svdtools
svdtools is a set of tools for modifying vendor-supplied, often buggy SVD files. It can be imported as a library for use in other applications, or run directly via the included svdtools
CLI utility.
A common use case is patching vendor-supplied SVD files, then applying svd2rust to the resulting patched SVD.
This project is developed and maintained by the Tools team.
Getting Started with Python version
Python 3.6 or newer is required to install and use svdtools
. To install:
$ pip3 install --upgrade --user svdtools
Once installation has completed, the svd
utility can be called from the command line.
An example is given in make example
, which calls svd patch example/incomplete-stm32l4x2.yaml
and generates a patched SVD file example/stm32l4x2.svd.patched
.
See Device and Peripheral YAML Format for more information on creating patches.
Getting Started with Rust version
This crate is guaranteed to compile on stable Rust 1.58.0 and up. To install:
$ cargo install svdtools
Once installation has completed, the svdtools
utility can be called from the command line. Command line interface is same as CLI for Python version.
Develop
To each their own, but the intended workflow is as follows:
- Setup a virtual environment via
make setup
; this also installs thesvd
CLI - Activate the virtual environment by running
source venv/bin/activate
(or use direnv) - Iterate, running
make check
andmake fix
as necessary
Device and Peripheral YAML Format
The patch specifications are in YAML, and have the following general format:
# Path to the SVD file we're targeting. Relative to this file.
# This must be included only in the device YAML file.
_svd: "../svd/STM32F0x0.svd"
# Include other YAML files. Path relative to this file.
_include:
- "../peripherals/gpio_v2.yaml"
# Alter top-level information and peripherals for this device
_modify:
version: 1.1
description: bla bla
addressUnitBits: 8
width: 32
cpu:
revision: r1p2
mpuPresent: true
# Peripherals can either live directly at this level (but other top-level
# fields will name match first)
C_ADC:
name: ADC_Common
# Or they can be inside a _peripherals block, to avoid name conflicts.
_peripherals:
FSMC:
description: Flexible static memory controller
# Multiple address blocks are supported via the addressBlocks list
# use either addressBlock or addressBlocks, but not both
addressBlocks:
- offset: 0x0
size: 0x400
usage: "ADC base registers"
- offset: 0x1000
size: 0x400
usage: "ADC extra registers"
# Add whole new peripherals to this device.
# Incredibly this feature is required.
_add:
ADC_Common:
description: ADC Common registers
groupName: ADC
baseAddress: 0x40012300
addressBlock:
offset: 0x0
size: 0x400
usage: "All ADC registers"
# Multiple address blocks are supported via the addressBlocks list
addressBlocks:
- offset: 0x0
size: 0x400
usage: "ADC base registers"
- offset: 0x1000
size: 0x400
usage: "ADC extra registers"
registers:
CSR:
description: ADC Common status register
addressOffset: 0x0
access: read-only
resetValue: 0x00000000
fields:
OVR3:
description: Overrun flag of ADC3
bitOffset: 21
bitWidth: 1
interrupts:
ADC1_2:
description: ADC global interrupt
value: 18
# A whole new peripheral can also be created as derivedFrom another peripheral.
_add:
USART3:
derivedFrom: USART1
baseAddress: "0x40004800"
interrupts:
USART3:
description: USART3 global interrupt
value: 39
# A new peripheral can have all its registers copied from another, in case
# it cannot quite be derivedFrom (e.g. some fields need different enumerated
# values) but it's otherwise almost exactly the same.
# The registers are copied but not name or address or interrupts, which are
# preserved if the target already exists.
_copy:
ADC3:
from: ADC2
# The new peripheral can also be copied from another svd file for a different
# device. This is useful when a peripheral is missing in a device but the exact
# same peripheral already exist in another device.
# When copying from another file, all fields including interrupts are copied.
_copy:
TIM1:
from: ../svd/stm32f302.svd:TIM1
# Replace peripheral registers by a 'deriveFrom'.
# This is used when e.g. UART4 and UART5 are both independently defined,
# but you'd like to make UART5 be defined as derivedFrom UART4 instead.
_derive:
# The KEY peripheral looses all its elements but 'interrupt', 'name',
# and 'baseAddress', and it is derivedFrom the VALUE peripheral.
# Peripherals that were 'deriveFrom="KEY"' are now 'deriveFrom="VALUE"'.
UART5: UART4
# Reorder the hierarchy of peripherals with 'deriveFrom'.
# This is used when e.g. I2C1 is marked as derivedFrom I2C3,
# but you'd like to swap that so that I2C3 becomes derivedFrom I2C1.
_rebase:
# The KEY peripheral steals everything but 'interrupt', 'name',
# and 'baseAddress' elements from the VALUE peripheral.
# Peripherals that were 'deriveFrom="VALUE"' are now 'deriveFrom="KEY"'.
# The VALUE peripheral is marked as derivedFrom the updated KEY.
I2C1: I2C3
# An STM32 peripheral, matches an SVD <peripheral> tag.
# Does not match any tag with derivedFrom attribute set.
"GPIO*":
# We can include other YAML files inside this peripheral
_include:
- "path/to/file.yaml"
# Alter fields on existing registers inside this peripheral
_modify:
# Rename this badly named register. Takes effect before anything else.
# Don't use wildcard matches if you are changing the name!
# We could have specified name or description or other tags to update.
GPIOB_OSPEEDR:
name: OSPEEDR
# Equivalently the register could go in a '_registers' block
_registers:
GPIOB_OSPEEDR:
name: OSPEEDR
# Change the value of an interrupt in this peripheral
_interrupts:
EXTI0:
value: 101
# Add new registers and interrupts to this peripheral.
# Entries are registers by default, which can also go inside a '_registers'
# block, or interrupts go in an '_interrupts' block.
_add:
EXAMPLER:
description: An example register
addressOffset: 0x04
access: read-write
fields:
EXR1:
description: Example field
bitOffset: 16
bitWidth: 4
_registers:
EXAMPLR2:
description: Another example register
_interrupts:
EXAMPLEI:
description: An example interrupt
value: 100
# Anywhere you can '_add' something, you can also '_delete' it.
# Wildcards are supported. The value here can be a YAML list of registers
# to delete (supported for backwards compatibility), or a YAML mapping
# of lists of registers or interrupts.
_delete:
GPIO*_EXTRAR:
_registers:
- GPIO*_EXAMPLER
_interrupts:
- USART1
# If registers have unnecessary common prefix/postfix,
# you can clean it in all registers in peripheral by:
_strip:
- "PREFIX_*_"
_strip_end:
- "_POSTFIX_"
# You can collect several same registers into one register array
# that will be represented with svd2rust as array or elements
# with one type
# Minimal version:
_array:
ARRAY*: {}
# You can also use the modifiers shown below:
_array:
ARRAY*:
name: NEW_NAME%s
_modify:
FIELD:
description: NEWDESC
OTHER_ARRAY*: {}
# If you have registers that make up a group and can be repeated,
# you can collect them into cluster like this:
_cluster:
CLUSTER%s:
FIRST_REG: {}
SECOND_REG: {}
# clusters can be expanded into individual registers. The name of the resulting register will be the cluster name, concatenated with the register name.
_expand_cluster:
- CLUSTER_ONE*
- CLUSTER_TWO*
# When expanding clusters containing more than one element (as specified by <dim>), each register will have substutute [%] in the cluster name with its index number. If the cluster has a dimIndex element, a %s in the cluster name will be replaced by dimIndex element. [%] is not compatible with dimIndex, as according to SVD 1.3.10
# The SVD 1.3.10 does not specify a delimiter for expansion, so passing the following parameters to the cluster as a hash will allow you to set the delimiter before and after the index is applied (the default delimiters are "_". You can also force the cluster to apply a zero index to a cluster with a single element by passing in the _zeroindex: true parameter
_expand_cluster:
CLUSTER_ONE*:
CLUSTER_TWO*:
_preindex: "_"
_postindex: "_"
_zeroindex: true
# if you pass the _noprefix: true parameter to a cluster, the cluster name will not be prefixed with the peripheral name. This is only valid for single element clusters.
