Hello, I'm working on an industrial production test device that needs to be able to interact with parts under test via JTAG/SWD to:
* Write to RAM * Write to the flash * Start execution at specific addresses * Provide single step environment for the device under test (during development mode) * Support arm (cortex and later) and riscv targets We're developing this testing platform with a Zynq Ultrascale as the CPU running a scripting language and I hope/plan on using OpenOCD as my tool to interact with the parts during testing. To reduce test time, we plan on putting a JTAG/SWD hardware block into the programmable logic that will reliably and quickly clock bits in and out. It's unclear to me what the peripheral interface to userland needs to be and I was trying to find the layer in OpenOCD that best fits what I'm trying to do. I assume it's something like: * Write bitstream * Read bitstream * Write bits, then read bits (or the opposite, or both) * Force interface bits (TMS, TDO, TCLK, etc.) in a particular direction for a set time (or number of clocks) * Set clock speed But I can't quite find that interface. Looking through the jtag/drivers directory, I've found... The bitbang/gpiod stuff seems way too low level(though as a starting point I may start with this to get functionality going while I work on the accelerated interface.) A bunch of the others seem to be too high of level, or not the interface I'm interested in (ftdi, usb, etc.). Close is the xlnx-pcie-xvc one, but the interface to the JTAG hardware isn't PCIE(its memory mapped registers over AXI.) I'm not sure the nomenclature of what I'm trying to make here: * Interface? * Adapter? * Driver? * Transport? (I don't think so) Also, I see in many of the 'drivers' source code is code that navigates the JTAG state machine, which I don't think is something I care about (or, more to the point, something I would prefer to leverage from existing code). Apologies if I have missed finding the architectural documentation that would be helpful to me, but what's at the doxygen scrape doesn't seem to really show high level flow or how the blocks interact. Anyways, any guidance would be greatly appreciated, Russ Schultz Silicon Labs
