[Code generation] Added skeleton for expressions generation

This commit is contained in:
Philippe Tillet
2018-12-20 10:32:07 -05:00
parent 9247ed3714
commit eab275dc99
5 changed files with 374 additions and 63 deletions

View File

@@ -13,9 +13,7 @@ extern translation_unit *ast_root;
const char src[] =
"\
void test(fp32 *A, fp32 *B, fp32 *C){\
fp32 c[16, 16] = {0};\
int32 i = 0;\
i += 1;\
}\
";

View File

@@ -2,8 +2,9 @@
#define TDL_INCLUDE_AST_H
#include "parser.hpp"
#include "llvm/IR/IRBuilder.h"
#include <cassert>
#include <list>
#include <vector>
#include <string>
namespace llvm{
@@ -58,7 +59,7 @@ enum TYPE_T{
// AST
class node {
public:
virtual void codegen(module*) const { }
virtual llvm::Value* codegen(module*) const { return nullptr; }
};
template<class T>
@@ -71,100 +72,146 @@ public:
return this;
}
void codegen(module* mod) const
{ for(T x: values_){ x->codegen(mod); } }
llvm::Value* codegen(module* mod) const{
for(T x: values_){
x->codegen(mod);
}
return nullptr;
}
const std::list<T> &values() const
const std::vector<T> &values() const
{ return values_; }
private:
std::list<T> values_;
std::vector<T> values_;
};
class binary_operator: public node{
class expression: public node{
public:
virtual llvm::Value* codegen(module *) const = 0;
};
class binary_operator: public expression{
private:
llvm::Value* llvm_op(llvm::IRBuilder<> &bld, llvm::Value *lhs, llvm::Value *rhs, const std::string &name) const;
public:
binary_operator(BIN_OP_T op, node *lhs, node *rhs)
: op_(op), lhs_(lhs), rhs_(rhs) { }
: op_(op), lhs_((expression*)lhs), rhs_((expression*)rhs) { }
llvm::Value* codegen(module *) const;
private:
const BIN_OP_T op_;
const node *lhs_;
const node *rhs_;
const expression *lhs_;
const expression *rhs_;
};
class constant: public node{
class constant: public expression{
public:
constant(int value): value_(value) { }
llvm::Value* codegen(module *mod) const;
private:
const int value_;
};
class string_literal: public node{
class string_literal: public expression{
public:
string_literal(char *&value): value_(value) { }
llvm::Value* codegen(module *mod) const;
public:
std::string value_;
};
class unary_operator: public node{
class unary_operator: public expression{
private:
llvm::Value *llvm_op(llvm::IRBuilder<> &builder, llvm::Value *arg, const std::string &name) const;
public:
unary_operator(UNARY_OP_T op, node *arg)
: op_(op), arg_(arg) { }
: op_(op),
arg_((expression*)arg) { }
llvm::Value* codegen(module *mod) const;
private:
const UNARY_OP_T op_;
const node *arg_;
const expression *arg_;
};
class cast_operator: public node{
public:
cast_operator(node *type, node *arg): type_(type), arg_(arg) { }
class type_name;
class cast_operator: public expression{
private:
llvm::Value *llvm_op(llvm::IRBuilder<> &builder, llvm::Type *T, llvm::Value *arg, const std::string &name) const;
public:
const node *type_;
const node *arg_;
cast_operator(node *T, node *arg):
T_((type_name*)T),
arg_((expression*)arg) { }
llvm::Value* codegen(module *mod) const;
public:
const type_name *T_;
const expression *arg_;
};
class conditional_expression: public node{
class conditional_expression: public expression{
private:
llvm::Value *llvm_op(llvm::IRBuilder<> &builder,
llvm::Value *cond, llvm::Value *true_value, llvm::Value *false_value,
const std::string &name) const;
public:
conditional_expression(node *cond, node *true_value, node *false_value)
: cond_(cond), true_value_(true_value), false_value_(false_value) { }
: cond_((expression*)cond),
true_value_((expression*)true_value),
false_value_((expression*)false_value) { }
llvm::Value* codegen(module *mod) const;
public:
const node *cond_;
const node *true_value_;
const node *false_value_;
const expression *cond_;
const expression *true_value_;
const expression *false_value_;
};
class assignment_expression: public node{
class assignment_expression: public expression{
private:
llvm::Value *llvm_op(llvm::IRBuilder<> &builder,
llvm::Value *lvalue, llvm::Value *rvalue,
const std::string &name) const;
public:
assignment_expression(node *lvalue, ASSIGN_OP_T op, node *rvalue)
: lvalue_(lvalue), op_(op), rvalue_(rvalue) { }
: lvalue_((expression*)lvalue), op_(op), rvalue_((expression*)rvalue) { }
llvm::Value* codegen(module *mod) const;
public:
ASSIGN_OP_T op_;
const node *lvalue_;
const node *rvalue_;
const expression *lvalue_;
const expression *rvalue_;
};
class statement: public node{
};
class initializer;
class declaration_specifier;
class declaration: public node{
public:
declaration(node *spec, node *init)
: spec_(spec), init_((list<initializer*>*)init) { }
: spec_((declaration_specifier*)spec), init_((list<initializer*>*)init) { }
void codegen(module* mod) const;
llvm::Value* codegen(module* mod) const;
public:
const node *spec_;
const declaration_specifier *spec_;
const list<initializer*> *init_;
};
@@ -177,7 +224,7 @@ public:
compound_statement(node* decls, node* statements)
: decls_((declarations_t)decls), statements_((statements_t)statements) {}
virtual void codegen(module* mod) const;
llvm::Value* codegen(module* mod) const;
private:
declarations_t decls_;
@@ -230,6 +277,7 @@ public:
decl_((declarator*)decl) { }
llvm::Type* type(module *mod) const;
std::string name() const;
public:
const declaration_specifier *spec_;
@@ -267,7 +315,7 @@ class identifier: public declarator{
llvm::Type* type_impl(module*mod, llvm::Type *type) const;
public:
identifier(char *&name): declarator(nullptr), name_(name) { }
identifier(char *&name): declarator(this), name_(name) { }
const std::string &name() const;
private:
@@ -302,6 +350,8 @@ public:
function(node *id, node *args)
: declarator(id), args_((list<parameter*>*)args) { }
void bind_parameters(module *mod, llvm::Function *fn) const;
public:
const list<parameter*>* args_;
};
@@ -314,23 +364,27 @@ private:
public:
initializer(node *decl, node *init)
: declarator((node*)((declarator*)decl)->id()),
decl_((declarator*)decl), init_(init){ }
decl_((declarator*)decl), init_((expression*)init){ }
void codegen(module *) const;
void specifier(const declaration_specifier *spec);
llvm::Value* codegen(module *) const;
public:
const declaration_specifier *spec_;
const declarator *decl_;
const node *init_;
const expression *init_;
};
class type: public node{
class type_name: public node{
public:
type(TYPE_T spec, node * decl)
: spec_(spec), decl_((declarator*)decl) { }
type_name(node *spec, node * decl)
: spec_((declaration_specifier*)spec), decl_((declarator*)decl) { }
llvm::Type *type(module *mod) const;
public:
const TYPE_T spec_;
const declaration_specifier *spec_;
const declarator *decl_;
};
@@ -340,7 +394,7 @@ public:
function_definition(node *spec, node *header, node *body)
: spec_((declaration_specifier*)spec), header_((function *)header), body_((compound_statement*)body) { }
void codegen(module* mod) const;
llvm::Value* codegen(module* mod) const;
public:
const declaration_specifier *spec_;
@@ -359,7 +413,7 @@ public:
return this;
}
void codegen(module* mod) const;
llvm::Value* codegen(module* mod) const;
private:
list<node*>* decls_;

View File

@@ -1,3 +1,4 @@
#include <unordered_map>
#include "ast.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/IRBuilder.h"
@@ -19,10 +20,13 @@ public:
module(const std::string &name, context *ctx);
llvm::Module* handle();
llvm::IRBuilder<>& builder();
void value(ast::node* node, llvm::Value* value);
llvm::Value *value(ast::node* node);
private:
llvm::Module handle_;
llvm::IRBuilder<> builder_;
std::unordered_map<ast::node*, llvm::Value*> values_;
};

