view gcc/config/i386/i386.md @ 69:1b10fe6932e1

merge 69
author Nobuyasu Oshiro <dimolto@cr.ie.u-ryukyu.ac.jp>
date Sun, 21 Aug 2011 07:53:12 +0900
parents 326d9e06c2e3 f6334be47118
children ab0bcb71f44d
line wrap: on
line source

;; GCC machine description for IA-32 and x86-64.
;; Copyright (C) 1988, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
;; 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011
;; Free Software Foundation, Inc.
;; Mostly by William Schelter.
;; x86_64 support added by Jan Hubicka
;;
;; This file is part of GCC.
;;
;; GCC is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation; either version 3, or (at your option)
;; any later version.
;;
;; GCC is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;; GNU General Public License for more details.
;;
;; You should have received a copy of the GNU General Public License
;; along with GCC; see the file COPYING3.  If not see
;; <http://www.gnu.org/licenses/>.  */
;;
;; The original PO technology requires these to be ordered by speed,
;; so that assigner will pick the fastest.
;;
;; See file "rtl.def" for documentation on define_insn, match_*, et. al.
;;
;; The special asm out single letter directives following a '%' are:
;; L,W,B,Q,S,T -- print the opcode suffix for specified size of operand.
;; C -- print opcode suffix for set/cmov insn.
;; c -- like C, but print reversed condition
;; F,f -- likewise, but for floating-point.
;; O -- if HAVE_AS_IX86_CMOV_SUN_SYNTAX, expand to "w.", "l." or "q.",
;;      otherwise nothing
;; R -- print the prefix for register names.
;; z -- print the opcode suffix for the size of the current operand.
;; Z -- likewise, with special suffixes for x87 instructions.
;; * -- print a star (in certain assembler syntax)
;; A -- print an absolute memory reference.
;; w -- print the operand as if it's a "word" (HImode) even if it isn't.
;; s -- print a shift double count, followed by the assemblers argument
;;	delimiter.
;; b -- print the QImode name of the register for the indicated operand.
;;	%b0 would print %al if operands[0] is reg 0.
;; w --  likewise, print the HImode name of the register.
;; k --  likewise, print the SImode name of the register.
;; q --  likewise, print the DImode name of the register.
;; x --  likewise, print the V4SFmode name of the register.
;; t --  likewise, print the V8SFmode name of the register.
;; h -- print the QImode name for a "high" register, either ah, bh, ch or dh.
;; y -- print "st(0)" instead of "st" as a register.
;; d -- print duplicated register operand for AVX instruction.
;; D -- print condition for SSE cmp instruction.
;; P -- if PIC, print an @PLT suffix.
;; X -- don't print any sort of PIC '@' suffix for a symbol.
;; & -- print some in-use local-dynamic symbol name.
;; H -- print a memory address offset by 8; used for sse high-parts
;; Y -- print condition for XOP pcom* instruction.
;; + -- print a branch hint as 'cs' or 'ds' prefix
;; ; -- print a semicolon (after prefixes due to bug in older gas).
;; @ -- print a segment register of thread base pointer load

;; UNSPEC usage:

(define_c_enum "unspec" [
  ;; Relocation specifiers
  UNSPEC_GOT
  UNSPEC_GOTOFF
  UNSPEC_GOTPCREL
  UNSPEC_GOTTPOFF
  UNSPEC_TPOFF
  UNSPEC_NTPOFF
  UNSPEC_DTPOFF
  UNSPEC_GOTNTPOFF
  UNSPEC_INDNTPOFF
  UNSPEC_PLTOFF
  UNSPEC_MACHOPIC_OFFSET
  UNSPEC_PCREL

  ;; Prologue support
  UNSPEC_STACK_ALLOC
  UNSPEC_SET_GOT
  UNSPEC_REG_SAVE
  UNSPEC_DEF_CFA
  UNSPEC_SET_RIP
  UNSPEC_SET_GOT_OFFSET
  UNSPEC_MEMORY_BLOCKAGE
  UNSPEC_STACK_CHECK

  ;; TLS support
  UNSPEC_TP
  UNSPEC_TLS_GD
  UNSPEC_TLS_LD_BASE
  UNSPEC_TLSDESC
  UNSPEC_TLS_IE_SUN

  ;; Other random patterns
  UNSPEC_SCAS
  UNSPEC_FNSTSW
  UNSPEC_SAHF
  UNSPEC_PARITY
  UNSPEC_FSTCW
  UNSPEC_ADD_CARRY
  UNSPEC_FLDCW
  UNSPEC_REP
  UNSPEC_LD_MPIC	; load_macho_picbase
  UNSPEC_TRUNC_NOOP
  UNSPEC_DIV_ALREADY_SPLIT
  UNSPEC_CALL_NEEDS_VZEROUPPER

  ;; For SSE/MMX support:
  UNSPEC_FIX_NOTRUNC
  UNSPEC_MASKMOV
  UNSPEC_MOVMSK
  UNSPEC_MOVNT
  UNSPEC_MOVU
  UNSPEC_RCP
  UNSPEC_RSQRT
  UNSPEC_SFENCE
  UNSPEC_PFRCP
  UNSPEC_PFRCPIT1
  UNSPEC_PFRCPIT2
  UNSPEC_PFRSQRT
  UNSPEC_PFRSQIT1
  UNSPEC_MFENCE
  UNSPEC_LFENCE
  UNSPEC_PSADBW
  UNSPEC_LDDQU
  UNSPEC_MS_TO_SYSV_CALL

  ;; Generic math support
  UNSPEC_COPYSIGN
  UNSPEC_IEEE_MIN	; not commutative
  UNSPEC_IEEE_MAX	; not commutative

  ;; x87 Floating point
  UNSPEC_SIN
  UNSPEC_COS
  UNSPEC_FPATAN
  UNSPEC_FYL2X
  UNSPEC_FYL2XP1
  UNSPEC_FRNDINT
  UNSPEC_FIST
  UNSPEC_F2XM1
  UNSPEC_TAN
  UNSPEC_FXAM

  ;; x87 Rounding
  UNSPEC_FRNDINT_FLOOR
  UNSPEC_FRNDINT_CEIL
  UNSPEC_FRNDINT_TRUNC
  UNSPEC_FRNDINT_MASK_PM
  UNSPEC_FIST_FLOOR
  UNSPEC_FIST_CEIL

  ;; x87 Double output FP
  UNSPEC_SINCOS_COS
  UNSPEC_SINCOS_SIN
  UNSPEC_XTRACT_FRACT
  UNSPEC_XTRACT_EXP
  UNSPEC_FSCALE_FRACT
  UNSPEC_FSCALE_EXP
  UNSPEC_FPREM_F
  UNSPEC_FPREM_U
  UNSPEC_FPREM1_F
  UNSPEC_FPREM1_U

  UNSPEC_C2_FLAG
  UNSPEC_FXAM_MEM

  ;; SSP patterns
  UNSPEC_SP_SET
  UNSPEC_SP_TEST
  UNSPEC_SP_TLS_SET
  UNSPEC_SP_TLS_TEST

  ;; SSSE3
  UNSPEC_PSHUFB
  UNSPEC_PSIGN
  UNSPEC_PALIGNR

  ;; For SSE4A support
  UNSPEC_EXTRQI
  UNSPEC_EXTRQ
  UNSPEC_INSERTQI
  UNSPEC_INSERTQ

  ;; For SSE4.1 support
  UNSPEC_BLENDV
  UNSPEC_INSERTPS
  UNSPEC_DP
  UNSPEC_MOVNTDQA
  UNSPEC_MPSADBW
  UNSPEC_PHMINPOSUW
  UNSPEC_PTEST
  UNSPEC_ROUND

  ;; For SSE4.2 support
  UNSPEC_CRC32
  UNSPEC_PCMPESTR
  UNSPEC_PCMPISTR

  ;; For FMA4 support
  UNSPEC_FMADDSUB
  UNSPEC_XOP_UNSIGNED_CMP
  UNSPEC_XOP_TRUEFALSE
  UNSPEC_XOP_PERMUTE
  UNSPEC_FRCZ

  ;; For AES support
  UNSPEC_AESENC
  UNSPEC_AESENCLAST
  UNSPEC_AESDEC
  UNSPEC_AESDECLAST
  UNSPEC_AESIMC
  UNSPEC_AESKEYGENASSIST

  ;; For PCLMUL support
  UNSPEC_PCLMUL

  ;; For AVX support
  UNSPEC_PCMP
  UNSPEC_VPERMIL
  UNSPEC_VPERMIL2
  UNSPEC_VPERMIL2F128
  UNSPEC_MASKLOAD
  UNSPEC_MASKSTORE
  UNSPEC_CAST
  UNSPEC_VTESTP
  UNSPEC_VCVTPH2PS
  UNSPEC_VCVTPS2PH

  ;; For BMI support
  UNSPEC_BEXTR

  ;; For RDRAND support
  UNSPEC_RDRAND
])

(define_c_enum "unspecv" [
  UNSPECV_BLOCKAGE
  UNSPECV_STACK_PROBE
  UNSPECV_PROBE_STACK_RANGE
  UNSPECV_EMMS
  UNSPECV_LDMXCSR
  UNSPECV_STMXCSR
  UNSPECV_FEMMS
  UNSPECV_CLFLUSH
  UNSPECV_ALIGN
  UNSPECV_MONITOR
  UNSPECV_MWAIT
  UNSPECV_CMPXCHG
  UNSPECV_XCHG
  UNSPECV_LOCK
  UNSPECV_PROLOGUE_USE
  UNSPECV_CLD
  UNSPECV_NOPS
  UNSPECV_VZEROALL
  UNSPECV_VZEROUPPER
  UNSPECV_RDTSC
  UNSPECV_RDTSCP
  UNSPECV_RDPMC
  UNSPECV_LLWP_INTRINSIC
  UNSPECV_SLWP_INTRINSIC
  UNSPECV_LWPVAL_INTRINSIC
  UNSPECV_LWPINS_INTRINSIC
  UNSPECV_RDFSBASE
  UNSPECV_RDGSBASE
  UNSPECV_WRFSBASE
  UNSPECV_WRGSBASE
  UNSPECV_SPLIT_STACK_RETURN
])

;; Constants to represent pcomtrue/pcomfalse variants
(define_constants
  [(PCOM_FALSE			0)
   (PCOM_TRUE			1)
   (COM_FALSE_S			2)
   (COM_FALSE_P			3)
   (COM_TRUE_S			4)
   (COM_TRUE_P			5)
  ])

;; Constants used in the XOP pperm instruction
(define_constants
  [(PPERM_SRC			0x00)	/* copy source */
   (PPERM_INVERT		0x20)	/* invert source */
   (PPERM_REVERSE		0x40)	/* bit reverse source */
   (PPERM_REV_INV		0x60)	/* bit reverse & invert src */
   (PPERM_ZERO			0x80)	/* all 0's */
   (PPERM_ONES			0xa0)	/* all 1's */
   (PPERM_SIGN			0xc0)	/* propagate sign bit */
   (PPERM_INV_SIGN		0xe0)	/* invert & propagate sign */
   (PPERM_SRC1			0x00)	/* use first source byte */
   (PPERM_SRC2			0x10)	/* use second source byte */
   ])

;; Registers by name.
(define_constants
  [(AX_REG			 0)
   (DX_REG			 1)
   (CX_REG			 2)
   (BX_REG			 3)
   (SI_REG			 4)
   (DI_REG			 5)
   (BP_REG			 6)
   (SP_REG			 7)
   (ST0_REG			 8)
   (ST1_REG			 9)
   (ST2_REG			10)
   (ST3_REG			11)
   (ST4_REG			12)
   (ST5_REG			13)
   (ST6_REG			14)
   (ST7_REG			15)
   (FLAGS_REG			17)
   (FPSR_REG			18)
   (FPCR_REG			19)
   (XMM0_REG			21)
   (XMM1_REG			22)
   (XMM2_REG			23)
   (XMM3_REG			24)
   (XMM4_REG			25)
   (XMM5_REG			26)
   (XMM6_REG			27)
   (XMM7_REG			28)
   (MM0_REG			29)
   (MM1_REG			30)
   (MM2_REG			31)
   (MM3_REG			32)
   (MM4_REG			33)
   (MM5_REG			34)
   (MM6_REG			35)
   (MM7_REG			36)
   (R8_REG			37)
   (R9_REG			38)
   (R10_REG			39)
   (R11_REG			40)
   (R12_REG			41)
   (R13_REG			42)
   (XMM8_REG			45)
   (XMM9_REG			46)
   (XMM10_REG			47)
   (XMM11_REG			48)
   (XMM12_REG			49)
   (XMM13_REG			50)
   (XMM14_REG			51)
   (XMM15_REG			52)
  ])

;; Insns whose names begin with "x86_" are emitted by gen_FOO calls
;; from i386.c.

;; In C guard expressions, put expressions which may be compile-time
;; constants first.  This allows for better optimization.  For
;; example, write "TARGET_64BIT && reload_completed", not
;; "reload_completed && TARGET_64BIT".


;; Processor type.
(define_attr "cpu" "none,pentium,pentiumpro,geode,k6,athlon,k8,core2,corei7,
		    atom,generic64,amdfam10,bdver1,btver1"
  (const (symbol_ref "ix86_schedule")))

;; A basic instruction type.  Refinements due to arguments to be
;; provided in other attributes.
(define_attr "type"
  "other,multi,
   alu,alu1,negnot,imov,imovx,lea,
   incdec,ishift,ishift1,rotate,rotate1,imul,idiv,
   icmp,test,ibr,setcc,icmov,
   push,pop,call,callv,leave,
   str,bitmanip,
   fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,fisttp,frndint,
   sselog,sselog1,sseiadd,sseiadd1,sseishft,sseishft1,sseimul,
   sse,ssemov,sseadd,ssemul,ssecmp,ssecomi,ssecvt,ssecvt1,sseicvt,ssediv,sseins,
   ssemuladd,sse4arg,lwp,
   mmx,mmxmov,mmxadd,mmxmul,mmxcmp,mmxcvt,mmxshft"
  (const_string "other"))

;; Main data type used by the insn
(define_attr "mode"
  "unknown,none,QI,HI,SI,DI,TI,OI,SF,DF,XF,TF,V8SF,V4DF,V4SF,V2DF,V2SF,V1DF"
  (const_string "unknown"))

;; The CPU unit operations uses.
(define_attr "unit" "integer,i387,sse,mmx,unknown"
  (cond [(eq_attr "type" "fmov,fop,fsgn,fmul,fdiv,fpspc,fcmov,fcmp,fxch,fistp,fisttp,frndint")
	   (const_string "i387")
	 (eq_attr "type" "sselog,sselog1,sseiadd,sseiadd1,sseishft,sseishft1,sseimul,
			  sse,ssemov,sseadd,ssemul,ssecmp,ssecomi,ssecvt,
			  ssecvt1,sseicvt,ssediv,sseins,ssemuladd,sse4arg")
	   (const_string "sse")
	 (eq_attr "type" "mmx,mmxmov,mmxadd,mmxmul,mmxcmp,mmxcvt,mmxshft")
	   (const_string "mmx")
	 (eq_attr "type" "other")
	   (const_string "unknown")]
	 (const_string "integer")))

;; The (bounding maximum) length of an instruction immediate.
(define_attr "length_immediate" ""
  (cond [(eq_attr "type" "incdec,setcc,icmov,str,lea,other,multi,idiv,leave,
                          bitmanip")
	   (const_int 0)
	 (eq_attr "unit" "i387,sse,mmx")
	   (const_int 0)
	 (eq_attr "type" "alu,alu1,negnot,imovx,ishift,rotate,ishift1,rotate1,
			  imul,icmp,push,pop")
	   (symbol_ref "ix86_attr_length_immediate_default(insn,1)")
	 (eq_attr "type" "imov,test")
	   (symbol_ref "ix86_attr_length_immediate_default(insn,0)")
	 (eq_attr "type" "call")
	   (if_then_else (match_operand 0 "constant_call_address_operand" "")
	     (const_int 4)
	     (const_int 0))
	 (eq_attr "type" "callv")
	   (if_then_else (match_operand 1 "constant_call_address_operand" "")
	     (const_int 4)
	     (const_int 0))
	 ;; We don't know the size before shorten_branches.  Expect
	 ;; the instruction to fit for better scheduling.
	 (eq_attr "type" "ibr")
	   (const_int 1)
	 ]
	 (symbol_ref "/* Update immediate_length and other attributes! */
		      gcc_unreachable (),1")))

;; The (bounding maximum) length of an instruction address.
(define_attr "length_address" ""
  (cond [(eq_attr "type" "str,other,multi,fxch")
	   (const_int 0)
	 (and (eq_attr "type" "call")
	      (match_operand 0 "constant_call_address_operand" ""))
	     (const_int 0)
	 (and (eq_attr "type" "callv")
	      (match_operand 1 "constant_call_address_operand" ""))
	     (const_int 0)
	 ]
	 (symbol_ref "ix86_attr_length_address_default (insn)")))

;; Set when length prefix is used.
(define_attr "prefix_data16" ""
  (cond [(eq_attr "type" "ssemuladd,sse4arg,sseiadd1,ssecvt1")
	   (const_int 0)
	 (eq_attr "mode" "HI")
	   (const_int 1)
	 (and (eq_attr "unit" "sse") (eq_attr "mode" "V2DF,TI"))
	   (const_int 1)
	]
	(const_int 0)))

;; Set when string REP prefix is used.
(define_attr "prefix_rep" ""
  (cond [(eq_attr "type" "ssemuladd,sse4arg,sseiadd1,ssecvt1")
	   (const_int 0)
	 (and (eq_attr "unit" "sse") (eq_attr "mode" "SF,DF"))
	   (const_int 1)
	]
	(const_int 0)))

;; Set when 0f opcode prefix is used.
(define_attr "prefix_0f" ""
  (if_then_else
    (ior (eq_attr "type" "imovx,setcc,icmov,bitmanip")
	 (eq_attr "unit" "sse,mmx"))
    (const_int 1)
    (const_int 0)))

;; Set when REX opcode prefix is used.
(define_attr "prefix_rex" ""
  (cond [(eq (symbol_ref "TARGET_64BIT") (const_int 0))
	   (const_int 0)
	 (and (eq_attr "mode" "DI")
	      (and (eq_attr "type" "!push,pop,call,callv,leave,ibr")
		   (eq_attr "unit" "!mmx")))
	   (const_int 1)
	 (and (eq_attr "mode" "QI")
	      (ne (symbol_ref "x86_extended_QIreg_mentioned_p (insn)")
		  (const_int 0)))
	   (const_int 1)
	 (ne (symbol_ref "x86_extended_reg_mentioned_p (insn)")
	     (const_int 0))
	   (const_int 1)
	 (and (eq_attr "type" "imovx")
	      (match_operand:QI 1 "ext_QIreg_operand" ""))
	   (const_int 1)
	]
	(const_int 0)))

;; There are also additional prefixes in 3DNOW, SSSE3.
;; ssemuladd,sse4arg default to 0f24/0f25 and DREX byte,
;; sseiadd1,ssecvt1 to 0f7a with no DREX byte.
;; 3DNOW has 0f0f prefix, SSSE3 and SSE4_{1,2} 0f38/0f3a.
(define_attr "prefix_extra" ""
  (cond [(eq_attr "type" "ssemuladd,sse4arg")
	   (const_int 2)
	 (eq_attr "type" "sseiadd1,ssecvt1")
	   (const_int 1)
	]
	(const_int 0)))

;; Prefix used: original, VEX or maybe VEX.
(define_attr "prefix" "orig,vex,maybe_vex"
  (if_then_else (eq_attr "mode" "OI,V8SF,V4DF")
    (const_string "vex")
    (const_string "orig")))

;; VEX W bit is used.
(define_attr "prefix_vex_w" "" (const_int 0))

;; The length of VEX prefix
;; Only instructions with 0f prefix can have 2 byte VEX prefix,
;; 0f38/0f3a prefixes can't.  In i386.md 0f3[8a] is
;; still prefix_0f 1, with prefix_extra 1.
(define_attr "length_vex" ""
  (if_then_else (and (eq_attr "prefix_0f" "1")
		     (eq_attr "prefix_extra" "0"))
    (if_then_else (eq_attr "prefix_vex_w" "1")
      (symbol_ref "ix86_attr_length_vex_default (insn, 1, 1)")
      (symbol_ref "ix86_attr_length_vex_default (insn, 1, 0)"))
    (if_then_else (eq_attr "prefix_vex_w" "1")
      (symbol_ref "ix86_attr_length_vex_default (insn, 0, 1)")
      (symbol_ref "ix86_attr_length_vex_default (insn, 0, 0)"))))

;; Set when modrm byte is used.
(define_attr "modrm" ""
  (cond [(eq_attr "type" "str,leave")
	   (const_int 0)
	 (eq_attr "unit" "i387")
	   (const_int 0)
         (and (eq_attr "type" "incdec")
	      (and (eq (symbol_ref "TARGET_64BIT") (const_int 0))
		   (ior (match_operand:SI 1 "register_operand" "")
			(match_operand:HI 1 "register_operand" ""))))
	   (const_int 0)
	 (and (eq_attr "type" "push")
	      (not (match_operand 1 "memory_operand" "")))
	   (const_int 0)
	 (and (eq_attr "type" "pop")
	      (not (match_operand 0 "memory_operand" "")))
	   (const_int 0)
	 (and (eq_attr "type" "imov")
	      (and (not (eq_attr "mode" "DI"))
		   (ior (and (match_operand 0 "register_operand" "")
			     (match_operand 1 "immediate_operand" ""))
		        (ior (and (match_operand 0 "ax_reg_operand" "")
				  (match_operand 1 "memory_displacement_only_operand" ""))
			     (and (match_operand 0 "memory_displacement_only_operand" "")
				  (match_operand 1 "ax_reg_operand" ""))))))
	   (const_int 0)
	 (and (eq_attr "type" "call")
	      (match_operand 0 "constant_call_address_operand" ""))
	     (const_int 0)
	 (and (eq_attr "type" "callv")
	      (match_operand 1 "constant_call_address_operand" ""))
	     (const_int 0)
	 (and (eq_attr "type" "alu,alu1,icmp,test")
	      (match_operand 0 "ax_reg_operand" ""))
	     (symbol_ref "(get_attr_length_immediate (insn) <= (get_attr_mode (insn) != MODE_QI))")
	 ]
	 (const_int 1)))

;; The (bounding maximum) length of an instruction in bytes.
;; ??? fistp and frndint are in fact fldcw/{fistp,frndint}/fldcw sequences.
;; Later we may want to split them and compute proper length as for
;; other insns.
(define_attr "length" ""
  (cond [(eq_attr "type" "other,multi,fistp,frndint")
	   (const_int 16)
	 (eq_attr "type" "fcmp")
	   (const_int 4)
	 (eq_attr "unit" "i387")
	   (plus (const_int 2)
		 (plus (attr "prefix_data16")
		       (attr "length_address")))
	 (ior (eq_attr "prefix" "vex")
	      (and (eq_attr "prefix" "maybe_vex")
		    (ne (symbol_ref "TARGET_AVX") (const_int 0))))
	   (plus (attr "length_vex")
		 (plus (attr "length_immediate")
		       (plus (attr "modrm")
			     (attr "length_address"))))]
	 (plus (plus (attr "modrm")
		     (plus (attr "prefix_0f")
			   (plus (attr "prefix_rex")
				 (plus (attr "prefix_extra")
				       (const_int 1)))))
	       (plus (attr "prefix_rep")
		     (plus (attr "prefix_data16")
			   (plus (attr "length_immediate")
				 (attr "length_address")))))))

;; The `memory' attribute is `none' if no memory is referenced, `load' or
;; `store' if there is a simple memory reference therein, or `unknown'
;; if the instruction is complex.

(define_attr "memory" "none,load,store,both,unknown"
  (cond [(eq_attr "type" "other,multi,str,lwp")
	   (const_string "unknown")
	 (eq_attr "type" "lea,fcmov,fpspc")
	   (const_string "none")
	 (eq_attr "type" "fistp,leave")
	   (const_string "both")
	 (eq_attr "type" "frndint")
	   (const_string "load")
	 (eq_attr "type" "push")
	   (if_then_else (match_operand 1 "memory_operand" "")
	     (const_string "both")
	     (const_string "store"))
	 (eq_attr "type" "pop")
	   (if_then_else (match_operand 0 "memory_operand" "")
	     (const_string "both")
	     (const_string "load"))
	 (eq_attr "type" "setcc")
	   (if_then_else (match_operand 0 "memory_operand" "")
	     (const_string "store")
	     (const_string "none"))
	 (eq_attr "type" "icmp,test,ssecmp,ssecomi,mmxcmp,fcmp")
	   (if_then_else (ior (match_operand 0 "memory_operand" "")
			      (match_operand 1 "memory_operand" ""))
	     (const_string "load")
	     (const_string "none"))
	 (eq_attr "type" "ibr")
	   (if_then_else (match_operand 0 "memory_operand" "")
	     (const_string "load")
	     (const_string "none"))
	 (eq_attr "type" "call")
	   (if_then_else (match_operand 0 "constant_call_address_operand" "")
	     (const_string "none")
	     (const_string "load"))
	 (eq_attr "type" "callv")
	   (if_then_else (match_operand 1 "constant_call_address_operand" "")
	     (const_string "none")
	     (const_string "load"))
	 (and (eq_attr "type" "alu1,negnot,ishift1,sselog1")
	      (match_operand 1 "memory_operand" ""))
	   (const_string "both")
	 (and (match_operand 0 "memory_operand" "")
	      (match_operand 1 "memory_operand" ""))
	   (const_string "both")
	 (match_operand 0 "memory_operand" "")
	   (const_string "store")
	 (match_operand 1 "memory_operand" "")
	   (const_string "load")
	 (and (eq_attr "type"
		 "!alu1,negnot,ishift1,
		   imov,imovx,icmp,test,bitmanip,
		   fmov,fcmp,fsgn,
		   sse,ssemov,ssecmp,ssecomi,ssecvt,ssecvt1,sseicvt,sselog1,
		   sseiadd1,mmx,mmxmov,mmxcmp,mmxcvt")
	      (match_operand 2 "memory_operand" ""))
	   (const_string "load")
	 (and (eq_attr "type" "icmov,ssemuladd,sse4arg")
	      (match_operand 3 "memory_operand" ""))
	   (const_string "load")
	]
	(const_string "none")))

;; Indicates if an instruction has both an immediate and a displacement.

(define_attr "imm_disp" "false,true,unknown"
  (cond [(eq_attr "type" "other,multi")
	   (const_string "unknown")
	 (and (eq_attr "type" "icmp,test,imov,alu1,ishift1,rotate1")
	      (and (match_operand 0 "memory_displacement_operand" "")
		   (match_operand 1 "immediate_operand" "")))
	   (const_string "true")
	 (and (eq_attr "type" "alu,ishift,rotate,imul,idiv")
	      (and (match_operand 0 "memory_displacement_operand" "")
		   (match_operand 2 "immediate_operand" "")))
	   (const_string "true")
	]
	(const_string "false")))

;; Indicates if an FP operation has an integer source.

(define_attr "fp_int_src" "false,true"
  (const_string "false"))

;; Defines rounding mode of an FP operation.

(define_attr "i387_cw" "trunc,floor,ceil,mask_pm,uninitialized,any"
  (const_string "any"))

;; Define attribute to classify add/sub insns that consumes carry flag (CF)
(define_attr "use_carry" "0,1" (const_string "0"))

;; Define attribute to indicate unaligned ssemov insns
(define_attr "movu" "0,1" (const_string "0"))

;; Describe a user's asm statement.
(define_asm_attributes
  [(set_attr "length" "128")
   (set_attr "type" "multi")])

(define_code_iterator plusminus [plus minus])

(define_code_iterator sat_plusminus [ss_plus us_plus ss_minus us_minus])

;; Base name for define_insn
(define_code_attr plusminus_insn
  [(plus "add") (ss_plus "ssadd") (us_plus "usadd")
   (minus "sub") (ss_minus "sssub") (us_minus "ussub")])

;; Base name for insn mnemonic.
(define_code_attr plusminus_mnemonic
  [(plus "add") (ss_plus "adds") (us_plus "addus")
   (minus "sub") (ss_minus "subs") (us_minus "subus")])
(define_code_attr plusminus_carry_mnemonic
  [(plus "adc") (minus "sbb")])

;; Mark commutative operators as such in constraints.
(define_code_attr comm [(plus "%") (ss_plus "%") (us_plus "%")
			(minus "") (ss_minus "") (us_minus "")])

;; Mapping of signed max and min
(define_code_iterator smaxmin [smax smin])

;; Mapping of unsigned max and min
(define_code_iterator umaxmin [umax umin])

;; Base name for integer and FP insn mnemonic
(define_code_attr maxmin_int [(smax "maxs") (smin "mins")
			      (umax "maxu") (umin "minu")])
(define_code_attr maxmin_float [(smax "max") (smin "min")])

;; Mapping of logic operators
(define_code_iterator any_logic [and ior xor])
(define_code_iterator any_or [ior xor])

;; Base name for insn mnemonic.
(define_code_attr logic [(and "and") (ior "or") (xor "xor")])

;; Mapping of shift-right operators
(define_code_iterator any_shiftrt [lshiftrt ashiftrt])

;; Base name for define_insn
(define_code_attr shiftrt_insn [(lshiftrt "lshr") (ashiftrt "ashr")])

;; Base name for insn mnemonic.
(define_code_attr shiftrt [(lshiftrt "shr") (ashiftrt "sar")])

;; Mapping of rotate operators
(define_code_iterator any_rotate [rotate rotatert])

;; Base name for define_insn
(define_code_attr rotate_insn [(rotate "rotl") (rotatert "rotr")])

;; Base name for insn mnemonic.
(define_code_attr rotate [(rotate "rol") (rotatert "ror")])

;; Mapping of abs neg operators
(define_code_iterator absneg [abs neg])

;; Base name for x87 insn mnemonic.
(define_code_attr absneg_mnemonic [(abs "abs") (neg "chs")])

;; Used in signed and unsigned widening multiplications.
(define_code_iterator any_extend [sign_extend zero_extend])

;; Various insn prefixes for signed and unsigned operations.
(define_code_attr u [(sign_extend "") (zero_extend "u")
		     (div "") (udiv "u")])
(define_code_attr s [(sign_extend "s") (zero_extend "u")])

;; Used in signed and unsigned divisions.
(define_code_iterator any_div [div udiv])

;; Instruction prefix for signed and unsigned operations.
(define_code_attr sgnprefix [(sign_extend "i") (zero_extend "")
			     (div "i") (udiv "")])

;; 64bit single word integer modes.
(define_mode_iterator SWI1248x [QI HI SI DI])

;; 64bit single word integer modes without QImode and HImode.
(define_mode_iterator SWI48x [SI DI])

;; Single word integer modes.
(define_mode_iterator SWI [QI HI SI (DI "TARGET_64BIT")])

;; Single word integer modes without SImode and DImode.
(define_mode_iterator SWI12 [QI HI])

;; Single word integer modes without DImode.
(define_mode_iterator SWI124 [QI HI SI])

;; Single word integer modes without QImode and DImode.
(define_mode_iterator SWI24 [HI SI])

;; Single word integer modes without QImode.
(define_mode_iterator SWI248 [HI SI (DI "TARGET_64BIT")])

;; Single word integer modes without QImode and HImode.
(define_mode_iterator SWI48 [SI (DI "TARGET_64BIT")])

;; All math-dependant single and double word integer modes.
(define_mode_iterator SDWIM [(QI "TARGET_QIMODE_MATH")
			     (HI "TARGET_HIMODE_MATH")
			     SI DI (TI "TARGET_64BIT")])

;; Math-dependant single word integer modes.
(define_mode_iterator SWIM [(QI "TARGET_QIMODE_MATH")
			    (HI "TARGET_HIMODE_MATH")
			    SI (DI "TARGET_64BIT")])

;; Math-dependant single word integer modes without DImode.
(define_mode_iterator SWIM124 [(QI "TARGET_QIMODE_MATH")
			       (HI "TARGET_HIMODE_MATH")
			       SI])

;; Math-dependant single word integer modes without QImode.
(define_mode_iterator SWIM248 [(HI "TARGET_HIMODE_MATH")
		      	       SI (DI "TARGET_64BIT")])

;; Double word integer modes.
(define_mode_iterator DWI [(DI "!TARGET_64BIT")
			   (TI "TARGET_64BIT")])

;; Double word integer modes as mode attribute.
(define_mode_attr DWI [(SI "DI") (DI "TI")])
(define_mode_attr dwi [(SI "di") (DI "ti")])

;; Half mode for double word integer modes.
(define_mode_iterator DWIH [(SI "!TARGET_64BIT")
			    (DI "TARGET_64BIT")])

;; Instruction suffix for integer modes.
(define_mode_attr imodesuffix [(QI "b") (HI "w") (SI "l") (DI "q")])

;; Pointer size prefix for integer modes (Intel asm dialect)
(define_mode_attr iptrsize [(QI "BYTE")
			    (HI "WORD")
			    (SI "DWORD")
			    (DI "QWORD")])

;; Register class for integer modes.
(define_mode_attr r [(QI "q") (HI "r") (SI "r") (DI "r")])

;; Immediate operand constraint for integer modes.
(define_mode_attr i [(QI "n") (HI "n") (SI "i") (DI "e")])

;; General operand constraint for word modes.
(define_mode_attr g [(QI "qmn") (HI "rmn") (SI "g") (DI "rme")])

;; Immediate operand constraint for double integer modes.
(define_mode_attr di [(SI "iF") (DI "e")])

;; Immediate operand constraint for shifts.
(define_mode_attr S [(QI "I") (HI "I") (SI "I") (DI "J") (TI "O")])

;; General operand predicate for integer modes.
(define_mode_attr general_operand
	[(QI "general_operand")
	 (HI "general_operand")
	 (SI "general_operand")
	 (DI "x86_64_general_operand")
	 (TI "x86_64_general_operand")])

;; General sign/zero extend operand predicate for integer modes.
(define_mode_attr general_szext_operand
	[(QI "general_operand")
	 (HI "general_operand")
	 (SI "general_operand")
	 (DI "x86_64_szext_general_operand")])

;; Immediate operand predicate for integer modes.
(define_mode_attr immediate_operand
	[(QI "immediate_operand")
	 (HI "immediate_operand")
	 (SI "immediate_operand")
	 (DI "x86_64_immediate_operand")])

;; Nonmemory operand predicate for integer modes.
(define_mode_attr nonmemory_operand
	[(QI "nonmemory_operand")
	 (HI "nonmemory_operand")
	 (SI "nonmemory_operand")
	 (DI "x86_64_nonmemory_operand")])

;; Operand predicate for shifts.
(define_mode_attr shift_operand
	[(QI "nonimmediate_operand")
	 (HI "nonimmediate_operand")
	 (SI "nonimmediate_operand")
	 (DI "shiftdi_operand")
	 (TI "register_operand")])

;; Operand predicate for shift argument.
(define_mode_attr shift_immediate_operand
	[(QI "const_1_to_31_operand")
	 (HI "const_1_to_31_operand")
	 (SI "const_1_to_31_operand")
	 (DI "const_1_to_63_operand")])

;; Input operand predicate for arithmetic left shifts.
(define_mode_attr ashl_input_operand
	[(QI "nonimmediate_operand")
	 (HI "nonimmediate_operand")
	 (SI "nonimmediate_operand")
	 (DI "ashldi_input_operand")
	 (TI "reg_or_pm1_operand")])

;; SSE and x87 SFmode and DFmode floating point modes
(define_mode_iterator MODEF [SF DF])

;; All x87 floating point modes
(define_mode_iterator X87MODEF [SF DF XF])

;; All integer modes handled by x87 fisttp operator.
(define_mode_iterator X87MODEI [HI SI DI])

;; All integer modes handled by integer x87 operators.
(define_mode_iterator X87MODEI12 [HI SI])

;; All integer modes handled by SSE cvtts?2si* operators.
(define_mode_iterator SSEMODEI24 [SI DI])

;; SSE asm suffix for floating point modes
(define_mode_attr ssemodefsuffix [(SF "s") (DF "d")])

;; SSE vector mode corresponding to a scalar mode
(define_mode_attr ssevecmode
  [(QI "V16QI") (HI "V8HI") (SI "V4SI") (DI "V2DI") (SF "V4SF") (DF "V2DF")])

;; Instruction suffix for REX 64bit operators.
(define_mode_attr rex64suffix [(SI "") (DI "{q}")])

;; This mode iterator allows :P to be used for patterns that operate on
;; pointer-sized quantities.  Exactly one of the two alternatives will match.
(define_mode_iterator P [(SI "Pmode == SImode") (DI "Pmode == DImode")])

;; Scheduling descriptions

(include "pentium.md")
(include "ppro.md")
(include "k6.md")
(include "athlon.md")
(include "bdver1.md")
(include "geode.md")
(include "atom.md")
(include "core2.md")


;; Operand and operator predicates and constraints

(include "predicates.md")
(include "constraints.md")


;; Compare and branch/compare and store instructions.

(define_expand "cbranch<mode>4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:SDWIM 1 "nonimmediate_operand" "")
		    (match_operand:SDWIM 2 "<general_operand>" "")))
   (set (pc) (if_then_else
	       (match_operator 0 "ordered_comparison_operator"
		[(reg:CC FLAGS_REG) (const_int 0)])
	       (label_ref (match_operand 3 "" ""))
	       (pc)))]
  ""
{
  if (MEM_P (operands[1]) && MEM_P (operands[2]))
    operands[1] = force_reg (<MODE>mode, operands[1]);
  ix86_expand_branch (GET_CODE (operands[0]),
		      operands[1], operands[2], operands[3]);
  DONE;
})

(define_expand "cstore<mode>4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:SWIM 2 "nonimmediate_operand" "")
		    (match_operand:SWIM 3 "<general_operand>" "")))
   (set (match_operand:QI 0 "register_operand" "")
	(match_operator 1 "ordered_comparison_operator"
	  [(reg:CC FLAGS_REG) (const_int 0)]))]
  ""
{
  if (MEM_P (operands[2]) && MEM_P (operands[3]))
    operands[2] = force_reg (<MODE>mode, operands[2]);
  ix86_expand_setcc (operands[0], GET_CODE (operands[1]),
		     operands[2], operands[3]);
  DONE;
})

(define_expand "cmp<mode>_1"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:SWI48 0 "nonimmediate_operand" "")
		    (match_operand:SWI48 1 "<general_operand>" "")))])

(define_insn "*cmp<mode>_ccno_1"
  [(set (reg FLAGS_REG)
	(compare (match_operand:SWI 0 "nonimmediate_operand" "<r>,?m<r>")
		 (match_operand:SWI 1 "const0_operand" "")))]
  "ix86_match_ccmode (insn, CCNOmode)"
  "@
   test{<imodesuffix>}\t%0, %0
   cmp{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "test,icmp")
   (set_attr "length_immediate" "0,1")
   (set_attr "mode" "<MODE>")])

(define_insn "*cmp<mode>_1"
  [(set (reg FLAGS_REG)
	(compare (match_operand:SWI 0 "nonimmediate_operand" "<r>m,<r>")
		 (match_operand:SWI 1 "<general_operand>" "<r><i>,<r>m")))]
  "ix86_match_ccmode (insn, CCmode)"
  "cmp{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "<MODE>")])

(define_insn "*cmp<mode>_minus_1"
  [(set (reg FLAGS_REG)
	(compare
	  (minus:SWI (match_operand:SWI 0 "nonimmediate_operand" "<r>m,<r>")
		     (match_operand:SWI 1 "<general_operand>" "<r><i>,<r>m"))
	  (const_int 0)))]
  "ix86_match_ccmode (insn, CCGOCmode)"
  "cmp{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "<MODE>")])

(define_insn "*cmpqi_ext_1"
  [(set (reg FLAGS_REG)
	(compare
	  (match_operand:QI 0 "general_operand" "Qm")
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)))]
  "!TARGET_64BIT && ix86_match_ccmode (insn, CCmode)"
  "cmp{b}\t{%h1, %0|%0, %h1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "QI")])

(define_insn "*cmpqi_ext_1_rex64"
  [(set (reg FLAGS_REG)
	(compare
	  (match_operand:QI 0 "register_operand" "Q")
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCmode)"
  "cmp{b}\t{%h1, %0|%0, %h1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "QI")])

(define_insn "*cmpqi_ext_2"
  [(set (reg FLAGS_REG)
	(compare
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)
	  (match_operand:QI 1 "const0_operand" "")))]
  "ix86_match_ccmode (insn, CCNOmode)"
  "test{b}\t%h0, %h0"
  [(set_attr "type" "test")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_expand "cmpqi_ext_3"
  [(set (reg:CC FLAGS_REG)
	(compare:CC
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "")
	      (const_int 8)
	      (const_int 8)) 0)
	  (match_operand:QI 1 "immediate_operand" "")))])

(define_insn "*cmpqi_ext_3_insn"
  [(set (reg FLAGS_REG)
	(compare
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)
	  (match_operand:QI 1 "general_operand" "Qmn")))]
  "!TARGET_64BIT && ix86_match_ccmode (insn, CCmode)"
  "cmp{b}\t{%1, %h0|%h0, %1}"
  [(set_attr "type" "icmp")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*cmpqi_ext_3_insn_rex64"
  [(set (reg FLAGS_REG)
	(compare
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)
	  (match_operand:QI 1 "nonmemory_operand" "Qn")))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCmode)"
  "cmp{b}\t{%1, %h0|%h0, %1}"
  [(set_attr "type" "icmp")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*cmpqi_ext_4"
  [(set (reg FLAGS_REG)
	(compare
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)
	  (subreg:QI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)) 0)))]
  "ix86_match_ccmode (insn, CCmode)"
  "cmp{b}\t{%h1, %h0|%h0, %h1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "QI")])

;; These implement float point compares.
;; %%% See if we can get away with VOIDmode operands on the actual insns,
;; which would allow mix and match FP modes on the compares.  Which is what
;; the old patterns did, but with many more of them.

(define_expand "cbranchxf4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:XF 1 "nonmemory_operand" "")
		    (match_operand:XF 2 "nonmemory_operand" "")))
   (set (pc) (if_then_else
              (match_operator 0 "ix86_fp_comparison_operator"
               [(reg:CC FLAGS_REG)
                (const_int 0)])
              (label_ref (match_operand 3 "" ""))
              (pc)))]
  "TARGET_80387"
{
  ix86_expand_branch (GET_CODE (operands[0]),
		      operands[1], operands[2], operands[3]);
  DONE;
})

(define_expand "cstorexf4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:XF 2 "nonmemory_operand" "")
		    (match_operand:XF 3 "nonmemory_operand" "")))
   (set (match_operand:QI 0 "register_operand" "")
              (match_operator 1 "ix86_fp_comparison_operator"
               [(reg:CC FLAGS_REG)
                (const_int 0)]))]
  "TARGET_80387"
{
  ix86_expand_setcc (operands[0], GET_CODE (operands[1]),
		     operands[2], operands[3]);
  DONE;
})

(define_expand "cbranch<mode>4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:MODEF 1 "cmp_fp_expander_operand" "")
		    (match_operand:MODEF 2 "cmp_fp_expander_operand" "")))
   (set (pc) (if_then_else
              (match_operator 0 "ix86_fp_comparison_operator"
               [(reg:CC FLAGS_REG)
                (const_int 0)])
              (label_ref (match_operand 3 "" ""))
              (pc)))]
  "TARGET_80387 || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)"
{
  ix86_expand_branch (GET_CODE (operands[0]),
		      operands[1], operands[2], operands[3]);
  DONE;
})

(define_expand "cstore<mode>4"
  [(set (reg:CC FLAGS_REG)
	(compare:CC (match_operand:MODEF 2 "cmp_fp_expander_operand" "")
		    (match_operand:MODEF 3 "cmp_fp_expander_operand" "")))
   (set (match_operand:QI 0 "register_operand" "")
              (match_operator 1 "ix86_fp_comparison_operator"
               [(reg:CC FLAGS_REG)
                (const_int 0)]))]
  "TARGET_80387 || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)"
{
  ix86_expand_setcc (operands[0], GET_CODE (operands[1]),
		     operands[2], operands[3]);
  DONE;
})

(define_expand "cbranchcc4"
  [(set (pc) (if_then_else
              (match_operator 0 "comparison_operator"
               [(match_operand 1 "flags_reg_operand" "")
                (match_operand 2 "const0_operand" "")])
              (label_ref (match_operand 3 "" ""))
              (pc)))]
  ""
{
  ix86_expand_branch (GET_CODE (operands[0]),
		      operands[1], operands[2], operands[3]);
  DONE;
})

(define_expand "cstorecc4"
  [(set (match_operand:QI 0 "register_operand" "")
              (match_operator 1 "comparison_operator"
               [(match_operand 2 "flags_reg_operand" "")
                (match_operand 3 "const0_operand" "")]))]
  ""
{
  ix86_expand_setcc (operands[0], GET_CODE (operands[1]),
		     operands[2], operands[3]);
  DONE;
})


;; FP compares, step 1:
;; Set the FP condition codes.
;;
;; CCFPmode	compare with exceptions
;; CCFPUmode	compare with no exceptions

;; We may not use "#" to split and emit these, since the REG_DEAD notes
;; used to manage the reg stack popping would not be preserved.

(define_insn "*cmpfp_0"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI
	  [(compare:CCFP
	     (match_operand 1 "register_operand" "f")
	     (match_operand 2 "const0_operand" ""))]
	UNSPEC_FNSTSW))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && GET_MODE (operands[1]) == GET_MODE (operands[2])"
  "* return output_fp_compare (insn, operands, 0, 0);"
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))])

(define_insn_and_split "*cmpfp_0_cc"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP
	  (match_operand 1 "register_operand" "f")
	  (match_operand 2 "const0_operand" "")))
   (clobber (match_operand:HI 0 "register_operand" "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_SAHF && !TARGET_CMOVE
   && GET_MODE (operands[1]) == GET_MODE (operands[2])"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(unspec:HI
	  [(compare:CCFP (match_dup 1)(match_dup 2))]
	UNSPEC_FNSTSW))
   (set (reg:CC FLAGS_REG)
	(unspec:CC [(match_dup 0)] UNSPEC_SAHF))]
  ""
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))])

(define_insn "*cmpfp_xf"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI
	  [(compare:CCFP
	     (match_operand:XF 1 "register_operand" "f")
	     (match_operand:XF 2 "register_operand" "f"))]
	  UNSPEC_FNSTSW))]
  "TARGET_80387"
  "* return output_fp_compare (insn, operands, 0, 0);"
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "mode" "XF")])

(define_insn_and_split "*cmpfp_xf_cc"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP
	  (match_operand:XF 1 "register_operand" "f")
	  (match_operand:XF 2 "register_operand" "f")))
   (clobber (match_operand:HI 0 "register_operand" "=a"))]
  "TARGET_80387
   && TARGET_SAHF && !TARGET_CMOVE"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(unspec:HI
	  [(compare:CCFP (match_dup 1)(match_dup 2))]
	UNSPEC_FNSTSW))
   (set (reg:CC FLAGS_REG)
	(unspec:CC [(match_dup 0)] UNSPEC_SAHF))]
  ""
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "mode" "XF")])

(define_insn "*cmpfp_<mode>"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI
	  [(compare:CCFP
	     (match_operand:MODEF 1 "register_operand" "f")
	     (match_operand:MODEF 2 "nonimmediate_operand" "fm"))]
	  UNSPEC_FNSTSW))]
  "TARGET_80387"
  "* return output_fp_compare (insn, operands, 0, 0);"
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "mode" "<MODE>")])

(define_insn_and_split "*cmpfp_<mode>_cc"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP
	  (match_operand:MODEF 1 "register_operand" "f")
	  (match_operand:MODEF 2 "nonimmediate_operand" "fm")))
   (clobber (match_operand:HI 0 "register_operand" "=a"))]
  "TARGET_80387
   && TARGET_SAHF && !TARGET_CMOVE"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(unspec:HI
	  [(compare:CCFP (match_dup 1)(match_dup 2))]
	UNSPEC_FNSTSW))
   (set (reg:CC FLAGS_REG)
	(unspec:CC [(match_dup 0)] UNSPEC_SAHF))]
  ""
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "mode" "<MODE>")])

(define_insn "*cmpfp_u"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI
	  [(compare:CCFPU
	     (match_operand 1 "register_operand" "f")
	     (match_operand 2 "register_operand" "f"))]
	  UNSPEC_FNSTSW))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && GET_MODE (operands[1]) == GET_MODE (operands[2])"
  "* return output_fp_compare (insn, operands, 0, 1);"
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))])

(define_insn_and_split "*cmpfp_u_cc"
  [(set (reg:CCFPU FLAGS_REG)
	(compare:CCFPU
	  (match_operand 1 "register_operand" "f")
	  (match_operand 2 "register_operand" "f")))
   (clobber (match_operand:HI 0 "register_operand" "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_SAHF && !TARGET_CMOVE
   && GET_MODE (operands[1]) == GET_MODE (operands[2])"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(unspec:HI
	  [(compare:CCFPU (match_dup 1)(match_dup 2))]
	UNSPEC_FNSTSW))
   (set (reg:CC FLAGS_REG)
	(unspec:CC [(match_dup 0)] UNSPEC_SAHF))]
  ""
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))])

(define_insn "*cmpfp_<mode>"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI
	  [(compare:CCFP
	     (match_operand 1 "register_operand" "f")
	     (match_operator 3 "float_operator"
	       [(match_operand:X87MODEI12 2 "memory_operand" "m")]))]
	  UNSPEC_FNSTSW))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && (TARGET_USE_<MODE>MODE_FIOP || optimize_function_for_size_p (cfun))
   && (GET_MODE (operands [3]) == GET_MODE (operands[1]))"
  "* return output_fp_compare (insn, operands, 0, 0);"
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "fp_int_src" "true")
   (set_attr "mode" "<MODE>")])

(define_insn_and_split "*cmpfp_<mode>_cc"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP
	  (match_operand 1 "register_operand" "f")
	  (match_operator 3 "float_operator"
	    [(match_operand:X87MODEI12 2 "memory_operand" "m")])))
   (clobber (match_operand:HI 0 "register_operand" "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_SAHF && !TARGET_CMOVE
   && (TARGET_USE_<MODE>MODE_FIOP || optimize_function_for_size_p (cfun))
   && (GET_MODE (operands [3]) == GET_MODE (operands[1]))"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(unspec:HI
	  [(compare:CCFP
	     (match_dup 1)
	     (match_op_dup 3 [(match_dup 2)]))]
	UNSPEC_FNSTSW))
   (set (reg:CC FLAGS_REG)
	(unspec:CC [(match_dup 0)] UNSPEC_SAHF))]
  ""
  [(set_attr "type" "multi")
   (set_attr "unit" "i387")
   (set_attr "fp_int_src" "true")
   (set_attr "mode" "<MODE>")])

;; FP compares, step 2
;; Move the fpsw to ax.

(define_insn "x86_fnstsw_1"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(unspec:HI [(reg:CCFP FPSR_REG)] UNSPEC_FNSTSW))]
  "TARGET_80387"
  "fnstsw\t%0"
  [(set (attr "length") (symbol_ref "ix86_attr_length_address_default (insn) + 2"))
   (set_attr "mode" "SI")
   (set_attr "unit" "i387")])

;; FP compares, step 3
;; Get ax into flags, general case.

(define_insn "x86_sahf_1"
  [(set (reg:CC FLAGS_REG)
	(unspec:CC [(match_operand:HI 0 "register_operand" "a")]
		   UNSPEC_SAHF))]
  "TARGET_SAHF"
{
#ifndef HAVE_AS_IX86_SAHF
  if (TARGET_64BIT)
    return ASM_BYTE "0x9e";
  else
#endif
  return "sahf";
}
  [(set_attr "length" "1")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "SI")])

;; Pentium Pro can do steps 1 through 3 in one go.
;; comi*, ucomi*, fcomi*, ficomi*,fucomi* (i387 instructions set condition codes)
(define_insn "*cmpfp_i_mixed"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP (match_operand 0 "register_operand" "f,x")
		      (match_operand 1 "nonimmediate_operand" "f,xm")))]
  "TARGET_MIX_SSE_I387
   && SSE_FLOAT_MODE_P (GET_MODE (operands[0]))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 0);"
  [(set_attr "type" "fcmp,ssecomi")
   (set_attr "prefix" "orig,maybe_vex")
   (set (attr "mode")
     (if_then_else (match_operand:SF 1 "" "")
        (const_string "SF")
        (const_string "DF")))
   (set (attr "prefix_rep")
	(if_then_else (eq_attr "type" "ssecomi")
		      (const_string "0")
		      (const_string "*")))
   (set (attr "prefix_data16")
	(cond [(eq_attr "type" "fcmp")
		 (const_string "*")
	       (eq_attr "mode" "DF")
		 (const_string "1")
	      ]
	      (const_string "0")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "double")])

(define_insn "*cmpfp_i_sse"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP (match_operand 0 "register_operand" "x")
		      (match_operand 1 "nonimmediate_operand" "xm")))]
  "TARGET_SSE_MATH
   && SSE_FLOAT_MODE_P (GET_MODE (operands[0]))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 0);"
  [(set_attr "type" "ssecomi")
   (set_attr "prefix" "maybe_vex")
   (set (attr "mode")
     (if_then_else (match_operand:SF 1 "" "")
        (const_string "SF")
        (const_string "DF")))
   (set_attr "prefix_rep" "0")
   (set (attr "prefix_data16")
	(if_then_else (eq_attr "mode" "DF")
		      (const_string "1")
		      (const_string "0")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "double")])

(define_insn "*cmpfp_i_i387"
  [(set (reg:CCFP FLAGS_REG)
	(compare:CCFP (match_operand 0 "register_operand" "f")
		      (match_operand 1 "register_operand" "f")))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[0]))
   && TARGET_CMOVE
   && !(SSE_FLOAT_MODE_P (GET_MODE (operands[0])) && TARGET_SSE_MATH)
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 0);"
  [(set_attr "type" "fcmp")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "double")])

(define_insn "*cmpfp_iu_mixed"
  [(set (reg:CCFPU FLAGS_REG)
	(compare:CCFPU (match_operand 0 "register_operand" "f,x")
		       (match_operand 1 "nonimmediate_operand" "f,xm")))]
  "TARGET_MIX_SSE_I387
   && SSE_FLOAT_MODE_P (GET_MODE (operands[0]))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 1);"
  [(set_attr "type" "fcmp,ssecomi")
   (set_attr "prefix" "orig,maybe_vex")
   (set (attr "mode")
     (if_then_else (match_operand:SF 1 "" "")
        (const_string "SF")
        (const_string "DF")))
   (set (attr "prefix_rep")
	(if_then_else (eq_attr "type" "ssecomi")
		      (const_string "0")
		      (const_string "*")))
   (set (attr "prefix_data16")
	(cond [(eq_attr "type" "fcmp")
		 (const_string "*")
	       (eq_attr "mode" "DF")
		 (const_string "1")
	      ]
	      (const_string "0")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "double")])

(define_insn "*cmpfp_iu_sse"
  [(set (reg:CCFPU FLAGS_REG)
	(compare:CCFPU (match_operand 0 "register_operand" "x")
		       (match_operand 1 "nonimmediate_operand" "xm")))]
  "TARGET_SSE_MATH
   && SSE_FLOAT_MODE_P (GET_MODE (operands[0]))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 1);"
  [(set_attr "type" "ssecomi")
   (set_attr "prefix" "maybe_vex")
   (set (attr "mode")
     (if_then_else (match_operand:SF 1 "" "")
        (const_string "SF")
        (const_string "DF")))
   (set_attr "prefix_rep" "0")
   (set (attr "prefix_data16")
	(if_then_else (eq_attr "mode" "DF")
		      (const_string "1")
		      (const_string "0")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "double")])

(define_insn "*cmpfp_iu_387"
  [(set (reg:CCFPU FLAGS_REG)
	(compare:CCFPU (match_operand 0 "register_operand" "f")
		       (match_operand 1 "register_operand" "f")))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[0]))
   && TARGET_CMOVE
   && !(SSE_FLOAT_MODE_P (GET_MODE (operands[0])) && TARGET_SSE_MATH)
   && GET_MODE (operands[0]) == GET_MODE (operands[1])"
  "* return output_fp_compare (insn, operands, 1, 1);"
  [(set_attr "type" "fcmp")
   (set (attr "mode")
     (cond [(match_operand:SF 1 "" "")
	      (const_string "SF")
	    (match_operand:DF 1 "" "")
	      (const_string "DF")
	   ]
	   (const_string "XF")))
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "direct")])

;; Push/pop instructions.

(define_insn "*push<mode>2"
  [(set (match_operand:DWI 0 "push_operand" "=<")
	(match_operand:DWI 1 "general_no_elim_operand" "riF*m"))]
  ""
  "#")

(define_split
  [(set (match_operand:TI 0 "push_operand" "")
        (match_operand:TI 1 "general_operand" ""))]
  "TARGET_64BIT && reload_completed
   && !SSE_REG_P (operands[1])"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*pushdi2_rex64"
  [(set (match_operand:DI 0 "push_operand" "=<,!<")
	(match_operand:DI 1 "general_no_elim_operand" "re*m,n"))]
  "TARGET_64BIT"
  "@
   push{q}\t%1
   #"
  [(set_attr "type" "push,multi")
   (set_attr "mode" "DI")])

;; Convert impossible pushes of immediate to existing instructions.
;; First try to get scratch register and go through it.  In case this
;; fails, push sign extended lower part first and then overwrite
;; upper part by 32bit move.
(define_peephole2
  [(match_scratch:DI 2 "r")
   (set (match_operand:DI 0 "push_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode)"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (match_dup 2))])

;; We need to define this as both peepholer and splitter for case
;; peephole2 pass is not run.
;; "&& 1" is needed to keep it from matching the previous pattern.
(define_peephole2
  [(set (match_operand:DI 0 "push_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode) && 1"
  [(set (match_dup 0) (match_dup 1))
   (set (match_dup 2) (match_dup 3))]
{
  split_double_mode (DImode, &operands[1], 1, &operands[2], &operands[3]);

  operands[1] = gen_lowpart (DImode, operands[2]);
  operands[2] = gen_rtx_MEM (SImode, gen_rtx_PLUS (DImode, stack_pointer_rtx,
						   GEN_INT (4)));
})

(define_split
  [(set (match_operand:DI 0 "push_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && ((optimize > 0 && flag_peephole2)
		    ? epilogue_completed : reload_completed)
   && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode)"
  [(set (match_dup 0) (match_dup 1))
   (set (match_dup 2) (match_dup 3))]
{
  split_double_mode (DImode, &operands[1], 1, &operands[2], &operands[3]);

  operands[1] = gen_lowpart (DImode, operands[2]);
  operands[2] = gen_rtx_MEM (SImode, gen_rtx_PLUS (DImode, stack_pointer_rtx,
						   GEN_INT (4)));
})

(define_split
  [(set (match_operand:DI 0 "push_operand" "")
        (match_operand:DI 1 "general_operand" ""))]
  "!TARGET_64BIT && reload_completed
   && !(MMX_REG_P (operands[1]) || SSE_REG_P (operands[1]))"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*pushsi2"
  [(set (match_operand:SI 0 "push_operand" "=<")
	(match_operand:SI 1 "general_no_elim_operand" "ri*m"))]
  "!TARGET_64BIT"
  "push{l}\t%1"
  [(set_attr "type" "push")
   (set_attr "mode" "SI")])

;; emit_push_insn when it calls move_by_pieces requires an insn to
;; "push a byte/word".  But actually we use pushl, which has the effect
;; of rounding the amount pushed up to a word.

;; For TARGET_64BIT we always round up to 8 bytes.
(define_insn "*push<mode>2_rex64"
  [(set (match_operand:SWI124 0 "push_operand" "=X")
	(match_operand:SWI124 1 "nonmemory_no_elim_operand" "r<i>"))]
  "TARGET_64BIT"
  "push{q}\t%q1"
  [(set_attr "type" "push")
   (set_attr "mode" "DI")])

(define_insn "*push<mode>2"
  [(set (match_operand:SWI12 0 "push_operand" "=X")
	(match_operand:SWI12 1 "nonmemory_no_elim_operand" "rn"))]
  "!TARGET_64BIT"
  "push{l}\t%k1"
  [(set_attr "type" "push")
   (set_attr "mode" "SI")])

(define_insn "*push<mode>2_prologue"
  [(set (match_operand:P 0 "push_operand" "=<")
	(match_operand:P 1 "general_no_elim_operand" "r<i>*m"))
   (clobber (mem:BLK (scratch)))]
  ""
  "push{<imodesuffix>}\t%1"
  [(set_attr "type" "push")
   (set_attr "mode" "<MODE>")])

(define_insn "*pop<mode>1"
  [(set (match_operand:P 0 "nonimmediate_operand" "=r*m")
	(match_operand:P 1 "pop_operand" ">"))]
  ""
  "pop{<imodesuffix>}\t%0"
  [(set_attr "type" "pop")
   (set_attr "mode" "<MODE>")])

(define_insn "*pop<mode>1_epilogue"
  [(set (match_operand:P 0 "nonimmediate_operand" "=r*m")
	(match_operand:P 1 "pop_operand" ">"))
   (clobber (mem:BLK (scratch)))]
  ""
  "pop{<imodesuffix>}\t%0"
  [(set_attr "type" "pop")
   (set_attr "mode" "<MODE>")])

;; Move instructions.

(define_expand "movoi"
  [(set (match_operand:OI 0 "nonimmediate_operand" "")
	(match_operand:OI 1 "general_operand" ""))]
  "TARGET_AVX"
  "ix86_expand_move (OImode, operands); DONE;")

(define_expand "movti"
  [(set (match_operand:TI 0 "nonimmediate_operand" "")
	(match_operand:TI 1 "nonimmediate_operand" ""))]
  "TARGET_64BIT || TARGET_SSE"
{
  if (TARGET_64BIT)
    ix86_expand_move (TImode, operands);
  else if (push_operand (operands[0], TImode))
    ix86_expand_push (TImode, operands[1]);
  else
    ix86_expand_vector_move (TImode, operands);
  DONE;
})

;; This expands to what emit_move_complex would generate if we didn't
;; have a movti pattern.  Having this avoids problems with reload on
;; 32-bit targets when SSE is present, but doesn't seem to be harmful
;; to have around all the time.
(define_expand "movcdi"
  [(set (match_operand:CDI 0 "nonimmediate_operand" "")
	(match_operand:CDI 1 "general_operand" ""))]
  ""
{
  if (push_operand (operands[0], CDImode))
    emit_move_complex_push (CDImode, operands[0], operands[1]);
  else
    emit_move_complex_parts (operands[0], operands[1]);
  DONE;
})

(define_expand "mov<mode>"
  [(set (match_operand:SWI1248x 0 "nonimmediate_operand" "")
	(match_operand:SWI1248x 1 "general_operand" ""))]
  ""
  "ix86_expand_move (<MODE>mode, operands); DONE;")

(define_insn "*mov<mode>_xor"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
	(match_operand:SWI48 1 "const0_operand" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  "xor{l}\t%k0, %k0"
  [(set_attr "type" "alu1")
   (set_attr "mode" "SI")
   (set_attr "length_immediate" "0")])

(define_insn "*mov<mode>_or"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
	(match_operand:SWI48 1 "const_int_operand" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed
   && operands[1] == constm1_rtx"
  "or{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "<MODE>")
   (set_attr "length_immediate" "1")])

(define_insn "*movoi_internal_avx"
  [(set (match_operand:OI 0 "nonimmediate_operand" "=x,x,m")
	(match_operand:OI 1 "vector_move_operand" "C,xm,x"))]
  "TARGET_AVX && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (which_alternative)
    {
    case 0:
      return "vxorps\t%0, %0, %0";
    case 1:
    case 2:
      if (misaligned_operand (operands[0], OImode)
	  || misaligned_operand (operands[1], OImode))
	return "vmovdqu\t{%1, %0|%0, %1}";
      else
	return "vmovdqa\t{%1, %0|%0, %1}";
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "sselog1,ssemov,ssemov")
   (set_attr "prefix" "vex")
   (set_attr "mode" "OI")])

(define_insn "*movti_internal_rex64"
  [(set (match_operand:TI 0 "nonimmediate_operand" "=!r,o,x,x,xm")
	(match_operand:TI 1 "general_operand" "riFo,riF,C,xm,x"))]
  "TARGET_64BIT && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return "#";
    case 2:
      if (get_attr_mode (insn) == MODE_V4SF)
	return "%vxorps\t%0, %d0";
      else
	return "%vpxor\t%0, %d0";
    case 3:
    case 4:
      /* TDmode values are passed as TImode on the stack.  Moving them
	 to stack may result in unaligned memory access.  */
      if (misaligned_operand (operands[0], TImode)
	  || misaligned_operand (operands[1], TImode))
	{
	  if (get_attr_mode (insn) == MODE_V4SF)
	    return "%vmovups\t{%1, %0|%0, %1}";
	 else
	   return "%vmovdqu\t{%1, %0|%0, %1}";
	}
      else
	{
	  if (get_attr_mode (insn) == MODE_V4SF)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	 else
	   return "%vmovdqa\t{%1, %0|%0, %1}";
	}
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "*,*,sselog1,ssemov,ssemov")
   (set_attr "prefix" "*,*,maybe_vex,maybe_vex,maybe_vex")
   (set (attr "mode")
   	(cond [(eq_attr "alternative" "2,3")
		 (if_then_else
		   (ne (symbol_ref "optimize_function_for_size_p (cfun)")
		       (const_int 0))
		   (const_string "V4SF")
		   (const_string "TI"))
	       (eq_attr "alternative" "4")
		 (if_then_else
		   (ior (ne (symbol_ref "TARGET_SSE_TYPELESS_STORES")
			    (const_int 0))
			(ne (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0)))
		   (const_string "V4SF")
		   (const_string "TI"))]
	       (const_string "DI")))])

(define_split
  [(set (match_operand:TI 0 "nonimmediate_operand" "")
	(match_operand:TI 1 "general_operand" ""))]
  "reload_completed
   && !SSE_REG_P (operands[0]) && !SSE_REG_P (operands[1])"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*movti_internal_sse"
  [(set (match_operand:TI 0 "nonimmediate_operand" "=x,x,m")
	(match_operand:TI 1 "vector_move_operand" "C,xm,x"))]
  "TARGET_SSE && !TARGET_64BIT
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (which_alternative)
    {
    case 0:
      if (get_attr_mode (insn) == MODE_V4SF)
	return "%vxorps\t%0, %d0";
      else
	return "%vpxor\t%0, %d0";
    case 1:
    case 2:
      /* TDmode values are passed as TImode on the stack.  Moving them
	 to stack may result in unaligned memory access.  */
      if (misaligned_operand (operands[0], TImode)
	  || misaligned_operand (operands[1], TImode))
	{
	  if (get_attr_mode (insn) == MODE_V4SF)
	    return "%vmovups\t{%1, %0|%0, %1}";
	 else
	   return "%vmovdqu\t{%1, %0|%0, %1}";
	}
      else
	{
	  if (get_attr_mode (insn) == MODE_V4SF)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	 else
	   return "%vmovdqa\t{%1, %0|%0, %1}";
	}
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "sselog1,ssemov,ssemov")
   (set_attr "prefix" "maybe_vex")
   (set (attr "mode")
	(cond [(ior (eq (symbol_ref "TARGET_SSE2") (const_int 0))
		    (ne (symbol_ref "optimize_function_for_size_p (cfun)")
			(const_int 0)))
		 (const_string "V4SF")
	       (and (eq_attr "alternative" "2")
		    (ne (symbol_ref "TARGET_SSE_TYPELESS_STORES")
			(const_int 0)))
		 (const_string "V4SF")]
	      (const_string "TI")))])

(define_insn "*movdi_internal_rex64"
  [(set (match_operand:DI 0 "nonimmediate_operand"
	  "=r,r  ,r,m ,!m,*y,*y,?r ,m ,?*Ym,?*y,*x,*x,?r ,m,?*Yi,*x,?*x,?*Ym")
	(match_operand:DI 1 "general_operand"
	  "Z ,rem,i,re,n ,C ,*y,*Ym,*y,r   ,m  ,C ,*x,*Yi,*x,r  ,m ,*Ym,*x"))]
  "TARGET_64BIT && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_SSECVT:
      if (SSE_REG_P (operands[0]))
	return "movq2dq\t{%1, %0|%0, %1}";
      else
	return "movdq2q\t{%1, %0|%0, %1}";

    case TYPE_SSEMOV:
      if (TARGET_AVX)
	{
	  if (get_attr_mode (insn) == MODE_TI)
	    return "vmovdqa\t{%1, %0|%0, %1}";
	  else
	    return "vmovq\t{%1, %0|%0, %1}";
	}

      if (get_attr_mode (insn) == MODE_TI)
	return "movdqa\t{%1, %0|%0, %1}";
      /* FALLTHRU */

    case TYPE_MMXMOV:
      /* Moves from and into integer register is done using movd
	 opcode with REX prefix.  */
      if (GENERAL_REG_P (operands[0]) || GENERAL_REG_P (operands[1]))
	return "movd\t{%1, %0|%0, %1}";
      return "movq\t{%1, %0|%0, %1}";

    case TYPE_SSELOG1:
      return "%vpxor\t%0, %d0";

    case TYPE_MMX:
      return "pxor\t%0, %0";

    case TYPE_MULTI:
      return "#";

    case TYPE_LEA:
      return "lea{q}\t{%a1, %0|%0, %a1}";

    default:
      gcc_assert (!flag_pic || LEGITIMATE_PIC_OPERAND_P (operands[1]));
      if (get_attr_mode (insn) == MODE_SI)
	return "mov{l}\t{%k1, %k0|%k0, %k1}";
      else if (which_alternative == 2)
	return "movabs{q}\t{%1, %0|%0, %1}";
      else
	return "mov{q}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "5")
	      (const_string "mmx")
	    (eq_attr "alternative" "6,7,8,9,10")
	      (const_string "mmxmov")
	    (eq_attr "alternative" "11")
	      (const_string "sselog1")
	    (eq_attr "alternative" "12,13,14,15,16")
	      (const_string "ssemov")
	    (eq_attr "alternative" "17,18")
	      (const_string "ssecvt")
	    (eq_attr "alternative" "4")
	      (const_string "multi")
 	    (match_operand:DI 1 "pic_32bit_operand" "")
	      (const_string "lea")
	   ]
	   (const_string "imov")))
   (set (attr "modrm")
     (if_then_else
       (and (eq_attr "alternative" "2") (eq_attr "type" "imov"))
	 (const_string "0")
	 (const_string "*")))
   (set (attr "length_immediate")
     (if_then_else
       (and (eq_attr "alternative" "2") (eq_attr "type" "imov"))
	 (const_string "8")
	 (const_string "*")))
   (set_attr "prefix_rex" "*,*,*,*,*,*,*,1,*,1,*,*,*,*,*,*,*,*,*")
   (set_attr "prefix_data16" "*,*,*,*,*,*,*,*,*,*,*,*,*,*,*,1,*,*,*")
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "11,12,13,14,15,16")
       (const_string "maybe_vex")
       (const_string "orig")))
   (set_attr "mode" "SI,DI,DI,DI,SI,DI,DI,DI,DI,DI,DI,TI,TI,DI,DI,DI,DI,DI,DI")])

;; Convert impossible stores of immediate to existing instructions.
;; First try to get scratch register and go through it.  In case this
;; fails, move by 32bit parts.
(define_peephole2
  [(match_scratch:DI 2 "r")
   (set (match_operand:DI 0 "memory_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode)"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (match_dup 2))])

;; We need to define this as both peepholer and splitter for case
;; peephole2 pass is not run.
;; "&& 1" is needed to keep it from matching the previous pattern.
(define_peephole2
  [(set (match_operand:DI 0 "memory_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode) && 1"
  [(set (match_dup 2) (match_dup 3))
   (set (match_dup 4) (match_dup 5))]
  "split_double_mode (DImode, &operands[0], 2, &operands[2], &operands[4]);")

(define_split
  [(set (match_operand:DI 0 "memory_operand" "")
        (match_operand:DI 1 "immediate_operand" ""))]
  "TARGET_64BIT && ((optimize > 0 && flag_peephole2)
		    ? epilogue_completed : reload_completed)
   && !symbolic_operand (operands[1], DImode)
   && !x86_64_immediate_operand (operands[1], DImode)"
  [(set (match_dup 2) (match_dup 3))
   (set (match_dup 4) (match_dup 5))]
  "split_double_mode (DImode, &operands[0], 2, &operands[2], &operands[4]);")

(define_insn "*movdi_internal"
  [(set (match_operand:DI 0 "nonimmediate_operand"
			"=r  ,o  ,*y,m*y,*y,*Y2,m  ,*Y2,*Y2,*x,m ,*x,*x")
	(match_operand:DI 1 "general_operand"
			"riFo,riF,C ,*y ,m ,C  ,*Y2,*Y2,m  ,C ,*x,*x,m "))]
  "!TARGET_64BIT && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "@
   #
   #
   pxor\t%0, %0
   movq\t{%1, %0|%0, %1}
   movq\t{%1, %0|%0, %1}
   %vpxor\t%0, %d0
   %vmovq\t{%1, %0|%0, %1}
   %vmovdqa\t{%1, %0|%0, %1}
   %vmovq\t{%1, %0|%0, %1}
   xorps\t%0, %0
   movlps\t{%1, %0|%0, %1}
   movaps\t{%1, %0|%0, %1}
   movlps\t{%1, %0|%0, %1}"
  [(set_attr "type" "*,*,mmx,mmxmov,mmxmov,sselog1,ssemov,ssemov,ssemov,sselog1,ssemov,ssemov,ssemov")
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "5,6,7,8")
       (const_string "vex")
       (const_string "orig")))
   (set_attr "mode" "DI,DI,DI,DI,DI,TI,DI,TI,DI,V4SF,V2SF,V4SF,V2SF")])

(define_split
  [(set (match_operand:DI 0 "nonimmediate_operand" "")
        (match_operand:DI 1 "general_operand" ""))]
  "!TARGET_64BIT && reload_completed
   && !(MMX_REG_P (operands[0]) || SSE_REG_P (operands[0]))
   && !(MMX_REG_P (operands[1]) || SSE_REG_P (operands[1]))"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*movsi_internal"
  [(set (match_operand:SI 0 "nonimmediate_operand"
			"=r,m ,*y,*y,?rm,?*y,*x,*x,?r ,m ,?*Yi,*x")
	(match_operand:SI 1 "general_operand"
			"g ,ri,C ,*y,*y ,rm ,C ,*x,*Yi,*x,r   ,m "))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_SSELOG1:
      if (get_attr_mode (insn) == MODE_TI)
        return "%vpxor\t%0, %d0";
      return "%vxorps\t%0, %d0";

    case TYPE_SSEMOV:
      switch (get_attr_mode (insn))
	{
	case MODE_TI:
	  return "%vmovdqa\t{%1, %0|%0, %1}";
	case MODE_V4SF:
	  return "%vmovaps\t{%1, %0|%0, %1}";
	case MODE_SI:
          return "%vmovd\t{%1, %0|%0, %1}";
	case MODE_SF:
          return "%vmovss\t{%1, %0|%0, %1}";
	default:
	  gcc_unreachable ();
	}

    case TYPE_MMX:
      return "pxor\t%0, %0";

    case TYPE_MMXMOV:
      if (get_attr_mode (insn) == MODE_DI)
	return "movq\t{%1, %0|%0, %1}";
      return "movd\t{%1, %0|%0, %1}";

    case TYPE_LEA:
      return "lea{l}\t{%a1, %0|%0, %a1}";

    default:
      gcc_assert (!flag_pic || LEGITIMATE_PIC_OPERAND_P (operands[1]));
      return "mov{l}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "2")
	      (const_string "mmx")
	    (eq_attr "alternative" "3,4,5")
	      (const_string "mmxmov")
	    (eq_attr "alternative" "6")
	      (const_string "sselog1")
	    (eq_attr "alternative" "7,8,9,10,11")
	      (const_string "ssemov")
 	    (match_operand:DI 1 "pic_32bit_operand" "")
	      (const_string "lea")
	   ]
	   (const_string "imov")))
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "0,1,2,3,4,5")
       (const_string "orig")
       (const_string "maybe_vex")))
   (set (attr "prefix_data16")
     (if_then_else (and (eq_attr "type" "ssemov") (eq_attr "mode" "SI"))
       (const_string "1")
       (const_string "*")))
   (set (attr "mode")
     (cond [(eq_attr "alternative" "2,3")
	      (const_string "DI")
	    (eq_attr "alternative" "6,7")
	      (if_then_else
	        (eq (symbol_ref "TARGET_SSE2") (const_int 0))
	        (const_string "V4SF")
	        (const_string "TI"))
	    (and (eq_attr "alternative" "8,9,10,11")
	         (eq (symbol_ref "TARGET_SSE2") (const_int 0)))
	      (const_string "SF")
	   ]
	   (const_string "SI")))])

(define_insn "*movhi_internal"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=r,r,r,m")
	(match_operand:HI 1 "general_operand" "r,rn,rm,rn"))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      /* movzwl is faster than movw on p2 due to partial word stalls,
	 though not as fast as an aligned movl.  */
      return "movz{wl|x}\t{%1, %k0|%k0, %1}";
    default:
      if (get_attr_mode (insn) == MODE_SI)
        return "mov{l}\t{%k1, %k0|%k0, %k1}";
      else
        return "mov{w}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (cond [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
		(const_int 0))
	      (const_string "imov")
	    (and (eq_attr "alternative" "0")
		 (ior (eq (symbol_ref "TARGET_PARTIAL_REG_STALL")
			  (const_int 0))
		      (eq (symbol_ref "TARGET_HIMODE_MATH")
			  (const_int 0))))
	      (const_string "imov")
	    (and (eq_attr "alternative" "1,2")
		 (match_operand:HI 1 "aligned_operand" ""))
	      (const_string "imov")
	    (and (ne (symbol_ref "TARGET_MOVX")
		     (const_int 0))
		 (eq_attr "alternative" "0,2"))
	      (const_string "imovx")
	   ]
	   (const_string "imov")))
    (set (attr "mode")
      (cond [(eq_attr "type" "imovx")
	       (const_string "SI")
	     (and (eq_attr "alternative" "1,2")
		  (match_operand:HI 1 "aligned_operand" ""))
	       (const_string "SI")
	     (and (eq_attr "alternative" "0")
		  (ior (eq (symbol_ref "TARGET_PARTIAL_REG_STALL")
			   (const_int 0))
		       (eq (symbol_ref "TARGET_HIMODE_MATH")
			   (const_int 0))))
	       (const_string "SI")
	    ]
	    (const_string "HI")))])

;; Situation is quite tricky about when to choose full sized (SImode) move
;; over QImode moves.  For Q_REG -> Q_REG move we use full size only for
;; partial register dependency machines (such as AMD Athlon), where QImode
;; moves issue extra dependency and for partial register stalls machines
;; that don't use QImode patterns (and QImode move cause stall on the next
;; instruction).
;;
;; For loads of Q_REG to NONQ_REG we use full sized moves except for partial
;; register stall machines with, where we use QImode instructions, since
;; partial register stall can be caused there.  Then we use movzx.
(define_insn "*movqi_internal"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=q,q ,q ,r,r ,?r,m")
	(match_operand:QI 1 "general_operand"      " q,qn,qm,q,rn,qm,qn"))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      gcc_assert (ANY_QI_REG_P (operands[1]) || MEM_P (operands[1]));
      return "movz{bl|x}\t{%1, %k0|%k0, %1}";
    default:
      if (get_attr_mode (insn) == MODE_SI)
        return "mov{l}\t{%k1, %k0|%k0, %k1}";
      else
        return "mov{b}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (cond [(and (eq_attr "alternative" "5")
		 (not (match_operand:QI 1 "aligned_operand" "")))
	      (const_string "imovx")
	    (ne (symbol_ref "optimize_function_for_size_p (cfun)")
		(const_int 0))
	      (const_string "imov")
	    (and (eq_attr "alternative" "3")
		 (ior (eq (symbol_ref "TARGET_PARTIAL_REG_STALL")
			  (const_int 0))
		      (eq (symbol_ref "TARGET_QIMODE_MATH")
			  (const_int 0))))
	      (const_string "imov")
	    (eq_attr "alternative" "3,5")
	      (const_string "imovx")
	    (and (ne (symbol_ref "TARGET_MOVX")
		     (const_int 0))
		 (eq_attr "alternative" "2"))
	      (const_string "imovx")
	   ]
	   (const_string "imov")))
   (set (attr "mode")
      (cond [(eq_attr "alternative" "3,4,5")
	       (const_string "SI")
	     (eq_attr "alternative" "6")
	       (const_string "QI")
	     (eq_attr "type" "imovx")
	       (const_string "SI")
	     (and (eq_attr "type" "imov")
		  (and (eq_attr "alternative" "0,1")
		       (and (ne (symbol_ref "TARGET_PARTIAL_REG_DEPENDENCY")
				(const_int 0))
			    (and (eq (symbol_ref "optimize_function_for_size_p (cfun)")
				     (const_int 0))
				 (eq (symbol_ref "TARGET_PARTIAL_REG_STALL")
				     (const_int 0))))))
	       (const_string "SI")
	     ;; Avoid partial register stalls when not using QImode arithmetic
	     (and (eq_attr "type" "imov")
		  (and (eq_attr "alternative" "0,1")
		       (and (ne (symbol_ref "TARGET_PARTIAL_REG_STALL")
				(const_int 0))
			    (eq (symbol_ref "TARGET_QIMODE_MATH")
				(const_int 0)))))
	       (const_string "SI")
	   ]
	   (const_string "QI")))])

;; Stores and loads of ax to arbitrary constant address.
;; We fake an second form of instruction to force reload to load address
;; into register when rax is not available
(define_insn "*movabs<mode>_1"
  [(set (mem:SWI1248x (match_operand:DI 0 "x86_64_movabs_operand" "i,r"))
	(match_operand:SWI1248x 1 "nonmemory_operand" "a,er"))]
  "TARGET_64BIT && ix86_check_movabs (insn, 0)"
  "@
   movabs{<imodesuffix>}\t{%1, %P0|%P0, %1}
   mov{<imodesuffix>}\t{%1, %a0|%a0, %1}"
  [(set_attr "type" "imov")
   (set_attr "modrm" "0,*")
   (set_attr "length_address" "8,0")
   (set_attr "length_immediate" "0,*")
   (set_attr "memory" "store")
   (set_attr "mode" "<MODE>")])

(define_insn "*movabs<mode>_2"
  [(set (match_operand:SWI1248x 0 "register_operand" "=a,r")
        (mem:SWI1248x (match_operand:DI 1 "x86_64_movabs_operand" "i,r")))]
  "TARGET_64BIT && ix86_check_movabs (insn, 1)"
  "@
   movabs{<imodesuffix>}\t{%P1, %0|%0, %P1}
   mov{<imodesuffix>}\t{%a1, %0|%0, %a1}"
  [(set_attr "type" "imov")
   (set_attr "modrm" "0,*")
   (set_attr "length_address" "8,0")
   (set_attr "length_immediate" "0")
   (set_attr "memory" "load")
   (set_attr "mode" "<MODE>")])

(define_insn "*swap<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "+r")
	(match_operand:SWI48 1 "register_operand" "+r"))
   (set (match_dup 1)
	(match_dup 0))]
  ""
  "xchg{<imodesuffix>}\t%1, %0"
  [(set_attr "type" "imov")
   (set_attr "mode" "<MODE>")
   (set_attr "pent_pair" "np")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "double")])

(define_insn "*swap<mode>_1"
  [(set (match_operand:SWI12 0 "register_operand" "+r")
	(match_operand:SWI12 1 "register_operand" "+r"))
   (set (match_dup 1)
	(match_dup 0))]
  "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)"
  "xchg{l}\t%k1, %k0"
  [(set_attr "type" "imov")
   (set_attr "mode" "SI")
   (set_attr "pent_pair" "np")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "double")])

;; Not added amdfam10_decode since TARGET_PARTIAL_REG_STALL
;; is disabled for AMDFAM10
(define_insn "*swap<mode>_2"
  [(set (match_operand:SWI12 0 "register_operand" "+<r>")
	(match_operand:SWI12 1 "register_operand" "+<r>"))
   (set (match_dup 1)
	(match_dup 0))]
  "TARGET_PARTIAL_REG_STALL"
  "xchg{<imodesuffix>}\t%1, %0"
  [(set_attr "type" "imov")
   (set_attr "mode" "<MODE>")
   (set_attr "pent_pair" "np")
   (set_attr "athlon_decode" "vector")])

(define_expand "movstrict<mode>"
  [(set (strict_low_part (match_operand:SWI12 0 "nonimmediate_operand" ""))
	(match_operand:SWI12 1 "general_operand" ""))]
  ""
{
  if (TARGET_PARTIAL_REG_STALL && optimize_function_for_speed_p (cfun))
    FAIL;
  /* Don't generate memory->memory moves, go through a register */
  if (MEM_P (operands[0]) && MEM_P (operands[1]))
    operands[1] = force_reg (<MODE>mode, operands[1]);
})

(define_insn "*movstrict<mode>_1"
  [(set (strict_low_part
	  (match_operand:SWI12 0 "nonimmediate_operand" "+<r>m,<r>"))
	(match_operand:SWI12 1 "general_operand" "<r>n,m"))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "mov{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "imov")
   (set_attr "mode" "<MODE>")])

(define_insn "*movstrict<mode>_xor"
  [(set (strict_low_part (match_operand:SWI12 0 "register_operand" "+<r>"))
	(match_operand:SWI12 1 "const0_operand" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  "xor{<imodesuffix>}\t%0, %0"
  [(set_attr "type" "alu1")
   (set_attr "mode" "<MODE>")
   (set_attr "length_immediate" "0")])

(define_insn "*mov<mode>_extv_1"
  [(set (match_operand:SWI24 0 "register_operand" "=R")
	(sign_extract:SWI24 (match_operand 1 "ext_register_operand" "Q")
			    (const_int 8)
			    (const_int 8)))]
  ""
  "movs{bl|x}\t{%h1, %k0|%k0, %h1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")])

(define_insn "*movqi_extv_1_rex64"
  [(set (match_operand:QI 0 "register_operand" "=Q,?R")
        (sign_extract:QI (match_operand 1 "ext_register_operand" "Q,Q")
                         (const_int 8)
                         (const_int 8)))]
  "TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      return "movs{bl|x}\t{%h1, %k0|%k0, %h1}";
    default:
      return "mov{b}\t{%h1, %0|%0, %h1}";
    }
}
  [(set (attr "type")
     (if_then_else (and (match_operand:QI 0 "register_operand" "")
			(ior (not (match_operand:QI 0 "q_regs_operand" ""))
			     (ne (symbol_ref "TARGET_MOVX")
				 (const_int 0))))
	(const_string "imovx")
	(const_string "imov")))
   (set (attr "mode")
     (if_then_else (eq_attr "type" "imovx")
	(const_string "SI")
	(const_string "QI")))])

(define_insn "*movqi_extv_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=Qm,?r")
        (sign_extract:QI (match_operand 1 "ext_register_operand" "Q,Q")
                         (const_int 8)
                         (const_int 8)))]
  "!TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      return "movs{bl|x}\t{%h1, %k0|%k0, %h1}";
    default:
      return "mov{b}\t{%h1, %0|%0, %h1}";
    }
}
  [(set (attr "type")
     (if_then_else (and (match_operand:QI 0 "register_operand" "")
			(ior (not (match_operand:QI 0 "q_regs_operand" ""))
			     (ne (symbol_ref "TARGET_MOVX")
				 (const_int 0))))
	(const_string "imovx")
	(const_string "imov")))
   (set (attr "mode")
     (if_then_else (eq_attr "type" "imovx")
	(const_string "SI")
	(const_string "QI")))])

(define_insn "*mov<mode>_extzv_1"
  [(set (match_operand:SWI48 0 "register_operand" "=R")
	(zero_extract:SWI48 (match_operand 1 "ext_register_operand" "Q")
			    (const_int 8)
			    (const_int 8)))]
  ""
  "movz{bl|x}\t{%h1, %k0|%k0, %h1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")])

(define_insn "*movqi_extzv_2_rex64"
  [(set (match_operand:QI 0 "register_operand" "=Q,?R")
        (subreg:QI
	  (zero_extract:SI (match_operand 1 "ext_register_operand" "Q,Q")
			   (const_int 8)
			   (const_int 8)) 0))]
  "TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      return "movz{bl|x}\t{%h1, %k0|%k0, %h1}";
    default:
      return "mov{b}\t{%h1, %0|%0, %h1}";
    }
}
  [(set (attr "type")
     (if_then_else (ior (not (match_operand:QI 0 "q_regs_operand" ""))
			(ne (symbol_ref "TARGET_MOVX")
			    (const_int 0)))
	(const_string "imovx")
	(const_string "imov")))
   (set (attr "mode")
     (if_then_else (eq_attr "type" "imovx")
	(const_string "SI")
	(const_string "QI")))])

(define_insn "*movqi_extzv_2"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=Qm,?R")
        (subreg:QI
	  (zero_extract:SI (match_operand 1 "ext_register_operand" "Q,Q")
			   (const_int 8)
			   (const_int 8)) 0))]
  "!TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      return "movz{bl|x}\t{%h1, %k0|%k0, %h1}";
    default:
      return "mov{b}\t{%h1, %0|%0, %h1}";
    }
}
  [(set (attr "type")
     (if_then_else (and (match_operand:QI 0 "register_operand" "")
			(ior (not (match_operand:QI 0 "q_regs_operand" ""))
			     (ne (symbol_ref "TARGET_MOVX")
				 (const_int 0))))
	(const_string "imovx")
	(const_string "imov")))
   (set (attr "mode")
     (if_then_else (eq_attr "type" "imovx")
	(const_string "SI")
	(const_string "QI")))])

(define_expand "mov<mode>_insv_1"
  [(set (zero_extract:SWI48 (match_operand 0 "ext_register_operand" "")
			    (const_int 8)
			    (const_int 8))
	(match_operand:SWI48 1 "nonmemory_operand" ""))])

(define_insn "*mov<mode>_insv_1_rex64"
  [(set (zero_extract:SWI48x (match_operand 0 "ext_register_operand" "+Q")
			     (const_int 8)
			     (const_int 8))
	(match_operand:SWI48x 1 "nonmemory_operand" "Qn"))]
  "TARGET_64BIT"
  "mov{b}\t{%b1, %h0|%h0, %b1}"
  [(set_attr "type" "imov")
   (set_attr "mode" "QI")])

(define_insn "*movsi_insv_1"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "+Q")
			 (const_int 8)
			 (const_int 8))
	(match_operand:SI 1 "general_operand" "Qmn"))]
  "!TARGET_64BIT"
  "mov{b}\t{%b1, %h0|%h0, %b1}"
  [(set_attr "type" "imov")
   (set_attr "mode" "QI")])

(define_insn "*movqi_insv_2"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "+Q")
			 (const_int 8)
			 (const_int 8))
	(lshiftrt:SI (match_operand:SI 1 "register_operand" "Q")
		     (const_int 8)))]
  ""
  "mov{b}\t{%h1, %h0|%h0, %h1}"
  [(set_attr "type" "imov")
   (set_attr "mode" "QI")])

;; Floating point push instructions.

(define_insn "*pushtf"
  [(set (match_operand:TF 0 "push_operand" "=<,<,<")
	(match_operand:TF 1 "general_no_elim_operand" "x,Fo,*r"))]
  "TARGET_SSE2"
{
  /* This insn should be already split before reg-stack.  */
  gcc_unreachable ();
}
  [(set_attr "type" "multi")
   (set_attr "unit" "sse,*,*")
   (set_attr "mode" "TF,SI,SI")])

(define_split
  [(set (match_operand:TF 0 "push_operand" "")
	(match_operand:TF 1 "sse_reg_operand" ""))]
  "TARGET_SSE2 && reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (const_int -16)))
   (set (mem:TF (reg:P SP_REG)) (match_dup 1))])

(define_split
  [(set (match_operand:TF 0 "push_operand" "")
	(match_operand:TF 1 "general_operand" ""))]
  "TARGET_SSE2 && reload_completed
   && !SSE_REG_P (operands[1])"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*pushxf"
  [(set (match_operand:XF 0 "push_operand" "=<,<")
	(match_operand:XF 1 "general_no_elim_operand" "f,ro"))]
  "optimize_function_for_speed_p (cfun)"
{
  /* This insn should be already split before reg-stack.  */
  gcc_unreachable ();
}
  [(set_attr "type" "multi")
   (set_attr "unit" "i387,*")
   (set_attr "mode" "XF,SI")])

;; Size of pushxf is 3 (for sub) + 2 (for fstp) + memory operand size.
;; Size of pushxf using integer instructions is 3+3*memory operand size
;; Pushing using integer instructions is longer except for constants
;; and direct memory references (assuming that any given constant is pushed
;; only once, but this ought to be handled elsewhere).

(define_insn "*pushxf_nointeger"
  [(set (match_operand:XF 0 "push_operand" "=X,X,X")
	(match_operand:XF 1 "general_no_elim_operand" "f,Fo,*r"))]
  "optimize_function_for_size_p (cfun)"
{
  /* This insn should be already split before reg-stack.  */
  gcc_unreachable ();
}
  [(set_attr "type" "multi")
   (set_attr "unit" "i387,*,*")
   (set_attr "mode" "XF,SI,SI")])

(define_split
  [(set (match_operand:XF 0 "push_operand" "")
	(match_operand:XF 1 "fp_register_operand" ""))]
  "reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (match_dup 2)))
   (set (mem:XF (reg:P SP_REG)) (match_dup 1))]
  "operands[2] = GEN_INT (-GET_MODE_SIZE (XFmode));")

(define_split
  [(set (match_operand:XF 0 "push_operand" "")
	(match_operand:XF 1 "general_operand" ""))]
  "reload_completed
   && !FP_REG_P (operands[1])"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*pushdf"
  [(set (match_operand:DF 0 "push_operand" "=<,<,<")
	(match_operand:DF 1 "general_no_elim_operand" "f,rFo,Y2"))]
  "TARGET_64BIT || TARGET_INTEGER_DFMODE_MOVES"
{
  /* This insn should be already split before reg-stack.  */
  gcc_unreachable ();
}
  [(set_attr "type" "multi")
   (set_attr "unit" "i387,*,*")
   (set_attr "mode" "DF,SI,DF")])

;; Size of pushdf is 3 (for sub) + 2 (for fstp) + memory operand size.
;; Size of pushdf using integer instructions is 2+2*memory operand size
;; On the average, pushdf using integers can be still shorter.  Allow this
;; pattern for optimize_size too.

(define_insn "*pushdf_nointeger"
  [(set (match_operand:DF 0 "push_operand" "=<,<,<,<")
	(match_operand:DF 1 "general_no_elim_operand" "f,Fo,*r,Y2"))]
  "!(TARGET_64BIT || TARGET_INTEGER_DFMODE_MOVES)"
{
  /* This insn should be already split before reg-stack.  */
  gcc_unreachable ();
}
  [(set_attr "type" "multi")
   (set_attr "unit" "i387,*,*,*")
   (set_attr "mode" "DF,SI,SI,DF")])

;; %%% Kill this when call knows how to work this out.
(define_split
  [(set (match_operand:DF 0 "push_operand" "")
	(match_operand:DF 1 "any_fp_register_operand" ""))]
  "reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (const_int -8)))
   (set (mem:DF (reg:P SP_REG)) (match_dup 1))])

(define_split
  [(set (match_operand:DF 0 "push_operand" "")
	(match_operand:DF 1 "general_operand" ""))]
  "reload_completed
   && !ANY_FP_REG_P (operands[1])"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*pushsf_rex64"
  [(set (match_operand:SF 0 "push_operand" "=X,X,X")
	(match_operand:SF 1 "nonmemory_no_elim_operand" "f,rF,x"))]
  "TARGET_64BIT"
{
  /* Anything else should be already split before reg-stack.  */
  gcc_assert (which_alternative == 1);
  return "push{q}\t%q1";
}
  [(set_attr "type" "multi,push,multi")
   (set_attr "unit" "i387,*,*")
   (set_attr "mode" "SF,DI,SF")])

(define_insn "*pushsf"
  [(set (match_operand:SF 0 "push_operand" "=<,<,<")
	(match_operand:SF 1 "general_no_elim_operand" "f,rFm,x"))]
  "!TARGET_64BIT"
{
  /* Anything else should be already split before reg-stack.  */
  gcc_assert (which_alternative == 1);
  return "push{l}\t%1";
}
  [(set_attr "type" "multi,push,multi")
   (set_attr "unit" "i387,*,*")
   (set_attr "mode" "SF,SI,SF")])

(define_split
  [(set (match_operand:SF 0 "push_operand" "")
	(match_operand:SF 1 "memory_operand" ""))]
  "reload_completed
   && MEM_P (operands[1])
   && (operands[2] = find_constant_src (insn))"
  [(set (match_dup 0)
	(match_dup 2))])

;; %%% Kill this when call knows how to work this out.
(define_split
  [(set (match_operand:SF 0 "push_operand" "")
	(match_operand:SF 1 "any_fp_register_operand" ""))]
  "reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (match_dup 2)))
   (set (mem:SF (reg:P SP_REG)) (match_dup 1))]
  "operands[2] = GEN_INT (-GET_MODE_SIZE (<MODE>mode));")

;; Floating point move instructions.

(define_expand "movtf"
  [(set (match_operand:TF 0 "nonimmediate_operand" "")
	(match_operand:TF 1 "nonimmediate_operand" ""))]
  "TARGET_SSE2"
{
  ix86_expand_move (TFmode, operands);
  DONE;
})

(define_expand "mov<mode>"
  [(set (match_operand:X87MODEF 0 "nonimmediate_operand" "")
	(match_operand:X87MODEF 1 "general_operand" ""))]
  ""
  "ix86_expand_move (<MODE>mode, operands); DONE;")

(define_insn "*movtf_internal"
  [(set (match_operand:TF 0 "nonimmediate_operand" "=x,m,x,?r,?o")
	(match_operand:TF 1 "general_operand" "xm,x,C,roF,Fr"))]
  "TARGET_SSE2
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      if (get_attr_mode (insn) == MODE_V4SF)
	return "%vmovaps\t{%1, %0|%0, %1}";
      else
	return "%vmovdqa\t{%1, %0|%0, %1}";
    case 2:
      if (get_attr_mode (insn) == MODE_V4SF)
	return "%vxorps\t%0, %d0";
      else
	return "%vpxor\t%0, %d0";
    case 3:
    case 4:
	return "#";
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "ssemov,ssemov,sselog1,*,*")
   (set_attr "prefix" "maybe_vex,maybe_vex,maybe_vex,*,*")
   (set (attr "mode")
        (cond [(eq_attr "alternative" "0,2")
		 (if_then_else
		   (ne (symbol_ref "optimize_function_for_size_p (cfun)")
		       (const_int 0))
		   (const_string "V4SF")
		   (const_string "TI"))
	       (eq_attr "alternative" "1")
		 (if_then_else
		   (ior (ne (symbol_ref "TARGET_SSE_TYPELESS_STORES")
			    (const_int 0))
			(ne (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0)))
		   (const_string "V4SF")
		   (const_string "TI"))]
	       (const_string "DI")))])

(define_split
  [(set (match_operand:TF 0 "nonimmediate_operand" "")
        (match_operand:TF 1 "general_operand" ""))]
  "reload_completed
   && !(SSE_REG_P (operands[0]) || SSE_REG_P (operands[1]))"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*movxf_internal"
  [(set (match_operand:XF 0 "nonimmediate_operand" "=f,m,f,r,o")
	(match_operand:XF 1 "general_operand" "fm,f,G,roF,Fr"))]
  "optimize_function_for_speed_p (cfun)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))
   && (reload_in_progress || reload_completed
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || memory_operand (operands[0], XFmode))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3: case 4:
      return "#";

    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,multi,multi")
   (set_attr "mode" "XF,XF,XF,SI,SI")])

;; Do not use integer registers when optimizing for size
(define_insn "*movxf_internal_nointeger"
  [(set (match_operand:XF 0 "nonimmediate_operand" "=f,m,f,*r,o")
	(match_operand:XF 1 "general_operand" "fm,f,G,*roF,F*r"))]
  "optimize_function_for_size_p (cfun)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))
   && (reload_in_progress || reload_completed
       || standard_80387_constant_p (operands[1])
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || memory_operand (operands[0], XFmode))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3: case 4:
      return "#";
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,multi,multi")
   (set_attr "mode" "XF,XF,XF,SI,SI")])

(define_split
  [(set (match_operand:XF 0 "nonimmediate_operand" "")
	(match_operand:XF 1 "general_operand" ""))]
  "reload_completed
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))
   && ! (FP_REG_P (operands[0]) ||
	 (GET_CODE (operands[0]) == SUBREG
	  && FP_REG_P (SUBREG_REG (operands[0]))))
   && ! (FP_REG_P (operands[1]) ||
	 (GET_CODE (operands[1]) == SUBREG
	  && FP_REG_P (SUBREG_REG (operands[1]))))"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*movdf_internal_rex64"
  [(set (match_operand:DF 0 "nonimmediate_operand"
		"=f,m,f,r  ,m ,Y2*x,Y2*x,Y2*x,m   ,Yi,r ")
	(match_operand:DF 1 "general_operand"
		"fm,f,G,rmF,Fr,C   ,Y2*x,m   ,Y2*x,r ,Yi"))]
  "TARGET_64BIT && !(MEM_P (operands[0]) && MEM_P (operands[1]))
   && (reload_in_progress || reload_completed
       || (ix86_cmodel == CM_MEDIUM || ix86_cmodel == CM_LARGE)
       || (!(TARGET_SSE2 && TARGET_SSE_MATH)
           && optimize_function_for_size_p (cfun)
	   && standard_80387_constant_p (operands[1]))
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || memory_operand (operands[0], DFmode))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3:
    case 4:
      return "#";

    case 5:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "%vxorps\t%0, %d0";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vxorps\t%0, %d0";
	  else
	    return "%vxorpd\t%0, %d0";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vxorps\t%0, %d0";
	  else
	    return "%vpxor\t%0, %d0";
	default:
	  gcc_unreachable ();
	}
    case 6:
    case 7:
    case 8:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "%vmovaps\t{%1, %0|%0, %1}";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	  else
	    return "%vmovapd\t{%1, %0|%0, %1}";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	  else
	    return "%vmovdqa\t{%1, %0|%0, %1}";
	case MODE_DI:
	  return "%vmovq\t{%1, %0|%0, %1}";
	case MODE_DF:
	  if (TARGET_AVX)
	    {
	      if (REG_P (operands[0]) && REG_P (operands[1]))
		return "vmovsd\t{%1, %0, %0|%0, %0, %1}";
	      else
		return "vmovsd\t{%1, %0|%0, %1}";
	    }
	  else
	    return "movsd\t{%1, %0|%0, %1}";
	case MODE_V1DF:
	  return "%vmovlpd\t{%1, %d0|%d0, %1}";
	case MODE_V2SF:
	  return "%vmovlps\t{%1, %d0|%d0, %1}";
	default:
	  gcc_unreachable ();
	}

    case 9:
    case 10:
    return "%vmovd\t{%1, %0|%0, %1}";

    default:
      gcc_unreachable();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,multi,multi,sselog1,ssemov,ssemov,ssemov,ssemov,ssemov")
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "0,1,2,3,4")
       (const_string "orig")
       (const_string "maybe_vex")))
   (set (attr "prefix_data16")
     (if_then_else (eq_attr "mode" "V1DF")
       (const_string "1")
       (const_string "*")))
   (set (attr "mode")
        (cond [(eq_attr "alternative" "0,1,2")
		 (const_string "DF")
	       (eq_attr "alternative" "3,4,9,10")
		 (const_string "DI")

	       /* For SSE1, we have many fewer alternatives.  */
	       (eq (symbol_ref "TARGET_SSE2") (const_int 0))
		 (cond [(eq_attr "alternative" "5,6")
			  (const_string "V4SF")
		       ]
		   (const_string "V2SF"))

	       /* xorps is one byte shorter.  */
	       (eq_attr "alternative" "5")
		 (cond [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0))
			  (const_string "V4SF")
			(ne (symbol_ref "TARGET_SSE_LOAD0_BY_PXOR")
			    (const_int 0))
			  (const_string "TI")
		       ]
		       (const_string "V2DF"))

	       /* For architectures resolving dependencies on
		  whole SSE registers use APD move to break dependency
		  chains, otherwise use short move to avoid extra work.

		  movaps encodes one byte shorter.  */
	       (eq_attr "alternative" "6")
		 (cond
		   [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
		        (const_int 0))
		      (const_string "V4SF")
		    (ne (symbol_ref "TARGET_SSE_PARTIAL_REG_DEPENDENCY")
		        (const_int 0))
		      (const_string "V2DF")
		   ]
		   (const_string "DF"))
	       /* For architectures resolving dependencies on register
		  parts we may avoid extra work to zero out upper part
		  of register.  */
	       (eq_attr "alternative" "7")
		 (if_then_else
		   (ne (symbol_ref "TARGET_SSE_SPLIT_REGS")
		       (const_int 0))
		   (const_string "V1DF")
		   (const_string "DF"))
	      ]
	      (const_string "DF")))])

(define_insn "*movdf_internal"
  [(set (match_operand:DF 0 "nonimmediate_operand"
		"=f,m,f,r  ,o ,Y2*x,Y2*x,Y2*x,m   ")
	(match_operand:DF 1 "general_operand"
		"fm,f,G,roF,Fr,C   ,Y2*x,m   ,Y2*x"))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))
   && optimize_function_for_speed_p (cfun)
   && TARGET_INTEGER_DFMODE_MOVES
   && (reload_in_progress || reload_completed
       || (ix86_cmodel == CM_MEDIUM || ix86_cmodel == CM_LARGE)
       || (!(TARGET_SSE2 && TARGET_SSE_MATH)
           && optimize_function_for_size_p (cfun)
	   && standard_80387_constant_p (operands[1]))
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || memory_operand (operands[0], DFmode))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3:
    case 4:
      return "#";

    case 5:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "xorps\t%0, %0";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "xorps\t%0, %0";
	  else
	    return "xorpd\t%0, %0";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "xorps\t%0, %0";
	  else
	    return "pxor\t%0, %0";
	default:
	  gcc_unreachable ();
	}
    case 6:
    case 7:
    case 8:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "movaps\t{%1, %0|%0, %1}";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "movaps\t{%1, %0|%0, %1}";
	  else
	    return "movapd\t{%1, %0|%0, %1}";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "movaps\t{%1, %0|%0, %1}";
	  else
	    return "movdqa\t{%1, %0|%0, %1}";
	case MODE_DI:
	  return "movq\t{%1, %0|%0, %1}";
	case MODE_DF:
	  return "movsd\t{%1, %0|%0, %1}";
	case MODE_V1DF:
	  return "movlpd\t{%1, %0|%0, %1}";
	case MODE_V2SF:
	  return "movlps\t{%1, %0|%0, %1}";
	default:
	  gcc_unreachable ();
	}

    default:
      gcc_unreachable();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,multi,multi,sselog1,ssemov,ssemov,ssemov")
   (set (attr "prefix_data16")
     (if_then_else (eq_attr "mode" "V1DF")
       (const_string "1")
       (const_string "*")))
   (set (attr "mode")
        (cond [(eq_attr "alternative" "0,1,2")
		 (const_string "DF")
	       (eq_attr "alternative" "3,4")
		 (const_string "SI")

	       /* For SSE1, we have many fewer alternatives.  */
	       (eq (symbol_ref "TARGET_SSE2") (const_int 0))
		 (cond [(eq_attr "alternative" "5,6")
			  (const_string "V4SF")
		       ]
		   (const_string "V2SF"))

	       /* xorps is one byte shorter.  */
	       (eq_attr "alternative" "5")
		 (cond [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0))
			  (const_string "V4SF")
			(ne (symbol_ref "TARGET_SSE_LOAD0_BY_PXOR")
			    (const_int 0))
			  (const_string "TI")
		       ]
		       (const_string "V2DF"))

	       /* For architectures resolving dependencies on
		  whole SSE registers use APD move to break dependency
		  chains, otherwise use short move to avoid extra work.

		  movaps encodes one byte shorter.  */
	       (eq_attr "alternative" "6")
		 (cond
		   [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
		        (const_int 0))
		      (const_string "V4SF")
		    (ne (symbol_ref "TARGET_SSE_PARTIAL_REG_DEPENDENCY")
		        (const_int 0))
		      (const_string "V2DF")
		   ]
		   (const_string "DF"))
	       /* For architectures resolving dependencies on register
		  parts we may avoid extra work to zero out upper part
		  of register.  */
	       (eq_attr "alternative" "7")
		 (if_then_else
		   (ne (symbol_ref "TARGET_SSE_SPLIT_REGS")
		       (const_int 0))
		   (const_string "V1DF")
		   (const_string "DF"))
	      ]
	      (const_string "DF")))])

;; Moving is usually shorter when only FP registers are used. This separate
;; movdf pattern avoids the use of integer registers for FP operations
;; when optimizing for size.

(define_insn "*movdf_internal_nointeger"
  [(set (match_operand:DF 0 "nonimmediate_operand"
			"=f,m,f,*r  ,o  ,Y2*x,Y2*x,Y2*x ,m  ")
	(match_operand:DF 1 "general_operand"
			"fm,f,G,*roF,*Fr,C   ,Y2*x,mY2*x,Y2*x"))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))
   && ((optimize_function_for_size_p (cfun)
       || !TARGET_INTEGER_DFMODE_MOVES) && !TARGET_64BIT)
   && (reload_in_progress || reload_completed
       || (ix86_cmodel == CM_MEDIUM || ix86_cmodel == CM_LARGE)
       || (!(TARGET_SSE2 && TARGET_SSE_MATH)
           && optimize_function_for_size_p (cfun)
           && !memory_operand (operands[0], DFmode)
	   && standard_80387_constant_p (operands[1]))
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || ((optimize_function_for_size_p (cfun)
            || !TARGET_MEMORY_MISMATCH_STALL
	    || reload_in_progress || reload_completed)
 	   && memory_operand (operands[0], DFmode)))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3:
    case 4:
      return "#";

    case 5:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "%vxorps\t%0, %d0";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vxorps\t%0, %d0";
	  else
	    return "%vxorpd\t%0, %d0";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vxorps\t%0, %d0";
	  else
	    return "%vpxor\t%0, %d0";
	default:
	  gcc_unreachable ();
	}
    case 6:
    case 7:
    case 8:
      switch (get_attr_mode (insn))
	{
	case MODE_V4SF:
	  return "%vmovaps\t{%1, %0|%0, %1}";
	case MODE_V2DF:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	  else
	    return "%vmovapd\t{%1, %0|%0, %1}";
	case MODE_TI:
	  if (TARGET_SSE_PACKED_SINGLE_INSN_OPTIMAL)
	    return "%vmovaps\t{%1, %0|%0, %1}";
	  else
	    return "%vmovdqa\t{%1, %0|%0, %1}";
	case MODE_DI:
	  return "%vmovq\t{%1, %0|%0, %1}";
	case MODE_DF:
	  if (TARGET_AVX)
	    {
	      if (REG_P (operands[0]) && REG_P (operands[1]))
		return "vmovsd\t{%1, %0, %0|%0, %0, %1}";
	      else
		return "vmovsd\t{%1, %0|%0, %1}";
	    }
	  else
	    return "movsd\t{%1, %0|%0, %1}";
	case MODE_V1DF:
	  if (TARGET_AVX)
	    {
	      if (REG_P (operands[0]))
		return "vmovlpd\t{%1, %0, %0|%0, %0, %1}";
	      else
		return "vmovlpd\t{%1, %0|%0, %1}";
	    }
	  else
	    return "movlpd\t{%1, %0|%0, %1}";
	case MODE_V2SF:
	  if (TARGET_AVX)
	    {
	      if (REG_P (operands[0]))
		return "vmovlps\t{%1, %0, %0|%0, %0, %1}";
	      else
		return "vmovlps\t{%1, %0|%0, %1}";
	    }
	  else
	    return "movlps\t{%1, %0|%0, %1}";
	default:
	  gcc_unreachable ();
	}

    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,multi,multi,sselog1,ssemov,ssemov,ssemov")
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "0,1,2,3,4")
       (const_string "orig")
       (const_string "maybe_vex")))
   (set (attr "prefix_data16")
     (if_then_else (eq_attr "mode" "V1DF")
       (const_string "1")
       (const_string "*")))
   (set (attr "mode")
        (cond [(eq_attr "alternative" "0,1,2")
		 (const_string "DF")
	       (eq_attr "alternative" "3,4")
		 (const_string "SI")

	       /* For SSE1, we have many fewer alternatives.  */
	       (eq (symbol_ref "TARGET_SSE2") (const_int 0))
		 (cond [(eq_attr "alternative" "5,6")
			  (const_string "V4SF")
		       ]
		   (const_string "V2SF"))

	       /* xorps is one byte shorter.  */
	       (eq_attr "alternative" "5")
		 (cond [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0))
			  (const_string "V4SF")
			(ne (symbol_ref "TARGET_SSE_LOAD0_BY_PXOR")
			    (const_int 0))
			  (const_string "TI")
		       ]
		       (const_string "V2DF"))

	       /* For architectures resolving dependencies on
		  whole SSE registers use APD move to break dependency
		  chains, otherwise use short move to avoid extra work.

		  movaps encodes one byte shorter.  */
	       (eq_attr "alternative" "6")
		 (cond
		   [(ne (symbol_ref "optimize_function_for_size_p (cfun)")
		        (const_int 0))
		      (const_string "V4SF")
		    (ne (symbol_ref "TARGET_SSE_PARTIAL_REG_DEPENDENCY")
		        (const_int 0))
		      (const_string "V2DF")
		   ]
		   (const_string "DF"))
	       /* For architectures resolving dependencies on register
		  parts we may avoid extra work to zero out upper part
		  of register.  */
	       (eq_attr "alternative" "7")
		 (if_then_else
		   (ne (symbol_ref "TARGET_SSE_SPLIT_REGS")
		       (const_int 0))
		   (const_string "V1DF")
		   (const_string "DF"))
	      ]
	      (const_string "DF")))])

(define_split
  [(set (match_operand:DF 0 "nonimmediate_operand" "")
	(match_operand:DF 1 "general_operand" ""))]
  "reload_completed
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))
   && ! (ANY_FP_REG_P (operands[0]) ||
	 (GET_CODE (operands[0]) == SUBREG
	  && ANY_FP_REG_P (SUBREG_REG (operands[0]))))
   && ! (ANY_FP_REG_P (operands[1]) ||
	 (GET_CODE (operands[1]) == SUBREG
	  && ANY_FP_REG_P (SUBREG_REG (operands[1]))))"
  [(const_int 0)]
  "ix86_split_long_move (operands); DONE;")

(define_insn "*movsf_internal"
  [(set (match_operand:SF 0 "nonimmediate_operand"
	  "=f,m,f,r  ,m ,x,x,x ,m,!*y,!m,!*y,?Yi,?r,!*Ym,!r")
	(match_operand:SF 1 "general_operand"
	  "fm,f,G,rmF,Fr,C,x,xm,x,m  ,*y,*y ,r  ,Yi,r   ,*Ym"))]
  "!(MEM_P (operands[0]) && MEM_P (operands[1]))
   && (reload_in_progress || reload_completed
       || (ix86_cmodel == CM_MEDIUM || ix86_cmodel == CM_LARGE)
       || (!TARGET_SSE_MATH && optimize_function_for_size_p (cfun)
	   && standard_80387_constant_p (operands[1]))
       || GET_CODE (operands[1]) != CONST_DOUBLE
       || memory_operand (operands[0], SFmode))"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return standard_80387_constant_opcode (operands[1]);

    case 3:
    case 4:
      return "mov{l}\t{%1, %0|%0, %1}";
    case 5:
      if (get_attr_mode (insn) == MODE_TI)
	return "%vpxor\t%0, %d0";
      else
	return "%vxorps\t%0, %d0";
    case 6:
      if (get_attr_mode (insn) == MODE_V4SF)
	return "%vmovaps\t{%1, %0|%0, %1}";
      else
	return "%vmovss\t{%1, %d0|%d0, %1}";
    case 7:
      if (TARGET_AVX)
	return REG_P (operands[1]) ? "vmovss\t{%1, %0, %0|%0, %0, %1}"
				   : "vmovss\t{%1, %0|%0, %1}";
      else
	return "movss\t{%1, %0|%0, %1}";
    case 8:
      return "%vmovss\t{%1, %0|%0, %1}";

    case 9: case 10: case 14: case 15:
      return "movd\t{%1, %0|%0, %1}";
    case 12: case 13:
      return "%vmovd\t{%1, %0|%0, %1}";

    case 11:
      return "movq\t{%1, %0|%0, %1}";

    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,fmov,fmov,imov,imov,sselog1,ssemov,ssemov,ssemov,mmxmov,mmxmov,mmxmov,ssemov,ssemov,mmxmov,mmxmov")
   (set (attr "prefix")
     (if_then_else (eq_attr "alternative" "5,6,7,8,12,13")
       (const_string "maybe_vex")
       (const_string "orig")))
   (set (attr "mode")
        (cond [(eq_attr "alternative" "3,4,9,10")
		 (const_string "SI")
	       (eq_attr "alternative" "5")
		 (if_then_else
		   (and (and (ne (symbol_ref "TARGET_SSE_LOAD0_BY_PXOR")
			    	 (const_int 0))
			     (ne (symbol_ref "TARGET_SSE2")
				 (const_int 0)))
			(eq (symbol_ref "optimize_function_for_size_p (cfun)")
			    (const_int 0)))
		   (const_string "TI")
		   (const_string "V4SF"))
	       /* For architectures resolving dependencies on
		  whole SSE registers use APS move to break dependency
		  chains, otherwise use short move to avoid extra work.

		  Do the same for architectures resolving dependencies on
		  the parts.  While in DF mode it is better to always handle
		  just register parts, the SF mode is different due to lack
		  of instructions to load just part of the register.  It is
		  better to maintain the whole registers in single format
		  to avoid problems on using packed logical operations.  */
	       (eq_attr "alternative" "6")
		 (if_then_else
		   (ior (ne (symbol_ref "TARGET_SSE_PARTIAL_REG_DEPENDENCY")
			    (const_int 0))
			(ne (symbol_ref "TARGET_SSE_SPLIT_REGS")
			    (const_int 0)))
		   (const_string "V4SF")
		   (const_string "SF"))
	       (eq_attr "alternative" "11")
		 (const_string "DI")]
	       (const_string "SF")))])

(define_split
  [(set (match_operand 0 "register_operand" "")
	(match_operand 1 "memory_operand" ""))]
  "reload_completed
   && MEM_P (operands[1])
   && (GET_MODE (operands[0]) == TFmode
       || GET_MODE (operands[0]) == XFmode
       || GET_MODE (operands[0]) == DFmode
       || GET_MODE (operands[0]) == SFmode)
   && (operands[2] = find_constant_src (insn))"
  [(set (match_dup 0) (match_dup 2))]
{
  rtx c = operands[2];
  rtx r = operands[0];

  if (GET_CODE (r) == SUBREG)
    r = SUBREG_REG (r);

  if (SSE_REG_P (r))
    {
      if (!standard_sse_constant_p (c))
	FAIL;
    }
  else if (FP_REG_P (r))
    {
      if (!standard_80387_constant_p (c))
	FAIL;
    }
  else if (MMX_REG_P (r))
    FAIL;
})

(define_split
  [(set (match_operand 0 "register_operand" "")
	(float_extend (match_operand 1 "memory_operand" "")))]
  "reload_completed
   && MEM_P (operands[1])
   && (GET_MODE (operands[0]) == TFmode
       || GET_MODE (operands[0]) == XFmode
       || GET_MODE (operands[0]) == DFmode
       || GET_MODE (operands[0]) == SFmode)
   && (operands[2] = find_constant_src (insn))"
  [(set (match_dup 0) (match_dup 2))]
{
  rtx c = operands[2];
  rtx r = operands[0];

  if (GET_CODE (r) == SUBREG)
    r = SUBREG_REG (r);

  if (SSE_REG_P (r))
    {
      if (!standard_sse_constant_p (c))
	FAIL;
    }
  else if (FP_REG_P (r))
    {
      if (!standard_80387_constant_p (c))
	FAIL;
    }
  else if (MMX_REG_P (r))
    FAIL;
})

;; Split the load of -0.0 or -1.0 into fldz;fchs or fld1;fchs sequence
(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(match_operand:X87MODEF 1 "immediate_operand" ""))]
  "reload_completed && FP_REGNO_P (REGNO (operands[0]))
   && (standard_80387_constant_p (operands[1]) == 8
       || standard_80387_constant_p (operands[1]) == 9)"
  [(set (match_dup 0)(match_dup 1))
   (set (match_dup 0)
	(neg:X87MODEF (match_dup 0)))]
{
  REAL_VALUE_TYPE r;

  REAL_VALUE_FROM_CONST_DOUBLE (r, operands[1]);
  if (real_isnegzero (&r))
    operands[1] = CONST0_RTX (<MODE>mode);
  else
    operands[1] = CONST1_RTX (<MODE>mode);
})

(define_insn "swapxf"
  [(set (match_operand:XF 0 "register_operand" "+f")
	(match_operand:XF 1 "register_operand" "+f"))
   (set (match_dup 1)
	(match_dup 0))]
  "TARGET_80387"
{
  if (STACK_TOP_P (operands[0]))
    return "fxch\t%1";
  else
    return "fxch\t%0";
}
  [(set_attr "type" "fxch")
   (set_attr "mode" "XF")])

(define_insn "*swap<mode>"
  [(set (match_operand:MODEF 0 "fp_register_operand" "+f")
	(match_operand:MODEF 1 "fp_register_operand" "+f"))
   (set (match_dup 1)
	(match_dup 0))]
  "TARGET_80387 || reload_completed"
{
  if (STACK_TOP_P (operands[0]))
    return "fxch\t%1";
  else
    return "fxch\t%0";
}
  [(set_attr "type" "fxch")
   (set_attr "mode" "<MODE>")])

;; Zero extension instructions

(define_expand "zero_extendsidi2"
  [(set (match_operand:DI 0 "nonimmediate_operand" "")
	(zero_extend:DI (match_operand:SI 1 "nonimmediate_operand" "")))]
  ""
{
  if (!TARGET_64BIT)
    {
      emit_insn (gen_zero_extendsidi2_1 (operands[0], operands[1]));
      DONE;
    }
})

(define_insn "*zero_extendsidi2_rex64"
  [(set (match_operand:DI 0 "nonimmediate_operand"  "=r,o,?*Ym,?*y,?*Yi,*Y2")
	(zero_extend:DI
	 (match_operand:SI 1 "nonimmediate_operand" "rm,0,r   ,m  ,r   ,m")))]
  "TARGET_64BIT"
  "@
   mov\t{%k1, %k0|%k0, %k1}
   #
   movd\t{%1, %0|%0, %1}
   movd\t{%1, %0|%0, %1}
   %vmovd\t{%1, %0|%0, %1}
   %vmovd\t{%1, %0|%0, %1}"
  [(set_attr "type" "imovx,imov,mmxmov,mmxmov,ssemov,ssemov")
   (set_attr "prefix" "orig,*,orig,orig,maybe_vex,maybe_vex")
   (set_attr "prefix_0f" "0,*,*,*,*,*")
   (set_attr "mode" "SI,DI,DI,DI,TI,TI")])

(define_split
  [(set (match_operand:DI 0 "memory_operand" "")
     	(zero_extend:DI (match_dup 0)))]
  "TARGET_64BIT"
  [(set (match_dup 4) (const_int 0))]
  "split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);")

;; %%% Kill me once multi-word ops are sane.
(define_insn "zero_extendsidi2_1"
  [(set (match_operand:DI 0 "nonimmediate_operand" "=r,?r,?o,?*Ym,?*y,?*Yi,*Y2")
	(zero_extend:DI
	 (match_operand:SI 1 "nonimmediate_operand" "0,rm,r ,r   ,m  ,r   ,m")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT"
  "@
   #
   #
   #
   movd\t{%1, %0|%0, %1}
   movd\t{%1, %0|%0, %1}
   %vmovd\t{%1, %0|%0, %1}
   %vmovd\t{%1, %0|%0, %1}"
  [(set_attr "type" "multi,multi,multi,mmxmov,mmxmov,ssemov,ssemov")
   (set_attr "prefix" "*,*,*,orig,orig,maybe_vex,maybe_vex")
   (set_attr "mode" "SI,SI,SI,DI,DI,TI,TI")])

(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(zero_extend:DI (match_operand:SI 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT && reload_completed
   && true_regnum (operands[0]) == true_regnum (operands[1])"
  [(set (match_dup 4) (const_int 0))]
  "split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);")

(define_split
  [(set (match_operand:DI 0 "nonimmediate_operand" "")
	(zero_extend:DI (match_operand:SI 1 "general_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT && reload_completed
   && !(MMX_REG_P (operands[0]) || SSE_REG_P (operands[0]))"
  [(set (match_dup 3) (match_dup 1))
   (set (match_dup 4) (const_int 0))]
  "split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);")

(define_insn "zero_extend<mode>di2"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	 (match_operand:SWI12 1 "nonimmediate_operand" "<r>m")))]
  "TARGET_64BIT"
  "movz{<imodesuffix>l|x}\t{%1, %k0|%k0, %1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")])

(define_expand "zero_extendhisi2"
  [(set (match_operand:SI 0 "register_operand" "")
	(zero_extend:SI (match_operand:HI 1 "nonimmediate_operand" "")))]
  ""
{
  if (TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun))
    {
      operands[1] = force_reg (HImode, operands[1]);
      emit_insn (gen_zero_extendhisi2_and (operands[0], operands[1]));
      DONE;
    }
})

(define_insn_and_split "zero_extendhisi2_and"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(zero_extend:SI (match_operand:HI 1 "register_operand" "0")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun)"
  "#"
  "&& reload_completed"
  [(parallel [(set (match_dup 0) (and:SI (match_dup 0) (const_int 65535)))
	      (clobber (reg:CC FLAGS_REG))])]
  ""
  [(set_attr "type" "alu1")
   (set_attr "mode" "SI")])

(define_insn "*zero_extendhisi2_movzwl"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(zero_extend:SI (match_operand:HI 1 "nonimmediate_operand" "rm")))]
  "!TARGET_ZERO_EXTEND_WITH_AND
   || optimize_function_for_size_p (cfun)"
  "movz{wl|x}\t{%1, %0|%0, %1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")])

(define_expand "zero_extendqi<mode>2"
  [(parallel
    [(set (match_operand:SWI24 0 "register_operand" "")
	  (zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "")))
     (clobber (reg:CC FLAGS_REG))])])

(define_insn "*zero_extendqi<mode>2_and"
  [(set (match_operand:SWI24 0 "register_operand" "=r,?&q")
	(zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "0,qm")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun)"
  "#"
  [(set_attr "type" "alu1")
   (set_attr "mode" "<MODE>")])

;; When source and destination does not overlap, clear destination
;; first and then do the movb
(define_split
  [(set (match_operand:SWI24 0 "register_operand" "")
	(zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed
   && (TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun))
   && ANY_QI_REG_P (operands[0])
   && (ANY_QI_REG_P (operands[1]) || MEM_P (operands[1]))
   && !reg_overlap_mentioned_p (operands[0], operands[1])"
  [(set (strict_low_part (match_dup 2)) (match_dup 1))]
{
  operands[2] = gen_lowpart (QImode, operands[0]);
  ix86_expand_clear (operands[0]);
})

(define_insn "*zero_extendqi<mode>2_movzbl_and"
  [(set (match_operand:SWI24 0 "register_operand" "=r,r")
	(zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "qm,0")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_ZERO_EXTEND_WITH_AND || optimize_function_for_size_p (cfun)"
  "#"
  [(set_attr "type" "imovx,alu1")
   (set_attr "mode" "<MODE>")])

;; For the movzbl case strip only the clobber
(define_split
  [(set (match_operand:SWI24 0 "register_operand" "")
	(zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed
   && (!TARGET_ZERO_EXTEND_WITH_AND || optimize_function_for_size_p (cfun))
   && (!REG_P (operands[1]) || ANY_QI_REG_P (operands[1]))"
  [(set (match_dup 0)
	(zero_extend:SWI24 (match_dup 1)))])

; zero extend to SImode to avoid partial register stalls
(define_insn "*zero_extendqi<mode>2_movzbl"
  [(set (match_operand:SWI24 0 "register_operand" "=r")
	(zero_extend:SWI24 (match_operand:QI 1 "nonimmediate_operand" "qm")))]
  "reload_completed
   && (!TARGET_ZERO_EXTEND_WITH_AND || optimize_function_for_size_p (cfun))"
  "movz{bl|x}\t{%1, %k0|%k0, %1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")])

;; Rest is handled by single and.
(define_split
  [(set (match_operand:SWI24 0 "register_operand" "")
	(zero_extend:SWI24 (match_operand:QI 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed
   && true_regnum (operands[0]) == true_regnum (operands[1])"
  [(parallel [(set (match_dup 0) (and:SWI24 (match_dup 0) (const_int 255)))
	      (clobber (reg:CC FLAGS_REG))])])

;; Sign extension instructions

(define_expand "extendsidi2"
  [(set (match_operand:DI 0 "register_operand" "")
	(sign_extend:DI (match_operand:SI 1 "register_operand" "")))]
  ""
{
  if (!TARGET_64BIT)
    {
      emit_insn (gen_extendsidi2_1 (operands[0], operands[1]));
      DONE;
    }
})

(define_insn "*extendsidi2_rex64"
  [(set (match_operand:DI 0 "register_operand" "=*a,r")
	(sign_extend:DI (match_operand:SI 1 "nonimmediate_operand" "*0,rm")))]
  "TARGET_64BIT"
  "@
   {cltq|cdqe}
   movs{lq|x}\t{%1, %0|%0, %1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "DI")
   (set_attr "prefix_0f" "0")
   (set_attr "modrm" "0,1")])

(define_insn "extendsidi2_1"
  [(set (match_operand:DI 0 "nonimmediate_operand" "=*A,r,?r,?*o")
	(sign_extend:DI (match_operand:SI 1 "register_operand" "0,0,r,r")))
   (clobber (reg:CC FLAGS_REG))
   (clobber (match_scratch:SI 2 "=X,X,X,&r"))]
  "!TARGET_64BIT"
  "#")

;; Extend to memory case when source register does die.
(define_split
  [(set (match_operand:DI 0 "memory_operand" "")
	(sign_extend:DI (match_operand:SI 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))
   (clobber (match_operand:SI 2 "register_operand" ""))]
  "(reload_completed
    && dead_or_set_p (insn, operands[1])
    && !reg_mentioned_p (operands[1], operands[0]))"
  [(set (match_dup 3) (match_dup 1))
   (parallel [(set (match_dup 1) (ashiftrt:SI (match_dup 1) (const_int 31)))
	      (clobber (reg:CC FLAGS_REG))])
   (set (match_dup 4) (match_dup 1))]
  "split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);")

;; Extend to memory case when source register does not die.
(define_split
  [(set (match_operand:DI 0 "memory_operand" "")
	(sign_extend:DI (match_operand:SI 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))
   (clobber (match_operand:SI 2 "register_operand" ""))]
  "reload_completed"
  [(const_int 0)]
{
  split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);

  emit_move_insn (operands[3], operands[1]);

  /* Generate a cltd if possible and doing so it profitable.  */
  if ((optimize_function_for_size_p (cfun) || TARGET_USE_CLTD)
      && true_regnum (operands[1]) == AX_REG
      && true_regnum (operands[2]) == DX_REG)
    {
      emit_insn (gen_ashrsi3_cvt (operands[2], operands[1], GEN_INT (31)));
    }
  else
    {
      emit_move_insn (operands[2], operands[1]);
      emit_insn (gen_ashrsi3_cvt (operands[2], operands[2], GEN_INT (31)));
    }
  emit_move_insn (operands[4], operands[2]);
  DONE;
})

;; Extend to register case.  Optimize case where source and destination
;; registers match and cases where we can use cltd.
(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(sign_extend:DI (match_operand:SI 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))
   (clobber (match_scratch:SI 2 ""))]
  "reload_completed"
  [(const_int 0)]
{
  split_double_mode (DImode, &operands[0], 1, &operands[3], &operands[4]);

  if (true_regnum (operands[3]) != true_regnum (operands[1]))
    emit_move_insn (operands[3], operands[1]);

  /* Generate a cltd if possible and doing so it profitable.  */
  if ((optimize_function_for_size_p (cfun) || TARGET_USE_CLTD)
      && true_regnum (operands[3]) == AX_REG
      && true_regnum (operands[4]) == DX_REG)
    {
      emit_insn (gen_ashrsi3_cvt (operands[4], operands[3], GEN_INT (31)));
      DONE;
    }

  if (true_regnum (operands[4]) != true_regnum (operands[1]))
    emit_move_insn (operands[4], operands[1]);

  emit_insn (gen_ashrsi3_cvt (operands[4], operands[4], GEN_INT (31)));
  DONE;
})

(define_insn "extend<mode>di2"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(sign_extend:DI
	 (match_operand:SWI12 1 "nonimmediate_operand" "<r>m")))]
  "TARGET_64BIT"
  "movs{<imodesuffix>q|x}\t{%1, %0|%0, %1}"
  [(set_attr "type" "imovx")
   (set_attr "mode" "DI")])

(define_insn "extendhisi2"
  [(set (match_operand:SI 0 "register_operand" "=*a,r")
	(sign_extend:SI (match_operand:HI 1 "nonimmediate_operand" "*0,rm")))]
  ""
{
  switch (get_attr_prefix_0f (insn))
    {
    case 0:
      return "{cwtl|cwde}";
    default:
      return "movs{wl|x}\t{%1, %0|%0, %1}";
    }
}
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")
   (set (attr "prefix_0f")
     ;; movsx is short decodable while cwtl is vector decoded.
     (if_then_else (and (eq_attr "cpu" "!k6")
			(eq_attr "alternative" "0"))
	(const_string "0")
	(const_string "1")))
   (set (attr "modrm")
     (if_then_else (eq_attr "prefix_0f" "0")
	(const_string "0")
	(const_string "1")))])

(define_insn "*extendhisi2_zext"
  [(set (match_operand:DI 0 "register_operand" "=*a,r")
	(zero_extend:DI
	 (sign_extend:SI
	  (match_operand:HI 1 "nonimmediate_operand" "*0,rm"))))]
  "TARGET_64BIT"
{
  switch (get_attr_prefix_0f (insn))
    {
    case 0:
      return "{cwtl|cwde}";
    default:
      return "movs{wl|x}\t{%1, %k0|%k0, %1}";
    }
}
  [(set_attr "type" "imovx")
   (set_attr "mode" "SI")
   (set (attr "prefix_0f")
     ;; movsx is short decodable while cwtl is vector decoded.
     (if_then_else (and (eq_attr "cpu" "!k6")
			(eq_attr "alternative" "0"))
	(const_string "0")
	(const_string "1")))
   (set (attr "modrm")
     (if_then_else (eq_attr "prefix_0f" "0")
	(const_string "0")
	(const_string "1")))])

(define_insn "extendqisi2"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(sign_extend:SI (match_operand:QI 1 "nonimmediate_operand" "qm")))]
  ""
  "movs{bl|x}\t{%1, %0|%0, %1}"
   [(set_attr "type" "imovx")
    (set_attr "mode" "SI")])

(define_insn "*extendqisi2_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (sign_extend:SI (match_operand:QI 1 "nonimmediate_operand" "qm"))))]
  "TARGET_64BIT"
  "movs{bl|x}\t{%1, %k0|%k0, %1}"
   [(set_attr "type" "imovx")
    (set_attr "mode" "SI")])

(define_insn "extendqihi2"
  [(set (match_operand:HI 0 "register_operand" "=*a,r")
	(sign_extend:HI (match_operand:QI 1 "nonimmediate_operand" "*0,qm")))]
  ""
{
  switch (get_attr_prefix_0f (insn))
    {
    case 0:
      return "{cbtw|cbw}";
    default:
      return "movs{bw|x}\t{%1, %0|%0, %1}";
    }
}
  [(set_attr "type" "imovx")
   (set_attr "mode" "HI")
   (set (attr "prefix_0f")
     ;; movsx is short decodable while cwtl is vector decoded.
     (if_then_else (and (eq_attr "cpu" "!k6")
			(eq_attr "alternative" "0"))
	(const_string "0")
	(const_string "1")))
   (set (attr "modrm")
     (if_then_else (eq_attr "prefix_0f" "0")
	(const_string "0")
	(const_string "1")))])

;; Conversions between float and double.

;; These are all no-ops in the model used for the 80387.
;; So just emit moves.

;; %%% Kill these when call knows how to work out a DFmode push earlier.
(define_split
  [(set (match_operand:DF 0 "push_operand" "")
	(float_extend:DF (match_operand:SF 1 "fp_register_operand" "")))]
  "reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (const_int -8)))
   (set (mem:DF (reg:P SP_REG)) (float_extend:DF (match_dup 1)))])

(define_split
  [(set (match_operand:XF 0 "push_operand" "")
	(float_extend:XF (match_operand:MODEF 1 "fp_register_operand" "")))]
  "reload_completed"
  [(set (reg:P SP_REG) (plus:P (reg:P SP_REG) (match_dup 2)))
   (set (mem:XF (reg:P SP_REG)) (float_extend:XF (match_dup 1)))]
  "operands[2] = GEN_INT (-GET_MODE_SIZE (XFmode));")

(define_expand "extendsfdf2"
  [(set (match_operand:DF 0 "nonimmediate_operand" "")
        (float_extend:DF (match_operand:SF 1 "general_operand" "")))]
  "TARGET_80387 || (TARGET_SSE2 && TARGET_SSE_MATH)"
{
  /* ??? Needed for compress_float_constant since all fp constants
     are LEGITIMATE_CONSTANT_P.  */
  if (GET_CODE (operands[1]) == CONST_DOUBLE)
    {
      if ((!TARGET_SSE2 || TARGET_MIX_SSE_I387)
	  && standard_80387_constant_p (operands[1]) > 0)
	{
	  operands[1] = simplify_const_unary_operation
	    (FLOAT_EXTEND, DFmode, operands[1], SFmode);
	  emit_move_insn_1 (operands[0], operands[1]);
	  DONE;
	}
      operands[1] = validize_mem (force_const_mem (SFmode, operands[1]));
    }
})

/* For converting SF(xmm2) to DF(xmm1), use the following code instead of
   cvtss2sd:
      unpcklps xmm2,xmm2   ; packed conversion might crash on signaling NaNs
      cvtps2pd xmm2,xmm1
   We do the conversion post reload to avoid producing of 128bit spills
   that might lead to ICE on 32bit target.  The sequence unlikely combine
   anyway.  */
(define_split
  [(set (match_operand:DF 0 "register_operand" "")
        (float_extend:DF
	  (match_operand:SF 1 "nonimmediate_operand" "")))]
  "TARGET_USE_VECTOR_FP_CONVERTS
   && optimize_insn_for_speed_p ()
   && reload_completed && SSE_REG_P (operands[0])"
   [(set (match_dup 2)
	 (float_extend:V2DF
	   (vec_select:V2SF
	     (match_dup 3)
	     (parallel [(const_int 0) (const_int 1)]))))]
{
  operands[2] = simplify_gen_subreg (V2DFmode, operands[0], DFmode, 0);
  operands[3] = simplify_gen_subreg (V4SFmode, operands[0], DFmode, 0);
  /* Use movss for loading from memory, unpcklps reg, reg for registers.
     Try to avoid move when unpacking can be done in source.  */
  if (REG_P (operands[1]))
    {
      /* If it is unsafe to overwrite upper half of source, we need
	 to move to destination and unpack there.  */
      if ((ORIGINAL_REGNO (operands[1]) < FIRST_PSEUDO_REGISTER
	   || PSEUDO_REGNO_BYTES (ORIGINAL_REGNO (operands[1])) > 4)
	  && true_regnum (operands[0]) != true_regnum (operands[1]))
	{
	  rtx tmp = gen_rtx_REG (SFmode, true_regnum (operands[0]));
	  emit_move_insn (tmp, operands[1]);
	}
      else
	operands[3] = simplify_gen_subreg (V4SFmode, operands[1], SFmode, 0);
      emit_insn (gen_vec_interleave_lowv4sf (operands[3], operands[3],
      		 			     operands[3]));
    }
  else
    emit_insn (gen_vec_setv4sf_0 (operands[3],
				  CONST0_RTX (V4SFmode), operands[1]));
})

(define_insn "*extendsfdf2_mixed"
  [(set (match_operand:DF 0 "nonimmediate_operand" "=f,m,x")
        (float_extend:DF
	  (match_operand:SF 1 "nonimmediate_operand" "fm,f,xm")))]
  "TARGET_SSE2 && TARGET_MIX_SSE_I387"
{
  switch (which_alternative)
    {
    case 0:
    case 1:
      return output_387_reg_move (insn, operands);

    case 2:
      return "%vcvtss2sd\t{%1, %d0|%d0, %1}";

    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,fmov,ssecvt")
   (set_attr "prefix" "orig,orig,maybe_vex")
   (set_attr "mode" "SF,XF,DF")])

(define_insn "*extendsfdf2_sse"
  [(set (match_operand:DF 0 "nonimmediate_operand" "=x")
        (float_extend:DF (match_operand:SF 1 "nonimmediate_operand" "xm")))]
  "TARGET_SSE2 && TARGET_SSE_MATH"
  "%vcvtss2sd\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "ssecvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "mode" "DF")])

(define_insn "*extendsfdf2_i387"
  [(set (match_operand:DF 0 "nonimmediate_operand" "=f,m")
        (float_extend:DF (match_operand:SF 1 "nonimmediate_operand" "fm,f")))]
  "TARGET_80387"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "SF,XF")])

(define_expand "extend<mode>xf2"
  [(set (match_operand:XF 0 "nonimmediate_operand" "")
        (float_extend:XF (match_operand:MODEF 1 "general_operand" "")))]
  "TARGET_80387"
{
  /* ??? Needed for compress_float_constant since all fp constants
     are LEGITIMATE_CONSTANT_P.  */
  if (GET_CODE (operands[1]) == CONST_DOUBLE)
    {
      if (standard_80387_constant_p (operands[1]) > 0)
	{
	  operands[1] = simplify_const_unary_operation
	    (FLOAT_EXTEND, XFmode, operands[1], <MODE>mode);
	  emit_move_insn_1 (operands[0], operands[1]);
	  DONE;
	}
      operands[1] = validize_mem (force_const_mem (<MODE>mode, operands[1]));
    }
})

(define_insn "*extend<mode>xf2_i387"
  [(set (match_operand:XF 0 "nonimmediate_operand" "=f,m")
        (float_extend:XF
	  (match_operand:MODEF 1 "nonimmediate_operand" "fm,f")))]
  "TARGET_80387"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "<MODE>,XF")])

;; %%% This seems bad bad news.
;; This cannot output into an f-reg because there is no way to be sure
;; of truncating in that case.  Otherwise this is just like a simple move
;; insn.  So we pretend we can output to a reg in order to get better
;; register preferencing, but we really use a stack slot.

;; Conversion from DFmode to SFmode.

(define_expand "truncdfsf2"
  [(set (match_operand:SF 0 "nonimmediate_operand" "")
	(float_truncate:SF
	  (match_operand:DF 1 "nonimmediate_operand" "")))]
  "TARGET_80387 || (TARGET_SSE2 && TARGET_SSE_MATH)"
{
  if (TARGET_SSE2 && TARGET_SSE_MATH && !TARGET_MIX_SSE_I387)
    ;
  else if (flag_unsafe_math_optimizations)
    ;
  else
    {
      enum ix86_stack_slot slot = (virtuals_instantiated
				   ? SLOT_TEMP
				   : SLOT_VIRTUAL);
      rtx temp = assign_386_stack_local (SFmode, slot);
      emit_insn (gen_truncdfsf2_with_temp (operands[0], operands[1], temp));
      DONE;
    }
})

/* For converting DF(xmm2) to SF(xmm1), use the following code instead of
   cvtsd2ss:
      unpcklpd xmm2,xmm2   ; packed conversion might crash on signaling NaNs
      cvtpd2ps xmm2,xmm1
   We do the conversion post reload to avoid producing of 128bit spills
   that might lead to ICE on 32bit target.  The sequence unlikely combine
   anyway.  */
(define_split
  [(set (match_operand:SF 0 "register_operand" "")
        (float_truncate:SF
	  (match_operand:DF 1 "nonimmediate_operand" "")))]
  "TARGET_USE_VECTOR_FP_CONVERTS
   && optimize_insn_for_speed_p ()
   && reload_completed && SSE_REG_P (operands[0])"
   [(set (match_dup 2)
	 (vec_concat:V4SF
	   (float_truncate:V2SF
	     (match_dup 4))
	   (match_dup 3)))]
{
  operands[2] = simplify_gen_subreg (V4SFmode, operands[0], SFmode, 0);
  operands[3] = CONST0_RTX (V2SFmode);
  operands[4] = simplify_gen_subreg (V2DFmode, operands[0], SFmode, 0);
  /* Use movsd for loading from memory, unpcklpd for registers.
     Try to avoid move when unpacking can be done in source, or SSE3
     movddup is available.  */
  if (REG_P (operands[1]))
    {
      if (!TARGET_SSE3
	  && true_regnum (operands[0]) != true_regnum (operands[1])
	  && (ORIGINAL_REGNO (operands[1]) < FIRST_PSEUDO_REGISTER
	      || PSEUDO_REGNO_BYTES (ORIGINAL_REGNO (operands[1])) > 8))
	{
	  rtx tmp = simplify_gen_subreg (DFmode, operands[0], SFmode, 0);
	  emit_move_insn (tmp, operands[1]);
	  operands[1] = tmp;
	}
      else if (!TARGET_SSE3)
	operands[4] = simplify_gen_subreg (V2DFmode, operands[1], DFmode, 0);
      emit_insn (gen_vec_dupv2df (operands[4], operands[1]));
    }
  else
    emit_insn (gen_sse2_loadlpd (operands[4],
				 CONST0_RTX (V2DFmode), operands[1]));
})

(define_expand "truncdfsf2_with_temp"
  [(parallel [(set (match_operand:SF 0 "" "")
		   (float_truncate:SF (match_operand:DF 1 "" "")))
	      (clobber (match_operand:SF 2 "" ""))])])

(define_insn "*truncdfsf_fast_mixed"
  [(set (match_operand:SF 0 "nonimmediate_operand"   "=fm,x")
        (float_truncate:SF
          (match_operand:DF 1 "nonimmediate_operand" "f  ,xm")))]
  "TARGET_SSE2 && TARGET_MIX_SSE_I387 && flag_unsafe_math_optimizations"
{
  switch (which_alternative)
    {
    case 0:
      return output_387_reg_move (insn, operands);
    case 1:
      return "%vcvtsd2ss\t{%1, %d0|%d0, %1}";
    default:
      gcc_unreachable ();
    }
}
  [(set_attr "type" "fmov,ssecvt")
   (set_attr "prefix" "orig,maybe_vex")
   (set_attr "mode" "SF")])

;; Yes, this one doesn't depend on flag_unsafe_math_optimizations,
;; because nothing we do here is unsafe.
(define_insn "*truncdfsf_fast_sse"
  [(set (match_operand:SF 0 "nonimmediate_operand"   "=x")
        (float_truncate:SF
          (match_operand:DF 1 "nonimmediate_operand" "xm")))]
  "TARGET_SSE2 && TARGET_SSE_MATH"
  "%vcvtsd2ss\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "ssecvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "mode" "SF")])

(define_insn "*truncdfsf_fast_i387"
  [(set (match_operand:SF 0 "nonimmediate_operand"   "=fm")
        (float_truncate:SF
          (match_operand:DF 1 "nonimmediate_operand" "f")))]
  "TARGET_80387 && flag_unsafe_math_optimizations"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "SF")])

(define_insn "*truncdfsf_mixed"
  [(set (match_operand:SF 0 "nonimmediate_operand"   "=m,Y2 ,?f,?x,?*r")
	(float_truncate:SF
	  (match_operand:DF 1 "nonimmediate_operand" "f ,Y2m,f ,f ,f")))
   (clobber (match_operand:SF 2 "memory_operand"     "=X,X  ,m ,m ,m"))]
  "TARGET_MIX_SSE_I387"
{
  switch (which_alternative)
    {
    case 0:
      return output_387_reg_move (insn, operands);
    case 1:
      return "%vcvtsd2ss\t{%1, %d0|%d0, %1}";

    default:
      return "#";
    }
}
  [(set_attr "type" "fmov,ssecvt,multi,multi,multi")
   (set_attr "unit" "*,*,i387,i387,i387")
   (set_attr "prefix" "orig,maybe_vex,orig,orig,orig")
   (set_attr "mode" "SF")])

(define_insn "*truncdfsf_i387"
  [(set (match_operand:SF 0 "nonimmediate_operand"   "=m,?f,?x,?*r")
	(float_truncate:SF
	  (match_operand:DF 1 "nonimmediate_operand" "f ,f ,f ,f")))
   (clobber (match_operand:SF 2 "memory_operand"     "=X,m ,m ,m"))]
  "TARGET_80387"
{
  switch (which_alternative)
    {
    case 0:
      return output_387_reg_move (insn, operands);

    default:
      return "#";
    }
}
  [(set_attr "type" "fmov,multi,multi,multi")
   (set_attr "unit" "*,i387,i387,i387")
   (set_attr "mode" "SF")])

(define_insn "*truncdfsf2_i387_1"
  [(set (match_operand:SF 0 "memory_operand" "=m")
	(float_truncate:SF
	  (match_operand:DF 1 "register_operand" "f")))]
  "TARGET_80387
   && !(TARGET_SSE2 && TARGET_SSE_MATH)
   && !TARGET_MIX_SSE_I387"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "SF")])

(define_split
  [(set (match_operand:SF 0 "register_operand" "")
	(float_truncate:SF
	 (match_operand:DF 1 "fp_register_operand" "")))
   (clobber (match_operand 2 "" ""))]
  "reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (match_dup 2))]
  "operands[1] = gen_rtx_REG (SFmode, true_regnum (operands[1]));")

;; Conversion from XFmode to {SF,DF}mode

(define_expand "truncxf<mode>2"
  [(parallel [(set (match_operand:MODEF 0 "nonimmediate_operand" "")
		   (float_truncate:MODEF
		     (match_operand:XF 1 "register_operand" "")))
	      (clobber (match_dup 2))])]
  "TARGET_80387"
{
  if (flag_unsafe_math_optimizations)
    {
      rtx reg = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (<MODE>mode);
      emit_insn (gen_truncxf<mode>2_i387_noop (reg, operands[1]));
      if (reg != operands[0])
	emit_move_insn (operands[0], reg);
      DONE;
    }
  else
    {
      enum ix86_stack_slot slot = (virtuals_instantiated
				   ? SLOT_TEMP
				   : SLOT_VIRTUAL);
      operands[2] = assign_386_stack_local (<MODE>mode, slot);
    }
})

(define_insn "*truncxfsf2_mixed"
  [(set (match_operand:SF 0 "nonimmediate_operand" "=m,?f,?x,?*r")
	(float_truncate:SF
	  (match_operand:XF 1 "register_operand"   "f ,f ,f ,f")))
   (clobber (match_operand:SF 2 "memory_operand"   "=X,m ,m ,m"))]
  "TARGET_80387"
{
  gcc_assert (!which_alternative);
  return output_387_reg_move (insn, operands);
}
  [(set_attr "type" "fmov,multi,multi,multi")
   (set_attr "unit" "*,i387,i387,i387")
   (set_attr "mode" "SF")])

(define_insn "*truncxfdf2_mixed"
  [(set (match_operand:DF 0 "nonimmediate_operand" "=m,?f,?Y2,?*r")
	(float_truncate:DF
	  (match_operand:XF 1 "register_operand"   "f ,f ,f  ,f")))
   (clobber (match_operand:DF 2 "memory_operand"   "=X,m ,m  ,m"))]
  "TARGET_80387"
{
  gcc_assert (!which_alternative);
  return output_387_reg_move (insn, operands);
}
  [(set_attr "type" "fmov,multi,multi,multi")
   (set_attr "unit" "*,i387,i387,i387")
   (set_attr "mode" "DF")])

(define_insn "truncxf<mode>2_i387_noop"
  [(set (match_operand:MODEF 0 "register_operand" "=f")
	(float_truncate:MODEF
	  (match_operand:XF 1 "register_operand" "f")))]
  "TARGET_80387 && flag_unsafe_math_optimizations"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "<MODE>")])

(define_insn "*truncxf<mode>2_i387"
  [(set (match_operand:MODEF 0 "memory_operand" "=m")
	(float_truncate:MODEF
	  (match_operand:XF 1 "register_operand" "f")))]
  "TARGET_80387"
  "* return output_387_reg_move (insn, operands);"
  [(set_attr "type" "fmov")
   (set_attr "mode" "<MODE>")])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float_truncate:MODEF
	  (match_operand:XF 1 "register_operand" "")))
   (clobber (match_operand:MODEF 2 "memory_operand" ""))]
  "TARGET_80387 && reload_completed"
  [(set (match_dup 2) (float_truncate:MODEF (match_dup 1)))
   (set (match_dup 0) (match_dup 2))])

(define_split
  [(set (match_operand:MODEF 0 "memory_operand" "")
	(float_truncate:MODEF
	  (match_operand:XF 1 "register_operand" "")))
   (clobber (match_operand:MODEF 2 "memory_operand" ""))]
  "TARGET_80387"
  [(set (match_dup 0) (float_truncate:MODEF (match_dup 1)))])

;; Signed conversion to DImode.

(define_expand "fix_truncxfdi2"
  [(parallel [(set (match_operand:DI 0 "nonimmediate_operand" "")
                   (fix:DI (match_operand:XF 1 "register_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_80387"
{
  if (TARGET_FISTTP)
   {
     emit_insn (gen_fix_truncdi_fisttp_i387_1 (operands[0], operands[1]));
     DONE;
   }
})

(define_expand "fix_trunc<mode>di2"
  [(parallel [(set (match_operand:DI 0 "nonimmediate_operand" "")
                   (fix:DI (match_operand:MODEF 1 "register_operand" "")))
              (clobber (reg:CC FLAGS_REG))])]
  "TARGET_80387 || (TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode))"
{
  if (TARGET_FISTTP
      && !(TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))
   {
     emit_insn (gen_fix_truncdi_fisttp_i387_1 (operands[0], operands[1]));
     DONE;
   }
  if (TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode))
   {
     rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (DImode);
     emit_insn (gen_fix_trunc<mode>di_sse (out, operands[1]));
     if (out != operands[0])
	emit_move_insn (operands[0], out);
     DONE;
   }
})

;; Signed conversion to SImode.

(define_expand "fix_truncxfsi2"
  [(parallel [(set (match_operand:SI 0 "nonimmediate_operand" "")
                   (fix:SI (match_operand:XF 1 "register_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_80387"
{
  if (TARGET_FISTTP)
   {
     emit_insn (gen_fix_truncsi_fisttp_i387_1 (operands[0], operands[1]));
     DONE;
   }
})

(define_expand "fix_trunc<mode>si2"
  [(parallel [(set (match_operand:SI 0 "nonimmediate_operand" "")
	           (fix:SI (match_operand:MODEF 1 "register_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_80387 || SSE_FLOAT_MODE_P (<MODE>mode)"
{
  if (TARGET_FISTTP
      && !(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))
   {
     emit_insn (gen_fix_truncsi_fisttp_i387_1 (operands[0], operands[1]));
     DONE;
   }
  if (SSE_FLOAT_MODE_P (<MODE>mode))
   {
     rtx out = REG_P (operands[0]) ? operands[0] : gen_reg_rtx (SImode);
     emit_insn (gen_fix_trunc<mode>si_sse (out, operands[1]));
     if (out != operands[0])
	emit_move_insn (operands[0], out);
     DONE;
   }
})

;; Signed conversion to HImode.

(define_expand "fix_trunc<mode>hi2"
  [(parallel [(set (match_operand:HI 0 "nonimmediate_operand" "")
	           (fix:HI (match_operand:X87MODEF 1 "register_operand" "")))
              (clobber (reg:CC FLAGS_REG))])]
  "TARGET_80387
   && !(SSE_FLOAT_MODE_P (<MODE>mode) && (!TARGET_FISTTP || TARGET_SSE_MATH))"
{
  if (TARGET_FISTTP)
   {
     emit_insn (gen_fix_trunchi_fisttp_i387_1 (operands[0], operands[1]));
     DONE;
   }
})

;; Unsigned conversion to SImode.

(define_expand "fixuns_trunc<mode>si2"
  [(parallel
    [(set (match_operand:SI 0 "register_operand" "")
	  (unsigned_fix:SI
	    (match_operand:MODEF 1 "nonimmediate_operand" "")))
     (use (match_dup 2))
     (clobber (match_scratch:<ssevecmode> 3 ""))
     (clobber (match_scratch:<ssevecmode> 4 ""))])]
  "!TARGET_64BIT && TARGET_SSE2 && TARGET_SSE_MATH"
{
  enum machine_mode mode = <MODE>mode;
  enum machine_mode vecmode = <ssevecmode>mode;
  REAL_VALUE_TYPE TWO31r;
  rtx two31;

  if (optimize_insn_for_size_p ())
    FAIL;

  real_ldexp (&TWO31r, &dconst1, 31);
  two31 = const_double_from_real_value (TWO31r, mode);
  two31 = ix86_build_const_vector (vecmode, true, two31);
  operands[2] = force_reg (vecmode, two31);
})

(define_insn_and_split "*fixuns_trunc<mode>_1"
  [(set (match_operand:SI 0 "register_operand" "=&x,&x")
	(unsigned_fix:SI
	  (match_operand:MODEF 3 "nonimmediate_operand" "xm,xm")))
   (use (match_operand:<ssevecmode> 4  "nonimmediate_operand" "m,x"))
   (clobber (match_scratch:<ssevecmode> 1 "=x,&x"))
   (clobber (match_scratch:<ssevecmode> 2 "=x,x"))]
  "!TARGET_64BIT && TARGET_SSE2 && TARGET_SSE_MATH
   && optimize_function_for_speed_p (cfun)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
{
  ix86_split_convert_uns_si_sse (operands);
  DONE;
})

;; Unsigned conversion to HImode.
;; Without these patterns, we'll try the unsigned SI conversion which
;; is complex for SSE, rather than the signed SI conversion, which isn't.

(define_expand "fixuns_trunc<mode>hi2"
  [(set (match_dup 2)
	(fix:SI (match_operand:MODEF 1 "nonimmediate_operand" "")))
   (set (match_operand:HI 0 "nonimmediate_operand" "")
	(subreg:HI (match_dup 2) 0))]
  "SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH"
  "operands[2] = gen_reg_rtx (SImode);")

;; When SSE is available, it is always faster to use it!
(define_insn "fix_trunc<mode>di_sse"
  [(set (match_operand:DI 0 "register_operand" "=r,r")
	(fix:DI (match_operand:MODEF 1 "nonimmediate_operand" "x,m")))]
  "TARGET_64BIT && SSE_FLOAT_MODE_P (<MODE>mode)
   && (!TARGET_FISTTP || TARGET_SSE_MATH)"
  "%vcvtts<ssemodefsuffix>2si{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "sseicvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "prefix_rex" "1")
   (set_attr "mode" "<MODE>")
   (set_attr "athlon_decode" "double,vector")
   (set_attr "amdfam10_decode" "double,double")
   (set_attr "bdver1_decode" "double,double")])

(define_insn "fix_trunc<mode>si_sse"
  [(set (match_operand:SI 0 "register_operand" "=r,r")
	(fix:SI (match_operand:MODEF 1 "nonimmediate_operand" "x,m")))]
  "SSE_FLOAT_MODE_P (<MODE>mode)
   && (!TARGET_FISTTP || TARGET_SSE_MATH)"
  "%vcvtts<ssemodefsuffix>2si\t{%1, %0|%0, %1}"
  [(set_attr "type" "sseicvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "mode" "<MODE>")
   (set_attr "athlon_decode" "double,vector")
   (set_attr "amdfam10_decode" "double,double")
   (set_attr "bdver1_decode" "double,double")])

;; Shorten x87->SSE reload sequences of fix_trunc?f?i_sse patterns.
(define_peephole2
  [(set (match_operand:MODEF 0 "register_operand" "")
	(match_operand:MODEF 1 "memory_operand" ""))
   (set (match_operand:SSEMODEI24 2 "register_operand" "")
	(fix:SSEMODEI24 (match_dup 0)))]
  "TARGET_SHORTEN_X87_SSE
   && !(TARGET_AVOID_VECTOR_DECODE && optimize_insn_for_speed_p ())
   && peep2_reg_dead_p (2, operands[0])"
  [(set (match_dup 2) (fix:SSEMODEI24 (match_dup 1)))])

;; Avoid vector decoded forms of the instruction.
(define_peephole2
  [(match_scratch:DF 2 "Y2")
   (set (match_operand:SSEMODEI24 0 "register_operand" "")
	(fix:SSEMODEI24 (match_operand:DF 1 "memory_operand" "")))]
  "TARGET_AVOID_VECTOR_DECODE && optimize_insn_for_speed_p ()"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (fix:SSEMODEI24 (match_dup 2)))])

(define_peephole2
  [(match_scratch:SF 2 "x")
   (set (match_operand:SSEMODEI24 0 "register_operand" "")
	(fix:SSEMODEI24 (match_operand:SF 1 "memory_operand" "")))]
  "TARGET_AVOID_VECTOR_DECODE && optimize_insn_for_speed_p ()"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (fix:SSEMODEI24 (match_dup 2)))])

(define_insn_and_split "fix_trunc<mode>_fisttp_i387_1"
  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "")
	(fix:X87MODEI (match_operand 1 "register_operand" "")))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_FISTTP
   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
	 && (TARGET_64BIT || <MODE>mode != DImode))
	&& TARGET_SSE_MATH)
   && can_create_pseudo_p ()"
  "#"
  "&& 1"
  [(const_int 0)]
{
  if (memory_operand (operands[0], VOIDmode))
    emit_insn (gen_fix_trunc<mode>_i387_fisttp (operands[0], operands[1]));
  else
    {
      operands[2] = assign_386_stack_local (<MODE>mode, SLOT_TEMP);
      emit_insn (gen_fix_trunc<mode>_i387_fisttp_with_temp (operands[0],
							    operands[1],
							    operands[2]));
    }
  DONE;
}
  [(set_attr "type" "fisttp")
   (set_attr "mode" "<MODE>")])

(define_insn "fix_trunc<mode>_i387_fisttp"
  [(set (match_operand:X87MODEI 0 "memory_operand" "=m")
	(fix:X87MODEI (match_operand 1 "register_operand" "f")))
   (clobber (match_scratch:XF 2 "=&1f"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_FISTTP
   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
	 && (TARGET_64BIT || <MODE>mode != DImode))
	&& TARGET_SSE_MATH)"
  "* return output_fix_trunc (insn, operands, 1);"
  [(set_attr "type" "fisttp")
   (set_attr "mode" "<MODE>")])

(define_insn "fix_trunc<mode>_i387_fisttp_with_temp"
  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "=m,?r")
	(fix:X87MODEI (match_operand 1 "register_operand" "f,f")))
   (clobber (match_operand:X87MODEI 2 "memory_operand" "=X,m"))
   (clobber (match_scratch:XF 3 "=&1f,&1f"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && TARGET_FISTTP
   && !((SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
	&& (TARGET_64BIT || <MODE>mode != DImode))
	&& TARGET_SSE_MATH)"
  "#"
  [(set_attr "type" "fisttp")
   (set_attr "mode" "<MODE>")])

(define_split
  [(set (match_operand:X87MODEI 0 "register_operand" "")
	(fix:X87MODEI (match_operand 1 "register_operand" "")))
   (clobber (match_operand:X87MODEI 2 "memory_operand" ""))
   (clobber (match_scratch 3 ""))]
  "reload_completed"
  [(parallel [(set (match_dup 2) (fix:X87MODEI (match_dup 1)))
	      (clobber (match_dup 3))])
   (set (match_dup 0) (match_dup 2))])

(define_split
  [(set (match_operand:X87MODEI 0 "memory_operand" "")
	(fix:X87MODEI (match_operand 1 "register_operand" "")))
   (clobber (match_operand:X87MODEI 2 "memory_operand" ""))
   (clobber (match_scratch 3 ""))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (fix:X87MODEI (match_dup 1)))
	      (clobber (match_dup 3))])])

;; See the comments in i386.h near OPTIMIZE_MODE_SWITCHING for the description
;; of the machinery. Please note the clobber of FLAGS_REG. In i387 control
;; word calculation (inserted by LCM in mode switching pass) a FLAGS_REG
;; clobbering insns can be used. Look at emit_i387_cw_initialization ()
;; function in i386.c.
(define_insn_and_split "*fix_trunc<mode>_i387_1"
  [(set (match_operand:X87MODEI 0 "nonimmediate_operand" "")
	(fix:X87MODEI (match_operand 1 "register_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && !TARGET_FISTTP
   && !(SSE_FLOAT_MODE_P (GET_MODE (operands[1]))
	 && (TARGET_64BIT || <MODE>mode != DImode))
   && can_create_pseudo_p ()"
  "#"
  "&& 1"
  [(const_int 0)]
{
  ix86_optimize_mode_switching[I387_TRUNC] = 1;

  operands[2] = assign_386_stack_local (HImode, SLOT_CW_STORED);
  operands[3] = assign_386_stack_local (HImode, SLOT_CW_TRUNC);
  if (memory_operand (operands[0], VOIDmode))
    emit_insn (gen_fix_trunc<mode>_i387 (operands[0], operands[1],
					 operands[2], operands[3]));
  else
    {
      operands[4] = assign_386_stack_local (<MODE>mode, SLOT_TEMP);
      emit_insn (gen_fix_trunc<mode>_i387_with_temp (operands[0], operands[1],
						     operands[2], operands[3],
						     operands[4]));
    }
  DONE;
}
  [(set_attr "type" "fistp")
   (set_attr "i387_cw" "trunc")
   (set_attr "mode" "<MODE>")])

(define_insn "fix_truncdi_i387"
  [(set (match_operand:DI 0 "memory_operand" "=m")
	(fix:DI (match_operand 1 "register_operand" "f")))
   (use (match_operand:HI 2 "memory_operand" "m"))
   (use (match_operand:HI 3 "memory_operand" "m"))
   (clobber (match_scratch:XF 4 "=&1f"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && !TARGET_FISTTP
   && !(TARGET_64BIT && SSE_FLOAT_MODE_P (GET_MODE (operands[1])))"
  "* return output_fix_trunc (insn, operands, 0);"
  [(set_attr "type" "fistp")
   (set_attr "i387_cw" "trunc")
   (set_attr "mode" "DI")])

(define_insn "fix_truncdi_i387_with_temp"
  [(set (match_operand:DI 0 "nonimmediate_operand" "=m,?r")
	(fix:DI (match_operand 1 "register_operand" "f,f")))
   (use (match_operand:HI 2 "memory_operand" "m,m"))
   (use (match_operand:HI 3 "memory_operand" "m,m"))
   (clobber (match_operand:DI 4 "memory_operand" "=X,m"))
   (clobber (match_scratch:XF 5 "=&1f,&1f"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && !TARGET_FISTTP
   && !(TARGET_64BIT && SSE_FLOAT_MODE_P (GET_MODE (operands[1])))"
  "#"
  [(set_attr "type" "fistp")
   (set_attr "i387_cw" "trunc")
   (set_attr "mode" "DI")])

(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(fix:DI (match_operand 1 "register_operand" "")))
   (use (match_operand:HI 2 "memory_operand" ""))
   (use (match_operand:HI 3 "memory_operand" ""))
   (clobber (match_operand:DI 4 "memory_operand" ""))
   (clobber (match_scratch 5 ""))]
  "reload_completed"
  [(parallel [(set (match_dup 4) (fix:DI (match_dup 1)))
	      (use (match_dup 2))
	      (use (match_dup 3))
	      (clobber (match_dup 5))])
   (set (match_dup 0) (match_dup 4))])

(define_split
  [(set (match_operand:DI 0 "memory_operand" "")
	(fix:DI (match_operand 1 "register_operand" "")))
   (use (match_operand:HI 2 "memory_operand" ""))
   (use (match_operand:HI 3 "memory_operand" ""))
   (clobber (match_operand:DI 4 "memory_operand" ""))
   (clobber (match_scratch 5 ""))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (fix:DI (match_dup 1)))
	      (use (match_dup 2))
	      (use (match_dup 3))
	      (clobber (match_dup 5))])])

(define_insn "fix_trunc<mode>_i387"
  [(set (match_operand:X87MODEI12 0 "memory_operand" "=m")
	(fix:X87MODEI12 (match_operand 1 "register_operand" "f")))
   (use (match_operand:HI 2 "memory_operand" "m"))
   (use (match_operand:HI 3 "memory_operand" "m"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && !TARGET_FISTTP
   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
  "* return output_fix_trunc (insn, operands, 0);"
  [(set_attr "type" "fistp")
   (set_attr "i387_cw" "trunc")
   (set_attr "mode" "<MODE>")])

(define_insn "fix_trunc<mode>_i387_with_temp"
  [(set (match_operand:X87MODEI12 0 "nonimmediate_operand" "=m,?r")
	(fix:X87MODEI12 (match_operand 1 "register_operand" "f,f")))
   (use (match_operand:HI 2 "memory_operand" "m,m"))
   (use (match_operand:HI 3 "memory_operand" "m,m"))
   (clobber (match_operand:X87MODEI12 4 "memory_operand" "=X,m"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && !TARGET_FISTTP
   && !SSE_FLOAT_MODE_P (GET_MODE (operands[1]))"
  "#"
  [(set_attr "type" "fistp")
   (set_attr "i387_cw" "trunc")
   (set_attr "mode" "<MODE>")])

(define_split
  [(set (match_operand:X87MODEI12 0 "register_operand" "")
	(fix:X87MODEI12 (match_operand 1 "register_operand" "")))
   (use (match_operand:HI 2 "memory_operand" ""))
   (use (match_operand:HI 3 "memory_operand" ""))
   (clobber (match_operand:X87MODEI12 4 "memory_operand" ""))]
  "reload_completed"
  [(parallel [(set (match_dup 4) (fix:X87MODEI12 (match_dup 1)))
	      (use (match_dup 2))
	      (use (match_dup 3))])
   (set (match_dup 0) (match_dup 4))])

(define_split
  [(set (match_operand:X87MODEI12 0 "memory_operand" "")
	(fix:X87MODEI12 (match_operand 1 "register_operand" "")))
   (use (match_operand:HI 2 "memory_operand" ""))
   (use (match_operand:HI 3 "memory_operand" ""))
   (clobber (match_operand:X87MODEI12 4 "memory_operand" ""))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (fix:X87MODEI12 (match_dup 1)))
	      (use (match_dup 2))
	      (use (match_dup 3))])])

(define_insn "x86_fnstcw_1"
  [(set (match_operand:HI 0 "memory_operand" "=m")
	(unspec:HI [(reg:HI FPCR_REG)] UNSPEC_FSTCW))]
  "TARGET_80387"
  "fnstcw\t%0"
  [(set (attr "length")
	(symbol_ref "ix86_attr_length_address_default (insn) + 2"))
   (set_attr "mode" "HI")
   (set_attr "unit" "i387")
   (set_attr "bdver1_decode" "vector")])

(define_insn "x86_fldcw_1"
  [(set (reg:HI FPCR_REG)
	(unspec:HI [(match_operand:HI 0 "memory_operand" "m")] UNSPEC_FLDCW))]
  "TARGET_80387"
  "fldcw\t%0"
  [(set (attr "length")
	(symbol_ref "ix86_attr_length_address_default (insn) + 2"))
   (set_attr "mode" "HI")
   (set_attr "unit" "i387")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "vector")
   (set_attr "bdver1_decode" "vector")])

;; Conversion between fixed point and floating point.

;; Even though we only accept memory inputs, the backend _really_
;; wants to be able to do this between registers.

(define_expand "floathi<mode>2"
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:HI 1 "nonimmediate_operand" "")))]
  "TARGET_80387
   && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
       || TARGET_MIX_SSE_I387)")

;; Pre-reload splitter to add memory clobber to the pattern.
(define_insn_and_split "*floathi<mode>2_1"
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:HI 1 "register_operand" "")))]
  "TARGET_80387
   && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
       || TARGET_MIX_SSE_I387)
   && can_create_pseudo_p ()"
  "#"
  "&& 1"
  [(parallel [(set (match_dup 0)
	      (float:X87MODEF (match_dup 1)))
   (clobber (match_dup 2))])]
  "operands[2] = assign_386_stack_local (HImode, SLOT_TEMP);")

(define_insn "*floathi<mode>2_i387_with_temp"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f,f")
	(float:X87MODEF (match_operand:HI 1 "nonimmediate_operand" "m,?r")))
  (clobber (match_operand:HI 2 "memory_operand" "=m,m"))]
  "TARGET_80387
   && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
       || TARGET_MIX_SSE_I387)"
  "#"
  [(set_attr "type" "fmov,multi")
   (set_attr "mode" "<MODE>")
   (set_attr "unit" "*,i387")
   (set_attr "fp_int_src" "true")])

(define_insn "*floathi<mode>2_i387"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f")
	(float:X87MODEF (match_operand:HI 1 "memory_operand" "m")))]
  "TARGET_80387
   && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
       || TARGET_MIX_SSE_I387)"
  "fild%Z1\t%1"
  [(set_attr "type" "fmov")
   (set_attr "mode" "<MODE>")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:HI 1 "register_operand" "")))
   (clobber (match_operand:HI 2 "memory_operand" ""))]
  "TARGET_80387
   && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
       || TARGET_MIX_SSE_I387)
   && reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (float:X87MODEF (match_dup 2)))])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:HI 1 "memory_operand" "")))
   (clobber (match_operand:HI 2 "memory_operand" ""))]
   "TARGET_80387
    && (!(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
        || TARGET_MIX_SSE_I387)
    && reload_completed"
  [(set (match_dup 0) (float:X87MODEF (match_dup 1)))])

(define_expand "float<SSEMODEI24:mode><X87MODEF:mode>2"
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "")))]
  "TARGET_80387
   || ((<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
       && SSE_FLOAT_MODE_P (<X87MODEF:MODE>mode) && TARGET_SSE_MATH)"
{
  if (!((<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
	&& SSE_FLOAT_MODE_P (<X87MODEF:MODE>mode) && TARGET_SSE_MATH)
      && !X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode))
    {
      rtx reg = gen_reg_rtx (XFmode);
      rtx insn;

      emit_insn (gen_float<SSEMODEI24:mode>xf2 (reg, operands[1]));

      if (<X87MODEF:MODE>mode == SFmode)
	insn = gen_truncxfsf2 (operands[0], reg);
      else if (<X87MODEF:MODE>mode == DFmode)
	insn = gen_truncxfdf2 (operands[0], reg);
      else
	gcc_unreachable ();

      emit_insn (insn);
      DONE;
    }
})

;; Pre-reload splitter to add memory clobber to the pattern.
(define_insn_and_split "*float<SSEMODEI24:mode><X87MODEF:mode>2_1"
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:SSEMODEI24 1 "register_operand" "")))]
  "((TARGET_80387
     && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode)
     && (!((<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
	   && SSE_FLOAT_MODE_P (<X87MODEF:MODE>mode) && TARGET_SSE_MATH)
	 || TARGET_MIX_SSE_I387))
    || ((<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
	&& SSE_FLOAT_MODE_P (<X87MODEF:MODE>mode) && TARGET_SSE_MATH
	&& ((<SSEMODEI24:MODE>mode == SImode
	     && TARGET_SSE2 && TARGET_USE_VECTOR_CONVERTS
	     && optimize_function_for_speed_p (cfun)
	     && flag_trapping_math)
	    || !(TARGET_INTER_UNIT_CONVERSIONS
	         || optimize_function_for_size_p (cfun)))))
   && can_create_pseudo_p ()"
  "#"
  "&& 1"
  [(parallel [(set (match_dup 0) (float:X87MODEF (match_dup 1)))
	      (clobber (match_dup 2))])]
{
  operands[2] = assign_386_stack_local (<SSEMODEI24:MODE>mode, SLOT_TEMP);

  /* Avoid store forwarding (partial memory) stall penalty
     by passing DImode value through XMM registers.  */
  if (<SSEMODEI24:MODE>mode == DImode && !TARGET_64BIT
      && TARGET_80387 && TARGET_SSE2 && TARGET_INTER_UNIT_MOVES
      && optimize_function_for_speed_p (cfun))
    {
      emit_insn (gen_floatdi<X87MODEF:mode>2_i387_with_xmm (operands[0],
							    operands[1],
							    operands[2]));
      DONE;
    }
})

(define_insn "*floatsi<mode>2_vector_mixed_with_temp"
  [(set (match_operand:MODEF 0 "register_operand" "=f,f,x,x,x")
	(float:MODEF
	  (match_operand:SI 1 "nonimmediate_operand" "m,?r,r,m,!x")))
   (clobber (match_operand:SI 2 "memory_operand" "=X,m,m,X,m"))]
  "TARGET_SSE2 && TARGET_MIX_SSE_I387
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)"
  "#"
  [(set_attr "type" "fmov,multi,sseicvt,sseicvt,sseicvt")
   (set_attr "mode" "<MODE>,<MODE>,<MODE>,<MODE>,<ssevecmode>")
   (set_attr "unit" "*,i387,*,*,*")
   (set_attr "athlon_decode" "*,*,double,direct,double")
   (set_attr "amdfam10_decode" "*,*,vector,double,double")
   (set_attr "bdver1_decode" "*,*,double,direct,double")
   (set_attr "fp_int_src" "true")])

(define_insn "*floatsi<mode>2_vector_mixed"
  [(set (match_operand:MODEF 0 "register_operand" "=f,x")
	(float:MODEF (match_operand:SI 1 "memory_operand" "m,m")))]
  "TARGET_SSE2 && TARGET_MIX_SSE_I387
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)"
  "@
   fild%Z1\t%1
   #"
  [(set_attr "type" "fmov,sseicvt")
   (set_attr "mode" "<MODE>,<ssevecmode>")
   (set_attr "unit" "i387,*")
   (set_attr "athlon_decode" "*,direct")
   (set_attr "amdfam10_decode" "*,double")
   (set_attr "bdver1_decode" "*,direct")
   (set_attr "fp_int_src" "true")])

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_mixed_with_temp"
  [(set (match_operand:MODEF 0 "register_operand" "=f,f,x,x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "m,?r,r,m")))
  (clobber (match_operand:SSEMODEI24 2 "memory_operand" "=X,m,m,X"))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_MIX_SSE_I387"
  "#"
  [(set_attr "type" "fmov,multi,sseicvt,sseicvt")
   (set_attr "mode" "<MODEF:MODE>")
   (set_attr "unit" "*,i387,*,*")
   (set_attr "athlon_decode" "*,*,double,direct")
   (set_attr "amdfam10_decode" "*,*,vector,double")
   (set_attr "bdver1_decode" "*,*,double,direct")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SSEMODEI24 1 "register_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_MIX_SSE_I387
   && TARGET_INTER_UNIT_CONVERSIONS
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(set (match_dup 0) (float:MODEF (match_dup 1)))])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SSEMODEI24 1 "register_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_MIX_SSE_I387
   && !(TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (float:MODEF (match_dup 2)))])

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_mixed_interunit"
  [(set (match_operand:MODEF 0 "register_operand" "=f,x,x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "m,r,m")))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_MIX_SSE_I387
   && (TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))"
  "@
   fild%Z1\t%1
   %vcvtsi2s<MODEF:ssemodefsuffix><SSEMODEI24:rex64suffix>\t{%1, %d0|%d0, %1}
   %vcvtsi2s<MODEF:ssemodefsuffix><SSEMODEI24:rex64suffix>\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "fmov,sseicvt,sseicvt")
   (set_attr "prefix" "orig,maybe_vex,maybe_vex")
   (set_attr "mode" "<MODEF:MODE>")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "prefix" "maybe_vex")
	    (ne (symbol_ref "<SSEMODEI24:MODE>mode == DImode") (const_int 0)))
       (const_string "1")
       (const_string "*")))
   (set_attr "unit" "i387,*,*")
   (set_attr "athlon_decode" "*,double,direct")
   (set_attr "amdfam10_decode" "*,vector,double")
   (set_attr "bdver1_decode" "*,double,direct")
   (set_attr "fp_int_src" "true")])

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_mixed_nointerunit"
  [(set (match_operand:MODEF 0 "register_operand" "=f,x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "memory_operand" "m,m")))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_MIX_SSE_I387
   && !(TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))"
  "@
   fild%Z1\t%1
   %vcvtsi2s<MODEF:ssemodefsuffix><SSEMODEI24:rex64suffix>\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "fmov,sseicvt")
   (set_attr "prefix" "orig,maybe_vex")
   (set_attr "mode" "<MODEF:MODE>")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "prefix" "maybe_vex")
	    (ne (symbol_ref "<SSEMODEI24:MODE>mode == DImode") (const_int 0)))
       (const_string "1")
       (const_string "*")))
   (set_attr "athlon_decode" "*,direct")
   (set_attr "amdfam10_decode" "*,double")
   (set_attr "bdver1_decode" "*,direct")
   (set_attr "fp_int_src" "true")])

(define_insn "*floatsi<mode>2_vector_sse_with_temp"
  [(set (match_operand:MODEF 0 "register_operand" "=x,x,x")
	(float:MODEF
	  (match_operand:SI 1 "nonimmediate_operand" "r,m,!x")))
   (clobber (match_operand:SI 2 "memory_operand" "=m,X,m"))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)"
  "#"
  [(set_attr "type" "sseicvt")
   (set_attr "mode" "<MODE>,<MODE>,<ssevecmode>")
   (set_attr "athlon_decode" "double,direct,double")
   (set_attr "amdfam10_decode" "vector,double,double")
   (set_attr "bdver1_decode" "double,direct,double")
   (set_attr "fp_int_src" "true")])

(define_insn "*floatsi<mode>2_vector_sse"
  [(set (match_operand:MODEF 0 "register_operand" "=x")
	(float:MODEF (match_operand:SI 1 "memory_operand" "m")))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)"
  "#"
  [(set_attr "type" "sseicvt")
   (set_attr "mode" "<MODE>")
   (set_attr "athlon_decode" "direct")
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SI 1 "register_operand" "")))
   (clobber (match_operand:SI 2 "memory_operand" ""))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(const_int 0)]
{
  rtx op1 = operands[1];

  operands[3] = simplify_gen_subreg (<ssevecmode>mode, operands[0],
				     <MODE>mode, 0);
  if (GET_CODE (op1) == SUBREG)
    op1 = SUBREG_REG (op1);

  if (GENERAL_REG_P (op1) && TARGET_INTER_UNIT_MOVES)
    {
      operands[4] = simplify_gen_subreg (V4SImode, operands[0], <MODE>mode, 0);
      emit_insn (gen_sse2_loadld (operands[4],
				  CONST0_RTX (V4SImode), operands[1]));
    }
  /* We can ignore possible trapping value in the
     high part of SSE register for non-trapping math. */
  else if (SSE_REG_P (op1) && !flag_trapping_math)
    operands[4] = simplify_gen_subreg (V4SImode, operands[1], SImode, 0);
  else
    {
      operands[4] = simplify_gen_subreg (V4SImode, operands[0], <MODE>mode, 0);
      emit_move_insn (operands[2], operands[1]);
      emit_insn (gen_sse2_loadld (operands[4],
				  CONST0_RTX (V4SImode), operands[2]));
    }
  emit_insn
    (gen_sse2_cvtdq2p<ssemodefsuffix> (operands[3], operands[4]));
  DONE;
})

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SI 1 "memory_operand" "")))
   (clobber (match_operand:SI 2 "memory_operand" ""))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(const_int 0)]
{
  operands[3] = simplify_gen_subreg (<ssevecmode>mode, operands[0],
				     <MODE>mode, 0);
  operands[4] = simplify_gen_subreg (V4SImode, operands[0], <MODE>mode, 0);

  emit_insn (gen_sse2_loadld (operands[4],
			      CONST0_RTX (V4SImode), operands[1]));
  emit_insn
    (gen_sse2_cvtdq2p<ssemodefsuffix> (operands[3], operands[4]));
  DONE;
})

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SI 1 "register_operand" "")))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(const_int 0)]
{
  rtx op1 = operands[1];

  operands[3] = simplify_gen_subreg (<ssevecmode>mode, operands[0],
				     <MODE>mode, 0);
  if (GET_CODE (op1) == SUBREG)
    op1 = SUBREG_REG (op1);

  if (GENERAL_REG_P (op1) && TARGET_INTER_UNIT_MOVES)
    {
      operands[4] = simplify_gen_subreg (V4SImode, operands[0], <MODE>mode, 0);
      emit_insn (gen_sse2_loadld (operands[4],
				  CONST0_RTX (V4SImode), operands[1]));
    }
  /* We can ignore possible trapping value in the
     high part of SSE register for non-trapping math. */
  else if (SSE_REG_P (op1) && !flag_trapping_math)
    operands[4] = simplify_gen_subreg (V4SImode, operands[1], SImode, 0);
  else
    gcc_unreachable ();
  emit_insn
    (gen_sse2_cvtdq2p<ssemodefsuffix> (operands[3], operands[4]));
  DONE;
})

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SI 1 "memory_operand" "")))]
  "TARGET_SSE2 && TARGET_SSE_MATH
   && TARGET_USE_VECTOR_CONVERTS && optimize_function_for_speed_p (cfun)
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(const_int 0)]
{
  operands[3] = simplify_gen_subreg (<ssevecmode>mode, operands[0],
				     <MODE>mode, 0);
  operands[4] = simplify_gen_subreg (V4SImode, operands[0], <MODE>mode, 0);

  emit_insn (gen_sse2_loadld (operands[4],
			      CONST0_RTX (V4SImode), operands[1]));
  emit_insn
    (gen_sse2_cvtdq2p<ssemodefsuffix> (operands[3], operands[4]));
  DONE;
})

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_sse_with_temp"
  [(set (match_operand:MODEF 0 "register_operand" "=x,x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "r,m")))
  (clobber (match_operand:SSEMODEI24 2 "memory_operand" "=m,X"))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH"
  "#"
  [(set_attr "type" "sseicvt")
   (set_attr "mode" "<MODEF:MODE>")
   (set_attr "athlon_decode" "double,direct")
   (set_attr "amdfam10_decode" "vector,double")
   (set_attr "bdver1_decode" "double,direct")
   (set_attr "fp_int_src" "true")])

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_sse_interunit"
  [(set (match_operand:MODEF 0 "register_operand" "=x,x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "r,m")))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH
   && (TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))"
  "%vcvtsi2s<MODEF:ssemodefsuffix><SSEMODEI24:rex64suffix>\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "sseicvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "mode" "<MODEF:MODE>")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "prefix" "maybe_vex")
	    (ne (symbol_ref "<SSEMODEI24:MODE>mode == DImode") (const_int 0)))
       (const_string "1")
       (const_string "*")))
   (set_attr "athlon_decode" "double,direct")
   (set_attr "amdfam10_decode" "vector,double")
   (set_attr "bdver1_decode" "double,direct")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SSEMODEI24 1 "nonimmediate_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH
   && (TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(set (match_dup 0) (float:MODEF (match_dup 1)))])

(define_insn "*float<SSEMODEI24:mode><MODEF:mode>2_sse_nointerunit"
  [(set (match_operand:MODEF 0 "register_operand" "=x")
	(float:MODEF
	  (match_operand:SSEMODEI24 1 "memory_operand" "m")))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH
   && !(TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))"
  "%vcvtsi2s<MODEF:ssemodefsuffix><SSEMODEI24:rex64suffix>\t{%1, %d0|%d0, %1}"
  [(set_attr "type" "sseicvt")
   (set_attr "prefix" "maybe_vex")
   (set_attr "mode" "<MODEF:MODE>")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "prefix" "maybe_vex")
	    (ne (symbol_ref "<SSEMODEI24:MODE>mode == DImode") (const_int 0)))
       (const_string "1")
       (const_string "*")))
   (set_attr "athlon_decode" "direct")
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SSEMODEI24 1 "register_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH
   && !(TARGET_INTER_UNIT_CONVERSIONS || optimize_function_for_size_p (cfun))
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (float:MODEF (match_dup 2)))])

(define_split
  [(set (match_operand:MODEF 0 "register_operand" "")
	(float:MODEF (match_operand:SSEMODEI24 1 "memory_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "(<SSEMODEI24:MODE>mode != DImode || TARGET_64BIT)
   && SSE_FLOAT_MODE_P (<MODEF:MODE>mode) && TARGET_SSE_MATH
   && reload_completed
   && (SSE_REG_P (operands[0])
       || (GET_CODE (operands[0]) == SUBREG
	   && SSE_REG_P (operands[0])))"
  [(set (match_dup 0) (float:MODEF (match_dup 1)))])

(define_insn "*float<SSEMODEI24:mode><X87MODEF:mode>2_i387_with_temp"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f,f")
	(float:X87MODEF
	  (match_operand:SSEMODEI24 1 "nonimmediate_operand" "m,?r")))
  (clobber (match_operand:SSEMODEI24 2 "memory_operand" "=X,m"))]
  "TARGET_80387
   && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode)"
  "@
   fild%Z1\t%1
   #"
  [(set_attr "type" "fmov,multi")
   (set_attr "mode" "<X87MODEF:MODE>")
   (set_attr "unit" "*,i387")
   (set_attr "fp_int_src" "true")])

(define_insn "*float<SSEMODEI24:mode><X87MODEF:mode>2_i387"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f")
	(float:X87MODEF
	  (match_operand:SSEMODEI24 1 "memory_operand" "m")))]
  "TARGET_80387
   && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode)"
  "fild%Z1\t%1"
  [(set_attr "type" "fmov")
   (set_attr "mode" "<X87MODEF:MODE>")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:SSEMODEI24 1 "register_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "TARGET_80387
   && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode)
   && reload_completed
   && FP_REG_P (operands[0])"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (float:X87MODEF (match_dup 2)))])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:SSEMODEI24 1 "memory_operand" "")))
   (clobber (match_operand:SSEMODEI24 2 "memory_operand" ""))]
  "TARGET_80387
   && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, <SSEMODEI24:MODE>mode)
   && reload_completed
   && FP_REG_P (operands[0])"
  [(set (match_dup 0) (float:X87MODEF (match_dup 1)))])

;; Avoid store forwarding (partial memory) stall penalty
;; by passing DImode value through XMM registers.  */

(define_insn "floatdi<X87MODEF:mode>2_i387_with_xmm"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f,f")
	(float:X87MODEF
	  (match_operand:DI 1 "nonimmediate_operand" "m,?r")))
   (clobber (match_scratch:V4SI 3 "=X,x"))
   (clobber (match_scratch:V4SI 4 "=X,x"))
   (clobber (match_operand:DI 2 "memory_operand" "=X,m"))]
  "TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE2 && TARGET_INTER_UNIT_MOVES
   && !TARGET_64BIT && optimize_function_for_speed_p (cfun)"
  "#"
  [(set_attr "type" "multi")
   (set_attr "mode" "<X87MODEF:MODE>")
   (set_attr "unit" "i387")
   (set_attr "fp_int_src" "true")])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:DI 1 "register_operand" "")))
   (clobber (match_scratch:V4SI 3 ""))
   (clobber (match_scratch:V4SI 4 ""))
   (clobber (match_operand:DI 2 "memory_operand" ""))]
  "TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE2 && TARGET_INTER_UNIT_MOVES
   && !TARGET_64BIT && optimize_function_for_speed_p (cfun)
   && reload_completed
   && FP_REG_P (operands[0])"
  [(set (match_dup 2) (match_dup 3))
   (set (match_dup 0) (float:X87MODEF (match_dup 2)))]
{
  /* The DImode arrived in a pair of integral registers (e.g. %edx:%eax).
     Assemble the 64-bit DImode value in an xmm register.  */
  emit_insn (gen_sse2_loadld (operands[3], CONST0_RTX (V4SImode),
			      gen_rtx_SUBREG (SImode, operands[1], 0)));
  emit_insn (gen_sse2_loadld (operands[4], CONST0_RTX (V4SImode),
			      gen_rtx_SUBREG (SImode, operands[1], 4)));
  emit_insn (gen_vec_interleave_lowv4si (operands[3], operands[3],
  	    				 operands[4]));

  operands[3] = gen_rtx_REG (DImode, REGNO (operands[3]));
})

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(float:X87MODEF (match_operand:DI 1 "memory_operand" "")))
   (clobber (match_scratch:V4SI 3 ""))
   (clobber (match_scratch:V4SI 4 ""))
   (clobber (match_operand:DI 2 "memory_operand" ""))]
  "TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE2 && TARGET_INTER_UNIT_MOVES
   && !TARGET_64BIT && optimize_function_for_speed_p (cfun)
   && reload_completed
   && FP_REG_P (operands[0])"
  [(set (match_dup 0) (float:X87MODEF (match_dup 1)))])

;; Avoid store forwarding (partial memory) stall penalty by extending
;; SImode value to DImode through XMM register instead of pushing two
;; SImode values to stack. Note that even !TARGET_INTER_UNIT_MOVES
;; targets benefit from this optimization. Also note that fild
;; loads from memory only.

(define_insn "*floatunssi<mode>2_1"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f,f")
	(unsigned_float:X87MODEF
	  (match_operand:SI 1 "nonimmediate_operand" "x,m")))
   (clobber (match_operand:DI 2 "memory_operand" "=m,m"))
   (clobber (match_scratch:SI 3 "=X,x"))]
  "!TARGET_64BIT
   && TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE"
  "#"
  [(set_attr "type" "multi")
   (set_attr "mode" "<MODE>")])

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(unsigned_float:X87MODEF
	  (match_operand:SI 1 "register_operand" "")))
   (clobber (match_operand:DI 2 "memory_operand" ""))
   (clobber (match_scratch:SI 3 ""))]
  "!TARGET_64BIT
   && TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE
   && reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0)
	(float:X87MODEF (match_dup 2)))]
  "operands[1] = simplify_gen_subreg (DImode, operands[1], SImode, 0);")

(define_split
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(unsigned_float:X87MODEF
	  (match_operand:SI 1 "memory_operand" "")))
   (clobber (match_operand:DI 2 "memory_operand" ""))
   (clobber (match_scratch:SI 3 ""))]
  "!TARGET_64BIT
   && TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
   && TARGET_SSE
   && reload_completed"
  [(set (match_dup 2) (match_dup 3))
   (set (match_dup 0)
	(float:X87MODEF (match_dup 2)))]
{
  emit_move_insn (operands[3], operands[1]);
  operands[3] = simplify_gen_subreg (DImode, operands[3], SImode, 0);
})

(define_expand "floatunssi<mode>2"
  [(parallel
     [(set (match_operand:X87MODEF 0 "register_operand" "")
	   (unsigned_float:X87MODEF
	     (match_operand:SI 1 "nonimmediate_operand" "")))
      (clobber (match_dup 2))
      (clobber (match_scratch:SI 3 ""))])]
  "!TARGET_64BIT
   && ((TARGET_80387 && X87_ENABLE_FLOAT (<X87MODEF:MODE>mode, DImode)
	&& TARGET_SSE)
       || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))"
{
  if (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
    {
      ix86_expand_convert_uns_si<mode>_sse (operands[0], operands[1]);
      DONE;
    }
  else
    {
      enum ix86_stack_slot slot = (virtuals_instantiated
				   ? SLOT_TEMP
				   : SLOT_VIRTUAL);
      operands[2] = assign_386_stack_local (DImode, slot);
    }
})

(define_expand "floatunsdisf2"
  [(use (match_operand:SF 0 "register_operand" ""))
   (use (match_operand:DI 1 "nonimmediate_operand" ""))]
  "TARGET_64BIT && TARGET_SSE_MATH"
  "x86_emit_floatuns (operands); DONE;")

(define_expand "floatunsdidf2"
  [(use (match_operand:DF 0 "register_operand" ""))
   (use (match_operand:DI 1 "nonimmediate_operand" ""))]
  "(TARGET_64BIT || TARGET_KEEPS_VECTOR_ALIGNED_STACK)
   && TARGET_SSE2 && TARGET_SSE_MATH"
{
  if (TARGET_64BIT)
    x86_emit_floatuns (operands);
  else
    ix86_expand_convert_uns_didf_sse (operands[0], operands[1]);
  DONE;
})

;; Add instructions

(define_expand "add<mode>3"
  [(set (match_operand:SDWIM 0 "nonimmediate_operand" "")
	(plus:SDWIM (match_operand:SDWIM 1 "nonimmediate_operand" "")
		    (match_operand:SDWIM 2 "<general_operand>" "")))]
  ""
  "ix86_expand_binary_operator (PLUS, <MODE>mode, operands); DONE;")

(define_insn_and_split "*add<dwi>3_doubleword"
  [(set (match_operand:<DWI> 0 "nonimmediate_operand" "=r,o")
	(plus:<DWI>
	  (match_operand:<DWI> 1 "nonimmediate_operand" "%0,0")
	  (match_operand:<DWI> 2 "<general_operand>" "ro<di>,r<di>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (PLUS, <DWI>mode, operands)"
  "#"
  "reload_completed"
  [(parallel [(set (reg:CC FLAGS_REG)
		   (unspec:CC [(match_dup 1) (match_dup 2)]
			      UNSPEC_ADD_CARRY))
	      (set (match_dup 0)
		   (plus:DWIH (match_dup 1) (match_dup 2)))])
   (parallel [(set (match_dup 3)
		   (plus:DWIH
		     (match_dup 4)
		     (plus:DWIH
		       (ltu:DWIH (reg:CC FLAGS_REG) (const_int 0))
		       (match_dup 5))))
	      (clobber (reg:CC FLAGS_REG))])]
  "split_double_mode (<DWI>mode, &operands[0], 3, &operands[0], &operands[3]);")

(define_insn "*add<mode>3_cc"
  [(set (reg:CC FLAGS_REG)
	(unspec:CC
	  [(match_operand:SWI48 1 "nonimmediate_operand" "%0,0")
	   (match_operand:SWI48 2 "<general_operand>" "r<i>,rm")]
	  UNSPEC_ADD_CARRY))
   (set (match_operand:SWI48 0 "nonimmediate_operand" "=rm,r")
	(plus:SWI48 (match_dup 1) (match_dup 2)))]
  "ix86_binary_operator_ok (PLUS, <MODE>mode, operands)"
  "add{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "addqi3_cc"
  [(set (reg:CC FLAGS_REG)
	(unspec:CC
	  [(match_operand:QI 1 "nonimmediate_operand" "%0,0")
	   (match_operand:QI 2 "general_operand" "qn,qm")]
	  UNSPEC_ADD_CARRY))
   (set (match_operand:QI 0 "nonimmediate_operand" "=qm,q")
	(plus:QI (match_dup 1) (match_dup 2)))]
  "ix86_binary_operator_ok (PLUS, QImode, operands)"
  "add{b}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "QI")])

(define_insn "*lea_1"
  [(set (match_operand:P 0 "register_operand" "=r")
	(match_operand:P 1 "no_seg_address_operand" "p"))]
  ""
  "lea{<imodesuffix>}\t{%a1, %0|%0, %a1}"
  [(set_attr "type" "lea")
   (set_attr "mode" "<MODE>")])

(define_insn "*lea_2"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(subreg:SI (match_operand:DI 1 "no_seg_address_operand" "p") 0))]
  "TARGET_64BIT"
  "lea{l}\t{%a1, %0|%0, %a1}"
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn "*lea_2_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (subreg:SI (match_operand:DI 1 "no_seg_address_operand" "p") 0)))]
  "TARGET_64BIT"
  "lea{l}\t{%a1, %k0|%k0, %a1}"
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn "*add<mode>_1"
  [(set (match_operand:SWI48 0 "nonimmediate_operand" "=r,rm,r,r")
	(plus:SWI48
	  (match_operand:SWI48 1 "nonimmediate_operand" "%0,0,r,r")
	  (match_operand:SWI48 2 "<general_operand>" "<g>,r<i>,0,l<i>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (PLUS, <MODE>mode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_INCDEC:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (operands[2] == const1_rtx)
        return "inc{<imodesuffix>}\t%0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{<imodesuffix>}\t%0";
	}

    default:
      /* For most processors, ADD is faster than LEA.  This alternative
	 was added to use ADD as much as possible.  */
      if (which_alternative == 2)
	{
	  rtx tmp;
	  tmp = operands[1], operands[1] = operands[2], operands[2] = tmp;
	}
        
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (x86_maybe_negate_const_int (&operands[2], <MODE>mode))
        return "sub{<imodesuffix>}\t{%2, %0|%0, %2}";

      return "add{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "3")
              (const_string "lea")
	    (match_operand:SWI48 2 "incdec_operand" "")
	      (const_string "incdec")
	   ]
	   (const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "<MODE>")])

;; It may seem that nonimmediate operand is proper one for operand 1.
;; The addsi_1 pattern allows nonimmediate operand at that place and
;; we take care in ix86_binary_operator_ok to not allow two memory
;; operands so proper swapping will be done in reload.  This allow
;; patterns constructed from addsi_1 to match.

(define_insn "*addsi_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r,r,r")
	(zero_extend:DI
	  (plus:SI (match_operand:SI 1 "nonimmediate_operand" "%0,r,r")
		   (match_operand:SI 2 "general_operand" "g,0,li"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (PLUS, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{l}\t%k0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{l}\t%k0";
	}

    default:
      /* For most processors, ADD is faster than LEA.  This alternative
	 was added to use ADD as much as possible.  */
      if (which_alternative == 1)
	{
	  rtx tmp;
	  tmp = operands[1], operands[1] = operands[2], operands[2] = tmp;
	}

      if (x86_maybe_negate_const_int (&operands[2], SImode))
        return "sub{l}\t{%2, %k0|%k0, %2}";

      return "add{l}\t{%2, %k0|%k0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "2")
	      (const_string "lea")
	    (match_operand:SI 2 "incdec_operand" "")
	      (const_string "incdec")
	   ]
	   (const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*addhi_1"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=rm,r")
	(plus:HI (match_operand:HI 1 "nonimmediate_operand" "%0,0")
		 (match_operand:HI 2 "general_operand" "rn,rm")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (PLUS, HImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
	return "inc{w}\t%0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
	  return "dec{w}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], HImode))
	return "sub{w}\t{%2, %0|%0, %2}";

      return "add{w}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:HI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "HI")])

(define_insn "*addhi_1_lea"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=r,rm,r,r")
	(plus:HI (match_operand:HI 1 "nonimmediate_operand" "%0,0,r,r")
		 (match_operand:HI 2 "general_operand" "rmn,rn,0,ln")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (PLUS, HImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_INCDEC:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (operands[2] == const1_rtx)
	return "inc{w}\t%0";
      else
	{
	  gcc_assert (operands[2] == constm1_rtx);
	  return "dec{w}\t%0";
	}

    default:
      /* For most processors, ADD is faster than LEA.  This alternative
	 was added to use ADD as much as possible.  */
      if (which_alternative == 2)
	{
	  rtx tmp;
	  tmp = operands[1], operands[1] = operands[2], operands[2] = tmp;
	}

      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (x86_maybe_negate_const_int (&operands[2], HImode))
	return "sub{w}\t{%2, %0|%0, %2}";

      return "add{w}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "3")
              (const_string "lea")
	    (match_operand:HI 2 "incdec_operand" "")
	      (const_string "incdec")
	   ]
	   (const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "HI,HI,HI,SI")])

;; %%% Potential partial reg stall on alternative 2.  What to do?
(define_insn "*addqi_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm,q,r")
	(plus:QI (match_operand:QI 1 "nonimmediate_operand" "%0,0,0")
		 (match_operand:QI 2 "general_operand" "qn,qmn,rn")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (PLUS, QImode, operands)"
{
  int widen = (which_alternative == 2);
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
	return widen ? "inc{l}\t%k0" : "inc{b}\t%0";
      else
	{
	  gcc_assert (operands[2] == constm1_rtx);
	  return widen ? "dec{l}\t%k0" : "dec{b}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], QImode))
	{
	  if (widen)
	    return "sub{l}\t{%2, %k0|%k0, %2}";
	  else
	    return "sub{b}\t{%2, %0|%0, %2}";
	}
      if (widen)
        return "add{l}\t{%k2, %k0|%k0, %k2}";
      else
        return "add{b}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:QI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "QI,QI,SI")])

;; %%% Potential partial reg stall on alternatives 3 and 4.  What to do?
(define_insn "*addqi_1_lea"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=q,qm,q,r,r,r")
	(plus:QI (match_operand:QI 1 "nonimmediate_operand" "%0,0,q,0,r,r")
		 (match_operand:QI 2 "general_operand" "qmn,qn,0,rn,0,ln")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (PLUS, QImode, operands)"
{
  int widen = (which_alternative == 3 || which_alternative == 4);

  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_INCDEC:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (operands[2] == const1_rtx)
	return widen ? "inc{l}\t%k0" : "inc{b}\t%0";
      else
	{
	  gcc_assert (operands[2] == constm1_rtx);
	  return widen ? "dec{l}\t%k0" : "dec{b}\t%0";
	}

    default:
      /* For most processors, ADD is faster than LEA.  These alternatives
	 were added to use ADD as much as possible.  */
      if (which_alternative == 2 || which_alternative == 4)
	{
	  rtx tmp;
	  tmp = operands[1], operands[1] = operands[2], operands[2] = tmp;
	}

      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (x86_maybe_negate_const_int (&operands[2], QImode))
	{
	  if (widen)
	    return "sub{l}\t{%2, %k0|%k0, %2}";
	  else
	    return "sub{b}\t{%2, %0|%0, %2}";
	}
      if (widen)
        return "add{l}\t{%k2, %k0|%k0, %k2}";
      else
        return "add{b}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "5")
              (const_string "lea")
	    (match_operand:QI 2 "incdec_operand" "")
	      (const_string "incdec")
	   ]
	   (const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "QI,QI,QI,SI,SI,SI")])

(define_insn "*addqi_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm,q"))
	(plus:QI (match_dup 0)
		 (match_operand:QI 1 "general_operand" "qn,qnm")))
   (clobber (reg:CC FLAGS_REG))]
  "(! TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[1] == const1_rtx)
	return "inc{b}\t%0";
      else
	{
	  gcc_assert (operands[1] == constm1_rtx);
	  return "dec{b}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[1], QImode))
	return "sub{b}\t{%1, %0|%0, %1}";

      return "add{b}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:QI 1 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu1")))
   (set (attr "memory")
     (if_then_else (match_operand 1 "memory_operand" "")
        (const_string "load")
        (const_string "none")))
   (set_attr "mode" "QI")])

;; Convert lea to the lea pattern to avoid flags dependency.
(define_split
  [(set (match_operand 0 "register_operand" "")
	(plus (match_operand 1 "register_operand" "")
              (match_operand 2 "nonmemory_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed && ix86_lea_for_add_ok (insn, operands)" 
  [(const_int 0)]
{
  rtx pat;
  enum machine_mode mode = GET_MODE (operands[0]);

  /* In -fPIC mode the constructs like (const (unspec [symbol_ref]))
     may confuse gen_lowpart.  */
  if (mode != Pmode)
    {
      operands[1] = gen_lowpart (Pmode, operands[1]);
      operands[2] = gen_lowpart (Pmode, operands[2]);
    }

  pat = gen_rtx_PLUS (Pmode, operands[1], operands[2]);

  if (GET_MODE_SIZE (mode) < GET_MODE_SIZE (SImode))
    operands[0] = gen_lowpart (SImode, operands[0]);

  if (TARGET_64BIT && mode != Pmode)
    pat = gen_rtx_SUBREG (SImode, pat, 0);

  emit_insn (gen_rtx_SET (VOIDmode, operands[0], pat));
  DONE;
})

;; Convert lea to the lea pattern to avoid flags dependency.
;; ??? This pattern handles immediate operands that do not satisfy immediate
;; operand predicate (LEGITIMATE_CONSTANT_P) in the previous pattern.
(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(plus:DI (match_operand:DI 1 "register_operand" "")
		 (match_operand:DI 2 "x86_64_immediate_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && reload_completed 
   && true_regnum (operands[0]) != true_regnum (operands[1])"
  [(set (match_dup 0)
	(plus:DI (match_dup 1) (match_dup 2)))])

;; Convert lea to the lea pattern to avoid flags dependency.
(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(zero_extend:DI
	  (plus:SI (match_operand:SI 1 "register_operand" "")
		   (match_operand:SI 2 "nonmemory_operand" ""))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && reload_completed
   && ix86_lea_for_add_ok (insn, operands)"
  [(set (match_dup 0)
	(zero_extend:DI (subreg:SI (plus:DI (match_dup 1) (match_dup 2)) 0)))]
{
  operands[1] = gen_lowpart (DImode, operands[1]);
  operands[2] = gen_lowpart (DImode, operands[2]);
})

(define_insn "*add<mode>_2"
  [(set (reg FLAGS_REG)
	(compare
	  (plus:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "%0,0")
	    (match_operand:SWI 2 "<general_operand>" "<g>,<r><i>"))
	  (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>,<r>m")
	(plus:SWI (match_dup 1) (match_dup 2)))]
  "ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (PLUS, <MODE>mode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{<imodesuffix>}\t%0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{<imodesuffix>}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], <MODE>mode))
        return "sub{<imodesuffix>}\t{%2, %0|%0, %2}";

      return "add{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:SWI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "<MODE>")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
(define_insn "*addsi_2_zext"
  [(set (reg FLAGS_REG)
	(compare
	  (plus:SI (match_operand:SI 1 "nonimmediate_operand" "%0")
		   (match_operand:SI 2 "general_operand" "g"))
	  (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (plus:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (PLUS, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{l}\t%k0";
      else
	{
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{l}\t%k0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], SImode))
        return "sub{l}\t{%2, %k0|%k0, %2}";

      return "add{l}\t{%2, %k0|%k0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:SI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*add<mode>_3"
  [(set (reg FLAGS_REG)
	(compare
	  (neg:SWI (match_operand:SWI 2 "<general_operand>" "<g>"))
	  (match_operand:SWI 1 "nonimmediate_operand" "%0")))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "ix86_match_ccmode (insn, CCZmode)
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{<imodesuffix>}\t%0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{<imodesuffix>}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], <MODE>mode))
        return "sub{<imodesuffix>}\t{%2, %0|%0, %2}";

      return "add{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:SWI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "<MODE>")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
(define_insn "*addsi_3_zext"
  [(set (reg FLAGS_REG)
	(compare
	  (neg:SI (match_operand:SI 2 "general_operand" "g"))
	  (match_operand:SI 1 "nonimmediate_operand" "%0")))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (plus:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCZmode)
   && ix86_binary_operator_ok (PLUS, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{l}\t%k0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{l}\t%k0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], SImode))
        return "sub{l}\t{%2, %k0|%k0, %2}";

      return "add{l}\t{%2, %k0|%k0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:SI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "SI")])

; For comparisons against 1, -1 and 128, we may generate better code
; by converting cmp to add, inc or dec as done by peephole2.  This pattern
; is matched then.  We can't accept general immediate, because for
; case of overflows,  the result is messed up.
; Also carry flag is reversed compared to cmp, so this conversion is valid
; only for comparisons not depending on it.

(define_insn "*adddi_4"
  [(set (reg FLAGS_REG)
	(compare
	  (match_operand:DI 1 "nonimmediate_operand" "0")
	  (match_operand:DI 2 "x86_64_immediate_operand" "e")))
   (clobber (match_scratch:DI 0 "=rm"))]
  "TARGET_64BIT
   && ix86_match_ccmode (insn, CCGCmode)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == constm1_rtx)
        return "inc{q}\t%0";
      else
        {
	  gcc_assert (operands[2] == const1_rtx);
          return "dec{q}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], DImode))
	return "add{q}\t{%2, %0|%0, %2}";

      return "sub{q}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:DI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "DI")])

; For comparisons against 1, -1 and 128, we may generate better code
; by converting cmp to add, inc or dec as done by peephole2.  This pattern
; is matched then.  We can't accept general immediate, because for
; case of overflows,  the result is messed up.
; Also carry flag is reversed compared to cmp, so this conversion is valid
; only for comparisons not depending on it.

(define_insn "*add<mode>_4"
  [(set (reg FLAGS_REG)
	(compare
	  (match_operand:SWI124 1 "nonimmediate_operand" "0")
	  (match_operand:SWI124 2 "const_int_operand" "n")))
   (clobber (match_scratch:SWI124 0 "=<r>m"))]
  "ix86_match_ccmode (insn, CCGCmode)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == constm1_rtx)
        return "inc{<imodesuffix>}\t%0";
      else
        {
	  gcc_assert (operands[2] == const1_rtx);
          return "dec{<imodesuffix>}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], <MODE>mode))
	return "add{<imodesuffix>}\t{%2, %0|%0, %2}";

      return "sub{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:<MODE> 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*add<mode>_5"
  [(set (reg FLAGS_REG)
	(compare
	  (plus:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "%0")
	    (match_operand:SWI 2 "<general_operand>" "<g>"))
	  (const_int 0)))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "ix86_match_ccmode (insn, CCGOCmode)
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
        return "inc{<imodesuffix>}\t%0";
      else
        {
          gcc_assert (operands[2] == constm1_rtx);
          return "dec{<imodesuffix>}\t%0";
	}

    default:
      if (x86_maybe_negate_const_int (&operands[2], <MODE>mode))
        return "sub{<imodesuffix>}\t{%2, %0|%0, %2}";

      return "add{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:SWI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set (attr "length_immediate")
      (if_then_else
	(and (eq_attr "type" "alu") (match_operand 2 "const128_operand" ""))
	(const_string "1")
	(const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*addqi_ext_1_rex64"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(plus:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (match_operand:QI 2 "nonmemory_operand" "Qn")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
	return "inc{b}\t%h0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{b}\t%h0";
        }

    default:
      return "add{b}\t{%2, %h0|%h0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:QI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "addqi_ext_1"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(plus:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (match_operand:QI 2 "general_operand" "Qmn")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT"
{
  switch (get_attr_type (insn))
    {
    case TYPE_INCDEC:
      if (operands[2] == const1_rtx)
	return "inc{b}\t%h0";
      else
        {
	  gcc_assert (operands[2] == constm1_rtx);
          return "dec{b}\t%h0";
	}

    default:
      return "add{b}\t{%2, %h0|%h0, %2}";
    }
}
  [(set (attr "type")
     (if_then_else (match_operand:QI 2 "incdec_operand" "")
	(const_string "incdec")
	(const_string "alu")))
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*addqi_ext_2"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(plus:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "%0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extract:SI
	    (match_operand 2 "ext_register_operand" "Q")
	    (const_int 8)
	    (const_int 8))))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "add{b}\t{%h2, %h0|%h0, %h2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "QI")])

;; The lea patterns for non-Pmodes needs to be matched by
;; several insns converted to real lea by splitters.

(define_insn_and_split "*lea_general_1"
  [(set (match_operand 0 "register_operand" "=r")
	(plus (plus (match_operand 1 "index_register_operand" "l")
		    (match_operand 2 "register_operand" "r"))
	      (match_operand 3 "immediate_operand" "i")))]
  "(GET_MODE (operands[0]) == QImode || GET_MODE (operands[0]) == HImode
    || (TARGET_64BIT && GET_MODE (operands[0]) == SImode))
   && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])
   && GET_MODE (operands[0]) == GET_MODE (operands[2])
   && (GET_MODE (operands[0]) == GET_MODE (operands[3])
       || GET_MODE (operands[3]) == VOIDmode)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
{
  rtx pat;
  operands[0] = gen_lowpart (SImode, operands[0]);
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[2] = gen_lowpart (Pmode, operands[2]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
  pat = gen_rtx_PLUS (Pmode, gen_rtx_PLUS (Pmode, operands[1], operands[2]),
  		      operands[3]);
  if (Pmode != SImode)
    pat = gen_rtx_SUBREG (SImode, pat, 0);
  emit_insn (gen_rtx_SET (VOIDmode, operands[0], pat));
  DONE;
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn_and_split "*lea_general_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (plus:SI (plus:SI
		     (match_operand:SI 1 "index_register_operand" "l")
		     (match_operand:SI 2 "register_operand" "r"))
		   (match_operand:SI 3 "immediate_operand" "i"))))]
  "TARGET_64BIT"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(zero_extend:DI (subreg:SI (plus:DI (plus:DI (match_dup 1)
						     (match_dup 2))
					    (match_dup 3)) 0)))]
{
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[2] = gen_lowpart (Pmode, operands[2]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn_and_split "*lea_general_2"
  [(set (match_operand 0 "register_operand" "=r")
	(plus (mult (match_operand 1 "index_register_operand" "l")
		    (match_operand 2 "const248_operand" "i"))
	      (match_operand 3 "nonmemory_operand" "ri")))]
  "(GET_MODE (operands[0]) == QImode || GET_MODE (operands[0]) == HImode
    || (TARGET_64BIT && GET_MODE (operands[0]) == SImode))
   && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])
   && (GET_MODE (operands[0]) == GET_MODE (operands[3])
       || GET_MODE (operands[3]) == VOIDmode)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
{
  rtx pat;
  operands[0] = gen_lowpart (SImode, operands[0]);
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
  pat = gen_rtx_PLUS (Pmode, gen_rtx_MULT (Pmode, operands[1], operands[2]),
  		      operands[3]);
  if (Pmode != SImode)
    pat = gen_rtx_SUBREG (SImode, pat, 0);
  emit_insn (gen_rtx_SET (VOIDmode, operands[0], pat));
  DONE;
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn_and_split "*lea_general_2_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (plus:SI (mult:SI
		     (match_operand:SI 1 "index_register_operand" "l")
		     (match_operand:SI 2 "const248_operand" "n"))
		   (match_operand:SI 3 "nonmemory_operand" "ri"))))]
  "TARGET_64BIT"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(zero_extend:DI (subreg:SI (plus:DI (mult:DI (match_dup 1)
						     (match_dup 2))
					    (match_dup 3)) 0)))]
{
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn_and_split "*lea_general_3"
  [(set (match_operand 0 "register_operand" "=r")
	(plus (plus (mult (match_operand 1 "index_register_operand" "l")
			  (match_operand 2 "const248_operand" "i"))
		    (match_operand 3 "register_operand" "r"))
	      (match_operand 4 "immediate_operand" "i")))]
  "(GET_MODE (operands[0]) == QImode || GET_MODE (operands[0]) == HImode
    || (TARGET_64BIT && GET_MODE (operands[0]) == SImode))
   && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && GET_MODE (operands[0]) == GET_MODE (operands[1])
   && GET_MODE (operands[0]) == GET_MODE (operands[3])"
  "#"
  "&& reload_completed"
  [(const_int 0)]
{
  rtx pat;
  operands[0] = gen_lowpart (SImode, operands[0]);
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
  operands[4] = gen_lowpart (Pmode, operands[4]);
  pat = gen_rtx_PLUS (Pmode,
  		      gen_rtx_PLUS (Pmode, gen_rtx_MULT (Pmode, operands[1],
		      					 operands[2]),
				    operands[3]),
  		      operands[4]);
  if (Pmode != SImode)
    pat = gen_rtx_SUBREG (SImode, pat, 0);
  emit_insn (gen_rtx_SET (VOIDmode, operands[0], pat));
  DONE;
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

(define_insn_and_split "*lea_general_3_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (plus:SI (plus:SI
		     (mult:SI
		       (match_operand:SI 1 "index_register_operand" "l")
		       (match_operand:SI 2 "const248_operand" "n"))
		     (match_operand:SI 3 "register_operand" "r"))
		   (match_operand:SI 4 "immediate_operand" "i"))))]
  "TARGET_64BIT"
  "#"
  "&& reload_completed"
  [(set (match_dup 0)
	(zero_extend:DI (subreg:SI (plus:DI (plus:DI (mult:DI (match_dup 1)
							      (match_dup 2))
						     (match_dup 3))
					    (match_dup 4)) 0)))]
{
  operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[3] = gen_lowpart (Pmode, operands[3]);
  operands[4] = gen_lowpart (Pmode, operands[4]);
}
  [(set_attr "type" "lea")
   (set_attr "mode" "SI")])

;; Subtract instructions

(define_expand "sub<mode>3"
  [(set (match_operand:SDWIM 0 "nonimmediate_operand" "")
	(minus:SDWIM (match_operand:SDWIM 1 "nonimmediate_operand" "")
		     (match_operand:SDWIM 2 "<general_operand>" "")))]
  ""
  "ix86_expand_binary_operator (MINUS, <MODE>mode, operands); DONE;")

(define_insn_and_split "*sub<dwi>3_doubleword"
  [(set (match_operand:<DWI> 0 "nonimmediate_operand" "=r,o")
	(minus:<DWI>
	  (match_operand:<DWI> 1 "nonimmediate_operand" "0,0")
	  (match_operand:<DWI> 2 "<general_operand>" "ro<di>,r<di>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (MINUS, <MODE>mode, operands)"
  "#"
  "reload_completed"
  [(parallel [(set (reg:CC FLAGS_REG)
		   (compare:CC (match_dup 1) (match_dup 2)))
	      (set (match_dup 0)
		   (minus:DWIH (match_dup 1) (match_dup 2)))])
   (parallel [(set (match_dup 3)
		   (minus:DWIH
		     (match_dup 4)
		     (plus:DWIH
		       (ltu:DWIH (reg:CC FLAGS_REG) (const_int 0))
		       (match_dup 5))))
	      (clobber (reg:CC FLAGS_REG))])]
  "split_double_mode (<DWI>mode, &operands[0], 3, &operands[0], &operands[3]);")

(define_insn "*sub<mode>_1"
  [(set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m,<r>")
	(minus:SWI
	  (match_operand:SWI 1 "nonimmediate_operand" "0,0")
	  (match_operand:SWI 2 "<general_operand>" "<r><i>,<r>m")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (MINUS, <MODE>mode, operands)"
  "sub{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*subsi_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (minus:SI (match_operand:SI 1 "register_operand" "0")
		    (match_operand:SI 2 "general_operand" "g"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (MINUS, SImode, operands)"
  "sub{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*subqi_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm,q"))
	(minus:QI (match_dup 0)
		  (match_operand:QI 1 "general_operand" "qn,qm")))
   (clobber (reg:CC FLAGS_REG))]
  "(! TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "sub{b}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "QI")])

(define_insn "*sub<mode>_2"
  [(set (reg FLAGS_REG)
	(compare
	  (minus:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "0,0")
	    (match_operand:SWI 2 "<general_operand>" "<r><i>,<r>m"))
	  (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m,<r>")
	(minus:SWI (match_dup 1) (match_dup 2)))]
  "ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (MINUS, <MODE>mode, operands)"
  "sub{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*subsi_2_zext"
  [(set (reg FLAGS_REG)
	(compare
	  (minus:SI (match_operand:SI 1 "register_operand" "0")
		    (match_operand:SI 2 "general_operand" "g"))
	  (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (minus:SI (match_dup 1)
		    (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (MINUS, SImode, operands)"
  "sub{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*sub<mode>_3"
  [(set (reg FLAGS_REG)
	(compare (match_operand:SWI 1 "nonimmediate_operand" "0,0")
		 (match_operand:SWI 2 "<general_operand>" "<r><i>,<r>m")))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m,<r>")
	(minus:SWI (match_dup 1) (match_dup 2)))]
  "ix86_match_ccmode (insn, CCmode)
   && ix86_binary_operator_ok (MINUS, <MODE>mode, operands)"
  "sub{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*subsi_3_zext"
  [(set (reg FLAGS_REG)
	(compare (match_operand:SI 1 "register_operand" "0")
		 (match_operand:SI 2 "general_operand" "g")))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (minus:SI (match_dup 1)
		    (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCmode)
   && ix86_binary_operator_ok (MINUS, SImode, operands)"
  "sub{l}\t{%2, %1|%1, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

;; Add with carry and subtract with borrow

(define_expand "<plusminus_insn><mode>3_carry"
  [(parallel
    [(set (match_operand:SWI 0 "nonimmediate_operand" "")
	  (plusminus:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "")
	    (plus:SWI (match_operator:SWI 4 "ix86_carry_flag_operator"
		       [(match_operand 3 "flags_reg_operand" "")
			(const_int 0)])
		      (match_operand:SWI 2 "<general_operand>" ""))))
     (clobber (reg:CC FLAGS_REG))])]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)")

(define_insn "*<plusminus_insn><mode>3_carry"
  [(set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m,<r>")
	(plusminus:SWI
	  (match_operand:SWI 1 "nonimmediate_operand" "<comm>0,0")
	  (plus:SWI
	    (match_operator 3 "ix86_carry_flag_operator"
	     [(reg FLAGS_REG) (const_int 0)])
	    (match_operand:SWI 2 "<general_operand>" "<r><i>,<r>m"))))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (PLUS, <MODE>mode, operands)"
  "<plusminus_carry_mnemonic>{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "use_carry" "1")
   (set_attr "pent_pair" "pu")
   (set_attr "mode" "<MODE>")])

(define_insn "*addsi3_carry_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (plus:SI (match_operand:SI 1 "nonimmediate_operand" "%0")
		   (plus:SI (match_operator 3 "ix86_carry_flag_operator"
			     [(reg FLAGS_REG) (const_int 0)])
			    (match_operand:SI 2 "general_operand" "g")))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (PLUS, SImode, operands)"
  "adc{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "use_carry" "1")
   (set_attr "pent_pair" "pu")
   (set_attr "mode" "SI")])

(define_insn "*subsi3_carry_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (minus:SI (match_operand:SI 1 "register_operand" "0")
		    (plus:SI (match_operator 3 "ix86_carry_flag_operator"
			      [(reg FLAGS_REG) (const_int 0)])
			     (match_operand:SI 2 "general_operand" "g")))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (MINUS, SImode, operands)"
  "sbb{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "pent_pair" "pu")
   (set_attr "mode" "SI")])

;; Overflow setting add and subtract instructions

(define_insn "*add<mode>3_cconly_overflow"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (plus:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "%0")
	    (match_operand:SWI 2 "<general_operand>" "<g>"))
	  (match_dup 1)))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "!(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "add{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*sub<mode>3_cconly_overflow"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (minus:SWI
	    (match_operand:SWI 0 "nonimmediate_operand" "<r>m,<r>")
	    (match_operand:SWI 1 "<general_operand>" "<r><i>,<r>m"))
	  (match_dup 0)))]
  ""
  "cmp{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "icmp")
   (set_attr "mode" "<MODE>")])

(define_insn "*<plusminus_insn><mode>3_cc_overflow"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	    (plusminus:SWI
		(match_operand:SWI 1 "nonimmediate_operand" "<comm>0,0")
		(match_operand:SWI 2 "<general_operand>" "<r><i>,<r>m"))
	    (match_dup 1)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m,<r>")
	(plusminus:SWI (match_dup 1) (match_dup 2)))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
  "<plusminus_mnemonic>{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*<plusminus_insn>si3_zext_cc_overflow"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (plusminus:SI
	    (match_operand:SI 1 "nonimmediate_operand" "<comm>0")
	    (match_operand:SI 2 "general_operand" "g"))
	  (match_dup 1)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (plusminus:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT && ix86_binary_operator_ok (<CODE>, SImode, operands)"
  "<plusminus_mnemonic>{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

;; The patterns that match these are at the end of this file.

(define_expand "<plusminus_insn>xf3"
  [(set (match_operand:XF 0 "register_operand" "")
	(plusminus:XF
	  (match_operand:XF 1 "register_operand" "")
	  (match_operand:XF 2 "register_operand" "")))]
  "TARGET_80387")

(define_expand "<plusminus_insn><mode>3"
  [(set (match_operand:MODEF 0 "register_operand" "")
	(plusminus:MODEF
	  (match_operand:MODEF 1 "register_operand" "")
	  (match_operand:MODEF 2 "nonimmediate_operand" "")))]
  "(TARGET_80387 && X87_ENABLE_ARITH (<MODE>mode))
    || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)")

;; Multiply instructions

(define_expand "mul<mode>3"
  [(parallel [(set (match_operand:SWIM248 0 "register_operand" "")
		   (mult:SWIM248
		     (match_operand:SWIM248 1 "register_operand" "")
		     (match_operand:SWIM248 2 "<general_operand>" "")))
	      (clobber (reg:CC FLAGS_REG))])])

(define_expand "mulqi3"
  [(parallel [(set (match_operand:QI 0 "register_operand" "")
		   (mult:QI
		     (match_operand:QI 1 "register_operand" "")
		     (match_operand:QI 2 "nonimmediate_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_QIMODE_MATH")

;; On AMDFAM10
;; IMUL reg32/64, reg32/64, imm8 	Direct
;; IMUL reg32/64, mem32/64, imm8 	VectorPath
;; IMUL reg32/64, reg32/64, imm32 	Direct
;; IMUL reg32/64, mem32/64, imm32 	VectorPath
;; IMUL reg32/64, reg32/64 		Direct
;; IMUL reg32/64, mem32/64 		Direct
;;
;; On BDVER1, all above IMULs use DirectPath

(define_insn "*mul<mode>3_1"
  [(set (match_operand:SWI48 0 "register_operand" "=r,r,r")
	(mult:SWI48
	  (match_operand:SWI48 1 "nonimmediate_operand" "%rm,rm,0")
	  (match_operand:SWI48 2 "<general_operand>" "K,<i>,mr")))
   (clobber (reg:CC FLAGS_REG))]
  "!(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "@
   imul{<imodesuffix>}\t{%2, %1, %0|%0, %1, %2}
   imul{<imodesuffix>}\t{%2, %1, %0|%0, %1, %2}
   imul{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "imul")
   (set_attr "prefix_0f" "0,0,1")
   (set (attr "athlon_decode")
	(cond [(eq_attr "cpu" "athlon")
		  (const_string "vector")
	       (eq_attr "alternative" "1")
		  (const_string "vector")
	       (and (eq_attr "alternative" "2")
		    (match_operand 1 "memory_operand" ""))
		  (const_string "vector")]
	      (const_string "direct")))
   (set (attr "amdfam10_decode")
	(cond [(and (eq_attr "alternative" "0,1")
		    (match_operand 1 "memory_operand" ""))
		  (const_string "vector")]
	      (const_string "direct")))
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "<MODE>")])

(define_insn "*mulsi3_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r,r,r")
	(zero_extend:DI
	  (mult:SI (match_operand:SI 1 "nonimmediate_operand" "%rm,rm,0")
		   (match_operand:SI 2 "general_operand" "K,i,mr"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "@
   imul{l}\t{%2, %1, %k0|%k0, %1, %2}
   imul{l}\t{%2, %1, %k0|%k0, %1, %2}
   imul{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "imul")
   (set_attr "prefix_0f" "0,0,1")
   (set (attr "athlon_decode")
	(cond [(eq_attr "cpu" "athlon")
		  (const_string "vector")
	       (eq_attr "alternative" "1")
		  (const_string "vector")
	       (and (eq_attr "alternative" "2")
		    (match_operand 1 "memory_operand" ""))
		  (const_string "vector")]
	      (const_string "direct")))
   (set (attr "amdfam10_decode")
	(cond [(and (eq_attr "alternative" "0,1")
		    (match_operand 1 "memory_operand" ""))
		  (const_string "vector")]
	      (const_string "direct")))
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "SI")])

;; On AMDFAM10
;; IMUL reg16, reg16, imm8 	VectorPath
;; IMUL reg16, mem16, imm8 	VectorPath
;; IMUL reg16, reg16, imm16 	VectorPath
;; IMUL reg16, mem16, imm16 	VectorPath
;; IMUL reg16, reg16 		Direct
;; IMUL reg16, mem16 		Direct
;;
;; On BDVER1, all HI MULs use DoublePath

(define_insn "*mulhi3_1"
  [(set (match_operand:HI 0 "register_operand" "=r,r,r")
	(mult:HI (match_operand:HI 1 "nonimmediate_operand" "%rm,rm,0")
		 (match_operand:HI 2 "general_operand" "K,n,mr")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_HIMODE_MATH
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "@
   imul{w}\t{%2, %1, %0|%0, %1, %2}
   imul{w}\t{%2, %1, %0|%0, %1, %2}
   imul{w}\t{%2, %0|%0, %2}"
  [(set_attr "type" "imul")
   (set_attr "prefix_0f" "0,0,1")
   (set (attr "athlon_decode")
	(cond [(eq_attr "cpu" "athlon")
		  (const_string "vector")
	       (eq_attr "alternative" "1,2")
		  (const_string "vector")]
	      (const_string "direct")))
   (set (attr "amdfam10_decode")
	(cond [(eq_attr "alternative" "0,1")
		  (const_string "vector")]
	      (const_string "direct")))
   (set_attr "bdver1_decode" "double")
   (set_attr "mode" "HI")])

;;On AMDFAM10 and BDVER1
;; MUL reg8 	Direct
;; MUL mem8 	Direct

(define_insn "*mulqi3_1"
  [(set (match_operand:QI 0 "register_operand" "=a")
	(mult:QI (match_operand:QI 1 "nonimmediate_operand" "%0")
		 (match_operand:QI 2 "nonimmediate_operand" "qm")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_QIMODE_MATH
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "mul{b}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "direct")))
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "QI")])

(define_expand "<u>mul<mode><dwi>3"
  [(parallel [(set (match_operand:<DWI> 0 "register_operand" "")
		   (mult:<DWI>
		     (any_extend:<DWI>
		       (match_operand:DWIH 1 "nonimmediate_operand" ""))
		     (any_extend:<DWI>
		       (match_operand:DWIH 2 "register_operand" ""))))
	      (clobber (reg:CC FLAGS_REG))])])

(define_expand "<u>mulqihi3"
  [(parallel [(set (match_operand:HI 0 "register_operand" "")
		   (mult:HI
		     (any_extend:HI
		       (match_operand:QI 1 "nonimmediate_operand" ""))
		     (any_extend:HI
		       (match_operand:QI 2 "register_operand" ""))))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_QIMODE_MATH")

(define_insn "*<u>mul<mode><dwi>3_1"
  [(set (match_operand:<DWI> 0 "register_operand" "=A")
	(mult:<DWI>
	  (any_extend:<DWI>
	    (match_operand:DWIH 1 "nonimmediate_operand" "%0"))
	  (any_extend:<DWI>
	    (match_operand:DWIH 2 "nonimmediate_operand" "rm"))))
   (clobber (reg:CC FLAGS_REG))]
  "!(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<sgnprefix>mul{<imodesuffix>}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "double")))
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "<MODE>")])

(define_insn "*<u>mulqihi3_1"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(mult:HI
	  (any_extend:HI
	    (match_operand:QI 1 "nonimmediate_operand" "%0"))
	  (any_extend:HI
	    (match_operand:QI 2 "nonimmediate_operand" "qm"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_QIMODE_MATH
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<sgnprefix>mul{b}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "direct")))
   (set_attr "amdfam10_decode" "direct")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "QI")])

(define_expand "<s>mul<mode>3_highpart"
  [(parallel [(set (match_operand:SWI48 0 "register_operand" "")
		   (truncate:SWI48
		     (lshiftrt:<DWI>
		       (mult:<DWI>
			 (any_extend:<DWI>
			   (match_operand:SWI48 1 "nonimmediate_operand" ""))
			 (any_extend:<DWI>
			   (match_operand:SWI48 2 "register_operand" "")))
		       (match_dup 4))))
	      (clobber (match_scratch:SWI48 3 ""))
	      (clobber (reg:CC FLAGS_REG))])]
  ""
  "operands[4] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode));")

(define_insn "*<s>muldi3_highpart_1"
  [(set (match_operand:DI 0 "register_operand" "=d")
	(truncate:DI
	  (lshiftrt:TI
	    (mult:TI
	      (any_extend:TI
		(match_operand:DI 1 "nonimmediate_operand" "%a"))
	      (any_extend:TI
		(match_operand:DI 2 "nonimmediate_operand" "rm")))
	    (const_int 64))))
   (clobber (match_scratch:DI 3 "=1"))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<sgnprefix>mul{q}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "double")))
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "DI")])

(define_insn "*<s>mulsi3_highpart_1"
  [(set (match_operand:SI 0 "register_operand" "=d")
	(truncate:SI
	  (lshiftrt:DI
	    (mult:DI
	      (any_extend:DI
		(match_operand:SI 1 "nonimmediate_operand" "%a"))
	      (any_extend:DI
		(match_operand:SI 2 "nonimmediate_operand" "rm")))
	    (const_int 32))))
   (clobber (match_scratch:SI 3 "=1"))
   (clobber (reg:CC FLAGS_REG))]
  "!(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<sgnprefix>mul{l}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "double")))
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "SI")])

(define_insn "*<s>mulsi3_highpart_zext"
  [(set (match_operand:DI 0 "register_operand" "=d")
	(zero_extend:DI (truncate:SI
	  (lshiftrt:DI
	    (mult:DI (any_extend:DI
		       (match_operand:SI 1 "nonimmediate_operand" "%a"))
		     (any_extend:DI
		       (match_operand:SI 2 "nonimmediate_operand" "rm")))
	    (const_int 32)))))
   (clobber (match_scratch:SI 3 "=1"))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<sgnprefix>mul{l}\t%2"
  [(set_attr "type" "imul")
   (set_attr "length_immediate" "0")
   (set (attr "athlon_decode")
     (if_then_else (eq_attr "cpu" "athlon")
        (const_string "vector")
        (const_string "double")))
   (set_attr "amdfam10_decode" "double")
   (set_attr "bdver1_decode" "direct")
   (set_attr "mode" "SI")])

;; The patterns that match these are at the end of this file.

(define_expand "mulxf3"
  [(set (match_operand:XF 0 "register_operand" "")
	(mult:XF (match_operand:XF 1 "register_operand" "")
		 (match_operand:XF 2 "register_operand" "")))]
  "TARGET_80387")

(define_expand "mul<mode>3"
  [(set (match_operand:MODEF 0 "register_operand" "")
	(mult:MODEF (match_operand:MODEF 1 "register_operand" "")
		    (match_operand:MODEF 2 "nonimmediate_operand" "")))]
  "(TARGET_80387 && X87_ENABLE_ARITH (<MODE>mode))
    || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)")

;; Divide instructions

;; The patterns that match these are at the end of this file.

(define_expand "divxf3"
  [(set (match_operand:XF 0 "register_operand" "")
	(div:XF (match_operand:XF 1 "register_operand" "")
		(match_operand:XF 2 "register_operand" "")))]
  "TARGET_80387")

(define_expand "divdf3"
  [(set (match_operand:DF 0 "register_operand" "")
 	(div:DF (match_operand:DF 1 "register_operand" "")
 		(match_operand:DF 2 "nonimmediate_operand" "")))]
   "(TARGET_80387 && X87_ENABLE_ARITH (DFmode))
    || (TARGET_SSE2 && TARGET_SSE_MATH)")

(define_expand "divsf3"
  [(set (match_operand:SF 0 "register_operand" "")
	(div:SF (match_operand:SF 1 "register_operand" "")
		(match_operand:SF 2 "nonimmediate_operand" "")))]
  "(TARGET_80387 && X87_ENABLE_ARITH (SFmode))
    || TARGET_SSE_MATH"
{
  if (TARGET_SSE_MATH && TARGET_RECIP && optimize_insn_for_speed_p ()
      && flag_finite_math_only && !flag_trapping_math
      && flag_unsafe_math_optimizations)
    {
      ix86_emit_swdivsf (operands[0], operands[1],
			 operands[2], SFmode);
      DONE;
    }
})

;; Divmod instructions.

(define_expand "divmod<mode>4"
  [(parallel [(set (match_operand:SWIM248 0 "register_operand" "")
		   (div:SWIM248
		     (match_operand:SWIM248 1 "register_operand" "")
		     (match_operand:SWIM248 2 "nonimmediate_operand" "")))
	      (set (match_operand:SWIM248 3 "register_operand" "")
		   (mod:SWIM248 (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])])

;; Split with 8bit unsigned divide:
;; 	if (dividend an divisor are in [0-255])
;;	   use 8bit unsigned integer divide
;;	 else
;;	   use original integer divide
(define_split
  [(set (match_operand:SWI48 0 "register_operand" "")
	(div:SWI48 (match_operand:SWI48 2 "register_operand" "")
		    (match_operand:SWI48 3 "nonimmediate_operand" "")))
   (set (match_operand:SWI48 1 "register_operand" "")
	(mod:SWI48 (match_dup 2) (match_dup 3)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_USE_8BIT_IDIV
   && TARGET_QIMODE_MATH
   && can_create_pseudo_p ()
   && !optimize_insn_for_size_p ()"
  [(const_int 0)]
  "ix86_split_idivmod (<MODE>mode, operands, true); DONE;")

(define_insn_and_split "divmod<mode>4_1"
  [(set (match_operand:SWI48 0 "register_operand" "=a")
	(div:SWI48 (match_operand:SWI48 2 "register_operand" "0")
		   (match_operand:SWI48 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWI48 1 "register_operand" "=&d")
	(mod:SWI48 (match_dup 2) (match_dup 3)))
   (unspec [(const_int 0)] UNSPEC_DIV_ALREADY_SPLIT)
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  "reload_completed"
  [(parallel [(set (match_dup 1)
		   (ashiftrt:SWI48 (match_dup 4) (match_dup 5)))
	      (clobber (reg:CC FLAGS_REG))])
   (parallel [(set (match_dup 0)
	           (div:SWI48 (match_dup 2) (match_dup 3)))
	      (set (match_dup 1)
		   (mod:SWI48 (match_dup 2) (match_dup 3)))
	      (use (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
{
  operands[5] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode)-1);

  if (optimize_function_for_size_p (cfun) || TARGET_USE_CLTD)
    operands[4] = operands[2];
  else
    {
      /* Avoid use of cltd in favor of a mov+shift.  */
      emit_move_insn (operands[1], operands[2]);
      operands[4] = operands[1];
    }
}
  [(set_attr "type" "multi")
   (set_attr "mode" "<MODE>")])

(define_insn_and_split "*divmod<mode>4"
  [(set (match_operand:SWIM248 0 "register_operand" "=a")
	(div:SWIM248 (match_operand:SWIM248 2 "register_operand" "0")
		    (match_operand:SWIM248 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWIM248 1 "register_operand" "=&d")
	(mod:SWIM248 (match_dup 2) (match_dup 3)))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  "reload_completed"
  [(parallel [(set (match_dup 1)
		   (ashiftrt:SWIM248 (match_dup 4) (match_dup 5)))
	      (clobber (reg:CC FLAGS_REG))])
   (parallel [(set (match_dup 0)
	           (div:SWIM248 (match_dup 2) (match_dup 3)))
	      (set (match_dup 1)
		   (mod:SWIM248 (match_dup 2) (match_dup 3)))
	      (use (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
{
  operands[5] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode)-1);

  if (<MODE>mode != HImode
      && (optimize_function_for_size_p (cfun) || TARGET_USE_CLTD))
    operands[4] = operands[2];
  else
    {
      /* Avoid use of cltd in favor of a mov+shift.  */
      emit_move_insn (operands[1], operands[2]);
      operands[4] = operands[1];
    }
}
  [(set_attr "type" "multi")
   (set_attr "mode" "<MODE>")])

(define_insn "*divmod<mode>4_noext"
  [(set (match_operand:SWIM248 0 "register_operand" "=a")
	(div:SWIM248 (match_operand:SWIM248 2 "register_operand" "0")
		    (match_operand:SWIM248 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWIM248 1 "register_operand" "=d")
	(mod:SWIM248 (match_dup 2) (match_dup 3)))
   (use (match_operand:SWIM248 4 "register_operand" "1"))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "idiv{<imodesuffix>}\t%3"
  [(set_attr "type" "idiv")
   (set_attr "mode" "<MODE>")])

(define_expand "divmodqi4"
  [(parallel [(set (match_operand:QI 0 "register_operand" "")
		   (div:QI
		     (match_operand:QI 1 "register_operand" "")
		     (match_operand:QI 2 "nonimmediate_operand" "")))
	      (set (match_operand:QI 3 "register_operand" "")
		   (mod:QI (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_QIMODE_MATH"
{
  rtx div, mod, insn;
  rtx tmp0, tmp1;
  
  tmp0 = gen_reg_rtx (HImode);
  tmp1 = gen_reg_rtx (HImode);

  /* Extend operands[1] to HImode.  Generate 8bit divide.  Result is
     in AX.  */
  emit_insn (gen_extendqihi2 (tmp1, operands[1]));
  emit_insn (gen_divmodhiqi3 (tmp0, tmp1, operands[2]));

  /* Extract remainder from AH.  */
  tmp1 = gen_rtx_SIGN_EXTRACT (QImode, tmp0, GEN_INT (8), GEN_INT (8));
  insn = emit_move_insn (operands[3], tmp1);

  mod = gen_rtx_MOD (QImode, operands[1], operands[2]);
  set_unique_reg_note (insn, REG_EQUAL, mod);

  /* Extract quotient from AL.  */
  insn = emit_move_insn (operands[0], gen_lowpart (QImode, tmp0));

  div = gen_rtx_DIV (QImode, operands[1], operands[2]);
  set_unique_reg_note (insn, REG_EQUAL, div);

  DONE;
})

;; Divide AX by r/m8, with result stored in
;; AL <- Quotient
;; AH <- Remainder
;; Change div/mod to HImode and extend the second argument to HImode
;; so that mode of div/mod matches with mode of arguments.  Otherwise
;; combine may fail.
(define_insn "divmodhiqi3"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(ior:HI
	  (ashift:HI
	    (zero_extend:HI
	      (truncate:QI
		(mod:HI (match_operand:HI 1 "register_operand" "0")
			(sign_extend:HI
			  (match_operand:QI 2 "nonimmediate_operand" "qm")))))
	    (const_int 8))
	  (zero_extend:HI
	    (truncate:QI
	      (div:HI (match_dup 1) (sign_extend:HI (match_dup 2)))))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_QIMODE_MATH"
  "idiv{b}\t%2"
  [(set_attr "type" "idiv")
   (set_attr "mode" "QI")])

(define_expand "udivmod<mode>4"
  [(parallel [(set (match_operand:SWIM248 0 "register_operand" "")
		   (udiv:SWIM248
		     (match_operand:SWIM248 1 "register_operand" "")
		     (match_operand:SWIM248 2 "nonimmediate_operand" "")))
	      (set (match_operand:SWIM248 3 "register_operand" "")
		   (umod:SWIM248 (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])])

;; Split with 8bit unsigned divide:
;; 	if (dividend an divisor are in [0-255])
;;	   use 8bit unsigned integer divide
;;	 else
;;	   use original integer divide
(define_split
  [(set (match_operand:SWI48 0 "register_operand" "")
	(udiv:SWI48 (match_operand:SWI48 2 "register_operand" "")
		    (match_operand:SWI48 3 "nonimmediate_operand" "")))
   (set (match_operand:SWI48 1 "register_operand" "")
	(umod:SWI48 (match_dup 2) (match_dup 3)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_USE_8BIT_IDIV
   && TARGET_QIMODE_MATH
   && can_create_pseudo_p ()
   && !optimize_insn_for_size_p ()"
  [(const_int 0)]
  "ix86_split_idivmod (<MODE>mode, operands, false); DONE;")

(define_insn_and_split "udivmod<mode>4_1"
  [(set (match_operand:SWI48 0 "register_operand" "=a")
	(udiv:SWI48 (match_operand:SWI48 2 "register_operand" "0")
		    (match_operand:SWI48 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWI48 1 "register_operand" "=&d")
	(umod:SWI48 (match_dup 2) (match_dup 3)))
   (unspec [(const_int 0)] UNSPEC_DIV_ALREADY_SPLIT)
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  "reload_completed"
  [(set (match_dup 1) (const_int 0))
   (parallel [(set (match_dup 0)
		   (udiv:SWI48 (match_dup 2) (match_dup 3)))
	      (set (match_dup 1)
		   (umod:SWI48 (match_dup 2) (match_dup 3)))
	      (use (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
  ""
  [(set_attr "type" "multi")
   (set_attr "mode" "<MODE>")])

(define_insn_and_split "*udivmod<mode>4"
  [(set (match_operand:SWIM248 0 "register_operand" "=a")
	(udiv:SWIM248 (match_operand:SWIM248 2 "register_operand" "0")
		      (match_operand:SWIM248 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWIM248 1 "register_operand" "=&d")
	(umod:SWIM248 (match_dup 2) (match_dup 3)))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  "reload_completed"
  [(set (match_dup 1) (const_int 0))
   (parallel [(set (match_dup 0)
		   (udiv:SWIM248 (match_dup 2) (match_dup 3)))
	      (set (match_dup 1)
		   (umod:SWIM248 (match_dup 2) (match_dup 3)))
	      (use (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
  ""
  [(set_attr "type" "multi")
   (set_attr "mode" "<MODE>")])

(define_insn "*udivmod<mode>4_noext"
  [(set (match_operand:SWIM248 0 "register_operand" "=a")
	(udiv:SWIM248 (match_operand:SWIM248 2 "register_operand" "0")
		      (match_operand:SWIM248 3 "nonimmediate_operand" "rm")))
   (set (match_operand:SWIM248 1 "register_operand" "=d")
	(umod:SWIM248 (match_dup 2) (match_dup 3)))
   (use (match_operand:SWIM248 4 "register_operand" "1"))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "div{<imodesuffix>}\t%3"
  [(set_attr "type" "idiv")
   (set_attr "mode" "<MODE>")])

(define_expand "udivmodqi4"
  [(parallel [(set (match_operand:QI 0 "register_operand" "")
		   (udiv:QI
		     (match_operand:QI 1 "register_operand" "")
		     (match_operand:QI 2 "nonimmediate_operand" "")))
	      (set (match_operand:QI 3 "register_operand" "")
		   (umod:QI (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_QIMODE_MATH"
{
  rtx div, mod, insn;
  rtx tmp0, tmp1;
  
  tmp0 = gen_reg_rtx (HImode);
  tmp1 = gen_reg_rtx (HImode);

  /* Extend operands[1] to HImode.  Generate 8bit divide.  Result is
     in AX.  */
  emit_insn (gen_zero_extendqihi2 (tmp1, operands[1]));
  emit_insn (gen_udivmodhiqi3 (tmp0, tmp1, operands[2]));

  /* Extract remainder from AH.  */
  tmp1 = gen_rtx_ZERO_EXTRACT (SImode, tmp0, GEN_INT (8), GEN_INT (8));
  tmp1 = simplify_gen_subreg (QImode, tmp1, SImode, 0);
  insn = emit_move_insn (operands[3], tmp1);

  mod = gen_rtx_UMOD (QImode, operands[1], operands[2]);
  set_unique_reg_note (insn, REG_EQUAL, mod);

  /* Extract quotient from AL.  */
  insn = emit_move_insn (operands[0], gen_lowpart (QImode, tmp0));

  div = gen_rtx_UDIV (QImode, operands[1], operands[2]);
  set_unique_reg_note (insn, REG_EQUAL, div);

  DONE;
})

(define_insn "udivmodhiqi3"
  [(set (match_operand:HI 0 "register_operand" "=a")
	(ior:HI
	  (ashift:HI
	    (zero_extend:HI
	      (truncate:QI
		(mod:HI (match_operand:HI 1 "register_operand" "0")
			(zero_extend:HI
			  (match_operand:QI 2 "nonimmediate_operand" "qm")))))
	    (const_int 8))
	  (zero_extend:HI
	    (truncate:QI
	      (div:HI (match_dup 1) (zero_extend:HI (match_dup 2)))))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_QIMODE_MATH"
  "div{b}\t%2"
  [(set_attr "type" "idiv")
   (set_attr "mode" "QI")])

;; We cannot use div/idiv for double division, because it causes
;; "division by zero" on the overflow and that's not what we expect
;; from truncate.  Because true (non truncating) double division is
;; never generated, we can't create this insn anyway.
;
;(define_insn ""
;  [(set (match_operand:SI 0 "register_operand" "=a")
;	(truncate:SI
;	  (udiv:DI (match_operand:DI 1 "register_operand" "A")
;		   (zero_extend:DI
;		     (match_operand:SI 2 "nonimmediate_operand" "rm")))))
;   (set (match_operand:SI 3 "register_operand" "=d")
;	(truncate:SI
;	  (umod:DI (match_dup 1) (zero_extend:DI (match_dup 2)))))
;   (clobber (reg:CC FLAGS_REG))]
;  ""
;  "div{l}\t{%2, %0|%0, %2}"
;  [(set_attr "type" "idiv")])

;;- Logical AND instructions

;; On Pentium, "test imm, reg" is pairable only with eax, ax, and al.
;; Note that this excludes ah.

(define_expand "testsi_ccno_1"
  [(set (reg:CCNO FLAGS_REG)
	(compare:CCNO
	  (and:SI (match_operand:SI 0 "nonimmediate_operand" "")
		  (match_operand:SI 1 "nonmemory_operand" ""))
	  (const_int 0)))])

(define_expand "testqi_ccz_1"
  [(set (reg:CCZ FLAGS_REG)
        (compare:CCZ (and:QI (match_operand:QI 0 "nonimmediate_operand" "")
			     (match_operand:QI 1 "nonmemory_operand" ""))
		 (const_int 0)))])

(define_expand "testdi_ccno_1"
  [(set (reg:CCNO FLAGS_REG)
	(compare:CCNO
	  (and:DI (match_operand:DI 0 "nonimmediate_operand" "")
		  (match_operand:DI 1 "x86_64_szext_general_operand" ""))
	  (const_int 0)))]
  "TARGET_64BIT && !(MEM_P (operands[0]) && MEM_P (operands[1]))")

(define_insn "*testdi_1"
  [(set (reg FLAGS_REG)
	(compare
	 (and:DI
	  (match_operand:DI 0 "nonimmediate_operand" "%!*a,r,!*a,r,rm")
	  (match_operand:DI 1 "x86_64_szext_general_operand" "Z,Z,e,e,re"))
	 (const_int 0)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "@
   test{l}\t{%k1, %k0|%k0, %k1}
   test{l}\t{%k1, %k0|%k0, %k1}
   test{q}\t{%1, %0|%0, %1}
   test{q}\t{%1, %0|%0, %1}
   test{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "test")
   (set_attr "modrm" "0,1,0,1,1")
   (set_attr "mode" "SI,SI,DI,DI,DI")])

(define_insn "*testqi_1_maybe_si"
  [(set (reg FLAGS_REG)
        (compare
	  (and:QI
	    (match_operand:QI 0 "nonimmediate_operand" "%!*a,q,qm,r")
	    (match_operand:QI 1 "general_operand" "n,n,qn,n"))
	  (const_int 0)))]
   "!(MEM_P (operands[0]) && MEM_P (operands[1]))
    && ix86_match_ccmode (insn,
 			 CONST_INT_P (operands[1])
 			 && INTVAL (operands[1]) >= 0 ? CCNOmode : CCZmode)"
{
  if (which_alternative == 3)
    {
      if (CONST_INT_P (operands[1]) && INTVAL (operands[1]) < 0)
	operands[1] = GEN_INT (INTVAL (operands[1]) & 0xff);
      return "test{l}\t{%1, %k0|%k0, %1}";
    }
  return "test{b}\t{%1, %0|%0, %1}";
}
  [(set_attr "type" "test")
   (set_attr "modrm" "0,1,1,1")
   (set_attr "mode" "QI,QI,QI,SI")
   (set_attr "pent_pair" "uv,np,uv,np")])

(define_insn "*test<mode>_1"
  [(set (reg FLAGS_REG)
	(compare
	 (and:SWI124
	  (match_operand:SWI124 0 "nonimmediate_operand" "%!*a,<r>,<r>m")
	  (match_operand:SWI124 1 "general_operand" "<i>,<i>,<r><i>"))
	 (const_int 0)))]
  "ix86_match_ccmode (insn, CCNOmode)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "test{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "test")
   (set_attr "modrm" "0,1,1")
   (set_attr "mode" "<MODE>")
   (set_attr "pent_pair" "uv,np,uv")])

(define_expand "testqi_ext_ccno_0"
  [(set (reg:CCNO FLAGS_REG)
	(compare:CCNO
	  (and:SI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "")
	      (const_int 8)
	      (const_int 8))
	    (match_operand 1 "const_int_operand" ""))
	  (const_int 0)))])

(define_insn "*testqi_ext_0"
  [(set (reg FLAGS_REG)
	(compare
	  (and:SI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8))
	    (match_operand 1 "const_int_operand" "n"))
	  (const_int 0)))]
  "ix86_match_ccmode (insn, CCNOmode)"
  "test{b}\t{%1, %h0|%h0, %1}"
  [(set_attr "type" "test")
   (set_attr "mode" "QI")
   (set_attr "length_immediate" "1")
   (set_attr "modrm" "1")
   (set_attr "pent_pair" "np")])

(define_insn "*testqi_ext_1_rex64"
  [(set (reg FLAGS_REG)
	(compare
	  (and:SI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8))
	    (zero_extend:SI
	      (match_operand:QI 1 "register_operand" "Q")))
	  (const_int 0)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)"
  "test{b}\t{%1, %h0|%h0, %1}"
  [(set_attr "type" "test")
   (set_attr "mode" "QI")])

(define_insn "*testqi_ext_1"
  [(set (reg FLAGS_REG)
	(compare
	  (and:SI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8))
	    (zero_extend:SI
	      (match_operand:QI 1 "general_operand" "Qm")))
	  (const_int 0)))]
  "!TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)"
  "test{b}\t{%1, %h0|%h0, %1}"
  [(set_attr "type" "test")
   (set_attr "mode" "QI")])

(define_insn "*testqi_ext_2"
  [(set (reg FLAGS_REG)
	(compare
	  (and:SI
	    (zero_extract:SI
	      (match_operand 0 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8))
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "Q")
	      (const_int 8)
	      (const_int 8)))
	  (const_int 0)))]
  "ix86_match_ccmode (insn, CCNOmode)"
  "test{b}\t{%h1, %h0|%h0, %h1}"
  [(set_attr "type" "test")
   (set_attr "mode" "QI")])

(define_insn "*testqi_ext_3_rex64"
  [(set (reg FLAGS_REG)
        (compare (zero_extract:DI
		   (match_operand 0 "nonimmediate_operand" "rm")
		   (match_operand:DI 1 "const_int_operand" "")
		   (match_operand:DI 2 "const_int_operand" ""))
		 (const_int 0)))]
  "TARGET_64BIT
   && ix86_match_ccmode (insn, CCNOmode)
   && INTVAL (operands[1]) > 0
   && INTVAL (operands[2]) >= 0
   /* Ensure that resulting mask is zero or sign extended operand.  */
   && (INTVAL (operands[1]) + INTVAL (operands[2]) <= 32
       || (INTVAL (operands[1]) + INTVAL (operands[2]) == 64
	   && INTVAL (operands[1]) > 32))
   && (GET_MODE (operands[0]) == SImode
       || GET_MODE (operands[0]) == DImode
       || GET_MODE (operands[0]) == HImode
       || GET_MODE (operands[0]) == QImode)"
  "#")

;; Combine likes to form bit extractions for some tests.  Humor it.
(define_insn "*testqi_ext_3"
  [(set (reg FLAGS_REG)
        (compare (zero_extract:SI
		   (match_operand 0 "nonimmediate_operand" "rm")
		   (match_operand:SI 1 "const_int_operand" "")
		   (match_operand:SI 2 "const_int_operand" ""))
		 (const_int 0)))]
  "ix86_match_ccmode (insn, CCNOmode)
   && INTVAL (operands[1]) > 0
   && INTVAL (operands[2]) >= 0
   && INTVAL (operands[1]) + INTVAL (operands[2]) <= 32
   && (GET_MODE (operands[0]) == SImode
       || (TARGET_64BIT && GET_MODE (operands[0]) == DImode)
       || GET_MODE (operands[0]) == HImode
       || GET_MODE (operands[0]) == QImode)"
  "#")

(define_split
  [(set (match_operand 0 "flags_reg_operand" "")
        (match_operator 1 "compare_operator"
	  [(zero_extract
	     (match_operand 2 "nonimmediate_operand" "")
	     (match_operand 3 "const_int_operand" "")
	     (match_operand 4 "const_int_operand" ""))
	   (const_int 0)]))]
  "ix86_match_ccmode (insn, CCNOmode)"
  [(set (match_dup 0) (match_op_dup 1 [(match_dup 2) (const_int 0)]))]
{
  rtx val = operands[2];
  HOST_WIDE_INT len = INTVAL (operands[3]);
  HOST_WIDE_INT pos = INTVAL (operands[4]);
  HOST_WIDE_INT mask;
  enum machine_mode mode, submode;

  mode = GET_MODE (val);
  if (MEM_P (val))
    {
      /* ??? Combine likes to put non-volatile mem extractions in QImode
	 no matter the size of the test.  So find a mode that works.  */
      if (! MEM_VOLATILE_P (val))
	{
	  mode = smallest_mode_for_size (pos + len, MODE_INT);
	  val = adjust_address (val, mode, 0);
	}
    }
  else if (GET_CODE (val) == SUBREG
	   && (submode = GET_MODE (SUBREG_REG (val)),
	       GET_MODE_BITSIZE (mode) > GET_MODE_BITSIZE (submode))
	   && pos + len <= GET_MODE_BITSIZE (submode)
	   && GET_MODE_CLASS (submode) == MODE_INT)
    {
      /* Narrow a paradoxical subreg to prevent partial register stalls.  */
      mode = submode;
      val = SUBREG_REG (val);
    }
  else if (mode == HImode && pos + len <= 8)
    {
      /* Small HImode tests can be converted to QImode.  */
      mode = QImode;
      val = gen_lowpart (QImode, val);
    }

  if (len == HOST_BITS_PER_WIDE_INT)
    mask = -1;
  else
    mask = ((HOST_WIDE_INT)1 << len) - 1;
  mask <<= pos;

  operands[2] = gen_rtx_AND (mode, val, gen_int_mode (mask, mode));
})

;; Convert HImode/SImode test instructions with immediate to QImode ones.
;; i386 does not allow to encode test with 8bit sign extended immediate, so
;; this is relatively important trick.
;; Do the conversion only post-reload to avoid limiting of the register class
;; to QI regs.
(define_split
  [(set (match_operand 0 "flags_reg_operand" "")
	(match_operator 1 "compare_operator"
	  [(and (match_operand 2 "register_operand" "")
	        (match_operand 3 "const_int_operand" ""))
	   (const_int 0)]))]
   "reload_completed
    && QI_REG_P (operands[2])
    && GET_MODE (operands[2]) != QImode
    && ((ix86_match_ccmode (insn, CCZmode)
    	 && !(INTVAL (operands[3]) & ~(255 << 8)))
	|| (ix86_match_ccmode (insn, CCNOmode)
	    && !(INTVAL (operands[3]) & ~(127 << 8))))"
  [(set (match_dup 0)
	(match_op_dup 1
	  [(and:SI (zero_extract:SI (match_dup 2) (const_int 8) (const_int 8))
		   (match_dup 3))
	   (const_int 0)]))]
  "operands[2] = gen_lowpart (SImode, operands[2]);
   operands[3] = gen_int_mode (INTVAL (operands[3]) >> 8, SImode);")

(define_split
  [(set (match_operand 0 "flags_reg_operand" "")
	(match_operator 1 "compare_operator"
	  [(and (match_operand 2 "nonimmediate_operand" "")
	        (match_operand 3 "const_int_operand" ""))
	   (const_int 0)]))]
   "reload_completed
    && GET_MODE (operands[2]) != QImode
    && (!REG_P (operands[2]) || ANY_QI_REG_P (operands[2]))
    && ((ix86_match_ccmode (insn, CCZmode)
	 && !(INTVAL (operands[3]) & ~255))
	|| (ix86_match_ccmode (insn, CCNOmode)
	    && !(INTVAL (operands[3]) & ~127)))"
  [(set (match_dup 0)
	(match_op_dup 1 [(and:QI (match_dup 2) (match_dup 3))
			 (const_int 0)]))]
  "operands[2] = gen_lowpart (QImode, operands[2]);
   operands[3] = gen_lowpart (QImode, operands[3]);")

;; %%% This used to optimize known byte-wide and operations to memory,
;; and sometimes to QImode registers.  If this is considered useful,
;; it should be done with splitters.

(define_expand "and<mode>3"
  [(set (match_operand:SWIM 0 "nonimmediate_operand" "")
	(and:SWIM (match_operand:SWIM 1 "nonimmediate_operand" "")
		  (match_operand:SWIM 2 "<general_szext_operand>" "")))]
  ""
  "ix86_expand_binary_operator (AND, <MODE>mode, operands); DONE;")

(define_insn "*anddi_1"
  [(set (match_operand:DI 0 "nonimmediate_operand" "=r,rm,r,r")
	(and:DI
	 (match_operand:DI 1 "nonimmediate_operand" "%0,0,0,qm")
	 (match_operand:DI 2 "x86_64_szext_general_operand" "Z,re,rm,L")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (AND, DImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      {
	enum machine_mode mode;

	gcc_assert (CONST_INT_P (operands[2]));
        if (INTVAL (operands[2]) == 0xff)
	  mode = QImode;
	else
	  {
	    gcc_assert (INTVAL (operands[2]) == 0xffff);
	    mode = HImode;
	  }

	operands[1] = gen_lowpart (mode, operands[1]);
	if (mode == QImode)
	  return "movz{bl|x}\t{%1, %k0|%k0, %1}";
	else
	  return "movz{wl|x}\t{%1, %k0|%k0, %1}";
      }

    default:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      if (get_attr_mode (insn) == MODE_SI)
	return "and{l}\t{%k2, %k0|%k0, %k2}";
      else
	return "and{q}\t{%2, %0|%0, %2}";
    }
}
  [(set_attr "type" "alu,alu,alu,imovx")
   (set_attr "length_immediate" "*,*,*,0")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "type" "imovx")
	    (and (ne (symbol_ref "INTVAL (operands[2]) == 0xff") (const_int 0))
		 (match_operand 1 "ext_QIreg_nomode_operand" "")))
       (const_string "1")
       (const_string "*")))
   (set_attr "mode" "SI,DI,DI,SI")])

(define_insn "*andsi_1"
  [(set (match_operand:SI 0 "nonimmediate_operand" "=rm,r,r")
	(and:SI (match_operand:SI 1 "nonimmediate_operand" "%0,0,qm")
		(match_operand:SI 2 "general_operand" "ri,rm,L")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (AND, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      {
	enum machine_mode mode;

	gcc_assert (CONST_INT_P (operands[2]));
        if (INTVAL (operands[2]) == 0xff)
	  mode = QImode;
	else
	  {
	    gcc_assert (INTVAL (operands[2]) == 0xffff);
	    mode = HImode;
	  }

	operands[1] = gen_lowpart (mode, operands[1]);
	if (mode == QImode)
	  return "movz{bl|x}\t{%1, %0|%0, %1}";
	else
	  return "movz{wl|x}\t{%1, %0|%0, %1}";
      }

    default:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      return "and{l}\t{%2, %0|%0, %2}";
    }
}
  [(set_attr "type" "alu,alu,imovx")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "type" "imovx")
	    (and (ne (symbol_ref "INTVAL (operands[2]) == 0xff") (const_int 0))
		 (match_operand 1 "ext_QIreg_nomode_operand" "")))
       (const_string "1")
       (const_string "*")))
   (set_attr "length_immediate" "*,*,0")
   (set_attr "mode" "SI")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
(define_insn "*andsi_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (and:SI (match_operand:SI 1 "nonimmediate_operand" "%0")
		  (match_operand:SI 2 "general_operand" "g"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (AND, SImode, operands)"
  "and{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*andhi_1"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=rm,r,r")
	(and:HI (match_operand:HI 1 "nonimmediate_operand" "%0,0,qm")
		(match_operand:HI 2 "general_operand" "rn,rm,L")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (AND, HImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_IMOVX:
      gcc_assert (CONST_INT_P (operands[2]));
      gcc_assert (INTVAL (operands[2]) == 0xff);
      return "movz{bl|x}\t{%b1, %k0|%k0, %b1}";

    default:
      gcc_assert (rtx_equal_p (operands[0], operands[1]));

      return "and{w}\t{%2, %0|%0, %2}";
    }
}
  [(set_attr "type" "alu,alu,imovx")
   (set_attr "length_immediate" "*,*,0")
   (set (attr "prefix_rex")
     (if_then_else
       (and (eq_attr "type" "imovx")
	    (match_operand 1 "ext_QIreg_nomode_operand" ""))
       (const_string "1")
       (const_string "*")))
   (set_attr "mode" "HI,HI,SI")])

;; %%% Potential partial reg stall on alternative 2.  What to do?
(define_insn "*andqi_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm,q,r")
	(and:QI (match_operand:QI 1 "nonimmediate_operand" "%0,0,0")
		(match_operand:QI 2 "general_operand" "qn,qmn,rn")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (AND, QImode, operands)"
  "@
   and{b}\t{%2, %0|%0, %2}
   and{b}\t{%2, %0|%0, %2}
   and{l}\t{%k2, %k0|%k0, %k2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "QI,QI,SI")])

(define_insn "*andqi_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm,q"))
	(and:QI (match_dup 0)
		(match_operand:QI 1 "general_operand" "qn,qmn")))
   (clobber (reg:CC FLAGS_REG))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "and{b}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "QI")])

(define_split
  [(set (match_operand 0 "register_operand" "")
	(and (match_dup 0)
	     (const_int -65536)))
   (clobber (reg:CC FLAGS_REG))]
  "(TARGET_FAST_PREFIX && !TARGET_PARTIAL_REG_STALL)
    || optimize_function_for_size_p (cfun)"
  [(set (strict_low_part (match_dup 1)) (const_int 0))]
  "operands[1] = gen_lowpart (HImode, operands[0]);")

(define_split
  [(set (match_operand 0 "ext_register_operand" "")
	(and (match_dup 0)
	     (const_int -256)))
   (clobber (reg:CC FLAGS_REG))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && reload_completed"
  [(set (strict_low_part (match_dup 1)) (const_int 0))]
  "operands[1] = gen_lowpart (QImode, operands[0]);")

(define_split
  [(set (match_operand 0 "ext_register_operand" "")
	(and (match_dup 0)
	     (const_int -65281)))
   (clobber (reg:CC FLAGS_REG))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && reload_completed"
  [(parallel [(set (zero_extract:SI (match_dup 0)
				    (const_int 8)
				    (const_int 8))
		   (xor:SI
		     (zero_extract:SI (match_dup 0)
				      (const_int 8)
				      (const_int 8))
		     (zero_extract:SI (match_dup 0)
				      (const_int 8)
				      (const_int 8))))
	      (clobber (reg:CC FLAGS_REG))])]
  "operands[0] = gen_lowpart (SImode, operands[0]);")

(define_insn "*anddi_2"
  [(set (reg FLAGS_REG)
	(compare
	 (and:DI
	  (match_operand:DI 1 "nonimmediate_operand" "%0,0,0")
	  (match_operand:DI 2 "x86_64_szext_general_operand" "Z,rem,re"))
	 (const_int 0)))
   (set (match_operand:DI 0 "nonimmediate_operand" "=r,r,rm")
	(and:DI (match_dup 1) (match_dup 2)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (AND, DImode, operands)"
  "@
   and{l}\t{%k2, %k0|%k0, %k2}
   and{q}\t{%2, %0|%0, %2}
   and{q}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI,DI,DI")])

(define_insn "*andqi_2_maybe_si"
  [(set (reg FLAGS_REG)
	(compare (and:QI
		  (match_operand:QI 1 "nonimmediate_operand" "%0,0,0")
		  (match_operand:QI 2 "general_operand" "qmn,qn,n"))
		 (const_int 0)))
   (set (match_operand:QI 0 "nonimmediate_operand" "=q,qm,*r")
	(and:QI (match_dup 1) (match_dup 2)))]
  "ix86_binary_operator_ok (AND, QImode, operands)
   && ix86_match_ccmode (insn,
			 CONST_INT_P (operands[2])
			 && INTVAL (operands[2]) >= 0 ? CCNOmode : CCZmode)"
{
  if (which_alternative == 2)
    {
      if (CONST_INT_P (operands[2]) && INTVAL (operands[2]) < 0)
        operands[2] = GEN_INT (INTVAL (operands[2]) & 0xff);
      return "and{l}\t{%2, %k0|%k0, %2}";
    }
  return "and{b}\t{%2, %0|%0, %2}";
}
  [(set_attr "type" "alu")
   (set_attr "mode" "QI,QI,SI")])

(define_insn "*and<mode>_2"
  [(set (reg FLAGS_REG)
	(compare (and:SWI124
		  (match_operand:SWI124 1 "nonimmediate_operand" "%0,0")
		  (match_operand:SWI124 2 "general_operand" "<g>,<r><i>"))
		 (const_int 0)))
   (set (match_operand:SWI124 0 "nonimmediate_operand" "=<r>,<r>m")
	(and:SWI124 (match_dup 1) (match_dup 2)))]
  "ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (AND, <MODE>mode, operands)"
  "and{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
(define_insn "*andsi_2_zext"
  [(set (reg FLAGS_REG)
	(compare (and:SI
		  (match_operand:SI 1 "nonimmediate_operand" "%0")
		  (match_operand:SI 2 "general_operand" "g"))
		 (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (and:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (AND, SImode, operands)"
  "and{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*andqi_2_slp"
  [(set (reg FLAGS_REG)
	(compare (and:QI
		   (match_operand:QI 0 "nonimmediate_operand" "+q,qm")
		   (match_operand:QI 1 "nonimmediate_operand" "qmn,qn"))
		 (const_int 0)))
   (set (strict_low_part (match_dup 0))
	(and:QI (match_dup 0) (match_dup 1)))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && ix86_match_ccmode (insn, CCNOmode)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "and{b}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "QI")])

;; ??? A bug in recog prevents it from recognizing a const_int as an
;; operand to zero_extend in andqi_ext_1.  It was checking explicitly
;; for a QImode operand, which of course failed.
(define_insn "andqi_ext_0"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(and:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (match_operand 2 "const_int_operand" "n")))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "and{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "1")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

;; Generated by peephole translating test to and.  This shows up
;; often in fp comparisons.
(define_insn "*andqi_ext_0_cc"
  [(set (reg FLAGS_REG)
	(compare
	  (and:SI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "0")
	      (const_int 8)
	      (const_int 8))
	    (match_operand 2 "const_int_operand" "n"))
	  (const_int 0)))
   (set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(and:SI
	  (zero_extract:SI
	    (match_dup 1)
	    (const_int 8)
	    (const_int 8))
	  (match_dup 2)))]
  "ix86_match_ccmode (insn, CCNOmode)"
  "and{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "1")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*andqi_ext_1_rex64"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(and:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extend:SI
	    (match_operand 2 "ext_register_operand" "Q"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT"
  "and{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_insn "*andqi_ext_1"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(and:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extend:SI
	    (match_operand:QI 2 "general_operand" "Qm"))))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT"
  "and{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_insn "*andqi_ext_2"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(and:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "%0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extract:SI
	    (match_operand 2 "ext_register_operand" "Q")
	    (const_int 8)
	    (const_int 8))))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "and{b}\t{%h2, %h0|%h0, %h2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

;; Convert wide AND instructions with immediate operand to shorter QImode
;; equivalents when possible.
;; Don't do the splitting with memory operands, since it introduces risk
;; of memory mismatch stalls.  We may want to do the splitting for optimizing
;; for size, but that can (should?) be handled by generic code instead.
(define_split
  [(set (match_operand 0 "register_operand" "")
	(and (match_operand 1 "register_operand" "")
	     (match_operand 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
   "reload_completed
    && QI_REG_P (operands[0])
    && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
    && !(~INTVAL (operands[2]) & ~(255 << 8))
    && GET_MODE (operands[0]) != QImode"
  [(parallel [(set (zero_extract:SI (match_dup 0) (const_int 8) (const_int 8))
		   (and:SI (zero_extract:SI (match_dup 1)
					    (const_int 8) (const_int 8))
			   (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "operands[0] = gen_lowpart (SImode, operands[0]);
   operands[1] = gen_lowpart (SImode, operands[1]);
   operands[2] = gen_int_mode ((INTVAL (operands[2]) >> 8) & 0xff, SImode);")

;; Since AND can be encoded with sign extended immediate, this is only
;; profitable when 7th bit is not set.
(define_split
  [(set (match_operand 0 "register_operand" "")
	(and (match_operand 1 "general_operand" "")
	     (match_operand 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
   "reload_completed
    && ANY_QI_REG_P (operands[0])
    && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
    && !(~INTVAL (operands[2]) & ~255)
    && !(INTVAL (operands[2]) & 128)
    && GET_MODE (operands[0]) != QImode"
  [(parallel [(set (strict_low_part (match_dup 0))
		   (and:QI (match_dup 1)
			   (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "operands[0] = gen_lowpart (QImode, operands[0]);
   operands[1] = gen_lowpart (QImode, operands[1]);
   operands[2] = gen_lowpart (QImode, operands[2]);")

;; Logical inclusive and exclusive OR instructions

;; %%% This used to optimize known byte-wide and operations to memory.
;; If this is considered useful, it should be done with splitters.

(define_expand "<code><mode>3"
  [(set (match_operand:SWIM 0 "nonimmediate_operand" "")
	(any_or:SWIM (match_operand:SWIM 1 "nonimmediate_operand" "")
		     (match_operand:SWIM 2 "<general_operand>" "")))]
  ""
  "ix86_expand_binary_operator (<CODE>, <MODE>mode, operands); DONE;")

(define_insn "*<code><mode>_1"
  [(set (match_operand:SWI248 0 "nonimmediate_operand" "=r,rm")
	(any_or:SWI248
	 (match_operand:SWI248 1 "nonimmediate_operand" "%0,0")
	 (match_operand:SWI248 2 "<general_operand>" "<g>,r<i>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
  "<logic>{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

;; %%% Potential partial reg stall on alternative 2.  What to do?
(define_insn "*<code>qi_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=q,m,r")
	(any_or:QI (match_operand:QI 1 "nonimmediate_operand" "%0,0,0")
		   (match_operand:QI 2 "general_operand" "qmn,qn,rn")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, QImode, operands)"
  "@
   <logic>{b}\t{%2, %0|%0, %2}
   <logic>{b}\t{%2, %0|%0, %2}
   <logic>{l}\t{%k2, %k0|%k0, %k2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "QI,QI,SI")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
(define_insn "*<code>si_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	 (any_or:SI (match_operand:SI 1 "nonimmediate_operand" "%0")
		    (match_operand:SI 2 "general_operand" "g"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (<CODE>, SImode, operands)"
  "<logic>{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*<code>si_1_zext_imm"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(any_or:DI
	 (zero_extend:DI (match_operand:SI 1 "register_operand" "%0"))
	 (match_operand:DI 2 "x86_64_zext_immediate_operand" "Z")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (<CODE>, SImode, operands)"
  "<logic>{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*<code>qi_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+q,m"))
	(any_or:QI (match_dup 0)
		   (match_operand:QI 1 "general_operand" "qmn,qn")))
   (clobber (reg:CC FLAGS_REG))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "<logic>{b}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "QI")])

(define_insn "*<code><mode>_2"
  [(set (reg FLAGS_REG)
	(compare (any_or:SWI
		  (match_operand:SWI 1 "nonimmediate_operand" "%0,0")
		  (match_operand:SWI 2 "<general_operand>" "<g>,<r><i>"))
		 (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>,<r>m")
	(any_or:SWI (match_dup 1) (match_dup 2)))]
  "ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
  "<logic>{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

;; See comment for addsi_1_zext why we do use nonimmediate_operand
;; ??? Special case for immediate operand is missing - it is tricky.
(define_insn "*<code>si_2_zext"
  [(set (reg FLAGS_REG)
	(compare (any_or:SI (match_operand:SI 1 "nonimmediate_operand" "%0")
			    (match_operand:SI 2 "general_operand" "g"))
		 (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (any_or:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (<CODE>, SImode, operands)"
  "<logic>{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*<code>si_2_zext_imm"
  [(set (reg FLAGS_REG)
	(compare (any_or:SI
		  (match_operand:SI 1 "nonimmediate_operand" "%0")
		  (match_operand:SI 2 "x86_64_zext_immediate_operand" "Z"))
		 (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(any_or:DI (zero_extend:DI (match_dup 1)) (match_dup 2)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && ix86_binary_operator_ok (<CODE>, SImode, operands)"
  "<logic>{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "SI")])

(define_insn "*<code>qi_2_slp"
  [(set (reg FLAGS_REG)
	(compare (any_or:QI (match_operand:QI 0 "nonimmediate_operand" "+q,qm")
			    (match_operand:QI 1 "general_operand" "qmn,qn"))
		 (const_int 0)))
   (set (strict_low_part (match_dup 0))
	(any_or:QI (match_dup 0) (match_dup 1)))]
  "(!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
   && ix86_match_ccmode (insn, CCNOmode)
   && !(MEM_P (operands[0]) && MEM_P (operands[1]))"
  "<logic>{b}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "mode" "QI")])

(define_insn "*<code><mode>_3"
  [(set (reg FLAGS_REG)
	(compare (any_or:SWI
		  (match_operand:SWI 1 "nonimmediate_operand" "%0")
		  (match_operand:SWI 2 "<general_operand>" "<g>"))
		 (const_int 0)))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "ix86_match_ccmode (insn, CCNOmode)
   && !(MEM_P (operands[1]) && MEM_P (operands[2]))"
  "<logic>{<imodesuffix>}\t{%2, %0|%0, %2}"
  [(set_attr "type" "alu")
   (set_attr "mode" "<MODE>")])

(define_insn "*<code>qi_ext_0"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(any_or:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (match_operand 2 "const_int_operand" "n")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)"
  "<logic>{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "1")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*<code>qi_ext_1_rex64"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(any_or:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extend:SI
	    (match_operand 2 "ext_register_operand" "Q"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT
   && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))"
  "<logic>{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_insn "*<code>qi_ext_1"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(any_or:SI
	  (zero_extract:SI
	    (match_operand 1 "ext_register_operand" "0")
	    (const_int 8)
	    (const_int 8))
	  (zero_extend:SI
	    (match_operand:QI 2 "general_operand" "Qm"))))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT
   && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))"
  "<logic>{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_insn "*<code>qi_ext_2"
  [(set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(any_or:SI
	  (zero_extract:SI (match_operand 1 "ext_register_operand" "0")
	  		   (const_int 8)
			   (const_int 8))
	  (zero_extract:SI (match_operand 2 "ext_register_operand" "Q")
	  		   (const_int 8)
			   (const_int 8))))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun)"
  "<logic>{b}\t{%h2, %h0|%h0, %h2}"
  [(set_attr "type" "alu")
   (set_attr "length_immediate" "0")
   (set_attr "mode" "QI")])

(define_split
  [(set (match_operand 0 "register_operand" "")
	(any_or (match_operand 1 "register_operand" "")
		(match_operand 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
   "reload_completed
    && QI_REG_P (operands[0])
    && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
    && !(INTVAL (operands[2]) & ~(255 << 8))
    && GET_MODE (operands[0]) != QImode"
  [(parallel [(set (zero_extract:SI (match_dup 0) (const_int 8) (const_int 8))
		   (any_or:SI (zero_extract:SI (match_dup 1)
					       (const_int 8) (const_int 8))
			      (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "operands[0] = gen_lowpart (SImode, operands[0]);
   operands[1] = gen_lowpart (SImode, operands[1]);
   operands[2] = gen_int_mode ((INTVAL (operands[2]) >> 8) & 0xff, SImode);")

;; Since OR can be encoded with sign extended immediate, this is only
;; profitable when 7th bit is set.
(define_split
  [(set (match_operand 0 "register_operand" "")
	(any_or (match_operand 1 "general_operand" "")
		(match_operand 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
   "reload_completed
    && ANY_QI_REG_P (operands[0])
    && (!TARGET_PARTIAL_REG_STALL || optimize_function_for_size_p (cfun))
    && !(INTVAL (operands[2]) & ~255)
    && (INTVAL (operands[2]) & 128)
    && GET_MODE (operands[0]) != QImode"
  [(parallel [(set (strict_low_part (match_dup 0))
		   (any_or:QI (match_dup 1)
			      (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
  "operands[0] = gen_lowpart (QImode, operands[0]);
   operands[1] = gen_lowpart (QImode, operands[1]);
   operands[2] = gen_lowpart (QImode, operands[2]);")

(define_expand "xorqi_cc_ext_1"
  [(parallel [
     (set (reg:CCNO FLAGS_REG)
	  (compare:CCNO
	    (xor:SI
	      (zero_extract:SI
		(match_operand 1 "ext_register_operand" "")
		(const_int 8)
		(const_int 8))
	      (match_operand:QI 2 "general_operand" ""))
	    (const_int 0)))
     (set (zero_extract:SI (match_operand 0 "ext_register_operand" "")
			   (const_int 8)
			   (const_int 8))
	  (xor:SI
	    (zero_extract:SI
	     (match_dup 1)
	     (const_int 8)
	     (const_int 8))
	    (match_dup 2)))])])

(define_insn "*xorqi_cc_ext_1_rex64"
  [(set (reg FLAGS_REG)
	(compare
	  (xor:SI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "0")
	      (const_int 8)
	      (const_int 8))
	    (match_operand:QI 2 "nonmemory_operand" "Qn"))
	  (const_int 0)))
   (set (zero_extract:SI (match_operand 0 "ext_register_operand" "=Q")
			 (const_int 8)
			 (const_int 8))
	(xor:SI
	  (zero_extract:SI
	   (match_dup 1)
	   (const_int 8)
	   (const_int 8))
	  (match_dup 2)))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)"
  "xor{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

(define_insn "*xorqi_cc_ext_1"
  [(set (reg FLAGS_REG)
	(compare
	  (xor:SI
	    (zero_extract:SI
	      (match_operand 1 "ext_register_operand" "0")
	      (const_int 8)
	      (const_int 8))
	    (match_operand:QI 2 "general_operand" "qmn"))
	  (const_int 0)))
   (set (zero_extract:SI (match_operand 0 "ext_register_operand" "=q")
			 (const_int 8)
			 (const_int 8))
	(xor:SI
	  (zero_extract:SI
	   (match_dup 1)
	   (const_int 8)
	   (const_int 8))
	  (match_dup 2)))]
  "!TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)"
  "xor{b}\t{%2, %h0|%h0, %2}"
  [(set_attr "type" "alu")
   (set_attr "modrm" "1")
   (set_attr "mode" "QI")])

;; Negation instructions

(define_expand "neg<mode>2"
  [(set (match_operand:SDWIM 0 "nonimmediate_operand" "")
	(neg:SDWIM (match_operand:SDWIM 1 "nonimmediate_operand" "")))]
  ""
  "ix86_expand_unary_operator (NEG, <MODE>mode, operands); DONE;")

(define_insn_and_split "*neg<dwi>2_doubleword"
  [(set (match_operand:<DWI> 0 "nonimmediate_operand" "=ro")
	(neg:<DWI> (match_operand:<DWI> 1 "nonimmediate_operand" "0")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_unary_operator_ok (NEG, <DWI>mode, operands)"
  "#"
  "reload_completed"
  [(parallel
    [(set (reg:CCZ FLAGS_REG)
	  (compare:CCZ (neg:DWIH (match_dup 1)) (const_int 0)))
     (set (match_dup 0) (neg:DWIH (match_dup 1)))])
   (parallel
    [(set (match_dup 2)
	  (plus:DWIH (match_dup 3)
		     (plus:DWIH (ltu:DWIH (reg:CC FLAGS_REG) (const_int 0))
				(const_int 0))))
     (clobber (reg:CC FLAGS_REG))])
   (parallel
    [(set (match_dup 2)
	  (neg:DWIH (match_dup 2)))
     (clobber (reg:CC FLAGS_REG))])]
  "split_double_mode (<DWI>mode, &operands[0], 2, &operands[0], &operands[2]);")

(define_insn "*neg<mode>2_1"
  [(set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(neg:SWI (match_operand:SWI 1 "nonimmediate_operand" "0")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_unary_operator_ok (NEG, <MODE>mode, operands)"
  "neg{<imodesuffix>}\t%0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "<MODE>")])

;; Combine is quite creative about this pattern.
(define_insn "*negsi2_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(lshiftrt:DI
	  (neg:DI (ashift:DI (match_operand:DI 1 "register_operand" "0")
			     (const_int 32)))
	(const_int 32)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_unary_operator_ok (NEG, SImode, operands)"
  "neg{l}\t%k0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "SI")])

;; The problem with neg is that it does not perform (compare x 0),
;; it really performs (compare 0 x), which leaves us with the zero
;; flag being the only useful item.

(define_insn "*neg<mode>2_cmpz"
  [(set (reg:CCZ FLAGS_REG)
	(compare:CCZ
	  (neg:SWI (match_operand:SWI 1 "nonimmediate_operand" "0"))
		   (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(neg:SWI (match_dup 1)))]
  "ix86_unary_operator_ok (NEG, <MODE>mode, operands)"
  "neg{<imodesuffix>}\t%0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "<MODE>")])

(define_insn "*negsi2_cmpz_zext"
  [(set (reg:CCZ FLAGS_REG)
	(compare:CCZ
	  (lshiftrt:DI
	    (neg:DI (ashift:DI
		      (match_operand:DI 1 "register_operand" "0")
		      (const_int 32)))
	    (const_int 32))
	  (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(lshiftrt:DI (neg:DI (ashift:DI (match_dup 1)
					(const_int 32)))
		     (const_int 32)))]
  "TARGET_64BIT && ix86_unary_operator_ok (NEG, SImode, operands)"
  "neg{l}\t%k0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "SI")])

;; Changing of sign for FP values is doable using integer unit too.

(define_expand "<code><mode>2"
  [(set (match_operand:X87MODEF 0 "register_operand" "")
	(absneg:X87MODEF (match_operand:X87MODEF 1 "register_operand" "")))]
  "TARGET_80387 || (SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)"
  "ix86_expand_fp_absneg_operator (<CODE>, <MODE>mode, operands); DONE;")

(define_insn "*absneg<mode>2_mixed"
  [(set (match_operand:MODEF 0 "register_operand" "=x,x,f,!r")
	(match_operator:MODEF 3 "absneg_operator"
	  [(match_operand:MODEF 1 "register_operand" "0,x,0,0")]))
   (use (match_operand:<ssevecmode> 2 "nonimmediate_operand" "xm,0,X,X"))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_MIX_SSE_I387 && SSE_FLOAT_MODE_P (<MODE>mode)"
  "#")

(define_insn "*absneg<mode>2_sse"
  [(set (match_operand:MODEF 0 "register_operand" "=x,x,!r")
	(match_operator:MODEF 3 "absneg_operator"
	  [(match_operand:MODEF 1 "register_operand" "0 ,x,0")]))
   (use (match_operand:<ssevecmode> 2 "register_operand" "xm,0,X"))
   (clobber (reg:CC FLAGS_REG))]
  "SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH"
  "#")

(define_insn "*absneg<mode>2_i387"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f,!r")
	(match_operator:X87MODEF 3 "absneg_operator"
	  [(match_operand:X87MODEF 1 "register_operand" "0,0")]))
   (use (match_operand 2 "" ""))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_80387 && !(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)"
  "#")

(define_expand "<code>tf2"
  [(set (match_operand:TF 0 "register_operand" "")
	(absneg:TF (match_operand:TF 1 "register_operand" "")))]
  "TARGET_SSE2"
  "ix86_expand_fp_absneg_operator (<CODE>, TFmode, operands); DONE;")

(define_insn "*absnegtf2_sse"
  [(set (match_operand:TF 0 "register_operand" "=x,x")
	(match_operator:TF 3 "absneg_operator"
	  [(match_operand:TF 1 "register_operand" "0,x")]))
   (use (match_operand:TF 2 "nonimmediate_operand" "xm,0"))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_SSE2"
  "#")

;; Splitters for fp abs and neg.

(define_split
  [(set (match_operand 0 "fp_register_operand" "")
	(match_operator 1 "absneg_operator" [(match_dup 0)]))
   (use (match_operand 2 "" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  [(set (match_dup 0) (match_op_dup 1 [(match_dup 0)]))])

(define_split
  [(set (match_operand 0 "register_operand" "")
	(match_operator 3 "absneg_operator"
	  [(match_operand 1 "register_operand" "")]))
   (use (match_operand 2 "nonimmediate_operand" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed && SSE_REG_P (operands[0])"
  [(set (match_dup 0) (match_dup 3))]
{
  enum machine_mode mode = GET_MODE (operands[0]);
  enum machine_mode vmode = GET_MODE (operands[2]);
  rtx tmp;

  operands[0] = simplify_gen_subreg (vmode, operands[0], mode, 0);
  operands[1] = simplify_gen_subreg (vmode, operands[1], mode, 0);
  if (operands_match_p (operands[0], operands[2]))
    {
      tmp = operands[1];
      operands[1] = operands[2];
      operands[2] = tmp;
    }
  if (GET_CODE (operands[3]) == ABS)
    tmp = gen_rtx_AND (vmode, operands[1], operands[2]);
  else
    tmp = gen_rtx_XOR (vmode, operands[1], operands[2]);
  operands[3] = tmp;
})

(define_split
  [(set (match_operand:SF 0 "register_operand" "")
	(match_operator:SF 1 "absneg_operator" [(match_dup 0)]))
   (use (match_operand:V4SF 2 "" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
{
  rtx tmp;
  operands[0] = gen_lowpart (SImode, operands[0]);
  if (GET_CODE (operands[1]) == ABS)
    {
      tmp = gen_int_mode (0x7fffffff, SImode);
      tmp = gen_rtx_AND (SImode, operands[0], tmp);
    }
  else
    {
      tmp = gen_int_mode (0x80000000, SImode);
      tmp = gen_rtx_XOR (SImode, operands[0], tmp);
    }
  operands[1] = tmp;
})

(define_split
  [(set (match_operand:DF 0 "register_operand" "")
	(match_operator:DF 1 "absneg_operator" [(match_dup 0)]))
   (use (match_operand 2 "" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
{
  rtx tmp;
  if (TARGET_64BIT)
    {
      tmp = gen_lowpart (DImode, operands[0]);
      tmp = gen_rtx_ZERO_EXTRACT (DImode, tmp, const1_rtx, GEN_INT (63));
      operands[0] = tmp;

      if (GET_CODE (operands[1]) == ABS)
	tmp = const0_rtx;
      else
	tmp = gen_rtx_NOT (DImode, tmp);
    }
  else
    {
      operands[0] = gen_highpart (SImode, operands[0]);
      if (GET_CODE (operands[1]) == ABS)
	{
	  tmp = gen_int_mode (0x7fffffff, SImode);
	  tmp = gen_rtx_AND (SImode, operands[0], tmp);
	}
      else
	{
	  tmp = gen_int_mode (0x80000000, SImode);
	  tmp = gen_rtx_XOR (SImode, operands[0], tmp);
	}
    }
  operands[1] = tmp;
})

(define_split
  [(set (match_operand:XF 0 "register_operand" "")
	(match_operator:XF 1 "absneg_operator" [(match_dup 0)]))
   (use (match_operand 2 "" ""))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed"
  [(parallel [(set (match_dup 0) (match_dup 1))
	      (clobber (reg:CC FLAGS_REG))])]
{
  rtx tmp;
  operands[0] = gen_rtx_REG (SImode,
			     true_regnum (operands[0])
			     + (TARGET_64BIT ? 1 : 2));
  if (GET_CODE (operands[1]) == ABS)
    {
      tmp = GEN_INT (0x7fff);
      tmp = gen_rtx_AND (SImode, operands[0], tmp);
    }
  else
    {
      tmp = GEN_INT (0x8000);
      tmp = gen_rtx_XOR (SImode, operands[0], tmp);
    }
  operands[1] = tmp;
})

;; Conditionalize these after reload. If they match before reload, we
;; lose the clobber and ability to use integer instructions.

(define_insn "*<code><mode>2_1"
  [(set (match_operand:X87MODEF 0 "register_operand" "=f")
	(absneg:X87MODEF (match_operand:X87MODEF 1 "register_operand" "0")))]
  "TARGET_80387
   && (reload_completed
       || !(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH))"
  "f<absneg_mnemonic>"
  [(set_attr "type" "fsgn")
   (set_attr "mode" "<MODE>")])

(define_insn "*<code>extendsfdf2"
  [(set (match_operand:DF 0 "register_operand" "=f")
	(absneg:DF (float_extend:DF
		     (match_operand:SF 1 "register_operand" "0"))))]
  "TARGET_80387 && (!TARGET_SSE_MATH || TARGET_MIX_SSE_I387)"
  "f<absneg_mnemonic>"
  [(set_attr "type" "fsgn")
   (set_attr "mode" "DF")])

(define_insn "*<code>extendsfxf2"
  [(set (match_operand:XF 0 "register_operand" "=f")
	(absneg:XF (float_extend:XF
		     (match_operand:SF 1 "register_operand" "0"))))]
  "TARGET_80387"
  "f<absneg_mnemonic>"
  [(set_attr "type" "fsgn")
   (set_attr "mode" "XF")])

(define_insn "*<code>extenddfxf2"
  [(set (match_operand:XF 0 "register_operand" "=f")
	(absneg:XF (float_extend:XF
		     (match_operand:DF 1 "register_operand" "0"))))]
  "TARGET_80387"
  "f<absneg_mnemonic>"
  [(set_attr "type" "fsgn")
   (set_attr "mode" "XF")])

;; Copysign instructions

(define_mode_iterator CSGNMODE [SF DF TF])
(define_mode_attr CSGNVMODE [(SF "V4SF") (DF "V2DF") (TF "TF")])

(define_expand "copysign<mode>3"
  [(match_operand:CSGNMODE 0 "register_operand" "")
   (match_operand:CSGNMODE 1 "nonmemory_operand" "")
   (match_operand:CSGNMODE 2 "register_operand" "")]
  "(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
   || (TARGET_SSE2 && (<MODE>mode == TFmode))"
  "ix86_expand_copysign (operands); DONE;")

(define_insn_and_split "copysign<mode>3_const"
  [(set (match_operand:CSGNMODE 0 "register_operand" "=x")
	(unspec:CSGNMODE
	  [(match_operand:<CSGNVMODE> 1 "vector_move_operand" "xmC")
	   (match_operand:CSGNMODE 2 "register_operand" "0")
	   (match_operand:<CSGNVMODE> 3 "nonimmediate_operand" "xm")]
	  UNSPEC_COPYSIGN))]
  "(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
   || (TARGET_SSE2 && (<MODE>mode == TFmode))"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_copysign_const (operands); DONE;")

(define_insn "copysign<mode>3_var"
  [(set (match_operand:CSGNMODE 0 "register_operand" "=x,x,x,x,x")
	(unspec:CSGNMODE
	  [(match_operand:CSGNMODE 2 "register_operand"	"x,0,0,x,x")
	   (match_operand:CSGNMODE 3 "register_operand"	"1,1,x,1,x")
	   (match_operand:<CSGNVMODE> 4 "nonimmediate_operand" "X,xm,xm,0,0")
	   (match_operand:<CSGNVMODE> 5 "nonimmediate_operand" "0,xm,1,xm,1")]
	  UNSPEC_COPYSIGN))
   (clobber (match_scratch:<CSGNVMODE> 1 "=x,x,x,x,x"))]
  "(SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
   || (TARGET_SSE2 && (<MODE>mode == TFmode))"
  "#")

(define_split
  [(set (match_operand:CSGNMODE 0 "register_operand" "")
	(unspec:CSGNMODE
	  [(match_operand:CSGNMODE 2 "register_operand" "")
	   (match_operand:CSGNMODE 3 "register_operand" "")
	   (match_operand:<CSGNVMODE> 4 "" "")
	   (match_operand:<CSGNVMODE> 5 "" "")]
	  UNSPEC_COPYSIGN))
   (clobber (match_scratch:<CSGNVMODE> 1 ""))]
  "((SSE_FLOAT_MODE_P (<MODE>mode) && TARGET_SSE_MATH)
    || (TARGET_SSE2 && (<MODE>mode == TFmode)))
   && reload_completed"
  [(const_int 0)]
  "ix86_split_copysign_var (operands); DONE;")

;; One complement instructions

(define_expand "one_cmpl<mode>2"
  [(set (match_operand:SWIM 0 "nonimmediate_operand" "")
	(not:SWIM (match_operand:SWIM 1 "nonimmediate_operand" "")))]
  ""
  "ix86_expand_unary_operator (NOT, <MODE>mode, operands); DONE;")

(define_insn "*one_cmpl<mode>2_1"
  [(set (match_operand:SWI248 0 "nonimmediate_operand" "=rm")
	(not:SWI248 (match_operand:SWI248 1 "nonimmediate_operand" "0")))]
  "ix86_unary_operator_ok (NOT, <MODE>mode, operands)"
  "not{<imodesuffix>}\t%0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "<MODE>")])

;; %%% Potential partial reg stall on alternative 1.  What to do?
(define_insn "*one_cmplqi2_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm,r")
	(not:QI (match_operand:QI 1 "nonimmediate_operand" "0,0")))]
  "ix86_unary_operator_ok (NOT, QImode, operands)"
  "@
   not{b}\t%0
   not{l}\t%k0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "QI,SI")])

;; ??? Currently never generated - xor is used instead.
(define_insn "*one_cmplsi2_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (not:SI (match_operand:SI 1 "register_operand" "0"))))]
  "TARGET_64BIT && ix86_unary_operator_ok (NOT, SImode, operands)"
  "not{l}\t%k0"
  [(set_attr "type" "negnot")
   (set_attr "mode" "SI")])

(define_insn "*one_cmpl<mode>2_2"
  [(set (reg FLAGS_REG)
	(compare (not:SWI (match_operand:SWI 1 "nonimmediate_operand" "0"))
		 (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(not:SWI (match_dup 1)))]
  "ix86_match_ccmode (insn, CCNOmode)
   && ix86_unary_operator_ok (NOT, <MODE>mode, operands)"
  "#"
  [(set_attr "type" "alu1")
   (set_attr "mode" "<MODE>")])

(define_split
  [(set (match_operand 0 "flags_reg_operand" "")
	(match_operator 2 "compare_operator"
	  [(not:SWI (match_operand:SWI 3 "nonimmediate_operand" ""))
	   (const_int 0)]))
   (set (match_operand:SWI 1 "nonimmediate_operand" "")
	(not:SWI (match_dup 3)))]
  "ix86_match_ccmode (insn, CCNOmode)"
  [(parallel [(set (match_dup 0)
		   (match_op_dup 2 [(xor:SWI (match_dup 3) (const_int -1))
				    (const_int 0)]))
	      (set (match_dup 1)
		   (xor:SWI (match_dup 3) (const_int -1)))])])

;; ??? Currently never generated - xor is used instead.
(define_insn "*one_cmplsi2_2_zext"
  [(set (reg FLAGS_REG)
	(compare (not:SI (match_operand:SI 1 "register_operand" "0"))
		 (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (not:SI (match_dup 1))))]
  "TARGET_64BIT && ix86_match_ccmode (insn, CCNOmode)
   && ix86_unary_operator_ok (NOT, SImode, operands)"
  "#"
  [(set_attr "type" "alu1")
   (set_attr "mode" "SI")])

(define_split
  [(set (match_operand 0 "flags_reg_operand" "")
	(match_operator 2 "compare_operator"
	  [(not:SI (match_operand:SI 3 "register_operand" ""))
	   (const_int 0)]))
   (set (match_operand:DI 1 "register_operand" "")
	(zero_extend:DI (not:SI (match_dup 3))))]
  "ix86_match_ccmode (insn, CCNOmode)"
  [(parallel [(set (match_dup 0)
		   (match_op_dup 2 [(xor:SI (match_dup 3) (const_int -1))
				    (const_int 0)]))
	      (set (match_dup 1)
		   (zero_extend:DI (xor:SI (match_dup 3) (const_int -1))))])])

;; Shift instructions

;; DImode shifts are implemented using the i386 "shift double" opcode,
;; which is written as "sh[lr]d[lw] imm,reg,reg/mem".  If the shift count
;; is variable, then the count is in %cl and the "imm" operand is dropped
;; from the assembler input.
;;
;; This instruction shifts the target reg/mem as usual, but instead of
;; shifting in zeros, bits are shifted in from reg operand.  If the insn
;; is a left shift double, bits are taken from the high order bits of
;; reg, else if the insn is a shift right double, bits are taken from the
;; low order bits of reg.  So if %eax is "1234" and %edx is "5678",
;; "shldl $8,%edx,%eax" leaves %edx unchanged and sets %eax to "2345".
;;
;; Since sh[lr]d does not change the `reg' operand, that is done
;; separately, making all shifts emit pairs of shift double and normal
;; shift.  Since sh[lr]d does not shift more than 31 bits, and we wish to
;; support a 63 bit shift, each shift where the count is in a reg expands
;; to a pair of shifts, a branch, a shift by 32 and a label.
;;
;; If the shift count is a constant, we need never emit more than one
;; shift pair, instead using moves and sign extension for counts greater
;; than 31.

