Extending LLVM: Adding instructions, intrinsics, types, etc.

  1. Introduction and Warning
  2. Adding a new intrinsic function
  3. Adding a new instruction
  4. Adding a new SelectionDAG node
  5. Adding a new type
    1. Adding a new fundamental type
    2. Adding a new derived type

Written by Misha Brukman, Brad Jones, Nate Begeman, and Chris Lattner

Introduction and Warning

During the course of using LLVM, you may wish to customize it for your research project or for experimentation. At this point, you may realize that you need to add something to LLVM, whether it be a new fundamental type, a new intrinsic function, or a whole new instruction.

When you come to this realization, stop and think. Do you really need to extend LLVM? Is it a new fundamental capability that LLVM does not support at its current incarnation or can it be synthesized from already pre-existing LLVM elements? If you are not sure, ask on the LLVM-dev list. The reason is that extending LLVM will get involved as you need to update all the different passes that you intend to use with your extension, and there are many LLVM analyses and transformations, so it may be quite a bit of work.

Adding an intrinsic function is far easier than adding an instruction, and is transparent to optimization passes. If your added functionality can be expressed as a function call, an intrinsic function is the method of choice for LLVM extension.

Before you invest a significant amount of effort into a non-trivial extension, ask on the list if what you are looking to do can be done with already-existing infrastructure, or if maybe someone else is already working on it. You will save yourself a lot of time and effort by doing so.

Adding a new intrinsic function

Adding a new intrinsic function to LLVM is much easier than adding a new instruction. Almost all extensions to LLVM should start as an intrinsic function and then be turned into an instruction if warranted.

  1. llvm/docs/LangRef.html: Document the intrinsic. Decide whether it is code generator specific and what the restrictions are. Talk to other people about it so that you are sure it's a good idea.
  2. llvm/include/llvm/Intrinsics*.td: Add an entry for your intrinsic. Describe its memory access characteristics for optimization (this controls whether it will be DCE'd, CSE'd, etc). Note that any intrinsic using the llvm_int_ty type for an argument will be deemed by tblgen as overloaded and the corresponding suffix will be required on the intrinsic's name.
  3. llvm/lib/Analysis/ConstantFolding.cpp: If it is possible to constant fold your intrinsic, add support to it in the canConstantFoldCallTo and ConstantFoldCall functions.
  4. llvm/test/Regression/*: Add test cases for your test cases to the test suite

Once the intrinsic has been added to the system, you must add code generator support for it. Generally you must do the following steps:

Add support to the C backend in lib/Target/CBackend/
Depending on the intrinsic, there are a few ways to implement this. For most intrinsics, it makes sense to add code to lower your intrinsic in LowerIntrinsicCall in lib/CodeGen/IntrinsicLowering.cpp. Second, if it makes sense to lower the intrinsic to an expanded sequence of C code in all cases, just emit the expansion in visitCallInst in Writer.cpp. If the intrinsic has some way to express it with GCC (or any other compiler) extensions, it can be conditionally supported based on the compiler compiling the CBE output (see llvm.prefetch for an example). Third, if the intrinsic really has no way to be lowered, just have the code generator emit code that prints an error message and calls abort if executed.
Add support to the .td file for the target(s) of your choice in lib/Target/*/*.td.
This is usually a matter of adding a pattern to the .td file that matches the intrinsic, though it may obviously require adding the instructions you want to generate as well. There are lots of examples in the PowerPC and X86 backend to follow.

Adding a new SelectionDAG node

As with intrinsics, adding a new SelectionDAG node to LLVM is much easier than adding a new instruction. New nodes are often added to help represent instructions common to many targets. These nodes often map to an LLVM instruction (add, sub) or intrinsic (byteswap, population count). In other cases, new nodes have been added to allow many targets to perform a common task (converting between floating point and integer representation) or capture more complicated behavior in a single node (rotate).

  1. include/llvm/CodeGen/SelectionDAGNodes.h: Add an enum value for the new SelectionDAG node.
  2. lib/CodeGen/SelectionDAG/SelectionDAG.cpp: Add code to print the node to getOperationName. If your new node can be evaluated at compile time when given constant arguments (such as an add of a constant with another constant), find the getNode method that takes the appropriate number of arguments, and add a case for your node to the switch statement that performs constant folding for nodes that take the same number of arguments as your new node.
  3. lib/CodeGen/SelectionDAG/LegalizeDAG.cpp: Add code to legalize, promote, and expand the node as necessary. At a minimum, you will need to add a case statement for your node in LegalizeOp which calls LegalizeOp on the node's operands, and returns a new node if any of the operands changed as a result of being legalized. It is likely that not all targets supported by the SelectionDAG framework will natively support the new node. In this case, you must also add code in your node's case statement in LegalizeOp to Expand your node into simpler, legal operations. The case for ISD::UREM for expanding a remainder into a divide, multiply, and a subtract is a good example.
  4. lib/CodeGen/SelectionDAG/LegalizeDAG.cpp: If targets may support the new node being added only at certain sizes, you will also need to add code to your node's case statement in LegalizeOp to Promote your node's operands to a larger size, and perform the correct operation. You will also need to add code to PromoteOp to do this as well. For a good example, see ISD::BSWAP, which promotes its operand to a wider size, performs the byteswap, and then shifts the correct bytes right to emulate the narrower byteswap in the wider type.
  5. lib/CodeGen/SelectionDAG/LegalizeDAG.cpp: Add a case for your node in ExpandOp to teach the legalizer how to perform the action represented by the new node on a value that has been split into high and low halves. This case will be used to support your node with a 64 bit operand on a 32 bit target.
  6. lib/CodeGen/SelectionDAG/DAGCombiner.cpp: If your node can be combined with itself, or other existing nodes in a peephole-like fashion, add a visit function for it, and call that function from . There are several good examples for simple combines you can do; visitFABS and visitSRL are good starting places.
  7. lib/Target/PowerPC/PPCISelLowering.cpp: Each target has an implementation of the TargetLowering class, usually in its own file (although some targets include it in the same file as the DAGToDAGISel). The default behavior for a target is to assume that your new node is legal for all types that are legal for that target. If this target does not natively support your node, then tell the target to either Promote it (if it is supported at a larger type) or Expand it. This will cause the code you wrote in LegalizeOp above to decompose your new node into other legal nodes for this target.
  8. lib/Target/TargetSelectionDAG.td: Most current targets supported by LLVM generate code using the DAGToDAG method, where SelectionDAG nodes are pattern matched to target-specific nodes, which represent individual instructions. In order for the targets to match an instruction to your new node, you must add a def for that node to the list in this file, with the appropriate type constraints. Look at add, bswap, and fadd for examples.
  9. lib/Target/PowerPC/PPCInstrInfo.td: Each target has a tablegen file that describes the target's instruction set. For targets that use the DAGToDAG instruction selection framework, add a pattern for your new node that uses one or more target nodes. Documentation for this is a bit sparse right now, but there are several decent examples. See the patterns for rotl in PPCInstrInfo.td.
  10. TODO: document complex patterns.
  11. llvm/test/Regression/CodeGen/*: Add test cases for your new node to the test suite. llvm/test/Regression/CodeGen/X86/bswap.ll is a good example.

Adding a new instruction

WARNING: adding instructions changes the bitcode format, and it will take some effort to maintain compatibility with the previous version. Only add an instruction if it is absolutely necessary.

  1. llvm/include/llvm/Instruction.def: add a number for your instruction and an enum name
  2. llvm/include/llvm/Instructions.h: add a definition for the class that will represent your instruction
  3. llvm/include/llvm/Support/InstVisitor.h: add a prototype for a visitor to your new instruction type
  4. llvm/lib/AsmParser/Lexer.l: add a new token to parse your instruction from assembly text file
  5. llvm/lib/AsmParser/llvmAsmParser.y: add the grammar on how your instruction can be read and what it will construct as a result
  6. llvm/lib/Bitcode/Reader/Reader.cpp: add a case for your instruction and how it will be parsed from bitcode
  7. llvm/lib/VMCore/Instruction.cpp: add a case for how your instruction will be printed out to assembly
  8. llvm/lib/VMCore/Instructions.cpp: implement the class you defined in llvm/include/llvm/Instructions.h
  9. Test your instruction
  10. llvm/lib/Target/*: Add support for your instruction to code generators, or add a lowering pass.
  11. llvm/test/Regression/*: add your test cases to the test suite.

Also, you need to implement (or modify) any analyses or passes that you want to understand this new instruction.

Adding a new type

WARNING: adding new types changes the bitcode format, and will break compatibility with currently-existing LLVM installations. Only add new types if it is absolutely necessary.

Adding a fundamental type

  1. llvm/include/llvm/Type.h: add enum for the new type; add static Type* for this type
  2. llvm/lib/VMCore/Type.cpp: add mapping from TypeID => Type*; initialize the static Type*
  3. llvm/lib/AsmReader/Lexer.l: add ability to parse in the type from text assembly
  4. llvm/lib/AsmReader/llvmAsmParser.y: add a token for that type

Adding a derived type

  1. llvm/include/llvm/Type.h: add enum for the new type; add a forward declaration of the type also
  2. llvm/include/llvm/DerivedTypes.h: add new class to represent new class in the hierarchy; add forward declaration to the TypeMap value type
  3. llvm/lib/VMCore/Type.cpp: add support for derived type to:
    std::string getTypeDescription(const Type &Ty,
      std::vector<const Type*> &TypeStack)
    bool TypesEqual(const Type *Ty, const Type *Ty2,
      std::map<const Type*, const Type*> & EqTypes)
    
    add necessary member functions for type, and factory methods
  4. llvm/lib/AsmReader/Lexer.l: add ability to parse in the type from text assembly
  5. llvm/lib/BitCode/Writer/Writer.cpp: modify void BitcodeWriter::outputType(const Type *T) to serialize your type
  6. llvm/lib/BitCode/Reader/Reader.cpp: modify const Type *BitcodeReader::ParseType() to read your data type
  7. llvm/lib/VMCore/AsmWriter.cpp: modify
    void calcTypeName(const Type *Ty,
                      std::vector<const Type*> &TypeStack,
                      std::map<const Type*,std::string> &TypeNames,
                      std::string & Result)
    
    to output the new derived type

Valid CSS Valid HTML 4.01 The LLVM Compiler Infrastructure
Last modified: $Date$