s)
1780 used, if known, followed by either "cfi-flash" or "jedec-flash"
1781 - reg : Address range of the flash chip
1782 - bank-width : Width (in bytes) of the flash bank. Equal to the
1783 device width times the number of interleaved chips.
1784 - device-width : (optional) Width of a single flash chip. If
1785 omitted, assumed to be equal to 'bank-width'.
1786 - #address-cells, #size-cells : Must be present if the flash has
1787 sub-nodes representing partitions (see below). In this case
1788 both #address-cells and #size-cells must be equal to 1.
1790 For JEDEC compatible devices, the following additional properties
1793 - vendor-id : Contains the flash chip's vendor id (1 byte).
1794 - device-id : Contains the flash chip's device id (1 byte).
1796 In addition to the information on the flash bank itself, the
1797 device tree may optionally contain additional information
1798 describing partitions of the flash address space. This can be
1799 used on platforms which have strong conventions about which
1800 portions of the flash are used for what purposes, but which don't
1801 use an on-flash partition table such as RedBoot.
1803 Each partition is represented as a sub-node of the flash device.
1804 Each node's name represents the name of the corresponding
1805 partition of the flash device.
1808 - reg : The partition's offset and size within the flash bank.
1809 - label : (optional) The label / name for this flash partition.
1810 If omitted, the label is taken from the node name (excluding
1812 - read-only : (optional) This parameter, if present, is a hint to
1813 Linux that this flash partition should only be mounted
1814 read-only. This is usually used for flash partitions
1815 containing early-boot firmware images or data which should not
1821 compatible = "amd,am29lv128ml", "cfi-flash";
1822 reg = <ff000000 01000000>;
1825 #address-cells = <1>;
1833 reg = <f80000 80000>;
1838 k) Global Utilities Block
1840 The global utilities block controls power management, I/O device
1841 enabling, power-on-reset configuration monitoring, general-purpose
1842 I/O signal configuration, alternate function selection for multiplexed
1843 signals, and clock control.
1845 Required properties:
1847 - compatible : Should define the compatible device type for
1849 - reg : Offset and length of the register set for the device.
1851 Recommended properties:
1853 - fsl,has-rstcr : Indicates that the global utilities register set
1854 contains a functioning "reset control register" (i.e. the board
1855 is wired to reset upon setting the HRESET_REQ bit in this register).
1859 global-utilities@e0000 { /* global utilities block */
1860 compatible = "fsl,mpc8548-guts";
1865 l) Freescale Communications Processor Module
1867 NOTE: This is an interim binding, and will likely change slightly,
1868 as more devices are supported. The QE bindings especially are
1874 - compatible : "fsl,cpm1", "fsl,cpm2", or "fsl,qe".
1875 - reg : A 48-byte region beginning with CPCR.
1879 #address-cells = <1>;
1881 #interrupt-cells = <2>;
1882 compatible = "fsl,mpc8272-cpm", "fsl,cpm2";
1886 ii) Properties common to mulitple CPM/QE devices
1888 - fsl,cpm-command : This value is ORed with the opcode and command flag
1889 to specify the device on which a CPM command operates.
1891 - fsl,cpm-brg : Indicates which baud rate generator the device
1892 is associated with. If absent, an unused BRG
1893 should be dynamically allocated. If zero, the
1894 device uses an external clock rather than a BRG.
1896 - reg : Unless otherwise specified, the first resource represents the
1897 scc/fcc/ucc registers, and the second represents the device's
1898 parameter RAM region (if it has one).
