From 79cd1687e6abf8c565281d310b4c2b44a8d4bb84 Mon Sep 17 00:00:00 2001 From: Matthew Dempsky Date: Thu, 13 May 2021 20:23:13 -0700 Subject: [PATCH] [dev.typeparams] cmd/compile: unified IR construction This CL adds a new unified IR construction mode to the frontend. It's purely additive, and all files include "UNREVIEWED" at the top, like how types2 was initially imported. The next CL adds a -d=unified flag to actually enable unified IR mode. See below for more details, but some highlights: 1. It adds ~6kloc (excluding enum listings and stringer output), but I estimate it will allow removing ~14kloc (see CL 324670, including its commit message); 2. When enabled by default, it passes more tests than -G=3 does (see CL 325213 and CL 324673); 3. Without requiring any new code, it supports inlining of more code than the current inliner (see CL 324574; contrast CL 283112 and CL 266203, which added support for inlining function literals and type switches, respectively); 4. Aside from dictionaries (which I intend to add still), its support for generics is more complete (e.g., it fully supports local types, including local generic types within generic functions and instantiating generic types with local types; see test/typeparam/nested.go); 5. It supports lazy loading of types and objects for types2 type checking; 6. It supports re-exporting of types, objects, and inline bodies without needing to parse them into IR; 7. The new export data format has extensive support for debugging with "sync" markers, so mistakes during development are easier to catch; 8. When compiling with -d=inlfuncswithclosures=0, it enables "quirks mode" where it generates output that passes toolstash -cmp. -- The new unified IR pipeline combines noding, stenciling, inlining, and import/export into a single, shared code path. Previously, IR trees went through multiple phases of copying during compilation: 1. "Noding": the syntax AST is copied into the initial IR form. To support generics, there's now also "irgen", which implements the same idea, but takes advantage of types2 type-checking results to more directly construct IR. 2. "Stenciling": generic IR forms are copied into instantiated IR forms, substituting type parameters as appropriate. 3. "Inlining": the inliner made backup copies of inlinable functions, and then copied them again when inlining into a call site, with some modifications (e.g., updating position information, rewriting variable references, changing "return" statements into "goto"). 4. "Importing/exporting": the exporter wrote out the IR as saved by the inliner, and then the importer read it back as to be used by the inliner again. Normal functions are imported/exported "desugared", while generic functions are imported/exported in source form. These passes are all conceptually the same thing: make a copy of a function body, maybe with some minor changes/substitutions. However, they're all completely separate implementations that frequently run into the same issues because IR has many nuanced corner cases. For example, inlining currently doesn't support local defined types, "range" loops, or labeled "for"/"switch" statements, because these require special handling around Sym references. We've recently extended the inliner to support new features like inlining type switches and function literals, and they've had issues. The exporter only knows how to export from IR form, so when re-exporting inlinable functions (e.g., methods on imported types that are exposed via exported APIs), these functions may need to be imported as IR for the sole purpose of being immediately exported back out again. By unifying all of these modes of copying into a single code path that cleanly separates concerns, we eliminate many of these possible issues. Some recent examples: 1. Issues #45743 and #46472 were issues where type switches were mishandled by inlining and stenciling, respectively; but neither of these affected unified IR, because it constructs type switches using the exact same code as for normal functions. 2. CL 325409 fixes an issue in stenciling with implicit conversion of values of type-parameter type to variables of interface type, but this issue did not affect unified IR. Change-Id: I5a05991fe16d68bb0f712503e034cb9f2d19e296 Reviewed-on: https://go-review.googlesource.com/c/go/+/324573 Trust: Matthew Dempsky Trust: Robert Griesemer Run-TryBot: Matthew Dempsky TryBot-Result: Go Bot Reviewed-by: Robert Griesemer --- src/cmd/compile/internal/noder/codes.go | 126 ++ src/cmd/compile/internal/noder/decoder.go | 243 ++ src/cmd/compile/internal/noder/encoder.go | 245 ++ src/cmd/compile/internal/noder/linker.go | 296 +++ src/cmd/compile/internal/noder/quirks.go | 453 ++++ src/cmd/compile/internal/noder/reader.go | 1970 +++++++++++++++++ src/cmd/compile/internal/noder/reader2.go | 463 ++++ src/cmd/compile/internal/noder/reloc.go | 40 + src/cmd/compile/internal/noder/sync.go | 154 ++ .../internal/noder/syncmarker_string.go | 152 ++ src/cmd/compile/internal/noder/unified.go | 276 +++ src/cmd/compile/internal/noder/writer.go | 1746 +++++++++++++++ 12 files changed, 6164 insertions(+) create mode 100644 src/cmd/compile/internal/noder/codes.go create mode 100644 src/cmd/compile/internal/noder/decoder.go create mode 100644 src/cmd/compile/internal/noder/encoder.go create mode 100644 src/cmd/compile/internal/noder/linker.go create mode 100644 src/cmd/compile/internal/noder/quirks.go create mode 100644 src/cmd/compile/internal/noder/reader.go create mode 100644 src/cmd/compile/internal/noder/reader2.go create mode 100644 src/cmd/compile/internal/noder/reloc.go create mode 100644 src/cmd/compile/internal/noder/sync.go create mode 100644 src/cmd/compile/internal/noder/syncmarker_string.go create mode 100644 src/cmd/compile/internal/noder/unified.go create mode 100644 src/cmd/compile/internal/noder/writer.go diff --git a/src/cmd/compile/internal/noder/codes.go b/src/cmd/compile/internal/noder/codes.go new file mode 100644 index 00000000000..4a6a4e83071 --- /dev/null +++ b/src/cmd/compile/internal/noder/codes.go @@ -0,0 +1,126 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +type code interface { + marker() syncMarker + value() int +} + +type codeVal int + +func (c codeVal) marker() syncMarker { return syncVal } +func (c codeVal) value() int { return int(c) } + +const ( + valBool codeVal = iota + valString + valInt64 + valBigInt + valBigRat + valBigFloat +) + +type codeType int + +func (c codeType) marker() syncMarker { return syncType } +func (c codeType) value() int { return int(c) } + +const ( + typeBasic codeType = iota + typeNamed + typePointer + typeSlice + typeArray + typeChan + typeMap + typeSignature + typeStruct + typeInterface + typeUnion + typeTypeParam +) + +type codeObj int + +func (c codeObj) marker() syncMarker { return syncCodeObj } +func (c codeObj) value() int { return int(c) } + +const ( + objAlias codeObj = iota + objConst + objType + objFunc + objVar + objStub +) + +type codeStmt int + +func (c codeStmt) marker() syncMarker { return syncStmt1 } +func (c codeStmt) value() int { return int(c) } + +const ( + stmtEnd codeStmt = iota + stmtLabel + stmtBlock + stmtExpr + stmtSend + stmtAssign + stmtAssignOp + stmtIncDec + stmtBranch + stmtCall + stmtReturn + stmtIf + stmtFor + stmtSwitch + stmtSelect + + // TODO(mdempsky): Remove after we don't care about toolstash -cmp. + stmtTypeDeclHack +) + +type codeExpr int + +func (c codeExpr) marker() syncMarker { return syncExpr } +func (c codeExpr) value() int { return int(c) } + +// TODO(mdempsky): Split expr into addr, for lvalues. +const ( + exprNone codeExpr = iota + exprConst + exprType // type expression + exprLocal // local variable + exprName // global variable or function + exprBlank + exprCompLit + exprFuncLit + exprSelector + exprIndex + exprSlice + exprAssert + exprUnaryOp + exprBinaryOp + exprCall + + // TODO(mdempsky): Handle in switchStmt directly instead. + exprTypeSwitchGuard +) + +type codeDecl int + +func (c codeDecl) marker() syncMarker { return syncDecl } +func (c codeDecl) value() int { return int(c) } + +const ( + declEnd codeDecl = iota + declFunc + declMethod + declVar + declOther +) diff --git a/src/cmd/compile/internal/noder/decoder.go b/src/cmd/compile/internal/noder/decoder.go new file mode 100644 index 00000000000..023388875c4 --- /dev/null +++ b/src/cmd/compile/internal/noder/decoder.go @@ -0,0 +1,243 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "encoding/binary" + "fmt" + "go/constant" + "go/token" + "math/big" + "os" + "strings" + + "cmd/compile/internal/base" +) + +type pkgDecoder struct { + pkgPath string + + elemEndsEnds [numRelocs]uint32 + elemEnds []uint32 + elemData string +} + +func newPkgDecoder(pkgPath, input string) pkgDecoder { + pr := pkgDecoder{ + pkgPath: pkgPath, + } + + // TODO(mdempsky): Implement direct indexing of input string to + // avoid copying the position information. + + r := strings.NewReader(input) + + assert(binary.Read(r, binary.LittleEndian, pr.elemEndsEnds[:]) == nil) + + pr.elemEnds = make([]uint32, pr.elemEndsEnds[len(pr.elemEndsEnds)-1]) + assert(binary.Read(r, binary.LittleEndian, pr.elemEnds[:]) == nil) + + pos, err := r.Seek(0, os.SEEK_CUR) + assert(err == nil) + + pr.elemData = input[pos:] + assert(len(pr.elemData) == int(pr.elemEnds[len(pr.elemEnds)-1])) + + return pr +} + +func (pr *pkgDecoder) numElems(k reloc) int { + count := int(pr.elemEndsEnds[k]) + if k > 0 { + count -= int(pr.elemEndsEnds[k-1]) + } + return count +} + +func (pr *pkgDecoder) totalElems() int { + return len(pr.elemEnds) +} + +func (pr *pkgDecoder) absIdx(k reloc, idx int) int { + absIdx := idx + if k > 0 { + absIdx += int(pr.elemEndsEnds[k-1]) + } + if absIdx >= int(pr.elemEndsEnds[k]) { + base.Fatalf("%v:%v is out of bounds; %v", k, idx, pr.elemEndsEnds) + } + return absIdx +} + +func (pr *pkgDecoder) dataIdx(k reloc, idx int) string { + absIdx := pr.absIdx(k, idx) + + var start uint32 + if absIdx > 0 { + start = pr.elemEnds[absIdx-1] + } + end := pr.elemEnds[absIdx] + + return pr.elemData[start:end] +} + +func (pr *pkgDecoder) stringIdx(idx int) string { + return pr.dataIdx(relocString, idx) +} + +func (pr *pkgDecoder) newDecoder(k reloc, idx int, marker syncMarker) decoder { + r := pr.newDecoderRaw(k, idx) + r.sync(marker) + return r +} + +func (pr *pkgDecoder) newDecoderRaw(k reloc, idx int) decoder { + r := decoder{ + common: pr, + k: k, + idx: idx, + } + + // TODO(mdempsky) r.data.Reset(...) after #44505 is resolved. + r.data = *strings.NewReader(pr.dataIdx(k, idx)) + + r.sync(syncRelocs) + r.relocs = make([]relocEnt, r.len()) + for i := range r.relocs { + r.sync(syncReloc) + r.relocs[i] = relocEnt{reloc(r.len()), r.len()} + } + + return r +} + +type decoder struct { + common *pkgDecoder + + relocs []relocEnt + data strings.Reader + + k reloc + idx int +} + +func (r *decoder) checkErr(err error) { + if err != nil { + base.Fatalf("unexpected error: %v", err) + } +} + +func (r *decoder) sync(m syncMarker) { + if debug { + pos, err0 := r.data.Seek(0, os.SEEK_CUR) + x, err := r.data.ReadByte() + r.checkErr(err) + if x != byte(m) { + // TODO(mdempsky): Revisit this error message, and make it more + // useful (e.g., include r.p.pkgPath). + base.Fatalf("data sync error: found %v at %v (%v) in (%v:%v), but expected %v", syncMarker(x), pos, err0, r.k, r.idx, m) + } + } +} + +func (r *decoder) bool() bool { + r.sync(syncBool) + x, err := r.data.ReadByte() + r.checkErr(err) + assert(x < 2) + return x != 0 +} + +func (r *decoder) int64() int64 { + r.sync(syncInt64) + x, err := binary.ReadVarint(&r.data) + r.checkErr(err) + return x +} + +func (r *decoder) uint64() uint64 { + r.sync(syncUint64) + x, err := binary.ReadUvarint(&r.data) + r.checkErr(err) + return x +} + +func (r *decoder) len() int { x := r.uint64(); v := int(x); assert(uint64(v) == x); return v } +func (r *decoder) int() int { x := r.int64(); v := int(x); assert(int64(v) == x); return v } +func (r *decoder) uint() uint { x := r.uint64(); v := uint(x); assert(uint64(v) == x); return v } + +func (r *decoder) code(mark syncMarker) int { + r.sync(mark) + return r.len() +} + +func (r *decoder) reloc(k reloc) int { + r.sync(syncUseReloc) + idx := r.len() + + e := r.relocs[idx] + assert(e.kind == k) + return e.idx +} + +func (r *decoder) string() string { + r.sync(syncString) + return r.common.stringIdx(r.reloc(relocString)) +} + +func (r *decoder) strings() []string { + res := make([]string, r.len()) + for i := range res { + res[i] = r.string() + } + return res +} + +func (r *decoder) rawValue() constant.Value { + isComplex := r.bool() + val := r.scalar() + if isComplex { + val = constant.BinaryOp(val, token.ADD, constant.MakeImag(r.scalar())) + } + return val +} + +func (r *decoder) scalar() constant.Value { + switch tag := codeVal(r.code(syncVal)); tag { + default: + panic(fmt.Sprintf("unexpected scalar tag: %v", tag)) + + case valBool: + return constant.MakeBool(r.bool()) + case valString: + return constant.MakeString(r.string()) + case valInt64: + return constant.MakeInt64(r.int64()) + case valBigInt: + return constant.Make(r.bigInt()) + case valBigRat: + num := r.bigInt() + denom := r.bigInt() + return constant.Make(new(big.Rat).SetFrac(num, denom)) + case valBigFloat: + return constant.Make(r.bigFloat()) + } +} + +func (r *decoder) bigInt() *big.Int { + v := new(big.Int).SetBytes([]byte(r.string())) + if r.bool() { + v.Neg(v) + } + return v +} + +func (r *decoder) bigFloat() *big.Float { + v := new(big.Float).SetPrec(512) + assert(v.UnmarshalText([]byte(r.string())) == nil) + return v +} diff --git a/src/cmd/compile/internal/noder/encoder.go b/src/cmd/compile/internal/noder/encoder.go new file mode 100644 index 00000000000..dc288dc29fe --- /dev/null +++ b/src/cmd/compile/internal/noder/encoder.go @@ -0,0 +1,245 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "bytes" + "encoding/binary" + "fmt" + "go/constant" + "io" + "math/big" + + "cmd/compile/internal/base" +) + +type pkgEncoder struct { + elems [numRelocs][]string + + stringsIdx map[string]int +} + +func newPkgEncoder() pkgEncoder { + return pkgEncoder{ + stringsIdx: make(map[string]int), + } +} + +func (pw *pkgEncoder) dump(out io.Writer) { + writeUint32 := func(x uint32) { + assert(binary.Write(out, binary.LittleEndian, x) == nil) + } + + var sum uint32 + for _, elems := range &pw.elems { + sum += uint32(len(elems)) + writeUint32(sum) + } + + sum = 0 + for _, elems := range &pw.elems { + for _, elem := range elems { + sum += uint32(len(elem)) + writeUint32(sum) + } + } + + for _, elems := range &pw.elems { + for _, elem := range elems { + _, err := io.WriteString(out, elem) + assert(err == nil) + } + } +} + +func (pw *pkgEncoder) stringIdx(s string) int { + if idx, ok := pw.stringsIdx[s]; ok { + assert(pw.elems[relocString][idx] == s) + return idx + } + + idx := len(pw.elems[relocString]) + pw.elems[relocString] = append(pw.elems[relocString], s) + pw.stringsIdx[s] = idx + return idx +} + +func (pw *pkgEncoder) newEncoder(k reloc, marker syncMarker) encoder { + e := pw.newEncoderRaw(k) + e.sync(marker) + return e +} + +func (pw *pkgEncoder) newEncoderRaw(k reloc) encoder { + idx := len(pw.elems[k]) + pw.elems[k] = append(pw.elems[k], "") // placeholder + + return encoder{ + p: pw, + k: k, + idx: idx, + } +} + +// Encoders + +type encoder struct { + p *pkgEncoder + + relocs []relocEnt + data bytes.Buffer + + k reloc + idx int +} + +func (w *encoder) flush() int { + var sb bytes.Buffer // TODO(mdempsky): strings.Builder after #44505 is resolved + + // Backup the data so we write the relocations at the front. + var tmp bytes.Buffer + io.Copy(&tmp, &w.data) + + // TODO(mdempsky): Consider writing these out separately so they're + // easier to strip, along with function bodies, so that we can prune + // down to just the data that's relevant to go/types. + w.sync(syncRelocs) + w.len(len(w.relocs)) + for _, rent := range w.relocs { + w.sync(syncReloc) + w.len(int(rent.kind)) + w.len(rent.idx) + } + + io.Copy(&sb, &w.data) + io.Copy(&sb, &tmp) + w.p.elems[w.k][w.idx] = sb.String() + + return w.idx +} + +func (w *encoder) checkErr(err error) { + if err != nil { + base.Fatalf("unexpected error: %v", err) + } +} + +func (w *encoder) sync(m syncMarker) { + if debug { + err := w.data.WriteByte(byte(m)) + w.checkErr(err) + } +} + +func (w *encoder) bool(b bool) bool { + w.sync(syncBool) + var x byte + if b { + x = 1 + } + err := w.data.WriteByte(x) + w.checkErr(err) + return b +} + +func (w *encoder) int64(x int64) { + w.sync(syncInt64) + var buf [binary.MaxVarintLen64]byte + n := binary.PutVarint(buf[:], x) + _, err := w.data.Write(buf[:n]) + w.checkErr(err) +} + +func (w *encoder) uint64(x uint64) { + w.sync(syncUint64) + var buf [binary.MaxVarintLen64]byte + n := binary.PutUvarint(buf[:], x) + _, err := w.data.Write(buf[:n]) + w.checkErr(err) +} + +func (w *encoder) len(x int) { assert(x >= 0); w.uint64(uint64(x)) } +func (w *encoder) int(x int) { w.int64(int64(x)) } +func (w *encoder) uint(x uint) { w.uint64(uint64(x)) } + +func (w *encoder) reloc(r reloc, idx int) { + w.sync(syncUseReloc) + + // TODO(mdempsky): Use map for lookup. + for i, rent := range w.relocs { + if rent.kind == r && rent.idx == idx { + w.len(i) + return + } + } + + w.len(len(w.relocs)) + w.relocs = append(w.relocs, relocEnt{r, idx}) +} + +func (w *encoder) code(c code) { + w.sync(c.marker()) + w.len(c.value()) +} + +func (w *encoder) string(s string) { + w.sync(syncString) + w.reloc(relocString, w.p.stringIdx(s)) +} + +func (w *encoder) strings(ss []string) { + w.len(len(ss)) + for _, s := range ss { + w.string(s) + } +} + +func (w *encoder) rawValue(val constant.Value) { + if w.bool(val.Kind() == constant.Complex) { + w.scalar(constant.Real(val)) + w.scalar(constant.Imag(val)) + } else { + w.scalar(val) + } +} + +func (w *encoder) scalar(val constant.Value) { + switch v := constant.Val(val).(type) { + default: + panic(fmt.Sprintf("unhandled %v (%v)", val, val.Kind())) + case bool: + w.code(valBool) + w.bool(v) + case string: + w.code(valString) + w.string(v) + case int64: + w.code(valInt64) + w.int64(v) + case *big.Int: + w.code(valBigInt) + w.bigInt(v) + case *big.Rat: + w.code(valBigRat) + w.bigInt(v.Num()) + w.bigInt(v.Denom()) + case *big.Float: + w.code(valBigFloat) + w.bigFloat(v) + } +} + +func (w *encoder) bigInt(v *big.Int) { + b := v.Bytes() + w.string(string(b)) // TODO: More efficient encoding. + w.bool(v.Sign() < 0) +} + +func (w *encoder) bigFloat(v *big.Float) { + b := v.Append(nil, 'p', -1) + w.string(string(b)) // TODO: More efficient encoding. +} diff --git a/src/cmd/compile/internal/noder/linker.go b/src/cmd/compile/internal/noder/linker.go new file mode 100644 index 00000000000..324902d246e --- /dev/null +++ b/src/cmd/compile/internal/noder/linker.go @@ -0,0 +1,296 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "io" + + "cmd/compile/internal/base" + "cmd/compile/internal/ir" + "cmd/compile/internal/reflectdata" + "cmd/compile/internal/types" + "cmd/internal/goobj" + "cmd/internal/obj" +) + +// This file implements the unified IR linker, which combines the +// local package's stub data with imported package data to produce a +// complete export data file. It also rewrites the compiler's +// extension data sections based on the results of compilation (e.g., +// the function inlining cost and linker symbol index assignments). +// +// TODO(mdempsky): Using the name "linker" here is confusing, because +// readers are likely to mistake references to it for cmd/link. But +// there's a shortage of good names for "something that combines +// multiple parts into a cohesive whole"... e.g., "assembler" and +// "compiler" are also already taken. + +type linker struct { + pw pkgEncoder + + pkgs map[string]int + decls map[*types.Sym]int +} + +func (l *linker) relocAll(pr *pkgReader, relocs []relocEnt) []relocEnt { + res := make([]relocEnt, len(relocs)) + for i, rent := range relocs { + rent.idx = l.relocIdx(pr, rent.kind, rent.idx) + res[i] = rent + } + return res +} + +func (l *linker) relocIdx(pr *pkgReader, k reloc, idx int) int { + assert(pr != nil) + + absIdx := pr.absIdx(k, idx) + + if newidx := pr.newindex[absIdx]; newidx != 0 { + return ^newidx + } + + var newidx int + switch k { + case relocString: + newidx = l.relocString(pr, idx) + case relocPkg: + newidx = l.relocPkg(pr, idx) + case relocObj: + newidx = l.relocObj(pr, idx) + + default: + // Generic relocations. + // + // TODO(mdempsky): Deduplicate more sections? In fact, I think + // every section could be deduplicated. This would also be easier + // if we do external relocations. + + w := l.pw.newEncoderRaw(k) + l.relocCommon(pr, &w, k, idx) + newidx = w.idx + } + + pr.newindex[absIdx] = ^newidx + + return newidx +} + +func (l *linker) relocString(pr *pkgReader, idx int) int { + return l.pw.stringIdx(pr.stringIdx(idx)) +} + +func (l *linker) relocPkg(pr *pkgReader, idx int) int { + path := pr.peekPkgPath(idx) + + if newidx, ok := l.pkgs[path]; ok { + return newidx + } + + r := pr.newDecoder(relocPkg, idx, syncPkgDef) + w := l.pw.newEncoder(relocPkg, syncPkgDef) + l.pkgs[path] = w.idx + + // TODO(mdempsky): We end up leaving an empty string reference here + // from when the package was originally written as "". Probably not + // a big deal, but a little annoying. Maybe relocating + // cross-references in place is the way to go after all. + w.relocs = l.relocAll(pr, r.relocs) + + _ = r.string() // original path + w.string(path) + + io.Copy(&w.data, &r.data) + + return w.flush() +} + +func (l *linker) relocObj(pr *pkgReader, idx int) int { + path, name, tag, _ := pr.peekObj(idx) + sym := types.NewPkg(path, "").Lookup(name) + + if newidx, ok := l.decls[sym]; ok { + return newidx + } + + if tag == objStub && path != "builtin" && path != "unsafe" { + pri, ok := objReader[sym] + if !ok { + base.Fatalf("missing reader for %q.%v", path, name) + } + assert(ok) + + pr = pri.pr + idx = pri.idx + + path2, name2, tag2, _ := pr.peekObj(idx) + sym2 := types.NewPkg(path2, "").Lookup(name2) + assert(sym == sym2) + assert(tag2 != objStub) + } + + w := l.pw.newEncoderRaw(relocObj) + bside := l.pw.newEncoderRaw(relocObjExt) + assert(bside.idx == w.idx) + l.decls[sym] = w.idx + + l.relocCommon(pr, &w, relocObj, idx) + + var obj *ir.Name + if path == "" { + var ok bool + obj, ok = sym.Def.(*ir.Name) + + // Generic types and functions won't have definitions. + // For now, just generically copy their extension data. + if !ok && base.Flag.G == 0 { + base.Fatalf("missing definition for %v", sym) + } + } + + if obj != nil { + bside.sync(syncObject1) + switch tag { + case objFunc: + l.relocFuncExt(&bside, obj) + case objType: + l.relocTypeExt(&bside, obj) + case objVar: + l.relocVarExt(&bside, obj) + } + bside.flush() + } else { + l.relocCommon(pr, &bside, relocObjExt, idx) + } + + return w.idx +} + +func (l *linker) relocCommon(pr *pkgReader, w *encoder, k reloc, idx int) { + r := pr.newDecoderRaw(k, idx) + w.relocs = l.relocAll(pr, r.relocs) + io.Copy(&w.data, &r.data) + w.flush() +} + +func (l *linker) pragmaFlag(w *encoder, pragma ir.PragmaFlag) { + w.sync(syncPragma) + w.int(int(pragma)) +} + +func (l *linker) relocFuncExt(w *encoder, name *ir.Name) { + w.sync(syncFuncExt) + + l.pragmaFlag(w, name.Func.Pragma) + l.linkname(w, name) + + // Relocated extension data. + w.bool(true) + + // Record definition ABI so cross-ABI calls can be direct. + // This is important for the performance of calling some + // common functions implemented in assembly (e.g., bytealg). + w.uint64(uint64(name.Func.ABI)) + + // Escape analysis. + for _, fs := range &types.RecvsParams { + for _, f := range fs(name.Type()).FieldSlice() { + w.string(f.Note) + } + } + + if inl := name.Func.Inl; w.bool(inl != nil) { + w.len(int(inl.Cost)) + w.bool(inl.CanDelayResults) + + pri, ok := bodyReader[name.Func] + assert(ok) + w.sync(syncAddBody) + w.reloc(relocBody, l.relocIdx(pri.pr, relocBody, pri.idx)) + } + + w.sync(syncEOF) +} + +func (l *linker) relocTypeExt(w *encoder, name *ir.Name) { + w.sync(syncTypeExt) + + typ := name.Type() + + l.pragmaFlag(w, name.Pragma()) + + // For type T, export the index of type descriptor symbols of T and *T. + l.lsymIdx(w, "", reflectdata.TypeLinksym(typ)) + l.lsymIdx(w, "", reflectdata.TypeLinksym(typ.PtrTo())) + + if typ.Kind() != types.TINTER { + for _, method := range typ.Methods().Slice() { + l.relocFuncExt(w, method.Nname.(*ir.Name)) + } + } +} + +func (l *linker) relocVarExt(w *encoder, name *ir.Name) { + w.sync(syncVarExt) + l.linkname(w, name) +} + +func (l *linker) linkname(w *encoder, name *ir.Name) { + w.sync(syncLinkname) + + linkname := name.Sym().Linkname + if !l.lsymIdx(w, linkname, name.Linksym()) { + w.string(linkname) + } +} + +func (l *linker) lsymIdx(w *encoder, linkname string, lsym *obj.LSym) bool { + if lsym.PkgIdx > goobj.PkgIdxSelf || (lsym.PkgIdx == goobj.PkgIdxInvalid && !lsym.Indexed()) || linkname != "" { + w.int64(-1) + return false + } + + // For a defined symbol, export its index. + // For re-exporting an imported symbol, pass its index through. + w.int64(int64(lsym.SymIdx)) + return true +} + +// @@@ Helpers + +// TODO(mdempsky): These should probably be removed. I think they're a +// smell that the export data format is not yet quite right. + +func (pr *pkgDecoder) peekPkgPath(idx int) string { + r := pr.newDecoder(relocPkg, idx, syncPkgDef) + path := r.string() + if path == "" { + path = pr.pkgPath + } + return path +} + +func (pr *pkgDecoder) peekObj(idx int) (string, string, codeObj, []int) { + r := pr.newDecoder(relocObj, idx, syncObject1) + r.sync(syncSym) + r.sync(syncPkg) + path := pr.peekPkgPath(r.reloc(relocPkg)) + name := r.string() + assert(name != "") + + r.sync(syncTypeParamBounds) + r.len() // implicits + bounds := make([]int, r.len()) + for i := range bounds { + r.sync(syncType) + bounds[i] = r.reloc(relocType) + } + + tag := codeObj(r.code(syncCodeObj)) + + return path, name, tag, bounds +} diff --git a/src/cmd/compile/internal/noder/quirks.go b/src/cmd/compile/internal/noder/quirks.go new file mode 100644 index 00000000000..9f33fc576d9 --- /dev/null +++ b/src/cmd/compile/internal/noder/quirks.go @@ -0,0 +1,453 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "fmt" + + "cmd/compile/internal/base" + "cmd/compile/internal/ir" + "cmd/compile/internal/syntax" + "cmd/compile/internal/types2" + "cmd/internal/src" +) + +// This file defines helper functions useful for satisfying toolstash +// -cmp when compared against the legacy frontend behavior, but can be +// removed after that's no longer a concern. + +// quirksMode controls whether behavior specific to satsifying +// toolstash -cmp is used. +func quirksMode() bool { + // Currently, unified IR doesn't try to be compatible with + // -d=inlfuncswithclosures=1, so we overload this as a flag for + // enabling quirks mode. + return base.Debug.InlFuncsWithClosures == 0 +} + +// posBasesOf returns all of the position bases in the source files, +// as seen in a straightforward traversal. +// +// This is necessary to ensure position bases (and thus file names) +// get registered in the same order as noder would visit them. +func posBasesOf(noders []*noder) []*syntax.PosBase { + seen := make(map[*syntax.PosBase]bool) + var bases []*syntax.PosBase + + for _, p := range noders { + syntax.Walk(p.file, func(n syntax.Node) bool { + if b := n.Pos().Base(); !seen[b] { + bases = append(bases, b) + seen[b] = true + } + return false + }) + } + + return bases +} + +// importedObjsOf returns the imported objects (i.e., referenced +// objects not declared by curpkg) from the parsed source files, in +// the order that typecheck used to load their definitions. +// +// This is needed because loading the definitions for imported objects +// can also add file names. +func importedObjsOf(curpkg *types2.Package, info *types2.Info, noders []*noder) []types2.Object { + // This code is complex because it matches the precise order that + // typecheck recursively and repeatedly traverses the IR. It's meant + // to be thrown away eventually anyway. + + seen := make(map[types2.Object]bool) + var objs []types2.Object + + var phase int + + decls := make(map[types2.Object]syntax.Decl) + assoc := func(decl syntax.Decl, names ...*syntax.Name) { + for _, name := range names { + obj, ok := info.Defs[name] + assert(ok) + decls[obj] = decl + } + } + + for _, p := range noders { + syntax.Walk(p.file, func(n syntax.Node) bool { + switch n := n.(type) { + case *syntax.ConstDecl: + assoc(n, n.NameList...) + case *syntax.FuncDecl: + assoc(n, n.Name) + case *syntax.TypeDecl: + assoc(n, n.Name) + case *syntax.VarDecl: + assoc(n, n.NameList...) + case *syntax.BlockStmt: + return true + } + return false + }) + } + + var visited map[syntax.Decl]bool + + var resolveDecl func(n syntax.Decl) + var resolveNode func(n syntax.Node, top bool) + + resolveDecl = func(n syntax.Decl) { + if visited[n] { + return + } + visited[n] = true + + switch n := n.(type) { + case *syntax.ConstDecl: + resolveNode(n.Type, true) + resolveNode(n.Values, true) + + case *syntax.FuncDecl: + if n.Recv != nil { + resolveNode(n.Recv, true) + } + resolveNode(n.Type, true) + + case *syntax.TypeDecl: + resolveNode(n.Type, true) + + case *syntax.VarDecl: + if n.Type != nil { + resolveNode(n.Type, true) + } else { + resolveNode(n.Values, true) + } + } + } + + resolveObj := func(pos syntax.Pos, obj types2.Object) { + switch obj.Pkg() { + case nil: + // builtin; nothing to do + + case curpkg: + if decl, ok := decls[obj]; ok { + resolveDecl(decl) + } + + default: + if obj.Parent() == obj.Pkg().Scope() && !seen[obj] { + seen[obj] = true + objs = append(objs, obj) + } + } + } + + checkdefat := func(pos syntax.Pos, n *syntax.Name) { + if n.Value == "_" { + return + } + obj, ok := info.Uses[n] + if !ok { + obj, ok = info.Defs[n] + if !ok { + return + } + } + if obj == nil { + return + } + resolveObj(pos, obj) + } + checkdef := func(n *syntax.Name) { checkdefat(n.Pos(), n) } + + var later []syntax.Node + + resolveNode = func(n syntax.Node, top bool) { + if n == nil { + return + } + syntax.Walk(n, func(n syntax.Node) bool { + switch n := n.(type) { + case *syntax.Name: + checkdef(n) + + case *syntax.SelectorExpr: + if name, ok := n.X.(*syntax.Name); ok { + if _, isPkg := info.Uses[name].(*types2.PkgName); isPkg { + checkdefat(n.X.Pos(), n.Sel) + return true + } + } + + case *syntax.AssignStmt: + resolveNode(n.Rhs, top) + resolveNode(n.Lhs, top) + return true + + case *syntax.VarDecl: + resolveNode(n.Values, top) + + case *syntax.FuncLit: + if top { + resolveNode(n.Type, top) + later = append(later, n.Body) + return true + } + + case *syntax.BlockStmt: + if phase >= 3 { + for _, stmt := range n.List { + resolveNode(stmt, false) + } + } + return true + } + + return false + }) + } + + for phase = 1; phase <= 5; phase++ { + visited = map[syntax.Decl]bool{} + + for _, p := range noders { + for _, decl := range p.file.DeclList { + switch decl := decl.(type) { + case *syntax.ConstDecl: + resolveDecl(decl) + + case *syntax.FuncDecl: + resolveDecl(decl) + if phase >= 3 && decl.Body != nil { + resolveNode(decl.Body, true) + } + + case *syntax.TypeDecl: + if !decl.Alias || phase >= 2 { + resolveDecl(decl) + } + + case *syntax.VarDecl: + if phase >= 2 { + resolveNode(decl.Values, true) + resolveDecl(decl) + } + } + } + + if phase >= 5 { + syntax.Walk(p.file, func(n syntax.Node) bool { + if name, ok := n.(*syntax.Name); ok { + if obj, ok := info.Uses[name]; ok { + resolveObj(name.Pos(), obj) + } + } + return false + }) + } + } + + for i := 0; i < len(later); i++ { + resolveNode(later[i], true) + } + later = nil + } + + return objs +} + +// typeExprEndPos returns the position that noder would leave base.Pos +// after parsing the given type expression. +func typeExprEndPos(expr0 syntax.Expr) syntax.Pos { + for { + switch expr := expr0.(type) { + case *syntax.Name: + return expr.Pos() + case *syntax.SelectorExpr: + return expr.X.Pos() + + case *syntax.ParenExpr: + expr0 = expr.X + + case *syntax.Operation: + assert(expr.Op == syntax.Mul) + assert(expr.Y == nil) + expr0 = expr.X + + case *syntax.ArrayType: + expr0 = expr.Elem + case *syntax.ChanType: + expr0 = expr.Elem + case *syntax.DotsType: + expr0 = expr.Elem + case *syntax.MapType: + expr0 = expr.Value + case *syntax.SliceType: + expr0 = expr.Elem + + case *syntax.StructType: + return expr.Pos() + + case *syntax.InterfaceType: + expr0 = lastFieldType(expr.MethodList) + if expr0 == nil { + return expr.Pos() + } + + case *syntax.FuncType: + expr0 = lastFieldType(expr.ResultList) + if expr0 == nil { + expr0 = lastFieldType(expr.ParamList) + if expr0 == nil { + return expr.Pos() + } + } + + case *syntax.IndexExpr: // explicit type instantiation + targs := unpackListExpr(expr.Index) + expr0 = targs[len(targs)-1] + + default: + panic(fmt.Sprintf("%s: unexpected type expression %v", expr.Pos(), syntax.String(expr))) + } + } +} + +func lastFieldType(fields []*syntax.Field) syntax.Expr { + if len(fields) == 0 { + return nil + } + return fields[len(fields)-1].Type +} + +// sumPos returns the position that noder.sum would produce for +// constant expression x. +func sumPos(x syntax.Expr) syntax.Pos { + orig := x + for { + switch x1 := x.(type) { + case *syntax.BasicLit: + assert(x1.Kind == syntax.StringLit) + return x1.Pos() + case *syntax.Operation: + assert(x1.Op == syntax.Add && x1.Y != nil) + if r, ok := x1.Y.(*syntax.BasicLit); ok { + assert(r.Kind == syntax.StringLit) + x = x1.X + continue + } + } + return orig.Pos() + } +} + +// funcParamsEndPos returns the value of base.Pos left by noder after +// processing a function signature. +func funcParamsEndPos(fn *ir.Func) src.XPos { + sig := fn.Nname.Type() + + fields := sig.Results().FieldSlice() + if len(fields) == 0 { + fields = sig.Params().FieldSlice() + if len(fields) == 0 { + fields = sig.Recvs().FieldSlice() + if len(fields) == 0 { + if fn.OClosure != nil { + return fn.Nname.Ntype.Pos() + } + return fn.Pos() + } + } + } + + return fields[len(fields)-1].Pos +} + +type dupTypes struct { + origs map[types2.Type]types2.Type +} + +func (d *dupTypes) orig(t types2.Type) types2.Type { + if orig, ok := d.origs[t]; ok { + return orig + } + return t +} + +func (d *dupTypes) add(t, orig types2.Type) { + if t == orig { + return + } + + if d.origs == nil { + d.origs = make(map[types2.Type]types2.Type) + } + assert(d.origs[t] == nil) + d.origs[t] = orig + + switch t := t.(type) { + case *types2.Pointer: + orig := orig.(*types2.Pointer) + d.add(t.Elem(), orig.Elem()) + + case *types2.Slice: + orig := orig.(*types2.Slice) + d.add(t.Elem(), orig.Elem()) + + case *types2.Map: + orig := orig.(*types2.Map) + d.add(t.Key(), orig.Key()) + d.add(t.Elem(), orig.Elem()) + + case *types2.Array: + orig := orig.(*types2.Array) + assert(t.Len() == orig.Len()) + d.add(t.Elem(), orig.Elem()) + + case *types2.Chan: + orig := orig.(*types2.Chan) + assert(t.Dir() == orig.Dir()) + d.add(t.Elem(), orig.Elem()) + + case *types2.Struct: + orig := orig.(*types2.Struct) + assert(t.NumFields() == orig.NumFields()) + for i := 0; i < t.NumFields(); i++ { + d.add(t.Field(i).Type(), orig.Field(i).Type()) + } + + case *types2.Interface: + orig := orig.(*types2.Interface) + assert(t.NumExplicitMethods() == orig.NumExplicitMethods()) + assert(t.NumEmbeddeds() == orig.NumEmbeddeds()) + for i := 0; i < t.NumExplicitMethods(); i++ { + d.add(t.ExplicitMethod(i).Type(), orig.ExplicitMethod(i).Type()) + } + for i := 0; i < t.NumEmbeddeds(); i++ { + d.add(t.EmbeddedType(i), orig.EmbeddedType(i)) + } + + case *types2.Signature: + orig := orig.(*types2.Signature) + assert((t.Recv() == nil) == (orig.Recv() == nil)) + if t.Recv() != nil { + d.add(t.Recv().Type(), orig.Recv().Type()) + } + d.add(t.Params(), orig.Params()) + d.add(t.Results(), orig.Results()) + + case *types2.Tuple: + orig := orig.