Some of these are also tripping up a python 2 / python 3 warning
in mypy in the way that '{}'.format(b'foo') works, which we silence by
explicitly requesting the python 3 behavior.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
The way that _init_tokens() is manipulating globals() defeats static
analyses of the file that are trying to infer a type for the 'tok_id'
variable in assignment expressions like 'tok_id = _T_INCLUDE'.
To make it easier on the analyser, define the token types as an
enum.IntEnum named _T. This means we can write e.g. '_T.INCLUDE'
instead of '_T_INCLUDE', avoiding line length increases in the lexing
code.
While we're here, use '==' and '!=' instead of 'is' and 'is not'
when comparing a tok_id that is obtained from an re.Match.lastindex
with a _T.FOO value.
This is now necessary since an int object and a _T object definitely
don't point to the same memory. It worked previously because CPython
interns all integer instances from -5 to 256, but that's an
implementation detail and not a language feature. Since we're getting
the ints from an re.Match.lastindex instead of putting the exact
_T_FOO values into some list, this code probably should not strictly
speaking have been using 'is'.
Explicitly initialize the global _token_re also, to make it more
visible for static analysis.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Continue annotating the module. In some cases mypy will miss that
_err() calls means the function will not return, so we return an
unnecessary local variable to appease it.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add a _MarkerType enum. A subsequent commit will use it for type
annotations in the Property class.
Fix an incorrect type in a comment while we're here.
We continue to use 'marker_type is _MarkerType.FOO' instead of
'marker_type == _MarkerType.FOO' because we are adding those actual
_MarkerType.FOO objects to each property, so 'is' comparison
is legitimate.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
A step along the way towards typing the whole module.
Fix an incorrect (or at best misleading) comment while we're here,
which was noticed by the type checker when I originally annotated
'props' as a Dict[str, bytes].
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Marking this NoReturn helps the type checker figure out that functions
which call it are only returning valid values or failing to
return. (It unfortunately doesn't always work as mypy's control flow
analysis seems to treat a direct 'raise DTError(...)' differently than
calling _err() in some situations, but it helps.)
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Move the DTError, Node, Type, and Property definitions to the top.
This way, class definitions occur before methods which use those
classes. This will be useful to avoid string literals in type
annotations that will be added later. Some can't be avoided due to
circular dependencies, but this will help.
Adjust whitespace.
No functional changes expected.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
It can be convenient to "iterate" over the elements of a property, in
the same way it is convenient to "iterate" over enabled instances.
Add a new macro for doing this, along with a DT_INST_FOREACH_PROP_ELEM
variant.
This is likely to be more convenient than UTIL_LISTIFY or FOR_EACH in
some situations because:
- it handles inputs of any length
- compiler error messages will be shorter and more self-contained
- it is easier to use with phandle-array type properties, which
require more complicated macro boilerplate when used with
util_macro.h APIs
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
We already have support for handling the Zephyr binding "path" type in
edtlib.Node._prop_val(), but the binding inference code isn't making
use of that. Handle this type as well, as it is just as convenient as
Type.PHANDLE and can be more idiomatic depending on the situation.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Instead of hard-coding constants, use an IntEnum.
These is still a subclass of 'int', but is both easier to import and
easier to read during debugging.
For example, compare:
>>> Type.BYTES
<Type.BYTES: 1>
with:
>>> TYPE_BYTES
1
However, 'Type.BYTES == 1' is still True, and the enum values
otherwise behave like you would expect.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add the ability to filter which properties get imported when we do an
include. We add a new YAML form for this:
include:
- name: other.yaml
property-blocklist:
- prop-to-block
or
include:
- name: other.yaml
property-allowlist:
- prop-to-allow
These lists can intermix simple file names with maps, like:
include:
- foo.yaml
- name: bar.yaml
property-allowlist:
- prop-to-allow
And you can filter from child bindings like this:
include:
- name: bar.yaml
child-binding:
property-allowlist:
- child-prop-to-allow
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
We are now in the process of extracting edtlib and dtlib into a
standalone source code library that we intend to share with other
projects.
Links related to the work making this standalone:
https://pypi.org/project/devicetree/https://python-devicetree.readthedocs.io/en/latest/https://github.com/zephyrproject-rtos/python-devicetree
This standalone repo includes the same features as what we have in
Zephyr, but in its own 'devicetree' python package with PyPI
integration, etc.
To avoid making this a hard fork, move the code that's being made
standalone around in Zephyr into a new scripts/dts/python-devicetree
subdirectory, and handle the package and sys.path changes in the
various places in the tree that use it.
From now on, it will be possible to update the standalone repository
by just recursively copying scripts/dts/python-devicetree's contents
into it and committing the results.
This is an interim step; do NOT 'pip install devicetree' yet.
The code in the zephyr repository is still the canonical location.
(In the long term, people will get the devicetree package from PyPI
just like they do the 'yaml' package today, but that won't happen for
the foreseeable future.)
This commit is purely intended to avoid a hard fork for the standalone
code, and no functional changes besides the package structure and
location of the code itself are expected.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Whenever a child-binding: dict has a compatible, we ought to make
that available in EDT._compat2binding. If we don't, we can't look it
up later.
This has adverse effects like missing child bindings which describe
buses.
Fixes: #32071
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Commit 0d4dca10b2 ("scripts: edtlib:
child binding compatibles match parents") was a hack meant to keep the
edtlib.Binding class in place without modifying some twister behavior
that needed further changes to work properly with first-class binding
objects.
This is a hack and is no longer necessary, so back out of this change.
Child Binding objects now have None compatible properties unless the
binding YAML explicitly sets a compatible.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Allow users to set the new EDT constructor argument which errors out
on deprecated properties via a command line argument.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
We'd like to start eliminating deprecated properties from upstream
Zephyr devicetrees. To make that possible in the build system, add an
EDT kwarg that does just that.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Commit 6bf761fc0a ("dts: Remove support
for deprecated DTS binding syntax") removed most of the support for
the 'legacy' bindings syntax.
A few straggler keys are still around in the bindings check code,
though. This allows some legacy keys which should cause errors to pass
silently instead.
Fix the error handling and print good errors for cases which are
removed, just in case someone is still using them somewhere.
Clean up some other error messages in the same function while we're
here.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
This macro returns a node's name with unit-adddress, given its node
identifier.
The node name is useful information for the user to utilize for debug
information, similar to DT_NODE_PATH, or DT_LABEL.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Change the use of a f-string to a normal string as there is nothing
that needs formatting in the particular instance.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
While using the encoded path to a device tree node guarantees a unique
identifier for the corresponding device there is a limit on the number
of characters of that name that can be captured when looking up a
device by name from user mode, and the path can exceed that limit.
Synthesize a unique name from the node dependency ordinal instead, and
update the gen_defines script to record the name associated with the
full path in the extern declaration.
Add a build-time check that no device is created with a name that
violates the user mode requirement.
Also update the network device DTS helper functions to use the same
inference for dev_name and label that the real one does, since they
bypass the real one.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Before we had a bindings index in the documentation, the generated
header file was (somewhat unfortunately) often our best reference for
what a particular binding or property within a binding ends up doing,
so it made good sense to put the description in the generated file.
Now that we have HTML documentation that's a bit more digestible than
the generated file, though, we can just point users at that. Do that
and remove the inline description from the generated file.
This makes it possible to put C-style multiline comments in the
descriptions themselves, which will be done in subsequent patches.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
This macro returns a node's full path, given its node identifier.
The entire path to a node is useful information for the user which can
be added to build-time error messages.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Generate a header (device_extern.h) that handles extern of possible
device structs that would come from devicetree. This removes the need
for DEVICE_DT_DECLARE and DEVICE_DT_INST_DECLARE which we can remove.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
There are some drivers in the tree that support devices on multiple
different buses, although so far this has not been represented in
device tree using the bus concept. In order to convert these drivers &
bindings to refer to a formal bus in device tree we need to be able to
match bindings which lack an explicit "on-bus: ..." value against any
parent bus. This will also be needed for any external bindings, since
those would not be aware of on-bus (as it's a Zephyhr-specific
extension).
