1 .. SPDX-License-Identifier: GPL-2.0
2
3 DeviceTree Booting
4 ------------------
5
6 During the development of the Linux/ppc64 kern
7 addition of new platform types outside of the
8 was decided to enforce some strict rules regar
9 bootloader <-> kernel interfaces, in order to
10 become the ppc32 kernel entry point and the wa
11 to the kernel. The legacy iSeries platform bre
12 this scheme, but no new board support will be
13 doesn't follow them properly. In addition, si
14 merged architecture for ppc32 and ppc64, new 3
15 platforms which move into arch/powerpc will be
16 well.
17
18 The main requirement that will be defined in m
19 of a device-tree whose format is defined after
20 However, in order to make life easier to embed
21 doesn't require the device-tree to represent e
22 requires some nodes and properties to be prese
23 not require you to create a node for every PCI
24 requirement to have a node for PCI host bridge
25 routing information and memory/IO ranges, amon
26 to define nodes for on chip devices and other
27 in an existing OF specification. This creates
28 kernel can then probe those and match drivers
29 code all sorts of tables. It also makes it mor
30 minor hardware upgrades without significantly
31 cluttering it with special cases.
32
33
34 Entry point
35 ~~~~~~~~~~~
36
37 There is one single entry point to the kernel,
38 of the kernel image. That entry point supports
39 conventions:
40
41 a) Boot from Open Firmware. If your fi
42 with Open Firmware (IEEE 1275) or prov
43 client interface API (support for "int
44 forth words isn't required), you can e
45
46 r5 : OF callback pointer as defi
47 bindings to powerpc. Only the 32
48 is currently supported
49
50 r3, r4 : address & length of an
51
52 The MMU is either on or off; the
53 trampoline located in arch/power
54 extract the device-tree and othe
55 firmware and build a flattened d
56 in b). prom_init() will then re-
57 the second method. This trampoli
58 context of the firmware, which i
59 exceptions during that time.
60
61 b) Direct entry with a flattened devic
62 point is called by a) after the OF tra
63 called directly by a bootloader that d
64 Firmware client interface. It is also
65 implement "hot" booting of a new kerne
66 running one. This method is what I wil
67 details in this document, as method a)
68 Firmware, and thus should be implement
69 various standard documents defining it
70 PowerPC platform. The entry point defi
71
72 r3 : physical pointer to the d
73 (defined in chapter II) in RAM
74
75 r4 : physical pointer to the k
76 used by the assembly code to p
77 in case you are entering the k
78 and a non-1:1 mapping.
79
80 r5 : NULL (as to differentiate
81
82 Note about SMP entry: Either your firmware put
83 CPUs in some sleep loop or spin loop in ROM wh
84 them out via a soft reset or some other means,
85 you don't need to care, or you'll have to ente
86 with all CPUs. The way to do that with method
87 described in a later revision of this document
88
89 Board supports (platforms) are not exclusive c
90 arbitrary set of board supports can be built i
91 image. The kernel will "know" what set of func
92 given platform based on the content of the dev
93 should:
94
95 a) add your platform support as a _boo
96 arch/powerpc/Kconfig, following the ex
97 PPC_PMAC and PPC_MAPLE. The latter is
98 example of a board support to start fr
99
100 b) create your main platform file as
101 "arch/powerpc/platforms/myplatform/myb
102 to the Makefile under the condition of
103 option. This file will define a struct
104 containing the various callbacks that
105 use to get to your platform specific c
106
107 A kernel image may support multiple platforms,
108 platforms feature the same core architecture.
109 cannot support both configurations with Book E
110 with classic Powerpc architectures.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.