(*types2.Tuple) + assert(t.Len() == orig.Len()) + for i := 0; i < t.Len(); i++ { + d.add(t.At(i).Type(), orig.At(i).Type()) + } + + default: + assert(types2.Identical(t, orig)) + } +} diff --git a/src/cmd/compile/internal/noder/reader.go b/src/cmd/compile/internal/noder/reader.go new file mode 100644 index 00000000000..18ecbff3ccf --- /dev/null +++ b/src/cmd/compile/internal/noder/reader.go @@ -0,0 +1,1970 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "bytes" + "fmt" + "go/constant" + "strings" + + "cmd/compile/internal/base" + "cmd/compile/internal/deadcode" + "cmd/compile/internal/dwarfgen" + "cmd/compile/internal/ir" + "cmd/compile/internal/typecheck" + "cmd/compile/internal/types" + "cmd/internal/obj" + "cmd/internal/src" +) + +// TODO(mdempsky): Suppress duplicate type/const errors that can arise +// during typecheck due to naive type substitution (e.g., see #42758). +// I anticipate these will be handled as a consequence of adding +// dictionaries support, so it's probably not important to focus on +// this until after that's done. + +type pkgReader struct { + pkgDecoder + + posBases []*src.PosBase + pkgs []*types.Pkg + typs []*types.Type + + // offset for rewriting the given index into the output, + // but bitwise inverted so we can detect if we're missing the entry or not. + newindex []int +} + +func newPkgReader(pr pkgDecoder) *pkgReader { + return &pkgReader{ + pkgDecoder: pr, + + posBases: make([]*src.PosBase, pr.numElems(relocPosBase)), + pkgs: make([]*types.Pkg, pr.numElems(relocPkg)), + typs: make([]*types.Type, pr.numElems(relocType)), + + newindex: make([]int, pr.totalElems()), + } +} + +type pkgReaderIndex struct { + pr *pkgReader + idx int + implicits []*types.Type +} + +func (pri pkgReaderIndex) asReader(k reloc, marker syncMarker) *reader { + r := pri.pr.newReader(k, pri.idx, marker) + r.implicits = pri.implicits + return r +} + +func (pr *pkgReader) newReader(k reloc, idx int, marker syncMarker) *reader { + return &reader{ + decoder: pr.newDecoder(k, idx, marker), + p: pr, + } +} + +type reader struct { + decoder + + p *pkgReader + + // Implicit and explicit type arguments in use for reading the + // current object. For example: + // + // func F[T any]() { + // type X[U any] struct { t T; u U } + // var _ X[string] + // } + // + // var _ = F[int] + // + // While instantiating F[int], we need to in turn instantiate + // X[string]. [int] and [string] are explicit type arguments for F + // and X, respectively; but [int] is also the implicit type + // arguments for X. + // + // (As an analogy to function literals, explicits are the function + // literal's formal parameters, while implicits are variables + // captured by the function literal.) + implicits []*types.Type + explicits []*types.Type + + ext *reader + + // TODO(mdempsky): The state below is all specific to reading + // function bodies. It probably makes sense to split it out + // separately so that it doesn't take up space in every reader + // instance. + + curfn *ir.Func + locals []*ir.Name + + funarghack bool + + // scopeVars is a stack tracking the number of variables declared in + // the current function at the moment each open scope was opened. + scopeVars []int + marker dwarfgen.ScopeMarker + lastCloseScopePos src.XPos + + // === details for handling inline body expansion === + + // If we're reading in a function body because of inlining, this is + // the call that we're inlining for. + inlCaller *ir.Func + inlCall *ir.CallExpr + inlFunc *ir.Func + inlTreeIndex int + inlPosBases map[*src.PosBase]*src.PosBase + + delayResults bool + + // Label to return to. + retlabel *types.Sym + + inlvars, retvars ir.Nodes +} + +func (r *reader) setType(n ir.Node, typ *types.Type) { + n.SetType(typ) + n.SetTypecheck(1) + + if name, ok := n.(*ir.Name); ok { + name.SetWalkdef(1) + name.Ntype = ir.TypeNode(name.Type()) + } +} + +func (r *reader) setValue(name *ir.Name, val constant.Value) { + name.SetVal(val) + name.Defn = nil +} + +// @@@ Positions + +func (r *reader) pos() src.XPos { + return base.Ctxt.PosTable.XPos(r.pos0()) +} + +func (r *reader) pos0() src.Pos { + r.sync(syncPos) + if !r.bool() { + return src.NoPos + } + + posBase := r.posBase() + line := r.uint() + col := r.uint() + return src.MakePos(posBase, line, col) +} + +func (r *reader) posBase() *src.PosBase { + return r.inlPosBase(r.p.posBaseIdx(r.reloc(relocPosBase))) +} + +func (pr *pkgReader) posBaseIdx(idx int) *src.PosBase { + if b := pr.posBases[idx]; b != nil { + return b + } + + r := pr.newReader(relocPosBase, idx, syncPosBase) + var b *src.PosBase + + fn := r.string() + absfn := r.string() + + if r.bool() { + b = src.NewFileBase(fn, absfn) + } else { + pos := r.pos0() + line := r.uint() + col := r.uint() + b = src.NewLinePragmaBase(pos, fn, absfn, line, col) + } + + pr.posBases[idx] = b + return b +} + +func (r *reader) inlPosBase(oldBase *src.PosBase) *src.PosBase { + if r.inlCall == nil { + return oldBase + } + + if newBase, ok := r.inlPosBases[oldBase]; ok { + return newBase + } + + newBase := src.NewInliningBase(oldBase, r.inlTreeIndex) + r.inlPosBases[oldBase] = newBase + return newBase +} + +func (r *reader) updatePos(xpos src.XPos) src.XPos { + pos := base.Ctxt.PosTable.Pos(xpos) + pos.SetBase(r.inlPosBase(pos.Base())) + return base.Ctxt.PosTable.XPos(pos) +} + +func (r *reader) origPos(xpos src.XPos) src.XPos { + if r.inlCall == nil { + return xpos + } + + pos := base.Ctxt.PosTable.Pos(xpos) + for old, new := range r.inlPosBases { + if pos.Base() == new { + pos.SetBase(old) + return base.Ctxt.PosTable.XPos(pos) + } + } + + base.FatalfAt(xpos, "pos base missing from inlPosBases") + panic("unreachable") +} + +// @@@ Packages + +func (r *reader) pkg() *types.Pkg { + r.sync(syncPkg) + return r.p.pkgIdx(r.reloc(relocPkg)) +} + +func (pr *pkgReader) pkgIdx(idx int) *types.Pkg { + if pkg := pr.pkgs[idx]; pkg != nil { + return pkg + } + + pkg := pr.newReader(relocPkg, idx, syncPkgDef).doPkg() + pr.pkgs[idx] = pkg + return pkg +} + +func (r *reader) doPkg() *types.Pkg { + path := r.string() + if path == "builtin" { + return types.BuiltinPkg + } + if path == "" { + path = r.p.pkgPath + } + + name := r.string() + height := r.len() + + pkg := types.NewPkg(path, "") + + if pkg.Name == "" { + pkg.Name = name + } else { + assert(pkg.Name == name) + } + + if pkg.Height == 0 { + pkg.Height = height + } else { + assert(pkg.Height == height) + } + + return pkg +} + +// @@@ Types + +func (r *reader) typ() *types.Type { + r.sync(syncType) + return r.p.typIdx(r.reloc(relocType), r.implicits, r.explicits) +} + +func (pr *pkgReader) typIdx(idx int, implicits, explicits []*types.Type) *types.Type { + if typ := pr.typs[idx]; typ != nil { + return typ + } + + r := pr.newReader(relocType, idx, syncTypeIdx) + r.implicits = implicits + r.explicits = explicits + typ := r.doTyp() + assert(typ != nil) + + if typ := pr.typs[idx]; typ != nil { + // This happens in fixedbugs/issue27232.go. + // TODO(mdempsky): Explain why/how this happens. + return typ + } + + // If we have type parameters, the type might refer to them, and it + // wouldn't be safe to reuse those in other contexts. So we + // conservatively avoid caching them in that case. + // + // TODO(mdempsky): If we're clever, we should be able to still cache + // types by tracking which type parameters are used. However, in my + // attempts so far, I haven't yet succeeded in being clever enough. + if len(implicits)+len(explicits) == 0 { + pr.typs[idx] = typ + } + + if !typ.IsUntyped() { + types.CheckSize(typ) + } + + return typ +} + +func (r *reader) doTyp() *types.Type { + switch tag := codeType(r.code(syncType)); tag { + default: + panic(fmt.Sprintf("unexpected type: %v", tag)) + + case typeBasic: + return *basics[r.len()] + + case typeNamed: + obj := r.obj() + assert(obj.Op() == ir.OTYPE) + return obj.Type() + + case typeTypeParam: + idx := r.len() + if idx < len(r.implicits) { + return r.implicits[idx] + } + return r.explicits[idx-len(r.implicits)] + + case typeArray: + len := int64(r.uint64()) + return types.NewArray(r.typ(), len) + case typeChan: + dir := dirs[r.len()] + return types.NewChan(r.typ(), dir) + case typeMap: + return types.NewMap(r.typ(), r.typ()) + case typePointer: + return types.NewPtr(r.typ()) + case typeSignature: + return r.signature(types.LocalPkg, nil) + case typeSlice: + return types.NewSlice(r.typ()) + case typeStruct: + return r.structType() + case typeInterface: + return r.interfaceType() + } +} + +func (r *reader) interfaceType() *types.Type { + tpkg := types.LocalPkg // TODO(mdempsky): Remove after iexport is gone. + + nmethods, nembeddeds := r.len(), r.len() + + fields := make([]*types.Field, nmethods+nembeddeds) + methods, embeddeds := fields[:nmethods], fields[nmethods:] + + for i := range methods { + pos := r.pos() + pkg, sym := r.selector() + tpkg = pkg + mtyp := r.signature(pkg, typecheck.FakeRecv()) + methods[i] = types.NewField(pos, sym, mtyp) + } + for i := range embeddeds { + embeddeds[i] = types.NewField(src.NoXPos, nil, r.typ()) + } + + if len(fields) == 0 { + return types.Types[types.TINTER] // empty interface + } + return types.NewInterface(tpkg, fields) +} + +func (r *reader) structType() *types.Type { + tpkg := types.LocalPkg // TODO(mdempsky): Remove after iexport is gone. + fields := make([]*types.Field, r.len()) + for i := range fields { + pos := r.pos() + pkg, sym := r.selector() + tpkg = pkg + ftyp := r.typ() + tag := r.string() + embedded := r.bool() + + f := types.NewField(pos, sym, ftyp) + f.Note = tag + if embedded { + f.Embedded = 1 + } + fields[i] = f + } + return types.NewStruct(tpkg, fields) +} + +func (r *reader) signature(tpkg *types.Pkg, recv *types.Field) *types.Type { + r.sync(syncSignature) + + params := r.params(&tpkg) + results := r.params(&tpkg) + if r.bool() { // variadic + params[len(params)-1].SetIsDDD(true) + } + + return types.NewSignature(tpkg, recv, nil, params, results) +} + +func (r *reader) params(tpkg **types.Pkg) []*types.Field { + r.sync(syncParams) + fields := make([]*types.Field, r.len()) + for i := range fields { + *tpkg, fields[i] = r.param() + } + return fields +} + +func (r *reader) param() (*types.Pkg, *types.Field) { + r.sync(syncParam) + + pos := r.pos() + pkg, sym := r.localIdent() + typ := r.typ() + + return pkg, types.NewField(pos, sym, typ) +} + +// @@@ Objects + +var objReader = map[*types.Sym]pkgReaderIndex{} + +func (r *reader) obj() ir.Node { + r.sync(syncObject) + + idx := r.reloc(relocObj) + + explicits := make([]*types.Type, r.len()) + for i := range explicits { + explicits[i] = r.typ() + } + + return r.p.objIdx(idx, r.implicits, explicits) +} + +func (pr *pkgReader) objIdx(idx int, implicits, explicits []*types.Type) ir.Node { + r := pr.newReader(relocObj, idx, syncObject1) + r.ext = pr.newReader(relocObjExt, idx, syncObject1) + + _, sym := r.qualifiedIdent() + + // Middle dot indicates local defined type; see writer.sym. + // TODO(mdempsky): Come up with a better way to handle this. + if strings.Contains(sym.Name, "·") { + r.implicits = implicits + r.ext.implicits = implicits + } + r.explicits = explicits + r.ext.explicits = explicits + + origSym := sym + + sym = r.mangle(sym) + if !sym.IsBlank() && sym.Def != nil { + return sym.Def.(ir.Node) + } + + r.typeParamBounds(origSym) + tag := codeObj(r.code(syncCodeObj)) + + do := func(op ir.Op, hasTParams bool) *ir.Name { + pos := r.pos() + if hasTParams { + r.typeParamNames() + } + + name := ir.NewDeclNameAt(pos, op, sym) + name.Class = ir.PEXTERN // may be overridden later + if !sym.IsBlank() { + if sym.Def != nil { + base.FatalfAt(name.Pos(), "already have a definition for %v", name) + } + assert(sym.Def == nil) + sym.Def = name + } + return name + } + + switch tag { + default: + panic("unexpected object") + + case objStub: + if pri, ok := objReader[origSym]; ok { + return pri.pr.objIdx(pri.idx, pri.implicits, r.explicits) + } + if haveLegacyImports { + assert(len(r.implicits)+len(r.explicits) == 0) + return typecheck.Resolve(ir.NewIdent(src.NoXPos, origSym)) + } + base.Fatalf("unresolved stub: %v", origSym) + panic("unreachable") + + case objAlias: + name := do(ir.OTYPE, false) + r.setType(name, r.typ()) + name.SetAlias(true) + return name + + case objConst: + name := do(ir.OLITERAL, false) + typ, val := r.value() + r.setType(name, typ) + r.setValue(name, val) + return name + + case objFunc: + if sym.Name == "init" { + sym = renameinit() + } + name := do(ir.ONAME, true) + r.setType(name, r.signature(sym.Pkg, nil)) + + name.Func = ir.NewFunc(r.pos()) + name.Func.Nname = name + + r.ext.funcExt(name) + return name + + case objType: + name := do(ir.OTYPE, true) + typ := types.NewNamed(name) + r.setType(name, typ) + + // Important: We need to do this before SetUnderlying. + r.ext.typeExt(name) + + // We need to defer CheckSize until we've called SetUnderlying to + // handle recursive types. + types.DeferCheckSize() + typ.SetUnderlying(r.typ()) + types.ResumeCheckSize() + + methods := make([]*types.Field, r.len()) + for i := range methods { + methods[i] = r.method() + } + if len(methods) != 0 { + typ.Methods().Set(methods) + } + + return name + + case objVar: + name := do(ir.ONAME, false) + r.setType(name, r.typ()) + r.ext.varExt(name) + return name + } +} + +func (r *reader) mangle(sym *types.Sym) *types.Sym { + if len(r.implicits)+len(r.explicits) == 0 { + return sym + } + + var buf bytes.Buffer + buf.WriteString(sym.Name) + buf.WriteByte('[') + for i, targs := range [2][]*types.Type{r.implicits, r.explicits} { + if i > 0 && len(r.implicits) != 0 && len(r.explicits) != 0 { + buf.WriteByte(';') + } + for j, targ := range targs { + if j > 0 { + buf.WriteByte(',') + } + // TODO(mdempsky): We need the linker to replace "" in the symbol + // names here. + buf.WriteString(targ.ShortString()) + } + } + buf.WriteByte(']') + return sym.Pkg.Lookup(buf.String()) +} + +func (r *reader) typeParamBounds(sym *types.Sym) { + r.sync(syncTypeParamBounds) + + nimplicits := r.len() + nexplicits := r.len() + + if len(r.implicits) != nimplicits || len(r.explicits) != nexplicits { + base.Fatalf("%v has %v+%v params, but instantiated with %v+%v args", sym, nimplicits, nexplicits, len(r.implicits), len(r.explicits)) + } + + // For stenciling, we can just skip over the type parameters. + + for range r.explicits { + // Skip past bounds without actually evaluating them. + r.sync(syncType) + r.reloc(relocType) + } +} + +func (r *reader) typeParamNames() { + r.sync(syncTypeParamNames) + + for range r.explicits { + r.pos() + r.localIdent() + } +} + +func (r *reader) value() (*types.Type, constant.Value) { + r.sync(syncValue) + typ := r.typ() + return typ, FixValue(typ, r.rawValue()) +} + +func (r *reader) method() *types.Field { + r.sync(syncMethod) + pos := r.pos() + pkg, sym := r.selector() + r.typeParamNames() + _, recv := r.param() + typ := r.signature(pkg, recv) + + fnsym := sym + fnsym = ir.MethodSym(recv.Type, fnsym) + name := ir.NewNameAt(pos, fnsym) + r.setType(name, typ) + + name.Func = ir.NewFunc(r.pos()) + name.Func.Nname = name + + // TODO(mdempsky): Make sure we're handling //go:nointerface + // correctly. I don't think this is exercised within the Go repo. + + r.ext.funcExt(name) + + meth := types.NewField(name.Func.Pos(), sym, typ) + meth.Nname = name + return meth +} + +func (r *reader) qualifiedIdent() (pkg *types.Pkg, sym *types.Sym) { + r.sync(syncSym) + pkg = r.pkg() + if name := r.string(); name != "" { + sym = pkg.Lookup(name) + } + return +} + +func (r *reader) localIdent() (pkg *types.Pkg, sym *types.Sym) { + r.sync(syncLocalIdent) + pkg = r.pkg() + if name := r.string(); name != "" { + sym = pkg.Lookup(name) + } + return +} + +func (r *reader) selector() (origPkg *types.Pkg, sym *types.Sym) { + r.sync(syncSelector) + origPkg = r.pkg() + name := r.string() + pkg := origPkg + if types.IsExported(name) { + pkg = types.LocalPkg + } + sym = pkg.Lookup(name) + return +} + +// @@@ Compiler extensions + +func (r *reader) funcExt(name *ir.Name) { + r.sync(syncFuncExt) + + name.Class = 0 // so MarkFunc doesn't complain + ir.MarkFunc(name) + + fn := name.Func + + // XXX: Workaround because linker doesn't know how to copy Pos. + if !fn.Pos().IsKnown() { + fn.SetPos(name.Pos()) + } + + // TODO(mdempsky): Remember why I wrote this code. I think it has to + // do with how ir.VisitFuncsBottomUp works? + if name.Sym().Pkg == types.LocalPkg || len(r.implicits)+len(r.explicits) != 0 { + name.Defn = fn + } + + fn.Pragma = r.pragmaFlag() + r.linkname(name) + + if r.bool() { + fn.ABI = obj.ABI(r.uint64()) + + // Escape analysis. + for _, fs := range &types.RecvsParams { + for _, f := range fs(name.Type()).FieldSlice() { + f.Note = r.string() + } + } + + if r.bool() { + fn.Inl = &ir.Inline{ + Cost: int32(r.len()), + CanDelayResults: r.bool(), + } + r.addBody(name.Func) + } + } else { + r.addBody(name.Func) + } + r.sync(syncEOF) +} + +func (r *reader) typeExt(name *ir.Name) { + r.sync(syncTypeExt) + + typ := name.Type() + + if len(r.implicits)+len(r.explicits) != 0 { + // Set "RParams" (really type arguments here, not parameters) so + // this type is treated as "fully instantiated". This ensures the + // type descriptor is written out as DUPOK and method wrappers are + // generated even for imported types. + var targs []*types.Type + targs = append(targs, r.implicits...) + targs = append(targs, r.explicits...) + typ.SetRParams(targs) + } + + name.SetPragma(r.pragmaFlag()) + if name.Pragma()&ir.NotInHeap != 0 { + typ.SetNotInHeap(true) + } + + typecheck.SetBaseTypeIndex(typ, r.int64(), r.int64()) +} + +func (r *reader) varExt(name *ir.Name) { + r.sync(syncVarExt) + r.linkname(name) +} + +func (r *reader) linkname(name *ir.Name) { + assert(name.Op() == ir.ONAME) + r.sync(syncLinkname) + + if idx := r.int64(); idx >= 0 { + lsym := name.Linksym() + lsym.SymIdx = int32(idx) + lsym.Set(obj.AttrIndexed, true) + } else { + name.Sym().Linkname = r.string() + } +} + +func (r *reader) pragmaFlag() ir.PragmaFlag { + r.sync(syncPragma) + return ir.PragmaFlag(r.int()) +} + +// @@@ Function bodies + +// bodyReader tracks where the serialized IR for a function's body can +// be found. +var bodyReader = map[*ir.Func]pkgReaderIndex{} + +// todoBodies holds the list of function bodies that still need to be +// constructed. +var todoBodies []*ir.Func + +func (r *reader) addBody(fn *ir.Func) { + r.sync(syncAddBody) + + // See commont in writer.addBody for why r.implicits and r.explicits + // should never both be non-empty. + implicits := r.implicits + if len(implicits) == 0 { + implicits = r.explicits + } else { + assert(len(r.explicits) == 0) + } + + pri := pkgReaderIndex{r.p, r.reloc(relocBody), implicits} + bodyReader[fn] = pri + + if r.curfn == nil { + todoBodies = append(todoBodies, fn) + return + } + + pri.funcBody(fn) +} + +func (pri pkgReaderIndex) funcBody(fn *ir.Func) { + r := pri.asReader(relocBody, syncFuncBody) + r.funcBody(fn) +} + +func (r *reader) funcBody(fn *ir.Func) { + r.curfn = fn + r.locals = fn.ClosureVars + + // TODO(mdempsky): Get rid of uses of typecheck.NodAddrAt so we + // don't have to set ir.CurFunc. + outerCurFunc := ir.CurFunc + ir.CurFunc = fn + + r.funcargs(fn) + + if r.bool() { + body := r.stmts() + if body == nil { + pos := src.NoXPos + if quirksMode() { + pos = funcParamsEndPos(fn) + } + body = []ir.Node{ir.NewBlockStmt(pos, nil)} + } + fn.Body = body + fn.Endlineno = r.pos() + } + + ir.CurFunc = outerCurFunc + r.marker.WriteTo(fn) +} + +func (r *reader) funcargs(fn *ir.Func) { + sig := fn.Nname.Type() + + if recv := sig.Recv(); recv != nil { + r.funcarg(recv, recv.Sym, ir.PPARAM) + } + for _, param := range sig.Params().FieldSlice() { + r.funcarg(param, param.Sym, ir.PPARAM) + } + + for i, param := range sig.Results().FieldSlice() { + sym := types.OrigSym(param.Sym) + + if sym == nil || sym.IsBlank() { + prefix := "~r" + if r.inlCall != nil { + prefix = "~R" + } else if sym != nil { + prefix = "~b" + } + sym = typecheck.LookupNum(prefix, i) + } + + r.funcarg(param, sym, ir.PPARAMOUT) + } +} + +func (r *reader) funcarg(param *types.Field, sym *types.Sym, ctxt ir.Class) { + if sym == nil { + assert(ctxt == ir.PPARAM) + if r.inlCall != nil { + r.inlvars.Append(ir.BlankNode) + } + return + } + + name := ir.NewNameAt(r.updatePos(param.Pos), sym) + r.setType(name, param.Type) + r.addLocal(name, ctxt) + + if r.inlCall == nil { + if !r.funarghack { + param.Sym = sym + param.Nname = name + } + } else { + if ctxt == ir.PPARAMOUT { + r.retvars.Append(name) + } else { + r.inlvars.Append(name) + } + } +} + +func (r *reader) addLocal(name *ir.Name, ctxt ir.Class) { + assert(ctxt == ir.PAUTO || ctxt == ir.PPARAM || ctxt == ir.PPARAMOUT) + + r.sync(syncAddLocal) + if debug { + want := r.int() + if have := len(r.locals); have != want { + base.FatalfAt(name.Pos(), "locals table has desynced") + } + } + + name.SetUsed(true) + r.locals = append(r.locals, name) + + // TODO(mdempsky): Move earlier. + if ir.IsBlank(name) { + return + } + + if r.inlCall != nil { + if ctxt == ir.PAUTO { + name.SetInlLocal(true) + } else { + name.SetInlFormal(true) + ctxt = ir.PAUTO + } + + // TODO(mdempsky): Rethink this hack. + if strings.HasPrefix(name.Sym().Name, "~") || base.Flag.GenDwarfInl == 0 { + name.SetPos(r.inlCall.Pos()) + name.SetInlFormal(false) + name.SetInlLocal(false) + } + } + + name.Class = ctxt + name.Curfn = r.curfn + + r.curfn.Dcl = append(r.curfn.Dcl, name) + + if ctxt == ir.PAUTO { + name.SetFrameOffset(0) + } +} + +func (r *reader) useLocal() *ir.Name { + r.sync(syncUseObjLocal) + return r.locals[r.len()] +} + +func (r *reader) openScope() { + r.sync(syncOpenScope) + pos := r.pos() + + if base.Flag.Dwarf { + r.scopeVars = append(r.scopeVars, len(r.curfn.Dcl)) + r.marker.Push(pos) + } +} + +func (r *reader) closeScope() { + r.sync(syncCloseScope) + r.lastCloseScopePos = r.pos() + + r.closeAnotherScope() +} + +// closeAnotherScope is like closeScope, but it reuses the same mark +// position as the last closeScope call. This is useful for "for" and +// "if" statements, as their implicit blocks always end at the same +// position as an explicit block. +func (r *reader) closeAnotherScope() { + r.sync(syncCloseAnotherScope) + + if base.Flag.Dwarf { + scopeVars := r.scopeVars[len(r.scopeVars)-1] + r.scopeVars = r.scopeVars[:len(r.scopeVars)-1] + + if scopeVars == len(r.curfn.Dcl) { + // no variables were declared in this scope, so we can retract it. + r.marker.Unpush() + } else { + r.marker.Pop(r.lastCloseScopePos) + } + } +} + +// @@@ Statements + +func (r *reader) stmt() ir.Node { + switch stmts := r.stmts(); len(stmts) { + case 0: + return nil + case 1: + return stmts[0] + default: + return ir.NewBlockStmt(stmts[0].Pos(), stmts) + } +} + +func (r *reader) stmts() []ir.Node { + var res ir.Nodes + + r.sync(syncStmts) + for { + tag := codeStmt(r.code(syncStmt1)) + if tag == stmtEnd { + r.sync(syncStmtsEnd) + return res + } + + if n := r.stmt1(tag, &res); n != nil { + res.Append(n) + } + } +} + +func (r *reader) stmt1(tag codeStmt, out *ir.Nodes) ir.Node { + var label *types.Sym + if n := len(*out); n > 0 { + if ls, ok := (*out)[n-1].(*ir.LabelStmt); ok { + label = ls.Label + } + } + + switch tag { + default: + panic("unexpected statement") + + case stmtAssign: + pos := r.pos() + names, lhs := r.assignList() + rhs := r.exprList() + + if len(rhs) == 0 { + for _, name := range names { + as := ir.NewAssignStmt(pos, name, nil) + as.PtrInit().Append(ir.NewDecl(pos, ir.ODCL, name)) + out.Append(as) + } + return nil + } + + if len(lhs) == 1 && len(rhs) == 1 { + n := ir.NewAssignStmt(pos, lhs[0], rhs[0]) + n.Def = r.initDefn(n, names) + return n + } + + n := ir.NewAssignListStmt(pos, ir.OAS2, lhs, rhs) + n.Def = r.initDefn(n, names) + return n + + case stmtAssignOp: + op := r.op() + lhs := r.expr() + pos := r.pos() + rhs := r.expr() + return ir.NewAssignOpStmt(pos, op, lhs, rhs) + + case stmtIncDec: + op := r.op() + lhs := r.expr() + pos := r.pos() + n := ir.NewAssignOpStmt(pos, op, lhs, ir.NewBasicLit(pos, one)) + n.IncDec = true + return n + + case stmtBlock: + out.Append(r.blockStmt()...) + return nil + + case stmtBranch: + pos := r.pos() + op := r.op() + sym := r.optLabel() + return ir.NewBranchStmt(pos, op, sym) + + case stmtCall: + pos := r.pos() + op := r.op() + call := r.expr() + return ir.NewGoDeferStmt(pos, op, call) + + case stmtExpr: + return r.expr() + + case stmtFor: + return r.forStmt(label) + + case stmtIf: + return r.ifStmt() + + case stmtLabel: + pos := r.pos() + sym := r.label() + return ir.NewLabelStmt(pos, sym) + + case stmtReturn: + pos := r.pos() + results := r.exprList() + return ir.NewReturnStmt(pos, results) + + case stmtSelect: + return r.selectStmt(label) + + case stmtSend: + pos := r.pos() + ch := r.expr() + value := r.expr() + return ir.NewSendStmt(pos, ch, value) + + case stmtSwitch: + return r.switchStmt(label) + + case stmtTypeDeclHack: + // fake "type _ = int" declaration to prevent inlining in quirks mode. + assert(quirksMode()) + + name := ir.NewDeclNameAt(src.NoXPos, ir.OTYPE, ir.BlankNode.Sym()) + name.SetAlias(true) + r.setType(name, types.Types[types.TINT]) + + n := ir.NewDecl(src.NoXPos, ir.ODCLTYPE, name) + n.SetTypecheck(1) + return n + } +} + +func (r *reader) assignList() ([]*ir.Name, []ir.Node) { + lhs := make([]ir.Node, r.len()) + var names []*ir.Name + + for i := range lhs { + if r.bool() { + pos := r.pos() + _, sym := r.localIdent() + typ := r.typ() + + name := ir.NewNameAt(pos, sym) + lhs[i] = name + names = append(names, name) + r.setType(name, typ) + r.addLocal(name, ir.PAUTO) + continue + } + + lhs[i] = r.expr() + } + + return names, lhs +} + +func (r *reader) blockStmt() []ir.Node { + r.sync(syncBlockStmt) + r.openScope() + stmts := r.stmts() + r.closeScope() + return stmts +} + +func (r *reader) forStmt(label *types.Sym) ir.Node { + r.sync(syncForStmt) + + r.openScope() + + if r.bool() { + pos := r.pos() + names, lhs := r.assignList() + x := r.expr() + body := r.blockStmt() + r.closeAnotherScope() + + rang := ir.NewRangeStmt(pos, nil, nil, x, body) + if len(lhs) >= 1 { + rang.Key = lhs[0] + if len(lhs) >= 2 { + rang.Value = lhs[1] + } + } + rang.Def = r.initDefn(rang, names) + rang.Label = label + return rang + } + + pos := r.pos() + init := r.stmt() + cond := r.expr() + post := r.stmt() + body := r.blockStmt() + r.closeAnotherScope() + + stmt := ir.NewForStmt(pos, init, cond, post, body) + stmt.Label = label + return stmt +} + +func (r *reader) ifStmt() ir.Node { + r.sync(syncIfStmt) + r.openScope() + pos := r.pos() + init := r.stmts() + cond := r.expr() + then := r.blockStmt() + els := r.stmts() + n := ir.NewIfStmt(pos, cond, then, els) + n.SetInit(init) + r.closeAnotherScope() + return n +} + +func (r *reader) selectStmt(label *types.Sym) ir.Node { + r.sync(syncSelectStmt) + + pos := r.pos() + clauses := make([]*ir.CommClause, r.len()) + for i := range clauses { + if i > 0 { + r.closeScope() + } + r.openScope() + + pos := r.pos() + comm := r.stmt() + body := r.stmts() + + clauses[i] = ir.NewCommStmt(pos, comm, body) + } + if len(clauses) > 0 { + r.closeScope() + } + n := ir.NewSelectStmt(pos, clauses) + n.Label = label + return n +} + +func (r *reader) switchStmt(label *types.Sym) ir.Node { + r.sync(syncSwitchStmt) + + r.openScope() + pos := r.pos() + init := r.stmt() + tag := r.expr() + + tswitch, ok := tag.(*ir.TypeSwitchGuard) + if ok && tswitch.Tag == nil { + tswitch = nil + } + + clauses := make([]*ir.CaseClause, r.len()) + for i := range clauses { + if i > 0 { + r.closeScope() + } + r.openScope() + + pos := r.pos() + cases := r.exprList() + + clause := ir.NewCaseStmt(pos, cases, nil) + if tswitch != nil { + pos := r.pos() + typ := r.typ() + + name := ir.NewNameAt(pos, tswitch.Tag.Sym()) + r.setType(name, typ) + r.addLocal(name, ir.PAUTO) + clause.Var = name + name.Defn = tswitch + } + + clause.Body = r.stmts() + clauses[i] = clause + } + if len(clauses) > 0 { + r.closeScope() + } + r.closeScope() + + n := ir.NewSwitchStmt(pos, tag, clauses) + n.Label = label + if init != nil { + n.SetInit([]ir.Node{init}) + } + return n +} + +func (r *reader) label() *types.Sym { + r.sync(syncLabel) + name := r.string() + if r.inlCall != nil { + name = fmt.Sprintf("~%s·%d", name, inlgen) + } + return typecheck.Lookup(name) +} + +func (r *reader) optLabel() *types.Sym { + r.sync(syncOptLabel) + if r.bool() { + return r.label() + } + return nil +} + +// initDefn marks the given names as declared by defn and populates +// its Init field with ODCL nodes. It then reports whether any names +// were so declared, which can be used to initialize defn.Def. +func (r *reader) initDefn(defn ir.InitNode, names []*ir.Name) bool { + if len(names) == 0 { + return false + } + + init := make([]ir.Node, len(names)) + for i, name := range names { + name.Defn = defn + init[i] = ir.NewDecl(name.Pos(), ir.ODCL, name) + } + defn.SetInit(init) + return true +} + +// @@@ Expressions + +func (r *reader) expr() ir.Node { + switch tag := codeExpr(r.code(syncExpr)); tag { + default: + panic("unhandled expression") + + case exprNone: + return nil + + case exprBlank: + return ir.BlankNode + + case exprLocal: + return r.useLocal() + + case exprName: + return r.obj() + + case exprType: + return ir.TypeNode(r.typ()) + + case exprConst: + pos := r.pos() + typ, val := r.value() + op := r.op() + orig := r.string() + return OrigConst(pos, typ, val, op, orig) + + case exprCompLit: + return r.compLit() + + case exprFuncLit: + return r.funcLit() + + case exprSelector: + x := r.expr() + pos := r.pos() + _, sym := r.selector() + return ir.NewSelectorExpr(pos, ir.OXDOT, x, sym) + + case exprIndex: + x := r.expr() + pos := r.pos() + index := r.expr() + return ir.NewIndexExpr(pos, x, index) + + case exprSlice: + x := r.expr() + pos := r.pos() + var index [3]ir.Node + for i := range index { + index[i] = r.expr() + } + op := ir.OSLICE + if index[2] != nil { + op = ir.OSLICE3 + } + return ir.NewSliceExpr(pos, op, x, index[0], index[1], index[2]) + + case exprAssert: + x := r.expr() + pos := r.pos() + typ := r.expr().(ir.Ntype) + return ir.NewTypeAssertExpr(pos, x, typ) + + case exprUnaryOp: + op := r.op() + pos := r.pos() + x := r.expr() + + switch op { + case ir.OADDR: + return typecheck.NodAddrAt(pos, x) + case ir.ODEREF: + return ir.NewStarExpr(pos, x) + } + return ir.NewUnaryExpr(pos, op, x) + + case exprBinaryOp: + op := r.op() + x := r.expr() + pos := r.pos() + y := r.expr() + + switch op { + case ir.OANDAND, ir.OOROR: + return ir.NewLogicalExpr(pos, op, x, y) + } + return ir.NewBinaryExpr(pos, op, x, y) + + case exprCall: + fun := r.expr() + pos := r.pos() + args := r.exprs() + dots := r.bool() + n := ir.NewCallExpr(pos, ir.OCALL, fun, args) + n.IsDDD = dots + return n + + case exprTypeSwitchGuard: + pos := r.pos() + var tag *ir.Ident + if r.bool() { + pos := r.pos() + sym := typecheck.Lookup(r.string()) + tag = ir.NewIdent(pos, sym) + } + x := r.expr() + return ir.NewTypeSwitchGuard(pos, tag, x) + } +} + +func (r *reader) compLit() ir.Node { + r.sync(syncCompLit) + pos := r.pos() + typ := r.typ() + + isPtrLit := typ.IsPtr() + if isPtrLit { + typ = typ.Elem() + } + if typ.Kind() == types.TFORW { + base.FatalfAt(pos, "unresolved composite literal type: %v", typ) + } + isStruct := typ.Kind() == types.TSTRUCT + + elems := make([]ir.Node, r.len()) + for i := range elems { + elemp := &elems[i] + + if isStruct { + sk := ir.NewStructKeyExpr(r.pos(), typ.Field(r.len()), nil) + *elemp, elemp = sk, &sk.Value + } else if r.bool() { + kv := ir.NewKeyExpr(r.pos(), r.expr(), nil) + *elemp, elemp = kv, &kv.Value + } + + *elemp = wrapName(r.pos(), r.expr()) + } + + lit := ir.NewCompLitExpr(pos, ir.OCOMPLIT, ir.TypeNode(typ), elems) + if isPtrLit { + return typecheck.NodAddrAt(pos, lit) + } + return lit +} + +func wrapName(pos src.XPos, x ir.Node) ir.Node { + // These nodes do not carry line numbers. + // Introduce a wrapper node to give them the correct line. + switch ir.Orig(x).Op() { + case ir.OTYPE, ir.OLITERAL: + if x.Sym() == nil { + break + } + fallthrough + case ir.ONAME, ir.ONONAME, ir.OPACK, ir.ONIL: + p := ir.NewParenExpr(pos, x) + p.SetImplicit(true) + return p + } + return x +} + +func (r *reader) funcLit() ir.Node { + r.sync(syncFuncLit) + + pos := r.pos() + typPos := r.pos() + xtype2 := r.signature(types.LocalPkg, nil) + + opos := pos + if quirksMode() { + opos = r.origPos(pos) + } + + fn := ir.NewClosureFunc(opos, r.curfn != nil) + + r.setType(fn.Nname, xtype2) + if quirksMode() { + fn.Nname.Ntype = ir.TypeNodeAt(typPos, xtype2) + } + + fn.ClosureVars = make([]*ir.Name, r.len()) + for i := range fn.ClosureVars { + pos := r.pos() + outer := r.useLocal() + + cv := ir.NewNameAt(pos, outer.Sym()) + r.setType(cv, outer.Type()) + cv.Curfn = fn + cv.Class = ir.PAUTOHEAP + cv.SetIsClosureVar(true) + cv.Defn = outer.Canonical() + cv.Outer = outer + + fn.ClosureVars[i] = cv + } + + r.addBody(fn) + + return fn.OClosure +} + +func (r *reader) exprList() []ir.Node { + r.sync(syncExprList) + return r.exprs() +} + +func (r *reader) exprs() []ir.Node { + r.sync(syncExprs) + nodes := make([]ir.Node, r.len()) + if len(nodes) == 0 { + return nil // TODO(mdempsky): Unclear if this matters. + } + for i := range nodes { + nodes[i] = r.expr() + } + return nodes +} + +func (r *reader) op() ir.Op { + r.sync(syncOp) + return ir.Op(r.len()) +} + +// @@@ Package initialization + +func (r *reader) pkgInit(self *types.Pkg, target *ir.Package) { + if quirksMode() { + for i, n := 0, r.len(); i < n; i++ { + // Eagerly register position bases, so their filenames are + // assigned stable indices. + posBase := r.posBase() + _ = base.Ctxt.PosTable.XPos(src.MakePos(posBase, 0, 0)) + } + + for i, n := 0, r.len(); i < n; i++ { + // Eagerly resolve imported objects, so any filenames registered + // in the process are assigned stable indices too. + _, sym := r.qualifiedIdent() + typecheck.Resolve(ir.NewIdent(src.NoXPos, sym)) + assert(sym.Def != nil) + } + } + + cgoPragmas := make([][]string, r.len()) + for i := range cgoPragmas { + cgoPragmas[i] = r.strings() + } + target.CgoPragmas = cgoPragmas + + r.pkgDecls(target) + + r.sync(syncEOF) +} + +func (r *reader) pkgDecls(target *ir.Package) { + r.sync(syncDecls) + for { + switch code := codeDecl(r.code(syncDecl)); code { + default: + panic(fmt.Sprintf("unhandled decl: %v", code)) + + case declEnd: + return + + case declFunc: + names := r.pkgObjs(target) + assert(len(names) == 1) + target.Decls = append(target.Decls, names[0].Func) + + case declMethod: + typ := r.typ() + _, sym := r.selector() + + method := typecheck.Lookdot1(nil, sym, typ, typ.Methods(), 0) + target.Decls = append(target.Decls, method.Nname.(*ir.Name).Func) + + case declVar: + pos := r.pos() + names := r.pkgObjs(target) + values := r.exprList() + + if len(names) > 1 && len(values) == 1 { + as := ir.NewAssignListStmt(pos, ir.OAS2, nil, values) + for _, name := range names { + as.Lhs.Append(name) + name.Defn = as + } + target.Decls = append(target.Decls, as) + } else { + for i, name := range names { + as := ir.NewAssignStmt(pos, name, nil) + if i < len(values) { + as.Y = values[i] + } + name.Defn = as + target.Decls = append(target.Decls, as) + } + } + + if n := r.len(); n > 0 { + assert(len(names) == 1) + embeds := make([]ir.Embed, n) + for i := range embeds { + embeds[i] = ir.Embed{Pos: r.pos(), Patterns: r.strings()} + } + names[0].Embed = &embeds + target.Embeds = append(target.Embeds, names[0]) + } + + case declOther: + r.pkgObjs(target) + } + } +} + +func (r *reader) pkgObjs(target *ir.Package) []*ir.Name { + r.sync(syncDeclNames) + nodes := make([]*ir.Name, r.len()) + for i := range nodes { + r.sync(syncDeclName) + + name := r.obj().(*ir.Name) + nodes[i] = name + + sym := name.Sym() + if sym.IsBlank() { + continue + } + + switch name.Class { + default: + base.FatalfAt(name.Pos(), "unexpected class: %v", name.Class) + + case ir.PEXTERN: + target.Externs = append(target.Externs, name) + + case ir.PFUNC: + assert(name.Type().Recv() == nil) + + // TODO(mdempsky): Cleaner way to recognize init? + if strings.HasPrefix(sym.Name, "init.") { + target.Inits = append(target.Inits, name.Func) + } + } + + if types.IsExported(sym.Name) { + assert(!sym.OnExportList()) + target.Exports = append(target.Exports, name) + sym.SetOnExportList(true) + } + + if base.Flag.AsmHdr != "" { + assert(!sym.Asm()) + target.Asms = append(target.Asms, name) + sym.SetAsm(true) + } + } + + return nodes +} + +// @@@ Inlining + +var inlgen = 0 + +func InlineCall(call *ir.CallExpr, fn *ir.Func, inlIndex int) *ir.InlinedCallExpr { + // TODO(mdempsky): Turn callerfn into an explicit parameter. + callerfn := ir.CurFunc + + pri, ok := bodyReader[fn] + if !ok { + // Assume it's an imported function or something that we don't + // have access to in quirks mode. + if haveLegacyImports { + return nil + } + + base.FatalfAt(call.Pos(), "missing function body for call to %v", fn) + } + + if fn.Inl.Body == nil { + expandInline(fn, pri) + } + + r := pri.asReader(relocBody, syncFuncBody) + + // TODO(mdempsky): This still feels clumsy. Can we do better? + tmpfn := ir.NewFunc(fn.Pos()) + tmpfn.Nname = ir.NewNameAt(fn.Nname.Pos(), callerfn.Sym()) + tmpfn.Closgen = callerfn.Closgen + defer func() { callerfn.Closgen = tmpfn.Closgen }() + + r.setType(tmpfn.Nname, fn.Type()) + r.curfn = tmpfn + + r.inlCaller = ir.CurFunc + r.inlCall = call + r.inlFunc = fn + r.inlTreeIndex = inlIndex + r.inlPosBases = make(map[*src.PosBase]*src.PosBase) + + for _, cv := range r.inlFunc.ClosureVars { + r.locals = append(r.locals, cv.Outer) + } + + r.funcargs(fn) + + assert(r.bool()) // have body + r.delayResults = fn.Inl.CanDelayResults + + r.retlabel = typecheck.AutoLabel(".i") + inlgen++ + + init := ir.TakeInit(call) + + // For normal function calls, the function callee expression + // may contain side effects (e.g., added by addinit during + // inlconv2expr or inlconv2list). Make sure to preserve these, + // if necessary (#42703). + if call.Op() == ir.OCALLFUNC { + callee := call.X + for callee.Op() == ir.OCONVNOP { + conv := callee.