View File

@@ -96,8 +96,8 @@ constant_list
;
type_name
: type_specifier { $$ = new type(get_type_spec($1), nullptr); }
| type_specifier abstract_declarator { $$ = new type(get_type_spec($1), $2); }
: declaration_specifiers { $$ = new type_name($1, nullptr); }
| declaration_specifiers abstract_declarator { $$ = new type_name($1, $2); }
;
/* -------------------------- */
@@ -228,6 +228,13 @@ expression
: assignment_expression { $$ = $1; }
;
/* Initialization */
initialization_expression
: assignment_expression { $$ = $1; }
| '{' constant '}' { $$ = $2; }
;
/* -------------------------- */
/* Statements */
/* -------------------------- */
@@ -315,14 +322,10 @@ declarator
| direct_declarator { $$ = $1; }
;
initializer
: assignment_expression { $$ = $1; }
| '{' constant '}' { $$ = $2; }
;
init_declarator
: declarator { $$ = new initializer($1, nullptr); }
| declarator '=' initializer { $$ = new initializer($1, $3); }
| declarator '=' initialization_expression { $$ = new initializer($1, $3); }
;
/* -------------------------- */

View File

@@ -1,3 +1,4 @@
#include <functional>
#include "ast.h"
#include "codegen.h"
#include "llvm/IR/DerivedTypes.h"
@@ -27,12 +28,21 @@ llvm::IRBuilder<>& module::builder() {
return builder_;
}
void module::value(ast::node* node, llvm::Value* value){
values_[node] = value;
}
llvm::Value *module::value(ast::node* node){
return values_[node];
}
namespace ast{
/* Translation unit */
void translation_unit::codegen(module *mod) const{
Value* translation_unit::codegen(module *mod) const{
decls_->codegen(mod);
return nullptr;
}
/* Declaration specifier */
@@ -55,6 +65,12 @@ Type* parameter::type(module *mod) const {
return decl_->type(mod, spec_->type(mod));
}
std::string parameter::name() const {
if(auto id = decl_->id())
return id->name();
return "";
}
/* Declarators */
Type* declarator::type(module *mod, Type *type) const{
if(ptr_)
@@ -88,6 +104,17 @@ Type* pointer::type_impl(module*, Type *type) const{
}
// Function
void function::bind_parameters(module *mod, Function *fn) const{
std::vector<llvm::Value*> args;
std::transform(fn->arg_begin(), fn->arg_end(), std::back_inserter(args), [&](llvm::Argument& x){ return &x;});
assert(args.size() == args_->values().size());
for(size_t i = 0; i < args.size(); i++){
parameter *param_i = args_->values().at(i);
args[i]->setName(param_i->name());
mod->value(param_i, args[i]);
}
}
Type* function::type_impl(module*mod, Type *type) const{
SmallVector<Type*, 8> types;
for(parameter* param: args_->values()){
@@ -97,31 +124,256 @@ Type* function::type_impl(module*mod, Type *type) const{
}
/* Function definition */
void function_definition::codegen(module *mod) const{
Value* function_definition::codegen(module *mod) const{
FunctionType *prototype = (FunctionType *)header_->type(mod, spec_->type(mod));
const std::string &name = header_->id()->name();
Function *fn = Function::Create(prototype, Function::ExternalLinkage, name, mod->handle());
header_->bind_parameters(mod, fn);
BasicBlock *entry = BasicBlock::Create(mod->handle()->getContext(), "entry", fn);
mod->builder().SetInsertPoint(entry);
body_->codegen(mod);
return nullptr;
}
/* Statements */
void compound_statement::codegen(module* mod) const{
Value* compound_statement::codegen(module* mod) const{
decls_->codegen(mod);
statements_->codegen(mod);
// statements_->codegen(mod);
return nullptr;
}
/* Declaration */
void declaration::codegen(module* mod) const{
Value* declaration::codegen(module* mod) const{
for(initializer *init: init_->values())
init->specifier(spec_);
init_->codegen(mod);
return nullptr;
}
/* Initializat */
void initializer::codegen(module *) const{
/* Initializer */
void initializer::specifier(const declaration_specifier *spec) {
spec_ = spec;
}
Value* initializer::codegen(module * mod) const{
Type *ty = decl_->type(mod, spec_->type(mod));
std::string name = decl_->id()->name();
Value *value = llvm::UndefValue::get(ty);
value->setName(name);
return nullptr;
}
/*------------------*/
/* Expression */
/*------------------*/
/* Binary operator */
Value *binary_operator::llvm_op(llvm::IRBuilder<> &builder, Value *lhs, Value *rhs, const std::string &name) const
{
Type *ltype = lhs->getType();
Type *rtype = rhs->getType();
bool is_float = ltype->isFloatingPointTy() || rtype->isFloatingPointTy();
bool is_ptr = ltype->isPointerTy() || rtype->isPointerTy();
bool is_int = ltype->isIntegerTy() || rtype->isIntegerTy();
bool is_signed = false;
// Mul
if(op_==MUL && is_float)
return builder.CreateFMul(lhs, rhs, name);
if(op_==MUL && is_int)
return builder.CreateMul(lhs, rhs, name);
// Div
if(op_==DIV && is_float)
return builder.CreateFDiv(lhs, rhs, name);
if(op_==DIV && is_int && is_signed)
return builder.CreateSDiv(lhs, rhs, name);
if(op_==DIV && is_int && !is_signed)
return builder.CreateUDiv(lhs, rhs, name);
// Mod
if(op_==MOD && is_float)
return builder.CreateFRem(lhs, rhs, name);
if(op_==MOD && is_int && is_signed)
return builder.CreateSRem(lhs, rhs, name);
if(op_==MOD && is_int && !is_signed)
return builder.CreateURem(lhs, rhs, name);
// Add
if(op_==ADD && is_float)
return builder.CreateFAdd(lhs, rhs, name);
if(op_==ADD && is_int)
return builder.CreateAdd(lhs, rhs);
if(op_==ADD && is_ptr)
return builder.CreateGEP(lhs, {rhs});
// Sub
if(op_==SUB && is_float)
return builder.CreateFSub(lhs, rhs, name);
if(op_==SUB && is_int)
return builder.CreateSub(lhs, rhs, name);
if(op_==SUB && is_ptr)
return builder.CreateGEP(lhs, {builder.CreateNeg(rhs)});
// Left shift
if(op_==LEFT_SHIFT){
assert(is_int);
return builder.CreateLShr(lhs, rhs, name);
}
// Right shift
if(op_==RIGHT_SHIFT){
assert(is_int);
return builder.CreateAShr(lhs, rhs, name);
}
// LT
if(op_ == LT && is_float)
return builder.CreateFCmpOLT(lhs, rhs, name);
if(op_ == LT && is_int && is_signed)
return builder.CreateICmpSLT(lhs, rhs, name);
if(op_ == LT && is_int && !is_signed)
return builder.CreateICmpULT(lhs, rhs, name);
// GT
if(op_ == GT && is_float)
return builder.CreateFCmpOGT(lhs, rhs, name);
if(op_ == GT && is_int && is_signed)
return builder.CreateICmpSGT(lhs, rhs, name);
if(op_ == GT && is_int && !is_signed)
return builder.CreateICmpUGT(lhs, rhs, name);
// LE
if(op_ == LE && is_float)
return builder.CreateFCmpOLE(lhs, rhs, name);
if(op_ == LE && is_int && is_signed)
return builder.CreateICmpSLE(lhs, rhs, name);
if(op_ == LE && is_int && !is_signed)
return builder.CreateICmpULE(lhs, rhs, name);
// GE
if(op_ == GE && is_float)
return builder.CreateFCmpOGE(lhs, rhs, name);
if(op_ == GE && is_int && is_signed)
return builder.CreateICmpSGE(lhs, rhs, name);
if(op_ == GE && is_int && !is_signed)
return builder.CreateICmpUGE(lhs, rhs, name);
// EQ
if(op_ == EQ && is_float)