The two drivers I'm particularly targeting is the ns16550 UART driver
(drivers/serial/uart_ns16550.c) and the DW I2C driver
(drivers/i2c/i2c_dw.c). They both support devices with a fixed MMIO
address as well as devices connected and discovered over PCIe. The
only issue is that instead of encoding the bus information the proper
DT way these bindings use a special "pcie" property in the DT node
entries to indicate whether the node is on the PCIe bus or not.
Being able to convert the above two drivers to use the DT bus concept
allow the removal of "hacks" like this:
if DT_INST_PROP(0, pcie) || \
DT_INST_PROP(1, pcie) || \
DT_INST_PROP(2, pcie) || \
DT_INST_PROP(3, pcie)
to the more intuitive:
if DT_ANY_INST_ON_BUS_STATUS_OKAY(pcie)
This also has the benefit that the driver doesn't need to make any
arbitrary assumptions of how many matching devices there may be but
works for any number of matches. This is already a problem now since
e.g. the ns16550 driver assumes a maximum of 4 nodes, whereas
dts/x86/elkhart_lake.dtsi defines up to 9 different ns16550 nodes.
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The DTS language permits zeroing out phandles in a phandle array to
say "there's nothing at this index", and dtlib manages that correctly,
but edtlib and gen_defines.py aren't equipped to do so.
Fix this by allowing None elements in the lists of ControllerAndData
values returned by edtlib for such properties.
Handle that in gen_defines.py by setting the generated
DT_N_<node>_P_<prop>_IDX_<i>_EXISTS macro to 0 in such cases.
The DT_N_<node>_P_<prop>_LEN macro still accounts for the entire
length of the phandle-array; it's just that some indexes may be
missing data.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
These are likely enough defined by mistake to emit a warning for.
Adjust tests to match, tweaking the test_warnings() setup: now that
we've got several test cases, it's a bit cleaner not to have to
copy/paste the ('edtlib', WARNING, ...) part of every expected log
record tuple.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
The edtlib strategy for emitting warnings is to print directly to
standard error. This in turn requires hacks to drop stored references
to stderr in various _warn_file attributes so the EDT objects can be
pickled.
In general, I think it's not really appropriate for library modules
like edtlib to be printing to stderr directly. The user should be able
to configure logging for general utility data munging modules like
this as they please, and not just deciding what file to print to.
Move this around so the standard logging module is used instead. We
can preserve backwards compatibility in gen_defines by customizing the
'edtlib' logging module behavior so it prints the exact same thing it
always has.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Whenever a devicetree binding defines a string property whose
enumerated values are all tokenizable, generate C macros for each
property value that are the corresponding tokens.
Note that "token" is distinct from "identifier": both 'foo' and '123'
are valid tokens, but only 'foo' is a valid identifier. We permit some
strings which are not valid identifiers in anticipation that the
generalization may be useful, e.g. when defining macros that paste the
token onto a prefix that makes the whole thing an identifier.
Fixes: #21273
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add enum_tokenizable and enum_upper_tokenizable to PropertySpec. These
allow a PropertySpec to declare that it both has an enumeration of
values and all of them are strings which are "tokenizable". Don't
bother extending Property with these; the user can access the
information through Property.spec now, so the extra delegation is
unnecessary.
See the docstrings for details on what "tokenizable" means. The basic
idea is that we should be able to use the DT binding's enum values as
C 'enum' enumerators in a "reasonable way".
Add val_as_token to Property. This produces a canonical token for the
property value.
Add tests for this feature in particular and property enumerations in
general.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
In the 'normal' case of a property whose definition is taken from a
binding YAML file, a fair number (three at present) of the attributes
available on Property objects are directly taken from the
corresponding PropertySpec object.
Refactor the internals of how a Property gets initialized so that it
has a direct reference to its PropertySpec, and make those attributes
properties which just delegate to the PropertySpec (which in turn just
delegate to the binding). Additionally, expose the PropertySpec
directly.
This will make it easier to extend the Property class with additional
attributes that normally come from the PropertySpec without having to
touch all the locations where Property.__init__ is called.
In the case of the 'default' properties, we handle this by dummying
out some PropertySpec objects. These dummy PropertySpecs in turn
require a dummy Binding.
