cmd/asm, cmd/internal/obj/s390x, math: add LGDR and LDGR instructions

The instructions allow moves between floating point and general
purpose registers without any conversion taking place.

Change-Id: I82c6f3ad9c841a83783b5be80dcf5cd538ff49e6
Reviewed-on: https://go-review.googlesource.com/38777
Run-TryBot: Michael Munday <munday@ca.ibm.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

src/cmd/asm/internal/asm/testdata/s390x.s

@@ -244,6 +244,9 @@ TEXT main·foo(SB),DUPOK|NOSPLIT,$16-0 // TEXT main.foo(SB), DUPOK|NOSPLIT, $16-
 	CMPUBGT	R9, $256, 0(PC)        // ec920000007d
 	CMPUBGE	R2, $0, 0(PC)          // ec2a0000007d
 
+	LGDR	F1, R12                // b3cd00c1
+	LDGR	R2, F15                // b3c100f2
+
 	CEFBRA	R0, F15                // b39400f0
 	CDFBRA	R1, F14                // b39500e1
 	CEGBRA	R2, F13                // b3a400d2

src/cmd/internal/obj/s390x/a.out.go

@@ -291,6 +291,10 @@ const (
 	AFIEBR
 	AFIDBR
 
+	// move from GPR to FPR and vice versa
+	ALDGR
+	ALGDR
+
 	// convert from int32/int64 to float/float64
 	ACEFBRA
 	ACDFBRA

src/cmd/internal/obj/s390x/anames.go

@@ -87,6 +87,8 @@ var Anames = []string{
 	"FSQRTS",
 	"FIEBR",
 	"FIDBR",
+	"LDGR",
+	"LGDR",
 	"CEFBRA",
 	"CDFBRA",
 	"CEGBRA",

src/cmd/internal/obj/s390x/asmz.go

@@ -201,6 +201,8 @@ var optab = []Optab{
 	Optab{AFMOVD, C_FREG, C_NONE, C_NONE, C_LOREG, 35, 0},
 	Optab{AFMOVD, C_FREG, C_NONE, C_NONE, C_ADDR, 74, 0},
 	Optab{AFMOVD, C_ZCON, C_NONE, C_NONE, C_FREG, 67, 0},
+	Optab{ALDGR, C_REG, C_NONE, C_NONE, C_FREG, 81, 0},
+	Optab{ALGDR, C_FREG, C_NONE, C_NONE, C_REG, 81, 0},
 	Optab{ACEFBRA, C_REG, C_NONE, C_NONE, C_FREG, 82, 0},
 	Optab{ACFEBRA, C_FREG, C_NONE, C_NONE, C_REG, 83, 0},
 	Optab{AFIEBR, C_SCON, C_FREG, C_NONE, C_FREG, 48, 0},
@@ -266,7 +268,7 @@ var optab = []Optab{
 	Optab{ADWORD, C_DCON, C_NONE, C_NONE, C_NONE, 31, 0},
 
 	// fast synchronization
-	Optab{ASYNC, C_NONE, C_NONE, C_NONE, C_NONE, 81, 0},
+	Optab{ASYNC, C_NONE, C_NONE, C_NONE, C_NONE, 80, 0},
 
 	// store clock
 	Optab{ASTCK, C_NONE, C_NONE, C_NONE, C_SAUTO, 88, REGSP},
@@ -3474,9 +3476,17 @@ func (c *ctxtz) asmout(p *obj.Prog, asm *[]byte) {
 			zRSY(op_CSG, uint32(p.From.Reg), uint32(p.Reg), uint32(p.To.Reg), uint32(v), asm)
 		}
 
-	case 81: // sync
+	case 80: // sync
 		zRR(op_BCR, 0xE, 0, asm)
 
