Add UL/ULL suffixes to constants in encoding.h

[heathdutton]

Adds _UL and _ULL macros (following the Linux kernel pattern) and wraps all 64-bit constants in encoding.h with _ULL().

This ensures the constants have the correct type in C while remaining compatible with assembly code.

Fixes #173

[aswaterman]

OK with me after my comment is addressed, but I'd like feedback from others before merging.

Thanks for the contribution, @heathdutton.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.08%. Comparing base (383cbca) to head (5f120ba).
⚠️ Report is 49 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #398      +/-   ##
==========================================
+ Coverage   96.53%   97.08%   +0.55%     
==========================================
  Files          10       14       +4     
  Lines         750      926     +176     
==========================================
+ Hits          724      899     +175     
- Misses         26       27       +1     

☔ 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.

[aswaterman]

Should we not wrap all of the remaining numerical constants (the ones that fit within 32 bits) with _RISCV_CSR_UL?

[heathdutton]

Done - wrapped all 32-bit constants with _RISCV_CSR_UL as well for consistency.

[Timmmm]

LGTM. I guess the 32-bit version isn't strictly necessary (I think it would only make a difference on 16-bit CPUs) but it's nice for symmetry. Also if it is necessary then there are some 32-bit valeus in the generated code that maybe should have this applied too.

Also I'd maybe rename to just _RISCV_UL, _RISCV_ULL since the macros aren't really CSR-specific. Very minor though.

[heathdutton]

Renamed to _RISCV_UL/_RISCV_ULL as suggested.

[aswaterman]

@Timmmm it isn't only about the width; it's also about the signedness. For example, 0x40000000 << 1 >> 1 and 0x40000000U << 1 >> 1 don't have the same value on a system with 32-bit int. This is a bigger source of surprise than the width matter, given C's rules for typing of constants.