cmd/asm: add amd64 PDEP, PEXT, and related integer VEX instructions

Requested off-list.
Trivial to add and more importantly trivial to test.

ANDNL
ANDNQ
BEXTRL
BEXTRQ
BZHIL
BZHIQ
MULXL
MULXQ
PDEPL
PDEPQ
PEXTL
PEXTQ
SARXL
SARXQ
SHRXL
SHRXQ

Change-Id: I3d46a0f653b81dd003ff6d2a394d8ce96a573b63
Reviewed-on: https://go-review.googlesource.com/18857Reviewed-by: Rob Pike <r@golang.org>

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

@@ -510,10 +510,22 @@ const (
 	AADDPS
 	AADDSD
 	AADDSS
+	AANDNL
+	AANDNQ
 	AANDNPD
 	AANDNPS
 	AANDPD
 	AANDPS
+	ABEXTRL
+	ABEXTRQ
+	ABLSIL
+	ABLSIQ
+	ABLSMSKL
+	ABLSMSKQ
+	ABLSRL
+	ABLSRQ
+	ABZHIL
+	ABZHIQ
 	ACMPPD
 	ACMPPS
 	ACMPSD
@@ -588,6 +600,8 @@ const (
 	AMULPS
 	AMULSD
 	AMULSS
+	AMULXL
+	AMULXQ
 	AORPD
 	AORPS
 	APACKSSLW
@@ -618,6 +632,10 @@ const (
 	APCMPGTB
 	APCMPGTL
 	APCMPGTW
+	APDEPL
+	APDEPQ
+	APEXTL
+	APEXTQ
 	APEXTRB
 	APEXTRD
 	APEXTRQ
@@ -714,6 +732,12 @@ const (
 	ARCPSS
 	ARSQRTPS
 	ARSQRTSS
+	ASARXL
+	ASARXQ
+	ASHLXL
+	ASHLXQ
+	ASHRXL
+	ASHRXQ
 	ASHUFPD
 	ASHUFPS
 	ASQRTPD

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

@@ -459,10 +459,22 @@ var Anames = []string{
 	"ADDPS",
 	"ADDSD",
 	"ADDSS",
+	"ANDNL",
+	"ANDNQ",
 	"ANDNPD",
 	"ANDNPS",
 	"ANDPD",
 	"ANDPS",
+	"BEXTRL",
+	"BEXTRQ",
+	"BLSIL",
+	"BLSIQ",
+	"BLSMSKL",
+	"BLSMSKQ",
+	"BLSRL",
+	"BLSRQ",
+	"BZHIL",
+	"BZHIQ",
 	"CMPPD",
 	"CMPPS",
 	"CMPSD",
@@ -537,6 +549,8 @@ var Anames = []string{
 	"MULPS",
 	"MULSD",
 	"MULSS",
+	"MULXL",
+	"MULXQ",
 	"ORPD",
 	"ORPS",
 	"PACKSSLW",
@@ -567,6 +581,10 @@ var Anames = []string{
 	"PCMPGTB",
 	"PCMPGTL",
 	"PCMPGTW",
+	"PDEPL",
+	"PDEPQ",
+	"PEXTL",
+	"PEXTQ",
 	"PEXTRB",
 	"PEXTRD",
 	"PEXTRQ",
@@ -663,6 +681,12 @@ var Anames = []string{
 	"RCPSS",
 	"RSQRTPS",
 	"RSQRTSS",
+	"SARXL",
+	"SARXQ",
+	"SHLXL",
+	"SHLXQ",
+	"SHRXL",
+	"SHRXQ",
 	"SHUFPD",
 	"SHUFPS",
 	"SQRTPD",

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

@@ -207,6 +207,7 @@ const (
 	Zbyte
 	Zvex_rm_v_r
 	Zvex_r_v_rm
+	Zvex_v_rm_r
 	Zmax
 )
 
@@ -850,6 +851,16 @@ var yvex_xy3 = []ytab{
 	{Yym, Yyr, Yyr, Zvex_rm_v_r, 2},
 }
 
+var yvex_r3 = []ytab{
+	{Yml, Yrl, Yrl, Zvex_rm_v_r, 2},
+	{Yml, Yrl, Yrl, Zvex_rm_v_r, 2},
+}
+
+var yvex_vmr3 = []ytab{
+	{Yrl, Yml, Yrl, Zvex_v_rm_r, 2},
+	{Yrl, Yml, Yrl, Zvex_v_rm_r, 2},
+}
+
 var yvex_xy2 = []ytab{
 	{Yxm, Ynone, Yxr, Zvex_rm_v_r, 2},
 	{Yym, Ynone, Yyr, Zvex_rm_v_r, 2},
@@ -1669,6 +1680,25 @@ var optab =
 	{APSHUFD, yxshuf, Pq, [23]uint8{0x70, 0}},
 	{APCLMULQDQ, yxshuf, Pq, [23]uint8{0x3a, 0x44, 0}},
 
+	{AANDNL, yvex_r3, Pvex, [23]uint8{VEX_LZ_0F38_W0, 0xF2}},
+	{AANDNQ, yvex_r3, Pvex, [23]uint8{VEX_LZ_0F38_W1, 0xF2}},
+	{ABEXTRL, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_0F38_W0, 0xF7}},
+	{ABEXTRQ, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_0F38_W1, 0xF7}},
+	{ABZHIL, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_0F38_W0, 0xF5}},
+	{ABZHIQ, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_0F38_W1, 0xF5}},
+	{AMULXL, yvex_r3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W0, 0xF6}},
+	{AMULXQ, yvex_r3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W1, 0xF6}},
+	{APDEPL, yvex_r3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W0, 0xF5}},
+	{APDEPQ, yvex_r3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W1, 0xF5}},
+	{APEXTL, yvex_r3, Pvex, [23]uint8{VEX_LZ_F3_0F38_W0, 0xF5}},
+	{APEXTQ, yvex_r3, Pvex, [23]uint8{VEX_LZ_F3_0F38_W1, 0xF5}},
+	{ASARXL, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_F3_0F38_W0, 0xF7}},
+	{ASARXQ, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_F3_0F38_W1, 0xF7}},
+	{ASHLXL, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_66_0F38_W0, 0xF7}},
+	{ASHLXQ, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_66_0F38_W1, 0xF7}},
+	{ASHRXL, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W0, 0xF7}},
+	{ASHRXQ, yvex_vmr3, Pvex, [23]uint8{VEX_LZ_F2_0F38_W1, 0xF7}},
+
 	{AVZEROUPPER, ynone, Px, [23]uint8{0xc5, 0xf8, 0x77}},
 	{AVMOVDQU, yvex_vmovdqa, Pvex, [23]uint8{VEX_128_F3_0F_WIG, 0x6F, VEX_128_F3_0F_WIG, 0x7F, VEX_256_F3_0F_WIG, 0x6F, VEX_256_F3_0F_WIG, 0x7F}},
 	{AVMOVDQA, yvex_vmovdqa, Pvex, [23]uint8{VEX_128_66_0F_WIG, 0x6F, VEX_128_66_0F_WIG, 0x7F, VEX_256_66_0F_WIG, 0x6F, VEX_256_66_0F_WIG, 0x7F}},
@@ -3542,6 +3572,10 @@ func doasm(ctxt *obj.Link, p *obj.Prog) {
 				asmvex(ctxt, &p.From, p.From3, &p.To, o.op[z], o.op[z+1])
 				asmand(ctxt, p, &p.From, &p.To)
 
+			case Zvex_v_rm_r:
+				asmvex(ctxt, p.From3, &p.From, &p.To, o.op[z], o.op[z+1])
+				asmand(ctxt, p, p.From3, &p.To)
+
 			case Zvex_r_v_rm:
 				asmvex(ctxt, &p.To, p.From3, &p.From, o.op[z], o.op[z+1])
 				asmand(ctxt, p, &p.To, &p.From)