+	case 81: // float to fixed and fixed to float moves (no conversion)
+		switch p.As {
+		case ALDGR:
+			zRRE(op_LDGR, uint32(p.To.Reg), uint32(p.From.Reg), asm)
+		case ALGDR:
+			zRRE(op_LGDR, uint32(p.To.Reg), uint32(p.From.Reg), asm)
+		}
+
 	case 82: // fixed to float conversion
 		var opcode uint32
 		switch p.As {

src/math/cosh_s390x.s

@@ -109,7 +109,7 @@ L14:
 	FMOVD   0(R1), F3
 	WFMADB  V1, V7, V3, V7
 	FNEG    F2, F3
-	WORD    $0xB3CD0015     //lgdr %r1,%f5
+	LGDR    F5, R1
 	MOVD    $coshe2<>+0(SB), R3
 	WFCEDBS V4, V0, V0
 	FMOVD   0(R3), F5
@@ -134,7 +134,7 @@ L17:
 	WORD    $0xEC21000F     //risbgn %r2,%r1,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
-	WORD    $0xB3C10022     //ldgr %f2,%r2
+	LDGR    R2, F2
 	FMADD   F2, F6, F2
 	MOVD    $coshx4ff<>+0(SB), R1
 	FMOVD   0(R1), F0
@@ -173,7 +173,7 @@ L20:
 	MOVD    $coshe1<>+0(SB), R1
 	FMOVD   0(R1), F5
 	WFMADB  V1, V2, V5, V1
-	WORD    $0xB3CD0013     //lgdr  %r1,%f3
+	LGDR    F3, R1
 	MOVD    $coshtab<>+0(SB), R5
 	WFMADB  V4, V6, V1, V3
 	WORD    $0xEC4139BC     //risbg %r4,%r1,57,128+60,3
@@ -197,8 +197,8 @@ L20:
 	WORD    $0xEC34000F     //risbgn %r3,%r4,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
-	WORD    $0xB3C10022     //ldgr %f2,%r2
-	WORD    $0xB3C10003     //ldgr %f0,%r3
+	LDGR    R2, F2
+	LDGR    R3, F0
 	FMADD   F2, F1, F2
 	FMADD   F0, F6, F0
 	FADD    F2, F0

src/math/log10_s390x.s

@@ -108,7 +108,7 @@ L8:
 	ORW     $0x45000000, R2
 L4:
 	FMOVD   log10rodataL19<>+120(SB), F2
-	WORD    $0xB3C10041     //ldgr  %f4,%r1
+	LDGR    R1, F4
 	WFMADB  V4, V0, V2, V0
 	FMOVD   log10rodataL19<>+112(SB), F4
 	FMOVD   log10rodataL19<>+104(SB), F6

src/math/sin_s390x.s

@@ -108,7 +108,7 @@ L2:
 	WFMDB   V0, V0, V1
 	FMOVD   0(R1), F7
 	WFMDB   V1, V1, V2
-	WORD    $0xB3CD0013     //lgdr  %r1,%f3
+	LGDR    F3, R1
 	MOVD    $sincosxlim<>+0(SB), R2
 	WORD    $0xA7110001     //tmll  %r1,1
 	BEQ     L6
@@ -245,7 +245,7 @@ L21:
 	WFMSDB  V0, V2, V3, V2
 	FMOVD   0(R1), F3
 	WFCHDBS V3, V1, V3
-	WORD    $0xB3CD0012     //lgdr %r1,%f2
+	LGDR    F2, R1
 	BEQ     L36
 	MOVD    $sincosxadd<>+0(SB), R2
 	FMOVD   0(R2), F4

