[all-commits] [llvm/llvm-project] 394c8d: [AggressiveInstCombine] Fold low-bits mask table l...

[Chris Kennelly via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 394c8dd1493308cb9add77b18987aee2219a65c1
      https://github.com/llvm/llvm-project/commit/394c8dd1493308cb9add77b18987aee2219a65c1
  Author: Chris Kennelly <ckennelly at ckennelly.com>
  Date:   2026-09-22 (Tue, 22 Sep 2026)

  Changed paths:
    M llvm/lib/Transforms/AggressiveInstCombine/AggressiveInstCombine.cpp
    A llvm/test/Transforms/AggressiveInstCombine/lower-table-based-lowbits-mask-n32.ll
    A llvm/test/Transforms/AggressiveInstCombine/lower-table-based-lowbits-mask.ll

  Log Message:
  -----------
  [AggressiveInstCombine] Fold low-bits mask table loads to arithmetic (#222509)

Recognize a load from a constant "low bits mask" table

```
  static const uintN_t tbl[K] = { 0, 1, 3, 7, ... }; // tbl[j] == (1 << j) - 1
  ... x & tbl[i] ...
```

and replace the load with the equivalent arithmetic mask (1 << i) - 1.
This is the target-independent form of the X86 combineAndLoadToBZHI DAG
combine: on X86 with BMI2 the surrounding and still selects to a single
bzhi.

Assisted-by: Claude Code



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