[ibex,tracer] Uniquify trace log with module hierarchy

[Razer6]

In case of having multiple Ibexes, this caues a conflict in openening the same filename from different instances. Add the unique module hierarchy into the filename.

[nasahlpa]

Do you know if the trace log is automatically parsed by some other tool where we also should adapt the path after your change? If not, LGTM.

[Razer6]

Not that I am aware of.

[Razer6]

@GregAC or anyone with access, can you look into the failing CI?

[andreaskurth]

Do you know if the trace log is automatically parsed by some other tool where we also should adapt the path after your change? If not, LGTM.

Yes, it gets parsed by DV, and that's why CI checks fail. The newly added _%m suffix will have to be added to dv/uvm/core_ibex/scripts/run_rtl.py as well as to documentation.

Before proceeding with that, I have a question for @Razer6: Is the hart ID, which gets appended to the file name, not sufficient for separating the trace log files? That is, do you have multiple harts with the same ID?

[Razer6]

Before proceeding with that, I have a question for @Razer6: Is the hart ID, which gets appended to the file name, not sufficient for separating the trace log files? That is, do you have multiple harts with the same ID?

We have multiple Ibexes in the system. These are independent systems, thus the RV spec specifies all of them having hart id 0. Hart ID only becomes different, if let's say, Ibex has multiple harts within itself.

[andreaskurth]

We have multiple Ibexes in the system. These are independent systems, thus the RV spec specifies all of them having hart id 0. Hart ID only becomes different, if let's say, Ibex has multiple harts within itself.

RISC-V just specifies that "at least one hart must have a hart ID of zero" and that "hart IDs must be unique within the execution environment." So this change would address the case when there are actually multiple disjoint execution environments in the same RTL simulation.