Add directed fp unit tests, and fix bugs found

[dhower-qc]

No description provided.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 72.23%. Comparing base (a1225d7) to head (209afcd).
⚠️ Report is 9 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1811      +/-   ##
==========================================
- Coverage   72.24%   72.23%   -0.02%     
==========================================
  Files          52       52              
  Lines       27671    27671              
  Branches     6009     6009              
==========================================
- Hits        19992    19988       -4     
- Misses       7679     7683       +4     

Flag	Coverage Δ
idlc	76.18% <ø> (ø)
udb	66.23% <ø> (-0.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Copilot]

Pull request overview

Adds a directed floating-point stimulus corpus plus a new C++/SoftFloat comparison test to validate generated fp.idl helpers, and applies several fixes to fp.idl and C++ support code uncovered by those tests.

Changes:

• Add an FPgen-style Ruby generator scaffold and check in an expanded directed f32 JSONL vector corpus.
• Add a new test_softfloat_fp Catch2 test (built + runnable via new rake tasks) that compares generated hart FP ops vs Berkeley SoftFloat.
• Fix multiple floating-point corner cases in spec/std/isa/isa/fp.idl (sNaN NV flag setting, sqrt exponent handling, recip approximation changes) and fix 3-valued logic operators in SatisfiedResult.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tools/scripts/fpgen_generate.rb	Adds a Ruby utility to emit directed f32 vectors and optional Z3 “all-types” vectors.
tests/data/fp/directed/f32_fpgen_expanded.jsonl	Adds checked-in directed f32 vector corpus used by the new C++ test.
tests/data/fp/directed/REUSE.toml	Declares licensing metadata for generated test corpus files in this directory.
spec/std/isa/isa/fp.idl	Fixes FP semantics (sNaN NV flagging, sqrt/div corner cases, recip helper, exp width changes).
backends/cpp_hart_gen/templates/db_data.hxx.erb	Fixes SatisfiedResult && / `
backends/cpp_hart_gen/tasks.rake	Adds rake tasks to build/run the new SoftFloat FP tests.
backends/cpp_hart_gen/cpp/test/test_softfloat_fp.cpp	New Catch2 test comparing hart FP ops vs Berkeley SoftFloat on directed vectors.
backends/cpp_hart_gen/CMakeLists.txt	Adds build dependencies for nlohmann-json and Berkeley SoftFloat; wires up test_softfloat_fp.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[dhower-qc]

Good point — fixing this now. I'll replace the two hard-coded Linux-x86_64-GCC strings with a SOFTFLOAT_BUILD_SPEC cache variable that is auto-derived from CMake's host/toolchain introspection variables (CMAKE_HOST_SYSTEM_NAME, CMAKE_HOST_SYSTEM_PROCESSOR, CMAKE_C_COMPILER_ID). The default value on Linux/x86_64/GCC will remain identical to what is there today, so there is no behaviour change for the current CI environment. Users on other hosts (e.g. macOS/ARM64/Clang) will get the correct spec automatically, and anyone on an unusual toolchain can still pass -DSOFTFLOAT_BUILD_SPEC=<value> to override. I'll push the change shortly.