Skip to content

Fixed Coverage Holes for RV32F #670

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

Merged
merged 4 commits into from
Jul 28, 2025

Conversation

sudo-apt-abdullah
Copy link
Contributor

The coverage was initially dropped due to the incorrect operand values. Each test has an associated dataset at its end, but the dataset values were incorrect.
op1val: 0x7f800001
op2val: 0xff800000
But the dataset incorrectly records:
NAN_BOXED(0x7f800001139095041,32,FLEN)
NAN_BOXED(0xff800000286578688,32,FLEN)
To correct this and achieve full coverage, I regenerated the tests with the correct boxing logic. Here's the difference in between the initial and final coverage can be seen: https://docs.google.com/spreadsheets/d/1V1TRYfl7aYsMEXnGEAn7FwO558mKAwf9a3cCnG6dBow/edit?gid=0#gid=0

@allenjbaum
Copy link
Collaborator

allenjbaum commented Jun 28, 2025 via email

@sudo-apt-abdullah
Copy link
Contributor Author

Yes, I worked to bring the tests up to 100% coverage, which was initially very low because the val_comb coverpoints weren’t being hit due to incorrect dataset values.

At first, I regenerated the tests with a corrected dataset. However, in the latest commit, I simply fixed the dataset using a Python script, since the test cases themselves were already correct.

@allenjbaum
Copy link
Collaborator

allenjbaum commented Jun 29, 2025 via email

@sudo-apt-abdullah
Copy link
Contributor Author

Thank you for pointing that out, and I apologize for the confusion caused by GitHub showing the entire file as changed.

To clarify, the test cases themselves were not regenerated — they remain exactly the same. The only modification was at the end of each file, where the dataset values are defined. These were updated using a Python script to ensure that the val_comb coverpoints are now correctly hit. The original test logic was valid; the issue was solely with incorrect operand values in the dataset section, which led to poor coverage initially.

Unfortunately, due to line-ending normalization or whitespace changes (possibly introduced by tooling or script formatting), GitHub's diff view shows the entire file as changed, even though the actual update is limited to the dataset portion.

You can confirm this by comparing the beginning and middle sections of each file — only the dataset at the end has been updated.

@UmerShahidengr
Copy link
Collaborator

I have checked the test files from @sudo-apt-abdullah system directly, and it looks fine to me. A python script is doing the same change in all tests, no other change was observed in any test. So I am merging this one.

Copy link
Collaborator

@UmerShahidengr UmerShahidengr left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@UmerShahidengr UmerShahidengr merged commit 0810992 into riscv-non-isa:dev Jul 28, 2025
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants