cmd/6a, cmd/8a, cmd/6l, cmd/8l: add AES instructions

Instructions for use in AES hashing.  See CL#7543043

R=rsc
CC=golang-dev
https://golang.org/cl/7548043

src/cmd/6a/lex.c

@@ -932,6 +932,8 @@ struct
 	"PCMPGTW",	LTYPE3,	APCMPGTW,
 	"PEXTRW",	LTYPEX,	APEXTRW,
 	"PINSRW",	LTYPEX,	APINSRW,
+	"PINSRD",	LTYPEX,	APINSRD,
+	"PINSRQ",	LTYPEX,	APINSRQ,
 	"PMADDWL",	LTYPE3,	APMADDWL,
 	"PMAXSW",	LTYPE3,	APMAXSW,
 	"PMAXUB",	LTYPE3,	APMAXUB,
@@ -949,6 +951,7 @@ struct
 	"PSHUFL",	LTYPEX,	APSHUFL,
 	"PSHUFLW",	LTYPEX,	APSHUFLW,
 	"PSHUFW",	LTYPEX, APSHUFW,
+	"PSHUFB",	LTYPEM, APSHUFB,
 	"PSLLO",	LTYPE3,	APSLLO,
 	"PSLLDQ",	LTYPE3,	APSLLO,	/* syn */
 	"PSLLL",	LTYPE3,	APSLLL,

src/cmd/6l/6.out.h

@@ -654,6 +654,8 @@ enum	as
 	APFSUB,
 	APFSUBR,
 	APINSRW,
+	APINSRD,
+	APINSRQ,
 	APMADDWL,
 	APMAXSW,
 	APMAXUB,
@@ -671,6 +673,7 @@ enum	as
 	APSHUFL,
 	APSHUFLW,
 	APSHUFW,
+	APSHUFB,
 	APSLLO,
 	APSLLL,
 	APSLLQ,

src/cmd/6l/l.h

@@ -263,6 +263,7 @@ enum
 	Zo_iw,
 	Zm_o,
 	Zm_r,
+	Zm2_r,
 	Zm_r_xm,
 	Zm_r_i_xm,
 	Zm_r_3d,
@@ -292,10 +293,11 @@ enum
 	P32		= 0x32,	/* 32-bit only */
 	Pe		= 0x66,	/* operand escape */
 	Pm		= 0x0f,	/* 2byte opcode escape */
-	Pq		= 0xff,	/* both escape */
+	Pq		= 0xff,	/* both escapes: 66 0f */
 	Pb		= 0xfe,	/* byte operands */
-	Pf2		= 0xf2,	/* xmm escape 1 */
-	Pf3		= 0xf3,	/* xmm escape 2 */
+	Pf2		= 0xf2,	/* xmm escape 1: f2 0f */
+	Pf3		= 0xf3,	/* xmm escape 2: f3 0f */
+	Pq3		= 0x67, /* xmm escape 3: 66 48 0f */
 	Pw		= 0x48,	/* Rex.w */
 	Py		= 0x80,	/* defaults to 64-bit mode */
 

src/cmd/6l/optab.c

@@ -522,6 +522,11 @@ uchar	ymshuf[] =
 	Ymm,	Ymr,	Zibm_r,	2,
 	0
 };
+uchar	ymshufb[] =
+{
+	Yxm,	Yxr,	Zm2_r,	2,
+	0
+};
 uchar	yxshuf[] =
 {
 	Yxm,	Yxr,	Zibm_r,	2,
@@ -532,6 +537,11 @@ uchar	yextrw[] =
 	Yxr,	Yrl,	Zibm_r,	2,
 	0
 };
+uchar	yinsr[] =
+{
+	Ymm,	Yxr,	Zibm_r,	3,
+	0
+};
 uchar	ypsdq[] =
 {
 	Yi8,	Yxr,	Zibo_m,	2,
@@ -983,6 +993,8 @@ Optab optab[] =
 	{ APFSUB,	ymfp,	Px, 0x9a },
 	{ APFSUBR,	ymfp,	Px, 0xaa },
 	{ APINSRW,	yextrw,	Pq, 0xc4,(00) },
+	{ APINSRD,	yinsr,	Pq, 0x3a, 0x22, (00) },
+	{ APINSRQ,	yinsr,	Pq3, 0x3a, 0x22, (00) },
 	{ APMADDWL,	ymm,	Py, 0xf5,Pe,0xf5 },
 	{ APMAXSW,	yxm,	Pe, 0xee },
 	{ APMAXUB,	yxm,	Pe, 0xde },
@@ -1008,6 +1020,7 @@ Optab optab[] =
 	{ APSHUFL,	yxshuf,	Pq, 0x70,(00) },
 	{ APSHUFLW,	yxshuf,	Pf2, 0x70,(00) },
 	{ APSHUFW,	ymshuf,	Pm, 0x70,(00) },
+	{ APSHUFB,	ymshufb,Pq, 0x38, 0x00 },
 	{ APSLLO,	ypsdq,	Pq, 0x73,(07) },
 	{ APSLLL,	yps,	Py, 0xf2, 0x72,(06), Pe,0xf2, Pe,0x72,(06) },
 	{ APSLLQ,	yps,	Py, 0xf3, 0x73,(06), Pe,0xf3, Pe,0x73,(06) },

