priv/host-x86/hdefs.c

/*---------------------------------------------------------------*/
/*---                                                         ---*/
/*--- This file (host-x86/hdefs.c) is                         ---*/
/*--- Copyright (C) OpenWorks LLP.  All rights reserved.      ---*/
/*---                                                         ---*/
/*---------------------------------------------------------------*/

/*
   This file is part of LibVEX, a library for dynamic binary
   instrumentation and translation.

   Copyright (C) 2004-2005 OpenWorks LLP.  All rights reserved.

   This library is made available under a dual licensing scheme.

   If you link LibVEX against other code all of which is itself
   licensed under the GNU General Public License, version 2 dated June
   1991 ("GPL v2"), then you may use LibVEX under the terms of the GPL
   v2, as appearing in the file LICENSE.GPL.  If the file LICENSE.GPL
   is missing, you can obtain a copy of the GPL v2 from the Free
   Software Foundation Inc., 51 Franklin St, Fifth Floor, Boston, MA
   02110-1301, USA.

   For any other uses of LibVEX, you must first obtain a commercial
   license from OpenWorks LLP.  Please contact info@open-works.co.uk
   for information about commercial licensing.

   This software is provided by OpenWorks LLP "as is" and any express
   or implied warranties, including, but not limited to, the implied
   warranties of merchantability and fitness for a particular purpose
   are disclaimed.  In no event shall OpenWorks LLP be liable for any
   direct, indirect, incidental, special, exemplary, or consequential
   damages (including, but not limited to, procurement of substitute
   goods or services; loss of use, data, or profits; or business
   interruption) however caused and on any theory of liability,
   whether in contract, strict liability, or tort (including
   negligence or otherwise) arising in any way out of the use of this
   software, even if advised of the possibility of such damage.

   Neither the names of the U.S. Department of Energy nor the
   University of California nor the names of its contributors may be
   used to endorse or promote products derived from this software
   without prior written permission.
*/

#include "libvex_basictypes.h"
#include "libvex.h"
#include "libvex_trc_values.h"

#include "main/vex_util.h"
#include "host-generic/h_generic_regs.h"
#include "host-x86/hdefs.h"


/* --------- Registers. --------- */

void ppHRegX86 ( HReg reg ) 
{
   Int r;
   static HChar* ireg32_names[8] 
     = { "%eax", "%ecx", "%edx", "%ebx", "%esp", "%ebp", "%esi", "%edi" };
   /* Be generic for all virtual regs. */
   if (hregIsVirtual(reg)) {
      ppHReg(reg);
      return;
   }
   /* But specific for real regs. */
   switch (hregClass(reg)) {
      case HRcInt32:
         r = hregNumber(reg);
         vassert(r >= 0 && r < 8);
         vex_printf("%s", ireg32_names[r]);
         return;
      case HRcFlt64:
         r = hregNumber(reg);
         vassert(r >= 0 && r < 6);
         vex_printf("%%fake%d", r);
         return;
      case HRcVec128:
         r = hregNumber(reg);
         vassert(r >= 0 && r < 8);
         vex_printf("%%xmm%d", r);
         return;
      default:
         vpanic("ppHRegX86");
   }
}

HReg hregX86_EAX ( void ) { return mkHReg(0, HRcInt32, False); }
HReg hregX86_ECX ( void ) { return mkHReg(1, HRcInt32, False); }
HReg hregX86_EDX ( void ) { return mkHReg(2, HRcInt32, False); }
HReg hregX86_EBX ( void ) { return mkHReg(3, HRcInt32, False); }
HReg hregX86_ESP ( void ) { return mkHReg(4, HRcInt32, False); }
HReg hregX86_EBP ( void ) { return mkHReg(5, HRcInt32, False); }
HReg hregX86_ESI ( void ) { return mkHReg(6, HRcInt32, False); }
HReg hregX86_EDI ( void ) { return mkHReg(7, HRcInt32, False); }

HReg hregX86_FAKE0 ( void ) { return mkHReg(0, HRcFlt64, False); }
HReg hregX86_FAKE1 ( void ) { return mkHReg(1, HRcFlt64, False); }
HReg hregX86_FAKE2 ( void ) { return mkHReg(2, HRcFlt64, False); }
HReg hregX86_FAKE3 ( void ) { return mkHReg(3, HRcFlt64, False); }
HReg hregX86_FAKE4 ( void ) { return mkHReg(4, HRcFlt64, False); }
HReg hregX86_FAKE5 ( void ) { return mkHReg(5, HRcFlt64, False); }

HReg hregX86_XMM0 ( void ) { return mkHReg(0, HRcVec128, False); }
HReg hregX86_XMM1 ( void ) { return mkHReg(1, HRcVec128, False); }
HReg hregX86_XMM2 ( void ) { return mkHReg(2, HRcVec128, False); }
HReg hregX86_XMM3 ( void ) { return mkHReg(3, HRcVec128, False); }
HReg hregX86_XMM4 ( void ) { return mkHReg(4, HRcVec128, False); }
HReg hregX86_XMM5 ( void ) { return mkHReg(5, HRcVec128, False); }
HReg hregX86_XMM6 ( void ) { return mkHReg(6, HRcVec128, False); }
HReg hregX86_XMM7 ( void ) { return mkHReg(7, HRcVec128, False); }


void getAllocableRegs_X86 ( Int* nregs, HReg** arr )
{
   *nregs = 20;
   *arr = LibVEX_Alloc(*nregs * sizeof(HReg));
   (*arr)[0] = hregX86_EAX();
   (*arr)[1] = hregX86_EBX();
   (*arr)[2] = hregX86_ECX();
   (*arr)[3] = hregX86_EDX();
   (*arr)[4] = hregX86_ESI();
   (*arr)[5] = hregX86_EDI();
   (*arr)[6] = hregX86_FAKE0();
   (*arr)[7] = hregX86_FAKE1();
   (*arr)[8] = hregX86_FAKE2();
   (*arr)[9] = hregX86_FAKE3();
   (*arr)[10] = hregX86_FAKE4();
   (*arr)[11] = hregX86_FAKE5();
   (*arr)[12] = hregX86_XMM0();
   (*arr)[13] = hregX86_XMM1();
   (*arr)[14] = hregX86_XMM2();
   (*arr)[15] = hregX86_XMM3();
   (*arr)[16] = hregX86_XMM4();
   (*arr)[17] = hregX86_XMM5();
   (*arr)[18] = hregX86_XMM6();
   (*arr)[19] = hregX86_XMM7();
}


/* --------- Condition codes, Intel encoding. --------- */

HChar* showX86CondCode ( X86CondCode cond )
{
   switch (cond) {
      case Xcc_O:      return "o";
      case Xcc_NO:     return "no";
      case Xcc_B:      return "b";
      case Xcc_NB:     return "nb";
      case Xcc_Z:      return "z";
      case Xcc_NZ:     return "nz";
      case Xcc_BE:     return "be";
      case Xcc_NBE:    return "nbe";
      case Xcc_S:      return "s";
      case Xcc_NS:     return "ns";
      case Xcc_P:      return "p";
      case Xcc_NP:     return "np";
      case Xcc_L:      return "l";
      case Xcc_NL:     return "nl";
      case Xcc_LE:     return "le";
      case Xcc_NLE:    return "nle";
      case Xcc_ALWAYS: return "ALWAYS";
      default: vpanic("ppX86CondCode");
   }
}


/* --------- X86AMode: memory address expressions. --------- */

X86AMode* X86AMode_IR ( UInt imm32, HReg reg ) {
   X86AMode* am = LibVEX_Alloc(sizeof(X86AMode));
   am->tag = Xam_IR;
   am->Xam.IR.imm = imm32;
   am->Xam.IR.reg = reg;
   return am;
}
X86AMode* X86AMode_IRRS ( UInt imm32, HReg base, HReg indEx, Int shift ) {
   X86AMode* am = LibVEX_Alloc(sizeof(X86AMode));
   am->tag = Xam_IRRS;
   am->Xam.IRRS.imm = imm32;
   am->Xam.IRRS.base = base;
   am->Xam.IRRS.index = indEx;
   am->Xam.IRRS.shift = shift;
   vassert(shift >= 0 && shift <= 3);
   return am;
}

X86AMode* dopyX86AMode ( X86AMode* am ) {
   switch (am->tag) {
      case Xam_IR: 
         return X86AMode_IR( am->Xam.IR.imm, am->Xam.IR.reg );
      case Xam_IRRS: 
         return X86AMode_IRRS( am->Xam.IRRS.imm, am->Xam.IRRS.base, 
                               am->Xam.IRRS.index, am->Xam.IRRS.shift );
      default:
         vpanic("dopyX86AMode");
   }
}

void ppX86AMode ( X86AMode* am ) {
   switch (am->tag) {
      case Xam_IR: 
         if (am->Xam.IR.imm == 0)
            vex_printf("(");
         else
            vex_printf("0x%x(", am->Xam.IR.imm);
         ppHRegX86(am->Xam.IR.reg);
         vex_printf(")");
         return;
      case Xam_IRRS:
         vex_printf("0x%x(", am->Xam.IRRS.imm);
         ppHRegX86(am->Xam.IRRS.base);
         vex_printf(",");
         ppHRegX86(am->Xam.IRRS.index);
         vex_printf(",%d)", 1 << am->Xam.IRRS.shift);
         return;
      default:
         vpanic("ppX86AMode");
   }
}

static void addRegUsage_X86AMode ( HRegUsage* u, X86AMode* am ) {
   switch (am->tag) {
      case Xam_IR: 
         addHRegUse(u, HRmRead, am->Xam.IR.reg);
         return;
      case Xam_IRRS:
         addHRegUse(u, HRmRead, am->Xam.IRRS.base);
         addHRegUse(u, HRmRead, am->Xam.IRRS.index);
         return;
      default:
         vpanic("addRegUsage_X86AMode");
   }
}

static void mapRegs_X86AMode ( HRegRemap* m, X86AMode* am ) {
   switch (am->tag) {
      case Xam_IR: 
         am->Xam.IR.reg = lookupHRegRemap(m, am->Xam.IR.reg);
         return;
      case Xam_IRRS:
         am->Xam.IRRS.base = lookupHRegRemap(m, am->Xam.IRRS.base);
         am->Xam.IRRS.index = lookupHRegRemap(m, am->Xam.IRRS.index);
         return;
      default:
         vpanic("mapRegs_X86AMode");
   }
}

/* --------- Operand, which can be reg, immediate or memory. --------- */

X86RMI* X86RMI_Imm ( UInt imm32 ) {
   X86RMI* op         = LibVEX_Alloc(sizeof(X86RMI));
   op->tag            = Xrmi_Imm;
   op->Xrmi.Imm.imm32 = imm32;
   return op;
}
X86RMI* X86RMI_Reg ( HReg reg ) {
   X86RMI* op       = LibVEX_Alloc(sizeof(X86RMI));
   op->tag          = Xrmi_Reg;
   op->Xrmi.Reg.reg = reg;
   return op;
}
X86RMI* X86RMI_Mem ( X86AMode* am ) {
   X86RMI* op      = LibVEX_Alloc(sizeof(X86RMI));
   op->tag         = Xrmi_Mem;
   op->Xrmi.Mem.am = am;
   return op;
}

void ppX86RMI ( X86RMI* op ) {
   switch (op->tag) {
      case Xrmi_Imm: 
         vex_printf("$0x%x", op->Xrmi.Imm.imm32);
         return;
      case Xrmi_Reg: 
         ppHRegX86(op->Xrmi.Reg.reg);
         return;
      case Xrmi_Mem: 
         ppX86AMode(op->Xrmi.Mem.am);
         return;
     default: 
         vpanic("ppX86RMI");
   }
}

/* An X86RMI can only be used in a "read" context (what would it mean
   to write or modify a literal?) and so we enumerate its registers
   accordingly. */
static void addRegUsage_X86RMI ( HRegUsage* u, X86RMI* op ) {
   switch (op->tag) {
      case Xrmi_Imm: 
         return;
      case Xrmi_Reg: 
         addHRegUse(u, HRmRead, op->Xrmi.Reg.reg);
         return;
      case Xrmi_Mem: 
         addRegUsage_X86AMode(u, op->Xrmi.Mem.am);
         return;
      default: 
         vpanic("addRegUsage_X86RMI");
   }
}

static void mapRegs_X86RMI ( HRegRemap* m, X86RMI* op ) {
   switch (op->tag) {
      case Xrmi_Imm: 
         return;
      case Xrmi_Reg: 
         op->Xrmi.Reg.reg = lookupHRegRemap(m, op->Xrmi.Reg.reg);
         return;
      case Xrmi_Mem: 
         mapRegs_X86AMode(m, op->Xrmi.Mem.am);
         return;
      default: 
         vpanic("mapRegs_X86RMI");
   }
}


/* --------- Operand, which can be reg or immediate only. --------- */

X86RI* X86RI_Imm ( UInt imm32 ) {
   X86RI* op         = LibVEX_Alloc(sizeof(X86RI));
   op->tag           = Xri_Imm;
   op->Xri.Imm.imm32 = imm32;
   return op;
}
X86RI* X86RI_Reg ( HReg reg ) {
   X86RI* op       = LibVEX_Alloc(sizeof(X86RI));
   op->tag         = Xri_Reg;
   op->Xri.Reg.reg = reg;
   return op;
}

void ppX86RI ( X86RI* op ) {
   switch (op->tag) {
      case Xri_Imm: 
         vex_printf("$0x%x", op->Xri.Imm.imm32);
         return;
      case Xri_Reg: 
         ppHRegX86(op->Xri.Reg.reg);
         return;
     default: 
         vpanic("ppX86RI");
   }
}

/* An X86RI can only be used in a "read" context (what would it mean
   to write or modify a literal?) and so we enumerate its registers
   accordingly. */
static void addRegUsage_X86RI ( HRegUsage* u, X86RI* op ) {
   switch (op->tag) {
      case Xri_Imm: 
         return;
      case Xri_Reg: 
         addHRegUse(u, HRmRead, op->Xri.Reg.reg);
         return;
      default: 
         vpanic("addRegUsage_X86RI");
   }
}

static void mapRegs_X86RI ( HRegRemap* m, X86RI* op ) {
   switch (op->tag) {
      case Xri_Imm: 
         return;
      case Xri_Reg: 
         op->Xri.Reg.reg = lookupHRegRemap(m, op->Xri.Reg.reg);
         return;
      default: 
         vpanic("mapRegs_X86RI");
   }
}


/* --------- Operand, which can be reg or memory only. --------- */

X86RM* X86RM_Reg ( HReg reg ) {
   X86RM* op       = LibVEX_Alloc(sizeof(X86RM));
   op->tag         = Xrm_Reg;
   op->Xrm.Reg.reg = reg;
   return op;
}
X86RM* X86RM_Mem ( X86AMode* am ) {
   X86RM* op      = LibVEX_Alloc(sizeof(X86RM));
   op->tag        = Xrm_Mem;
   op->Xrm.Mem.am = am;
   return op;
}

void ppX86RM ( X86RM* op ) {
   switch (op->tag) {
      case Xrm_Mem: 
         ppX86AMode(op->Xrm.Mem.am);
         return;
      case Xrm_Reg: 
         ppHRegX86(op->Xrm.Reg.reg);
         return;
     default: 
         vpanic("ppX86RM");
   }
}

/* Because an X86RM can be both a source or destination operand, we
   have to supply a mode -- pertaining to the operand as a whole --
   indicating how it's being used. */
static void addRegUsage_X86RM ( HRegUsage* u, X86RM* op, HRegMode mode ) {
   switch (op->tag) {
      case Xrm_Mem: 
         /* Memory is read, written or modified.  So we just want to
            know the regs read by the amode. */
         addRegUsage_X86AMode(u, op->Xrm.Mem.am);
         return;
      case Xrm_Reg: 
         /* reg is read, written or modified.  Add it in the
            appropriate way. */
         addHRegUse(u, mode, op->Xrm.Reg.reg);
         return;
     default: 
         vpanic("addRegUsage_X86RM");
   }
}

static void mapRegs_X86RM ( HRegRemap* m, X86RM* op )
{
   switch (op->tag) {
      case Xrm_Mem: 
         mapRegs_X86AMode(m, op->Xrm.Mem.am);
         return;
      case Xrm_Reg: 
         op->Xrm.Reg.reg = lookupHRegRemap(m, op->Xrm.Reg.reg);
         return;
     default: 
         vpanic("mapRegs_X86RM");
   }
}


/* --------- Instructions. --------- */

HChar* showX86UnaryOp ( X86UnaryOp op ) {
   switch (op) {
      case Xun_NOT: return "not";
      case Xun_NEG: return "neg";
      default: vpanic("showX86UnaryOp");
   }
}

HChar* showX86AluOp ( X86AluOp op ) {
   switch (op) {
      case Xalu_MOV:  return "mov";
      case Xalu_CMP:  return "cmp";
      case Xalu_ADD:  return "add";
      case Xalu_SUB:  return "sub";
      case Xalu_ADC:  return "adc";
      case Xalu_SBB:  return "sbb";
      case Xalu_AND:  return "and";
      case Xalu_OR:   return "or";
      case Xalu_XOR:  return "xor";
      case Xalu_MUL:  return "mul";
      default: vpanic("showX86AluOp");
   }
}

HChar* showX86ShiftOp ( X86ShiftOp op ) {
   switch (op) {
      case Xsh_SHL: return "shl";
      case Xsh_SHR: return "shr";
      case Xsh_SAR: return "sar";
      default: vpanic("showX86ShiftOp");
   }
}

HChar* showX86FpOp ( X86FpOp op ) {
   switch (op) {
      case Xfp_ADD:    return "add";
      case Xfp_SUB:    return "sub";
      case Xfp_MUL:    return "mul";
      case Xfp_DIV:    return "div";
      case Xfp_SCALE:  return "scale";
      case Xfp_ATAN:   return "atan";
      case Xfp_YL2X:   return "yl2x";
      case Xfp_YL2XP1: return "yl2xp1";
      case Xfp_PREM:   return "prem";
      case Xfp_PREM1:  return "prem1";
      case Xfp_SQRT:   return "sqrt";
      case Xfp_ABS:    return "abs";
      case Xfp_NEG:    return "chs";
      case Xfp_MOV:    return "mov";
      case Xfp_SIN:    return "sin";
      case Xfp_COS:    return "cos";
      case Xfp_TAN:    return "tan";
      case Xfp_ROUND:  return "round";
      case Xfp_2XM1:   return "2xm1";
      default: vpanic("showX86FpOp");
   }
}

HChar* showX86SseOp ( X86SseOp op ) {
   switch (op) {
      case Xsse_MOV:      return "mov(?!)";
      case Xsse_ADDF:     return "add";
      case Xsse_SUBF:     return "sub";
      case Xsse_MULF:     return "mul";
      case Xsse_DIVF:     return "div";
      case Xsse_MAXF:     return "max";
      case Xsse_MINF:     return "min";
      case Xsse_CMPEQF:   return "cmpFeq";
      case Xsse_CMPLTF:   return "cmpFlt";
      case Xsse_CMPLEF:   return "cmpFle";
      case Xsse_CMPUNF:   return "cmpFun";
      case Xsse_RCPF:     return "rcp";
      case Xsse_RSQRTF:   return "rsqrt";
      case Xsse_SQRTF:    return "sqrt";
      case Xsse_AND:      return "and";
      case Xsse_OR:       return "or";
      case Xsse_XOR:      return "xor";
      case Xsse_ANDN:     return "andn";
      case Xsse_ADD8:     return "paddb";
      case Xsse_ADD16:    return "paddw";
      case Xsse_ADD32:    return "paddd";
      case Xsse_ADD64:    return "paddq";
      case Xsse_QADD8U:   return "paddusb";
      case Xsse_QADD16U:  return "paddusw";
      case Xsse_QADD8S:   return "paddsb";
      case Xsse_QADD16S:  return "paddsw";
      case Xsse_SUB8:     return "psubb";
      case Xsse_SUB16:    return "psubw";
      case Xsse_SUB32:    return "psubd";
      case Xsse_SUB64:    return "psubq";
      case Xsse_QSUB8U:   return "psubusb";
      case Xsse_QSUB16U:  return "psubusw";
      case Xsse_QSUB8S:   return "psubsb";
      case Xsse_QSUB16S:  return "psubsw";
      case Xsse_MUL16:    return "pmullw";
      case Xsse_MULHI16U: return "pmulhuw";
      case Xsse_MULHI16S: return "pmulhw";
      case Xsse_AVG8U:    return "pavgb";
      case Xsse_AVG16U:   return "pavgw";
      case Xsse_MAX16S:   return "pmaxw";
      case Xsse_MAX8U:    return "pmaxub";
      case Xsse_MIN16S:   return "pminw";
      case Xsse_MIN8U:    return "pminub";
      case Xsse_CMPEQ8:   return "pcmpeqb";
      case Xsse_CMPEQ16:  return "pcmpeqw";
      case Xsse_CMPEQ32:  return "pcmpeqd";
      case Xsse_CMPGT8S:  return "pcmpgtb";
      case Xsse_CMPGT16S: return "pcmpgtw";
      case Xsse_CMPGT32S: return "pcmpgtd";
      case Xsse_SHL16:    return "psllw";
      case Xsse_SHL32:    return "pslld";
      case Xsse_SHL64:    return "psllq";
      case Xsse_SHR16:    return "psrlw";
      case Xsse_SHR32:    return "psrld";
      case Xsse_SHR64:    return "psrlq";
      case Xsse_SAR16:    return "psraw";
      case Xsse_SAR32:    return "psrad";
      case Xsse_PACKSSD:  return "packssdw";
      case Xsse_PACKSSW:  return "packsswb";
      case Xsse_PACKUSW:  return "packuswb";
      case Xsse_UNPCKHB:  return "punpckhb";
      case Xsse_UNPCKHW:  return "punpckhw";
      case Xsse_UNPCKHD:  return "punpckhd";
      case Xsse_UNPCKHQ:  return "punpckhq";
      case Xsse_UNPCKLB:  return "punpcklb";
      case Xsse_UNPCKLW:  return "punpcklw";
      case Xsse_UNPCKLD:  return "punpckld";
      case Xsse_UNPCKLQ:  return "punpcklq";
      default: vpanic("showX86SseOp");
   }
}

