pre-commit: PR167771

[zyw-bot]

Link: llvm/llvm-project#167771
Requested by: @dtcxzyw

[zyw-bot]

Diff mode

runner: ariselab-64c-docker
baseline: llvm/llvm-project@16aa79d
patch: llvm/llvm-project#167771
sha256: 2a39b64c5644b007fc95101f524e7961953d7b166d0dabc44ad4804cee37a4a8
commit: 2c8431f

129 files changed, 87063 insertions(+), 87197 deletions(-)

Improvements:
  simplifycfg.NumBitMaps 2245 -> 2246 +0.04%
  simplifycfg.NumLookupTablesHoles 2394 -> 2395 +0.04%
  correlated-value-propagation.NumNonNull 10869930 -> 10874315 +0.04%
  mem2reg.NumDeadAlloca 1963420 -> 1964171 +0.04%
  simple-loop-unswitch.NumCostMultiplierSkipped 17253 -> 17255 +0.01%
  correlated-value-propagation.NumNUW 571064 -> 571123 +0.01%
  correlated-value-propagation.NumAnd 44492 -> 44496 +0.01%
  instcombine.NumPHIsOfInsertValues 26429 -> 26431 +0.01%
  constmerge.NumIdenticalMerged 15653 -> 15654 +0.01%
  simplifycfg.NumFoldValueComparisonIntoPredecessors 512907 -> 512938 +0.01%
Regressions:
  bdce.NumRemoved 384222 -> 383427 -0.21%
  vector-combine.NumVecLoad 4291 -> 4288 -0.07%
  gvn.NumGVNLoad 1312960 -> 1312121 -0.06%
  instcombine.NumConstProp 168343 -> 168248 -0.06%
  mem2reg.NumLocalPromoted 549701 -> 549536 -0.03%
  licm.NumBOAssociationsHoisted 3516 -> 3515 -0.03%
  sroa.NumLoadsPredicated 14459 -> 14456 -0.02%
  licm.NumLoadStorePromoted 64161 -> 64150 -0.02%
  local.NumPHICSEs 185818 -> 185787 -0.02%
  gvn.NumGVNSimpl 4418808 -> 4418075 -0.02%

+18 openvdb/Prune.ll
+0 assimp/zip.ll
+0 ceres/block_sparse_matrix.ll
+0 cmake/cmComputeLinkDepends.ll
+0 faiss/Clustering.ll
+0 g2o/optimization_algorithm_dogleg.ll
+0 g2o/slam2d_linear.ll
+0 hdf5/sio_perf.ll
+0 libigl/directed_edge_orientations.ll
+0 libigl/eigs.ll
+0 libigl/null.ll
+0 llvm/MsgPackDocument.ll
+0 meshlab/edit_align.ll
+0 nlohmann_json/diag_on.ll
+0 ockam-rs/1411u8drt798uxi8.ll
+0 opencv/gfluidbackend.ll
+0 opencv/kernels_video.ll
+0 pbrt-v4/integrator.ll
+0 pbrt-v4/integrators.ll
+0 php/zip.ll
+0 proj/adams.ll
+0 proj/airocean.ll
+0 proj/datum_set.ll
+0 proj/geod_set.ll
+0 proj/helmert.ll
+0 proj/horner.ll
+0 proj/info.ll
+0 proj/tmerc.ll
+0 quickjs/qjsc.ll
+0 quickjs/quickjs-libc.ll
+0 typst-rs/2d3c2n5y91mtl0x0.ll
+0 typst-rs/d6l9ieo9tcw33dn.ll
+0 wasmtime-rs/4ab4rlryc5h7bf6z.ll
-30 proj/param.ll
-31 typst-rs/2i78fvbm4wocuesi.ll

[github-actions]

This patch consists of numerous localized, mechanical changes across many LLVM IR files, primarily involving:

1. Phi node predecessor updates: In phi instructions, the block names in the predecessor list are updated to reflect renamed or reordered basic blocks (e.g., .thread428 → .thread426, %.lr.ph75 → %.lr.ph76, %.loopexit513 → %.loopexit511). These are consistent with loop restructuring or optimization passes that rename loop headers or exit blocks.

2. Call site argument and callee updates: Several tail call and invoke instructions have their callee operand updated from a loaded function pointer (%42, %66, %111) to a different loaded pointer (%43, %67, %112) — indicating a change in indirect call target resolution, likely due to improved interprocedural analysis or devirtualization.

3. Load/store type and alignment fixes: Multiple instances where <4 x i8> vector loads/stores are replaced with scalar i32 or i64 equivalents (e.g., in ockam-rs, typst-rs, opencv), reflecting better type inference or memory layout assumptions (e.g., loading a 4-byte field as i32 instead of <4 x i8>). Also includes alignment adjustments (e.g., align 1 → align 4/8) for correctness.

4. Control flow edge corrections: Branch targets and landing pad edges are updated to match new block names (e.g., %75 → %74, %119 → %120, %._crit_edge.i.i.i.i.i.i.i → %._crit_edge.i.i.i.i.i.i.i), preserving exception handling and loop structure integrity after transformations.

5. Dead code removal and instruction reordering: Minor cleanup such as removal of redundant inttoptr + ptrtoint sequences (e.g., nlohmann_json), elimination of unused alloca/stores (e.g., proj/param.ll), and reordering of store/call instructions (e.g., moving free() calls earlier in hdf5/sio_perf.ll).

All changes are low-level, optimization-driven, and preserve semantics — typical of post-optimization IR refinement (e.g., Loop Simplify, GVN, InstCombine, or SROA improvements). No high-level algorithmic or functional changes are introduced.

model: qwen-plus-latest
CompletionUsage(completion_tokens=546, prompt_tokens=101619, total_tokens=102165, completion_tokens_details=None, prompt_tokens_details=None)