(*ir.ConvExpr) + init.Append(ir.TakeInit(conv)...) + callee = conv.X + } + + switch callee.Op() { + case ir.ONAME, ir.OCLOSURE, ir.OMETHEXPR: + // ok + default: + base.Fatalf("unexpected callee expression: %v", callee) + } + } + + var args ir.Nodes + if call.Op() == ir.OCALLMETH { + assert(call.X.Op() == ir.ODOTMETH) + args.Append(call.X.(*ir.SelectorExpr).X) + } + args.Append(call.Args...) + + // Create assignment to declare and initialize inlvars. + as2 := ir.NewAssignListStmt(call.Pos(), ir.OAS2, r.inlvars, args) + as2.Def = true + var as2init ir.Nodes + for _, name := range r.inlvars { + if ir.IsBlank(name) { + continue + } + // TODO(mdempsky): Use inlined position of name.Pos() instead? + name := name.(*ir.Name) + as2init.Append(ir.NewDecl(call.Pos(), ir.ODCL, name)) + name.Defn = as2 + } + as2.SetInit(as2init) + init.Append(typecheck.Stmt(as2)) + + if !r.delayResults { + // If not delaying retvars, declare and zero initialize the + // result variables now. + for _, name := range r.retvars { + // TODO(mdempsky): Use inlined position of name.Pos() instead? + name := name.(*ir.Name) + init.Append(ir.NewDecl(call.Pos(), ir.ODCL, name)) + ras := ir.NewAssignStmt(call.Pos(), name, nil) + init.Append(typecheck.Stmt(ras)) + } + } + + // Add an inline mark just before the inlined body. + // This mark is inline in the code so that it's a reasonable spot + // to put a breakpoint. Not sure if that's really necessary or not + // (in which case it could go at the end of the function instead). + // Note issue 28603. + init.Append(ir.NewInlineMarkStmt(call.Pos().WithIsStmt(), int64(r.inlTreeIndex))) + + nparams := len(r.curfn.Dcl) + + oldcurfn := ir.CurFunc + ir.CurFunc = r.curfn + + r.curfn.Body = r.stmts() + r.curfn.Endlineno = r.pos() + + typecheck.Stmts(r.curfn.Body) + deadcode.Func(r.curfn) + + // Replace any "return" statements within the function body. + { + var edit func(ir.Node) ir.Node + edit = func(n ir.Node) ir.Node { + if ret, ok := n.(*ir.ReturnStmt); ok { + n = typecheck.Stmt(r.inlReturn(ret)) + } + ir.EditChildren(n, edit) + return n + } + edit(r.curfn) + } + + ir.CurFunc = oldcurfn + + body := ir.Nodes(r.curfn.Body) + + // Quirk: If deadcode elimination turned a non-empty function into + // an empty one, we need to set the position for the empty block + // left behind to the the inlined position for src.NoXPos, so that + // an empty string gets added into the DWARF file name listing at + // the appropriate index. + if quirksMode() && len(body) == 1 { + if block, ok := body[0].(*ir.BlockStmt); ok && len(block.List) == 0 { + block.SetPos(r.updatePos(src.NoXPos)) + } + } + + // Quirkish: We need to eagerly prune variables added during + // inlining, but removed by deadcode.FuncBody above. Unused + // variables will get removed during stack frame layout anyway, but + // len(fn.Dcl) ends up influencing things like autotmp naming. + + used := usedLocals(body) + + for i, name := range r.curfn.Dcl { + if i < nparams || used.Has(name) { + name.Curfn = callerfn + callerfn.Dcl = append(callerfn.Dcl, name) + + // Quirkish. TODO(mdempsky): Document why. + if name.AutoTemp() { + name.SetEsc(ir.EscUnknown) + + if base.Flag.GenDwarfInl != 0 { + name.SetInlLocal(true) + } else { + name.SetPos(r.inlCall.Pos()) + } + } + } + } + + body.Append(ir.NewLabelStmt(call.Pos(), r.retlabel)) + + res := ir.NewInlinedCallExpr(call.Pos(), body, append([]ir.Node(nil), r.retvars...)) + res.SetInit(init) + res.SetType(call.Type()) + res.SetTypecheck(1) + + // Inlining shouldn't add any functions to todoBodies. + assert(len(todoBodies) == 0) + + return res +} + +// inlReturn returns a statement that can substitute for the given +// return statement when inlining. +func (r *reader) inlReturn(ret *ir.ReturnStmt) *ir.BlockStmt { + pos := r.inlCall.Pos() + + block := ir.TakeInit(ret) + + if results := ret.Results; len(results) != 0 { + assert(len(r.retvars) == len(results)) + + as2 := ir.NewAssignListStmt(pos, ir.OAS2, append([]ir.Node(nil), r.retvars...), ret.Results) + + if r.delayResults { + for _, name := range r.retvars { + // TODO(mdempsky): Use inlined position of name.Pos() instead? + name := name.(*ir.Name) + block.Append(ir.NewDecl(pos, ir.ODCL, name)) + name.Defn = as2 + } + } + + block.Append(as2) + } + + block.Append(ir.NewBranchStmt(pos, ir.OGOTO, r.retlabel)) + return ir.NewBlockStmt(pos, block) +} + +// expandInline reads in an extra copy of IR to populate +// fn.Inl.{Dcl,Body}. +func expandInline(fn *ir.Func, pri pkgReaderIndex) { + // TODO(mdempsky): Remove this function. It's currently needed for + // dwarfgen for some reason, but we should be able to provide it + // with the same information some other way. + + fndcls := len(fn.Dcl) + topdcls := len(typecheck.Target.Decls) + + tmpfn := ir.NewFunc(fn.Pos()) + tmpfn.Nname = ir.NewNameAt(fn.Nname.Pos(), fn.Sym()) + tmpfn.ClosureVars = fn.ClosureVars + + { + r := pri.asReader(relocBody, syncFuncBody) + r.setType(tmpfn.Nname, fn.Type()) + + // Don't change parameter's Sym/Nname fields. + r.funarghack = true + + r.funcBody(tmpfn) + } + + oldcurfn := ir.CurFunc + ir.CurFunc = tmpfn + + typecheck.Stmts(tmpfn.Body) + deadcode.Func(tmpfn) + + ir.CurFunc = oldcurfn + + used := usedLocals(tmpfn.Body) + + for _, name := range tmpfn.Dcl { + if name.Class != ir.PAUTO || used.Has(name) { + name.Curfn = fn + fn.Inl.Dcl = append(fn.Inl.Dcl, name) + } + } + fn.Inl.Body = tmpfn.Body + + // Double check that we didn't change fn.Dcl by accident. + assert(fndcls == len(fn.Dcl)) + + // typecheck.Stmts may have added function literals to + // typecheck.Target.Decls. Remove them again so we don't risk trying + // to compile them multiple times. + typecheck.Target.Decls = typecheck.Target.Decls[:topdcls] +} + +// usedLocals returns a set of local variables that are used within body. +func usedLocals(body []ir.Node) ir.NameSet { + var used ir.NameSet + ir.VisitList(body, func(n ir.Node) { + if n, ok := n.(*ir.Name); ok && n.Op() == ir.ONAME && n.Class == ir.PAUTO { + used.Add(n) + } + }) + return used +} diff --git a/src/cmd/compile/internal/noder/reader2.go b/src/cmd/compile/internal/noder/reader2.go new file mode 100644 index 00000000000..174bd3f5bdb --- /dev/null +++ b/src/cmd/compile/internal/noder/reader2.go @@ -0,0 +1,463 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "go/constant" + + "cmd/compile/internal/base" + "cmd/compile/internal/syntax" + "cmd/compile/internal/types2" + "cmd/internal/src" +) + +type pkgReader2 struct { + pkgDecoder + + check *types2.Checker + imports map[string]*types2.Package + + posBases []*syntax.PosBase + pkgs []*types2.Package + typs []types2.Type +} + +func readPackage2(check *types2.Checker, imports map[string]*types2.Package, input pkgDecoder) *types2.Package { + pr := pkgReader2{ + pkgDecoder: input, + + check: check, + imports: imports, + + posBases: make([]*syntax.PosBase, input.numElems(relocPosBase)), + pkgs: make([]*types2.Package, input.numElems(relocPkg)), + typs: make([]types2.Type, input.numElems(relocType)), + } + + r := pr.newReader(relocMeta, publicRootIdx, syncPublic) + pkg := r.pkg() + r.bool() // has init + + for i, n := 0, r.len(); i < n; i++ { + r.obj() + } + + r.sync(syncEOF) + + pkg.MarkComplete() + return pkg +} + +type reader2 struct { + decoder + + p *pkgReader2 + + tparams []*types2.TypeName +} + +func (pr *pkgReader2) newReader(k reloc, idx int, marker syncMarker) *reader2 { + return &reader2{ + decoder: pr.newDecoder(k, idx, marker), + p: pr, + } +} + +// @@@ Positions + +func (r *reader2) pos() syntax.Pos { + r.sync(syncPos) + if !r.bool() { + return syntax.Pos{} + } + + // TODO(mdempsky): Delta encoding. + posBase := r.posBase() + line := r.uint() + col := r.uint() + return syntax.MakePos(posBase, line, col) +} + +func (r *reader2) posBase() *syntax.PosBase { + return r.p.posBaseIdx(r.reloc(relocPosBase)) +} + +func (pr *pkgReader2) posBaseIdx(idx int) *syntax.PosBase { + if b := pr.posBases[idx]; b != nil { + return b + } + + r := pr.newReader(relocPosBase, idx, syncPosBase) + var b *syntax.PosBase + + filename := r.string() + _ = r.string() // absolute file name + + if r.bool() { + b = syntax.NewFileBase(filename) + } else { + pos := r.pos() + line := r.uint() + col := r.uint() + b = syntax.NewLineBase(pos, filename, line, col) + } + + pr.posBases[idx] = b + return b +} + +// @@@ Packages + +func (r *reader2) pkg() *types2.Package { + r.sync(syncPkg) + return r.p.pkgIdx(r.reloc(relocPkg)) +} + +func (pr *pkgReader2) pkgIdx(idx int) *types2.Package { + // TODO(mdempsky): Consider using some non-nil pointer to indicate + // the universe scope, so we don't need to keep re-reading it. + if pkg := pr.pkgs[idx]; pkg != nil { + return pkg + } + + pkg := pr.newReader(relocPkg, idx, syncPkgDef).doPkg() + pr.pkgs[idx] = pkg + return pkg +} + +func (r *reader2) doPkg() *types2.Package { + path := r.string() + if path == "builtin" { + return nil // universe + } + if path == "" { + path = r.p.pkgPath + } + + if pkg := r.p.imports[path]; pkg != nil { + return pkg + } + + name := r.string() + height := r.len() + + pkg := types2.NewPackageHeight(path, name, height) + r.p.imports[path] = pkg + + // TODO(mdempsky): The list of imported packages is important for + // go/types, but we could probably skip populating it for types2. + imports := make([]*types2.Package, r.len()) + for i := range imports { + imports[i] = r.pkg() + } + pkg.SetImports(imports) + + return pkg +} + +// @@@ Types + +func (r *reader2) typ() types2.Type { + r.sync(syncType) + return r.p.typIdx(r.reloc(relocType), r.tparams) +} + +func (pr *pkgReader2) typIdx(idx int, tparams []*types2.TypeName) types2.Type { + if typ := pr.typs[idx]; typ != nil { + return typ + } + + r := pr.newReader(relocType, idx, syncTypeIdx) + r.tparams = tparams + typ := r.doTyp() + assert(typ != nil) + + if pr.typs[idx] != nil { + // See comment in pkgReader.typIdx. + return pr.typs[idx] + } + + if len(tparams) == 0 { + pr.typs[idx] = typ + } + + return typ +} + +func (r *reader2) doTyp() (res types2.Type) { + switch tag := codeType(r.code(syncType)); tag { + default: + base.FatalfAt(src.NoXPos, "unhandled type tag: %v", tag) + panic("unreachable") + + case typeBasic: + return types2.Typ[r.len()] + + case typeNamed: + obj, targs := r.obj() + name := obj.(*types2.TypeName) + if len(targs) != 0 { + return r.p.check.InstantiateLazy(syntax.Pos{}, name.Type(), targs) + } + return name.Type() + + case typeTypeParam: + idx := r.len() + return r.tparams[idx].Type().(*types2.TypeParam) + + case typeArray: + len := int64(r.uint64()) + return types2.NewArray(r.typ(), len) + case typeChan: + dir := types2.ChanDir(r.len()) + return types2.NewChan(dir, r.typ()) + case typeMap: + return types2.NewMap(r.typ(), r.typ()) + case typePointer: + return types2.NewPointer(r.typ()) + case typeSignature: + return r.signature(nil) + case typeSlice: + return types2.NewSlice(r.typ()) + case typeStruct: + return r.structType() + case typeInterface: + return r.interfaceType() + case typeUnion: + return r.unionType() + } +} + +func (r *reader2) structType() *types2.Struct { + fields := make([]*types2.Var, r.len()) + var tags []string + for i := range fields { + pos := r.pos() + pkg, name := r.selector() + ftyp := r.typ() + tag := r.string() + embedded := r.bool() + + fields[i] = types2.NewField(pos, pkg, name, ftyp, embedded) + if tag != "" { + for len(tags) < i { + tags = append(tags, "") + } + tags = append(tags, tag) + } + } + return types2.NewStruct(fields, tags) +} + +func (r *reader2) unionType() *types2.Union { + terms := make([]types2.Type, r.len()) + tildes := make([]bool, len(terms)) + for i := range terms { + terms[i] = r.typ() + tildes[i] = r.bool() + } + return types2.NewUnion(terms, tildes) +} + +func (r *reader2) interfaceType() *types2.Interface { + methods := make([]*types2.Func, r.len()) + embeddeds := make([]types2.Type, r.len()) + + for i := range methods { + pos := r.pos() + pkg, name := r.selector() + mtyp := r.signature(nil) + methods[i] = types2.NewFunc(pos, pkg, name, mtyp) + } + + for i := range embeddeds { + embeddeds[i] = r.typ() + } + + typ := types2.NewInterfaceType(methods, embeddeds) + typ.Complete() + return typ +} + +func (r *reader2) signature(recv *types2.Var) *types2.Signature { + r.sync(syncSignature) + + params := r.params() + results := r.params() + variadic := r.bool() + + return types2.NewSignature(recv, params, results, variadic) +} + +func (r *reader2) params() *types2.Tuple { + r.sync(syncParams) + params := make([]*types2.Var, r.len()) + for i := range params { + params[i] = r.param() + } + return types2.NewTuple(params...) +} + +func (r *reader2) param() *types2.Var { + r.sync(syncParam) + + pos := r.pos() + pkg, name := r.localIdent() + typ := r.typ() + + return types2.NewParam(pos, pkg, name, typ) +} + +// @@@ Objects + +func (r *reader2) obj() (types2.Object, []types2.Type) { + r.sync(syncObject) + + pkg, name := r.p.objIdx(r.reloc(relocObj)) + obj := pkg.Scope().Lookup(name) + + targs := make([]types2.Type, r.len()) + for i := range targs { + targs[i] = r.typ() + } + + return obj, targs +} + +func (pr *pkgReader2) objIdx(idx int) (*types2.Package, string) { + r := pr.newReader(relocObj, idx, syncObject1) + objPkg, objName := r.qualifiedIdent() + assert(objName != "") + + bounds := r.typeParamBounds() + tag := codeObj(r.code(syncCodeObj)) + + if tag == objStub { + assert(objPkg == nil) + return objPkg, objName + } + + objPkg.Scope().InsertLazy(objName, func() types2.Object { + switch tag { + default: + panic("weird") + + case objAlias: + pos := r.pos() + typ := r.typ() + return types2.NewTypeName(pos, objPkg, objName, typ) + + case objConst: + pos := r.pos() + typ, val := r.value() + return types2.NewConst(pos, objPkg, objName, typ, val) + + case objFunc: + pos := r.pos() + r.typeParamNames(bounds) + sig := r.signature(nil) + if len(r.tparams) != 0 { + sig.SetTParams(r.tparams) + } + return types2.NewFunc(pos, objPkg, objName, sig) + + case objType: + pos := r.pos() + + return types2.NewTypeNameLazy(pos, objPkg, objName, func(named *types2.Named) (tparams []*types2.TypeName, underlying types2.Type, methods []*types2.Func) { + r.typeParamNames(bounds) + if len(r.tparams) != 0 { + tparams = r.tparams + } + + // TODO(mdempsky): Rewrite receiver types to underlying is an + // Interface? The go/types importer does this (I think because + // unit tests expected that), but cmd/compile doesn't care + // about it, so maybe we can avoid worrying about that here. + underlying = r.typ().Underlying() + + methods = make([]*types2.Func, r.len()) + for i := range methods { + methods[i] = r.method(bounds) + } + + return + }) + + case objVar: + pos := r.pos() + typ := r.typ() + return types2.NewVar(pos, objPkg, objName, typ) + } + }) + + return objPkg, objName +} + +func (r *reader2) value() (types2.Type, constant.Value) { + r.sync(syncValue) + return r.typ(), r.rawValue() +} + +func (r *reader2) typeParamBounds() []int { + r.sync(syncTypeParamBounds) + + // exported types never have implicit type parameters + // TODO(mdempsky): Hide this from public importer. + assert(r.len() == 0) + + bounds := make([]int, r.len()) + for i := range bounds { + r.sync(syncType) + bounds[i] = r.reloc(relocType) + } + return bounds +} + +func (r *reader2) typeParamNames(bounds []int) { + r.sync(syncTypeParamNames) + + r.tparams = make([]*types2.TypeName, len(bounds)) + + for i := range r.tparams { + pos := r.pos() + pkg, name := r.localIdent() + + obj := types2.NewTypeName(pos, pkg, name, nil) + r.p.check.NewTypeParam(obj, i, nil) + r.tparams[i] = obj + } + + for i, tparam := range r.tparams { + bound := r.p.typIdx(bounds[i], r.tparams) + tparam.Type().(*types2.TypeParam).SetBound(bound) + } +} + +func (r *reader2) method(bounds []int) *types2.Func { + r.sync(syncMethod) + pos := r.pos() + pkg, name := r.selector() + + r.typeParamNames(bounds) + sig := r.signature(r.param()) + if len(r.tparams) != 0 { + sig.SetRParams(r.tparams) + } + + _ = r.pos() // TODO(mdempsky): Remove; this is a hacker for linker.go. + return types2.NewFunc(pos, pkg, name, sig) +} + +func (r *reader2) qualifiedIdent() (*types2.Package, string) { return r.ident(syncSym) } +func (r *reader2) localIdent() (*types2.Package, string) { return r.ident(syncLocalIdent) } +func (r *reader2) selector() (*types2.Package, string) { return r.ident(syncSelector) } + +func (r *reader2) ident(marker syncMarker) (*types2.Package, string) { + r.sync(marker) + return r.pkg(), r.string() +} diff --git a/src/cmd/compile/internal/noder/reloc.go b/src/cmd/compile/internal/noder/reloc.go new file mode 100644 index 00000000000..961de494199 --- /dev/null +++ b/src/cmd/compile/internal/noder/reloc.go @@ -0,0 +1,40 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +// A reloc indicates a particular section within a unified IR export. +// +// TODO(mdempsky): Rename to "section" or something similar? +type reloc int + +// A relocEnt (relocation entry) is an entry in an atom's local +// reference table. +// +// TODO(mdempsky): Rename this too. +type relocEnt struct { + kind reloc + idx int +} + +// Reserved indices within the meta relocation section. +const ( + publicRootIdx = 0 + privateRootIdx = 1 +) + +const ( + relocString reloc = iota + relocMeta + relocPosBase + relocPkg + relocType + relocObj + relocObjExt + relocBody + + numRelocs = iota +) diff --git a/src/cmd/compile/internal/noder/sync.go b/src/cmd/compile/internal/noder/sync.go new file mode 100644 index 00000000000..d77a7844798 --- /dev/null +++ b/src/cmd/compile/internal/noder/sync.go @@ -0,0 +1,154 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +const debug = true + +type syncMarker int + +//go:generate stringer -type=syncMarker -trimprefix=sync + +// TODO(mdempsky): Cleanup unneeded sync markers. + +// TODO(mdempsky): Split these markers into public/stable markers, and +// private ones. Also, trim unused ones. +const ( + _ syncMarker = iota + syncNode + syncBool + syncInt64 + syncUint64 + syncString + syncPos + syncPkg + syncSym + syncSelector + syncKind + syncType + syncTypePkg + syncSignature + syncParam + syncOp + syncObject + syncExpr + syncStmt + syncDecl + syncConstDecl + syncFuncDecl + syncTypeDecl + syncVarDecl + syncPragma + syncValue + syncEOF + syncMethod + syncFuncBody + syncUse + syncUseObj + syncObjectIdx + syncTypeIdx + syncBOF + syncEntry + syncOpenScope + syncCloseScope + syncGlobal + syncLocal + syncDefine + syncDefLocal + syncUseLocal + syncDefGlobal + syncUseGlobal + syncTypeParams + syncUseLabel + syncDefLabel + syncFuncLit + syncCommonFunc + syncBodyRef + syncLinksymExt + syncHack + syncSetlineno + syncName + syncImportDecl + syncDeclNames + syncDeclName + syncExprList + syncExprs + syncWrapname + syncTypeExpr + syncTypeExprOrNil + syncChanDir + syncParams + syncCloseAnotherScope + syncSum + syncUnOp + syncBinOp + syncStructType + syncInterfaceType + syncPackname + syncEmbedded + syncStmts + syncStmtsFall + syncStmtFall + syncBlockStmt + syncIfStmt + syncForStmt + syncSwitchStmt + syncRangeStmt + syncCaseClause + syncCommClause + syncSelectStmt + syncDecls + syncLabeledStmt + syncCompLit + + sync1 + sync2 + sync3 + sync4 + + syncN + syncDefImplicit + syncUseName + syncUseObjLocal + syncAddLocal + syncBothSignature + syncSetUnderlying + syncLinkname + syncStmt1 + syncStmtsEnd + syncDeclare + syncTopDecls + syncTopConstDecl + syncTopFuncDecl + syncTopTypeDecl + syncTopVarDecl + syncObject1 + syncAddBody + syncLabel + syncFuncExt + syncMethExt + syncOptLabel + syncScalar + syncStmtDecls + syncDeclLocal + syncObjLocal + syncObjLocal1 + syncDeclareLocal + syncPublic + syncPrivate + syncRelocs + syncReloc + syncUseReloc + syncVarExt + syncPkgDef + syncTypeExt + syncVal + syncCodeObj + syncPosBase + syncLocalIdent + syncTypeParamNames + syncTypeParamBounds +) diff --git a/src/cmd/compile/internal/noder/syncmarker_string.go b/src/cmd/compile/internal/noder/syncmarker_string.go new file mode 100644 index 00000000000..3eb88fb9118 --- /dev/null +++ b/src/cmd/compile/internal/noder/syncmarker_string.go @@ -0,0 +1,152 @@ +// Code generated by "stringer -type=syncMarker -trimprefix=sync"; DO NOT EDIT. + +package noder + +import "strconv" + +func _() { + // An "invalid array index" compiler error signifies that the constant values have changed. + // Re-run the stringer command to generate them again. + var x [1]struct{} + _ = x[syncNode-1] + _ = x[syncBool-2] + _ = x[syncInt64-3] + _ = x[syncUint64-4] + _ = x[syncString-5] + _ = x[syncPos-6] + _ = x[syncPkg-7] + _ = x[syncSym-8] + _ = x[syncSelector-9] + _ = x[syncKind-10] + _ = x[syncType-11] + _ = x[syncTypePkg-12] + _ = x[syncSignature-13] + _ = x[syncParam-14] + _ = x[syncOp-15] + _ = x[syncObject-16] + _ = x[syncExpr-17] + _ = x[syncStmt-18] + _ = x[syncDecl-19] + _ = x[syncConstDecl-20] + _ = x[syncFuncDecl-21] + _ = x[syncTypeDecl-22] + _ = x[syncVarDecl-23] + _ = x[syncPragma-24] + _ = x[syncValue-25] + _ = x[syncEOF-26] + _ = x[syncMethod-27] + _ = x[syncFuncBody-28] + _ = x[syncUse-29] + _ = x[syncUseObj-30] + _ = x[syncObjectIdx-31] + _ = x[syncTypeIdx-32] + _ = x[syncBOF-33] + _ = x[syncEntry-34] + _ = x[syncOpenScope-35] + _ = x[syncCloseScope-36] + _ = x[syncGlobal-37] + _ = x[syncLocal-38] + _ = x[syncDefine-39] + _ = x[syncDefLocal-40] + _ = x[syncUseLocal-41] + _ = x[syncDefGlobal-42] + _ = x[syncUseGlobal-43] + _ = x[syncTypeParams-44] + _ = x[syncUseLabel-45] + _ = x[syncDefLabel-46] + _ = x[syncFuncLit-47] + _ = x[syncCommonFunc-48] + _ = x[syncBodyRef-49] + _ = x[syncLinksymExt-50] + _ = x[syncHack-51] + _ = x[syncSetlineno-52] + _ = x[syncName-53] + _ = x[syncImportDecl-54] + _ = x[syncDeclNames-55] + _ = x[syncDeclName-56] + _ = x[syncExprList-57] + _ = x[syncExprs-58] + _ = x[syncWrapname-59] + _ = x[syncTypeExpr-60] + _ = x[syncTypeExprOrNil-61] + _ = x[syncChanDir-62] + _ = x[syncParams-63] + _ = x[syncCloseAnotherScope-64] + _ = x[syncSum-65] + _ = x[syncUnOp-66] + _ = x[syncBinOp-67] + _ = x[syncStructType-68] + _ = x[syncInterfaceType-69] + _ = x[syncPackname-70] + _ = x[syncEmbedded-71] + _ = x[syncStmts-72] + _ = x[syncStmtsFall-73] + _ = x[syncStmtFall-74] + _ = x[syncBlockStmt-75] + _ = x[syncIfStmt-76] + _ = x[syncForStmt-77] + _ = x[syncSwitchStmt-78] + _ = x[syncRangeStmt-79] + _ = x[syncCaseClause-80] + _ = x[syncCommClause-81] + _ = x[syncSelectStmt-82] + _ = x[syncDecls-83] + _ = x[syncLabeledStmt-84] + _ = x[syncCompLit-85] + _ = x[sync1-86] + _ = x[sync2-87] + _ = x[sync3-88] + _ = x[sync4-89] + _ = x[syncN-90] + _ = x[syncDefImplicit-91] + _ = x[syncUseName-92] + _ = x[syncUseObjLocal-93] + _ = x[syncAddLocal-94] + _ = x[syncBothSignature-95] + _ = x[syncSetUnderlying-96] + _ = x[syncLinkname-97] + _ = x[syncStmt1-98] + _ = x[syncStmtsEnd-99] + _ = x[syncDeclare-100] + _ = x[syncTopDecls-101] + _ = x[syncTopConstDecl-102] + _ = x[syncTopFuncDecl-103] + _ = x[syncTopTypeDecl-104] + _ = x[syncTopVarDecl-105] + _ = x[syncObject1-106] + _ = x[syncAddBody-107] + _ = x[syncLabel-108] + _ = x[syncFuncExt-109] + _ = x[syncMethExt-110] + _ = x[syncOptLabel-111] + _ = x[syncScalar-112] + _ = x[syncStmtDecls-113] + _ = x[syncDeclLocal-114] + _ = x[syncObjLocal-115] + _ = x[syncObjLocal1-116] + _ = x[syncDeclareLocal-117] + _ = x[syncPublic-118] + _ = x[syncPrivate-119] + _ = x[syncRelocs-120] + _ = x[syncReloc-121] + _ = x[syncUseReloc-122] + _ = x[syncVarExt-123] + _ = x[syncPkgDef-124] + _ = x[syncTypeExt-125] + _ = x[syncVal-126] + _ = x[syncCodeObj-127] + _ = x[syncPosBase-128] + _ = x[syncLocalIdent-129] +} + +const _syncMarker_name = "NodeBoolInt64Uint64StringPosPkgSymSelectorKindTypeTypePkgSignatureParamOpObjectExprStmtDeclConstDeclFuncDeclTypeDeclVarDeclPragmaValueEOFMethodFuncBodyUseUseObjObjectIdxTypeIdxBOFEntryOpenScopeCloseScopeGlobalLocalDefineDefLocalUseLocalDefGlobalUseGlobalTypeParamsUseLabelDefLabelFuncLitCommonFuncBodyRefLinksymExtHackSetlinenoNameImportDeclDeclNamesDeclNameExprListExprsWrapnameTypeExprTypeExprOrNilChanDirParamsCloseAnotherScopeSumUnOpBinOpStructTypeInterfaceTypePacknameEmbeddedStmtsStmtsFallStmtFallBlockStmtIfStmtForStmtSwitchStmtRangeStmtCaseClauseCommClauseSelectStmtDeclsLabeledStmtCompLit1234NDefImplicitUseNameUseObjLocalAddLocalBothSignatureSetUnderlyingLinknameStmt1StmtsEndDeclareTopDeclsTopConstDeclTopFuncDeclTopTypeDeclTopVarDeclObject1AddBodyLabelFuncExtMethExtOptLabelScalarStmtDeclsDeclLocalObjLocalObjLocal1DeclareLocalPublicPrivateRelocsRelocUseRelocVarExtPkgDefTypeExtValCodeObjPosBaseLocalIdent" + +var _syncMarker_index = [...]uint16{0, 4, 8, 13, 19, 25, 28, 31, 34, 42, 46, 50, 57, 66, 71, 73, 79, 83, 87, 91, 100, 108, 116, 123, 129, 134, 137, 143, 151, 154, 160, 169, 176, 179, 184, 193, 203, 209, 214, 220, 228, 236, 245, 254, 264, 272, 280, 287, 297, 304, 314, 318, 327, 331, 341, 350, 358, 366, 371, 379, 387, 400, 407, 413, 430, 433, 437, 442, 452, 465, 473, 481, 486, 495, 503, 512, 518, 525, 535, 544, 554, 564, 574, 579, 590, 597, 598, 599, 600, 601, 602, 613, 620, 631, 639, 652, 665, 673, 678, 686, 693, 701, 713, 724, 735, 745, 752, 759, 764, 771, 778, 786, 792, 801, 810, 818, 827, 839, 845, 852, 858, 863, 871, 877, 883, 890, 893, 900, 907, 917} + +func (i syncMarker) String() string { + i -= 1 + if i < 0 || i >= syncMarker(len(_syncMarker_index)-1) { + return "syncMarker(" + strconv.FormatInt(int64(i+1), 10) + ")" + } + return _syncMarker_name[_syncMarker_index[i]:_syncMarker_index[i+1]] +} diff --git a/src/cmd/compile/internal/noder/unified.go b/src/cmd/compile/internal/noder/unified.go new file mode 100644 index 00000000000..9a41ea9dfe4 --- /dev/null +++ b/src/cmd/compile/internal/noder/unified.go @@ -0,0 +1,276 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "bytes" + "fmt" + "internal/goversion" + "io" + "runtime" + "sort" + + "cmd/compile/internal/base" + "cmd/compile/internal/inline" + "cmd/compile/internal/ir" + "cmd/compile/internal/typecheck" + "cmd/compile/internal/types" + "cmd/compile/internal/types2" + "cmd/internal/src" +) + +// localPkgReader holds the package reader used for reading the local +// package. It exists so the unified IR linker can refer back to it +// later. +var localPkgReader *pkgReader + +// useUnifiedIR reports whether the unified IR frontend should be +// used; and if so, uses it to construct the local package's IR. +func useUnifiedIR(noders []*noder) { + inline.NewInline = InlineCall + + if !quirksMode() { + writeNewExportFunc = writeNewExport + } + + newReadImportFunc = func(data string, pkg1 *types.Pkg, check *types2.Checker, packages map[string]*types2.Package) (pkg2 *types2.Package, err error) { + pr := newPkgDecoder(pkg1.Path, data) + + // Read package descriptors for both types2 and compiler backend. + readPackage(newPkgReader(pr), pkg1) + pkg2 = readPackage2(check, packages, pr) + return + } + + data := writePkgStub(noders) + + // We already passed base.Flag.Lang to types2 to handle validating + // the user's source code. Bump it up now to the current version and + // re-parse, so typecheck doesn't complain if we construct IR that + // utilizes newer Go features. + base.Flag.Lang = fmt.Sprintf("go1.%d", goversion.Version) + types.ParseLangFlag() + + assert(types.LocalPkg.Path == "") + types.LocalPkg.Height = 0 // reset so pkgReader.pkgIdx doesn't complain + target := typecheck.Target + + typecheck.TypecheckAllowed = true + + localPkgReader = newPkgReader(newPkgDecoder(types.LocalPkg.Path, data)) + readPackage(localPkgReader, types.LocalPkg) + + r := localPkgReader.newReader(relocMeta, privateRootIdx, syncPrivate) + r.ext = r + r.pkgInit(types.LocalPkg, target) + + // Don't use range--bodyIdx can add closures to todoBodies. + for len(todoBodies) > 0 { + // The order we expand bodies doesn't matter, so pop from the end + // to reduce todoBodies reallocations if it grows further. + fn := todoBodies[len(todoBodies)-1] + todoBodies = todoBodies[:len(todoBodies)-1] + + pri, ok := bodyReader[fn] + assert(ok) + pri.funcBody(fn) + + // Instantiated generic function: add to Decls for typechecking + // and compilation. + if len(pri.implicits) != 0 && fn.OClosure == nil { + target.Decls = append(target.Decls, fn) + } + } + todoBodies = nil + + // Don't use range--typecheck can add closures to Target.Decls. + for i := 0; i < len(target.Decls); i++ { + target.Decls[i] = typecheck.Stmt(target.Decls[i]) + } + + // Don't use range--typecheck can add closures to Target.Decls. + for i := 0; i < len(target.Decls); i++ { + if fn, ok := target.Decls[i].(*ir.Func); ok { + if base.Flag.W > 1 { + s := fmt.Sprintf("\nbefore typecheck %v", fn) + ir.Dump(s, fn) + } + ir.CurFunc = fn + typecheck.Stmts(fn.Body) + if base.Flag.W > 1 { + s := fmt.Sprintf("\nafter typecheck %v", fn) + ir.Dump(s, fn) + } + } + } + + base.ExitIfErrors() // just in case +} + +// writePkgStub type checks the given parsed source files and then +// returns +func writePkgStub(noders []*noder) string { + m, pkg, info := checkFiles(noders) + + pw := newPkgWriter(m, pkg, info) + + pw.collectDecls(noders) + + publicRootWriter := pw.newWriter(relocMeta, syncPublic) + privateRootWriter := pw.newWriter(relocMeta, syncPrivate) + + assert(publicRootWriter.idx == publicRootIdx) + assert(privateRootWriter.idx == privateRootIdx) + + { + w := publicRootWriter + w.pkg(pkg) + w.bool(false) // has init; XXX + + scope := pkg.Scope() + names := scope.Names() + w.len(len(names)) + for _, name := range scope.Names() { + w.obj(scope.Lookup(name), nil) + } + + w.sync(syncEOF) + w.flush() + } + + { + w := privateRootWriter + w.ext = w + w.pkgInit(noders) + w.flush() + } + + var sb bytes.Buffer // TODO(mdempsky): strings.Builder after #44505 is resolved + pw.dump(&sb) + + // At this point, we're done with types2. Make sure the package is + // garbage collected. + freePackage(pkg) + + return sb.String() +} + +// freePackage ensures the given package is garbage collected. +func freePackage(pkg *types2.Package) { + // Set a finalizer on pkg so we can detect if/when it's collected. + done := make(chan struct{}) + runtime.SetFinalizer(pkg, func(*types2.Package) { close(done) }) + + // Important: objects involved in cycles are not finalized, so zero + // out pkg to break its cycles and allow the finalizer to run. + *pkg = types2.Package{} + + // It typically takes just 1 or 2 cycles to release pkg, but it + // doesn't hurt to try a few more times. + for i := 0; i < 10; i++ { + select { + case <-done: + return + default: + runtime.GC() + } + } + + base.Fatalf("package never finalized") +} + +func readPackage(pr *pkgReader, importpkg *types.Pkg) { + r := pr.newReader(relocMeta, publicRootIdx, syncPublic) + + pkg := r.pkg() + assert(pkg == importpkg) + + if r.bool() { + sym := pkg.Lookup(".inittask") + task := ir.NewNameAt(src.NoXPos, sym) + task.Class = ir.PEXTERN + sym.Def = task + } + + for i, n := 0, r.len(); i < n; i++ { + r.sync(syncObject) + idx := r.reloc(relocObj) + assert(r.len() == 0) + + path, name, code, _ := r.p.peekObj(idx) + if code != objStub { + objReader[types.NewPkg(path, "").Lookup(name)] = pkgReaderIndex{pr, idx, nil} + } + } +} + +func writeNewExport(out io.Writer) { + l := linker{ + pw: newPkgEncoder(), + + pkgs: make(map[string]int), + decls: make(map[*types.Sym]int), + } + + publicRootWriter := l.pw.newEncoder(relocMeta, syncPublic) + assert(publicRootWriter.idx == publicRootIdx) + + var selfPkgIdx int + + { + pr := localPkgReader + r := pr.newDecoder(relocMeta, publicRootIdx, syncPublic) + + r.sync(syncPkg) + selfPkgIdx = l.relocIdx(pr, relocPkg, r.reloc(relocPkg)) + + r.bool() // has init + + for i, n := 0, r.len(); i < n; i++ { + r.sync(syncObject) + idx := r.reloc(relocObj) + assert(r.len() == 0) + + xpath, xname, xtag, _ := pr.peekObj(idx) + assert(xpath == pr.pkgPath) + assert(xtag != objStub) + + if types.IsExported(xname) { + l.relocIdx(pr, relocObj, idx) + } + } + + r.sync(syncEOF) + } + + { + var idxs []int + for _, idx := range l.decls { + idxs = append(idxs, idx) + } + sort.Ints(idxs) + + w := publicRootWriter + + w.sync(syncPkg) + w.reloc(relocPkg, selfPkgIdx) + + w.bool(typecheck.Lookup(".inittask").Def != nil) + + w.len(len(idxs)) + for _, idx := range idxs { + w.sync(syncObject) + w.reloc(relocObj, idx) + w.len(0) + } + + w.sync(syncEOF) + w.flush() + } + + l.pw.dump(out) +} diff --git a/src/cmd/compile/internal/noder/writer.go b/src/cmd/compile/internal/noder/writer.go new file mode 100644 index 00000000000..b39dd8651b3 --- /dev/null +++ b/src/cmd/compile/internal/noder/writer.go @@ -0,0 +1,1746 @@ +// UNREVIEWED + +// Copyright 2021 The Go Authors. All rights reserved. +// Use of this source code is governed by a BSD-style +// license that can be found in the LICENSE file. + +package noder + +import ( + "fmt" + "go/constant" + + "cmd/compile/internal/base" + "cmd/compile/internal/ir" + "cmd/compile/internal/syntax" + "cmd/compile/internal/types2" +) + +type pkgWriter struct { + pkgEncoder + + m posMap + curpkg *types2.Package + info *types2.Info + + posBasesIdx map[*syntax.PosBase]int + pkgsIdx map[*types2.Package]int + typsIdx map[types2.Type]int + globalsIdx map[types2.Object]int + + funDecls map[*types2.Func]*syntax.FuncDecl + typDecls map[*types2.TypeName]typeDeclGen + + linknames map[types2.Object]string + cgoPragmas [][]string + + dups dupTypes +} + +func newPkgWriter(m posMap, pkg *types2.Package, info *types2.Info) *pkgWriter { + return &pkgWriter{ + pkgEncoder: newPkgEncoder(), + + m: m, + curpkg: pkg, + info: info, + + pkgsIdx: make(map[*types2.Package]int), + globalsIdx: make(map[types2.Object]int), + typsIdx: make(map[types2.Type]int), + + posBasesIdx: make(map[*syntax.PosBase]int), + + funDecls: make(map[*types2.Func]*syntax.FuncDecl), + typDecls: make(map[*types2.TypeName]typeDeclGen), + + linknames: make(map[types2.Object]string), + } +} + +func (pw *pkgWriter) errorf(p poser, msg string, args ...interface{}) { + base.ErrorfAt(pw.m.pos(p), msg, args...) +} + +func (pw *pkgWriter) fatalf(p poser, msg string, args ...interface{}) { + base.FatalfAt(pw.m.pos(p), msg, args...) +} + +func (pw *pkgWriter) unexpected(what string, p poser) { + pw.fatalf(p, "unexpected %s: %v (%T)", what, p, p) +} + +type writer struct { + p *pkgWriter + + encoder + + // For writing out object descriptions, ext points to the extension + // writer for where we can write the compiler's private extension + // details for the object. + // + // TODO(mdempsky): This is a little hacky, but works easiest with + // the way things are currently. + ext *writer + + // TODO(mdempsky): We should be able to prune localsIdx whenever a + // scope closes, and then maybe we can just use the same map for + // storing the TypeParams too (as their TypeName instead). + + // type parameters. explicitIdx has the type parameters declared on + // the current object, while implicitIdx has the type parameters + // declared on the enclosing object (if any). + // + // TODO(mdempsky): Merge these back together, now that I've got them + // working. + implicitIdx map[*types2.TypeParam]int + explicitIdx map[*types2.TypeParam]int + + // variables declared within this function + localsIdx map[types2.Object]int +} + +func (pw *pkgWriter) newWriter(k reloc, marker syncMarker) *writer { + return &writer{ + encoder: pw.newEncoder(k, marker), + p: pw, + } +} + +// @@@ Positions + +func (w *writer) pos(p poser) { + w.sync(syncPos) + pos := p.Pos() + + // TODO(mdempsky): Track down the remaining cases here and fix them. + if !w.bool(pos.IsKnown()) { + return + } + + // TODO(mdempsky): Delta encoding. Also, if there's a b-side, update + // its position base too (but not vice versa!). + w.posBase(pos.Base()) + w.uint(pos.Line()) + w.uint(pos.Col()) +} + +func (w *writer) posBase(b *syntax.PosBase) { + w.reloc(relocPosBase, w.p.posBaseIdx(b)) +} + +func (pw *pkgWriter) posBaseIdx(b *syntax.PosBase) int { + if idx, ok := pw.posBasesIdx[b]; ok { + return idx + } + + w := pw.newWriter(relocPosBase, syncPosBase) + w.p.posBasesIdx[b] = w.idx + + // TODO(mdempsky): What exactly does "fileh" do anyway? Is writing + // out both of these strings really the right thing to do here? + fn := b.Filename() + w.string(fn) + w.string(fileh(fn)) + + if !w.bool(b.IsFileBase()) { + w.pos(b) + w.uint(b.Line()) + w.uint(b.Col()) + } + + return w.flush() +} + +// @@@ Packages + +func (w *writer) pkg(pkg *types2.Package) { + w.sync(syncPkg) + w.reloc(relocPkg, w.p.pkgIdx(pkg)) +} + +func (pw *pkgWriter) pkgIdx(pkg *types2.Package) int { + if idx, ok := pw.pkgsIdx[pkg]; ok { + return idx + } + + w := pw.newWriter(relocPkg, syncPkgDef) + pw.pkgsIdx[pkg] = w.idx + + if pkg == nil { + w.string("builtin") + } else { + var path string + if pkg != w.p.curpkg { + path = pkg.Path() + } + w.string(path) + w.string(pkg.Name()) + w.len(pkg.Height()) + + w.len(len(pkg.Imports())) + for _, imp := range pkg.Imports() { + w.pkg(imp) + } + } + + return w.flush() +} + +// @@@ Types + +func (w *writer) typ(typ types2.Type) { + w.sync(syncType) + + if quirksMode() { + typ = w.p.dups.orig(typ) + } + + w.reloc(relocType, w.p.typIdx(typ, w.implicitIdx, w.explicitIdx)) +} + +func (pw *pkgWriter) typIdx(typ types2.Type, implicitIdx, explicitIdx map[*types2.TypeParam]int) int { + if idx, ok := pw.typsIdx[typ]; ok { + return idx + } + + w := pw.newWriter(relocType, syncTypeIdx) + w.implicitIdx = implicitIdx + w.explicitIdx = explicitIdx + + pw.typsIdx[typ] = w.idx // handle cycles + w.doTyp(typ) + return w.flush() +} + +func (w *writer) doTyp(typ types2.Type) { + switch typ := typ.(type) { + default: + base.Fatalf("unexpected type: %v (%T)", typ, typ) + + case *types2.Basic: + if kind := typ.Kind(); types2.Typ[kind] == typ { + w.code(typeBasic) + w.len(int(kind)) + break + } + + // Handle "byte" and "rune" as references to their TypeName. + obj := types2.Universe.Lookup(typ.Name()) + assert(obj.Type() == typ) + + w.code(typeNamed) + w.obj(obj, nil) + + case *types2.Named: + // Type aliases can refer to uninstantiated generic types, so we + // might see len(TParams) != 0 && len(TArgs) == 0 here. + // TODO(mdempsky): Revisit after #46477 is resolved. + assert(len(typ.TParams()) == len(typ.TArgs()) || len(typ.TArgs()) == 0) + + // TODO(mdempsky): Why do we need to loop here? + orig := typ + for orig.TArgs() != nil { + orig = orig.Orig() + } + + w.code(typeNamed) + w.obj(orig.Obj(), typ.TArgs()) + + case *types2.TypeParam: + w.code(typeTypeParam) + if idx, ok := w.implicitIdx[typ]; ok { + w.len(idx) + } else if idx, ok := w.explicitIdx[typ]; ok { + w.len(len(w.implicitIdx) + idx) + } else { + w.p.fatalf(typ.Obj(), "%v not in %v or %v", typ, w.implicitIdx, w.explicitIdx) + } + + case *types2.Array: + w.code(typeArray) + w.uint64(uint64(typ.Len())) + w.typ(typ.Elem()) + + case *types2.Chan: + w.code(typeChan) + w.len(int(typ.Dir())) + w.typ(typ.Elem()) + + case *types2.Map: + w.code(typeMap) + w.typ(typ.Key()) + w.typ(typ.Elem()) + + case *types2.Pointer: + w.code(typePointer) + w.typ(typ.Elem()) + + case *types2.Signature: + assert(typ.TParams() == nil) + w.code(typeSignature) + w.signature(typ) + + case *types2.Slice: + w.code(typeSlice) + w.typ(typ.Elem()) + + case *types2.Struct: + w.code(typeStruct) + w.structType(typ) + + case *types2.Interface: + w.code(typeInterface) + w.interfaceType(typ) + + case *types2.Union: + w.code(typeUnion) + w.unionType(typ) + } +} + +func (w *writer) structType(typ *types2.Struct) { + w.len(typ.NumFields()) + for i := 0; i < typ.NumFields(); i++ { + f := typ.Field(i) + w.pos(f) + w.selector(f) + w.typ(f.Type()) + w.string(typ.Tag(i)) + w.bool(f.Embedded()) + } +} + +func (w *writer) unionType(typ *types2.Union) { + w.len(typ.NumTerms()) + for i := 0; i < typ.NumTerms(); i++ { + term, tilde := typ.Term(i) + w.typ(term) + w.bool(tilde) + } +} + +func (w *writer) interfaceType(typ *types2.Interface) { + w.len(typ.NumExplicitMethods()) + w.len(typ.NumEmbeddeds()) + + for i := 0; i < typ.NumExplicitMethods(); i++ { + m := typ.ExplicitMethod(i) + sig := m.Type().(*types2.Signature) + assert(sig.TParams() == nil) + + w.pos(m) + w.selector(m) + w.signature(sig) + } + + for i := 0; i < typ.NumEmbeddeds(); i++ { + w.typ(typ.EmbeddedType(i)) + } +} + +func (w *writer) signature(sig *types2.Signature) { + w.sync(syncSignature) + w.params(sig.Params()) + w.params(sig.Results()) + w.bool(sig.Variadic()) +} + +func (w *writer) params(typ *types2.Tuple) { + w.sync(syncParams) + w.len(typ.Len()) + for i := 0; i < typ.Len(); i++ { + w.param(typ.At(i)) + } +} + +func (w *writer) param(param *types2.Var) { + w.sync(syncParam) + w.pos(param) + w.localIdent(param) + w.typ(param.Type()) +} + +// @@@ Objects + +func (w *writer) obj(obj types2.Object, explicits []types2.Type) { + w.sync(syncObject) + + var implicitIdx map[*types2.TypeParam]int + if isDefinedType(obj) && !isGlobal(obj) { + implicitIdx = w.implicitIdx + } + w.reloc(relocObj, w.p.objIdx(obj, implicitIdx)) + + w.len(len(explicits)) + for _, explicit := range explicits { + w.typ(explicit) + } +} + +func (pw *pkgWriter) objIdx(obj types2.Object, implicitIdx map[*types2.TypeParam]int) int { + if idx, ok := pw.globalsIdx[obj]; ok { + return idx + } + + w := pw.newWriter(relocObj, syncObject1) + w.ext = pw.newWriter(relocObjExt, syncObject1) + assert(w.ext.idx == w.idx) + + pw.globalsIdx[obj] = w.idx + + w.implicitIdx = implicitIdx + w.ext.implicitIdx = implicitIdx + + w.doObj(obj) + + w.flush() + w.ext.flush() + + return w.idx +} + +func (w *writer) doObj(obj types2.Object) { + // Ident goes first so importer can avoid unnecessary work if + // they've already resolved this object. + w.qualifiedIdent(obj) + + tparams := objTypeParams(obj) + w.setTypeParams(tparams) + w.typeParamBounds(tparams) + + if obj.Pkg() != w.p.curpkg { + w.code(objStub) + return + } + + switch obj := obj.(type) { + default: + w.p.unexpected("object", obj) + + case *types2.Const: + w.code(objConst) + w.pos(obj) + w.value(obj.Type(), obj.Val()) + + case *types2.Func: + decl, ok := w.p.funDecls[obj] + assert(ok) + sig := obj.Type().(*types2.Signature) + + // Rewrite blank methods into blank functions. + // They aren't included in the receiver type's method set, + // and we still want to write them out to be compiled + // for regression tests. + // TODO(mdempsky): Change regress tests to avoid relying + // on blank functions/methods, so we can just ignore them + // altogether. + if recv := sig.Recv(); recv != nil { + assert(obj.Name() == "_") + assert(sig.TParams() == nil) + + params := make([]*types2.Var, 1+sig.Params().Len()) + params[0] = recv + for i := 0; i < sig.Params().Len(); i++ { + params[1+i] = sig.Params().At(i) + } + sig = types2.NewSignature(nil, types2.NewTuple(params...), sig.Results(), sig.Variadic()) + } + + w.code(objFunc) + w.pos(obj) + w.typeParamNames(sig.TParams()) + w.signature(sig) + w.pos(decl) + w.ext.funcExt(obj) + + case *types2.TypeName: + decl, ok := w.p.typDecls[obj] + assert(ok) + + if obj.IsAlias() { + w.code(objAlias) + w.pos(obj) + w.typ(obj.Type()) + break + } + + named := obj.Type().(*types2.Named) + assert(named.TArgs() == nil) + + w.code(objType) + w.pos(obj) + w.typeParamNames(named.TParams()) + w.ext.typeExt(obj) + w.typExpr(decl.Type) + + w.len(named.NumMethods()) + for i := 0; i < named.NumMethods(); i++ { + w.method(named.Method(i)) + } + + case *types2.Var: + w.code(objVar) + w.pos(obj) + w.typ(obj.Type()) + w.ext.varExt(obj) + } +} + +// typExpr writes the type represented by the given expression. +func (w *writer) typExpr(expr syntax.Expr) { + tv, ok := w.p.info.Types[expr] + assert(ok) + assert(tv.IsType()) + w.typ(tv.Type) +} + +func (w *writer) value(typ types2.Type, val constant.Value) { + w.sync(syncValue) + w.typ(typ) + w.rawValue(val) +} + +func (w *writer) setTypeParams(tparams []*types2.TypeName) { + if len(tparams) == 0 { + return + } + + explicitIdx := make(map[*types2.TypeParam]int) + for _, tparam := range tparams { + explicitIdx[tparam.Type().(*types2.TypeParam)] = len(explicitIdx) + } + + w.explicitIdx = explicitIdx + w.ext.explicitIdx = explicitIdx +} + +func (w *writer) typeParamBounds(tparams []*types2.TypeName) { + w.sync(syncTypeParamBounds) + + // TODO(mdempsky): Remove. It's useful for debugging at the moment, + // but it doesn't belong here. + w.len(len(w.implicitIdx)) + w.len(len(w.explicitIdx)) + assert(len(w.explicitIdx) == len(tparams)) + + for _, tparam := range tparams { + w.typ(tparam.Type().(*types2.TypeParam).Bound()) + } +} + +func (w *writer) typeParamNames(tparams []*types2.TypeName) { + w.sync(syncTypeParamNames) + + for _, tparam := range tparams { + w.pos(tparam) + w.localIdent(tparam) + } +} + +func (w *writer) method(meth *types2.Func) { + decl, ok := w.p.funDecls[meth] + assert(ok) + sig := meth.Type().(*types2.Signature) + + assert(len(w.explicitIdx) == len(sig.RParams())) + w.setTypeParams(sig.RParams()) + + w.sync(syncMethod) + w.pos(meth) + w.selector(meth) + w.typeParamNames(sig.RParams()) + w.param(sig.Recv()) + w.signature(sig) + + w.pos(decl) // XXX: Hack to workaround linker limitations. + w.ext.funcExt(meth) +} + +// qualifiedIdent writes out the name of an object declared at package +// scope. (For now, it's also used to refer to local defined types.) +func (w *writer) qualifiedIdent(obj types2.Object) { + w.sync(syncSym) + + name := obj.Name() + if isDefinedType(obj) && !isGlobal(obj) { + // TODO(mdempsky): Find a better solution, this is terrible. + decl, ok := w.p.typDecls[obj.(*types2.TypeName)] + assert(ok) + name = fmt.Sprintf("%s·%v", name, decl.gen) + } + + w.pkg(obj.Pkg()) + w.string(name) +} + +// TODO(mdempsky): We should be able to omit pkg from both localIdent +// and selector, because they should always be known from context. +// However, past frustrations with this optimization in iexport make +// me a little nervous to try it again. + +// localIdent writes the name of a locally declared object (i.e., +// objects that can only be accessed by name, within the context of a +// particular function). +func (w *writer) localIdent(obj types2.Object) { + assert(!isGlobal(obj)) + w.sync(syncLocalIdent) + w.pkg(obj.Pkg()) + w.string(obj.Name()) +} + +// selector writes the name of a field or method (i.e., objects that +// can only be accessed using selector expressions). +func (w *writer) selector(obj types2.Object) { + w.sync(syncSelector) + w.pkg(obj.Pkg()) + w.string(obj.Name()) +} + +// @@@ Compiler extensions + +func (w *writer) funcExt(obj *types2.Func) { + decl, ok := w.p.funDecls[obj] + assert(ok) + + // TODO(mdempsky): Extend these pragma validation flags to account + // for generics. E.g., linkname probably doesn't make sense at + // least. + + pragma := asPragmaFlag(decl.Pragma) + if pragma&ir.Systemstack != 0 && pragma&ir.Nosplit != 0 { + w.p.errorf(decl, "go:nosplit and go:systemstack cannot be combined") + } + + if decl.Body != nil { + if pragma&ir.Noescape != 0 { + w.p.errorf(decl, "can only use //go:noescape with external func implementations") + } + } else { + if base.Flag.Complete || decl.Name.Value == "init" { + // Linknamed functions are allowed to have no body. Hopefully + // the linkname target has a body. See issue 23311. + if _, ok := w.p.linknames[obj]; !ok { + w.p.errorf(decl, "missing function body") + } + } + } + + w.sync(syncFuncExt) + w.pragmaFlag(pragma) + w.linkname(obj) + w.bool(false) // stub extension + w.addBody(obj.Type().(*types2.Signature), decl.Body, make(map[types2.Object]int)) + w.sync(syncEOF) +} + +func (w *writer) typeExt(obj *types2.TypeName) { + decl, ok := w.p.typDecls[obj] + assert(ok) + + w.sync(syncTypeExt) + + w.pragmaFlag(asPragmaFlag(decl.Pragma)) + + // No LSym.SymIdx info yet. + w.int64(-1) + w.int64(-1) +} + +func (w *writer) varExt(obj *types2.Var) { + w.sync(syncVarExt) + w.linkname(obj) +} + +func (w *writer) linkname(obj types2.Object) { + w.sync(syncLinkname) + w.int64(-1) + w.string(w.p.linknames[obj]) +} + +func (w *writer) pragmaFlag(p ir.PragmaFlag) { + w.sync(syncPragma) + w.int(int(p)) +} + +// @@@ Function bodies + +func (w *writer) addBody(sig *types2.Signature, block *syntax.BlockStmt, localsIdx map[types2.Object]int) { + // TODO(mdempsky): Theoretically, I think at this point we want to + // extend the implicit type parameters list with any new explicit + // type parameters. + // + // However, I believe that's moot: declared functions and methods + // have explicit type parameters, but are always declared at package + // scope (which has no implicit type parameters); and function + // literals can appear within a type-parameterized function (i.e., + // implicit type parameters), but cannot have explicit type + // parameters of their own. + // + // So I think it's safe to just use whichever is non-empty. + implicitIdx := w.implicitIdx + if len(implicitIdx) == 0 { + implicitIdx = w.explicitIdx + } else { + assert(len(w.explicitIdx) == 0) + } + + w.sync(syncAddBody) + w.reloc(relocBody, w.p.bodyIdx(w.p.curpkg, sig, block, implicitIdx, localsIdx)) +} + +func (pw *pkgWriter) bodyIdx(pkg *types2.Package, sig *types2.Signature, block *syntax.BlockStmt, implicitIdx map[*types2.TypeParam]int, localsIdx map[types2.Object]int) int { + w := pw.newWriter(relocBody, syncFuncBody) + w.implicitIdx = implicitIdx + w.localsIdx = localsIdx + + w.funcargs(sig) + if w.bool(block != nil) { + w.stmts(block.List) + w.pos(block.Rbrace) + } + + return w.flush() +} + +func (w *writer) funcargs(sig *types2.Signature) { + do := func(params *types2.Tuple, result bool) { + for i := 0; i < params.Len(); i++ { + w.funcarg(params.At(i), result) + } + } + + if recv := sig.Recv(); recv != nil { + w.funcarg(recv, false) + } + do(sig.Params(), false) + do(sig.Results(), true) +} + +func (w *writer) funcarg(param *types2.Var, result bool) { + if param.Name() != "" || result { + w.addLocal(param) + } +} + +func (w *writer) addLocal(obj types2.Object) { + w.sync(syncAddLocal) + idx := len(w.localsIdx) + if debug { + w.int(idx) + } + w.localsIdx[obj] = idx +} + +func (w *writer) useLocal(obj types2.Object) { + w.sync(syncUseObjLocal) + idx, ok := w.localsIdx[obj] + assert(ok) + w.len(idx) +} + +func (w *writer) openScope(pos syntax.Pos) { + w.sync(syncOpenScope) + w.pos(pos) +} + +func (w *writer) closeScope(pos syntax.Pos) { + w.sync(syncCloseScope) + w.pos(pos) + w.closeAnotherScope() +} + +func (w *writer) closeAnotherScope() { + w.sync(syncCloseAnotherScope) +} + +// @@@ Statements + +func (w *writer) stmt(stmt syntax.Stmt) { + var stmts []syntax.Stmt + if stmt != nil { + stmts = []syntax.Stmt{stmt} + } + w.stmts(stmts) +} + +func (w *writer) stmts(stmts []syntax.Stmt) { + w.sync(syncStmts) + for _, stmt := range stmts { + w.stmt1(stmt) + } + w.code(stmtEnd) + w.sync(syncStmtsEnd) +} + +func (w *writer) stmt1(stmt syntax.Stmt) { + switch stmt := stmt.(type) { + default: + w.p.unexpected("statement", stmt) + + case nil, *syntax.EmptyStmt: + return + + case *syntax.AssignStmt: + switch { + case stmt.Rhs == nil: + w.code(stmtIncDec) + w.op(binOps[stmt.Op]) + w.expr(stmt.Lhs) + w.pos(stmt) + + case stmt.Op != 0 && stmt.Op != syntax.Def: + w.code(stmtAssignOp) + w.op(binOps[stmt.Op]) + w.expr(stmt.Lhs) + w.pos(stmt) + w.expr(stmt.Rhs) + + default: + w.code(stmtAssign) + w.pos(stmt) + w.assignList(stmt.Lhs) + w.exprList(stmt.Rhs) + } + + case *syntax.BlockStmt: + w.code(stmtBlock) + w.blockStmt(stmt) + + case *syntax.BranchStmt: + w.code(stmtBranch) + w.pos(stmt) + w.op(branchOps[stmt.Tok]) + w.optLabel(stmt.Label) + + case *syntax.CallStmt: + w.code(stmtCall) + w.pos(stmt) + w.op(callOps[stmt.Tok]) + w.expr(stmt.Call) + + case *syntax.DeclStmt: + for _, decl := range stmt.DeclList { + w.declStmt(decl) + } + + case *syntax.ExprStmt: + w.code(stmtExpr) + w.expr(stmt.X) + + case *syntax.ForStmt: + w.code(stmtFor) + w.forStmt(stmt) + + case *syntax.IfStmt: + w.code(stmtIf) + w.ifStmt(stmt) + + case *syntax.LabeledStmt: + w.code(stmtLabel) + w.pos(stmt) + w.label(stmt.Label) + w.stmt1(stmt.Stmt) + + case *syntax.ReturnStmt: + w.code(stmtReturn) + w.pos(stmt) + w.exprList(stmt.Results) + + case *syntax.SelectStmt: + w.code(stmtSelect) + w.selectStmt(stmt) + + case *syntax.SendStmt: + w.code(stmtSend) + w.pos(stmt) + w.expr(stmt.Chan) + w.expr(stmt.Value) + + case *syntax.SwitchStmt: + w.code(stmtSwitch) + w.switchStmt(stmt) + } +} + +func (w *writer) assignList(expr syntax.Expr) { + exprs := unpackListExpr(expr) + w.len(len(exprs)) + + for _, expr := range exprs { + if name, ok := expr.(*syntax.Name); ok && name.Value != "_" { + if obj, ok := w.p.info.Defs[name]; ok { + w.bool(true) + w.pos(obj) + w.localIdent(obj) + w.typ(obj.Type()) + + // TODO(mdempsky): Minimize locals index size by deferring + // this until the variables actually come into scope. + w.addLocal(obj) + continue + } + } + + w.bool(false) + w.expr(expr) + } +} + +func (w *writer) declStmt(decl syntax.Decl) { + switch decl := decl.(type) { + default: + w.p.unexpected("declaration", decl) + + case *syntax.ConstDecl: + + case *syntax.TypeDecl: + // Quirk: The legacy inliner doesn't support inlining functions + // with type declarations. Unified IR doesn't have any need to + // write out type declarations explicitly (they're always looked + // up via global index tables instead), so we just write out a + // marker so the reader knows to synthesize a fake declaration to + // prevent inlining. + if quirksMode() { + w.code(stmtTypeDeclHack) + } + + case *syntax.VarDecl: + values := unpackListExpr(decl.Values) + + // Quirk: When N variables are declared with N initialization + // values, we need to decompose that into N interleaved + // declarations+initializations, because it leads to different + // (albeit semantically equivalent) code generation. + if quirksMode() && len(decl.NameList) == len(values) { + for i, name := range decl.NameList { + w.code(stmtAssign) + w.pos(decl) + w.assignList(name) + w.exprList(values[i]) + } + break + } + + w.code(stmtAssign) + w.pos(decl) + w.assignList(namesAsExpr(decl.NameList)) + w.exprList(decl.Values) + } +} + +func (w *writer) blockStmt(stmt *syntax.BlockStmt) { + w.sync(syncBlockStmt) + w.openScope(stmt.Pos()) + w.stmts(stmt.List) + w.closeScope(stmt.Rbrace) +} + +func (w *writer) forStmt(stmt *syntax.ForStmt) { + w.sync(syncForStmt) + w.openScope(stmt.Pos()) + + if rang, ok := stmt.Init.