_expand_cluster:
CLUSTER_ONE*:
_noprefix: true
# A register on this peripheral, matches an SVD <register> tag
MODER:
# As in the peripheral scope, rename or redescribe a field.
# Don't use wildcard matches if you are changing the name!
_modify:
FIELD:
description: NEWDESC
# Change the writeConstraint of a field to enumerateValues
_write_constraint: "enum"
# Remove any writeConstraint from this field
_write_constraint: "none"
# Change the writeConstraint of a field to a range of values
_write_constraint: [MINIMUM, MAXIMUM]
# Add new fields to this register
_add:
NEWFIELD:
description: DESCRIPTION
bitOffset: 12
bitWidth: 4
access: read-write
# Often fields that should be one contiguous integer are specified
# as a number of individual bits instead. This merges any matching
# registers into a single field with the combined bitwidth and lowest
# bit offset, and the shared description and access.
_merge:
- "FIELD*"
# You can also merge fields with different base name like this:
_merge:
FIELD: [FIELD1, FIELD_?]
# Or like this:
_merge:
FIELD:
- FIELD1
- FIELD_?
# Or even like this:
_merge:
NEW_FIELD: "FIELD*"
# A field in this register, matches an SVD <field> tag
FIELD:
# You can optionally specify name for `enumeratedValues`
_name: NAME
# By giving the field a dictionary we construct an enumerateValues
VARIANT: [VALUE, DESCRIPTION]
VARIANT: [VALUE, DESCRIPTION]
# Use `-1` for "default" variant which will be consider
# for all other values that are not listed explicitly
# usually datasheet marks them `0b0xxx`, `0b1x`, etc.
VARIANT: [-1, DESCRIPTION]
FIELD:
# If a field already has enumerateValues, drop them and
# replace them with entirely new ones.
_replace_enum:
VARIANT: [VALUE, DESCRIPTION]
VARIANT: [VALUE, DESCRIPTION]
# Another field. A list of two numbers gives a range writeConstraint.
FIELD: [MINIMUM, MAXIMUM]
# Another field with separate enumerated values for read and write
FIELD:
_read:
VARIANT: [VALUE, DESCRIPTION]
VARIANT: [VALUE, DESCRIPTION]
_write:
VARIANT: [VALUE, DESCRIPTION]
VARIANT: [VALUE, DESCRIPTION]
# Sometimes fields are to big so we need to split them into smaller fields
EXTI:
IMR:
# This would split MR into MRi where i = 0 ... bitlength
_split: [MR]
# This would split CHxFM into CHiFM where i = 0 ... bitlength
# and use the current bit for the description in each field
_split:
CHxFM:
name: CH%sFM
description: Processor 2 transmit channel %s free interrupt mask
# If fields have unnecessary common prefix/postfix,
# you can clean it in all registers in peripheral by:
_strip:
- "PREFIX_*_"
_strip_end:
- "_POSTFIX_"
# You can list glob-like rules separated by commas to cover more periperals or registers at time.
# If rule is optional (peripheral may be missing in some devices) add `?~` in the header.
# Don't abuse it. First test not optional rule.
"?~TIM[18],TIM20":
CR2:
# Fields also support collecting in arrays
_array:
OIS?:
description: Output Idle state (OC%s output)
# Optional rules are supported here too
"?~OIS?N":
description: Output Idle state (OC%sN output)
Name Matching
Peripheral, register, and field names can be specified:
- Directly (eg. the full name of the peripheral/register/field)
- Using
?
and*
for single- and multi- character wildcards - Using
[ABC]
to give a list of possible matching characters - Using commas to separate a list of possible matches
You must quote the name if using any special characters in YAML.
The enumerated values On
and Off
are treated as a boolean in YAML and Python will throw the error: AttributeError: 'bool' object has no attribute 'startswith'
, which does not give you much information about where the error is. To avoid it, surround the values with quotes like any other special character.
Style Guide
- Enumerated values should be named in the past tense (enabled, masked, etc.)
- Descriptions should start with capital letters and should not end with a period
License
svdtools is licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE)
- MIT License (LICENSE-MIT)
at your option.
Contribute
Pull requests are very welcome!
Please apply black
and isort
before committing. This can be accomplished by:
- running
make fix
- running
black svdtools/
andisort -y --recursive svdtools/
- installing an editor/IDE plugin
This avoids bikeshedding over formatting issues :)
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
Code of Conduct
Contribution to this crate is organized under the terms of the Rust Code of Conduct, the maintainer of this crate, the Tools team, promises to intervene to uphold that code of conduct.