// parseFunc parses the given function and attempts to construct an equivalent
// Go function declaration AST node.
func parseFunc(graph *dot.Graph, module llvm.Module, funcName string, hprims []*xprimitive.Primitive) (*ast.FuncDecl, error) {
llFunc := module.NamedFunction(funcName)
if llFunc.IsNil() {
return nil, errutil.Newf("unable to locate function %q", funcName)
}
if llFunc.IsDeclaration() {
return nil, errutil.Newf("unable to create AST for %q; expected function definition, got function declaration (e.g. no body)", funcName)
}
// Parse each basic block.
bbs := make(map[string]BasicBlock)
for _, llBB := range llFunc.BasicBlocks() {
bb, err := parseBasicBlock(llBB)
if err != nil {
return nil, err
}
bbs[bb.Name()] = bb
if flagVerbose && !flagQuiet {
printBB(bb)
}
}
// Replace PHI instructions with assignment statements in the appropriate
// basic blocks.
for _, bb := range bbs {
block, ok := bb.(*basicBlock)
if !ok {
return nil, errutil.Newf("invalid basic block type; expected *basicBlock, got %T", bb)
}
for ident, defs := range block.phis {
for _, def := range defs {
assign := &ast.AssignStmt{
Lhs: []ast.Expr{newIdent(ident)},
Tok: token.ASSIGN,
Rhs: []ast.Expr{def.expr},
}
bbSrc := bbs[def.bb]
stmts := bbSrc.Stmts()
stmts = append(stmts, assign)
bbSrc.SetStmts(stmts)
}
}
}
// Perform control flow analysis.
body, err := restructure(graph, bbs, hprims)
if err != nil {
return nil, errutil.Err(err)
}
sig := &ast.FuncType{
Params: &ast.FieldList{},
}
if funcName != "main" {
// TODO: Implement parsing of function signature.
}
return createFunc(funcName, sig, body)
}
// createCFG generates a control flow graph for the given function using one
// node per basic block.
func createCFG(module llvm.Module, funcName string) (*dot.Graph, error) {
f := module.NamedFunction(funcName)
if f.IsNil() {
return nil, errutil.Newf("unable to locate function %q", funcName)
}
if f.IsDeclaration() {
return nil, errutil.Newf("unable to generate CFG for %q; expected function definition, got function declaration (e.g. no body)", funcName)
}
// Create a new directed graph.
graph := dot.NewGraph()
graph.SetDir(true)
graph.SetName(funcName)
// Populate the graph with one node per basic block.
for _, bb := range f.BasicBlocks() {
// Add node (i.e. basic block) to the graph.
bbName, err := getBBName(bb.AsValue())
if err != nil {
return nil, errutil.Err(err)
}
if bb == f.EntryBasicBlock() {
attrs := map[string]string{"label": "entry"}
graph.AddNode(funcName, bbName, attrs)
} else {
graph.AddNode(funcName, bbName, nil)
}
// Add edges from node (i.e. target basic blocks) to the graph.
term := bb.LastInstruction()
nops := term.OperandsCount()
switch opcode := term.InstructionOpcode(); opcode {
case llvm.Ret:
// exit node.
// ret <type> <value>
// ret void
case llvm.Br:
switch nops {
case 1:
// unconditional branch.
// br label <target>
target := term.Operand(0)
targetName, err := getBBName(target)
if err != nil {
return nil, errutil.Err(err)
}
graph.AddEdge(bbName, targetName, true, nil)
case 3:
// 2-way conditional branch.
// br i1 <cond>, label <target_true>, label <target_false>
// NOTE: The LLVM library has a peculiar way of ordering the operands:
// term.Operand(0) refers to the 1st operand
// term.Operand(1) refers to the 3rd operand
// term.Operand(2) refers to the 2nd operand
// TODO: Make the order logical with the pure Go implementation of
// LLVM IR.
targetTrue, targetFalse := term.Operand(2), term.Operand(1)
targetTrueName, err := getBBName(targetTrue)
if err != nil {
return nil, errutil.Err(err)
}
targetFalseName, err := getBBName(targetFalse)
if err != nil {
return nil, errutil.Err(err)
}
attrs := map[string]string{"label": "false"}
graph.AddEdge(bbName, targetFalseName, true, attrs)
attrs = map[string]string{"label": "true"}
graph.AddEdge(bbName, targetTrueName, true, attrs)
default:
return nil, errutil.Newf("invalid number of operands (%d) for br instruction", nops)
}
case llvm.Switch:
// n-way conditional branch.
// switch <type> <value>, label <default_target> [
// <type> <case1>, label <case1_target>
// <type> <case2>, label <case2_target>
// ...
// ]
if nops < 2 {
return nil, errutil.Newf("invalid number of operands (%d) for switch instruction", nops)
}
// Default branch.
targetDefault := term.Operand(1)
targetDefaultName, err := getBBName(targetDefault)
if err != nil {
return nil, errutil.Err(err)
}
attrs := map[string]string{"label": "default"}
graph.AddEdge(bbName, targetDefaultName, true, attrs)
// Case branches.
for i := 3; i < nops; i += 2 {
// Case branch.
targetCase := term.Operand(i)
targetCaseName, err := getBBName(targetCase)
if err != nil {
return nil, errutil.Err(err)
}
caseID := (i - 3) / 2
label := fmt.Sprintf("case %d", caseID)
attrs := map[string]string{"label": label}
graph.AddEdge(bbName, targetCaseName, true, attrs)
}
case llvm.Unreachable:
// unreachable node.
// unreachable
default:
// TODO: Implement support for:
// - indirectbr
// - invoke
// - resume
panic(fmt.Sprintf("not yet implemented; support for terminator %v", opcode))
}
}
return graph, nil
}