-
Notifications
You must be signed in to change notification settings - Fork 762
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[EDN] V2 Review Items #11489
Comments
Signed-off-by: Steve Nelson <[email protected]>
Signed-off-by: Steve Nelson <[email protected]>
The latest nightly regression shows > 90% coverage for this block. Is there anything else pending to sign-off? |
I see FSM coverage is still below 90%. |
This might be related to branches that we can exclude (due to escalation), but I don't know whether the coverage exclusions have been applied to this block yet. |
If the FSM coverage is due to D2S related features, we could just waive it and address during V2S. |
I quickly glanced over the coverage results and saw that the uncovered arcs in @weicaiyang didn't we discuss an option that would automatically exclude reset transitions from coverage collection? |
@msfschaffner
Similar to edn_main_sm . *-> Idle can be triggered by disabling Disabling EDN in the middle of the seq may be not very important. Perhaps we can leave it for V3. |
actually that's quite important. So we definitely should test these. @senelson7 previously merged #14534 for this, so it's probably something that should be tested as well. |
Make sense. Thanks for pointing this out. |
I had code in a PR that forced the *->Error and *->Idle transitions but the consensus was it wasn't necessary just to get FSM transition coverage up. #14407 highlights this. Maybe this was just wrt reset and *->Error but I removed it all. Are you indicating you want the *->Idle due to not enabled to be put back? I probably had only done it for csrng so far but also planned to do it for edn. |
Hi @senelson7, |
For *-> error, we could keep idle -> error as the common sec_cm V2S test can cover it. And add exclusions for non-idle -> error. #14407 only exclude transition that only happens on reset. So it does't exclude all *->idle, as some can be triggered by |
The DV portion in #14407 is done. The transition that only happens on reset are automatically excluded now. It works for EDN. AES FSM coverage is improved to 90%+.
Indeed! |
The lint policy update in #14407 is pending (we have to wait for an updated policy file bundled with the next AscentLint release from RealIntent). However, I ran the new policy already locally and all FSMs in the entire chip conform to the new rule. |
Hence it sounds like the only thing that we still need to do for EDN is covering functional arcs to IDLE that are not due to the reset. Would the suggestion from @weicaiyang RE programming |
Maybe, but it'll be totally random which state we are in when we disable. Maybe we can define what edn will be doing when it is disabled? If not, there are several cases to try to hit, i.e. no/some/all endpoints requesting, waiting for entropy, entropy delivered, etc. Same for csrng? |
These sound very interesting cases to hit. @martin-lueker may have better idea on what cases are good to test. |
As discussed in the last Si meeting, here are the takeaway.
@senelson7 Let me know if you think we need more discussion on this issue. |
All items covered. Close this issue |
This issue captures AIs that came up in the V2 meeting for EDN:
All valid and ready combinations are covered in TB now: https://reports.opentitan.org/hw/ip/edn/dv/latest/cov_report/grp15.html
Two sequences: edn_disable, edn_disable_auto_req_mode
The text was updated successfully, but these errors were encountered: