compiler: disambiguate function-local named types (#5336)

* compiler: make getTypeCodeName a method on compilerContext

* testdata: add regression tests for function-local named types

* compiler: disambiguate function-local named types

* Some PR feedback
This commit is contained in:
Jake Bailey
2026-06-05 02:04:44 -07:00
committed by GitHub
parent 41f666c670
commit 1b9bb143bf
10 changed files with 644 additions and 19 deletions
+276 -19
View File
@@ -10,6 +10,7 @@ import (
"fmt"
"go/token"
"go/types"
"path/filepath"
"sort"
"strconv"
"strings"
@@ -165,7 +166,7 @@ func (c *compilerContext) getTypeCode(typ types.Type) llvm.Value {
}
}
typeCodeName, isLocal := getTypeCodeName(typ)
typeCodeName, isLocal := c.getTypeCodeName(typ)
globalName := "reflect/types.type:" + typeCodeName
var global llvm.Value
if isLocal {
@@ -580,20 +581,50 @@ var basicTypeNames = [...]string{
// getTypeCodeName returns a name for this type that can be used in the
// interface lowering pass to assign type codes as expected by the reflect
// package. See getTypeCodeNum.
func getTypeCodeName(t types.Type) (string, bool) {
//
// isLocal is true when the type is declared inside a function body.
// Such types need a per-declaration (or per instantiation) suffix
// because their printed names are not unique.
//
// Ordinary function-local types (TypeName.Parent() != nil) are
// disambiguated lazily from their declaration position: every
// declaration in the package has a distinct (file, line, column)
// triple, and the position is taken un-//line-adjusted so it is
// stable across builds. Such types are only nameable inside their
// declaring package, so the name does not need to agree with anything
// computed in another package.
//
// Synthetic locals (TypeName.Parent() == nil), produced by generic
// instantiation, are pre-registered by scanLocalTypes because their
// names must agree across packages that materialize the same instance.
func (c *compilerContext) getTypeCodeName(t types.Type) (name string, isLocal bool) {
switch t := types.Unalias(t).(type) {
case *types.Named:
if t.Obj().Parent() != t.Obj().Pkg().Scope() {
return "named:" + t.String() + "$local", true
tn := t.Obj()
if tn.Pkg() == nil || tn.Parent() == tn.Pkg().Scope() {
// Package-scope or builtin: the printed name is unique.
return "named:" + t.String(), false
}
return "named:" + t.String(), false
if tn.Parent() != nil {
// Ordinary function-local type. Use the un-//line-adjusted
// declaration position as the disambiguator.
pos := c.program.Fset.PositionFor(tn.Pos(), false)
return fmt.Sprintf("named:%s$%s:%d:%d", t.String(), filepath.Base(pos.Filename), pos.Line, pos.Column), true
}
// Synthetic local from generic instantiation: must have been
// pre-registered by scanLocalTypes.