1902 Currently defined compatibles:
1912 device_type = "serial";
1913 compatible = "fsl,mpc8272-scc-uart",
1914 "fsl,cpm2-scc-uart";
1915 reg = <11a00 20 8000 100>;
1916 interrupts = <28 8>;
1917 interrupt-parent = <&PIC>;
1919 fsl,cpm-command = <00800000>;
1924 Currently defined compatibles:
1928 - fsl,cpm2-fcc-enet (third resource is GFEMR)
1934 device_type = "network";
1935 compatible = "fsl,mpc8272-fcc-enet",
1936 "fsl,cpm2-fcc-enet";
1937 reg = <11300 20 8400 100 11390 1>;
1938 local-mac-address = [ 00 00 00 00 00 00 ];
1939 interrupts = <20 8>;
1940 interrupt-parent = <&PIC>;
1941 phy-handle = <&PHY0>;
1942 linux,network-index = <0>;
1943 fsl,cpm-command = <12000300>;
1948 Currently defined compatibles:
1949 fsl,pq1-fec-mdio (reg is same as first resource of FEC device)
1950 fsl,cpm2-mdio-bitbang (reg is port C registers)
1952 Properties for fsl,cpm2-mdio-bitbang:
1953 fsl,mdio-pin : pin of port C controlling mdio data
1954 fsl,mdc-pin : pin of port C controlling mdio clock
1959 device_type = "mdio";
1960 compatible = "fsl,mpc8272ads-mdio-bitbang",
1961 "fsl,mpc8272-mdio-bitbang",
1962 "fsl,cpm2-mdio-bitbang";
1964 #address-cells = <1>;
1966 fsl,mdio-pin = <12>;
1970 v) Baud Rate Generators
1972 Currently defined compatibles:
1978 - reg : There may be an arbitrary number of reg resources; BRG
1979 numbers are assigned to these in order.
1980 - clock-frequency : Specifies the base frequency driving
1986 compatible = "fsl,mpc8272-brg",
1989 reg = <119f0 10 115f0 10>;
1990 clock-frequency = <d#25000000>;
1993 vi) Interrupt Controllers
1995 Currently defined compatibles:
1997 - only one interrupt cell
2000 - second interrupt cell is level/sense:
2006 interrupt-controller@10c00 {
2007 #interrupt-cells = <2>;
2008 interrupt-controller;
2010 compatible = "mpc8272-pic", "fsl,cpm2-pic";
2013 vii) USB (Universal Serial Bus Controller)
2016 - compatible : "fsl,cpm1-usb", "fsl,cpm2-usb", "fsl,qe-usb"
2020 #address-cells = <1>;
2022 compatible = "fsl,cpm2-usb";
2023 reg = <11b60 18 8b00 100>;
2025 interrupt-parent = <&PIC>;
2026 fsl,cpm-command = <2e600000>;
2029 viii) Multi-User RAM (MURAM)
2031 The multi-user/dual-ported RAM is expressed as a bus under the CPM node.
2033 Ranges must be set up subject to the following restrictions:
2035 - Children's reg nodes must be offsets from the start of all muram, even
2036 if the user-data area does not begin at zero.
2037 - If multiple range entries are used, the difference between the parent
2038 address and the child address must be the same in all, so that a single
2039 mapping can cover them all while maintaining the ability to determine
2040 CPM-side offsets with pointer subtraction. It is recommended that
2041 multiple range entries not be used.
2042 - A child address of zero must be translatable, even if no reg resources
2045 A child "data" node must exist, compatible with "fsl,cpm-muram-data", to
2046 indicate the portion of muram that is usable by the OS for arbitrary
2047 purposes. The data node may have an arbitrary number of reg resources,
2048 all of which contribute to the allocatable muram pool.
2050 Example, based on mpc8272:
2053 #address-cells = <1>;
2055 ranges = <0 0 10000>;
2058 compatible = "fsl,cpm-muram-data";
2059 reg = <0 2000 9800 800>;
2063 m) Chipselect/Local Bus
2066 - name : Should be localbus
2067 - #address-cells : Should be either two or three. The first cell is the
2068 chipselect number, and the remaining cells are the
2069 offset into the chipselect.
2070 - #size-cells : Either one or two, depending on how large each chipselect
2072 - ranges : Each range corresponds to a single chipselect, and cover
2073 the entire access window as configured.