src/cmd/6l/span.c

@@ -1161,6 +1161,11 @@ found:
 		*andptr++ = Pe;
 		*andptr++ = Pm;
 		break;
+	case Pq3:	/* 16 bit escape, Rex.w, and opcode escape */
+		*andptr++ = Pe;
+		*andptr++ = Pw;
+		*andptr++ = Pm;
+		break;
 
 	case Pf2:	/* xmm opcode escape */
 	case Pf3:
@@ -1229,6 +1234,11 @@ found:
 		*andptr++ = op;
 		asmand(&p->from, &p->to);
 		break;
+	case Zm2_r:
+		*andptr++ = op;
+		*andptr++ = o->op[z+1];
+		asmand(&p->from, &p->to);
+		break;
 
 	case Zm_r_xm:
 		mediaop(o, op, t[3], z);

src/cmd/8a/a.y

@@ -54,7 +54,7 @@
 %left	'*' '/' '%'
 %token	<lval>	LTYPE0 LTYPE1 LTYPE2 LTYPE3 LTYPE4
 %token	<lval>	LTYPEC LTYPED LTYPEN LTYPER LTYPET LTYPES LTYPEM LTYPEI LTYPEG LTYPEXC
-%token	<lval>	LCONST LFP LPC LSB
+%token	<lval>	LTYPEX LCONST LFP LPC LSB
 %token	<lval>	LBREG LLREG LSREG LFREG LXREG
 %token	<dval>	LFCONST
 %token	<sval>	LSCONST LSP
@@ -63,7 +63,7 @@
 %type	<con2>	con2
 %type	<gen>	mem imm imm2 reg nam rel rem rim rom omem nmem
 %type	<gen2>	nonnon nonrel nonrem rimnon rimrem remrim
-%type	<gen2>	spec1 spec2 spec3 spec4 spec5 spec6 spec7 spec8 spec9
+%type	<gen2>	spec1 spec2 spec3 spec4 spec5 spec6 spec7 spec8 spec9 spec10
 %%
 prog:
 |	prog
@@ -117,6 +117,7 @@ inst:
 |	LTYPEI spec7	{ outcode($1, &$2); }
 |	LTYPEG spec8	{ outcode($1, &$2); }
 |	LTYPEXC spec9	{ outcode($1, &$2); }
+|	LTYPEX spec10	{ outcode($1, &$2); }
 
 nonnon:
 	{
@@ -296,6 +297,16 @@ spec9:	/* CMPPS/CMPPD */
 		$$.to.offset = $5;
 	}
 
+spec10:	/* PINSRD */
+	imm ',' rem ',' reg
+	{
+		$$.from = $3;
+		$$.to = $5;
+		if($1.type != D_CONST)
+			yyerror("illegal constant");
+		$$.to.offset = $1.offset;
+	}
+
 rem:
 	reg
 |	mem

src/cmd/8a/lex.c

@@ -306,6 +306,7 @@ struct
 	"CMPXCHGB",	LTYPE3,	ACMPXCHGB,
 	"CMPXCHGL",	LTYPE3,	ACMPXCHGL,
 	"CMPXCHGW",	LTYPE3,	ACMPXCHGW,
+	"CPUID",	LTYPE0,	ACPUID,
 	"DAA",		LTYPE0,	ADAA,
 	"DAS",		LTYPE0,	ADAS,
 	"DATA",		LTYPED,	ADATA,
@@ -408,6 +409,7 @@ struct
 	"MOVB",		LTYPE3,	AMOVB,
 	"MOVL",		LTYPEM,	AMOVL,
 	"MOVW",		LTYPEM,	AMOVW,
+	"MOVQ",		LTYPEM, AMOVQ,
 	"MOVBLSX",	LTYPE3, AMOVBLSX,
 	"MOVBLZX",	LTYPE3, AMOVBLZX,
 	"MOVBWSX",	LTYPE3, AMOVBWSX,
@@ -437,6 +439,7 @@ struct
 	"OUTSL",	LTYPE0,	AOUTSL,
 	"OUTSW",	LTYPE0,	AOUTSW,
 	"PAUSE",	LTYPEN,	APAUSE,
+	"PINSRD",	LTYPEX,	APINSRD,
 	"POPAL",	LTYPE0,	APOPAL,
 	"POPAW",	LTYPE0,	APOPAW,
 	"POPFL",	LTYPE0,	APOPFL,
@@ -687,6 +690,7 @@ struct
 	"ADDPS",	LTYPE3,	AADDPS,
 	"ADDSD",	LTYPE3,	AADDSD,
 	"ADDSS",	LTYPE3,	AADDSS,
+	"AESENC",	LTYPE3,	AAESENC,
 	"ANDNPD",	LTYPE3,	AANDNPD,
 	"ANDNPS",	LTYPE3,	AANDNPS,
 	"ANDPD",	LTYPE3,	AANDPD,
@@ -755,11 +759,13 @@ struct
 	"ORPD",		LTYPE3,	AORPD,
 	"ORPS",		LTYPE3,	AORPS,
 	"PADDQ",	LTYPE3,	APADDQ,
+	"PAND",		LTYPE3,	APAND,
 	"PMAXSW",	LTYPE3,	APMAXSW,
 	"PMAXUB",	LTYPE3,	APMAXUB,
 	"PMINSW",	LTYPE3,	APMINSW,
 	"PMINUB",	LTYPE3,	APMINUB,
 	"PSADBW",	LTYPE3,	APSADBW,
+	"PSHUFB",	LTYPE3, APSHUFB,
 	"PSUBB",	LTYPE3,	APSUBB,
 	"PSUBL",	LTYPE3,	APSUBL,
 	"PSUBQ",	LTYPE3,	APSUBQ,

src/cmd/8a/y.tab.h

-/* A Bison parser, made by GNU Bison 2.6.5.  */
+/* A Bison parser, made by GNU Bison 2.5.  */
 
 /* Bison interface for Yacc-like parsers in C
    
-      Copyright (C) 1984, 1989-1990, 2000-2012 Free Software Foundation, Inc.
+      Copyright (C) 1984, 1989-1990, 2000-2011 Free Software Foundation, Inc.
    
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -30,15 +30,6 @@
    This special exception was added by the Free Software Foundation in
    version 2.2 of Bison.  */
 
-#ifndef YY_YY_Y_TAB_H_INCLUDED
-# define YY_YY_Y_TAB_H_INCLUDED
-/* Enabling traces.  */
-#ifndef YYDEBUG
-# define YYDEBUG 0
-#endif
-#if YYDEBUG
-extern int yydebug;
-#endif
 
 /* Tokens.  */
 #ifndef YYTOKENTYPE
@@ -61,21 +52,22 @@ extern int yydebug;
      LTYPEI = 270,
      LTYPEG = 271,
      LTYPEXC = 272,
-     LCONST = 273,
-     LFP = 274,
-     LPC = 275,
-     LSB = 276,
-     LBREG = 277,
-     LLREG = 278,
-     LSREG = 279,
-     LFREG = 280,
-     LXREG = 281,
-     LFCONST = 282,
-     LSCONST = 283,
-     LSP = 284,
-     LNAME = 285,
-     LLAB = 286,
-     LVAR = 287
+     LTYPEX = 273,
+     LCONST = 274,
+     LFP = 275,
+     LPC = 276,
+     LSB = 277,
+     LBREG = 278,
+     LLREG = 279,
+     LSREG = 280,
+     LFREG = 281,
+     LXREG = 282,
+     LFCONST = 283,
+     LSCONST = 284,
+     LSP = 285,
+     LNAME = 286,
+     LLAB = 287,
+     LVAR = 288
    };
 #endif
 /* Tokens.  */
@@ -94,28 +86,31 @@ extern int yydebug;
 #define LTYPEI 270
 #define LTYPEG 271
 #define LTYPEXC 272
-#define LCONST 273
-#define LFP 274
-#define LPC 275
-#define LSB 276
-#define LBREG 277
-#define LLREG 278
-#define LSREG 279
-#define LFREG 280
-#define LXREG 281
-#define LFCONST 282
-#define LSCONST 283
-#define LSP 284
-#define LNAME 285
-#define LLAB 286
-#define LVAR 287
+#define LTYPEX 273
+#define LCONST 274
+#define LFP 275
+#define LPC 276
+#define LSB 277
+#define LBREG 278
+#define LLREG 279
+#define LSREG 280
+#define LFREG 281
+#define LXREG 282
+#define LFCONST 283
+#define LSCONST 284
+#define LSP 285
+#define LNAME 286
+#define LLAB 287
+#define LVAR 288
+
 
