Closed 0Grit closed 2 years ago
@ilg-ul (Not ARM affiliated) See below for why I copied.
Please note that I am not part of the Arm team.
The CMSIS Pack design has a lot of problems, and I cannot fully recommend it, but CMSIS SVD is more or less ok.
Sorry was a bit late when I opened this.
I copied @ilg-ul because he has an excellent toolset similar to a lot of the cmsis concepts but with significant improvements.
Several years ago, when I was intensely promoting a common software interface to RISC-V, I also suggested an improved version of SVD, which I used internally in my projects and in the Eclipse Embedded CDT RISC-V plug-ins.
Unfortunately the RISC-V International design team still did not agree on a microcontroller profile, and I don't know of any real efforts to standardise the software interface.
I think that, with some caution, the Arm CMSIS SVD specs can be also used to define the RISC-V peripherals, and I salute Espressif's initiative to start this project.
If more and more RISC-V vendors will use SVDs, it may become a de facto standard, and the RISC-V International team will probably accept it.
I apologize for getting to this issue so late, I have been preoccupied with other projects. Thank you all for your interest!
To answer the initial question, no I have not been in contact with anybody from the ARM team. I am unsure at this point if we will use anything else from CMSIS. We are producing these SVDs largely at the request of the community, and I'm admittedly not well versed with CMSIS in general. I suppose if somebody made a compelling argument for something in particular it could be considered, though.
Discussions are now enabled so please move any further conversation here: https://github.com/espressif/svd/discussions/17
Bravo on experimenting with an open standard.
I'd love to see this work evolve into using more of CMSIS and eventually enable CMSIS & RISC-V to also benefit from it.
Are you in contact with anybody from the ARM teams now that they are making the CMSIS packs concept more open?