2077 compatible = "fsl,mpc8272ads-localbus",
2078 "fsl,mpc8272-localbus",
2080 #address-cells = <2>;
2082 reg = <f0010100 40>;
2084 ranges = <0 0 fe000000 02000000
2085 1 0 f4500000 00008000>;
2088 compatible = "jedec-flash";
2089 reg = <0 0 2000000>;
2096 compatible = "fsl,mpc8272ads-bcsr";
2101 n) 4xx/Axon EMAC ethernet nodes
2103 The EMAC ethernet controller in IBM and AMCC 4xx chips, and also
2104 the Axon bridge. To operate this needs to interact with a ths
2105 special McMAL DMA controller, and sometimes an RGMII or ZMII
2106 interface. In addition to the nodes and properties described
2107 below, the node for the OPB bus on which the EMAC sits must have a
2108 correct clock-frequency property.
2110 i) The EMAC node itself
2112 Required properties:
2113 - device_type : "network"
2115 - compatible : compatible list, contains 2 entries, first is
2116 "ibm,emac-CHIP" where CHIP is the host ASIC (440gx,
2117 405gp, Axon) and second is either "ibm,emac" or
2118 "ibm,emac4". For Axon, thus, we have: "ibm,emac-axon",
2120 - interrupts : <interrupt mapping for EMAC IRQ and WOL IRQ>
2121 - interrupt-parent : optional, if needed for interrupt mapping
2122 - reg : <registers mapping>
2123 - local-mac-address : 6 bytes, MAC address
2124 - mal-device : phandle of the associated McMAL node
2125 - mal-tx-channel : 1 cell, index of the tx channel on McMAL associated
2127 - mal-rx-channel : 1 cell, index of the rx channel on McMAL associated
2129 - cell-index : 1 cell, hardware index of the EMAC cell on a given
2130 ASIC (typically 0x0 and 0x1 for EMAC0 and EMAC1 on
2132 - max-frame-size : 1 cell, maximum frame size supported in bytes
2133 - rx-fifo-size : 1 cell, Rx fifo size in bytes for 10 and 100 Mb/sec
2136 - tx-fifo-size : 1 cell, Tx fifo size in bytes for 10 and 100 Mb/sec
2139 - fifo-entry-size : 1 cell, size of a fifo entry (used to calculate
2141 For Axon, 0x00000010
2142 - mal-burst-size : 1 cell, MAL burst size (used to calculate thresholds)
2144 For Axon, 0x00000100 (I think ...)
2145 - phy-mode : string, mode of operations of the PHY interface.
2146 Supported values are: "mii", "rmii", "smii", "rgmii",
2147 "tbi", "gmii", rtbi", "sgmii".
2148 For Axon on CAB, it is "rgmii"
2149 - mdio-device : 1 cell, required iff using shared MDIO registers
2150 (440EP). phandle of the EMAC to use to drive the
2151 MDIO lines for the PHY used by this EMAC.
2152 - zmii-device : 1 cell, required iff connected to a ZMII. phandle of
2153 the ZMII device node
2154 - zmii-channel : 1 cell, required iff connected to a ZMII. Which ZMII
2155 channel or 0xffffffff if ZMII is only used for MDIO.
2156 - rgmii-device : 1 cell, required iff connected to an RGMII. phandle
2157 of the RGMII device node.
2158 For Axon: phandle of plb5/plb4/opb/rgmii
2159 - rgmii-channel : 1 cell, required iff connected to an RGMII. Which
2160 RGMII channel is used by this EMAC.
2161 Fox Axon: present, whatever value is appropriate for each
2162 EMAC, that is the content of the current (bogus) "phy-port"
2165 Recommended properties:
2166 - linux,network-index : This is the intended "index" of this
2167 network device. This is used by the bootwrapper to interpret
2168 MAC addresses passed by the firmware when no information other
2169 than indices is available to associate an address with a device.
2171 Optional properties:
2172 - phy-address : 1 cell, optional, MDIO address of the PHY. If absent,
2173 a search is performed.
2174 - phy-map : 1 cell, optional, bitmap of addresses to probe the PHY
2175 for, used if phy-address is absent. bit 0x00000001 is
2177 For Axon it can be absent, thouugh my current driver
2178 doesn't handle phy-address yet so for now, keep
2180 - rx-fifo-size-gige : 1 cell, Rx fifo size in bytes for 1000 Mb/sec
2181 operations (if absent the value is the same as
2182 rx-fifo-size). For Axon, either absent or 2048.
2183 - tx-fifo-size-gige : 1 cell, Tx fifo size in bytes for 1000 Mb/sec
2184 operations (if absent the value is the same as
2185 tx-fifo-size). For Axon, either absent or 2048.
2186 - tah-device : 1 cell, optional. If connected to a TAH engine for
2187 offload, phandle of the TAH device node.
2188 - tah-channel : 1 cell, optional. If appropriate, channel used on the
2193 EMAC0: ethernet@40000800 {
2194 linux,network-index = <0>;
2195 device_type = "network";
2196 compatible = "ibm,emac-440gp", "ibm,emac";
2197 interrupt-parent = <&UIC1>;
2198 interrupts = <1c 4 1d 4>;
2199 reg = <40000800 70>;
2200 local-mac-address = [00 04 AC E3 1B 1E];
2201 mal-device = <&MAL0>;
2202 mal-tx-channel = <0 1>;
2203 mal-rx-channel = <0>;
2205 max-frame-size = <5dc>;
2206 rx-fifo-size = <1000>;
2207 tx-fifo-size = <800>;
2209 phy-map = <00000001>;
2210 zmii-device = <&ZMII0>;
2216 Required properties:
2217 - device_type : "dma-controller"
2218 - compatible : compatible list, containing 2 entries, first is
2219 "ibm,mcmal-CHIP" where CHIP is the host ASIC (like
2220 emac) and the second is either "ibm,mcmal" or
2222 For Axon, "ibm,mcmal-axon","ibm,mcmal2"
2223 - interrupts : <interrupt mapping for the MAL interrupts sources:
2224 5 sources: tx_eob, rx_eob, serr, txde, rxde>.
2225 For Axon: This is _different_ from the current
2226 firmware. We use the "delayed" interrupts for txeob
2227 and rxeob. Thus we end up with mapping those 5 MPIC
2228 interrupts, all level positive sensitive: 10, 11, 32,
2230 - dcr-reg : < DCR registers range >
2231 - dcr-parent : if needed for dcr-reg
2232 - num-tx-chans : 1 cell, number of Tx channels
2233 - num-rx-chans : 1 cell, number of Rx channels
2237 Required properties:
2238 - compatible : compatible list, containing 2 entries, first is
2239 "ibm,zmii-CHIP" where CHIP is the host ASIC (like
2240 EMAC) and the second is "ibm,zmii".
2241 For Axon, there is no ZMII node.
2242 - reg : <registers mapping>
2246 Required properties:
2247 - compatible : compatible list, containing 2 entries, first is
2248 "ibm,rgmii-CHIP" where CHIP is the host ASIC (like
2249 EMAC) and the second is "ibm,rgmii".
2250 For Axon, "ibm,rgmii-axon","ibm,rgmii"
2251 - reg : <registers mapping>
2252 - revision : as provided by the RGMII new version register if
2254 For Axon: 0x0000012a
2258 The Xilinx EDK toolchain ships with a set of IP cores (devices) for use
2259 in Xilinx Spartan and Virtex FPGAs. The devices cover the whole range
2260 of standard device types (network, serial, etc.) and miscellanious
2261 devices (gpio, LCD, spi, etc). Also, since these devices are
2262 implemented within the fpga fabric every instance of the device can be
2263 synthesised with different options that change the behaviour.
2265 Each IP-core has a set of parameters which the FPGA designer can use to
2266 control how the core is synthesized. Historically, the EDK tool would
2267 extract the device parameters relevant to device drivers and copy them
2268 into an 'xparameters.h' in the form of #define symbols. This tells the
2269 device drivers how the IP cores are configured, but it requres the kernel
2270 to be recompiled every time the FPGA bitstream is resynthesized.
2272 The new approach is to export the parameters into the device tree and
2273 generate a new device tree each time the FPGA bitstream changes. The
2274 parameters which used to be exported as #defines will now become
2275 properties of the device node. In general, device nodes for IP-cores
2276 will take the following form:
2278 (name)@(base-address) {
2279 compatible = "xlnx,(ip-core-name)-(HW_VER)"
2280 [, (list of compatible devices), ...];
2281 reg = <(baseaddr) (size)>;
2282 interrupt-parent = <&interrupt-controller-phandle>;
2283 interrupts = < ... >;
2284 xlnx,(parameter1) = "(string-value)";
2285 xlnx,(parameter2) = <(int-value)>;
2288 (ip-core-name): the name of the ip block (given after the BEGIN
2289 directive in system.mhs). Should be in lowercase
2290 and all underscores '_' converted to dashes '-'.
2291 (name): is derived from the "PARAMETER INSTANCE" value.
2292 (parameter#): C_* parameters from system.mhs. The C_ prefix is
2293 dropped from the parameter name, the name is converted
2294 to lowercase and all underscore '_' characters are
2295 converted to dashes '-'.
2296 (baseaddr): the C_BASEADDR parameter.
2297 (HW_VER): from the HW_VER parameter.
2298 (size): equals C_HIGHADDR - C_BASEADDR + 1
2300 Typically, the compatible list will include the exact IP core version
2301 followed by an older IP core version which implements the same
2302 interface or any other device with the same interface.
2304 'reg', 'interrupt-parent' and 'interrupts' are all optional properties.
2306 For example, the following block from system.mhs:
2309 PARAMETER INSTANCE = opb_uartlite_0
2310 PARAMETER HW_VER = 1.00.b
2311 PARAMETER C_BAUDRATE = 115200
2312 PARAMETER C_DATA_BITS = 8
2313 PARAMETER C_ODD_PARITY = 0
2314 PARAMETER C_USE_PARITY = 0
2315 PARAMETER C_CLK_FREQ = 50000000
2316 PARAMETER C_BASEADDR = 0xEC100000
2317 PARAMETER C_HIGHADDR = 0xEC10FFFF
2318 BUS_INTERFACE SOPB = opb_7
2319 PORT OPB_Clk = CLK_50MHz
2320 PORT Interrupt = opb_uartlite_0_Interrupt
2321 PORT RX = opb_uartlite_0_RX
2322 PORT TX = opb_uartlite_0_TX
2323 PORT OPB_Rst = sys_bus_reset_0
2326 becomes the following device tree node:
2328 opb-uartlite-0@ec100000 {
2329 device_type = "serial";
2330 compatible = "xlnx,opb-uartlite-1.00.b";
2331 reg = <ec100000 10000>;
2332 interrupt-parent = <&opb-intc>;
2333 interrupts = <1 0>; // got this from the opb_intc parameters
2334 current-speed = <d#115200>; // standard serial device prop
2335 clock-frequency = <d#50000000>; // standard serial device prop
2336 xlnx,data-bits = <8>;
2337 xlnx,odd-parity = <0>;
2338 xlnx,use-parity = <0>;
2341 Some IP cores actually implement 2 or more logical devices. In this case,
2342 the device should still describe the whole IP core with a single node
2343 and add a child node for each logical device. The ranges property can
2344 be used to translate from parent IP-core to the registers of each device.
2345 (Note: this makes the assumption that both logical devices have the same
2346 bus binding. If this is not true, then separate nodes should be used for
2347 each logical device). The 'cell-index' property can be used to enumerate
2348 logical devices within an IP core. For example, the following is the
2349 system.mhs entry for the dual ps2 controller found on the ml403 reference
2352 BEGIN opb_ps2_dual_ref
2353 PARAMETER INSTANCE = opb_ps2_dual_ref_0
2354 PARAMETER HW_VER = 1.00.a
2355 PARAMETER C_BASEADDR = 0xA9000000
2356 PARAMETER C_HIGHADDR = 0xA9001FFF
2357 BUS_INTERFACE SOPB = opb_v20_0
2358 PORT Sys_Intr1 = ps2_1_intr
2359 PORT Sys_Intr2 = ps2_2_intr
2360 PORT Clkin1 = ps2_clk_rx_1
2361 PORT Clkin2 = ps2_clk_rx_2
2362 PORT Clkpd1 = ps2_clk_tx_1
2363 PORT Clkpd2 = ps2_clk_tx_2
2364 PORT Rx1 = ps2_d_rx_1
2365 PORT Rx2 = ps2_d_rx_2
2366 PORT Txpd1 = ps2_d_tx_1
2367 PORT Txpd2 = ps2_d_tx_2
2370 It would result in the following device tree nodes:
2372 opb_ps2_dual_ref_0@a9000000 {
2373 ranges = <0 a9000000 2000>;
2374 // If this device had extra parameters, then they would
2377 compatible = "xlnx,opb-ps2-dual-ref-1.00.a";
2379 interrupt-parent = <&opb-intc>;
2384 compatible = "xlnx,opb-ps2-dual-ref-1.00.a";
2386 interrupt-parent = <&opb-intc>;
2392 Also, the system.mhs file defines bus attachments from the processor
2393 to the devices. The device tree structure should reflect the bus
2394 attachments. Again an example; this system.mhs fragment:
2396 BEGIN ppc405_virtex4
2397 PARAMETER INSTANCE = ppc405_0
2398 PARAMETER HW_VER = 1.01.a
2399 BUS_INTERFACE DPLB = plb_v34_0
2400 BUS_INTERFACE IPLB = plb_v34_0
2404 PARAMETER INSTANCE = opb_intc_0
2405 PARAMETER HW_VER = 1.00.c
2406 PARAMETER C_BASEADDR = 0xD1000FC0
2407 PARAMETER C_HIGHADDR = 0xD1000FDF
2408 BUS_INTERFACE SOPB = opb_v20_0
2412 PARAMETER INSTANCE = opb_uart16550_0
2413 PARAMETER HW_VER = 1.00.d
2414 PARAMETER C_BASEADDR = 0xa0000000
2415 PARAMETER C_HIGHADDR = 0xa0001FFF
2416 BUS_INTERFACE SOPB = opb_v20_0
2420 PARAMETER INSTANCE = plb_v34_0
2421 PARAMETER HW_VER = 1.02.a
2424 BEGIN plb_bram_if_cntlr
2425 PARAMETER INSTANCE = plb_bram_if_cntlr_0
2426 PARAMETER HW_VER = 1.00.b
2427 PARAMETER C_BASEADDR = 0xFFFF0000
2428 PARAMETER C_HIGHADDR = 0xFFFFFFFF
2429 BUS_INTERFACE SPLB = plb_v34_0
2432 BEGIN plb2opb_bridge
2433 PARAMETER INSTANCE = plb2opb_bridge_0
2434 PARAMETER HW_VER = 1.01.a
2435 PARAMETER C_RNG0_BASEADDR = 0x20000000
2436 PARAMETER C_RNG0_HIGHADDR = 0x3FFFFFFF
2437 PARAMETER C_RNG1_BASEADDR = 0x60000000
2438 PARAMETER C_RNG1_HIGHADDR = 0x7FFFFFFF
2439 PARAMETER C_RNG2_BASEADDR = 0x80000000
2440 PARAMETER C_RNG2_HIGHADDR = 0xBFFFFFFF
2441 PARAMETER C_RNG3_BASEADDR = 0xC0000000
2442 PARAMETER C_RNG3_HIGHADDR = 0xDFFFFFFF
2443 BUS_INTERFACE SPLB = plb_v34_0
2444 BUS_INTERFACE MOPB = opb_v20_0
2447 Gives this device tree (some properties removed for clarity):
2450 #address-cells = <1>;
2452 device_type = "ibm,plb";
2453 ranges; // 1:1 translation
2455 plb-bram-if-cntrl-0@ffff0000 {
2456 reg = <ffff0000 10000>;
2460 #address-cells = <1>;
2462 ranges = <20000000 20000000 20000000
2463 60000000 60000000 20000000
2464 80000000 80000000 40000000
2465 c0000000 c0000000 20000000>;
2467 opb-uart16550-0@a0000000 {
2468 reg = <a00000000 2000>;
2471 opb-intc-0@d1000fc0 {
2472 reg = <d1000fc0 20>;
2477 That covers the general approach to binding xilinx IP cores into the
2478 device tree. The following are bindings for specific devices:
