cmd/asm: add x86 POPCNTW, POPCNTL

Fixes #4816.

Change-Id: Ibeaa69f57b7519d56df4ea357edf8d9dc2102ffe
Reviewed-on: https://go-review.googlesource.com/18851
Reviewed-by: Rob Pike <[email protected]>
Run-TryBot: Russ Cox <[email protected]>
TryBot-Result: Gobot Gobot <[email protected]>

Author: rsc

src/cmd/asm/internal/asm/testdata/amd64enc.s

@@ -4145,22 +4145,22 @@ TEXT asmtest(SB),7,$0
 	POPQ DX                                 // 8fc2 or 5a
 	PUSHQ AX
 	POPQ R11                                // 418fc3 or 415b
-	//TODO: POPCNTW (BX), DX                // 66f30fb813
-	//TODO: POPCNTW (R11), DX               // 66f3410fb813
-	//TODO: POPCNTW DX, DX                  // 66f30fb8d2
-	//TODO: POPCNTW R11, DX                 // 66f3410fb8d3
-	//TODO: POPCNTW (BX), R11               // 66f3440fb81b
-	//TODO: POPCNTW (R11), R11              // 66f3450fb81b
-	//TODO: POPCNTW DX, R11                 // 66f3440fb8da
-	//TODO: POPCNTW R11, R11                // 66f3450fb8db
-	//TODO: POPCNTL (BX), DX                // f30fb813
-	//TODO: POPCNTL (R11), DX               // f3410fb813
-	//TODO: POPCNTL DX, DX                  // f30fb8d2
-	//TODO: POPCNTL R11, DX                 // f3410fb8d3
-	//TODO: POPCNTL (BX), R11               // f3440fb81b
-	//TODO: POPCNTL (R11), R11              // f3450fb81b
-	//TODO: POPCNTL DX, R11                 // f3440fb8da
-	//TODO: POPCNTL R11, R11                // f3450fb8db
+	POPCNTW (BX), DX                        // 66f30fb813
+	POPCNTW (R11), DX                       // 66f3410fb813
+	POPCNTW DX, DX                          // 66f30fb8d2
+	POPCNTW R11, DX                         // 66f3410fb8d3
+	POPCNTW (BX), R11                       // 66f3440fb81b
+	POPCNTW (R11), R11                      // 66f3450fb81b
+	POPCNTW DX, R11                         // 66f3440fb8da
+	POPCNTW R11, R11                        // 66f3450fb8db
+	POPCNTL (BX), DX                        // f30fb813
+	POPCNTL (R11), DX                       // f3410fb813
+	POPCNTL DX, DX                          // f30fb8d2
+	POPCNTL R11, DX                         // f3410fb8d3
+	POPCNTL (BX), R11                       // f3440fb81b
+	POPCNTL (R11), R11                      // f3450fb81b
+	POPCNTL DX, R11                         // f3440fb8da
+	POPCNTL R11, R11                        // f3450fb8db
 	POPCNTQ (BX), DX                        // f3480fb813
 	POPCNTQ (R11), DX                       // f3490fb813
 	POPCNTQ DX, DX                          // f3480fb8d2

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

@@ -219,6 +219,7 @@ const (
 	Pb    = 0xfe /* byte operands */
 	Pf2   = 0xf2 /* xmm escape 1: f2 0f */
 	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 */
 	Pfw   = 0xf4 /* Pf3 with Rex.w: f3 48 0f */
 	Pvex1 = 0xc5 /* 66.0f escape, vex encoding */
@@ -1208,6 +1209,8 @@ var optab =
 	{APMULULQ, ymm, Py1, [23]uint8{0xf4, Pe, 0xf4}},
 	{APOPAL, ynone, P32, [23]uint8{0x61}},
 	{APOPAW, ynone, Pe, [23]uint8{0x61}},
+	{APOPCNTW, yml_rl, Pef3, [23]uint8{0xb8}},
+	{APOPCNTL, yml_rl, Pf3, [23]uint8{0xb8}},
 	{APOPCNTQ, yml_rl, Pfw, [23]uint8{0xb8}},
 	{APOPFL, ynone, P32, [23]uint8{0x9d}},
 	{APOPFQ, ynone, Py, [23]uint8{0x9d}},
@@ -3204,6 +3207,14 @@ func doasm(ctxt *obj.Link, p *obj.Prog) {
 				ctxt.Andptr[0] = Pm
 				ctxt.Andptr = ctxt.Andptr[1:]
 
+			case Pef3:
+				ctxt.Andptr[0] = Pe
+				ctxt.Andptr = ctxt.Andptr[1:]
+				ctxt.Andptr[0] = Pf3
+				ctxt.Andptr = ctxt.Andptr[1:]
+				ctxt.Andptr[0] = Pm
+				ctxt.Andptr = ctxt.Andptr[1:]
+
 			case Pfw: /* xmm opcode escape + REX.W */
 				ctxt.Rexflag |= Pw
 				ctxt.Andptr[0] = Pf3