[CodeGen][BranchFolding] Add options to control common hoisting and block reordering (#205704)
The BranchFolder pass performs several independent optimizations: tail
merging, common-code hoisting, and branch optimization (which includes
basic-block reordering). Currently only tail merging is individually
controllable -- it is disabled for targets that require a structured
CFG,
because tail merging can make the CFG irreducible. Common-code hoisting
and
basic-block reordering always run whenever the pass executes.
Those two sub-phases do not change CFG edges (so they cannot affect
reducibility); they only change block layout. A target whose register
allocation is sensitive to the final block layout may need the rest of
branch folding while suppressing reordering and/or hoisting. This is a
layout / register-allocation concern, distinct from the structured-CFG
(reducibility) concern that gates tail merging.
This change makes common-code hoisting and basic-block reordering
individually controllable:
- BranchFolder gains an EnableBasicBlockReordering flag (with a
setBasicBlockReordering() setter) that gates the two block-reordering
transforms in OptimizeBlock.
- BranchFolderLegacy gains constructor parameters for both sub-phases,
and a
createBranchFolder(EnableCommonHoist, EnableBasicBlockReordering)
factory
is exposed so a target can request a safe subset from its pass pipeline.
- Two hidden command-line flags, -branch-folder-hoist-common-code and
-branch-folder-reorder-blocks, override the configured values at the
start
of OptimizeFunction (mirroring -enable-tail-merge).
Defaults enable both sub-phases, so existing behavior is unchanged for
all
in-tree targets.
New MIR tests exercise each flag via -run-pass=branch-folder.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Welcome to the LLVM project!
This repository contains the source code for LLVM, a toolkit for the construction of highly optimized compilers, optimizers, and run-time environments.
The LLVM project has multiple components. The core of the project is itself called “LLVM”. This contains all of the tools, libraries, and header files needed to process intermediate representations and convert them into object files. Tools include an assembler, disassembler, bitcode analyzer, and bitcode optimizer.
C-like languages use the Clang frontend. This component compiles C, C++, Objective-C, and Objective-C++ code into LLVM bitcode -- and from there into object files, using LLVM.
Other components include: the libc++ C++ standard library, the LLD linker, and more.
Consult the Getting Started with LLVM page for information on building and running LLVM.
For information on how to contribute to the LLVM project, please take a look at the Contributing to LLVM guide.
Join the LLVM Discourse forums, Discord chat, LLVM Office Hours or Regular sync-ups.
The LLVM project has adopted a code of conduct for participants to all modes of communication within the project.