Pass discovered modules to fusesoc generators about opentitan HOT 8 CLOSED

Hey Phil, what is the plan around this issue? I recall you mentioning that the goal was to influence fusesoc directly. Is that still the plan? If so, or if not, can you keep this issue fresh with the methods that solve this need?

from opentitan.

We're evaluating whether it makes sense to contract Olof (the FuseSoC author) to work on this and other wishlist issues. I'm working through the related issues and docs today in order to move this forwards.

from opentitan.

Scrubbing stale issues... @imphil , has this wishlist item been addressed by FuseSoC? Thanks.

from opentitan.

No, still open and workaround is still in place.

from opentitan.

One step closer here. The cores in the dependency tree now have a property that list their resolved direct dependencies. The complete resolved dependency tree is also available in the EDAM structure. It should be possible to leverage this to also send that info to the generators. Is the VLNVs enough or do you also need core location?

from opentitan.

I'm looking into this now. Thanks for the first step @olofk!

from opentitan.

Fixed in https://github.com/lowRISC/fusesoc/commits/generators, used in #1679. Will need upstreaming to fusesoc once we're happy with it in OT for a bit.

from opentitan.

Upstream bug is at olofk/fusesoc#391 and waiting for a review from @olofk. We have been running with this patch in our ot branch for over half a year now and it works well.

from opentitan.