regzon is the name for the json output for reg'z data model. This PR includes a number of tests for serialization and deserialization. The intention of this format is to provide an accessible format for tooling, and to provide a medium for schema patching. Rather than errata, patching can be utilized to improve codegen such as type deduplication.
This format is subject to change.
* A quick fix for ATDF codegen and register groups
The current code was expecting register groups and peripherals to have a
1:1 ratio, however the register groups under `modules` are more like
templates which are used to instantiate peripherals. This does the
minimal amount of refactoring in order for this to work and unblock some
users.
* i386 -> x86 target name update
* add system registers for cortex m0 and m1, nvic register generation for the chip
* make warning log a debug log
* usize to u64
* register size is now max 2^16 - 1
* add template for PRs
* add CI workflow
* add test step and some empty tests
* reduce targets in workflow, add example svd
* added first test
* added line
* don't crosscompile from windows for now
* snapshotting so that I can work on this as I travel
* add atdf parsing
* update readme
* fix register generation edge case and skip fields named RESERVED
* usb: unify handling of length parameters
The length parameters in USB packets are strict and the existing
implementation does already hardcode the length within the serialize
functions. Therefore the user should not provide these lengths during
creation time. In order to make this more verbose use defaults values
within the structs and reference these defaults values within the
serialization functions. It's not possible to remove these fields
completly as this fields are used within deserialization from the
hardware itself.
Signed-off-by: Tobias Kohlbau <tobias@kohlbau.de>
* Remove length fields as they are obsolete. The lengths is only required
during the serialization and therefore is already hardcoded. The only
other convinience is the calculation of the descriptor size, but this is
already wrong. extern structs are padded and therefore do not have the
right sizes.
Signed-off-by: Tobias Kohlbau <tobias@kohlbau.de>
* cleanup some accidentially added changes
Signed-off-by: Tobias Kohlbau <tobias@kohlbau.de>
---------
Signed-off-by: Tobias Kohlbau <tobias@kohlbau.de>
This field is a bit tricky: It belongs to the EXECCTRL register, while
most of the other fields can determined based on the directives used in
the assembled program, this one has to be set explicitly. This makes it
so that we have to plumb some way to explicitly set the field. I did
this by making `LoadAndStartProgramOptions` take `ExecOptions`, but this
isn't perfect, because if a user sets other fields, they will be ignored
in favour of the values set in the program.
* Adds better docs to README.md
* Adds tar, gzip to bundle.py checked dependencies.
* Adds fancy rendering of .data/chip-families.svg in output for embedding.
Matt Knight
Ethan Frei
Matt Hall
Riccardo Binetti
Gregory Anders
Tobias Kohlbau
Grazfather
Felix Queißner