Hi, I created https://github.com/openrisc/or1k-tests a few days ago to collect all relevant test cases for OpenRISC 1000. The repo is currently empty, so I'm throwing out some ideas for how to populate it.
First thing is that we need a lot of test cases. We are currently using the ones from ORPSoCv2 (with some updated tests in mor1kx-dev-env), the GCC regression test suite and the or1ksim test suite. We are looking to deprecate orpsocv2 and mor1kx-dev-env, so the relevant testcases should just be copied into ork1k-tests. We need to figure out what to do with all the peripheral tests. Should they be part of the or1k test suite, and if so, should we mark them as "system" test cases somehow to inform users that they will need other peripherals, mapped to certain addresses. This will also be a good time to revisit the idea of a standardized address map for or1k peripherals, something that was discussed during orconf and afterwards. The GCC regression test suite is not something we want to copy into the or1k-tests repo, so I'm just suggesting some nice scripts to make it easy to run them from an external location The or1ksim tests is a bit of middle ground. They are currently integrated in or1ksim so that users can run make tests. Perhaps we want to keep them where they are and make some scripts to run them as well, or we start to decouple them from or1ksim. Both ways have pros and cons. Ideally, we want to be able to run all tests against sw models (or1ksim, jor1k, qemu) orpsoc (both verilator and event based sims) and real hardware. If that is too complicated, I say we start with orpsoc and or1ksim. We also want to get some kind of analysis from our tests. The GCC tests generate nice lists of passed/failed, but I'm not sure how or1ksim does it, and the orpsoc tests only report exit status. We probably want some code that can parse the output and generate some statistics. So my idea for a simple layout is src/ Source code for test cases scripts/ Scripts for running the different test suites against our targets doc/ Documentation This is deliberately underspecified to allow us to just dump all the existing stuff into the repo without getting stuck in structural discussions. Any thoughts? //Olof
_______________________________________________ OpenRISC mailing list [email protected] http://lists.openrisc.net/listinfo/openrisc
