[all-commits] [llvm/llvm-project] cac8d7: Speed up Sema virtual-method no-work paths

[Mehdi Amini via All-commits]

  Branch: refs/heads/users/mamini/clang-speed-sema-no-work
  Home:   https://github.com/llvm/llvm-project
  Commit: cac8d7f66458ce562034b073b14e56626d301fb7
      https://github.com/llvm/llvm-project/commit/cac8d7f66458ce562034b073b14e56626d301fb7
  Author: Mehdi Amini <joker.eph at gmail.com>
  Date:   2026-09-11 (Fri, 11 Sep 2026)

  Changed paths:
    M clang/lib/Sema/SemaDecl.cpp
    M clang/lib/Sema/SemaDeclCXX.cpp
    M clang/lib/Sema/SemaExpr.cpp

  Log Message:
  -----------
  Speed up Sema virtual-method no-work paths

Skip override and hidden-virtual analysis when class structure proves
there is no work to perform. Avoid swapping empty ODR-use maps when
leaving an expression evaluation context.

CTMark O0 (3 samples, CPU 6): 29.439800 s -> 29.619533 s (+0.611%).

Impact on significant TUs in MLIR build time:
- `mlir/lib/RegisterAllDialects.cpp`: 4.9758% fewer retired
  instructions.
- `mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp`: 1.8612% fewer retired
  instructions.

Assisted-by: Codex



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications