cmd/asm: add VLD[1-4]R vector instructions on arm64

This change adds VLD1R, VLD2R, VLD3R, VLD4R

Change-Id: Ie19e9ae02fdfc94b9344acde8c9938849efb0bf0
Reviewed-on: https://go-review.googlesource.com/c/go/+/181697
Run-TryBot: Cherry Zhang <cherryyz@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Cherry Zhang <cherryyz@google.com>

4 files changed, 53 insertions, 19 deletions

diff --git a/src/cmd/asm/internal/asm/testdata/arm64.s b/src/cmd/asm/internal/asm/testdata/arm64.s
index 9f19ff1e8d..93f70045b7 100644
--- a/src/cmd/asm/internal/asm/testdata/arm64.s
+++ b/src/cmd/asm/internal/asm/testdata/arm64.s
@@ -352,6 +352,18 @@ TEXT	foo(SB), DUPOK|NOSPLIT, $-8
 	VLD4	(R15), [V10.H4, V11.H4, V12.H4, V13.H4]         // ea05400c
 	VLD4.P	32(R24), [V31.B8, V0.B8, V1.B8, V2.B8]          // 1f03df0c
 	VLD4.P	(R13)(R9), [V14.S2, V15.S2, V16.S2, V17.S2]     // VLD4.P	(R13)(R9*1), [V14.S2,V15.S2,V16.S2,V17.S2] // ae09c90c
+	VLD1R	(R0), [V0.B16]					// 00c0404d
+	VLD1R.P	16(R0), [V0.B16]				// 00c0df4d
+	VLD1R.P	(R15)(R1), [V15.H4]				// VLD1R.P	(R15)(R1*1), [V15.H4] // efc5c10d
+	VLD2R	(R15), [V15.H4, V16.H4]				// efc5600d
+	VLD2R.P	32(R0), [V0.D2, V1.D2]				// 00ccff4d
+	VLD2R.P	(R0)(R5), [V31.D1, V0.D1]			// VLD2R.P	(R0)(R5*1), [V31.D1, V0.D1] // 1fcce50d
+	VLD3R	(RSP), [V31.S2, V0.S2, V1.S2]			// ffeb400d
+	VLD3R.P	24(R15), [V15.H4, V16.H4, V17.H4]		// efe5df0d
+	VLD3R.P	(R15)(R6), [V15.H8, V16.H8, V17.H8]		// VLD3R.P	(R15)(R6*1), [V15.H8, V16.H8, V17.H8] // efe5c64d
+	VLD4R	(R0), [V0.B8, V1.B8, V2.B8, V3.B8]		// 00e0600d
+	VLD4R.P	64(RSP), [V31.S4, V0.S4, V1.S4, V2.S4]		// ffebff4d
+	VLD4R.P	(R15)(R9), [V15.H4, V16.H4, V17.H4, V18.H4]	// VLD4R.P	(R15)(R9*1), [V15.H4, V16.H4, V17.H4, V18.H4] // efe5e90d
 	VST1.P	[V24.S2], 8(R2)                                 // 58789f0c
 	VST1	[V29.S2, V30.S2], (R29)                         // bdab000c
 	VST1	[V14.H4, V15.H4, V16.H4], (R27)                 // 6e67000c
diff --git a/src/cmd/internal/obj/arm64/a.out.go b/src/cmd/internal/obj/arm64/a.out.go
index f793cdc4f9..4e5eb75a22 100644
--- a/src/cmd/internal/obj/arm64/a.out.go
+++ b/src/cmd/internal/obj/arm64/a.out.go
@@ -956,6 +956,10 @@ const (
 	AVLD2
 	AVLD3
 	AVLD4
+	AVLD1R
+	AVLD2R
+	AVLD3R
+	AVLD4R
 	AVORR
 	AVREV32
 	AVREV64
diff --git a/src/cmd/internal/obj/arm64/anames.go b/src/cmd/internal/obj/arm64/anames.go
index 621af6c195..2c277dfb95 100644
--- a/src/cmd/internal/obj/arm64/anames.go
+++ b/src/cmd/internal/obj/arm64/anames.go
@@ -463,6 +463,10 @@ var Anames = []string{
 	"VLD2",
 	"VLD3",
 	"VLD4",
+	"VLD1R",
+	"VLD2R",
+	"VLD3R",
+	"VLD4R",
 	"VORR",
 	"VREV32",
 	"VREV64",
diff --git a/src/cmd/internal/obj/arm64/asm7.go b/src/cmd/internal/obj/arm64/asm7.go
index 47586ba262..c7fa943e7e 100644
--- a/src/cmd/internal/obj/arm64/asm7.go
+++ b/src/cmd/internal/obj/arm64/asm7.go
@@ -784,15 +784,9 @@ var optab = []Optab{
 	{AVLD1, C_ZOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, 0},
 	{AVLD1, C_LOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
 	{AVLD1, C_ROFF, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD2, C_ZOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, 0},
-	{AVLD2, C_LOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD2, C_ROFF, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD3, C_ZOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, 0},
-	{AVLD3, C_LOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD3, C_ROFF, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD4, C_ZOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, 0},
-	{AVLD4, C_LOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
-	{AVLD4, C_ROFF, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
+	{AVLD1R, C_ZOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, 0},
+	{AVLD1R, C_LOREG, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
+	{AVLD1R, C_ROFF, C_NONE, C_NONE, C_LIST, 81, 4, 0, 0, C_XPOST},
 	{AVLD1, C_LOREG, C_NONE, C_NONE, C_ELEM, 97, 4, 0, 0, C_XPOST},
 	{AVLD1, C_ROFF, C_NONE, C_NONE, C_ELEM, 97, 4, 0, 0, C_XPOST},
 	{AVLD1, C_LOREG, C_NONE, C_NONE, C_ELEM, 97, 4, 0, 0, 0},
@@ -2709,13 +2703,18 @@ func buildop(ctxt *obj.Link) {
 		case AVZIP1:
 			oprangeset(AVZIP2, t)
 
+		case AVLD1R:
+			oprangeset(AVLD2, t)
+			oprangeset(AVLD2R, t)
+			oprangeset(AVLD3, t)
+			oprangeset(AVLD3R, t)
+			oprangeset(AVLD4, t)
+			oprangeset(AVLD4R, t)
+
 		case ASHA1H,
 			AVCNT,
 			AVMOV,
 			AVLD1,
-			AVLD2,
-			AVLD3,
-			AVLD4,
 			AVST1,
 			AVST2,
 			AVST3,
@@ -2803,7 +2802,7 @@ func (c *ctxt7) checkindex(p *obj.Prog, index, maxindex int) {
 func (c *ctxt7) checkoffset(p *obj.Prog, as obj.As) {
 	var offset, list, n, expect int64
 	switch as {
-	case AVLD1, AVLD2, AVLD3, AVLD4:
+	case AVLD1, AVLD2, AVLD3, AVLD4, AVLD1R, AVLD2R, AVLD3R, AVLD4R:
 		offset = p.From.Offset
 		list = p.To.Offset
 	case AVST1, AVST2, AVST3, AVST4:
@@ -2836,11 +2835,13 @@ func (c *ctxt7) checkoffset(p *obj.Prog, as obj.As) {
 	switch as {
 	case AVLD1, AVST1:
 		return
-	case AVLD2, AVST2:
+	case AVLD1R:
+		expect = 1
+	case AVLD2, AVST2, AVLD2R:
 		expect = 2
-	case AVLD3, AVST3:
+	case AVLD3, AVST3, AVLD3R:
 		expect = 3
-	case AVLD4, AVST4:
+	case AVLD4, AVST4, AVLD4R:
 		expect = 4
 	}
 
@@ -4344,10 +4345,10 @@ func (c *ctxt7) asmout(p *obj.Prog, o *Optab, out []uint32) {
 		}
 		o1 |= (uint32(imm5&0x1f) << 16) | (uint32(rf&31) << 5) | uint32(rt&31)
 
-	case 81: /* vld[1-4] (Rn), [Vt1.<T>, Vt2.<T>, ...] */
+	case 81: /* vld[1-4]|vld[1-4]r (Rn), [Vt1.<T>, Vt2.<T>, ...] */
 		c.checkoffset(p, p.As)
 		r := int(p.From.Reg)
-		o1 = 3<<26 | 1<<22
+		o1 = c.oprrr(p, p.As)
 		if o.scond == C_XPOST {
 			o1 |= 1 << 23
 			if p.From.Index == 0 {
@@ -4358,7 +4359,7 @@ func (c *ctxt7) asmout(p *obj.Prog, o *Optab, out []uint32) {
 				if isRegShiftOrExt(&p.From) {
 					c.ctxt.Diag("invalid extended register op: %v\n", p)
 				}
-				o1 |= uint32(p.From.Index&31) << 16
+				o1 |= uint32(p.From.Index&0x1f) << 16
 			}
 		}
 		o1 |= uint32(p.To.Offset)
@@ -5591,6 +5592,15 @@ func (c *ctxt7) oprrr(p *obj.Prog, a obj.As) uint32 {
 
 	case AVRBIT:
 		return 0x2E<<24 | 1<<22 | 0x10<<17 | 5<<12 | 2<<10
+
+	case AVLD1, AVLD2, AVLD3, AVLD4:
+		return 3<<26 | 1<<22
+
+	case AVLD1R, AVLD3R:
+		return 0xD<<24 | 1<<22
+
+	case AVLD2R, AVLD4R:
+		return 0xD<<24 | 3<<21
 	}
 
 	c.ctxt.Diag("%v: bad rrr %d %v", p, a, a)
@@ -6779,6 +6789,10 @@ func (c *ctxt7) maskOpvldvst(p *obj.Prog, o1 uint32) uint32 {
 
 	o1 &^= 0xf000 // mask out "opcode" field (bit 12-15)
 	switch p.As {
+	case AVLD1R, AVLD2R:
+		o1 |= 0xC << 12
+	case AVLD3R, AVLD4R:
+		o1 |= 0xE << 12
 	case AVLD2, AVST2:
 		o1 |= 8 << 12
 	case AVLD3, AVST3: