fix(testgen): add frm cross-product support to cp_fs2_edges and cp_fs3_edges

[dev-aditya-hub]

Summary

cp_fs2_edges and cp_fs3_edges were missing the frm cross-product logic that cp_fs1_edges already has. When a coverpoint name contains _frm, the generator is expected to iterate over all six rounding modes (dyn, rdn, rmm, rne, rtz, rup) and emit one test bin per edge value per mode.

Without this, any test plan referencing cp_fs2_edges_frm or cp_fs3_edges_frm would silently produce tests with frm=None. The assembler defaults to DYN in that case, and since fcsr.frm is at its reset value of RNE, a DUT that hardwires RNE computes the correct result and passes every bin without being asked to use any other rounding mode.

Fix

Added cross_frm = "_frm" in coverpoint and the inner for frm_mode in frm_modes loop to both cp_fs2_edges and cp_fs3_edges, matching the structure already present in cp_fs1_edges. Bin names and description strings now include the frm suffix when a rounding mode is active, consistent with the existing pattern.

Files Changed

• coverpoints/cp_fp_reg_edges.py — add frm cross-product logic to make_fs2_edges and make_fs3_edges

Test plan

• Coverpoints containing _frm now produce six bins per edge value from cp_fs2_edges and cp_fs3_edges, matching the output of cp_fs1_edges for the same input
• Coverpoints without _frm still produce one bin per edge value with frm=None, identical to the previous behavior
• Existing YAML test plans that use cp_fs2_edges without the _frm suffix are unaffected

[davidharrishmc]

This was intentional to not include all cross products. An ideal suite would have cross of fs1, fs2, fs3, and rm, but the runtime explodes. There is a more systematic treatment of floating-point based on IBM coverpoints in progress that we expect to deploy this summer. I would suggest only fixing bugs in floating-point tests at this point, but not adding new coverage.

…

On May 7, 2026, at 8:09 AM, Aditya Kuchekar ***@***.***> wrote: Summary cp_fs2_edges and cp_fs3_edges were missing the frm cross-product logic that cp_fs1_edges already has. When a coverpoint name contains _frm, the generator is expected to iterate over all six rounding modes (dyn, rdn, rmm, rne, rtz, rup) and emit one test bin per edge value per mode. Without this, any test plan referencing cp_fs2_edges_frm or cp_fs3_edges_frm would silently produce tests with frm=None. The assembler defaults to DYN in that case, and since fcsr.frm is at its reset value of RNE, a DUT that hardwires RNE computes the correct result and passes every bin without being asked to use any other rounding mode. Fix Added cross_frm = "_frm" in coverpoint and the inner for frm_mode in frm_modes loop to both cp_fs2_edges and cp_fs3_edges, matching the structure already present in cp_fs1_edges. Bin names and description strings now include the frm suffix when a rounding mode is active, consistent with the existing pattern. Files Changed coverpoints/cp_fp_reg_edges.py — add frm cross-product logic to make_fs2_edges and make_fs3_edges Test plan Coverpoints containing _frm now produce six bins per edge value from cp_fs2_edges and cp_fs3_edges, matching the output of cp_fs1_edges for the same input Coverpoints without _frm still produce one bin per edge value with frm=None, identical to the previous behavior Existing YAML test plans that use cp_fs2_edges without the _frm suffix are unaffected You can view, comment on, or merge this pull request online at: #1450 Commit Summary cb9e81e <cb9e81e> fix(testgen): add frm cross-product support to cp_fs2_edges and cp_fs3_edges File Changes (1 file <https://github.com/riscv/riscv-arch-test/pull/1450/files>) M generators/testgen/src/testgen/coverpoints/cp_fp_reg_edges.py <https://github.com/riscv/riscv-arch-test/pull/1450/files#diff-5d5b8aa92bdea8d20ea324c98d9dc94664552d47149e35d924159732fd481ab9> (32) Patch Links: https://github.com/riscv/riscv-arch-test/pull/1450.patch https://github.com/riscv/riscv-arch-test/pull/1450.diff — Reply to this email directly, view it on GitHub <#1450>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AR4AA3ZZNJZSJFHVXEG3FRT4ZSRL7AVCNFSM6AAAAACYUXVZ76VHI2DSMVQWIX3LMV43ASLTON2WKOZUGM4TSNZYGE3DCNQ>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you are subscribed to this thread.