X86Instr* X86Instr_Alu32R ( X86AluOp op, X86RMI* src, HReg dst ) {
   X86Instr* i       = LibVEX_Alloc(sizeof(X86Instr));
   i->tag            = Xin_Alu32R;
   i->Xin.Alu32R.op  = op;
   i->Xin.Alu32R.src = src;
   i->Xin.Alu32R.dst = dst;
   return i;
}
X86Instr* X86Instr_Alu32M ( X86AluOp op, X86RI* src, X86AMode* dst ) {
   X86Instr* i       = LibVEX_Alloc(sizeof(X86Instr));
   i->tag            = Xin_Alu32M;
   i->Xin.Alu32M.op  = op;
   i->Xin.Alu32M.src = src;
   i->Xin.Alu32M.dst = dst;
   vassert(op != Xalu_MUL);
   return i;
}
X86Instr* X86Instr_Sh32 ( X86ShiftOp op, UInt src, HReg dst ) {
   X86Instr* i     = LibVEX_Alloc(sizeof(X86Instr));
   i->tag          = Xin_Sh32;
   i->Xin.Sh32.op  = op;
   i->Xin.Sh32.src = src;
   i->Xin.Sh32.dst = dst;
   return i;
}
X86Instr* X86Instr_Test32 ( UInt imm32, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Test32;
   i->Xin.Test32.imm32 = imm32;
   i->Xin.Test32.dst   = dst;
   return i;
}
X86Instr* X86Instr_Unary32 ( X86UnaryOp op, HReg dst ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_Unary32;
   i->Xin.Unary32.op  = op;
   i->Xin.Unary32.dst = dst;
   return i;
}
X86Instr* X86Instr_MulL ( Bool syned, X86RM* src ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_MulL;
   i->Xin.MulL.syned  = syned;
   i->Xin.MulL.src    = src;
   return i;
}
X86Instr* X86Instr_Div ( Bool syned, X86RM* src ) {
   X86Instr* i      = LibVEX_Alloc(sizeof(X86Instr));
   i->tag           = Xin_Div;
   i->Xin.Div.syned = syned;
   i->Xin.Div.src   = src;
   return i;
}
X86Instr* X86Instr_Sh3232  ( X86ShiftOp op, UInt amt, HReg src, HReg dst ) {
   X86Instr* i       = LibVEX_Alloc(sizeof(X86Instr));
   i->tag            = Xin_Sh3232;
   i->Xin.Sh3232.op  = op;
   i->Xin.Sh3232.amt = amt;
   i->Xin.Sh3232.src = src;
   i->Xin.Sh3232.dst = dst;
   vassert(op == Xsh_SHL || op == Xsh_SHR);
   return i;
}
X86Instr* X86Instr_Push( X86RMI* src ) {
   X86Instr* i     = LibVEX_Alloc(sizeof(X86Instr));
   i->tag          = Xin_Push;
   i->Xin.Push.src = src;
   return i;
}
X86Instr* X86Instr_Call ( X86CondCode cond, Addr32 target, Int regparms ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_Call;
   i->Xin.Call.cond     = cond;
   i->Xin.Call.target   = target;
   i->Xin.Call.regparms = regparms;
   vassert(regparms >= 0 && regparms <= 3);
   return i;
}
X86Instr* X86Instr_Goto ( IRJumpKind jk, X86CondCode cond, X86RI* dst ) {
   X86Instr* i      = LibVEX_Alloc(sizeof(X86Instr));
   i->tag           = Xin_Goto;
   i->Xin.Goto.cond = cond;
   i->Xin.Goto.dst  = dst;
   i->Xin.Goto.jk   = jk;
   return i;
}
X86Instr* X86Instr_CMov32  ( X86CondCode cond, X86RM* src, HReg dst ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_CMov32;
   i->Xin.CMov32.cond = cond;
   i->Xin.CMov32.src  = src;
   i->Xin.CMov32.dst  = dst;
   vassert(cond != Xcc_ALWAYS);
   return i;
}
X86Instr* X86Instr_LoadEX ( UChar szSmall, Bool syned,
                            X86AMode* src, HReg dst ) {
   X86Instr* i           = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                = Xin_LoadEX;
   i->Xin.LoadEX.szSmall = szSmall;
   i->Xin.LoadEX.syned   = syned;
   i->Xin.LoadEX.src     = src;
   i->Xin.LoadEX.dst     = dst;
   vassert(szSmall == 1 || szSmall == 2);
   return i;
}
X86Instr* X86Instr_Store ( UChar sz, HReg src, X86AMode* dst ) {
   X86Instr* i      = LibVEX_Alloc(sizeof(X86Instr));
   i->tag           = Xin_Store;
   i->Xin.Store.sz  = sz;
   i->Xin.Store.src = src;
   i->Xin.Store.dst = dst;
   vassert(sz == 1 || sz == 2);
   return i;
}
X86Instr* X86Instr_Set32 ( X86CondCode cond, HReg dst ) {
   X86Instr* i       = LibVEX_Alloc(sizeof(X86Instr));
   i->tag            = Xin_Set32;
   i->Xin.Set32.cond = cond;
   i->Xin.Set32.dst  = dst;
   return i;
}
X86Instr* X86Instr_Bsfr32 ( Bool isFwds, HReg src, HReg dst ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_Bsfr32;
   i->Xin.Bsfr32.isFwds = isFwds;
   i->Xin.Bsfr32.src    = src;
   i->Xin.Bsfr32.dst    = dst;
   return i;
}
X86Instr* X86Instr_MFence ( VexSubArch subarch )
{
   X86Instr* i           = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                = Xin_MFence;
   i->Xin.MFence.subarch = subarch;
   vassert(subarch == VexSubArchX86_sse0
           || subarch == VexSubArchX86_sse1
           || subarch == VexSubArchX86_sse2);
   return i;
}

X86Instr* X86Instr_FpUnary ( X86FpOp op, HReg src, HReg dst ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_FpUnary;
   i->Xin.FpUnary.op  = op;
   i->Xin.FpUnary.src = src;
   i->Xin.FpUnary.dst = dst;
   return i;
}
X86Instr* X86Instr_FpBinary ( X86FpOp op, HReg srcL, HReg srcR, HReg dst ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_FpBinary;
   i->Xin.FpBinary.op   = op;
   i->Xin.FpBinary.srcL = srcL;
   i->Xin.FpBinary.srcR = srcR;
   i->Xin.FpBinary.dst  = dst;
   return i;
}
X86Instr* X86Instr_FpLdSt ( Bool isLoad, UChar sz, HReg reg, X86AMode* addr ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_FpLdSt;
   i->Xin.FpLdSt.isLoad = isLoad;
   i->Xin.FpLdSt.sz     = sz;
   i->Xin.FpLdSt.reg    = reg;
   i->Xin.FpLdSt.addr   = addr;
   vassert(sz == 4 || sz == 8);
   return i;
}
X86Instr* X86Instr_FpLdStI ( Bool isLoad, UChar sz,  
                             HReg reg, X86AMode* addr ) {
   X86Instr* i           = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                = Xin_FpLdStI;
   i->Xin.FpLdStI.isLoad = isLoad;
   i->Xin.FpLdStI.sz     = sz;
   i->Xin.FpLdStI.reg    = reg;
   i->Xin.FpLdStI.addr   = addr;
   vassert(sz == 2 || sz == 4 || sz == 8);
   return i;
}
X86Instr* X86Instr_Fp64to32 ( HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Fp64to32;
   i->Xin.Fp64to32.src = src;
   i->Xin.Fp64to32.dst = dst;
   return i;
}
X86Instr* X86Instr_FpCMov ( X86CondCode cond, HReg src, HReg dst ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_FpCMov;
   i->Xin.FpCMov.cond = cond;
   i->Xin.FpCMov.src  = src;
   i->Xin.FpCMov.dst  = dst;
   vassert(cond != Xcc_ALWAYS);
   return i;
}
X86Instr* X86Instr_FpLdCW ( X86AMode* addr ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_FpLdCW;
   i->Xin.FpLdCW.addr   = addr;
   return i;
}
X86Instr* X86Instr_FpStSW_AX ( void ) {
   X86Instr* i = LibVEX_Alloc(sizeof(X86Instr));
   i->tag      = Xin_FpStSW_AX;
   return i;
}
X86Instr* X86Instr_FpCmp ( HReg srcL, HReg srcR, HReg dst ) {
   X86Instr* i       = LibVEX_Alloc(sizeof(X86Instr));
   i->tag            = Xin_FpCmp;
   i->Xin.FpCmp.srcL = srcL;
   i->Xin.FpCmp.srcR = srcR;
   i->Xin.FpCmp.dst  = dst;
   return i;
}

X86Instr* X86Instr_SseConst ( UShort con, HReg dst ) {
   X86Instr* i            = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                 = Xin_SseConst;
   i->Xin.SseConst.con    = con;
   i->Xin.SseConst.dst    = dst;
   vassert(hregClass(dst) == HRcVec128);
   return i;
}
X86Instr* X86Instr_SseLdSt ( Bool isLoad, HReg reg, X86AMode* addr ) {
   X86Instr* i           = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                = Xin_SseLdSt;
   i->Xin.SseLdSt.isLoad = isLoad;
   i->Xin.SseLdSt.reg    = reg;
   i->Xin.SseLdSt.addr   = addr;
   return i;
}
X86Instr* X86Instr_SseLdzLO  ( Int sz, HReg reg, X86AMode* addr )
{
   X86Instr* i           = LibVEX_Alloc(sizeof(X86Instr));
   i->tag                = Xin_SseLdzLO;
   i->Xin.SseLdzLO.sz    = toUChar(sz);
   i->Xin.SseLdzLO.reg   = reg;
   i->Xin.SseLdzLO.addr  = addr;
   vassert(sz == 4 || sz == 8);
   return i;
}
X86Instr* X86Instr_Sse32Fx4 ( X86SseOp op, HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Sse32Fx4;
   i->Xin.Sse32Fx4.op  = op;
   i->Xin.Sse32Fx4.src = src;
   i->Xin.Sse32Fx4.dst = dst;
   vassert(op != Xsse_MOV);
   return i;
}
X86Instr* X86Instr_Sse32FLo ( X86SseOp op, HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Sse32FLo;
   i->Xin.Sse32FLo.op  = op;
   i->Xin.Sse32FLo.src = src;
   i->Xin.Sse32FLo.dst = dst;
   vassert(op != Xsse_MOV);
   return i;
}
X86Instr* X86Instr_Sse64Fx2 ( X86SseOp op, HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Sse64Fx2;
   i->Xin.Sse64Fx2.op  = op;
   i->Xin.Sse64Fx2.src = src;
   i->Xin.Sse64Fx2.dst = dst;
   vassert(op != Xsse_MOV);
   return i;
}
X86Instr* X86Instr_Sse64FLo ( X86SseOp op, HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_Sse64FLo;
   i->Xin.Sse64FLo.op  = op;
   i->Xin.Sse64FLo.src = src;
   i->Xin.Sse64FLo.dst = dst;
   vassert(op != Xsse_MOV);
   return i;
}
X86Instr* X86Instr_SseReRg ( X86SseOp op, HReg re, HReg rg ) {
   X86Instr* i        = LibVEX_Alloc(sizeof(X86Instr));
   i->tag             = Xin_SseReRg;
   i->Xin.SseReRg.op  = op;
   i->Xin.SseReRg.src = re;
   i->Xin.SseReRg.dst = rg;
   return i;
}
X86Instr* X86Instr_SseCMov ( X86CondCode cond, HReg src, HReg dst ) {
   X86Instr* i         = LibVEX_Alloc(sizeof(X86Instr));
   i->tag              = Xin_SseCMov;
   i->Xin.SseCMov.cond = cond;
   i->Xin.SseCMov.src  = src;
   i->Xin.SseCMov.dst  = dst;
   vassert(cond != Xcc_ALWAYS);
   return i;
}
X86Instr* X86Instr_SseShuf ( Int order, HReg src, HReg dst ) {
   X86Instr* i          = LibVEX_Alloc(sizeof(X86Instr));
   i->tag               = Xin_SseShuf;
   i->Xin.SseShuf.order = order;
   i->Xin.SseShuf.src   = src;
   i->Xin.SseShuf.dst   = dst;
   vassert(order >= 0 && order <= 0xFF);
   return i;
}

