3 Writing DeviceTree Bindings in json-schema
4 ==========================================
6 Devicetree bindings are written using json-schema vocabulary. Schema files are
7 written in a JSON compatible subset of YAML. YAML is used instead of JSON as it
8 considered more human readable and has some advantages such as allowing
9 comments (Prefixed with '#').
14 Each schema doc is a structured json-schema which is defined by a set of
15 top-level properties. Generally, there is one binding defined per file. The
16 top-level json-schema properties used are:
19 A json-schema unique identifier string. The string must be a valid
20 URI typically containing the binding's filename and path. For DT schema, it must
21 begin with "http://devicetree.org/schemas/". The URL is used in constructing
22 references to other files specified in schema "$ref" properties. A $ref values
23 with a leading '/' will have the hostname prepended. A $ref value a relative
24 path or filename only will be prepended with the hostname and path components
25 of the current schema file's '$id' value. A URL is used even for local files,
26 but there may not actually be files present at those locations.
29 Indicates the meta-schema the schema file adheres to.
32 A one line description on the contents of the binding schema.
35 A DT specific property. Contains a list of email address(es)
36 for maintainers of this binding.
39 Optional. A multi-line text block containing any detailed
40 information about this binding. It should contain things such as what the block
41 or device does, standards the device conforms to, and links to datasheets for
45 Optional. A json-schema used to match nodes for applying the
46 schema. By default without 'select', nodes are matched against their possible
47 compatible string values or node name. Most bindings should not need select.
50 Optional. A list of other schemas to include. This is used to
51 include other schemas the binding conforms to. This may be schemas for a
52 particular class of devices such as I2C or SPI controllers.
55 A set of sub-schema defining all the DT properties for the
56 binding. The exact schema syntax depends on whether properties are known,
57 common properties (e.g. 'interrupts') or are binding/vendor specific properties.
59 A property can also define a child DT node with child properties defined
62 For more details on properties sections, see 'Property Schema' section.
65 Optional. Similar to 'properties', but names are regex.
68 A list of DT properties from the 'properties' section that
69 must always be present.
72 Optional. A list of one or more DTS hunks implementing the
73 binding. Note: YAML doesn't allow leading tabs, so spaces must be used instead.
75 Unless noted otherwise, all properties are required.
80 The 'properties' section of the schema contains all the DT properties for a
81 binding. Each property contains a set of constraints using json-schema
82 vocabulary for that property. The properties schemas are what is used for
83 validation of DT files.
85 For common properties, only additional constraints not covered by the common
86 binding schema need to be defined such as how many values are valid or what
87 possible values are valid.
89 Vendor specific properties will typically need more detailed schema. With the
90 exception of boolean properties, they should have a reference to a type in
91 schemas/types.yaml. A "description" property is always required.
93 The Devicetree schemas don't exactly match the YAML encoded DT data produced by
94 dtc. They are simplified to make them more compact and avoid a bunch of
95 boilerplate. The tools process the schema files to produce the final schema for
96 validation. There are currently 2 transformations the tools perform.
98 The default for arrays in json-schema is they are variable sized and allow more
99 entries than explicitly defined. This can be restricted by defining 'minItems',
100 'maxItems', and 'additionalItems'. However, for DeviceTree Schemas, a fixed
101 size is desired in most cases, so these properties are added based on the
102 number of entries in an 'items' list.
104 The YAML Devicetree format also makes all string values an array and scalar
105 values a matrix (in order to define groupings) even when only a single value
106 is present. Single entries in schemas are fixed up to match this encoding.
114 The DT schema project must be installed in order to validate the DT schema
115 binding documents and validate DTS files using the DT schema. The DT schema
116 project can be installed with pip::
118 pip3 install git+https://github.com/devicetree-org/dt-schema.git@master
120 dtc must also be built with YAML output support enabled. This requires that
121 libyaml and its headers be installed on the host system.
126 The DT schema binding documents must be validated using the meta-schema (the
127 schema for the schema) to ensure they are both valid json-schema and valid
128 binding schema. All of the DT binding documents can be validated using the
129 ``dt_binding_check`` target::
131 make dt_binding_check
133 In order to perform validation of DT source files, use the ``dtbs_check`` target::
137 Note that ``dtbs_check`` will skip any binding schema files with errors. It is
138 necessary to use ``dt_binding_check`` to get all the validation errors in the
139 binding schema files.
141 It is also possible to run checks with a single schema file by setting the
142 ``DT_SCHEMA_FILES`` variable to a specific schema file.
146 make dt_binding_check DT_SCHEMA_FILES=Documentation/devicetree/bindings/trivial-devices.yaml
147 make dtbs_check DT_SCHEMA_FILES=Documentation/devicetree/bindings/trivial-devices.yaml
150 json-schema Resources
151 ---------------------
154 `JSON-Schema Specifications <http://json-schema.org/>`_
156 `Using JSON Schema Book <http://usingjsonschema.com/>`_