(define_expand "ashl<mode>3"
  [(set (match_operand:SDWIM 0 "<shift_operand>" "")
	(ashift:SDWIM (match_operand:SDWIM 1 "<ashl_input_operand>" "")
		      (match_operand:QI 2 "nonmemory_operand" "")))]
  ""
  "ix86_expand_binary_operator (ASHIFT, <MODE>mode, operands); DONE;")

(define_insn "*ashl<mode>3_doubleword"
  [(set (match_operand:DWI 0 "register_operand" "=&r,r")
	(ashift:DWI (match_operand:DWI 1 "reg_or_pm1_operand" "n,0")
		    (match_operand:QI 2 "nonmemory_operand" "<S>c,<S>c")))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  [(set_attr "type" "multi")])

(define_split
  [(set (match_operand:DWI 0 "register_operand" "")
	(ashift:DWI (match_operand:DWI 1 "nonmemory_operand" "")
		    (match_operand:QI 2 "nonmemory_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "(optimize && flag_peephole2) ? epilogue_completed : reload_completed"
  [(const_int 0)]
  "ix86_split_ashl (operands, NULL_RTX, <MODE>mode); DONE;")

;; By default we don't ask for a scratch register, because when DWImode
;; values are manipulated, registers are already at a premium.  But if
;; we have one handy, we won't turn it away.

(define_peephole2
  [(match_scratch:DWIH 3 "r")
   (parallel [(set (match_operand:<DWI> 0 "register_operand" "")
		   (ashift:<DWI>
		     (match_operand:<DWI> 1 "nonmemory_operand" "")
		     (match_operand:QI 2 "nonmemory_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])
   (match_dup 3)]
  "TARGET_CMOVE"
  [(const_int 0)]
  "ix86_split_ashl (operands, operands[3], <DWI>mode); DONE;")

(define_insn "x86_64_shld"
  [(set (match_operand:DI 0 "nonimmediate_operand" "+r*m")
        (ior:DI (ashift:DI (match_dup 0)
		  (match_operand:QI 2 "nonmemory_operand" "Jc"))
		(lshiftrt:DI (match_operand:DI 1 "register_operand" "r")
		  (minus:QI (const_int 64) (match_dup 2)))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT"
  "shld{q}\t{%s2%1, %0|%0, %1, %2}"
  [(set_attr "type" "ishift")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "vector")
   (set_attr "bdver1_decode" "vector")])

(define_insn "x86_shld"
  [(set (match_operand:SI 0 "nonimmediate_operand" "+r*m")
        (ior:SI (ashift:SI (match_dup 0)
		  (match_operand:QI 2 "nonmemory_operand" "Ic"))
		(lshiftrt:SI (match_operand:SI 1 "register_operand" "r")
		  (minus:QI (const_int 32) (match_dup 2)))))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "shld{l}\t{%s2%1, %0|%0, %1, %2}"
  [(set_attr "type" "ishift")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "SI")
   (set_attr "pent_pair" "np")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "vector")
   (set_attr "bdver1_decode" "vector")])

(define_expand "x86_shift<mode>_adj_1"
  [(set (reg:CCZ FLAGS_REG)
	(compare:CCZ (and:QI (match_operand:QI 2 "register_operand" "")
			     (match_dup 4))
		     (const_int 0)))
   (set (match_operand:SWI48 0 "register_operand" "")
        (if_then_else:SWI48 (ne (reg:CCZ FLAGS_REG) (const_int 0))
			    (match_operand:SWI48 1 "register_operand" "")
			    (match_dup 0)))
   (set (match_dup 1)
	(if_then_else:SWI48 (ne (reg:CCZ FLAGS_REG) (const_int 0))
			    (match_operand:SWI48 3 "register_operand" "r")
			    (match_dup 1)))]
  "TARGET_CMOVE"
  "operands[4] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode));")

(define_expand "x86_shift<mode>_adj_2"
  [(use (match_operand:SWI48 0 "register_operand" ""))
   (use (match_operand:SWI48 1 "register_operand" ""))
   (use (match_operand:QI 2 "register_operand" ""))]
  ""
{
  rtx label = gen_label_rtx ();
  rtx tmp;

  emit_insn (gen_testqi_ccz_1 (operands[2],
			       GEN_INT (GET_MODE_BITSIZE (<MODE>mode))));

  tmp = gen_rtx_REG (CCZmode, FLAGS_REG);
  tmp = gen_rtx_EQ (VOIDmode, tmp, const0_rtx);
  tmp = gen_rtx_IF_THEN_ELSE (VOIDmode, tmp,
			      gen_rtx_LABEL_REF (VOIDmode, label),
			      pc_rtx);
  tmp = emit_jump_insn (gen_rtx_SET (VOIDmode, pc_rtx, tmp));
  JUMP_LABEL (tmp) = label;

  emit_move_insn (operands[0], operands[1]);
  ix86_expand_clear (operands[1]);

  emit_label (label);
  LABEL_NUSES (label) = 1;

  DONE;
})

;; Avoid useless masking of count operand.
(define_insn_and_split "*ashl<mode>3_mask"
  [(set (match_operand:SWI48 0 "nonimmediate_operand" "=rm")
	(ashift:SWI48
	  (match_operand:SWI48 1 "nonimmediate_operand" "0")
	  (subreg:QI
	    (and:SI
	      (match_operand:SI 2 "nonimmediate_operand" "c")
	      (match_operand:SI 3 "const_int_operand" "n")) 0)))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (ASHIFT, <MODE>mode, operands)
   && (INTVAL (operands[3]) & (GET_MODE_BITSIZE (<MODE>mode)-1))
      == GET_MODE_BITSIZE (<MODE>mode)-1"
  "#"
  "&& 1"
  [(parallel [(set (match_dup 0)
		   (ashift:SWI48 (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
{
  if (can_create_pseudo_p ())
    operands [2] = force_reg (SImode, operands[2]);

  operands[2] = simplify_gen_subreg (QImode, operands[2], SImode, 0);
}
  [(set_attr "type" "ishift")
   (set_attr "mode" "<MODE>")])

(define_insn "*ashl<mode>3_1"
  [(set (match_operand:SWI48 0 "nonimmediate_operand" "=rm,r")
	(ashift:SWI48 (match_operand:SWI48 1 "nonimmediate_operand" "0,l")
		      (match_operand:QI 2 "nonmemory_operand" "c<S>,M")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (ASHIFT, <MODE>mode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      gcc_assert (rtx_equal_p (operands[0], operands[1]));
      return "add{<imodesuffix>}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{<imodesuffix>}\t%0";
      else
	return "sal{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "1")
	      (const_string "lea")
            (and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*ashlsi3_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r,r")
	(zero_extend:DI
	  (ashift:SI (match_operand:SI 1 "register_operand" "0,l")
		     (match_operand:QI 2 "nonmemory_operand" "cI,M"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (ASHIFT, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{l}\t%k0, %k0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{l}\t%k0";
      else
	return "sal{l}\t{%2, %k0|%k0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "1")
	      (const_string "lea")
            (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		     (const_int 0))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*ashlhi3_1"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=rm")
	(ashift:HI (match_operand:HI 1 "nonimmediate_operand" "0")
		   (match_operand:QI 2 "nonmemory_operand" "cI")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (ASHIFT, HImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{w}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{w}\t%0";
      else
	return "sal{w}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "HI")])

(define_insn "*ashlhi3_1_lea"
  [(set (match_operand:HI 0 "nonimmediate_operand" "=rm,r")
	(ashift:HI (match_operand:HI 1 "nonimmediate_operand" "0,l")
		   (match_operand:QI 2 "nonmemory_operand" "cI,M")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (ASHIFT, HImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{w}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{w}\t%0";
      else
	return "sal{w}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "1")
	      (const_string "lea")
            (and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "HI,SI")])

(define_insn "*ashlqi3_1"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm,r")
	(ashift:QI (match_operand:QI 1 "nonimmediate_operand" "0,0")
		   (match_operand:QI 2 "nonmemory_operand" "cI,cI")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (ASHIFT, QImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      if (REG_P (operands[1]) && !ANY_QI_REG_P (operands[1]))
        return "add{l}\t%k0, %k0";
      else
        return "add{b}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	{
	  if (get_attr_mode (insn) == MODE_SI)
	    return "sal{l}\t%k0";
	  else
	    return "sal{b}\t%0";
	}
      else
	{
	  if (get_attr_mode (insn) == MODE_SI)
	    return "sal{l}\t{%2, %k0|%k0, %2}";
	  else
	    return "sal{b}\t{%2, %0|%0, %2}";
	}
    }
}
  [(set (attr "type")
     (cond [(and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "QI,SI")])

;; %%% Potential partial reg stall on alternative 2.  What to do?
(define_insn "*ashlqi3_1_lea"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm,r,r")
	(ashift:QI (match_operand:QI 1 "nonimmediate_operand" "0,0,l")
		   (match_operand:QI 2 "nonmemory_operand" "cI,cI,M")))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL
   && ix86_binary_operator_ok (ASHIFT, QImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_LEA:
      return "#";

    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      if (REG_P (operands[1]) && !ANY_QI_REG_P (operands[1]))
        return "add{l}\t%k0, %k0";
      else
        return "add{b}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	{
	  if (get_attr_mode (insn) == MODE_SI)
	    return "sal{l}\t%k0";
	  else
	    return "sal{b}\t%0";
	}
      else
	{
	  if (get_attr_mode (insn) == MODE_SI)
	    return "sal{l}\t{%2, %k0|%k0, %2}";
	  else
	    return "sal{b}\t{%2, %0|%0, %2}";
	}
    }
}
  [(set (attr "type")
     (cond [(eq_attr "alternative" "2")
	      (const_string "lea")
            (and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "QI,SI,SI")])

(define_insn "*ashlqi3_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm"))
	(ashift:QI (match_dup 0)
		   (match_operand:QI 1 "nonmemory_operand" "cI")))
   (clobber (reg:CC FLAGS_REG))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_FLAG_REG_STALL
    || (operands[1] == const1_rtx
	&& (TARGET_SHIFT1
	    || (TARGET_DOUBLE_WITH_ADD && REG_P (operands[0])))))"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[1] == const1_rtx);
      return "add{b}\t%0, %0";

    default:
      if (operands[1] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{b}\t%0";
      else
	return "sal{b}\t{%1, %0|%0, %1}";
    }
}
  [(set (attr "type")
     (cond [(and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 1 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift1")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift1")
		 (and (match_operand 1 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "QI")])

;; Convert lea to the lea pattern to avoid flags dependency.
(define_split
  [(set (match_operand 0 "register_operand" "")
	(ashift (match_operand 1 "index_register_operand" "")
                (match_operand:QI 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "reload_completed
   && true_regnum (operands[0]) != true_regnum (operands[1])"
  [(const_int 0)]
{
  rtx pat;
  enum machine_mode mode = GET_MODE (operands[0]);

  if (mode != Pmode)
    operands[1] = gen_lowpart (Pmode, operands[1]);
  operands[2] = gen_int_mode (1 << INTVAL (operands[2]), Pmode);

  pat = gen_rtx_MULT (Pmode, operands[1], operands[2]);

  if (GET_MODE_SIZE (mode) < GET_MODE_SIZE (SImode))
    operands[0] = gen_lowpart (SImode, operands[0]);

  if (TARGET_64BIT && mode != Pmode)
    pat = gen_rtx_SUBREG (SImode, pat, 0);

  emit_insn (gen_rtx_SET (VOIDmode, operands[0], pat));
  DONE;
})

;; Convert lea to the lea pattern to avoid flags dependency.
(define_split
  [(set (match_operand:DI 0 "register_operand" "")
	(zero_extend:DI
	  (ashift:SI (match_operand:SI 1 "index_register_operand" "")
		     (match_operand:QI 2 "const_int_operand" ""))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && reload_completed
   && true_regnum (operands[0]) != true_regnum (operands[1])"
  [(set (match_dup 0)
	(zero_extend:DI (subreg:SI (mult:DI (match_dup 1) (match_dup 2)) 0)))]
{
  operands[1] = gen_lowpart (DImode, operands[1]);
  operands[2] = gen_int_mode (1 << INTVAL (operands[2]), DImode);
})

;; This pattern can't accept a variable shift count, since shifts by
;; zero don't affect the flags.  We assume that shifts by constant
;; zero are optimized away.
(define_insn "*ashl<mode>3_cmp"
  [(set (reg FLAGS_REG)
	(compare
	  (ashift:SWI (match_operand:SWI 1 "nonimmediate_operand" "0")
		      (match_operand:QI 2 "<shift_immediate_operand>" "<S>"))
	  (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(ashift:SWI (match_dup 1) (match_dup 2)))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_FLAG_REG_STALL
    || (operands[2] == const1_rtx
	&& (TARGET_SHIFT1
	    || (TARGET_DOUBLE_WITH_ADD && REG_P (operands[0])))))
   && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (ASHIFT, <MODE>mode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{<imodesuffix>}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{<imodesuffix>}\t%0";
      else
	return "sal{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*ashlsi3_cmp_zext"
  [(set (reg FLAGS_REG)
	(compare
	  (ashift:SI (match_operand:SI 1 "register_operand" "0")
		     (match_operand:QI 2 "const_1_to_31_operand" "I"))
	  (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (ashift:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT
   && (optimize_function_for_size_p (cfun)
       || !TARGET_PARTIAL_FLAG_REG_STALL
       || (operands[2] == const1_rtx
	   && (TARGET_SHIFT1
	       || TARGET_DOUBLE_WITH_ADD)))
   && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (ASHIFT, SImode, operands)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{l}\t%k0, %k0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{l}\t%k0";
      else
	return "sal{l}\t{%2, %k0|%k0, %2}";
    }
}
  [(set (attr "type")
     (cond [(and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		     (const_int 0))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*ashl<mode>3_cconly"
  [(set (reg FLAGS_REG)
	(compare
	  (ashift:SWI (match_operand:SWI 1 "register_operand" "0")
		      (match_operand:QI 2 "<shift_immediate_operand>" "<S>"))
	  (const_int 0)))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_FLAG_REG_STALL
    || (operands[2] == const1_rtx
	&& (TARGET_SHIFT1
	    || TARGET_DOUBLE_WITH_ADD)))
   && ix86_match_ccmode (insn, CCGOCmode)"
{
  switch (get_attr_type (insn))
    {
    case TYPE_ALU:
      gcc_assert (operands[2] == const1_rtx);
      return "add{<imodesuffix>}\t%0, %0";

    default:
      if (operands[2] == const1_rtx
	  && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
	return "sal{<imodesuffix>}\t%0";
      else
	return "sal{<imodesuffix>}\t{%2, %0|%0, %2}";
    }
}
  [(set (attr "type")
     (cond [(and (and (ne (symbol_ref "TARGET_DOUBLE_WITH_ADD")
		          (const_int 0))
		      (match_operand 0 "register_operand" ""))
		 (match_operand 2 "const1_operand" ""))
	      (const_string "alu")
	   ]
	   (const_string "ishift")))
   (set (attr "length_immediate")
     (if_then_else
       (ior (eq_attr "type" "alu")
	    (and (eq_attr "type" "ishift")
		 (and (match_operand 2 "const1_operand" "")
		      (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
			  (const_int 0)))))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

;; See comment above `ashl<mode>3' about how this works.

(define_expand "<shiftrt_insn><mode>3"
  [(set (match_operand:SDWIM 0 "<shift_operand>" "")
	(any_shiftrt:SDWIM (match_operand:SDWIM 1 "<shift_operand>" "")
			   (match_operand:QI 2 "nonmemory_operand" "")))]
  ""
  "ix86_expand_binary_operator (<CODE>, <MODE>mode, operands); DONE;")

;; Avoid useless masking of count operand.
(define_insn_and_split "*<shiftrt_insn><mode>3_mask"
  [(set (match_operand:SWI48 0 "nonimmediate_operand" "=rm")
	(any_shiftrt:SWI48
	  (match_operand:SWI48 1 "nonimmediate_operand" "0")
	  (subreg:QI
	    (and:SI
	      (match_operand:SI 2 "nonimmediate_operand" "c")
	      (match_operand:SI 3 "const_int_operand" "n")) 0)))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)
   && (INTVAL (operands[3]) & (GET_MODE_BITSIZE (<MODE>mode)-1))
      == GET_MODE_BITSIZE (<MODE>mode)-1"
  "#"
  "&& 1"
  [(parallel [(set (match_dup 0)
		   (any_shiftrt:SWI48 (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
{
  if (can_create_pseudo_p ())
    operands [2] = force_reg (SImode, operands[2]);

  operands[2] = simplify_gen_subreg (QImode, operands[2], SImode, 0);
}
  [(set_attr "type" "ishift")
   (set_attr "mode" "<MODE>")])

(define_insn_and_split "*<shiftrt_insn><mode>3_doubleword"
  [(set (match_operand:DWI 0 "register_operand" "=r")
	(any_shiftrt:DWI (match_operand:DWI 1 "register_operand" "0")
			 (match_operand:QI 2 "nonmemory_operand" "<S>c")))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "#"
  "(optimize && flag_peephole2) ? epilogue_completed : reload_completed"
  [(const_int 0)]
  "ix86_split_<shiftrt_insn> (operands, NULL_RTX, <MODE>mode); DONE;"
  [(set_attr "type" "multi")])

;; By default we don't ask for a scratch register, because when DWImode
;; values are manipulated, registers are already at a premium.  But if
;; we have one handy, we won't turn it away.

(define_peephole2
  [(match_scratch:DWIH 3 "r")
   (parallel [(set (match_operand:<DWI> 0 "register_operand" "")
		   (any_shiftrt:<DWI>
		     (match_operand:<DWI> 1 "register_operand" "")
		     (match_operand:QI 2 "nonmemory_operand" "")))
	      (clobber (reg:CC FLAGS_REG))])
   (match_dup 3)]
  "TARGET_CMOVE"
  [(const_int 0)]
  "ix86_split_<shiftrt_insn> (operands, operands[3], <DWI>mode); DONE;")

(define_insn "x86_64_shrd"
  [(set (match_operand:DI 0 "nonimmediate_operand" "+r*m")
        (ior:DI (ashiftrt:DI (match_dup 0)
		  (match_operand:QI 2 "nonmemory_operand" "Jc"))
		(ashift:DI (match_operand:DI 1 "register_operand" "r")
		  (minus:QI (const_int 64) (match_dup 2)))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT"
  "shrd{q}\t{%s2%1, %0|%0, %1, %2}"
  [(set_attr "type" "ishift")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "vector")
   (set_attr "bdver1_decode" "vector")])

(define_insn "x86_shrd"
  [(set (match_operand:SI 0 "nonimmediate_operand" "+r*m")
        (ior:SI (ashiftrt:SI (match_dup 0)
		  (match_operand:QI 2 "nonmemory_operand" "Ic"))
		(ashift:SI (match_operand:SI 1 "register_operand" "r")
		  (minus:QI (const_int 32) (match_dup 2)))))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "shrd{l}\t{%s2%1, %0|%0, %1, %2}"
  [(set_attr "type" "ishift")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "SI")
   (set_attr "pent_pair" "np")
   (set_attr "athlon_decode" "vector")
   (set_attr "amdfam10_decode" "vector")
   (set_attr "bdver1_decode" "vector")])

(define_insn "ashrdi3_cvt"
  [(set (match_operand:DI 0 "nonimmediate_operand" "=*d,rm")
	(ashiftrt:DI (match_operand:DI 1 "nonimmediate_operand" "*a,0")
		     (match_operand:QI 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && INTVAL (operands[2]) == 63
   && (TARGET_USE_CLTD || optimize_function_for_size_p (cfun))
   && ix86_binary_operator_ok (ASHIFTRT, DImode, operands)"
  "@
   {cqto|cqo}
   sar{q}\t{%2, %0|%0, %2}"
  [(set_attr "type" "imovx,ishift")
   (set_attr "prefix_0f" "0,*")
   (set_attr "length_immediate" "0,*")
   (set_attr "modrm" "0,1")
   (set_attr "mode" "DI")])

(define_insn "ashrsi3_cvt"
  [(set (match_operand:SI 0 "nonimmediate_operand" "=*d,rm")
	(ashiftrt:SI (match_operand:SI 1 "nonimmediate_operand" "*a,0")
		     (match_operand:QI 2 "const_int_operand" "")))
   (clobber (reg:CC FLAGS_REG))]
  "INTVAL (operands[2]) == 31
   && (TARGET_USE_CLTD || optimize_function_for_size_p (cfun))
   && ix86_binary_operator_ok (ASHIFTRT, SImode, operands)"
  "@
   {cltd|cdq}
   sar{l}\t{%2, %0|%0, %2}"
  [(set_attr "type" "imovx,ishift")
   (set_attr "prefix_0f" "0,*")
   (set_attr "length_immediate" "0,*")
   (set_attr "modrm" "0,1")
   (set_attr "mode" "SI")])

(define_insn "*ashrsi3_cvt_zext"
  [(set (match_operand:DI 0 "register_operand" "=*d,r")
	(zero_extend:DI
	  (ashiftrt:SI (match_operand:SI 1 "register_operand" "*a,0")
		       (match_operand:QI 2 "const_int_operand" ""))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && INTVAL (operands[2]) == 31
   && (TARGET_USE_CLTD || optimize_function_for_size_p (cfun))
   && ix86_binary_operator_ok (ASHIFTRT, SImode, operands)"
  "@
   {cltd|cdq}
   sar{l}\t{%2, %k0|%k0, %2}"
  [(set_attr "type" "imovx,ishift")
   (set_attr "prefix_0f" "0,*")
   (set_attr "length_immediate" "0,*")
   (set_attr "modrm" "0,1")
   (set_attr "mode" "SI")])

(define_expand "x86_shift<mode>_adj_3"
  [(use (match_operand:SWI48 0 "register_operand" ""))
   (use (match_operand:SWI48 1 "register_operand" ""))
   (use (match_operand:QI 2 "register_operand" ""))]
  ""
{
  rtx label = gen_label_rtx ();
  rtx tmp;

  emit_insn (gen_testqi_ccz_1 (operands[2],
			       GEN_INT (GET_MODE_BITSIZE (<MODE>mode))));

  tmp = gen_rtx_REG (CCZmode, FLAGS_REG);
  tmp = gen_rtx_EQ (VOIDmode, tmp, const0_rtx);
  tmp = gen_rtx_IF_THEN_ELSE (VOIDmode, tmp,
			      gen_rtx_LABEL_REF (VOIDmode, label),
			      pc_rtx);
  tmp = emit_jump_insn (gen_rtx_SET (VOIDmode, pc_rtx, tmp));
  JUMP_LABEL (tmp) = label;

  emit_move_insn (operands[0], operands[1]);
  emit_insn (gen_ashr<mode>3_cvt (operands[1], operands[1],
				  GEN_INT (GET_MODE_BITSIZE (<MODE>mode)-1)));
  emit_label (label);
  LABEL_NUSES (label) = 1;

  DONE;
})

(define_insn "*<shiftrt_insn><mode>3_1"
  [(set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(any_shiftrt:SWI (match_operand:SWI 1 "nonimmediate_operand" "0")
			 (match_operand:QI 2 "nonmemory_operand" "c<S>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{<imodesuffix>}\t%0";
  else
    return "<shiftrt>{<imodesuffix>}\t{%2, %0|%0, %2}";
}
  [(set_attr "type" "ishift")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*<shiftrt_insn>si3_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (any_shiftrt:SI (match_operand:SI 1 "register_operand" "0")
			  (match_operand:QI 2 "nonmemory_operand" "cI"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (<CODE>, SImode, operands)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{l}\t%k0";
  else
    return "<shiftrt>{l}\t{%2, %k0|%k0, %2}";
}
  [(set_attr "type" "ishift")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*<shiftrt_insn>qi3_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm"))
	(any_shiftrt:QI (match_dup 0)
			(match_operand:QI 1 "nonmemory_operand" "cI")))
   (clobber (reg:CC FLAGS_REG))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_REG_STALL
    || (operands[1] == const1_rtx
	&& TARGET_SHIFT1))"
{
  if (operands[1] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{b}\t%0";
  else
    return "<shiftrt>{b}\t{%1, %0|%0, %1}";
}
  [(set_attr "type" "ishift1")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 1 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "QI")])

;; This pattern can't accept a variable shift count, since shifts by
;; zero don't affect the flags.  We assume that shifts by constant
;; zero are optimized away.
(define_insn "*<shiftrt_insn><mode>3_cmp"
  [(set (reg FLAGS_REG)
	(compare
	  (any_shiftrt:SWI
	    (match_operand:SWI 1 "nonimmediate_operand" "0")
	    (match_operand:QI 2 "<shift_immediate_operand>" "<S>"))
	  (const_int 0)))
   (set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(any_shiftrt:SWI (match_dup 1) (match_dup 2)))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_FLAG_REG_STALL
    || (operands[2] == const1_rtx
	&& TARGET_SHIFT1))
   && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{<imodesuffix>}\t%0";
  else
    return "<shiftrt>{<imodesuffix>}\t{%2, %0|%0, %2}";
}
  [(set_attr "type" "ishift")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*<shiftrt_insn>si3_cmp_zext"
  [(set (reg FLAGS_REG)
	(compare
	  (any_shiftrt:SI (match_operand:SI 1 "register_operand" "0")
			  (match_operand:QI 2 "const_1_to_31_operand" "I"))
	  (const_int 0)))
   (set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI (any_shiftrt:SI (match_dup 1) (match_dup 2))))]
  "TARGET_64BIT
   && (optimize_function_for_size_p (cfun)
       || !TARGET_PARTIAL_FLAG_REG_STALL
       || (operands[2] == const1_rtx
	   && TARGET_SHIFT1))
   && ix86_match_ccmode (insn, CCGOCmode)
   && ix86_binary_operator_ok (<CODE>, SImode, operands)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{l}\t%k0";
  else
    return "<shiftrt>{l}\t{%2, %k0|%k0, %2}";
}
  [(set_attr "type" "ishift")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*<shiftrt_insn><mode>3_cconly"
  [(set (reg FLAGS_REG)
	(compare
	  (any_shiftrt:SWI
	    (match_operand:SWI 1 "register_operand" "0")
	    (match_operand:QI 2 "<shift_immediate_operand>" "<S>"))
	  (const_int 0)))
   (clobber (match_scratch:SWI 0 "=<r>"))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_FLAG_REG_STALL
    || (operands[2] == const1_rtx
	&& TARGET_SHIFT1))
   && ix86_match_ccmode (insn, CCGOCmode)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<shiftrt>{<imodesuffix>}\t%0";
  else
    return "<shiftrt>{<imodesuffix>}\t{%2, %0|%0, %2}";
}
  [(set_attr "type" "ishift")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

;; Rotate instructions

(define_expand "<rotate_insn>ti3"
  [(set (match_operand:TI 0 "register_operand" "")
	(any_rotate:TI (match_operand:TI 1 "register_operand" "")
		       (match_operand:QI 2 "nonmemory_operand" "")))]
  "TARGET_64BIT"
{
  if (const_1_to_63_operand (operands[2], VOIDmode))
    emit_insn (gen_ix86_<rotate_insn>ti3_doubleword
		(operands[0], operands[1], operands[2]));
  else
    FAIL;

  DONE;
})

(define_expand "<rotate_insn>di3"
  [(set (match_operand:DI 0 "shiftdi_operand" "")
	(any_rotate:DI (match_operand:DI 1 "shiftdi_operand" "")
		       (match_operand:QI 2 "nonmemory_operand" "")))]
 ""
{
  if (TARGET_64BIT)
    ix86_expand_binary_operator (<CODE>, DImode, operands);
  else if (const_1_to_31_operand (operands[2], VOIDmode))
    emit_insn (gen_ix86_<rotate_insn>di3_doubleword
		(operands[0], operands[1], operands[2]));
  else
    FAIL;

  DONE;
})

(define_expand "<rotate_insn><mode>3"
  [(set (match_operand:SWIM124 0 "nonimmediate_operand" "")
	(any_rotate:SWIM124 (match_operand:SWIM124 1 "nonimmediate_operand" "")
			    (match_operand:QI 2 "nonmemory_operand" "")))]
  ""
  "ix86_expand_binary_operator (<CODE>, <MODE>mode, operands); DONE;")

;; Avoid useless masking of count operand.
(define_insn_and_split "*<rotate_insn><mode>3_mask"
  [(set (match_operand:SWI48 0 "nonimmediate_operand" "=rm")
	(any_rotate:SWI48
	  (match_operand:SWI48 1 "nonimmediate_operand" "0")
	  (subreg:QI
	    (and:SI
	      (match_operand:SI 2 "nonimmediate_operand" "c")
	      (match_operand:SI 3 "const_int_operand" "n")) 0)))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)
   && (INTVAL (operands[3]) & (GET_MODE_BITSIZE (<MODE>mode)-1))
      == GET_MODE_BITSIZE (<MODE>mode)-1"
  "#"
  "&& 1"
  [(parallel [(set (match_dup 0)
		   (any_rotate:SWI48 (match_dup 1) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
{
  if (can_create_pseudo_p ())
    operands [2] = force_reg (SImode, operands[2]);

  operands[2] = simplify_gen_subreg (QImode, operands[2], SImode, 0);
}
  [(set_attr "type" "rotate")
   (set_attr "mode" "<MODE>")])

;; Implement rotation using two double-precision
;; shift instructions and a scratch register.

(define_insn_and_split "ix86_rotl<dwi>3_doubleword"
 [(set (match_operand:<DWI> 0 "register_operand" "=r")
       (rotate:<DWI> (match_operand:<DWI> 1 "register_operand" "0")
		     (match_operand:QI 2 "<shift_immediate_operand>" "<S>")))
  (clobber (reg:CC FLAGS_REG))
  (clobber (match_scratch:DWIH 3 "=&r"))]
 ""
 "#"
 "reload_completed"
 [(set (match_dup 3) (match_dup 4))
  (parallel
   [(set (match_dup 4)
	 (ior:DWIH (ashift:DWIH (match_dup 4) (match_dup 2))
		   (lshiftrt:DWIH (match_dup 5)
				  (minus:QI (match_dup 6) (match_dup 2)))))
    (clobber (reg:CC FLAGS_REG))])
  (parallel
   [(set (match_dup 5)
	 (ior:DWIH (ashift:DWIH (match_dup 5) (match_dup 2))
		   (lshiftrt:DWIH (match_dup 3)
				  (minus:QI (match_dup 6) (match_dup 2)))))
    (clobber (reg:CC FLAGS_REG))])]
{
  operands[6] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode));

  split_double_mode (<DWI>mode, &operands[0], 1, &operands[4], &operands[5]);
})

(define_insn_and_split "ix86_rotr<dwi>3_doubleword"
 [(set (match_operand:<DWI> 0 "register_operand" "=r")
       (rotatert:<DWI> (match_operand:<DWI> 1 "register_operand" "0")
		       (match_operand:QI 2 "<shift_immediate_operand>" "<S>")))
  (clobber (reg:CC FLAGS_REG))
  (clobber (match_scratch:DWIH 3 "=&r"))]
 ""
 "#"
 "reload_completed"
 [(set (match_dup 3) (match_dup 4))
  (parallel
   [(set (match_dup 4)
	 (ior:DWIH (ashiftrt:DWIH (match_dup 4) (match_dup 2))
		   (ashift:DWIH (match_dup 5)
				(minus:QI (match_dup 6) (match_dup 2)))))
    (clobber (reg:CC FLAGS_REG))])
  (parallel
   [(set (match_dup 5)
	 (ior:DWIH (ashiftrt:DWIH (match_dup 5) (match_dup 2))
		   (ashift:DWIH (match_dup 3)
				(minus:QI (match_dup 6) (match_dup 2)))))
    (clobber (reg:CC FLAGS_REG))])]
{
  operands[6] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode));

  split_double_mode (<DWI>mode, &operands[0], 1, &operands[4], &operands[5]);
})

(define_insn "*<rotate_insn><mode>3_1"
  [(set (match_operand:SWI 0 "nonimmediate_operand" "=<r>m")
	(any_rotate:SWI (match_operand:SWI 1 "nonimmediate_operand" "0")
			(match_operand:QI 2 "nonmemory_operand" "c<S>")))
   (clobber (reg:CC FLAGS_REG))]
  "ix86_binary_operator_ok (<CODE>, <MODE>mode, operands)"
{
  if (operands[2] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<rotate>{<imodesuffix>}\t%0";
  else
    return "<rotate>{<imodesuffix>}\t{%2, %0|%0, %2}";
}
  [(set_attr "type" "rotate")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "<MODE>")])

(define_insn "*<rotate_insn>si3_1_zext"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(zero_extend:DI
	  (any_rotate:SI (match_operand:SI 1 "register_operand" "0")
			 (match_operand:QI 2 "nonmemory_operand" "cI"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && ix86_binary_operator_ok (<CODE>, SImode, operands)"
{
    if (operands[2] == const1_rtx
	&& (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<rotate>{l}\t%k0";
  else
    return "<rotate>{l}\t{%2, %k0|%k0, %2}";
}
  [(set_attr "type" "rotate")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 2 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "SI")])

(define_insn "*<rotate_insn>qi3_1_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm"))
	(any_rotate:QI (match_dup 0)
		       (match_operand:QI 1 "nonmemory_operand" "cI")))
   (clobber (reg:CC FLAGS_REG))]
  "(optimize_function_for_size_p (cfun)
    || !TARGET_PARTIAL_REG_STALL
    || (operands[1] == const1_rtx
	&& TARGET_SHIFT1))"
{
  if (operands[1] == const1_rtx
      && (TARGET_SHIFT1 || optimize_function_for_size_p (cfun)))
    return "<rotate>{b}\t%0";
  else
    return "<rotate>{b}\t{%1, %0|%0, %1}";
}
  [(set_attr "type" "rotate1")
   (set (attr "length_immediate")
     (if_then_else
       (and (match_operand 1 "const1_operand" "")
	    (ne (symbol_ref "TARGET_SHIFT1 || optimize_function_for_size_p (cfun)")
		(const_int 0)))
       (const_string "0")
       (const_string "*")))
   (set_attr "mode" "QI")])

(define_split
 [(set (match_operand:HI 0 "register_operand" "")
       (any_rotate:HI (match_dup 0) (const_int 8)))
  (clobber (reg:CC FLAGS_REG))]
 "reload_completed
  && (TARGET_USE_XCHGB || optimize_function_for_size_p (cfun))"
 [(parallel [(set (strict_low_part (match_dup 0))
		  (bswap:HI (match_dup 0)))
	     (clobber (reg:CC FLAGS_REG))])])

;; Bit set / bit test instructions

(define_expand "extv"
  [(set (match_operand:SI 0 "register_operand" "")
	(sign_extract:SI (match_operand:SI 1 "register_operand" "")
			 (match_operand:SI 2 "const8_operand" "")
			 (match_operand:SI 3 "const8_operand" "")))]
  ""
{
  /* Handle extractions from %ah et al.  */
  if (INTVAL (operands[2]) != 8 || INTVAL (operands[3]) != 8)
    FAIL;

  /* From mips.md: extract_bit_field doesn't verify that our source
     matches the predicate, so check it again here.  */
  if (! ext_register_operand (operands[1], VOIDmode))
    FAIL;
})

(define_expand "extzv"
  [(set (match_operand:SI 0 "register_operand" "")
	(zero_extract:SI (match_operand 1 "ext_register_operand" "")
			 (match_operand:SI 2 "const8_operand" "")
			 (match_operand:SI 3 "const8_operand" "")))]
  ""
{
  /* Handle extractions from %ah et al.  */
  if (INTVAL (operands[2]) != 8 || INTVAL (operands[3]) != 8)
    FAIL;

  /* From mips.md: extract_bit_field doesn't verify that our source
     matches the predicate, so check it again here.  */
  if (! ext_register_operand (operands[1], VOIDmode))
    FAIL;
})

(define_expand "insv"
  [(set (zero_extract (match_operand 0 "ext_register_operand" "")
		      (match_operand 1 "const8_operand" "")
		      (match_operand 2 "const8_operand" ""))
        (match_operand 3 "register_operand" ""))]
  ""
{
  rtx (*gen_mov_insv_1) (rtx, rtx);

  /* Handle insertions to %ah et al.  */
  if (INTVAL (operands[1]) != 8 || INTVAL (operands[2]) != 8)
    FAIL;

  /* From mips.md: insert_bit_field doesn't verify that our source
     matches the predicate, so check it again here.  */
  if (! ext_register_operand (operands[0], VOIDmode))
    FAIL;

  gen_mov_insv_1 = (TARGET_64BIT
		    ? gen_movdi_insv_1 : gen_movsi_insv_1);

  emit_insn (gen_mov_insv_1 (operands[0], operands[3]));
  DONE;
})

;; %%% bts, btr, btc, bt.
;; In general these instructions are *slow* when applied to memory,
;; since they enforce atomic operation.  When applied to registers,
;; it depends on the cpu implementation.  They're never faster than
;; the corresponding and/ior/xor operations, so with 32-bit there's
;; no point.  But in 64-bit, we can't hold the relevant immediates
;; within the instruction itself, so operating on bits in the high
;; 32-bits of a register becomes easier.
;;
;; These are slow on Nocona, but fast on Athlon64.  We do require the use
;; of btrq and btcq for corner cases of post-reload expansion of absdf and
;; negdf respectively, so they can never be disabled entirely.

(define_insn "*btsq"
  [(set (zero_extract:DI (match_operand:DI 0 "register_operand" "+r")
			 (const_int 1)
			 (match_operand:DI 1 "const_0_to_63_operand" ""))
	(const_int 1))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && (TARGET_USE_BT || reload_completed)"
  "bts{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")])

(define_insn "*btrq"
  [(set (zero_extract:DI (match_operand:DI 0 "register_operand" "+r")
			 (const_int 1)
			 (match_operand:DI 1 "const_0_to_63_operand" ""))
	(const_int 0))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && (TARGET_USE_BT || reload_completed)"
  "btr{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")])

(define_insn "*btcq"
  [(set (zero_extract:DI (match_operand:DI 0 "register_operand" "+r")
			 (const_int 1)
			 (match_operand:DI 1 "const_0_to_63_operand" ""))
	(not:DI (zero_extract:DI (match_dup 0) (const_int 1) (match_dup 1))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT && (TARGET_USE_BT || reload_completed)"
  "btc{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")])

;; Allow Nocona to avoid these instructions if a register is available.

(define_peephole2
  [(match_scratch:DI 2 "r")
   (parallel [(set (zero_extract:DI
		     (match_operand:DI 0 "register_operand" "")
		     (const_int 1)
		     (match_operand:DI 1 "const_0_to_63_operand" ""))
		   (const_int 1))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_64BIT && !TARGET_USE_BT"
  [(const_int 0)]
{
  HOST_WIDE_INT i = INTVAL (operands[1]), hi, lo;
  rtx op1;

  if (HOST_BITS_PER_WIDE_INT >= 64)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else if (i < HOST_BITS_PER_WIDE_INT)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else
    lo = 0, hi = (HOST_WIDE_INT)1 << (i - HOST_BITS_PER_WIDE_INT);

  op1 = immed_double_const (lo, hi, DImode);
  if (i >= 31)
    {
      emit_move_insn (operands[2], op1);
      op1 = operands[2];
    }

  emit_insn (gen_iordi3 (operands[0], operands[0], op1));
  DONE;
})

(define_peephole2
  [(match_scratch:DI 2 "r")
   (parallel [(set (zero_extract:DI
		     (match_operand:DI 0 "register_operand" "")
		     (const_int 1)
		     (match_operand:DI 1 "const_0_to_63_operand" ""))
		   (const_int 0))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_64BIT && !TARGET_USE_BT"
  [(const_int 0)]
{
  HOST_WIDE_INT i = INTVAL (operands[1]), hi, lo;
  rtx op1;

  if (HOST_BITS_PER_WIDE_INT >= 64)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else if (i < HOST_BITS_PER_WIDE_INT)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else
    lo = 0, hi = (HOST_WIDE_INT)1 << (i - HOST_BITS_PER_WIDE_INT);

  op1 = immed_double_const (~lo, ~hi, DImode);
  if (i >= 32)
    {
      emit_move_insn (operands[2], op1);
      op1 = operands[2];
    }

  emit_insn (gen_anddi3 (operands[0], operands[0], op1));
  DONE;
})

(define_peephole2
  [(match_scratch:DI 2 "r")
   (parallel [(set (zero_extract:DI
		     (match_operand:DI 0 "register_operand" "")
		     (const_int 1)
		     (match_operand:DI 1 "const_0_to_63_operand" ""))
	      (not:DI (zero_extract:DI
			(match_dup 0) (const_int 1) (match_dup 1))))
	      (clobber (reg:CC FLAGS_REG))])]
  "TARGET_64BIT && !TARGET_USE_BT"
  [(const_int 0)]
{
  HOST_WIDE_INT i = INTVAL (operands[1]), hi, lo;
  rtx op1;

  if (HOST_BITS_PER_WIDE_INT >= 64)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else if (i < HOST_BITS_PER_WIDE_INT)
    lo = (HOST_WIDE_INT)1 << i, hi = 0;
  else
    lo = 0, hi = (HOST_WIDE_INT)1 << (i - HOST_BITS_PER_WIDE_INT);

  op1 = immed_double_const (lo, hi, DImode);
  if (i >= 31)
    {
      emit_move_insn (operands[2], op1);
      op1 = operands[2];
    }

  emit_insn (gen_xordi3 (operands[0], operands[0], op1));
  DONE;
})

(define_insn "*bt<mode>"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (zero_extract:SWI48
	    (match_operand:SWI48 0 "register_operand" "r")
	    (const_int 1)
	    (match_operand:SWI48 1 "nonmemory_operand" "rN"))
	  (const_int 0)))]
  "TARGET_USE_BT || optimize_function_for_size_p (cfun)"
  "bt{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "<MODE>")])

;; Store-flag instructions.

;; For all sCOND expanders, also expand the compare or test insn that
;; generates cc0.  Generate an equality comparison if `seq' or `sne'.

(define_insn_and_split "*setcc_di_1"
  [(set (match_operand:DI 0 "register_operand" "=q")
	(match_operator:DI 1 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))]
  "TARGET_64BIT && !TARGET_PARTIAL_REG_STALL"
  "#"
  "&& reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (zero_extend:DI (match_dup 2)))]
{
  PUT_MODE (operands[1], QImode);
  operands[2] = gen_lowpart (QImode, operands[0]);
})

(define_insn_and_split "*setcc_si_1_and"
  [(set (match_operand:SI 0 "register_operand" "=q")
	(match_operator:SI 1 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_PARTIAL_REG_STALL
   && TARGET_ZERO_EXTEND_WITH_AND && optimize_function_for_speed_p (cfun)"
  "#"
  "&& reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (parallel [(set (match_dup 0) (zero_extend:SI (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])]
{
  PUT_MODE (operands[1], QImode);
  operands[2] = gen_lowpart (QImode, operands[0]);
})

(define_insn_and_split "*setcc_si_1_movzbl"
  [(set (match_operand:SI 0 "register_operand" "=q")
	(match_operator:SI 1 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))]
  "!TARGET_PARTIAL_REG_STALL
   && (!TARGET_ZERO_EXTEND_WITH_AND || optimize_function_for_size_p (cfun))"
  "#"
  "&& reload_completed"
  [(set (match_dup 2) (match_dup 1))
   (set (match_dup 0) (zero_extend:SI (match_dup 2)))]
{
  PUT_MODE (operands[1], QImode);
  operands[2] = gen_lowpart (QImode, operands[0]);
})

(define_insn "*setcc_qi"
  [(set (match_operand:QI 0 "nonimmediate_operand" "=qm")
	(match_operator:QI 1 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))]
  ""
  "set%C1\t%0"
  [(set_attr "type" "setcc")
   (set_attr "mode" "QI")])

(define_insn "*setcc_qi_slp"
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" "+qm"))
	(match_operator:QI 1 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))]
  ""
  "set%C1\t%0"
  [(set_attr "type" "setcc")
   (set_attr "mode" "QI")])

;; In general it is not safe to assume too much about CCmode registers,
;; so simplify-rtx stops when it sees a second one.  Under certain
;; conditions this is safe on x86, so help combine not create
;;
;;	seta	%al
;;	testb	%al, %al
;;	sete	%al

(define_split
  [(set (match_operand:QI 0 "nonimmediate_operand" "")
	(ne:QI (match_operator 1 "ix86_comparison_operator"
	         [(reg FLAGS_REG) (const_int 0)])
	    (const_int 0)))]
  ""
  [(set (match_dup 0) (match_dup 1))]
  "PUT_MODE (operands[1], QImode);")

(define_split
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" ""))
	(ne:QI (match_operator 1 "ix86_comparison_operator"
	         [(reg FLAGS_REG) (const_int 0)])
	    (const_int 0)))]
  ""
  [(set (match_dup 0) (match_dup 1))]
  "PUT_MODE (operands[1], QImode);")

(define_split
  [(set (match_operand:QI 0 "nonimmediate_operand" "")
	(eq:QI (match_operator 1 "ix86_comparison_operator"
	         [(reg FLAGS_REG) (const_int 0)])
	    (const_int 0)))]
  ""
  [(set (match_dup 0) (match_dup 1))]
{
  rtx new_op1 = copy_rtx (operands[1]);
  operands[1] = new_op1;
  PUT_MODE (new_op1, QImode);
  PUT_CODE (new_op1, ix86_reverse_condition (GET_CODE (new_op1),
					     GET_MODE (XEXP (new_op1, 0))));

  /* Make sure that (a) the CCmode we have for the flags is strong
     enough for the reversed compare or (b) we have a valid FP compare.  */
  if (! ix86_comparison_operator (new_op1, VOIDmode))
    FAIL;
})

(define_split
  [(set (strict_low_part (match_operand:QI 0 "nonimmediate_operand" ""))
	(eq:QI (match_operator 1 "ix86_comparison_operator"
	         [(reg FLAGS_REG) (const_int 0)])
	    (const_int 0)))]
  ""
  [(set (match_dup 0) (match_dup 1))]
{
  rtx new_op1 = copy_rtx (operands[1]);
  operands[1] = new_op1;
  PUT_MODE (new_op1, QImode);
  PUT_CODE (new_op1, ix86_reverse_condition (GET_CODE (new_op1),
					     GET_MODE (XEXP (new_op1, 0))));

  /* Make sure that (a) the CCmode we have for the flags is strong
     enough for the reversed compare or (b) we have a valid FP compare.  */
  if (! ix86_comparison_operator (new_op1, VOIDmode))
    FAIL;
})

;; The SSE store flag instructions saves 0 or 0xffffffff to the result.
;; subsequent logical operations are used to imitate conditional moves.
;; 0xffffffff is NaN, but not in normalized form, so we can't represent
;; it directly.

(define_insn "*avx_setcc<mode>"
  [(set (match_operand:MODEF 0 "register_operand" "=x")
	(match_operator:MODEF 1 "avx_comparison_float_operator"
	  [(match_operand:MODEF 2 "register_operand" "x")
	   (match_operand:MODEF 3 "nonimmediate_operand" "xm")]))]
  "TARGET_AVX"
  "vcmp%D1s<ssemodefsuffix>\t{%3, %2, %0|%0, %2, %3}"
  [(set_attr "type" "ssecmp")
   (set_attr "prefix" "vex")
   (set_attr "length_immediate" "1")
   (set_attr "mode" "<MODE>")])

(define_insn "*sse_setcc<mode>"
  [(set (match_operand:MODEF 0 "register_operand" "=x")
	(match_operator:MODEF 1 "sse_comparison_operator"
	  [(match_operand:MODEF 2 "register_operand" "0")
	   (match_operand:MODEF 3 "nonimmediate_operand" "xm")]))]
  "SSE_FLOAT_MODE_P (<MODE>mode)"
  "cmp%D1s<ssemodefsuffix>\t{%3, %0|%0, %3}"
  [(set_attr "type" "ssecmp")
   (set_attr "length_immediate" "1")
   (set_attr "mode" "<MODE>")])

;; Basic conditional jump instructions.
;; We ignore the overflow flag for signed branch instructions.

(define_insn "*jcc_1"
  [(set (pc)
	(if_then_else (match_operator 1 "ix86_comparison_operator"
				      [(reg FLAGS_REG) (const_int 0)])
		      (label_ref (match_operand 0 "" ""))
		      (pc)))]
  ""
  "%+j%C1\t%l0"
  [(set_attr "type" "ibr")
   (set_attr "modrm" "0")
   (set (attr "length")
	   (if_then_else (and (ge (minus (match_dup 0) (pc))
				  (const_int -126))
			      (lt (minus (match_dup 0) (pc))
				  (const_int 128)))
	     (const_int 2)
	     (const_int 6)))])

(define_insn "*jcc_2"
  [(set (pc)
	(if_then_else (match_operator 1 "ix86_comparison_operator"
				      [(reg FLAGS_REG) (const_int 0)])
		      (pc)
		      (label_ref (match_operand 0 "" ""))))]
  ""
  "%+j%c1\t%l0"
  [(set_attr "type" "ibr")
   (set_attr "modrm" "0")
   (set (attr "length")
	   (if_then_else (and (ge (minus (match_dup 0) (pc))
				  (const_int -126))
			      (lt (minus (match_dup 0) (pc))
				  (const_int 128)))
	     (const_int 2)
	     (const_int 6)))])

;; In general it is not safe to assume too much about CCmode registers,
;; so simplify-rtx stops when it sees a second one.  Under certain
;; conditions this is safe on x86, so help combine not create
;;
;;	seta	%al
;;	testb	%al, %al
;;	je	Lfoo

(define_split
  [(set (pc)
	(if_then_else (ne (match_operator 0 "ix86_comparison_operator"
				      [(reg FLAGS_REG) (const_int 0)])
			  (const_int 0))
		      (label_ref (match_operand 1 "" ""))
		      (pc)))]
  ""
  [(set (pc)
	(if_then_else (match_dup 0)
		      (label_ref (match_dup 1))
		      (pc)))]
  "PUT_MODE (operands[0], VOIDmode);")

(define_split
  [(set (pc)
	(if_then_else (eq (match_operator 0 "ix86_comparison_operator"
				      [(reg FLAGS_REG) (const_int 0)])
			  (const_int 0))
		      (label_ref (match_operand 1 "" ""))
		      (pc)))]
  ""
  [(set (pc)
	(if_then_else (match_dup 0)
		      (label_ref (match_dup 1))
		      (pc)))]
{
  rtx new_op0 = copy_rtx (operands[0]);
  operands[0] = new_op0;
  PUT_MODE (new_op0, VOIDmode);
  PUT_CODE (new_op0, ix86_reverse_condition (GET_CODE (new_op0),
					     GET_MODE (XEXP (new_op0, 0))));

  /* Make sure that (a) the CCmode we have for the flags is strong
     enough for the reversed compare or (b) we have a valid FP compare.  */
  if (! ix86_comparison_operator (new_op0, VOIDmode))
    FAIL;
})

;; zero_extend in SImode is correct also for DImode, since this is what combine
;; pass generates from shift insn with QImode operand.  Actually, the mode
;; of operand 2 (bit offset operand) doesn't matter since bt insn takes
;; appropriate modulo of the bit offset value.

(define_insn_and_split "*jcc_bt<mode>"
  [(set (pc)
  	(if_then_else (match_operator 0 "bt_comparison_operator"
			[(zero_extract:SWI48
			   (match_operand:SWI48 1 "register_operand" "r")
			   (const_int 1)
			   (zero_extend:SI
			     (match_operand:QI 2 "register_operand" "r")))
			 (const_int 0)])
		      (label_ref (match_operand 3 "" ""))
		      (pc)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_USE_BT || optimize_function_for_size_p (cfun)"
  "#"
  "&& 1"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (zero_extract:SWI48
	    (match_dup 1)
	    (const_int 1)
	    (match_dup 2))
	  (const_int 0)))
   (set (pc)
	(if_then_else (match_op_dup 0 [(reg:CCC FLAGS_REG) (const_int 0)])
		      (label_ref (match_dup 3))
		      (pc)))]
{
  operands[2] = simplify_gen_subreg (<MODE>mode, operands[2], QImode, 0);

  PUT_CODE (operands[0], reverse_condition (GET_CODE (operands[0])));
})

;; Avoid useless masking of bit offset operand.  "and" in SImode is correct
;; also for DImode, this is what combine produces.
(define_insn_and_split "*jcc_bt<mode>_mask"
  [(set (pc)
  	(if_then_else (match_operator 0 "bt_comparison_operator"
			[(zero_extract:SWI48
			   (match_operand:SWI48 1 "register_operand" "r")
			   (const_int 1)
			   (and:SI
			     (match_operand:SI 2 "register_operand" "r")
			     (match_operand:SI 3 "const_int_operand" "n")))])
		      (label_ref (match_operand 4 "" ""))
		      (pc)))
   (clobber (reg:CC FLAGS_REG))]
  "(TARGET_USE_BT || optimize_function_for_size_p (cfun))
   && (INTVAL (operands[3]) & (GET_MODE_BITSIZE (<MODE>mode)-1))
      == GET_MODE_BITSIZE (<MODE>mode)-1"
  "#"
  "&& 1"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (zero_extract:SWI48
	    (match_dup 1)
	    (const_int 1)
	    (match_dup 2))
	  (const_int 0)))
   (set (pc)
	(if_then_else (match_op_dup 0 [(reg:CCC FLAGS_REG) (const_int 0)])
		      (label_ref (match_dup 4))
		      (pc)))]
{
  operands[2] = simplify_gen_subreg (<MODE>mode, operands[2], SImode, 0);

  PUT_CODE (operands[0], reverse_condition (GET_CODE (operands[0])));
})

(define_insn_and_split "*jcc_btsi_1"
  [(set (pc)
  	(if_then_else (match_operator 0 "bt_comparison_operator"
			[(and:SI
			   (lshiftrt:SI
			     (match_operand:SI 1 "register_operand" "r")
			     (match_operand:QI 2 "register_operand" "r"))
			   (const_int 1))
			 (const_int 0)])
		      (label_ref (match_operand 3 "" ""))
		      (pc)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_USE_BT || optimize_function_for_size_p (cfun)"
  "#"
  "&& 1"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (zero_extract:SI
	    (match_dup 1)
	    (const_int 1)
	    (match_dup 2))
	  (const_int 0)))
   (set (pc)
	(if_then_else (match_op_dup 0 [(reg:CCC FLAGS_REG) (const_int 0)])
		      (label_ref (match_dup 3))
		      (pc)))]
{
  operands[2] = simplify_gen_subreg (SImode, operands[2], QImode, 0);

  PUT_CODE (operands[0], reverse_condition (GET_CODE (operands[0])));
})

;; avoid useless masking of bit offset operand
(define_insn_and_split "*jcc_btsi_mask_1"
  [(set (pc)
  	(if_then_else
	  (match_operator 0 "bt_comparison_operator"
	    [(and:SI
	       (lshiftrt:SI
		 (match_operand:SI 1 "register_operand" "r")
		 (subreg:QI
		   (and:SI
		     (match_operand:SI 2 "register_operand" "r")
		     (match_operand:SI 3 "const_int_operand" "n")) 0))
	       (const_int 1))
	     (const_int 0)])
	  (label_ref (match_operand 4 "" ""))
	  (pc)))
   (clobber (reg:CC FLAGS_REG))]
  "(TARGET_USE_BT || optimize_function_for_size_p (cfun))
   && (INTVAL (operands[3]) & 0x1f) == 0x1f"
  "#"
  "&& 1"
  [(set (reg:CCC FLAGS_REG)
	(compare:CCC
	  (zero_extract:SI
	    (match_dup 1)
	    (const_int 1)
	    (match_dup 2))
	  (const_int 0)))
   (set (pc)
	(if_then_else (match_op_dup 0 [(reg:CCC FLAGS_REG) (const_int 0)])
		      (label_ref (match_dup 4))
		      (pc)))]
  "PUT_CODE (operands[0], reverse_condition (GET_CODE (operands[0])));")

;; Define combination compare-and-branch fp compare instructions to help
;; combine.

(define_insn "*fp_jcc_1_387"
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "f")
			 (match_operand 2 "nonimmediate_operand" "fm")])
	  (label_ref (match_operand 3 "" ""))
	  (pc)))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 4 "=a"))]
  "TARGET_80387
   && (GET_MODE (operands[1]) == SFmode || GET_MODE (operands[1]) == DFmode)
   && GET_MODE (operands[1]) == GET_MODE (operands[2])
   && SELECT_CC_MODE (GET_CODE (operands[0]),
		      operands[1], operands[2]) == CCFPmode
   && !TARGET_CMOVE"
  "#")

(define_insn "*fp_jcc_1r_387"
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "f")
			 (match_operand 2 "nonimmediate_operand" "fm")])
	  (pc)
	  (label_ref (match_operand 3 "" ""))))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 4 "=a"))]
  "TARGET_80387
   && (GET_MODE (operands[1]) == SFmode || GET_MODE (operands[1]) == DFmode)
   && GET_MODE (operands[1]) == GET_MODE (operands[2])
   && SELECT_CC_MODE (GET_CODE (operands[0]),
		      operands[1], operands[2]) == CCFPmode
   && !TARGET_CMOVE"
  "#")

(define_insn "*fp_jcc_2_387"
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "f")
			 (match_operand 2 "register_operand" "f")])
	  (label_ref (match_operand 3 "" ""))
	  (pc)))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 4 "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && GET_MODE (operands[1]) == GET_MODE (operands[2])
   && !TARGET_CMOVE"
  "#")

(define_insn "*fp_jcc_2r_387"
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "f")
			 (match_operand 2 "register_operand" "f")])
	  (pc)
	  (label_ref (match_operand 3 "" ""))))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 4 "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && GET_MODE (operands[1]) == GET_MODE (operands[2])
   && !TARGET_CMOVE"
  "#")

(define_insn "*fp_jcc_3_387"
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "f")
			 (match_operand 2 "const0_operand" "")])
	  (label_ref (match_operand 3 "" ""))
	  (pc)))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 4 "=a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[1]))
   && GET_MODE (operands[1]) == GET_MODE (operands[2])
   && SELECT_CC_MODE (GET_CODE (operands[0]),
		      operands[1], operands[2]) == CCFPmode
   && !TARGET_CMOVE"
  "#")

(define_split
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "")
			 (match_operand 2 "nonimmediate_operand" "")])
	  (match_operand 3 "" "")
	  (match_operand 4 "" "")))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))]
  "reload_completed"
  [(const_int 0)]
{
  ix86_split_fp_branch (GET_CODE (operands[0]), operands[1], operands[2],
	                operands[3], operands[4], NULL_RTX, NULL_RTX);
  DONE;
})

(define_split
  [(set (pc)
	(if_then_else (match_operator 0 "ix86_fp_comparison_operator"
			[(match_operand 1 "register_operand" "")
			 (match_operand 2 "general_operand" "")])
	  (match_operand 3 "" "")
	  (match_operand 4 "" "")))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 5 "=a"))]
  "reload_completed"
  [(const_int 0)]
{
  ix86_split_fp_branch (GET_CODE (operands[0]), operands[1], operands[2],
	     		operands[3], operands[4], operands[5], NULL_RTX);
  DONE;
})

;; The order of operands in *fp_jcc_4_387 is forced by combine in
;; simplify_comparison () function. Float operator is treated as RTX_OBJ
;; with a precedence over other operators and is always put in the first
;; place. Swap condition and operands to match ficom instruction.

(define_insn "*fp_jcc_4_<mode>_387"
  [(set (pc)
	(if_then_else
	  (match_operator 0 "ix86_swapped_fp_comparison_operator"
	    [(match_operator 1 "float_operator"
	      [(match_operand:X87MODEI12 2 "nonimmediate_operand" "m,?r")])
	     (match_operand 3 "register_operand" "f,f")])
	  (label_ref (match_operand 4 "" ""))
	  (pc)))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 5 "=a,a"))]
  "X87_FLOAT_MODE_P (GET_MODE (operands[3]))
   && (TARGET_USE_<MODE>MODE_FIOP || optimize_function_for_size_p (cfun))
   && GET_MODE (operands[1]) == GET_MODE (operands[3])
   && ix86_fp_compare_mode (swap_condition (GET_CODE (operands[0]))) == CCFPmode
   && !TARGET_CMOVE"
  "#")

(define_split
  [(set (pc)
	(if_then_else
	  (match_operator 0 "ix86_swapped_fp_comparison_operator"
	    [(match_operator 1 "float_operator"
	      [(match_operand:X87MODEI12 2 "memory_operand" "")])
	     (match_operand 3 "register_operand" "")])
	  (match_operand 4 "" "")
	  (match_operand 5 "" "")))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 6 "=a"))]
  "reload_completed"
  [(const_int 0)]
{
  operands[7] = gen_rtx_FLOAT (GET_MODE (operands[1]), operands[2]);

  ix86_split_fp_branch (swap_condition (GET_CODE (operands[0])),
			operands[3], operands[7],
			operands[4], operands[5], operands[6], NULL_RTX);
  DONE;
})

;; %%% Kill this when reload knows how to do it.
(define_split
  [(set (pc)
	(if_then_else
	  (match_operator 0 "ix86_swapped_fp_comparison_operator"
	    [(match_operator 1 "float_operator"
	      [(match_operand:X87MODEI12 2 "register_operand" "")])
	     (match_operand 3 "register_operand" "")])
	  (match_operand 4 "" "")
	  (match_operand 5 "" "")))
   (clobber (reg:CCFP FPSR_REG))
   (clobber (reg:CCFP FLAGS_REG))
   (clobber (match_scratch:HI 6 "=a"))]
  "reload_completed"
  [(const_int 0)]
{
  operands[7] = ix86_force_to_memory (GET_MODE (operands[2]), operands[2]);
  operands[7] = gen_rtx_FLOAT (GET_MODE (operands[1]), operands[7]);

  ix86_split_fp_branch (swap_condition (GET_CODE (operands[0])),
			operands[3], operands[7],
			operands[4], operands[5], operands[6], operands[2]);
  DONE;
})

;; Unconditional and other jump instructions

(define_insn "jump"
  [(set (pc)
	(label_ref (match_operand 0 "" "")))]
  ""
  "jmp\t%l0"
  [(set_attr "type" "ibr")
   (set (attr "length")
	   (if_then_else (and (ge (minus (match_dup 0) (pc))
				  (const_int -126))
			      (lt (minus (match_dup 0) (pc))
				  (const_int 128)))
	     (const_int 2)
	     (const_int 5)))
   (set_attr "modrm" "0")])

(define_expand "indirect_jump"
  [(set (pc) (match_operand 0 "nonimmediate_operand" ""))]
  ""
  "")

(define_insn "*indirect_jump"
  [(set (pc) (match_operand:P 0 "nonimmediate_operand" "rm"))]
  ""
  "jmp\t%A0"
  [(set_attr "type" "ibr")
   (set_attr "length_immediate" "0")])

(define_expand "tablejump"
  [(parallel [(set (pc) (match_operand 0 "nonimmediate_operand" ""))
	      (use (label_ref (match_operand 1 "" "")))])]
  ""
{
  /* In PIC mode, the table entries are stored GOT (32-bit) or PC (64-bit)
     relative.  Convert the relative address to an absolute address.  */
  if (flag_pic)
    {
      rtx op0, op1;
      enum rtx_code code;

      /* We can't use @GOTOFF for text labels on VxWorks;
	 see gotoff_operand.  */
      if (TARGET_64BIT || TARGET_VXWORKS_RTP)
	{
	  code = PLUS;
	  op0 = operands[0];
	  op1 = gen_rtx_LABEL_REF (Pmode, operands[1]);
	}
      else if (TARGET_MACHO || HAVE_AS_GOTOFF_IN_DATA)
	{
	  code = PLUS;
	  op0 = operands[0];
	  op1 = pic_offset_table_rtx;
	}
      else
	{
	  code = MINUS;
	  op0 = pic_offset_table_rtx;
	  op1 = operands[0];
	}

      operands[0] = expand_simple_binop (Pmode, code, op0, op1, NULL_RTX, 0,
					 OPTAB_DIRECT);
    }
})

(define_insn "*tablejump_1"
  [(set (pc) (match_operand:P 0 "nonimmediate_operand" "rm"))
   (use (label_ref (match_operand 1 "" "")))]
  ""
  "jmp\t%A0"
  [(set_attr "type" "ibr")
   (set_attr "length_immediate" "0")])

;; Convert setcc + movzbl to xor + setcc if operands don't overlap.

(define_peephole2
  [(set (reg FLAGS_REG) (match_operand 0 "" ""))
   (set (match_operand:QI 1 "register_operand" "")
	(match_operator:QI 2 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))
   (set (match_operand 3 "q_regs_operand" "")
	(zero_extend (match_dup 1)))]
  "(peep2_reg_dead_p (3, operands[1])
    || operands_match_p (operands[1], operands[3]))
   && ! reg_overlap_mentioned_p (operands[3], operands[0])"
  [(set (match_dup 4) (match_dup 0))
   (set (strict_low_part (match_dup 5))
	(match_dup 2))]
{
  operands[4] = gen_rtx_REG (GET_MODE (operands[0]), FLAGS_REG);
  operands[5] = gen_lowpart (QImode, operands[3]);
  ix86_expand_clear (operands[3]);
})

;; Similar, but match zero_extendhisi2_and, which adds a clobber.

(define_peephole2
  [(set (reg FLAGS_REG) (match_operand 0 "" ""))
   (set (match_operand:QI 1 "register_operand" "")
	(match_operator:QI 2 "ix86_comparison_operator"
	  [(reg FLAGS_REG) (const_int 0)]))
   (parallel [(set (match_operand 3 "q_regs_operand" "")
		   (zero_extend (match_dup 1)))
	      (clobber (reg:CC FLAGS_REG))])]
  "(peep2_reg_dead_p (3, operands[1])
    || operands_match_p (operands[1], operands[3]))
   && ! reg_overlap_mentioned_p (operands[3], operands[0])"
  [(set (match_dup 4) (match_dup 0))
   (set (strict_low_part (match_dup 5))
	(match_dup 2))]
{
  operands[4] = gen_rtx_REG (GET_MODE (operands[0]), FLAGS_REG);
  operands[5] = gen_lowpart (QImode, operands[3]);
  ix86_expand_clear (operands[3]);
})

;; Call instructions.

;; The predicates normally associated with named expanders are not properly
;; checked for calls.  This is a bug in the generic code, but it isn't that
;; easy to fix.  Ignore it for now and be prepared to fix things up.

;; P6 processors will jump to the address after the decrement when %esp
;; is used as a call operand, so they will execute return address as a code.
;; See Pentium Pro errata 70, Pentium 2 errata A33 and Pentium 3 errata E17.
 
;; Call subroutine returning no value.

(define_expand "call_pop"
  [(parallel [(call (match_operand:QI 0 "" "")
		    (match_operand:SI 1 "" ""))
	      (set (reg:SI SP_REG)
		   (plus:SI (reg:SI SP_REG)
			    (match_operand:SI 3 "" "")))])]
  "!TARGET_64BIT"
{
  ix86_expand_call (NULL, operands[0], operands[1],
		    operands[2], operands[3], 0);
  DONE;
})

(define_insn_and_split "*call_pop_0_vzeroupper"
  [(parallel
    [(call (mem:QI (match_operand:SI 0 "constant_call_address_operand" ""))
	   (match_operand:SI 1 "" ""))
     (set (reg:SI SP_REG)
	  (plus:SI (reg:SI SP_REG)
		   (match_operand:SI 2 "immediate_operand" "")))])
   (unspec [(match_operand 3 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && !TARGET_64BIT"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[3]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_pop_0"
  [(call (mem:QI (match_operand:SI 0 "constant_call_address_operand" ""))
	 (match_operand:SI 1 "" ""))
   (set (reg:SI SP_REG)
	(plus:SI (reg:SI SP_REG)
		 (match_operand:SI 2 "immediate_operand" "")))]
  "!TARGET_64BIT"
{
  if (SIBLING_CALL_P (insn))
    return "jmp\t%P0";
  else
    return "call\t%P0";
}
  [(set_attr "type" "call")])

(define_insn_and_split "*call_pop_1_vzeroupper"
  [(parallel
    [(call (mem:QI (match_operand:SI 0 "call_insn_operand" "lsm"))
	   (match_operand:SI 1 "" ""))
     (set (reg:SI SP_REG)
	  (plus:SI (reg:SI SP_REG)
		   (match_operand:SI 2 "immediate_operand" "i")))])
   (unspec [(match_operand 3 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && !TARGET_64BIT && !SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[3]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_pop_1"
  [(call (mem:QI (match_operand:SI 0 "call_insn_operand" "lsm"))
	 (match_operand:SI 1 "" ""))
   (set (reg:SI SP_REG)
	(plus:SI (reg:SI SP_REG)
		 (match_operand:SI 2 "immediate_operand" "i")))]
  "!TARGET_64BIT && !SIBLING_CALL_P (insn)"
{
  if (constant_call_address_operand (operands[0], Pmode))
    return "call\t%P0";
  return "call\t%A0";
}
  [(set_attr "type" "call")])

(define_insn_and_split "*sibcall_pop_1_vzeroupper"
 [(parallel
   [(call (mem:QI (match_operand:SI 0 "sibcall_insn_operand" "s,U"))
	   (match_operand:SI 1 "" ""))
     (set (reg:SI SP_REG)
	  (plus:SI (reg:SI SP_REG)
		   (match_operand:SI 2 "immediate_operand" "i,i")))])
   (unspec [(match_operand 3 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && !TARGET_64BIT && SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[3]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*sibcall_pop_1"
  [(call (mem:QI (match_operand:SI 0 "sibcall_insn_operand" "s,U"))
	 (match_operand:SI 1 "" ""))
   (set (reg:SI SP_REG)
	(plus:SI (reg:SI SP_REG)
		 (match_operand:SI 2 "immediate_operand" "i,i")))]
  "!TARGET_64BIT && SIBLING_CALL_P (insn)"
  "@
   jmp\t%P0
   jmp\t%A0"
  [(set_attr "type" "call")])

(define_expand "call"
  [(call (match_operand:QI 0 "" "")
	 (match_operand 1 "" ""))
   (use (match_operand 2 "" ""))]
  ""
{
  ix86_expand_call (NULL, operands[0], operands[1], operands[2], NULL, 0);
  DONE;
})

(define_expand "sibcall"
  [(call (match_operand:QI 0 "" "")
	 (match_operand 1 "" ""))
   (use (match_operand 2 "" ""))]
  ""
{
  ix86_expand_call (NULL, operands[0], operands[1], operands[2], NULL, 1);
  DONE;
})

(define_insn_and_split "*call_0_vzeroupper"
  [(call (mem:QI (match_operand 0 "constant_call_address_operand" ""))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_0"
  [(call (mem:QI (match_operand 0 "constant_call_address_operand" ""))
	 (match_operand 1 "" ""))]
  ""
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*call_1_vzeroupper"
  [(call (mem:QI (match_operand:SI 0 "call_insn_operand" "lsm"))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && !TARGET_64BIT && !SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_1"
  [(call (mem:QI (match_operand:SI 0 "call_insn_operand" "lsm"))
	 (match_operand 1 "" ""))]
  "!TARGET_64BIT && !SIBLING_CALL_P (insn)"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*sibcall_1_vzeroupper"
  [(call (mem:QI (match_operand:SI 0 "sibcall_insn_operand" "s,U"))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && !TARGET_64BIT && SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*sibcall_1"
  [(call (mem:QI (match_operand:SI 0 "sibcall_insn_operand" "s,U"))
	 (match_operand 1 "" ""))]
  "!TARGET_64BIT && SIBLING_CALL_P (insn)"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*call_1_rex64_vzeroupper"
  [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rsm"))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && TARGET_64BIT && !SIBLING_CALL_P (insn)
   && ix86_cmodel != CM_LARGE && ix86_cmodel != CM_LARGE_PIC"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_1_rex64"
  [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rsm"))
	 (match_operand 1 "" ""))]
  "TARGET_64BIT && !SIBLING_CALL_P (insn)
   && ix86_cmodel != CM_LARGE && ix86_cmodel != CM_LARGE_PIC"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*call_1_rex64_ms_sysv_vzeroupper"
  [(parallel
    [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rsm"))
	   (match_operand 1 "" ""))
     (unspec [(const_int 0)] UNSPEC_MS_TO_SYSV_CALL)
     (clobber (reg:TI XMM6_REG))
     (clobber (reg:TI XMM7_REG))
     (clobber (reg:TI XMM8_REG))
     (clobber (reg:TI XMM9_REG))
     (clobber (reg:TI XMM10_REG))
     (clobber (reg:TI XMM11_REG))
     (clobber (reg:TI XMM12_REG))
     (clobber (reg:TI XMM13_REG))
     (clobber (reg:TI XMM14_REG))
     (clobber (reg:TI XMM15_REG))
     (clobber (reg:DI SI_REG))
     (clobber (reg:DI DI_REG))])
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && TARGET_64BIT && !SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_1_rex64_ms_sysv"
  [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rsm"))
	 (match_operand 1 "" ""))
   (unspec [(const_int 0)] UNSPEC_MS_TO_SYSV_CALL)
   (clobber (reg:TI XMM6_REG))
   (clobber (reg:TI XMM7_REG))
   (clobber (reg:TI XMM8_REG))
   (clobber (reg:TI XMM9_REG))
   (clobber (reg:TI XMM10_REG))
   (clobber (reg:TI XMM11_REG))
   (clobber (reg:TI XMM12_REG))
   (clobber (reg:TI XMM13_REG))
   (clobber (reg:TI XMM14_REG))
   (clobber (reg:TI XMM15_REG))
   (clobber (reg:DI SI_REG))
   (clobber (reg:DI DI_REG))]
  "TARGET_64BIT && !SIBLING_CALL_P (insn)"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*call_1_rex64_large_vzeroupper"
  [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rm"))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && TARGET_64BIT && !SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*call_1_rex64_large"
  [(call (mem:QI (match_operand:DI 0 "call_insn_operand" "rm"))
	 (match_operand 1 "" ""))]
  "TARGET_64BIT && !SIBLING_CALL_P (insn)"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

(define_insn_and_split "*sibcall_1_rex64_vzeroupper"
  [(call (mem:QI (match_operand:DI 0 "sibcall_insn_operand" "s,U"))
	 (match_operand 1 "" ""))
   (unspec [(match_operand 2 "const_int_operand" "")]
   	   UNSPEC_CALL_NEEDS_VZEROUPPER)]
  "TARGET_VZEROUPPER && TARGET_64BIT && SIBLING_CALL_P (insn)"
  "#"
  "&& reload_completed"
  [(const_int 0)]
  "ix86_split_call_vzeroupper (curr_insn, operands[2]); DONE;"
  [(set_attr "type" "call")])

(define_insn "*sibcall_1_rex64"
  [(call (mem:QI (match_operand:DI 0 "sibcall_insn_operand" "s,U"))
	 (match_operand 1 "" ""))]
  "TARGET_64BIT && SIBLING_CALL_P (insn)"
  { return ix86_output_call_insn (insn, operands[0], 0); }
  [(set_attr "type" "call")])

;; Call subroutine, returning value in operand 0
(define_expand "call_value_pop"
  [(parallel [(set (match_operand 0 "" "")
		   (call (match_operand:QI 1 "" "")
			 (match_operand:SI 2 "" "")))
	      (set (reg:SI SP_REG)
		   (plus:SI (reg:SI SP_REG)
			    (match_operand:SI 4 "" "")))])]
  "!TARGET_64BIT"
{
  ix86_expand_call (operands[0], operands[1], operands[2],
		    operands[3], operands[4], 0);
  DONE;
})

(define_expand "call_value"
  [(set (match_operand 0 "" "")
	(call (match_operand:QI 1 "" "")
	      (match_operand:SI 2 "" "")))
   (use (match_operand:SI 3 "" ""))]
  ;; Operand 3 is not used on the i386.
  ""
{
  ix86_expand_call (operands[0], operands[1], operands[2],
		    operands[3], NULL, 0);
  DONE;
})

(define_expand "sibcall_value"
  [(set (match_operand 0 "" "")
	(call (match_operand:QI 1 "" "")
	      (match_operand:SI 2 "" "")))
   (use (match_operand:SI 3 "" ""))]
  ;; Operand 3 is not used on the i386.
  ""
{
  ix86_expand_call (operands[0], operands[1], operands[2],
		    operands[3], NULL, 1);
  DONE;
})

(define_expand "sibcall_value_pop"
  [(parallel [(set (match_operand 0 "" "")
		   (call (match_operand:QI 1 "" "")
			 (match_operand:SI 2 "" "")))
	      (set (reg:SI SP_REG)
		   (plus:SI (reg:SI SP_REG)
			    (match_operand:SI 4 "" "")))])]
  "!TARGET_64BIT"
{
  ix86_expand_call (operands[0], operands[1], operands[2],
		    operands[3], operands[4], 1);
  DONE;
})


;; Call subroutine returning any type.

(define_expand "untyped_call"
  [(parallel [(call (match_operand 0 "" "")
		    (const_int 0))
	      (match_operand 1 "" "")
	      (match_operand 2 "" "")])]
  ""
{
  int i;

  /* In order to give reg-stack an easier job in validating two
     coprocessor registers as containing a possible return value,
     simply pretend the untyped call returns a complex long double
     value. 

     We can't use SSE_REGPARM_MAX here since callee is unprototyped
     and should have the default ABI.  */

  ix86_expand_call ((TARGET_FLOAT_RETURNS_IN_80387
		     ? gen_rtx_REG (XCmode, FIRST_FLOAT_REG) : NULL),
		    operands[0], const0_rtx,
		    GEN_INT ((TARGET_64BIT
			      ? (ix86_abi == SYSV_ABI
				 ? X86_64_SSE_REGPARM_MAX
				 : X86_64_MS_SSE_REGPARM_MAX)
			      : X86_32_SSE_REGPARM_MAX)
		    	     - 1),
		    NULL, 0);

  for (i = 0; i < XVECLEN (operands[2], 0); i++)
    {
      rtx set = XVECEXP (operands[2], 0, i);
      emit_move_insn (SET_DEST (set), SET_SRC (set));
    }

  /* The optimizer does not know that the call sets the function value
     registers we stored in the result block.  We avoid problems by
     claiming that all hard registers are used and clobbered at this
     point.  */
  emit_insn (gen_blockage ());

  DONE;
})

;; Prologue and epilogue instructions

;; UNSPEC_VOLATILE is considered to use and clobber all hard registers and
;; all of memory.  This blocks insns from being moved across this point.

(define_insn "blockage"
  [(unspec_volatile [(const_int 0)] UNSPECV_BLOCKAGE)]
  ""
  ""
  [(set_attr "length" "0")])

;; Do not schedule instructions accessing memory across this point.

(define_expand "memory_blockage"
  [(set (match_dup 0)
	(unspec:BLK [(match_dup 0)] UNSPEC_MEMORY_BLOCKAGE))]
  ""
{
  operands[0] = gen_rtx_MEM (BLKmode, gen_rtx_SCRATCH (Pmode));
  MEM_VOLATILE_P (operands[0]) = 1;
})

(define_insn "*memory_blockage"
  [(set (match_operand:BLK 0 "" "")
	(unspec:BLK [(match_dup 0)] UNSPEC_MEMORY_BLOCKAGE))]
  ""
  ""
  [(set_attr "length" "0")])

;; As USE insns aren't meaningful after reload, this is used instead
;; to prevent deleting instructions setting registers for PIC code
(define_insn "prologue_use"
  [(unspec_volatile [(match_operand 0 "" "")] UNSPECV_PROLOGUE_USE)]
  ""
  ""
  [(set_attr "length" "0")])

;; Insn emitted into the body of a function to return from a function.
;; This is only done if the function's epilogue is known to be simple.
;; See comments for ix86_can_use_return_insn_p in i386.c.

(define_expand "return"
  [(return)]
  "ix86_can_use_return_insn_p ()"
{
  if (crtl->args.pops_args)
    {
      rtx popc = GEN_INT (crtl->args.pops_args);
      emit_jump_insn (gen_return_pop_internal (popc));
      DONE;
    }
})

(define_insn "return_internal"
  [(return)]
  "reload_completed"
  "ret"
  [(set_attr "length" "1")
   (set_attr "atom_unit" "jeu")
   (set_attr "length_immediate" "0")
   (set_attr "modrm" "0")])

;; Used by x86_machine_dependent_reorg to avoid penalty on single byte RET
;; instruction Athlon and K8 have.

(define_insn "return_internal_long"
  [(return)
   (unspec [(const_int 0)] UNSPEC_REP)]
  "reload_completed"
  "rep\;ret"
  [(set_attr "length" "2")
   (set_attr "atom_unit" "jeu")
   (set_attr "length_immediate" "0")
   (set_attr "prefix_rep" "1")
   (set_attr "modrm" "0")])

(define_insn "return_pop_internal"
  [(return)
   (use (match_operand:SI 0 "const_int_operand" ""))]
  "reload_completed"
  "ret\t%0"
  [(set_attr "length" "3")
   (set_attr "atom_unit" "jeu")
   (set_attr "length_immediate" "2")
   (set_attr "modrm" "0")])

(define_insn "return_indirect_internal"
  [(return)
   (use (match_operand:SI 0 "register_operand" "r"))]
  "reload_completed"
  "jmp\t%A0"
  [(set_attr "type" "ibr")
   (set_attr "length_immediate" "0")])

(define_insn "nop"
  [(const_int 0)]
  ""
  "nop"
  [(set_attr "length" "1")
   (set_attr "length_immediate" "0")
   (set_attr "modrm" "0")])

;; Generate nops.  Operand 0 is the number of nops, up to 8.
(define_insn "nops"
  [(unspec_volatile [(match_operand 0 "const_int_operand" "")]
		    UNSPECV_NOPS)]
  "reload_completed"
{
  int num = INTVAL (operands[0]);

  gcc_assert (num >= 1 && num <= 8);

  while (num--)
    fputs ("\tnop\n", asm_out_file);

  return "";
}
  [(set (attr "length") (symbol_ref "INTVAL (operands[0])"))
   (set_attr "length_immediate" "0")
   (set_attr "modrm" "0")])

;; Pad to 16-byte boundary, max skip in op0.  Used to avoid
;; branch prediction penalty for the third jump in a 16-byte
;; block on K8.

(define_insn "pad"
  [(unspec_volatile [(match_operand 0 "" "")] UNSPECV_ALIGN)]
  ""
{
#ifdef ASM_OUTPUT_MAX_SKIP_PAD
  ASM_OUTPUT_MAX_SKIP_PAD (asm_out_file, 4, (int)INTVAL (operands[0]));
#else
  /* It is tempting to use ASM_OUTPUT_ALIGN here, but we don't want to do that.
     The align insn is used to avoid 3 jump instructions in the row to improve
     branch prediction and the benefits hardly outweigh the cost of extra 8
     nops on the average inserted by full alignment pseudo operation.  */
#endif
  return "";
}
  [(set_attr "length" "16")])

(define_expand "prologue"
  [(const_int 0)]
  ""
  "ix86_expand_prologue (); DONE;")

(define_insn "set_got"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(unspec:SI [(const_int 0)] UNSPEC_SET_GOT))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT"
  "* return output_set_got (operands[0], NULL_RTX);"
  [(set_attr "type" "multi")
   (set_attr "length" "12")])

(define_insn "set_got_labelled"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(unspec:SI [(label_ref (match_operand 1 "" ""))]
	 UNSPEC_SET_GOT))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_64BIT"
  "* return output_set_got (operands[0], operands[1]);"
  [(set_attr "type" "multi")
   (set_attr "length" "12")])

(define_insn "set_got_rex64"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(unspec:DI [(const_int 0)] UNSPEC_SET_GOT))]
  "TARGET_64BIT"
  "lea{q}\t{_GLOBAL_OFFSET_TABLE_(%%rip), %0|%0, _GLOBAL_OFFSET_TABLE_[rip]}"
  [(set_attr "type" "lea")
   (set_attr "length_address" "4")
   (set_attr "mode" "DI")])

(define_insn "set_rip_rex64"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(unspec:DI [(label_ref (match_operand 1 "" ""))] UNSPEC_SET_RIP))]
  "TARGET_64BIT"
  "lea{q}\t{%l1(%%rip), %0|%0, %l1[rip]}"
  [(set_attr "type" "lea")
   (set_attr "length_address" "4")
   (set_attr "mode" "DI")])

(define_insn "set_got_offset_rex64"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(unspec:DI
	  [(label_ref (match_operand 1 "" ""))]
	  UNSPEC_SET_GOT_OFFSET))]
  "TARGET_64BIT"
  "movabs{q}\t{$_GLOBAL_OFFSET_TABLE_-%l1, %0|%0, OFFSET FLAT:_GLOBAL_OFFSET_TABLE_-%l1}"
  [(set_attr "type" "imov")
   (set_attr "length_immediate" "0")
   (set_attr "length_address" "8")
   (set_attr "mode" "DI")])

(define_expand "epilogue"
  [(const_int 0)]
  ""
  "ix86_expand_epilogue (1); DONE;")

(define_expand "sibcall_epilogue"
  [(const_int 0)]
  ""
  "ix86_expand_epilogue (0); DONE;")

(define_expand "eh_return"
  [(use (match_operand 0 "register_operand" ""))]
  ""
{
  rtx tmp, sa = EH_RETURN_STACKADJ_RTX, ra = operands[0];

  /* Tricky bit: we write the address of the handler to which we will
     be returning into someone else's stack frame, one word below the
     stack address we wish to restore.  */
  tmp = gen_rtx_PLUS (Pmode, arg_pointer_rtx, sa);
  tmp = plus_constant (tmp, -UNITS_PER_WORD);
  tmp = gen_rtx_MEM (Pmode, tmp);
  emit_move_insn (tmp, ra);

  emit_jump_insn (gen_eh_return_internal ());
  emit_barrier ();
  DONE;
})

(define_insn_and_split "eh_return_internal"
  [(eh_return)]
  ""
  "#"
  "epilogue_completed"
  [(const_int 0)]
  "ix86_expand_epilogue (2); DONE;")

(define_insn "leave"
  [(set (reg:SI SP_REG) (plus:SI (reg:SI BP_REG) (const_int 4)))
   (set (reg:SI BP_REG) (mem:SI (reg:SI BP_REG)))
   (clobber (mem:BLK (scratch)))]
  "!TARGET_64BIT"
  "leave"
  [(set_attr "type" "leave")])

(define_insn "leave_rex64"
  [(set (reg:DI SP_REG) (plus:DI (reg:DI BP_REG) (const_int 8)))
   (set (reg:DI BP_REG) (mem:DI (reg:DI BP_REG)))
   (clobber (mem:BLK (scratch)))]
  "TARGET_64BIT"
  "leave"
  [(set_attr "type" "leave")])

;; Handle -fsplit-stack.

(define_expand "split_stack_prologue"
  [(const_int 0)]
  ""
{
  ix86_expand_split_stack_prologue ();
  DONE;
})

;; In order to support the call/return predictor, we use a return
;; instruction which the middle-end doesn't see.
(define_insn "split_stack_return"
  [(unspec_volatile [(match_operand:SI 0 "const_int_operand" "")]
		     UNSPECV_SPLIT_STACK_RETURN)]
  ""
{
  if (operands[0] == const0_rtx)
    return "ret";
  else
    return "ret\t%0";
}
  [(set_attr "atom_unit" "jeu")
   (set_attr "modrm" "0")
   (set (attr "length")
	(if_then_else (match_operand:SI 0 "const0_operand" "")
		      (const_int 1)
		      (const_int 3)))
   (set (attr "length_immediate")
	(if_then_else (match_operand:SI 0 "const0_operand" "")
		      (const_int 0)
		      (const_int 2)))])

;; If there are operand 0 bytes available on the stack, jump to
;; operand 1.

(define_expand "split_stack_space_check"
  [(set (pc) (if_then_else
	      (ltu (minus (reg SP_REG)
			  (match_operand 0 "register_operand" ""))
		   (unspec [(const_int 0)] UNSPEC_STACK_CHECK))
	      (label_ref (match_operand 1 "" ""))
	      (pc)))]
  ""
{
  rtx reg, size, limit;

  reg = gen_reg_rtx (Pmode);
  size = force_reg (Pmode, operands[0]);
  emit_insn (gen_sub3_insn (reg, stack_pointer_rtx, size));
  limit = gen_rtx_UNSPEC (Pmode, gen_rtvec (1, const0_rtx),
			  UNSPEC_STACK_CHECK);
  limit = gen_rtx_MEM (Pmode, gen_rtx_CONST (Pmode, limit));
  ix86_expand_branch (GEU, reg, limit, operands[1]);

  DONE;
})

;; Bit manipulation instructions.

(define_expand "ffs<mode>2"
  [(set (match_dup 2) (const_int -1))
   (parallel [(set (reg:CCZ FLAGS_REG)
		   (compare:CCZ
		     (match_operand:SWI48 1 "nonimmediate_operand" "")
		     (const_int 0)))
	      (set (match_operand:SWI48 0 "register_operand" "")
		   (ctz:SWI48 (match_dup 1)))])
   (set (match_dup 0) (if_then_else:SWI48
			(eq (reg:CCZ FLAGS_REG) (const_int 0))
			(match_dup 2)
			(match_dup 0)))
   (parallel [(set (match_dup 0) (plus:SWI48 (match_dup 0) (const_int 1)))
	      (clobber (reg:CC FLAGS_REG))])]
  ""
{
  if (<MODE>mode == SImode && !TARGET_CMOVE)
    {
      emit_insn (gen_ffssi2_no_cmove (operands[0], operands [1]));
      DONE;
    }
  operands[2] = gen_reg_rtx (<MODE>mode);
})

(define_insn_and_split "ffssi2_no_cmove"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(ffs:SI (match_operand:SI 1 "nonimmediate_operand" "rm")))
   (clobber (match_scratch:SI 2 "=&q"))
   (clobber (reg:CC FLAGS_REG))]
  "!TARGET_CMOVE"
  "#"
  "&& reload_completed"
  [(parallel [(set (reg:CCZ FLAGS_REG)
		   (compare:CCZ (match_dup 1) (const_int 0)))
	      (set (match_dup 0) (ctz:SI (match_dup 1)))])
   (set (strict_low_part (match_dup 3))
	(eq:QI (reg:CCZ FLAGS_REG) (const_int 0)))
   (parallel [(set (match_dup 2) (neg:SI (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])
   (parallel [(set (match_dup 0) (ior:SI (match_dup 0) (match_dup 2)))
	      (clobber (reg:CC FLAGS_REG))])
   (parallel [(set (match_dup 0) (plus:SI (match_dup 0) (const_int 1)))
	      (clobber (reg:CC FLAGS_REG))])]
{
  operands[3] = gen_lowpart (QImode, operands[2]);
  ix86_expand_clear (operands[2]);
})

(define_insn "*ffs<mode>_1"
  [(set (reg:CCZ FLAGS_REG)
	(compare:CCZ (match_operand:SWI48 1 "nonimmediate_operand" "rm")
		     (const_int 0)))
   (set (match_operand:SWI48 0 "register_operand" "=r")
	(ctz:SWI48 (match_dup 1)))]
  ""
  "bsf{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "<MODE>")])

(define_insn "ctz<mode>2"
  [(set (match_operand:SWI248 0 "register_operand" "=r")
	(ctz:SWI248 (match_operand:SWI248 1 "nonimmediate_operand" "rm")))
   (clobber (reg:CC FLAGS_REG))]
  ""
{
  if (TARGET_BMI)
    return "tzcnt{<imodesuffix>}\t{%1, %0|%0, %1}";
  else
    return "bsf{<imodesuffix>}\t{%1, %0|%0, %1}";
}
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set (attr "prefix_rep") (symbol_ref "TARGET_BMI"))
   (set_attr "mode" "<MODE>")])

(define_expand "clz<mode>2"
  [(parallel
     [(set (match_operand:SWI248 0 "register_operand" "")
	   (minus:SWI248
	     (match_dup 2)
	     (clz:SWI248 (match_operand:SWI248 1 "nonimmediate_operand" ""))))
      (clobber (reg:CC FLAGS_REG))])
   (parallel
     [(set (match_dup 0) (xor:SWI248 (match_dup 0) (match_dup 2)))
      (clobber (reg:CC FLAGS_REG))])]
  ""
{
  if (TARGET_ABM)
    {
      emit_insn (gen_clz<mode>2_abm (operands[0], operands[1]));
      DONE;
    }
  operands[2] = GEN_INT (GET_MODE_BITSIZE (<MODE>mode)-1);
})

(define_insn "clz<mode>2_abm"
  [(set (match_operand:SWI248 0 "register_operand" "=r")
	(clz:SWI248 (match_operand:SWI248 1 "nonimmediate_operand" "rm")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_ABM || TARGET_BMI"
  "lzcnt{<imodesuffix>}\t{%1, %0|%0, %1}"
  [(set_attr "prefix_rep" "1")
   (set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

;; BMI instructions.
(define_insn "*bmi_andn_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (not:SWI48
            (match_operand:SWI48 1 "register_operand" "r"))
            (match_operand:SWI48 2 "nonimmediate_operand" "rm")))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_BMI"
  "andn\t{%2, %1, %0|%0, %1, %2}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "bmi_bextr_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (unspec:SWI48 [(match_operand:SWI48 1 "nonimmediate_operand" "rm")
                       (match_operand:SWI48 2 "register_operand" "r")]
                       UNSPEC_BEXTR))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_BMI"
  "bextr\t{%2, %1, %0|%0, %1, %2}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*bmi_blsi_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (neg:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm"))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_BMI"
  "blsi\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*bmi_blsmsk_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (xor:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int -1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_BMI"
  "blsmsk\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*bmi_blsr_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int -1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_BMI"
   "blsr\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

;; TBM instructions.
(define_insn "tbm_bextri_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (zero_extract:SWI48
          (match_operand:SWI48 1 "nonimmediate_operand" "rm")
          (match_operand:SWI48 2 "const_0_to_255_operand" "n")
          (match_operand:SWI48 3 "const_0_to_255_operand" "n")))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
{
  operands[2] = GEN_INT (INTVAL (operands[2]) << 8 | INTVAL (operands[3]));
  return "bextr\t{%2, %1, %0|%0, %1, %2}";
}
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blcfill_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int 1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blcfill\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blci_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (ior:SWI48
          (not:SWI48
            (plus:SWI48
              (match_operand:SWI48 1 "nonimmediate_operand" "rm")
              (const_int 1)))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blci\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blcic_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int 1))
          (not:SWI48
            (match_dup 1))))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blcic\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blcmsk_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (xor:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int 1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blcmsk\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blcs_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (ior:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int 1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blcs\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blsfill_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (ior:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int -1))
          (match_dup 1)))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blsfill\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_blsic_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (ior:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int -1))
          (not:SWI48
            (match_dup 1))))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "blsic\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_t1mskc_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (ior:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int 1))
          (not:SWI48
            (match_dup 1))))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "t1mskc\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "*tbm_tzmsk_<mode>"
  [(set (match_operand:SWI48 0 "register_operand" "=r")
        (and:SWI48
          (plus:SWI48
            (match_operand:SWI48 1 "nonimmediate_operand" "rm")
            (const_int -1))
          (not:SWI48
            (match_dup 1))))
   (clobber (reg:CC FLAGS_REG))]
   "TARGET_TBM"
   "tzmsk\t{%1, %0|%0, %1}"
  [(set_attr "type" "bitmanip")
   (set_attr "mode" "<MODE>")])

(define_insn "bsr_rex64"
  [(set (match_operand:DI 0 "register_operand" "=r")
	(minus:DI (const_int 63)
		  (clz:DI (match_operand:DI 1 "nonimmediate_operand" "rm"))))
   (clobber (reg:CC FLAGS_REG))]
  "TARGET_64BIT"
  "bsr{q}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")
   (set_attr "prefix_0f" "1")
   (set_attr "mode" "DI")])

(define_insn "bsr"
  [(set (match_operand:SI 0 "register_operand" "=r")
	(minus:SI (const_int 31)
		  (clz:SI (match_operand:SI 1 "nonimmediate_operand" "rm"))))
   (clobber (reg:CC FLAGS_REG))]
  ""
  "bsr{l}\t{%1, %0|%0, %1}"
  [(set_attr "type" "alu1")