Passes

• Passes
  • -cir-call-conv-lowering
  • -cir-canonicalize
  • -cir-drop-ast
  • -cir-flatten-cfg
  • -cir-goto-solver
  • -cir-hoist-allocas
  • -cir-idiom-recognizer
    • Options
  • -cir-lib-opt
    • Options
  • -cir-lifetime-check
    • Options
  • -cir-lowering-prepare
  • -cir-mlir-scf-prepare
  • -cir-simplify

-cir-call-conv-lowering

Handle calling conventions for CIR functions

This pass lowers CIR function definitions and calls according to the calling conventions for the target architecture. This pass is necessary to properly lower CIR functions to LLVM IR.

-cir-canonicalize

Performs CIR canonicalization

Perform canonicalizations on CIR and removes some redundant operations.

This pass performs basic cleanup and canonicalization transformations that hopefully do not affect CIR-to-source fidelity and high-level code analysis passes too much. Example transformations performed in this pass include empty scope cleanup, trivial try cleanup, redundant branch cleanup, etc. Those more "heavyweight" transformations and those transformations that could significantly affect CIR-to-source fidelity are performed in the cir-simplify pass.

-cir-drop-ast

Remove clang AST nodes attached to CIR operations

Some CIR operations have references back to Clang AST, this is necessary to perform lots of useful checks without having to duplicate all rich AST information in CIR. As we move down in the pipeline (e.g. generating LLVM or other MLIR dialects), the need for such nodes diminish and AST information can be dropped.

Right now this is enabled by default in Clang prior to dialect codegen from CIR, but not before lifetime check, where AST is required to be present.

-cir-flatten-cfg

Produces flatten CFG

This pass transforms CIR and inline all the nested regions. Thus, the next post condtions are met after the pass applied:

• there is not any nested region in a function body
• all the blocks in a function belong to the parent region In other words, this pass removes such CIR operations like IfOp, LoopOp, ScopeOp and etc. and produces a flat CIR.

-cir-goto-solver

Replaces goto operations with branches

This pass transforms CIR and replaces goto-s with branch operations to the proper blocks.

-cir-hoist-allocas

Hoist allocas to the entry of the function

This pass hoist all non-dynamic allocas to the entry of the function. This is helpful for later code generation.

-cir-idiom-recognizer

Raise calls to C/C++ libraries to CIR operations

This pass recognize idiomatic C++ usage and incorporate C++ standard containers, library functions calls, and types into CIR operation, attributes and types.

Detections done by this pass can be inspected by users by using remarks. Currently supported are all and found-calls.

Options

-remarks : Diagnostic remarks to enable Supported styles: {all|found-calls}

-cir-lib-opt

Optimize C/C++ library calls

By using higher level information from cir-idiom-recognize, this pass apply transformations to CIR based on specific C/C++ library semantics.

Transformations done by this pass can be inspected by users by using remarks. Currently supported are all and transforms.

Options

-remarks : Diagnostic remarks to enable Supported styles: {all|transforms}

-cir-lifetime-check

Check lifetime safety and generate diagnostics

This pass relies on a lifetime analysis pass and uses the diagnostics mechanism to report to the user. It does not change any code.

A default ctor is specified but is solely in order to make tablegen happy, since this pass requires the presence of an ASTContext, one can set that up using mlir::createLifetimeCheckPass(clang::ASTContext &) instead.

Options

-history       : List of history styles to emit as part of diagnostics. Supported styles: {all|null|invalid}
-remarks       : List of remark styles to enable as part of diagnostics. Supported styles: {all|pset}
-history_limit : Max amount of diagnostics to emit on pointer history

-cir-lowering-prepare

Preparation work before lowering to LLVM dialect

This pass does preparation work for LLVM lowering. For example, it may expand the global variable initialziation in a more ABI-friendly form.

-cir-mlir-scf-prepare

Preparation work before lowering to SCF dialect

This pass does preparation work for SCF lowering. For example, it may hoist the loop invariant or canonicalize the loop comparison. Currently, the pass only be enabled for through MLIR pipeline.

-cir-simplify

Performs CIR simplification and code optimization

The pass performs code simplification and optimization on CIR.

Unlike the cir-canonicalize pass, this pass contains more aggresive code transformations that could significantly affect CIR-to-source fidelity. Example transformations performed in this pass include ternary folding, code hoisting, etc.