v := c.localTypeNames.At(t)
if v == nil {
panic("compiler: synthetic local type " + tn.Name() + " was not registered by scanLocalTypes")
}
return "named:" + v.(string), true
case *types.Array:
s, isLocal := getTypeCodeName(t.Elem())
s, isLocal := c.getTypeCodeName(t.Elem())
return "array:" + strconv.FormatInt(t.Len(), 10) + ":" + s, isLocal
case *types.Basic:
return "basic:" + basicTypeNames[t.Kind()], false
case *types.Chan:
s, isLocal := getTypeCodeName(t.Elem())
s, isLocal := c.getTypeCodeName(t.Elem())
var dir string
switch t.Dir() {
case types.SendOnly:
@@ -613,7 +644,7 @@ func getTypeCodeName(t types.Type) (string, bool) {
if !token.IsExported(name) {
name = t.Method(i).Pkg().Path() + "." + name
}
s, local := getTypeCodeName(t.Method(i).Type())
s, local := c.getTypeCodeName(t.Method(i).Type())
if local {
isLocal = true
}
@@ -621,17 +652,17 @@ func getTypeCodeName(t types.Type) (string, bool) {
}
return "interface:" + "{" + strings.Join(methods, ",") + "}", isLocal
case *types.Map:
keyType, keyLocal := getTypeCodeName(t.Key())
elemType, elemLocal := getTypeCodeName(t.Elem())
keyType, keyLocal := c.getTypeCodeName(t.Key())
elemType, elemLocal := c.getTypeCodeName(t.Elem())
return "map:" + "{" + keyType + "," + elemType + "}", keyLocal || elemLocal
case *types.Pointer:
s, isLocal := getTypeCodeName(t.Elem())
s, isLocal := c.getTypeCodeName(t.Elem())
return "pointer:" + s, isLocal
case *types.Signature:
isLocal := false
params := make([]string, t.Params().Len())
for i := 0; i < t.Params().Len(); i++ {
s, local := getTypeCodeName(t.Params().At(i).Type())
s, local := c.getTypeCodeName(t.Params().At(i).Type())
if local {
isLocal = true
}
@@ -639,7 +670,7 @@ func getTypeCodeName(t types.Type) (string, bool) {
}
results := make([]string, t.Results().Len())
for i := 0; i < t.Results().Len(); i++ {
s, local := getTypeCodeName(t.Results().At(i).Type())
s, local := c.getTypeCodeName(t.Results().At(i).Type())
if local {
isLocal = true
}
@@ -647,7 +678,7 @@ func getTypeCodeName(t types.Type) (string, bool) {
}
return "func:" + "{" + strings.Join(params, ",") + "}{" + strings.Join(results, ",") + "}", isLocal
case *types.Slice:
s, isLocal := getTypeCodeName(t.Elem())
s, isLocal := c.getTypeCodeName(t.Elem())
return "slice:" + s, isLocal
case *types.Struct:
elems := make([]string, t.NumFields())
@@ -657,7 +688,7 @@ func getTypeCodeName(t types.Type) (string, bool) {
if t.Field(i).Embedded() {
embedded = "#"
}
s, local := getTypeCodeName(t.Field(i).Type())
s, local := c.getTypeCodeName(t.Field(i).Type())
if local {
isLocal = true
}
@@ -672,6 +703,232 @@ func getTypeCodeName(t types.Type) (string, bool) {
}
}
// scanLocalTypes assigns names to every synthetic *types.TypeName
// (TypeName.Parent() == nil) reachable from this package and stores
// them in c.localTypeNames.
//
// Synthetic TypeNames are produced by generic instantiation: two
// instantiations of the same generic function (e.g. F[int] and
// F[string]) produce TypeNames with the same printed name and the
// same source position, so each is named with the enclosing
// instance's RelString as prefix. RelString encodes the type
// arguments, matching Go's runtime behavior, where F[int].Inner and
// F[string].Inner are distinct types even when Inner does not mention
// the type parameter.
//
// A given instance may be materialized by several packages (the body
// of F[int] is compiled in every package that calls F[int]); its
// reflect/types.type:* global has LinkOnceODRLinkage and is merged by
// name at link time. The chosen name therefore depends only on
// intrinsic SSA properties (RelString and the raw token.Pos used as a
// sort key), so any package compiling the same instance produces the
// same identifier.
//
// Ordinary function-local TypeNames (TypeName.Parent() != nil) are
// not handled here: they are nameable only inside their declaring
// package, and getTypeCodeName derives a stable per-declaration name
// for them directly from their source position.
func (c *compilerContext) scanLocalTypes(ssaPkg *ssa.Package) {
// Locate every generic instance reachable from this package
// (including instances declared in imported packages and any
// function reached through an instance subtree).
var instances []*ssa.Function
seen := map[*ssa.Function]struct{}{}
var walk func(fn *ssa.Function, inInstance bool)
walk = func(fn *ssa.Function, inInstance bool) {
if fn == nil {
return
}
if _, ok := seen[fn]; ok {
return
}
// fn belongs to an instance subtree if it is itself an
// instantiation or if we reached it from one.