void ppX86Instr ( X86Instr* i, Bool mode64 ) {
   vassert(mode64 == False);
   switch (i->tag) {
      case Xin_Alu32R:
         vex_printf("%sl ", showX86AluOp(i->Xin.Alu32R.op));
         ppX86RMI(i->Xin.Alu32R.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Alu32R.dst);
         return;
      case Xin_Alu32M:
         vex_printf("%sl ", showX86AluOp(i->Xin.Alu32M.op));
         ppX86RI(i->Xin.Alu32M.src);
         vex_printf(",");
         ppX86AMode(i->Xin.Alu32M.dst);
         return;
      case Xin_Sh32:
         vex_printf("%sl ", showX86ShiftOp(i->Xin.Sh32.op));
         if (i->Xin.Sh32.src == 0)
           vex_printf("%%cl,"); 
         else 
            vex_printf("$%d,", (Int)i->Xin.Sh32.src);
         ppHRegX86(i->Xin.Sh32.dst);
         return;
      case Xin_Test32:
         vex_printf("testl $%d,", (Int)i->Xin.Test32.imm32);
         ppHRegX86(i->Xin.Test32.dst);
         return;
      case Xin_Unary32:
         vex_printf("%sl ", showX86UnaryOp(i->Xin.Unary32.op));
         ppHRegX86(i->Xin.Unary32.dst);
         return;
      case Xin_MulL:
         vex_printf("%cmull ", i->Xin.MulL.syned ? 's' : 'u');
         ppX86RM(i->Xin.MulL.src);
         return;
      case Xin_Div:
         vex_printf("%cdivl ", i->Xin.Div.syned ? 's' : 'u');
         ppX86RM(i->Xin.Div.src);
         return;
      case Xin_Sh3232:
         vex_printf("%sdl ", showX86ShiftOp(i->Xin.Sh3232.op));
         if (i->Xin.Sh3232.amt == 0)
           vex_printf(" %%cl,"); 
         else 
            vex_printf(" $%d,", (Int)i->Xin.Sh3232.amt);
         ppHRegX86(i->Xin.Sh3232.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Sh3232.dst);
         return;
      case Xin_Push:
         vex_printf("pushl ");
         ppX86RMI(i->Xin.Push.src);
         return;
      case Xin_Call:
         vex_printf("call%s[%d] ", 
                    i->Xin.Call.cond==Xcc_ALWAYS 
                       ? "" : showX86CondCode(i->Xin.Call.cond), 
                    i->Xin.Call.regparms);
         vex_printf("0x%x", i->Xin.Call.target);
         break;
      case Xin_Goto:
         if (i->Xin.Goto.cond != Xcc_ALWAYS) {
            vex_printf("if (%%eflags.%s) { ", 
                       showX86CondCode(i->Xin.Goto.cond));
	 }
         if (i->Xin.Goto.jk != Ijk_Boring
             && i->Xin.Goto.jk != Ijk_Call
             && i->Xin.Goto.jk != Ijk_Ret) {
            vex_printf("movl $");
            ppIRJumpKind(i->Xin.Goto.jk);
            vex_printf(",%%ebp ; ");
         }
         vex_printf("movl ");
         ppX86RI(i->Xin.Goto.dst);
         vex_printf(",%%eax ; movl $dispatcher_addr,%%edx ; jmp *%%edx");
         if (i->Xin.Goto.cond != Xcc_ALWAYS) {
            vex_printf(" }");
	 }
         return;
      case Xin_CMov32:
         vex_printf("cmov%s ", showX86CondCode(i->Xin.CMov32.cond));
         ppX86RM(i->Xin.CMov32.src);
         vex_printf(",");
         ppHRegX86(i->Xin.CMov32.dst);
         return;
      case Xin_LoadEX:
         vex_printf("mov%c%cl ",
                    i->Xin.LoadEX.syned ? 's' : 'z',
                    i->Xin.LoadEX.szSmall==1 ? 'b' : 'w');
         ppX86AMode(i->Xin.LoadEX.src);
         vex_printf(",");
         ppHRegX86(i->Xin.LoadEX.dst);
         return;
      case Xin_Store:
         vex_printf("mov%c ", i->Xin.Store.sz==1 ? 'b' : 'w');
         ppHRegX86(i->Xin.Store.src);
         vex_printf(",");
         ppX86AMode(i->Xin.Store.dst);
         return;
      case Xin_Set32:
         vex_printf("setl%s ", showX86CondCode(i->Xin.Set32.cond));
         ppHRegX86(i->Xin.Set32.dst);
         return;
      case Xin_Bsfr32:
         vex_printf("bs%cl ", i->Xin.Bsfr32.isFwds ? 'f' : 'r');
         ppHRegX86(i->Xin.Bsfr32.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Bsfr32.dst);
         return;
      case Xin_MFence:
         vex_printf("mfence(%s)",
                    LibVEX_ppVexSubArch(i->Xin.MFence.subarch));
         return;
      case Xin_FpUnary:
         vex_printf("g%sD ", showX86FpOp(i->Xin.FpUnary.op));
         ppHRegX86(i->Xin.FpUnary.src);
         vex_printf(",");
         ppHRegX86(i->Xin.FpUnary.dst);
         break;
      case Xin_FpBinary:
         vex_printf("g%sD ", showX86FpOp(i->Xin.FpBinary.op));
         ppHRegX86(i->Xin.FpBinary.srcL);
         vex_printf(",");
         ppHRegX86(i->Xin.FpBinary.srcR);
         vex_printf(",");
         ppHRegX86(i->Xin.FpBinary.dst);
         break;
      case Xin_FpLdSt:
         if (i->Xin.FpLdSt.isLoad) {
            vex_printf("gld%c " , i->Xin.FpLdSt.sz==8 ? 'D' : 'F');
            ppX86AMode(i->Xin.FpLdSt.addr);
            vex_printf(", ");
            ppHRegX86(i->Xin.FpLdSt.reg);
         } else {
            vex_printf("gst%c " , i->Xin.FpLdSt.sz==8 ? 'D' : 'F');
            ppHRegX86(i->Xin.FpLdSt.reg);
            vex_printf(", ");
            ppX86AMode(i->Xin.FpLdSt.addr);
         }
         return;
      case Xin_FpLdStI:
         if (i->Xin.FpLdStI.isLoad) {
            vex_printf("gild%s ", i->Xin.FpLdStI.sz==8 ? "ll" : 
                                  i->Xin.FpLdStI.sz==4 ? "l" : "w");
            ppX86AMode(i->Xin.FpLdStI.addr);
            vex_printf(", ");
            ppHRegX86(i->Xin.FpLdStI.reg);
         } else {
            vex_printf("gist%s ", i->Xin.FpLdStI.sz==8 ? "ll" : 
                                  i->Xin.FpLdStI.sz==4 ? "l" : "w");
            ppHRegX86(i->Xin.FpLdStI.reg);
            vex_printf(", ");
            ppX86AMode(i->Xin.FpLdStI.addr);
         }
         return;
      case Xin_Fp64to32:
         vex_printf("gdtof ");
         ppHRegX86(i->Xin.Fp64to32.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Fp64to32.dst);
         return;
      case Xin_FpCMov:
         vex_printf("gcmov%s ", showX86CondCode(i->Xin.FpCMov.cond));
         ppHRegX86(i->Xin.FpCMov.src);
         vex_printf(",");
         ppHRegX86(i->Xin.FpCMov.dst);
         return;
      case Xin_FpLdCW:
         vex_printf("fldcw ");
         ppX86AMode(i->Xin.FpLdCW.addr);
         return;
      case Xin_FpStSW_AX:
         vex_printf("fstsw %%ax");
         return;
      case Xin_FpCmp:
         vex_printf("gcmp ");
         ppHRegX86(i->Xin.FpCmp.srcL);
         vex_printf(",");
         ppHRegX86(i->Xin.FpCmp.srcR);
         vex_printf(",");
         ppHRegX86(i->Xin.FpCmp.dst);
         break;
      case Xin_SseConst:
         vex_printf("const $0x%04x,", (Int)i->Xin.SseConst.con);
         ppHRegX86(i->Xin.SseConst.dst);
         break;
      case Xin_SseLdSt:
         vex_printf("movups ");
         if (i->Xin.SseLdSt.isLoad) {
            ppX86AMode(i->Xin.SseLdSt.addr);
            vex_printf(",");
            ppHRegX86(i->Xin.SseLdSt.reg);
         } else {
            ppHRegX86(i->Xin.SseLdSt.reg);
            vex_printf(",");
            ppX86AMode(i->Xin.SseLdSt.addr);
         }
         return;
      case Xin_SseLdzLO:
         vex_printf("movs%s ", i->Xin.SseLdzLO.sz==4 ? "s" : "d");
         ppX86AMode(i->Xin.SseLdzLO.addr);
         vex_printf(",");
         ppHRegX86(i->Xin.SseLdzLO.reg);
         return;
      case Xin_Sse32Fx4:
         vex_printf("%sps ", showX86SseOp(i->Xin.Sse32Fx4.op));
         ppHRegX86(i->Xin.Sse32Fx4.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Sse32Fx4.dst);
         return;
      case Xin_Sse32FLo:
         vex_printf("%sss ", showX86SseOp(i->Xin.Sse32FLo.op));
         ppHRegX86(i->Xin.Sse32FLo.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Sse32FLo.dst);
         return;
      case Xin_Sse64Fx2:
         vex_printf("%spd ", showX86SseOp(i->Xin.Sse64Fx2.op));
         ppHRegX86(i->Xin.Sse64Fx2.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Sse64Fx2.dst);
         return;
      case Xin_Sse64FLo:
         vex_printf("%ssd ", showX86SseOp(i->Xin.Sse64FLo.op));
         ppHRegX86(i->Xin.Sse64FLo.src);
         vex_printf(",");
         ppHRegX86(i->Xin.Sse64FLo.dst);
         return;
      case Xin_SseReRg:
         vex_printf("%s ", showX86SseOp(i->Xin.SseReRg.op));
         ppHRegX86(i->Xin.SseReRg.src);
         vex_printf(",");
         ppHRegX86(i->Xin.SseReRg.dst);
         return;
      case Xin_SseCMov:
         vex_printf("cmov%s ", showX86CondCode(i->Xin.SseCMov.cond));
         ppHRegX86(i->Xin.SseCMov.src);
         vex_printf(",");
         ppHRegX86(i->Xin.SseCMov.dst);
         return;
      case Xin_SseShuf:
         vex_printf("pshufd $0x%x,", i->Xin.SseShuf.order);
         ppHRegX86(i->Xin.SseShuf.src);
         vex_printf(",");
         ppHRegX86(i->Xin.SseShuf.dst);
         return;

