[dev.regabi] cmd/compile: reserve X15 as zero register on AMD64

In ABIInternal, reserve X15 as constant zero, and use it to zero
memory. (Maybe there can be more use of it?)

The register is zeroed when transition to ABIInternal from ABI0.

Caveat: using X15 generates longer instructions than using X0.
Maybe we want to use X0?

Change-Id: I12d5ee92a01fc0b59dad4e5ab023ac71bc2a8b7d
Reviewed-on: https://go-review.googlesource.com/c/go/+/288093
Trust: Cherry Zhang <cherryyz@google.com>
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: David Chase <drchase@google.com>

2 files changed, 38 insertions, 9 deletions

diff --git a/src/cmd/compile/internal/amd64/ggen.go b/src/cmd/compile/internal/amd64/ggen.go
index dacdb07a38..aefdb14a69 100644
--- a/src/cmd/compile/internal/amd64/ggen.go
+++ b/src/cmd/compile/internal/amd64/ggen.go
@@ -22,8 +22,8 @@ var isPlan9 = objabi.GOOS == "plan9"
 const (
 	dzBlocks    = 16 // number of MOV/ADD blocks
 	dzBlockLen  = 4  // number of clears per block
-	dzBlockSize = 19 // size of instructions in a single block
-	dzMovSize   = 4  // size of single MOV instruction w/ offset
+	dzBlockSize = 23 // size of instructions in a single block
+	dzMovSize   = 5  // size of single MOV instruction w/ offset
 	dzLeaqSize  = 4  // size of single LEAQ instruction
 	dzClearStep = 16 // number of bytes cleared by each MOV instruction
 
diff --git a/src/cmd/compile/internal/amd64/ssa.go b/src/cmd/compile/internal/amd64/ssa.go
index da355c49d1..d9c97183fd 100644
--- a/src/cmd/compile/internal/amd64/ssa.go
+++ b/src/cmd/compile/internal/amd64/ssa.go
@@ -813,6 +813,20 @@ func ssaGenValue(s *ssagen.State, v *ssa.Value) {
 		p.To.Type = obj.TYPE_MEM
 		p.To.Reg = v.Args[0].Reg()
 		ssagen.AddAux2(&p.To, v, sc.Off())
+	case ssa.OpAMD64MOVOstorezero:
+		if s.ABI != obj.ABIInternal {
+			v.Fatalf("MOVOstorezero can be only used in ABIInternal functions")
+		}
+		if !base.Flag.ABIWrap {
+			// zeroing X15 manually if wrappers are not used
+			opregreg(s, x86.AXORPS, x86.REG_X15, x86.REG_X15)
+		}
+		p := s.Prog(v.Op.Asm())
+		p.From.Type = obj.TYPE_REG
+		p.From.Reg = x86.REG_X15
+		p.To.Type = obj.TYPE_MEM
+		p.To.Reg = v.Args[0].Reg()
+		ssagen.AddAux(&p.To, v)
 	case ssa.OpAMD64MOVQstoreconstidx1, ssa.OpAMD64MOVQstoreconstidx8, ssa.OpAMD64MOVLstoreconstidx1, ssa.OpAMD64MOVLstoreconstidx4, ssa.OpAMD64MOVWstoreconstidx1, ssa.OpAMD64MOVWstoreconstidx2, ssa.OpAMD64MOVBstoreconstidx1,
 		ssa.OpAMD64ADDLconstmodifyidx1, ssa.OpAMD64ADDLconstmodifyidx4, ssa.OpAMD64ADDLconstmodifyidx8, ssa.OpAMD64ADDQconstmodifyidx1, ssa.OpAMD64ADDQconstmodifyidx8,
 		ssa.OpAMD64ANDLconstmodifyidx1, ssa.OpAMD64ANDLconstmodifyidx4, ssa.OpAMD64ANDLconstmodifyidx8, ssa.OpAMD64ANDQconstmodifyidx1, ssa.OpAMD64ANDQconstmodifyidx8,
@@ -900,6 +914,13 @@ func ssaGenValue(s *ssagen.State, v *ssa.Value) {
 			v.Fatalf("input[0] and output not in same register %s", v.LongString())
 		}
 	case ssa.OpAMD64DUFFZERO:
+		if s.ABI != obj.ABIInternal {
+			v.Fatalf("MOVOconst can be only used in ABIInternal functions")
+		}
+		if !base.Flag.ABIWrap {
+			// zeroing X15 manually if wrappers are not used
+			opregreg(s, x86.AXORPS, x86.REG_X15, x86.REG_X15)
+		}
 		off := duffStart(v.AuxInt)
 		adj := duffAdj(v.AuxInt)
 		var p *obj.Prog
@@ -915,12 +936,6 @@ func ssaGenValue(s *ssagen.State, v *ssa.Value) {
 		p.To.Type = obj.TYPE_ADDR
 		p.To.Sym = ir.Syms.Duffzero
 		p.To.Offset = off
-	case ssa.OpAMD64MOVOconst:
-		if v.AuxInt != 0 {
-			v.Fatalf("MOVOconst can only do constant=0")
-		}
-		r := v.Reg()
-		opregreg(s, x86.AXORPS, r, r)
 	case ssa.OpAMD64DUFFCOPY:
 		p := s.Prog(obj.ADUFFCOPY)
 		p.To.Type = obj.TYPE_ADDR
@@ -1000,7 +1015,17 @@ func ssaGenValue(s *ssagen.State, v *ssa.Value) {
 			q.To.Type = obj.TYPE_REG
 			q.To.Reg = r
 		}
-	case ssa.OpAMD64CALLstatic, ssa.OpAMD64CALLclosure, ssa.OpAMD64CALLinter:
+	case ssa.OpAMD64CALLstatic:
+		if s.ABI == obj.ABI0 && v.Aux.(*ssa.AuxCall).Fn.ABI() == obj.ABIInternal {
+			// zeroing X15 when entering ABIInternal from ABI0
+			opregreg(s, x86.AXORPS, x86.REG_X15, x86.REG_X15)
+		}
+		s.Call(v)
+		if s.ABI == obj.ABIInternal && v.Aux.(*ssa.AuxCall).Fn.ABI() == obj.ABI0 {
+			// zeroing X15 when entering ABIInternal from ABI0
+			opregreg(s, x86.AXORPS, x86.REG_X15, x86.REG_X15)
+		}
+	case ssa.OpAMD64CALLclosure, ssa.OpAMD64CALLinter:
 		s.Call(v)
 
 	case ssa.OpAMD64LoweredGetCallerPC:
@@ -1297,6 +1322,10 @@ func ssaGenBlock(s *ssagen.State, b, next *ssa.Block) {
 	case ssa.BlockRet:
 		s.Prog(obj.ARET)
 	case ssa.BlockRetJmp:
+		if s.ABI == obj.ABI0 && b.Aux.(*obj.LSym).ABI() == obj.ABIInternal {
+			// zeroing X15 when entering ABIInternal from ABI0
+			opregreg(s, x86.AXORPS, x86.REG_X15, x86.REG_X15)
+		}
 		p := s.Prog(obj.ARET)
 		p.To.Type = obj.TYPE_MEM
 		p.To.Name = obj.NAME_EXTERN