//
// len(TypeArgs()) is used instead of fn.Origin() because
// Origin() may call Build() on fn's declaring package, which
// would defeat per-package compilation.
isInstanceRoot := len(fn.TypeArgs()) > 0
if !isInstanceRoot && !inInstance && fn.Pkg != ssaPkg {
return
}
if fn.Blocks == nil && fn.AnonFuncs == nil {
return
}
seen[fn] = struct{}{}
isInInstance := inInstance || isInstanceRoot
if isInInstance {
instances = append(instances, fn)
}
for _, anon := range fn.AnonFuncs {
walk(anon, isInInstance)
}
var ops [10]*ssa.Value
for _, b := range fn.Blocks {
for _, instr := range b.Instrs {
for _, op := range instr.Operands(ops[:0]) {
if op == nil || *op == nil {
continue
}
if callee, ok := (*op).(*ssa.Function); ok {
walk(callee, isInInstance)
}
}
}
}
}
for _, member := range ssaPkg.Members {
switch m := member.(type) {
case *ssa.Function:
walk(m, false)
case *ssa.Type:
mset := c.program.MethodSets.MethodSet(m.Type())
for i := 0; i < mset.Len(); i++ {
walk(c.program.MethodValue(mset.At(i)), false)
}
pmset := c.program.MethodSets.MethodSet(types.NewPointer(m.Type()))
for i := 0; i < pmset.Len(); i++ {
walk(c.program.MethodValue(pmset.At(i)), false)
}
}
}
// Registration is first-writer-wins (a synthetic TypeName may be
// reachable from several instances), so visit instances in a
// deterministic order. Pos() is a defensive tiebreaker.
sort.Slice(instances, func(i, j int) bool {
ri, rj := instances[i].RelString(nil), instances[j].RelString(nil)
if ri != rj {
return ri < rj
}
return instances[i].Pos() < instances[j].Pos()
})
for _, fn := range instances {
c.registerSyntheticLocalTypes(fn)
}
}
// registerSyntheticLocalTypes walks every type reachable from fn's
// body and records each synthetic *types.Named (TypeName.Parent() ==
// nil) in c.localTypeNames. Each is named with fn.RelString as the
// owning function plus a per-function counter assigned in source
// order.
//
// First-writer-wins: a *types.Named already present in
// c.localTypeNames is left alone, so a synthetic type reachable from
// several instances keeps the name assigned by the first (in
// scanLocalTypes' deterministic order).
func (c *compilerContext) registerSyntheticLocalTypes(fn *ssa.Function) {
var found []*types.Named
seen := map[types.Type]struct{}{}
var visit func(t types.Type)
visit = func(t types.Type) {
if t == nil {
return
}
if _, ok := seen[t]; ok {
return
}
seen[t] = struct{}{}
switch t := t.(type) {
case *types.Alias:
visit(types.Unalias(t))
case *types.Named:
tn := t.Obj()
if tn.Pkg() != nil && tn.Parent() == nil {
if c.localTypeNames.At(t) == nil {
// Reserve the slot so later calls within this
// scanLocalTypes invocation skip it; the final
// name is filled in after sorting, before any
// getTypeCodeName lookups happen.
c.localTypeNames.Set(t, "")
found = append(found, t)
}
}
targs := t.TypeArgs()
for i := 0; i < targs.Len(); i++ {
visit(targs.At(i))
}
visit(t.Underlying())
case *types.Pointer:
visit(t.Elem())
case *types.Slice:
visit(t.Elem())
case *types.Array:
visit(t.Elem())
case *types.Chan:
visit(t.Elem())
case *types.Map:
visit(t.Key())
visit(t.Elem())
case *types.Struct:
for i := 0; i < t.NumFields(); i++ {
visit(t.Field(i).Type())
}
case *types.Signature:
if p := t.Params(); p != nil {
for i := 0; i < p.Len(); i++ {
visit(p.At(i).Type())
}
}
if r := t.Results(); r != nil {
for i := 0; i < r.Len(); i++ {
visit(r.At(i).Type())
}
}
case *types.Tuple:
for i := 0; i < t.Len(); i++ {
visit(t.At(i).Type())
}
case *types.Interface:
// A synthetic local type can be reachable only through a
// local interface's method signature, so descend into
// them. getTypeCodeName encodes those signatures into
// the interface's identifier, and the seen map breaks
// cycles formed by methods that mention the interface
// itself.
for i := 0; i < t.NumMethods(); i++ {
visit(t.Method(i).Type())
}
}
}
for _, p := range fn.Params {
visit(p.Type())
}
for _, fv := range fn.FreeVars {
visit(fv.Type())
}
for _, l := range fn.Locals {
visit(l.Type())
}
var ops [10]*ssa.Value
for _, b := range fn.Blocks {
for _, instr := range b.Instrs {
if v, ok := instr.(ssa.Value); ok {
visit(v.Type())
}
for _, op := range instr.Operands(ops[:0]) {
if op != nil && *op != nil {
visit((*op).Type())
}
}
}
}
if len(found) == 0 {
return
}
// Sort by raw token.Pos: this gives a total order on declarations
// that is stable across builds and unaffected by //line directives
// (which only adjust the human-facing position from Fset.Position).
sort.Slice(found, func(i, j int) bool {
return found[i].Obj().Pos() < found[j].Obj().Pos()
})
enclosing := fn.RelString(nil)
for i, named := range found {
c.localTypeNames.Set(named, fmt.Sprintf("%s.%s$%d", enclosing, named.Obj().Name(), i))
}
}
// getTypeMethodSet returns a reference (GEP) to a global method set. This
// method set should be unreferenced after the interface lowering pass.
func (c *compilerContext) getTypeMethodSet(typ types.Type) llvm.Value {
@@ -769,7 +1026,7 @@ func (b *builder) createTypeAssert(expr *ssa.TypeAssert) llvm.Value {
commaOk = b.createInterfaceTypeAssert(intf, actualTypeNum)
}
} else {
name, _ := getTypeCodeName(expr.AssertedType)
name, _ := b.getTypeCodeName(expr.AssertedType)
globalName := "reflect/types.typeid:" + name
assertedTypeCodeGlobal := b.mod.NamedGlobal(globalName)
if assertedTypeCodeGlobal.IsNil() {
@@ -882,7 +1139,7 @@ func (c *compilerContext) getMethodSetValue(methods []*types.Func) llvm.Value {
if !token.IsExported(name) {
name = method.Pkg().Path() + "." + name
}
s, _ := getTypeCodeName(method.Type())
s, _ := c.getTypeCodeName(method.Type())
globalName := "reflect/types.signature:" + name + ":" + s
value := c.mod.NamedGlobal(globalName)
if value.IsNil() {
@@ -926,7 +1183,7 @@ func (c *compilerContext) getMethodSetValue(methods []*types.Func) llvm.Value {
// thunk is declared, not defined: it will be defined by the interface lowering
// pass.
func (c *compilerContext) getInvokeFunction(instr *ssa.CallCommon) llvm.Value {
s, _ := getTypeCodeName(instr.Value.Type().Underlying())
s, _ := c.getTypeCodeName(instr.Value.Type().Underlying())
fnName := s + "." + instr.Method.Name() + "$invoke"
llvmFn := c.mod.NamedFunction(fnName)
if llvmFn.IsNil() {
@@ -951,7 +1208,7 @@ func (c *compilerContext) getInvokeFunction(instr *ssa.CallCommon) llvm.Value {
// by the interface lowering pass as a type-ID comparison chain, avoiding the
// need for runtime.typeImplementsMethodSet at compile time.
func (b *builder) createInterfaceTypeAssert(intf *types.Interface, actualType llvm.Value) llvm.Value {
s, _ := getTypeCodeName(intf)
s, _ := b.getTypeCodeName(intf)
fnName := s + ".$typeassert"
llvmFn := b.mod.NamedFunction(fnName)
if llvmFn.IsNil() {