[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 <mdempsky@google.com>
Trust: Robert Griesemer <gri@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Robert Griesemer <gri@golang.org>

1 files changed, 245 insertions, 0 deletions

diff --git a/src/cmd/compile/internal/noder/encoder.go b/src/cmd/compile/internal/noder/encoder.go
new file mode 100644
index 0000000000..dc288dc29f
--- /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.
+}