This change has the advantage that it improves the degree to which
these defaults are checked, e.g. it makes sure that 'status' is one of
the permitted values.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
We have a use case for checking the results of a DT_PROP_HAS_IDX()
call with COND_CODE_1(). That won't work because its expansion is an
integer comparison; COND_CODE_1() expects a literal 1 or 0.
Adjust the macro implementation so it expands to a literal 1 or 0.
Make this work even when the index argument needs an expansion while
we're at it.
Fixes: #29833
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Fixes: #29758
Commit 7165b77a81 ("scripts: edtlib:
refactor for first class bindings") introduced a Binding class.
Its child_binding instance attribute has a compatible which can be
None. Adjust this behavior so the child Binding object has the same
compatible as the parent binding which ultimately has a compatible.
Without this, sanitycheck's expr_parser is doing some matching on
compatibles in child nodes that is producing unexpected results.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
The child_binding object should default to having a path and
compatible that matches the parent's. Mark it as xfail because the
compatible part is failing.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
This is a convenience function for creating a bunch of Binding objects
from files in a directory.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add the ability to mark a property as 'deprecated' to get a warning that
it will be removed in the future.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
_prop_val comment referred to one of the arguments called "optional"
however the code has changed to call that argument "required" now. Fix
up the comment block to use the correct argument name and semantics of
that argument.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Add two new types: Binding and PropertySpec.
- Binding is a first-class abstraction for a devicetree binding
file as defined by a YAML file in the Zephyr syntax.
- PropertySpec is a helper type which represents a property
definition within a Binding.
Make the Binding constructor a new entry point to the library. This
enables users to deal with bindings as standalone entities, apart from
how they characterize a particular devicetree.
Rework the EDT and Node internals that deal with bindings as dict
values to use the Binding type instead. To make this less ambiguous,
use the variable name 'raw' when we're dealing with a binding as it's
parsed from YAML, and 'binding' when we're dealing with a Python
Binding object.
This commit drops support for the following legacy bindings keys
- '#cells': use '*-cells' instead (e.g. 'gpio-cells', 'pwm-cells')
- "child-bus: foo" and "child: bus: foo": use "bus:" instead
- "parent-bus" and "parent: bus: ": use "on-bus:" instead
Officially, legacy bindings have been gone since
6bf761fc0a ("dts: Remove support for deprecated DTS binding
syntax"), so this is vestigial code, and I couldn't find any in-tree
users.
It also drops the convention in this file that ""-strings are
preferred.
I honestly don't understand why this was ever enforced; the file
itself admits single quotes are common in Python and we use them
elsewhere in Zephyr's Python code.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add the first API functions that directly deal with node dependency
ordinals as determined by edtlib:
- DT_DEP_ORD(node_id): node_id's ordinal
- DT_REQUIRES_DEP_ORDS(node_id): list of dep ordinals for node_id's
direct dependencies
- DT_SUPPORTS_DEP_ORDS(node_id): list of dep ordinals for nodes
depending directly on node_id
- DT_INST_ equivalents
This is not meant to be an exhaustive set of macros related to
dependency ordinals; rather, it's a starting out point meant to enable
initial struct device dependency tracking work. We can add more if
needed.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
The legacy macros were first deprecated in Zephyr v2.3. Now that
Zephyr v2.4 has been released, that makes two releases where these
macros have been deprecated, so it's OK to remove them.
This leaves support for legacy binding syntax in place. Removing that
is left to future work.
We need to update various pieces of documentation related to flash
partitions that never got updated when the new API was introduced.
Consolidate this information in the flash_map.h API reference page,
since that's really where users will run into it. This also gives us
the opportunity to improve this documentation.
Adjust a couple of kconfigfunctions.py and sanitycheck bits to use
non-legacy edtlib APIs.
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Each controller node in a phandle-array may set the number of cells in
a specifier as any nonnegative integer. Currently, we don't allow
this in edtlib in the case where there are multiple controllers in a
phandle-array property all of which have 0 cells in the relevant
specifier, which is not correct. Fix this, add a regression test, and
improve the error message while we are here.
Fixes: #28709
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
Add a lookup table for finding a node by its dependency ordinal.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Signed-off-by: Martí Bolívar <marti.bolivar@nordicsemi.no>