src/math/sinh_s390x.s

@@ -112,7 +112,7 @@ L6:
 	MOVD    $sinhxadd<>+0(SB), R2
 	FMOVD   0(R2), F0
 	MOVD    sinhrlog2<>+0(SB), R2
-	WORD    $0xB3C10062     //ldgr  %f6,%r2
+	LDGR    R2, F6
 	WFMSDB  V4, V6, V0, V16
 	FMOVD   sinhrodataL21<>+8(SB), F6
 	WFADB   V0, V16, V0
@@ -168,7 +168,7 @@ L6:
 	WORD    $0xEC12000F     //risbgn %r1,%r2,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
-	WORD    $0xB3C10021     //ldgr %f2,%r1
+	LDGR    R1, F2
 	FMUL    F2, F0
 	FMOVD   F0, ret+8(FP)
 	RET
@@ -177,7 +177,7 @@ L20:
 	MOVD    $sinhxadd<>+0(SB), R2
 	FMOVD   0(R2), F2
 	MOVD    sinhrlog2<>+0(SB), R2
-	WORD    $0xB3C10002     //ldgr  %f0,%r2
+	LDGR    R2, F0
 	WFMSDB  V4, V0, V2, V6
 	FMOVD   sinhrodataL21<>+8(SB), F0
 	FADD    F6, F2
@@ -193,13 +193,13 @@ L20:
 	FMOVD   0(R2), F3
 	MOVD    $sinhe6<>+0(SB), R2
 	FMOVD   0(R2), F5
-	WORD    $0xB3CD0026     //lgdr %r2,%f6
+	LGDR    F6, R2
 	RLL     $3, R2, R2
 	WORD    $0xEC12000F     //risbgn %r1,%r2,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
 	WFMADB  V2, V1, V0, V1
-	WORD    $0xB3C10001     //ldgr  %f0,%r1
+	LDGR    R1, F0
 	MOVD    $sinhe5<>+0(SB), R1
 	WFMADB  V2, V3, V5, V3
 	FMOVD   0(R1), F5
@@ -214,7 +214,7 @@ L20:
 	WORD    $0xEC32000F     //risbgn %r3,%r2,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
-	WORD    $0xB3C10063     //ldgr  %f6,%r3
+	LDGR    R3, F6
 	WFADB   V0, V6, V16
 	MOVD    $sinhe4<>+0(SB), R1
 	WFMADB  V1, V7, V5, V1
@@ -245,7 +245,7 @@ L9:
 	WORD    $0xEC12000F     //risbgn %r1,%r2,64-64+0,64-64+0+16-1,64-0-16
 	BYTE    $0x30
 	BYTE    $0x59
-	WORD    $0xB3C10021     //ldgr %f2,%r1
+	LDGR    R1, F2
 	FMUL    F2, F0
 	FMOVD   F0, ret+8(FP)
 	RET

src/math/tanh_s390x.s

@@ -59,7 +59,7 @@ GLOBL tanhtab<>+0(SB), RODATA, $128
 
 TEXT ·tanhAsm(SB),NOSPLIT,$0-16
 	FMOVD   x+0(FP), F0
-	//specail case Tanh(±0) = ±0
+	// special case Tanh(±0) = ±0
 	FMOVD   $(0.0), F1
 	FCMPU   F0, F1
 	BEQ     tanhIsZero
@@ -72,10 +72,10 @@ L2:
 	MOVD    $tanhxadd<>+0(SB), R2
 	FMOVD   0(R2), F2
 	MOVD    tanhrlog2<>+0(SB), R2
-	WORD    $0xB3C10042     //ldgr %f4,%r2
+	LDGR    R2, F4
 	WFMSDB  V0, V4, V2, V4
 	MOVD    $tanhtab<>+0(SB), R3
-	WORD    $0xB3CD0024     //lgdr %r2,%f4
+	LGDR    F4, R2
 	WORD    $0xEC4239BC     //risbg %r4,%r2,57,128+60,3
 	BYTE    $0x03
 	BYTE    $0x55
@@ -86,7 +86,7 @@ L2:
 	BYTE    $0x30
 	BYTE    $0x59
 	WORD    $0x68543000     //ld %f5,0(%r4,%r3)
-	WORD    $0xB3C10061     //ldgr %f6,%r1
+	LDGR    R1, F6
 	BLT     L3
 	MOVD    $tanhxzero<>+0(SB), R1
 	FMOVD   0(R1), F2