cmd/asm: add requested amd64 instructions

Add amd64 instructions I promised to add for Go 1.6
at the beginning of January.

These may be the last instructions added by hand.
I intend to generate the whole set mechanically for Go 1.7.

Fixes #13822.

Change-Id: I8c6bae2efd25f717f9ec750402e50f408a911d2b
Reviewed-on: https://go-review.googlesource.com/18853Reviewed-by: Rob Pike <r@golang.org>

src/cmd/asm/internal/arch/arch.go

@@ -162,11 +162,11 @@ func archX86(linkArch *obj.LinkArch) *Arch {
 	instructions["MOVDQ2Q"] = x86.AMOVQ
 	instructions["MOVNTDQ"] = x86.AMOVNTO
 	instructions["MOVOA"] = x86.AMOVO
-	instructions["MOVOA"] = x86.AMOVO
 	instructions["PF2ID"] = x86.APF2IL
 	instructions["PI2FD"] = x86.API2FL
 	instructions["PSLLDQ"] = x86.APSLLO
 	instructions["PSRLDQ"] = x86.APSRLO
+	instructions["PADDD"] = x86.APADDL
 
 	return &Arch{
 		LinkArch:       linkArch,

src/cmd/asm/internal/asm/testdata/amd64.s

@@ -121,5 +121,11 @@ label:
 loop:
 	LOOP	loop // LOOP
 
+	// Intel pseudonyms for our own renamings.
+	PADDD	M2, M1 // PADDL M2, M1
+	MOVDQ2Q	X1, M1 // MOVQ X1, M1
+	MOVNTDQ	X1, (AX)	// MOVNTO X1, (AX)
+	MOVOA	(AX), X1	// MOVO (AX), X1
+
 // LTYPE0 nonnon	{ outcode($1, &$2); }
 	RET // c3

src/cmd/internal/obj/x86/a.out.go

@@ -601,15 +601,15 @@ const (
 	APADDUSB
 	APADDUSW
 	APADDW
+	APAND
 	APANDB
 	APANDL
+	APANDN
 	APANDSB
 	APANDSW
 	APANDUSB
 	APANDUSW
 	APANDW
-	APAND
-	APANDN
 	APAVGB
 	APAVGW
 	APCMPEQB
@@ -618,10 +618,10 @@ const (
 	APCMPGTB
 	APCMPGTL
 	APCMPGTW
-	APEXTRW
 	APEXTRB
 	APEXTRD
 	APEXTRQ
+	APEXTRW
 	APFACC
 	APFADD
 	APFCMPEQ
@@ -633,42 +633,63 @@ const (
 	APFNACC
 	APFPNACC
 	APFRCP
-	APFRCPIT1
 	APFRCPI2T
+	APFRCPIT1
 	APFRSQIT1
 	APFRSQRT
 	APFSUB
 	APFSUBR
-	APINSRW
+	APHADDD
+	APHADDSW
+	APHADDW
+	APHMINPOSUW
+	APHSUBD
+	APHSUBSW
+	APHSUBW
 	APINSRB
 	APINSRD
 	APINSRQ
+	APINSRW
 	APMADDWL
 	APMAXSW
 	APMAXUB
 	APMINSW
 	APMINUB
 	APMOVMSKB
+	APMOVSXBD
+	APMOVSXBQ
+	APMOVSXBW
+	APMOVSXDQ
+	APMOVSXWD
+	APMOVSXWQ
+	APMOVZXBD
+	APMOVZXBQ
+	APMOVZXBW
+	APMOVZXDQ
+	APMOVZXWD
+	APMOVZXWQ
+	APMULDQ
 	APMULHRW
 	APMULHUW
 	APMULHW
+	APMULLD
 	APMULLW
 	APMULULQ
 	APOR
 	APSADBW
+	APSHUFB
 	APSHUFHW
 	APSHUFL
 	APSHUFLW
 	APSHUFW
-	APSHUFB
-	APSLLO
 	APSLLL
+	APSLLO
 	APSLLQ
 	APSLLW
 	APSRAL
 	APSRAW
-	APSRLO
 	APSRLL
+	APSRLO
 	APSRLQ
 	APSRLW
 	APSUBB

src/cmd/internal/obj/x86/anames.go

@@ -550,15 +550,15 @@ var Anames = []string{
 	"PADDUSB",
 	"PADDUSW",
 	"PADDW",
+	"PAND",
 	"PANDB",
 	"PANDL",
+	"PANDN",
 	"PANDSB",
 	"PANDSW",
 	"PANDUSB",
 	"PANDUSW",
 	"PANDW",
-	"PAND",
-	"PANDN",
 	"PAVGB",
 	"PAVGW",
 	"PCMPEQB",
@@ -567,10 +567,10 @@ var Anames = []string{
 	"PCMPGTB",
 	"PCMPGTL",
 	"PCMPGTW",
-	"PEXTRW",
 	"PEXTRB",
 	"PEXTRD",
 	"PEXTRQ",
+	"PEXTRW",
 	"PFACC",
 	"PFADD",
 	"PFCMPEQ",
@@ -582,42 +582,63 @@ var Anames = []string{
 	"PFNACC",
 	"PFPNACC",
 	"PFRCP",
-	"PFRCPIT1",
 	"PFRCPI2T",
+	"PFRCPIT1",
 	"PFRSQIT1",
 	"PFRSQRT",
 	"PFSUB",
 	"PFSUBR",
-	"PINSRW",
+	"PHADDD",
+	"PHADDSW",
+	"PHADDW",
+	"PHMINPOSUW",
+	"PHSUBD",
+	"PHSUBSW",
+	"PHSUBW",
 	"PINSRB",
 	"PINSRD",
 	"PINSRQ",
+	"PINSRW",
 	"PMADDWL",
 	"PMAXSW",
 	"PMAXUB",
 	"PMINSW",
 	"PMINUB",
 	"PMOVMSKB",
+	"PMOVSXBD",
+	"PMOVSXBQ",
+	"PMOVSXBW",
+	"PMOVSXDQ",
+	"PMOVSXWD",
+	"PMOVSXWQ",
+	"PMOVZXBD",
+	"PMOVZXBQ",
+	"PMOVZXBW",
+	"PMOVZXDQ",
+	"PMOVZXWD",
+	"PMOVZXWQ",
+	"PMULDQ",
 	"PMULHRW",
 	"PMULHUW",
 	"PMULHW",
+	"PMULLD",
 	"PMULLW",
 	"PMULULQ",
 	"POR",
 	"PSADBW",
+	"PSHUFB",
 	"PSHUFHW",
 	"PSHUFL",
 	"PSHUFLW",
 	"PSHUFW",
-	"PSHUFB",
-	"PSLLO",
 	"PSLLL",
+	"PSLLO",
 	"PSLLQ",
 	"PSLLW",
 	"PSRAL",
 	"PSRAW",
-	"PSRLO",
 	"PSRLL",
+	"PSRLO",
 	"PSRLQ",
 	"PSRLW",
 	"PSUBB",

src/cmd/internal/obj/x86/asm6.go

@@ -222,6 +222,7 @@ const (
 	Pf3  = 0xf3 /* xmm escape 2: f3 0f */
 	Pef3 = 0xf5 /* xmm escape 2 with 16-bit prefix: 66 f3 0f */
 	Pq3  = 0x67 /* xmm escape 3: 66 48 0f */
+	Pq4  = 0x68 /* xmm escape 4: 66 0F 38 */
 	Pfw  = 0xf4 /* Pf3 with Rex.w: f3 48 0f */
 	Pw   = 0x48 /* Rex.w */
 	Pw8  = 0x90 // symbolic; exact value doesn't matter
@@ -675,6 +676,10 @@ var yxm = []ytab{
 	{Yxm, Ynone, Yxr, Zm_r_xm, 1},
 }
 
+var yxm_q4 = []ytab{
+	{Yxm, Ynone, Yxr, Zm_r, 1},
+}
+
 var yxcvm1 = []ytab{
 	{Yxm, Ynone, Yxr, Zm_r_xm, 2},
 	{Yxm, Ynone, Ymr, Zm_r_xm, 2},
@@ -817,6 +822,10 @@ var yxabort = []ytab{
 	{Yu8, Ynone, Ynone, Zib_, 1},
 }
 
+var ylddqu = []ytab{
+	{Ym, Ynone, Yxr, Zm_r, 1},
+}
+
 // VEX instructions that come in two forms:
 //	VTHING xmm2/m128, xmmV, xmm1
 //	VTHING ymm2/m256, ymmV, ymm1
@@ -873,6 +882,11 @@ var yvex_xxmyxm = []ytab{
 	{Yyr, Ynone, Yxm, Zvex_r_v_rm, 2},
 }
 
+var ymmxmm0f38 = []ytab{
+	{Ymm, Ynone, Ymr, Zlitm_r, 3},
+	{Yxm, Ynone, Yxr, Zlitm_r, 5},
+}
+
 /*
  * You are doasm, holding in your hand a Prog* with p->as set to, say, ACRC32,
  * and p->from and p->to as operands (Addr*).  The linker scans optab to find
@@ -1149,6 +1163,7 @@ var optab =
 	{ALAHF, ynone, Px, [23]uint8{0x9f}},
 	{ALARL, yml_rl, Pm, [23]uint8{0x02}},
 	{ALARW, yml_rl, Pq, [23]uint8{0x02}},
+	{ALDDQU, ylddqu, Pf2, [23]uint8{0xf0}},
 	{ALDMXCSR, ysvrs, Pm, [23]uint8{0xae, 02, 0xae, 02}},
 	{ALEAL, ym_rl, Px, [23]uint8{0x8d}},
 	{ALEAQ, ym_rl, Pw, [23]uint8{0x8d}},
@@ -1293,6 +1308,13 @@ var optab =
 	{APFRSQRT, ymfp, Px, [23]uint8{0x97}},
 	{APFSUB, ymfp, Px, [23]uint8{0x9a}},
 	{APFSUBR, ymfp, Px, [23]uint8{0xaa}},
+	{APHADDD, ymmxmm0f38, Px, [23]uint8{0x0F, 0x38, 0x02, 0, 0x66, 0x0F, 0x38, 0x02, 0}},
+	{APHADDSW, yxm_q4, Pq4, [23]uint8{0x03}},
+	{APHADDW, yxm_q4, Pq4, [23]uint8{0x01}},
+	{APHMINPOSUW, yxm_q4, Pq4, [23]uint8{0x41}},
+	{APHSUBD, yxm_q4, Pq4, [23]uint8{0x06}},
+	{APHSUBSW, yxm_q4, Pq4, [23]uint8{0x07}},
+	{APHSUBW, yxm_q4, Pq4, [23]uint8{0x05}},
 	{APINSRW, yinsrw, Pq, [23]uint8{0xc4, 00}},
 	{APINSRB, yinsr, Pq, [23]uint8{0x3a, 0x20, 00}},
 	{APINSRD, yinsr, Pq, [23]uint8{0x3a, 0x22, 00}},
@@ -1303,9 +1325,23 @@ var optab =
 	{APMINSW, yxm, Pe, [23]uint8{0xea}},
 	{APMINUB, yxm, Pe, [23]uint8{0xda}},
 	{APMOVMSKB, ymskb, Px, [23]uint8{Pe, 0xd7, 0xd7}},
+	{APMOVSXBD, yxm_q4, Pq4, [23]uint8{0x21}},
+	{APMOVSXBQ, yxm_q4, Pq4, [23]uint8{0x22}},
+	{APMOVSXBW, yxm_q4, Pq4, [23]uint8{0x20}},
+	{APMOVSXDQ, yxm_q4, Pq4, [23]uint8{0x25}},
+	{APMOVSXWD, yxm_q4, Pq4, [23]uint8{0x23}},
+	{APMOVSXWQ, yxm_q4, Pq4, [23]uint8{0x24}},
+	{APMOVZXBD, yxm_q4, Pq4, [23]uint8{0x31}},
+	{APMOVZXBQ, yxm_q4, Pq4, [23]uint8{0x32}},
+	{APMOVZXBW, yxm_q4, Pq4, [23]uint8{0x30}},
+	{APMOVZXDQ, yxm_q4, Pq4, [23]uint8{0x35}},
+	{APMOVZXWD, yxm_q4, Pq4, [23]uint8{0x33}},
+	{APMOVZXWQ, yxm_q4, Pq4, [23]uint8{0x34}},
+	{APMULDQ, yxm_q4, Pq4, [23]uint8{0x28}},
 	{APMULHRW, ymfp, Px, [23]uint8{0xb7}},
 	{APMULHUW, ymm, Py1, [23]uint8{0xe4, Pe, 0xe4}},
 	{APMULHW, ymm, Py1, [23]uint8{0xe5, Pe, 0xe5}},
+	{APMULLD, yxm_q4, Pq4, [23]uint8{0x40}},
 	{APMULLW, ymm, Py1, [23]uint8{0xd5, Pe, 0xd5}},
 	{APMULULQ, ymm, Py1, [23]uint8{0xf4, Pe, 0xf4}},
 	{APOPAL, ynone, P32, [23]uint8{0x61}},
@@ -3292,6 +3328,12 @@ func doasm(ctxt *obj.Link, p *obj.Prog) {
 				ctxt.Andptr[0] = Pm
 				ctxt.Andptr = ctxt.Andptr[1:]
 
+			case Pq4: /*  66 0F 38 */
+				ctxt.Andptr[0] = 0x66
+				ctxt.Andptr[1] = 0x0F
+				ctxt.Andptr[2] = 0x38
+				ctxt.Andptr = ctxt.Andptr[3:]
+
 			case Pf2, /* xmm opcode escape */
 				Pf3:
 				ctxt.Andptr[0] = byte(o.prefix)