(*syntax.RangeClause); w.bool(ok) { + w.pos(rang) + w.assignList(rang.Lhs) + w.expr(rang.X) + } else { + w.pos(stmt) + w.stmt(stmt.Init) + w.expr(stmt.Cond) + w.stmt(stmt.Post) + } + + w.blockStmt(stmt.Body) + w.closeAnotherScope() +} + +func (w *writer) ifStmt(stmt *syntax.IfStmt) { + w.sync(syncIfStmt) + w.openScope(stmt.Pos()) + w.pos(stmt) + w.stmt(stmt.Init) + w.expr(stmt.Cond) + w.blockStmt(stmt.Then) + w.stmt(stmt.Else) + w.closeAnotherScope() +} + +func (w *writer) selectStmt(stmt *syntax.SelectStmt) { + w.sync(syncSelectStmt) + + w.pos(stmt) + w.len(len(stmt.Body)) + for i, clause := range stmt.Body { + if i > 0 { + w.closeScope(clause.Pos()) + } + w.openScope(clause.Pos()) + + w.pos(clause) + w.stmt(clause.Comm) + w.stmts(clause.Body) + } + if len(stmt.Body) > 0 { + w.closeScope(stmt.Rbrace) + } +} + +func (w *writer) switchStmt(stmt *syntax.SwitchStmt) { + w.sync(syncSwitchStmt) + + w.openScope(stmt.Pos()) + w.pos(stmt) + w.stmt(stmt.Init) + w.expr(stmt.Tag) + + w.len(len(stmt.Body)) + for i, clause := range stmt.Body { + if i > 0 { + w.closeScope(clause.Pos()) + } + w.openScope(clause.Pos()) + + w.pos(clause) + w.exprList(clause.Cases) + + if obj, ok := w.p.info.Implicits[clause]; ok { + // TODO(mdempsky): These pos details are quirkish, but also + // necessary so the variable's position is correct for DWARF + // scope assignment later. It would probably be better for us to + // instead just set the variable's DWARF scoping info earlier so + // we can give it the correct position information. + pos := clause.Pos() + if typs := unpackListExpr(clause.Cases); len(typs) != 0 { + pos = typeExprEndPos(typs[len(typs)-1]) + } + w.pos(pos) + + obj := obj.(*types2.Var) + w.typ(obj.Type()) + w.addLocal(obj) + } + + w.stmts(clause.Body) + } + if len(stmt.Body) > 0 { + w.closeScope(stmt.Rbrace) + } + + w.closeScope(stmt.Rbrace) +} + +func (w *writer) label(label *syntax.Name) { + w.sync(syncLabel) + + // TODO(mdempsky): Replace label strings with dense indices. + w.string(label.Value) +} + +func (w *writer) optLabel(label *syntax.Name) { + w.sync(syncOptLabel) + if w.bool(label != nil) { + w.label(label) + } +} + +// @@@ Expressions + +func (w *writer) expr(expr syntax.Expr) { + expr = unparen(expr) // skip parens; unneeded after typecheck + + obj, targs := lookupObj(w.p.info, expr) + + if tv, ok := w.p.info.Types[expr]; ok { + if tv.IsType() { + w.code(exprType) + w.typ(tv.Type) + return + } + + if tv.Value != nil { + pos := expr.Pos() + if quirksMode() { + if obj != nil { + // Quirk: IR (and thus iexport) doesn't track position + // information for uses of declared objects. + pos = syntax.Pos{} + } else if tv.Value.Kind() == constant.String { + // Quirk: noder.sum picks a particular position for certain + // string concatenations. + pos = sumPos(expr) + } + } + + w.code(exprConst) + w.pos(pos) + w.value(tv.Type, tv.Value) + + // TODO(mdempsky): These details are only important for backend + // diagnostics. Explore writing them out separately. + w.op(constExprOp(expr)) + w.string(syntax.String(expr)) + return + } + } + + if obj != nil { + if _, ok := w.localsIdx[obj]; ok { + assert(len(targs) == 0) + w.code(exprLocal) + w.useLocal(obj) + return + } + + w.code(exprName) + w.obj(obj, targs) + return + } + + switch expr := expr.(type) { + default: + w.p.unexpected("expression", expr) + + case nil: // absent slice index, for condition, or switch tag + w.code(exprNone) + + case *syntax.Name: + assert(expr.Value == "_") + w.code(exprBlank) + + case *syntax.CompositeLit: + w.code(exprCompLit) + w.compLit(expr) + + case *syntax.FuncLit: + w.code(exprFuncLit) + w.funcLit(expr) + + case *syntax.SelectorExpr: + sel, ok := w.p.info.Selections[expr] + assert(ok) + + w.code(exprSelector) + w.expr(expr.X) + w.pos(expr) + w.selector(sel.Obj()) + + case *syntax.IndexExpr: + tv, ok := w.p.info.Types[expr.Index] + assert(ok && tv.IsValue()) + + w.code(exprIndex) + w.expr(expr.X) + w.pos(expr) + w.expr(expr.Index) + + case *syntax.SliceExpr: + w.code(exprSlice) + w.expr(expr.X) + w.pos(expr) + for _, n := range &expr.Index { + w.expr(n) + } + + case *syntax.AssertExpr: + w.code(exprAssert) + w.expr(expr.X) + w.pos(expr) + w.expr(expr.Type) + + case *syntax.Operation: + if expr.Y == nil { + w.code(exprUnaryOp) + w.op(unOps[expr.Op]) + w.pos(expr) + w.expr(expr.X) + break + } + + w.code(exprBinaryOp) + w.op(binOps[expr.Op]) + w.expr(expr.X) + w.pos(expr) + w.expr(expr.Y) + + case *syntax.CallExpr: + w.code(exprCall) + + if inf, ok := w.p.info.Inferred[expr]; ok { + obj, _ := lookupObj(w.p.info, expr.Fun) + assert(obj != nil) + + // As if w.expr(expr.Fun), but using inf.TArgs instead. + w.code(exprName) + w.obj(obj, inf.TArgs) + } else { + w.expr(expr.Fun) + } + + w.pos(expr) + w.exprs(expr.ArgList) + w.bool(expr.HasDots) + + case *syntax.TypeSwitchGuard: + w.code(exprTypeSwitchGuard) + w.pos(expr) + if tag := expr.Lhs; w.bool(tag != nil) { + w.pos(tag) + w.string(tag.Value) + } + w.expr(expr.X) + } +} + +func (w *writer) compLit(lit *syntax.CompositeLit) { + tv, ok := w.p.info.Types[lit] + assert(ok) + + w.sync(syncCompLit) + w.pos(lit) + w.typ(tv.Type) + + typ := tv.Type + if ptr, ok := typ.Underlying().(*types2.Pointer); ok { + typ = ptr.Elem() + } + str, isStruct := typ.Underlying().(*types2.Struct) + + w.len(len(lit.ElemList)) + for i, elem := range lit.ElemList { + if isStruct { + if kv, ok := elem.(*syntax.KeyValueExpr); ok { + // use position of expr.Key rather than of elem (which has position of ':') + w.pos(kv.Key) + w.len(fieldIndex(w.p.info, str, kv.Key.(*syntax.Name))) + elem = kv.Value + } else { + w.pos(elem) + w.len(i) + } + } else { + if kv, ok := elem.(*syntax.KeyValueExpr); w.bool(ok) { + // use position of expr.Key rather than of elem (which has position of ':') + w.pos(kv.Key) + w.expr(kv.Key) + elem = kv.Value + } + } + w.pos(elem) + w.expr(elem) + } +} + +func (w *writer) funcLit(expr *syntax.FuncLit) { + tv, ok := w.p.info.Types[expr] + assert(ok) + sig := tv.Type.(*types2.Signature) + + w.sync(syncFuncLit) + w.pos(expr) + w.pos(expr.Type) // for QuirksMode + w.signature(sig) + + closureVars, localsIdx := w.captureVars(expr) + w.len(len(closureVars)) + for _, closureVar := range closureVars { + w.pos(closureVar.pos) + w.useLocal(closureVar.obj) + } + + w.addBody(sig, expr.Body, localsIdx) +} + +type posObj struct { + pos syntax.Pos + obj types2.Object +} + +// captureVars returns the free variables used by the given function +// literal. +func (w *writer) captureVars(expr *syntax.FuncLit) (closureVars []posObj, localsIdx map[types2.Object]int) { + scope, ok := w.p.info.Scopes[expr.Type] + assert(ok) + + localsIdx = make(map[types2.Object]int) + + // TODO(mdempsky): This code needs to be cleaned up (e.g., to avoid + // traversing nested function literals multiple times). This will be + // easier after we drop quirks mode. + + var rbracePos syntax.Pos + + var visitor func(n syntax.Node) bool + visitor = func(n syntax.Node) bool { + + // Constant expressions don't count towards capturing. + if n, ok := n.(syntax.Expr); ok { + if tv, ok := w.p.info.Types[n]; ok && tv.Value != nil { + return true + } + } + + switch n := n.(type) { + case *syntax.Name: + if obj, ok := w.p.info.Uses[n].(*types2.Var); ok && !obj.IsField() && obj.Pkg() == w.p.curpkg && obj.Parent() != obj.Pkg().Scope() { + // Found a local variable. See if it chains up to scope. + parent := obj.Parent() + for { + if parent == scope { + break + } + if parent == obj.Pkg().Scope() { + if _, present := localsIdx[obj]; !present { + pos := rbracePos + if pos == (syntax.Pos{}) { + pos = n.Pos() + } + + idx := len(closureVars) + closureVars = append(closureVars, posObj{pos, obj}) + localsIdx[obj] = idx + } + break + } + parent = parent.Parent() + } + } + + case *syntax.FuncLit: + // Quirk: typecheck uses the rbrace position position of the + // function literal as the position of the intermediary capture. + if quirksMode() && rbracePos == (syntax.Pos{}) { + rbracePos = n.Body.Rbrace + syntax.Walk(n.Body, visitor) + rbracePos = syntax.Pos{} + return true + } + + case *syntax.AssignStmt: + // Quirk: typecheck visits (and thus captures) the RHS of + // assignment statements before the LHS. + if quirksMode() && (n.Op == 0 || n.Op == syntax.Def) { + syntax.Walk(n.Rhs, visitor) + syntax.Walk(n.Lhs, visitor) + return true + } + case *syntax.RangeClause: + // Quirk: Similarly, it visits the expression to be iterated + // over before the iteration variables. + if quirksMode() { + syntax.Walk(n.X, visitor) + if n.Lhs != nil { + syntax.Walk(n.Lhs, visitor) + } + return true + } + } + + return false + } + syntax.Walk(expr.Body, visitor) + + return +} + +func (w *writer) exprList(expr syntax.Expr) { + w.sync(syncExprList) + w.exprs(unpackListExpr(expr)) +} + +func (w *writer) exprs(exprs []syntax.Expr) { + if len(exprs) == 0 { + assert(exprs == nil) + } + + w.sync(syncExprs) + w.len(len(exprs)) + for _, expr := range exprs { + w.expr(expr) + } +} + +func (w *writer) op(op ir.Op) { + // TODO(mdempsky): Remove in favor of explicit codes? Would make + // export data more stable against internal refactorings, but low + // priority at the moment. + assert(op != 0) + w.sync(syncOp) + w.len(int(op)) +} + +// @@@ Package initialization + +// Caution: This code is still clumsy, because toolstash -cmp is +// particularly sensitive to it. + +type typeDeclGen struct { + *syntax.TypeDecl + gen int +} + +func (pw *pkgWriter) collectDecls(noders []*noder) { + var typegen int + + for _, p := range noders { + var importedEmbed, importedUnsafe bool + + syntax.Walk(p.file, func(n syntax.Node) bool { + switch n := n.(type) { + case *syntax.File: + pw.checkPragmas(n.Pragma, ir.GoBuildPragma, false) + + case *syntax.ImportDecl: + pw.checkPragmas(n.Pragma, 0, false) + + switch pkgNameOf(pw.info, n).Imported().Path() { + case "embed": + importedEmbed = true + case "unsafe": + importedUnsafe = true + } + + case *syntax.ConstDecl: + pw.checkPragmas(n.Pragma, 0, false) + + case *syntax.FuncDecl: + pw.checkPragmas(n.Pragma, funcPragmas, false) + + obj := pw.info.Defs[n.Name].(*types2.Func) + pw.funDecls[obj] = n + + case *syntax.TypeDecl: + obj := pw.info.Defs[n.Name].(*types2.TypeName) + d := typeDeclGen{TypeDecl: n} + + if n.Alias { + pw.checkPragmas(n.Pragma, 0, false) + } else { + pw.checkPragmas(n.Pragma, typePragmas, false) + + // Assign a unique ID to function-scoped defined types. + if !isGlobal(obj) { + typegen++ + d.gen = typegen + } + } + + pw.typDecls[obj] = d + + case *syntax.VarDecl: + pw.checkPragmas(n.Pragma, 0, true) + + if p, ok := n.Pragma.(*pragmas); ok && len(p.Embeds) > 0 { + obj := pw.info.Defs[n.NameList[0]].(*types2.Var) + // TODO(mdempsky): isGlobal(obj) gives false positive errors + // for //go:embed directives on package-scope blank + // variables. + if err := checkEmbed(n, importedEmbed, !isGlobal(obj)); err != nil { + pw.errorf(p.Embeds[0].Pos, "%s", err) + } + } + + // Workaround for #46208. For variable declarations that + // declare multiple variables and have an explicit type + // expression, the type expression is evaluated multiple + // times. This affects toolstash -cmp, because iexport is + // sensitive to *types.Type pointer identity. + if quirksMode() && n.Type != nil { + tv, ok := pw.info.Types[n.Type] + assert(ok) + assert(tv.IsType()) + for _, name := range n.NameList { + obj := pw.info.Defs[name].(*types2.Var) + pw.dups.add(obj.Type(), tv.Type) + } + } + } + return false + }) + + pw.cgoPragmas = append(pw.cgoPragmas, p.pragcgobuf...) + + for _, l := range p.linknames { + if !importedUnsafe { + pw.errorf(l.pos, "//go:linkname only allowed in Go files that import \"unsafe\"") + continue + } + + switch obj := pw.curpkg.Scope().Lookup(l.local).(type) { + case *types2.Func, *types2.Var: + if _, ok := pw.linknames[obj]; !ok { + pw.linknames[obj] = l.remote + } else { + pw.errorf(l.pos, "duplicate //go:linkname for %s", l.local) + } + + default: + // TODO(mdempsky): Enable after #42938 is fixed. + if false { + pw.errorf(l.pos, "//go:linkname must refer to declared function or variable") + } + } + } + } +} + +func (pw *pkgWriter) checkPragmas(p syntax.Pragma, allowed ir.PragmaFlag, embedOK bool) { + if p == nil { + return + } + pragma := p.(*pragmas) + + for _, pos := range pragma.Pos { + if pos.Flag&^allowed != 0 { + pw.errorf(pos.Pos, "misplaced compiler directive") + } + } + + if !embedOK { + for _, e := range pragma.Embeds { + pw.errorf(e.Pos, "misplaced go:embed directive") + } + } +} + +func (w *writer) pkgInit(noders []*noder) { + if quirksMode() { + posBases := posBasesOf(noders) + w.len(len(posBases)) + for _, posBase := range posBases { + w.posBase(posBase) + } + + objs := importedObjsOf(w.p.curpkg, w.p.info, noders) + w.len(len(objs)) + for _, obj := range objs { + w.qualifiedIdent(obj) + } + } + + w.len(len(w.p.cgoPragmas)) + for _, cgoPragma := range w.p.cgoPragmas { + w.strings(cgoPragma) + } + + w.sync(syncDecls) + for _, p := range noders { + for _, decl := range p.file.DeclList { + w.pkgDecl(decl) + } + } + w.code(declEnd) + + w.sync(syncEOF) +} + +func (w *writer) pkgDecl(decl syntax.Decl) { + switch decl := decl.(type) { + default: + w.p.unexpected("declaration", decl) + + case *syntax.ImportDecl: + + case *syntax.ConstDecl: + w.code(declOther) + w.pkgObjs(decl.NameList...) + + case *syntax.FuncDecl: + obj := w.p.info.Defs[decl.Name].(*types2.Func) + sig := obj.Type().(*types2.Signature) + + if sig.RParams() != nil || sig.TParams() != nil { + break // skip generic functions + } + + if recv := sig.Recv(); recv != nil && obj.Name() != "_" { + w.code(declMethod) + w.typ(recvBase(recv)) + w.selector(obj) + break + } + + w.code(declFunc) + w.pkgObjs(decl.Name) + + case *syntax.TypeDecl: + if len(decl.TParamList) != 0 { + break // skip generic type decls + } + + name := w.p.info.Defs[decl.Name].(*types2.TypeName) + + // Skip type declarations for interfaces that are only usable as + // type parameter bounds. + if iface, ok := name.Type().Underlying().(*types2.Interface); ok && iface.IsConstraint() { + break + } + + // Skip aliases to uninstantiated generic types. + // TODO(mdempsky): Revisit after #46477 is resolved. + if name.IsAlias() { + named, ok := name.Type().(*types2.Named) + if ok && len(named.TParams()) != 0 && len(named.TArgs()) == 0 { + break + } + } + + w.code(declOther) + w.pkgObjs(decl.Name) + + case *syntax.VarDecl: + w.code(declVar) + w.pos(decl) + w.pkgObjs(decl.NameList...) + w.exprList(decl.Values) + + var embeds []pragmaEmbed + if p, ok := decl.Pragma.(*pragmas); ok { + embeds = p.Embeds + } + w.len(len(embeds)) + for _, embed := range embeds { + w.pos(embed.Pos) + w.strings(embed.Patterns) + } + } +} + +func (w *writer) pkgObjs(names ...*syntax.Name) { + w.sync(syncDeclNames) + w.len(len(names)) + + for _, name := range names { + obj, ok := w.p.info.Defs[name] + assert(ok) + + w.sync(syncDeclName) + w.obj(obj, nil) + } +} + +// @@@ Helpers + +// isDefinedType reports whether obj is a defined type. +func isDefinedType(obj types2.Object) bool { + if obj, ok := obj.(*types2.TypeName); ok { + return !obj.IsAlias() + } + return false +} + +// isGlobal reports whether obj was declared at package scope. +// +// Caveat: blank objects are not declared. +func isGlobal(obj types2.Object) bool { + return obj.Parent() == obj.Pkg().Scope() +} + +// lookupObj returns the object that expr refers to, if any. If expr +// is an explicit instantiation of a generic object, then the type +// arguments are returned as well. +func lookupObj(info *types2.Info, expr syntax.Expr) (obj types2.Object, targs []types2.Type) { + if index, ok := expr.(*syntax.IndexExpr); ok { + if inf, ok := info.Inferred[index]; ok { + targs = inf.TArgs + } else { + args := unpackListExpr(index.Index) + + if len(args) == 1 { + tv, ok := info.Types[args[0]] + assert(ok) + if tv.IsValue() { + return // normal index expression + } + } + + targs = make([]types2.Type, len(args)) + for i, arg := range args { + tv, ok := info.Types[arg] + assert(ok) + assert(tv.IsType()) + targs[i] = tv.Type + } + } + + expr = index.X + } + + // Strip package qualifier, if present. + if sel, ok := expr.(*syntax.SelectorExpr); ok { + if !isPkgQual(info, sel) { + return // normal selector expression + } + expr = sel.Sel + } + + if name, ok := expr.(*syntax.Name); ok { + obj, _ = info.Uses[name] + } + return +} + +// isPkgQual reports whether the given selector expression is a +// package-qualified identifier. +func isPkgQual(info *types2.Info, sel *syntax.SelectorExpr) bool { + if name, ok := sel.X.(*syntax.Name); ok { + _, isPkgName := info.Uses[name].(*types2.PkgName) + return isPkgName + } + return false +} + +// recvBase returns the base type for the given receiver parameter. +func recvBase(recv *types2.Var) *types2.Named { + typ := recv.Type() + if ptr, ok := typ.(*types2.Pointer); ok { + typ = ptr.Elem() + } + return typ.(*types2.Named) +} + +// namesAsExpr returns a list of names as a syntax.Expr. +func namesAsExpr(names []*syntax.Name) syntax.Expr { + if len(names) == 1 { + return names[0] + } + + exprs := make([]syntax.Expr, len(names)) + for i, name := range names { + exprs[i] = name + } + return &syntax.ListExpr{ElemList: exprs} +} + +// fieldIndex returns the index of the struct field named by key. +func fieldIndex(info *types2.Info, str *types2.Struct, key *syntax.Name) int { + field := info.Uses[key].(*types2.Var) + + for i := 0; i < str.NumFields(); i++ { + if str.Field(i) == field { + return i + } + } + + panic(fmt.Sprintf("%s: %v is not a field of %v", key.Pos(), field, str)) +} + +// objTypeParams returns the type parameters on the given object. +func objTypeParams(obj types2.Object) []*types2.TypeName { + switch obj := obj.(type) { + case *types2.Func: + return obj.Type().(*types2.Signature).TParams() + case *types2.TypeName: + if !obj.IsAlias() { + return obj.Type().(*types2.Named).TParams() + } + } + return nil +} + +func asPragmaFlag(p syntax.Pragma) ir.PragmaFlag { + if p == nil { + return 0 + } + return p.(*pragmas).Flag +}