2480 i) Xilinx ML300 Framebuffer
2482 Simple framebuffer device from the ML300 reference design (also on the
2483 ML403 reference design as well as others).
2485 Optional properties:
2486 - resolution = <xres yres> : pixel resolution of framebuffer. Some
2487 implementations use a different resolution.
2488 Default is <d#640 d#480>
2489 - virt-resolution = <xvirt yvirt> : Size of framebuffer in memory.
2490 Default is <d#1024 d#480>.
2491 - rotate-display (empty) : rotate display 180 degrees.
2493 ii) Xilinx SystemACE
2495 The Xilinx SystemACE device is used to program FPGAs from an FPGA
2496 bitstream stored on a CF card. It can also be used as a generic CF
2499 Optional properties:
2500 - 8-bit (empty) : Set this property for SystemACE in 8 bit mode
2502 iii) Xilinx EMAC and Xilinx TEMAC
2504 Xilinx Ethernet devices. In addition to general xilinx properties
2505 listed above, nodes for these devices should include a phy-handle
2506 property, and may include other common network device properties
2507 like local-mac-address.
2511 Xilinx uartlite devices are simple fixed speed serial ports.
2514 - current-speed : Baud rate of uartlite
2516 More devices will be defined as this spec matures.
2518 VII - Specifying interrupt information for devices
2519 ===================================================
2521 The device tree represents the busses and devices of a hardware
2522 system in a form similar to the physical bus topology of the
2525 In addition, a logical 'interrupt tree' exists which represents the
2526 hierarchy and routing of interrupts in the hardware.
2528 The interrupt tree model is fully described in the
2529 document "Open Firmware Recommended Practice: Interrupt
2530 Mapping Version 0.9". The document is available at:
2531 <http://playground.sun.com/1275/practice>.
2533 1) interrupts property
2534 ----------------------
2536 Devices that generate interrupts to a single interrupt controller
2537 should use the conventional OF representation described in the
2538 OF interrupt mapping documentation.
2540 Each device which generates interrupts must have an 'interrupt'
2541 property. The interrupt property value is an arbitrary number of
2542 of 'interrupt specifier' values which describe the interrupt or
2543 interrupts for the device.
2545 The encoding of an interrupt specifier is determined by the
2546 interrupt domain in which the device is located in the
2547 interrupt tree. The root of an interrupt domain specifies in
2548 its #interrupt-cells property the number of 32-bit cells
2549 required to encode an interrupt specifier. See the OF interrupt
2550 mapping documentation for a detailed description of domains.
2552 For example, the binding for the OpenPIC interrupt controller
2553 specifies an #interrupt-cells value of 2 to encode the interrupt
2554 number and level/sense information. All interrupt children in an
2555 OpenPIC interrupt domain use 2 cells per interrupt in their interrupts
2558 The PCI bus binding specifies a #interrupt-cell value of 1 to encode
2559 which interrupt pin (INTA,INTB,INTC,INTD) is used.