 
 
 #if ! defined YYSTYPE && ! defined YYSTYPE_IS_DECLARED
 typedef union YYSTYPE
 {
-/* Line 2042 of yacc.c  */
+
+/* Line 2068 of yacc.c  */
 #line 37 "a.y"
 
 	Sym	*sym;
@@ -130,8 +125,9 @@ typedef union YYSTYPE
 	Gen2	gen2;
 
 
-/* Line 2042 of yacc.c  */
-#line 135 "y.tab.h"
+
+/* Line 2068 of yacc.c  */
+#line 131 "y.tab.h"
 } YYSTYPE;
 # define YYSTYPE_IS_TRIVIAL 1
 # define yystype YYSTYPE /* obsolescent; will be withdrawn */
@@ -140,18 +136,4 @@ typedef union YYSTYPE
 
 extern YYSTYPE yylval;
 
-#ifdef YYPARSE_PARAM
-#if defined __STDC__ || defined __cplusplus
-int yyparse (void *YYPARSE_PARAM);
-#else
-int yyparse ();
-#endif
-#else /* ! YYPARSE_PARAM */
-#if defined __STDC__ || defined __cplusplus
-int yyparse (void);
-#else
-int yyparse ();
-#endif
-#endif /* ! YYPARSE_PARAM */
 
-#endif /* !YY_YY_Y_TAB_H_INCLUDED  */

src/cmd/8l/8.out.h

@@ -153,6 +153,7 @@ enum	as
 	AMOVB,
 	AMOVL,
 	AMOVW,
+	AMOVQ,
 	AMOVBLSX,
 	AMOVBLZX,
 	AMOVBWSX,
@@ -397,6 +398,7 @@ enum	as
 	ACMPXCHGW,
 	ACMPXCHG8B,
 
+	ACPUID,
 	ARDTSC,
 
 	AXADDB,
@@ -531,6 +533,7 @@ enum	as
 	AORPD,
 	AORPS,
 	APADDQ,
+	APAND,
 	APMAXSW,
 	APMAXUB,
 	APMINSW,
@@ -566,7 +569,12 @@ enum	as
 	AUNPCKLPS,
 	AXORPD,
 	AXORPS,
-	
+
+	/* SSE 3+ */
+	AAESENC,
+	APINSRD,
+	APSHUFB,
+
 	AUSEFIELD,
 	ALOCALS,
 	ATYPE,

src/cmd/8l/l.h

@@ -172,7 +172,7 @@ struct	Optab
 	short	as;
 	uchar*	ytab;
 	uchar	prefix;
-	uchar	op[10];
+	uchar	op[12];
 };
 
 enum
@@ -217,6 +217,7 @@ enum
 	Zxxx		= 0,
 
 	Zlit,
+	Zlitm_r,
 	Z_rp,
 	Zbr,
 	Zcall,
@@ -233,6 +234,7 @@ enum
 	Zloop,
 	Zm_o,
 	Zm_r,
+	Zm2_r,
 	Zm_r_xm,
 	Zm_r_i_xm,
 	Zaut_r,
@@ -249,6 +251,7 @@ enum
 	Zib_rr,
 	Zil_rr,
 	Zclr,
+	Zibm_r,	/* mmx1,mmx2/mem64,imm8 */
 	Zbyte,
 	Zmov,
 	Zmax,

src/cmd/8l/optab.c

@@ -150,9 +150,16 @@ uchar	ymovl[] =
 //	Yi0,	Yml,	Zibo_m,	2,	// shorter but slower AND $0,dst
 	Yi32,	Yrl,	Zil_rp,	1,
 	Yi32,	Yml,	Zilo_m,	2,
+	Yml,	Yxr,	Zm_r_xm,	2,	// XMM MOVD (32 bit)
+	Yxr,	Yml,	Zr_m_xm,	2,	// XMM MOVD (32 bit)
 	Yiauto,	Yrl,	Zaut_r,	2,
 	0
 };
+uchar	ymovq[] =
+{
+	Yml,	Yxr,	Zm_r_xm,	2,
+	0
+};
 uchar	ym_rl[] =
 {
 	Ym,	Yrl,	Zm_r,	1,
@@ -435,6 +442,21 @@ uchar	yprefetch[] =
 	Ym,	Ynone,	Zm_o,	2,
 	0,
 };
+uchar	yaes[] =
+{
+	Yxm,	Yxr,	Zlitm_r,	2,
+	0
+};
+uchar	yinsrd[] =
+{
+	Yml,	Yxr,	Zibm_r,	2,
+	0
+};
+uchar	ymshufb[] =
+{
+	Yxm,	Yxr,	Zm2_r,	2,
+	0
+};
 
