Dear OpenOCD Community, I’m contributing to the RISC-V fork of OpenOCD ( https://github.com/riscv-collab/riscv-openocd) and have proposed improvements to enhance transport selection flexibility when communicating with targets. I’ve opened an issue to discuss this work: https://github.com/riscv-collab/riscv-openocd/discussions/1175.
My changes are in a branch of my fork: https://github.com/anicic-nikola/riscv-openocd-esperanto. The improvements include: - Support for selecting different transport mechanisms (e.g., socket DMI) to reduce JTAG dependency. - A Python-based target simulator to test OpenOCD with socket DMI communication. etc. These changes build on discussions in https://github.com/riscv-collab/riscv-openocd/discussions/1175#discussioncomment-12852481 and aim to make OpenOCD more flexible for diverse hardware setups. The simulator facilitates testing without physical hardware. I’d appreciate your feedback on: - The proposed transport selection approach and its alignment with OpenOCD’s goals. - Additional improvements or test cases needed. - Advices on how to tackle remaining challenges described in the discussion. I also request *permission to submit a pull request *to the OpenOCD repository, as I currently face a “must be a collaborator” restriction. Alternatively, I can submit patches to the openocd-devel list if preferred. Please share your thoughts via this thread, the GitHub issue, or by replying to me directly. I’m eager to refine this work based on community input. Thank you, Nikola Anicic Esperanto Technologies, Inc.