2561 2) interrupt-parent property
2562 ----------------------------
2564 The interrupt-parent property is specified to define an explicit
2565 link between a device node and its interrupt parent in
2566 the interrupt tree. The value of interrupt-parent is the
2567 phandle of the parent node.
2569 If the interrupt-parent property is not defined for a node, it's
2570 interrupt parent is assumed to be an ancestor in the node's
2571 _device tree_ hierarchy.
2573 3) OpenPIC Interrupt Controllers
2574 --------------------------------
2576 OpenPIC interrupt controllers require 2 cells to encode
2577 interrupt information. The first cell defines the interrupt
2578 number. The second cell defines the sense and level
2581 Sense and level information should be encoded as follows:
2583 0 = low to high edge sensitive type enabled
2584 1 = active low level sensitive type enabled
2585 2 = active high level sensitive type enabled
2586 3 = high to low edge sensitive type enabled
2588 4) ISA Interrupt Controllers
2589 ----------------------------
2591 ISA PIC interrupt controllers require 2 cells to encode
2592 interrupt information. The first cell defines the interrupt
2593 number. The second cell defines the sense and level
2596 ISA PIC interrupt controllers should adhere to the ISA PIC
2597 encodings listed below:
2599 0 = active low level sensitive type enabled
2600 1 = active high level sensitive type enabled
2601 2 = high to low edge sensitive type enabled
2602 3 = low to high edge sensitive type enabled
2605 Appendix A - Sample SOC node for MPC8540
2606 ========================================
2608 Note that the #address-cells and #size-cells for the SoC node
2609 in this example have been explicitly listed; these are likely
2610 not necessary as they are usually the same as the root node.
2613 #address-cells = <1>;
2615 #interrupt-cells = <2>;
2616 device_type = "soc";
2617 ranges = <00000000 e0000000 00100000>
2618 reg = <e0000000 00003000>;
2619 bus-frequency = <0>;
2623 device_type = "mdio";
2624 compatible = "gianfar";
2627 linux,phandle = <2452000>
2628 interrupt-parent = <40000>;
2629 interrupts = <35 1>;
2631 device_type = "ethernet-phy";
2635 linux,phandle = <2452001>
2636 interrupt-parent = <40000>;
2637 interrupts = <35 1>;
2639 device_type = "ethernet-phy";
2643 linux,phandle = <2452002>
2644 interrupt-parent = <40000>;
2645 interrupts = <35 1>;
2647 device_type = "ethernet-phy";
2654 device_type = "network";
2656 compatible = "gianfar";
2658 mac-address = [ 00 E0 0C 00 73 00 ];
2659 interrupts = <d 3 e 3 12 3>;
2660 interrupt-parent = <40000>;
2661 phy-handle = <2452000>;
2665 #address-cells = <1>;
2667 device_type = "network";
2669 compatible = "gianfar";
2671 mac-address = [ 00 E0 0C 00 73 01 ];
2672 interrupts = <13 3 14 3 18 3>;
2673 interrupt-parent = <40000>;
2674 phy-handle = <2452001>;
2678 #address-cells = <1>;
2680 device_type = "network";
2682 compatible = "gianfar";
2684 mac-address = [ 00 E0 0C 00 73 02 ];
2685 interrupts = <19 3>;
2686 interrupt-parent = <40000>;
2687 phy-handle = <2452002>;
2691 device_type = "serial";
2692 compatible = "ns16550";
2694 clock-frequency = <0>;
2695 interrupts = <1a 3>;
2696 interrupt-parent = <40000>;
2700 linux,phandle = <40000>;
2701 clock-frequency = <0>;
2702 interrupt-controller;
2703 #address-cells = <0>;
2704 reg = <40000 40000>;
2706 compatible = "chrp,open-pic";
2707 device_type = "open-pic";
2712 interrupt-parent = <40000>;
2713 interrupts = <1b 3>;
2715 device_type = "i2c";
2716 compatible = "fsl-i2c";
git.openpandora.org / pandora-kernel.git / blob