 Optab optab[] =
 /*	as, ytab, andproto, opcode */
@@ -552,8 +574,9 @@ Optab optab[] =
 	{ ALSLL,	yml_rl,	Pm, 0x03  },
 	{ ALSLW,	yml_rl,	Pq, 0x03  },
 	{ AMOVB,	ymovb,	Pb, 0x88,0x8a,0xb0,0xc6,(00) },
-	{ AMOVL,	ymovl,	Px, 0x89,0x8b,0x31,0x83,(04),0xb8,0xc7,(00) },
+	{ AMOVL,	ymovl,	Px, 0x89,0x8b,0x31,0x83,(04),0xb8,0xc7,(00),Pe,0x6e,Pe,0x7e },
 	{ AMOVW,	ymovl,	Pe, 0x89,0x8b,0x31,0x83,(04),0xb8,0xc7,(00) },
+	{ AMOVQ,	ymovq,	Pf3, 0x7e },
 	{ AMOVBLSX,	ymb_rl,	Pm, 0xbe },
 	{ AMOVBLZX,	ymb_rl,	Pm, 0xb6 },
 	{ AMOVBWSX,	ymb_rl,	Pq, 0xbe },
@@ -793,6 +816,7 @@ Optab optab[] =
 	{ ACMPXCHGW,	yrl_ml,	Pm, 0xb1 },
 	{ ACMPXCHG8B,	yscond,	Pm, 0xc7,(01) },
 
+	{ ACPUID,	ynone,	Pm, 0xa2 },
 	{ ARDTSC,	ynone,	Pm, 0x31 },
 
 	{ AXADDB,	yrb_mb,	Pb, 0x0f,0xc0 },
@@ -925,6 +949,7 @@ Optab optab[] =
 	{ AORPD,	yxm,	Pq, 0x56 },
 	{ AORPS,	yxm,	Pm, 0x56 },
 	{ APADDQ,	yxm,	Pe, 0xd4 },
+	{ APAND,	yxm,	Pe, 0xdb },
 	{ APMAXSW,	yxm,	Pe, 0xee },
 	{ APMAXUB,	yxm,	Pe, 0xde },
 	{ APMINSW,	yxm,	Pe, 0xea },
@@ -961,6 +986,10 @@ Optab optab[] =
 	{ AXORPD,	yxm,	Pe, 0x57 },
 	{ AXORPS,	yxm,	Pm, 0x57 },
 
+	{ AAESENC,	yaes,	Pq, 0x38,0xdc,(0) },
+	{ APINSRD,	yinsrd,	Pq, 0x3a, 0x22, (00) },
+	{ APSHUFB,	ymshufb,Pq, 0x38, 0x00 },
+
 	{ AUSEFIELD,	ynop,	Px, 0,0 },
 	{ ALOCALS },
 	{ ATYPE },

src/cmd/8l/span.c

@@ -1003,11 +1003,23 @@ found:
 			*andptr++ = op;
 		break;
 
+	case Zlitm_r:
+		for(; op = o->op[z]; z++)
+			*andptr++ = op;
+		asmand(&p->from, reg[p->to.type]);
+		break;
+
 	case Zm_r:
 		*andptr++ = op;
 		asmand(&p->from, reg[p->to.type]);
 		break;
 
+	case Zm2_r:
+		*andptr++ = op;
+		*andptr++ = o->op[z+1];
+		asmand(&p->from, reg[p->to.type]);
+		break;
+
 	case Zm_r_xm:
 		mediaop(o, op, t[3], z);
 		asmand(&p->from, reg[p->to.type]);
@@ -1019,6 +1031,13 @@ found:
 		*andptr++ = p->to.offset;
 		break;
 
+	case Zibm_r:
+		while ((op = o->op[z++]) != 0)
+			*andptr++ = op;
+		asmand(&p->from, reg[p->to.type]);
+		*andptr++ = p->to.offset;
+		break;
+
 	case Zaut_r:
 		*andptr++ = 0x8d;	/* leal */
 		if(p->from.type != D_ADDR)