Fix wrapping arithmetic triggering false overflow errors in MIR optimization

[Copilot]

Wrapping arithmetic intrinsics (wrapping_add, wrapping_sub, wrapping_mul) incorrectly triggered compile-time overflow errors when MIR optimization level ≥2. Example:

fn main() {
    println!("{}", 1_u32.wrapping_sub(2_u32));  // Error: arithmetic operation will overflow
}

Root Cause

LowerIntrinsics converts wrapping intrinsics to plain BinOp::{Add,Sub,Mul}. Later, KnownPanicsLint called wrapping_to_overflowing() on all plain arithmetic operations to check overflow, incorrectly flagging wrapping operations.

Changes

• compiler/rustc_mir_transform/src/known_panics_lint.rs: Modified check_binary_op() to only check operations already marked as overflow-checking (*WithOverflow variants). Removed the blanket wrapping_to_overflowing() conversion.

• tests/ui/mir/mir-opt-wrapping-arithmetic.rs: Added regression test covering wrapping operations with -Z mir-opt-level=2, including const evaluation.

Behavior Preserved

• Debug mode overflow checking: Regular arithmetic uses *WithOverflow variants (still checked)
• Const/static overflow checking: Always uses *WithOverflow variants (still checked)
• Release mode wrapping: Plain Add/Sub/Mul correctly wraps (not an error)
• Wrapping intrinsics: Now correctly skip overflow checks

Original prompt

Fix a compiler bug where u32::wrapping_sub (and likely other wrapping arithmetic) incorrectly triggers an overflow error when compiling with -Z mir-opt-level=2 or above.

Repro:

fn main() {
    println!("{}", 1_u32.wrapping_sub(2_u32));
}

Expected: Compiles and runs, printing 4294967295.
Actual (with -Z mir-opt-level=2): Compile-time error:

error: this arithmetic operation will overflow
...
attempt to compute `1_u32 - 2_u32`, which would overflow

Requirements:

• Make wrapping arithmetic calls (at least wrapping_sub) not be turned into checked/overflowing arithmetic during MIR optimizations.
• Add regression test(s) that fail before the fix and pass after.
• Ensure no new ICEs or behavior changes for non-wrapping arithmetic.
• The PR must be opened against the Enselic/rust repository (a fork), not upstream.
• Do not include links to the upstream issue in the PR description or commit messages.

Notes:

• The original report used rustc nightly 1.52.0-nightly (2021-02-18) and -Z mir-opt-level=2.
• Likely area: MIR constant propagation / SimplifyCfg / InstCombine / ConstProp optimizations incorrectly folding wrapping_sub into a plain subtraction and then applying overflow lint/diagnostics.

This pull request was created from Copilot chat.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.