cmd/compile: fold offset into address on Wasm

On Wasm, the offset was not folded into LoweredAddr, so it was
not rematerializeable. This led to the address-taken operation
in some cases generated too early, before the local variable
becoming live. The liveness code thinks the variable live when
the address is taken, then backs it up to live at function
entry, then complains about it, because nothing other than
arguments should be live on entry.

This CL folds the offset into the address operation, so it is
rematerializeable and so generated right before use, after the
variable actually becomes live.

It might be possible to relax the liveness code not to think a
variable live when its address being taken, but until the address
actually being used. But it would be quite complicated. As we're
late in Go 1.11 freeze, it would be better not to do it. Also,
I think the address operation is rematerializeable now on all
architectures, so this is probably less necessary.

This may also be a slight optimization, as the address+offset is
now rematerializeable, which can be generated on the Wasm stack,
without using any "registers" which are emulated by local
variables on Wasm. I don't know how to do benchmarks on Wasm. At
least, cmd/go binary size shrinks 9K.

Fixes #25966.

Change-Id: I01e5869515d6a3942fccdcb857f924a866876e57
Reviewed-on: https://go-review.googlesource.com/120599
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Richard Musiol <neelance@gmail.com>

src/cmd/compile/internal/ssa/gen/Wasm.rules

@@ -390,6 +390,7 @@
 (I64AddConst [0] x) -> x
 (I64Eqz (I64Eqz (I64Eqz x))) -> (I64Eqz x)
 
+// folding offset into load/store
 ((I64Load|I64Load32U|I64Load32S|I64Load16U|I64Load16S|I64Load8U|I64Load8S) [off] (I64AddConst [off2] ptr) mem)
 	&& isU32Bit(off+off2) ->
 	((I64Load|I64Load32U|I64Load32S|I64Load16U|I64Load16S|I64Load8U|I64Load8S) [off+off2] ptr mem)
@@ -397,3 +398,7 @@
 ((I64Store|I64Store32|I64Store16|I64Store8) [off] (I64AddConst [off2] ptr) val mem)
 	&& isU32Bit(off+off2) ->
 	((I64Store|I64Store32|I64Store16|I64Store8) [off+off2] ptr val mem)
+
+// folding offset into address
+(I64AddConst [off] (LoweredAddr {sym} [off2] base)) && isU32Bit(off+off2) ->
+	(LoweredAddr {sym} [off+off2] base)

src/cmd/compile/internal/ssa/gen/WasmOps.go

@@ -103,7 +103,7 @@ func init() {
 		{name: "LoweredClosureCall", argLength: 3, reg: regInfo{inputs: []regMask{gp, gp, 0}, clobbers: callerSave}, aux: "Int64", call: true}, // call function via closure. arg0=codeptr, arg1=closure, arg2=mem, auxint=argsize, returns mem
 		{name: "LoweredInterCall", argLength: 2, reg: regInfo{inputs: []regMask{gp}, clobbers: callerSave}, aux: "Int64", call: true},          // call fn by pointer. arg0=codeptr, arg1=mem, auxint=argsize, returns mem
 
-		{name: "LoweredAddr", argLength: 1, reg: gp11, aux: "SymOff", rematerializeable: true, symEffect: "Addr"}, // returns base+aux, arg0=base
+		{name: "LoweredAddr", argLength: 1, reg: gp11, aux: "SymOff", rematerializeable: true, symEffect: "Addr"}, // returns base+aux+auxint, arg0=base
 		{name: "LoweredMove", argLength: 3, reg: regInfo{inputs: []regMask{gp, gp}}, aux: "Int64"},                // large move. arg0=dst, arg1=src, arg2=mem, auxint=len/8, returns mem
 		{name: "LoweredZero", argLength: 2, reg: regInfo{inputs: []regMask{gp}}, aux: "Int64"},                    // large zeroing. arg0=start, arg1=mem, auxint=len/8, returns mem
 

src/cmd/compile/internal/ssa/rewriteWasm.go

@@ -5210,6 +5210,27 @@ func rewriteValueWasm_OpWasmI64AddConst_0(v *Value) bool {
 		v.AddArg(x)
 		return true
 	}
+	// match: (I64AddConst [off] (LoweredAddr {sym} [off2] base))
+	// cond: isU32Bit(off+off2)
+	// result: (LoweredAddr {sym} [off+off2] base)
+	for {
+		off := v.AuxInt
+		v_0 := v.Args[0]
+		if v_0.Op != OpWasmLoweredAddr {
+			break
+		}
+		off2 := v_0.AuxInt
+		sym := v_0.Aux
+		base := v_0.Args[0]
+		if !(isU32Bit(off + off2)) {
+			break
+		}
+		v.reset(OpWasmLoweredAddr)
+		v.AuxInt = off + off2
+		v.Aux = sym
+		v.AddArg(base)
+		return true
+	}
 	return false
 }
 func rewriteValueWasm_OpWasmI64And_0(v *Value) bool {

src/cmd/compile/internal/wasm/ssa.go

@@ -232,19 +232,13 @@ func ssaGenValueOnStack(s *gc.SSAGenState, v *ssa.Value) {
 
 	case ssa.OpWasmLoweredAddr:
 		p := s.Prog(wasm.AGet)
-		switch n := v.Aux.(type) {
+		p.From.Type = obj.TYPE_ADDR
+		switch v.Aux.(type) {
 		case *obj.LSym:
-			p.From = obj.Addr{Type: obj.TYPE_ADDR, Name: obj.NAME_EXTERN, Sym: n}
+			gc.AddAux(&p.From, v)
 		case *gc.Node:
-			p.From = obj.Addr{
-				Type:   obj.TYPE_ADDR,
-				Name:   obj.NAME_AUTO,
-				Reg:    v.Args[0].Reg(),
-				Offset: n.Xoffset,
-			}
-			if n.Class() == gc.PPARAM || n.Class() == gc.PPARAMOUT {
-				p.From.Name = obj.NAME_PARAM
-			}
+			p.From.Reg = v.Args[0].Reg()
+			gc.AddAux(&p.From, v)
 		default:
 			panic("wasm: bad LoweredAddr")
 		}

test/fixedbugs/issue25966.go

@@ -0,0 +1,24 @@
+// compile -N
+
+// Copyright 2018 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.
+
+// Issue 25966: liveness code complains autotmp live on
+// function entry.
+
+package p
+
+var F = []func(){
+	func() func() { return (func())(nil) }(),
+}
+
+var A = []int{}
+
+type ss struct {
+	string
+	float64
+	i int
+}
+
+var V = A[ss{}.i]