return builder.CreateFCmpOEQ(lhs, rhs, name);
if(op_ == EQ && is_int)
return builder.CreateICmpEQ(lhs, rhs, name);
// NE
if(op_ == NE && is_float)
return builder.CreateFCmpONE(lhs, rhs, name);
if(op_ == NE && is_int)
return builder.CreateICmpNE(lhs, rhs, name);
// AND
if(op_ == AND){
assert(is_int);
return builder.CreateAnd(lhs, rhs, name);
}
if(op_ == XOR){
assert(is_int);
return builder.CreateXor(lhs, rhs, name);
}
if(op_ == OR){
assert(is_int);
return builder.CreateOr(lhs, rhs, name);
}
if(op_ == LAND){
assert(is_int);
return builder.CreateAnd(lhs, rhs, name);
}
if(op_ == LOR){
assert(is_int);
return builder.CreateOr(lhs, rhs, name);
}
assert(false && "unreachable");
throw;
}
Value* binary_operator::codegen(module *mod) const{
Value *lhs = lhs_->codegen(mod);
Value *rhs = rhs_->codegen(mod);
Value *result = llvm_op(mod->builder(), lhs, rhs, "");
return result;
}
/* Unary operator */
Value *unary_operator::llvm_op(llvm::IRBuilder<> &builder, Value *arg, const std::string &name) const{
Type *atype = arg->getType();
bool is_float = atype->isFloatingPointTy();
bool is_int = atype->isIntegerTy();
if(op_ == INC){
assert(is_int);
return builder.CreateAdd(arg, builder.getInt32(1), name);
}
if(op_ == DEC){
assert(is_int);
return builder.CreateSub(arg, builder.getInt32(1), name);
}
if(op_ == PLUS)
return arg;
if(op_ == MINUS && is_float)
return builder.CreateFNeg(arg, name);
if(op_ == MINUS && is_int)
return builder.CreateNeg(arg, name);
if(op_ == ADDR)
throw std::runtime_error("not supported");
if(op_ == DEREF)
return builder.CreateLoad(arg, name);
if(op_ == COMPL)
throw std::runtime_error("not supported");
if(op_ == NOT)
return builder.CreateNot(arg, name);
assert(false && "unrechable");
throw;
}
Value* unary_operator::codegen(module *mod) const{
Value *arg = arg_->codegen(mod);
Value *result = llvm_op(mod->builder(), arg, "");
return result;
}
/* Cast operator */
Value *cast_operator::llvm_op(IRBuilder<> &builder, Type *T, Value *arg, const std::string &name) const{
return nullptr;
}
Value* cast_operator::codegen(module *mod) const{
Value *arg = arg_->codegen(mod);
Type *T = T_->type(mod);
return llvm_op(mod->builder(), T, arg, "");
}
/* Conditional expression */
Value *conditional_expression::llvm_op(IRBuilder<> &builder, Value *cond, Value *true_value, Value *false_value, const std::string &name) const{
return nullptr;
}
Value *conditional_expression::codegen(module *mod) const{
Value *cond = cond_->codegen(mod);
Value *true_value = true_value_->codegen(mod);
Value *false_value = false_value_->codegen(mod);
return llvm_op(mod->builder(), cond, true_value, false_value, "");
}
/* Assignment expression */
Value *assignment_expression::llvm_op(llvm::IRBuilder<> &builder, Value *lvalue, Value *rvalue, const std::string &name) const{
return nullptr;
}
Value *assignment_expression::codegen(module *mod) const{
Value *lvalue = lvalue_->codegen(mod);
Value *rvalue = rvalue_->codegen(mod);
return llvm_op(mod->builder(), lvalue, rvalue, "");
}
/* Type name */
llvm::Type *type_name::type(module *mod) const{
return decl_->type(mod, spec_->type(mod));
}
/* String literal */
llvm::Value* string_literal::codegen(module *mod) const{
return ConstantDataArray::getString(mod->handle()->getContext(), value_);
}
/* Constant */
llvm::Value* constant::codegen(module *mod) const{
return mod->builder().getInt32(value_);
}
}