      default:
         vpanic("ppX86Instr");
   }
}

/* --------- Helpers for register allocation. --------- */

void getRegUsage_X86Instr (HRegUsage* u, X86Instr* i, Bool mode64)
{
   Bool unary;
   vassert(mode64 == False);
   initHRegUsage(u);
   switch (i->tag) {
      case Xin_Alu32R:
         addRegUsage_X86RMI(u, i->Xin.Alu32R.src);
         if (i->Xin.Alu32R.op == Xalu_MOV) {
            addHRegUse(u, HRmWrite, i->Xin.Alu32R.dst);
            return;
         }
         if (i->Xin.Alu32R.op == Xalu_CMP) { 
            addHRegUse(u, HRmRead, i->Xin.Alu32R.dst);
            return;
         }
         addHRegUse(u, HRmModify, i->Xin.Alu32R.dst);
         return;
      case Xin_Alu32M:
         addRegUsage_X86RI(u, i->Xin.Alu32M.src);
         addRegUsage_X86AMode(u, i->Xin.Alu32M.dst);
         return;
      case Xin_Sh32:
         addHRegUse(u, HRmModify, i->Xin.Sh32.dst);
         if (i->Xin.Sh32.src == 0)
            addHRegUse(u, HRmRead, hregX86_ECX());
         return;
      case Xin_Test32:
         addHRegUse(u, HRmRead, i->Xin.Test32.dst);
         return;
      case Xin_Unary32:
         addHRegUse(u, HRmModify, i->Xin.Unary32.dst);
         return;
      case Xin_MulL:
         addRegUsage_X86RM(u, i->Xin.MulL.src, HRmRead);
         addHRegUse(u, HRmModify, hregX86_EAX());
         addHRegUse(u, HRmWrite, hregX86_EDX());
         return;
      case Xin_Div:
         addRegUsage_X86RM(u, i->Xin.Div.src, HRmRead);
         addHRegUse(u, HRmModify, hregX86_EAX());
         addHRegUse(u, HRmModify, hregX86_EDX());
         return;
      case Xin_Sh3232:
         addHRegUse(u, HRmRead, i->Xin.Sh3232.src);
         addHRegUse(u, HRmModify, i->Xin.Sh3232.dst);
         if (i->Xin.Sh3232.amt == 0)
            addHRegUse(u, HRmRead, hregX86_ECX());
         return;
      case Xin_Push:
         addRegUsage_X86RMI(u, i->Xin.Push.src);
         addHRegUse(u, HRmModify, hregX86_ESP());
         return;
      case Xin_Call:
         /* This is a bit subtle. */
         /* First off, claim it trashes all the caller-saved regs
            which fall within the register allocator's jurisdiction.
            These I believe to be %eax,%ecx,%edx. */
         addHRegUse(u, HRmWrite, hregX86_EAX());
         addHRegUse(u, HRmWrite, hregX86_ECX());
         addHRegUse(u, HRmWrite, hregX86_EDX());
         /* Now we have to state any parameter-carrying registers
            which might be read.  This depends on the regparmness. */
         switch (i->Xin.Call.regparms) {
            case 3: addHRegUse(u, HRmRead, hregX86_ECX()); /*fallthru*/
            case 2: addHRegUse(u, HRmRead, hregX86_EDX()); /*fallthru*/
            case 1: addHRegUse(u, HRmRead, hregX86_EAX()); break;
            case 0: break;
            default: vpanic("getRegUsage_X86Instr:Call:regparms");
         }
         /* Finally, there is the issue that the insn trashes a
            register because the literal target address has to be
            loaded into a register.  Fortunately, for the 0/1/2
            regparm case, we can use EAX, EDX and ECX respectively, so
            this does not cause any further damage.  For the 3-regparm
            case, we'll have to choose another register arbitrarily --
            since A, D and C are used for parameters -- and so we might
            as well choose EDI. */
         if (i->Xin.Call.regparms == 3)
            addHRegUse(u, HRmWrite, hregX86_EDI());
         /* Upshot of this is that the assembler really must observe
            the here-stated convention of which register to use as an
            address temporary, depending on the regparmness: 0==EAX,
            1==EDX, 2==ECX, 3==EDI. */
         return;
      case Xin_Goto:
         addRegUsage_X86RI(u, i->Xin.Goto.dst);
         addHRegUse(u, HRmWrite, hregX86_EAX()); /* used for next guest addr */
         addHRegUse(u, HRmWrite, hregX86_EDX()); /* used for dispatcher addr */
         if (i->Xin.Goto.jk != Ijk_Boring
             && i->Xin.Goto.jk != Ijk_Call
             && i->Xin.Goto.jk != Ijk_Ret)
            /* note, this is irrelevant since ebp is not actually
               available to the allocator.  But still .. */
            addHRegUse(u, HRmWrite, hregX86_EBP());
         return;
      case Xin_CMov32:
         addRegUsage_X86RM(u, i->Xin.CMov32.src, HRmRead);
         addHRegUse(u, HRmModify, i->Xin.CMov32.dst);
         return;
      case Xin_LoadEX:
         addRegUsage_X86AMode(u, i->Xin.LoadEX.src);
         addHRegUse(u, HRmWrite, i->Xin.LoadEX.dst);
         return;
      case Xin_Store:
         addHRegUse(u, HRmRead, i->Xin.Store.src);
         addRegUsage_X86AMode(u, i->Xin.Store.dst);
         return;
      case Xin_Set32:
         addHRegUse(u, HRmWrite, i->Xin.Set32.dst);
         return;
      case Xin_Bsfr32:
         addHRegUse(u, HRmRead, i->Xin.Bsfr32.src);
         addHRegUse(u, HRmWrite, i->Xin.Bsfr32.dst);
         return;
      case Xin_MFence:
         return;
      case Xin_FpUnary:
         addHRegUse(u, HRmRead, i->Xin.FpUnary.src);
         addHRegUse(u, HRmWrite, i->Xin.FpUnary.dst);
         return;
      case Xin_FpBinary:
         addHRegUse(u, HRmRead, i->Xin.FpBinary.srcL);
         addHRegUse(u, HRmRead, i->Xin.FpBinary.srcR);
         addHRegUse(u, HRmWrite, i->Xin.FpBinary.dst);
         return;
      case Xin_FpLdSt:
         addRegUsage_X86AMode(u, i->Xin.FpLdSt.addr);
         addHRegUse(u, i->Xin.FpLdSt.isLoad ? HRmWrite : HRmRead,
                       i->Xin.FpLdSt.reg);
         return;
      case Xin_FpLdStI:
         addRegUsage_X86AMode(u, i->Xin.FpLdStI.addr);
         addHRegUse(u, i->Xin.FpLdStI.isLoad ? HRmWrite : HRmRead,
                       i->Xin.FpLdStI.reg);
         return;
      case Xin_Fp64to32:
         addHRegUse(u, HRmRead,  i->Xin.Fp64to32.src);
         addHRegUse(u, HRmWrite, i->Xin.Fp64to32.dst);
         return;
      case Xin_FpCMov:
         addHRegUse(u, HRmRead,   i->Xin.FpCMov.src);
         addHRegUse(u, HRmModify, i->Xin.FpCMov.dst);
         return;
      case Xin_FpLdCW:
         addRegUsage_X86AMode(u, i->Xin.FpLdCW.addr);
         return;
      case Xin_FpStSW_AX:
         addHRegUse(u, HRmWrite, hregX86_EAX());
         return;
      case Xin_FpCmp:
         addHRegUse(u, HRmRead, i->Xin.FpCmp.srcL);
         addHRegUse(u, HRmRead, i->Xin.FpCmp.srcR);
         addHRegUse(u, HRmWrite, i->Xin.FpCmp.dst);
         addHRegUse(u, HRmWrite, hregX86_EAX());
         return;
      case Xin_SseLdSt:
         addRegUsage_X86AMode(u, i->Xin.SseLdSt.addr);
         addHRegUse(u, i->Xin.SseLdSt.isLoad ? HRmWrite : HRmRead,
                       i->Xin.SseLdSt.reg);
         return;
      case Xin_SseLdzLO:
         addRegUsage_X86AMode(u, i->Xin.SseLdzLO.addr);
         addHRegUse(u, HRmWrite, i->Xin.SseLdzLO.reg);
         return;
      case Xin_SseConst:
         addHRegUse(u, HRmWrite, i->Xin.SseConst.dst);
         return;
      case Xin_Sse32Fx4:
         vassert(i->Xin.Sse32Fx4.op != Xsse_MOV);
         unary = toBool( i->Xin.Sse32Fx4.op == Xsse_RCPF
                         || i->Xin.Sse32Fx4.op == Xsse_RSQRTF
                         || i->Xin.Sse32Fx4.op == Xsse_SQRTF );
         addHRegUse(u, HRmRead, i->Xin.Sse32Fx4.src);
         addHRegUse(u, unary ? HRmWrite : HRmModify, 
                       i->Xin.Sse32Fx4.dst);
         return;
      case Xin_Sse32FLo:
         vassert(i->Xin.Sse32FLo.op != Xsse_MOV);
         unary = toBool( i->Xin.Sse32FLo.op == Xsse_RCPF
                         || i->Xin.Sse32FLo.op == Xsse_RSQRTF
                         || i->Xin.Sse32FLo.op == Xsse_SQRTF );
         addHRegUse(u, HRmRead, i->Xin.Sse32FLo.src);
         addHRegUse(u, unary ? HRmWrite : HRmModify, 
                       i->Xin.Sse32FLo.dst);
         return;
      case Xin_Sse64Fx2:
         vassert(i->Xin.Sse64Fx2.op != Xsse_MOV);
         unary = toBool( i->Xin.Sse64Fx2.op == Xsse_RCPF
                         || i->Xin.Sse64Fx2.op == Xsse_RSQRTF
                         || i->Xin.Sse64Fx2.op == Xsse_SQRTF );
         addHRegUse(u, HRmRead, i->Xin.Sse64Fx2.src);
         addHRegUse(u, unary ? HRmWrite : HRmModify, 
                       i->Xin.Sse64Fx2.dst);
         return;
      case Xin_Sse64FLo:
         vassert(i->Xin.Sse64FLo.op != Xsse_MOV);
         unary = toBool( i->Xin.Sse64FLo.op == Xsse_RCPF
                         || i->Xin.Sse64FLo.op == Xsse_RSQRTF
                         || i->Xin.Sse64FLo.op == Xsse_SQRTF );
         addHRegUse(u, HRmRead, i->Xin.Sse64FLo.src);
         addHRegUse(u, unary ? HRmWrite : HRmModify, 
                       i->Xin.Sse64FLo.dst);
         return;
      case Xin_SseReRg:
         if (i->Xin.SseReRg.op == Xsse_XOR
             && i->Xin.SseReRg.src == i->Xin.SseReRg.dst) {
            /* reg-alloc needs to understand 'xor r,r' as a write of r */
            /* (as opposed to a rite of passage :-) */
            addHRegUse(u, HRmWrite, i->Xin.SseReRg.dst);
         } else {
            addHRegUse(u, HRmRead, i->Xin.SseReRg.src);
            addHRegUse(u, i->Xin.SseReRg.op == Xsse_MOV 
                             ? HRmWrite : HRmModify, 
                          i->Xin.SseReRg.dst);
         }
         return;
      case Xin_SseCMov:
         addHRegUse(u, HRmRead,   i->Xin.SseCMov.src);
         addHRegUse(u, HRmModify, i->Xin.SseCMov.dst);
         return;
      case Xin_SseShuf:
         addHRegUse(u, HRmRead,  i->Xin.SseShuf.src);
         addHRegUse(u, HRmWrite, i->Xin.SseShuf.dst);
         return;
      default:
         ppX86Instr(i, False);
         vpanic("getRegUsage_X86Instr");
   }
}

/* local helper */
static void mapReg( HRegRemap* m, HReg* r )
{
   *r = lookupHRegRemap(m, *r);
}

void mapRegs_X86Instr ( HRegRemap* m, X86Instr* i, Bool mode64 )
{
   vassert(mode64 == False);
   switch (i->tag) {
      case Xin_Alu32R:
         mapRegs_X86RMI(m, i->Xin.Alu32R.src);
         mapReg(m, &i->Xin.Alu32R.dst);
         return;
      case Xin_Alu32M:
         mapRegs_X86RI(m, i->Xin.Alu32M.src);
         mapRegs_X86AMode(m, i->Xin.Alu32M.dst);
         return;
      case Xin_Sh32:
         mapReg(m, &i->Xin.Sh32.dst);
         return;
      case Xin_Test32:
         mapReg(m, &i->Xin.Test32.dst);
         return;
      case Xin_Unary32:
         mapReg(m, &i->Xin.Unary32.dst);
         return;
      case Xin_MulL:
         mapRegs_X86RM(m, i->Xin.MulL.src);
         return;
      case Xin_Div:
         mapRegs_X86RM(m, i->Xin.Div.src);
         return;
      case Xin_Sh3232:
         mapReg(m, &i->Xin.Sh3232.src);
         mapReg(m, &i->Xin.Sh3232.dst);
         return;
      case Xin_Push:
         mapRegs_X86RMI(m, i->Xin.Push.src);
         return;
      case Xin_Call:
         return;
      case Xin_Goto:
         mapRegs_X86RI(m, i->Xin.Goto.dst);
         return;
      case Xin_CMov32:
         mapRegs_X86RM(m, i->Xin.CMov32.src);
         mapReg(m, &i->Xin.CMov32.dst);
         return;
      case Xin_LoadEX:
         mapRegs_X86AMode(m, i->Xin.LoadEX.src);
         mapReg(m, &i->Xin.LoadEX.dst);
         return;
      case Xin_Store:
         mapReg(m, &i->Xin.Store.src);
         mapRegs_X86AMode(m, i->Xin.Store.dst);
         return;
      case Xin_Set32:
         mapReg(m, &i->Xin.Set32.dst);
         return;
      case Xin_Bsfr32:
         mapReg(m, &i->Xin.Bsfr32.src);
         mapReg(m, &i->Xin.Bsfr32.dst);
         return;
      case Xin_MFence:
         return;
      case Xin_FpUnary:
         mapReg(m, &i->Xin.FpUnary.src);
         mapReg(m, &i->Xin.FpUnary.dst);
         return;
      case Xin_FpBinary:
         mapReg(m, &i->Xin.FpBinary.srcL);
         mapReg(m, &i->Xin.FpBinary.srcR);
         mapReg(m, &i->Xin.FpBinary.dst);
         return;
      case Xin_FpLdSt:
         mapRegs_X86AMode(m, i->Xin.FpLdSt.addr);
         mapReg(m, &i->Xin.FpLdSt.reg);
         return;
      case Xin_FpLdStI:
         mapRegs_X86AMode(m, i->Xin.FpLdStI.addr);
         mapReg(m, &i->Xin.FpLdStI.reg);
         return;
      case Xin_Fp64to32:
         mapReg(m, &i->Xin.Fp64to32.src);
         mapReg(m, &i->Xin.Fp64to32.dst);
         return;
      case Xin_FpCMov:
         mapReg(m, &i->Xin.FpCMov.src);
         mapReg(m, &i->Xin.FpCMov.dst);
         return;
      case Xin_FpLdCW:
         mapRegs_X86AMode(m, i->Xin.FpLdCW.addr);
         return;
      case Xin_FpStSW_AX:
         return;
      case Xin_FpCmp:
         mapReg(m, &i->Xin.FpCmp.srcL);
         mapReg(m, &i->Xin.FpCmp.srcR);
         mapReg(m, &i->Xin.FpCmp.dst);
         return;
      case Xin_SseConst:
         mapReg(m, &i->Xin.SseConst.dst);
         return;
      case Xin_SseLdSt:
         mapReg(m, &i->Xin.SseLdSt.reg);
         mapRegs_X86AMode(m, i->Xin.SseLdSt.addr);
         break;
      case Xin_SseLdzLO:
         mapReg(m, &i->Xin.SseLdzLO.reg);
         mapRegs_X86AMode(m, i->Xin.SseLdzLO.addr);
         break;
      case Xin_Sse32Fx4:
         mapReg(m, &i->Xin.Sse32Fx4.src);
         mapReg(m, &i->Xin.Sse32Fx4.dst);
         return;
      case Xin_Sse32FLo:
         mapReg(m, &i->Xin.Sse32FLo.src);
         mapReg(m, &i->Xin.Sse32FLo.dst);
         return;
      case Xin_Sse64Fx2:
         mapReg(m, &i->Xin.Sse64Fx2.src);
         mapReg(m, &i->Xin.Sse64Fx2.dst);
         return;
      case Xin_Sse64FLo:
         mapReg(m, &i->Xin.Sse64FLo.src);
         mapReg(m, &i->Xin.Sse64FLo.dst);
         return;
      case Xin_SseReRg:
         mapReg(m, &i->Xin.SseReRg.src);
         mapReg(m, &i->Xin.SseReRg.dst);
         return;
      case Xin_SseCMov:
         mapReg(m, &i->Xin.SseCMov.src);
         mapReg(m, &i->Xin.SseCMov.dst);
         return;
      case Xin_SseShuf:
         mapReg(m, &i->Xin.SseShuf.src);
         mapReg(m, &i->Xin.SseShuf.dst);
         return;
      default:
         ppX86Instr(i, mode64);
         vpanic("mapRegs_X86Instr");
   }
}

/* Figure out if i represents a reg-reg move, and if so assign the
   source and destination to *src and *dst.  If in doubt say No.  Used
   by the register allocator to do move coalescing. 
*/
Bool isMove_X86Instr ( X86Instr* i, HReg* src, HReg* dst )
{
   /* Moves between integer regs */
   if (i->tag == Xin_Alu32R) {
      if (i->Xin.Alu32R.op != Xalu_MOV)
         return False;
      if (i->Xin.Alu32R.src->tag != Xrmi_Reg)
         return False;
      *src = i->Xin.Alu32R.src->Xrmi.Reg.reg;
      *dst = i->Xin.Alu32R.dst;
      return True;
   }
   /* Moves between FP regs */
   if (i->tag == Xin_FpUnary) {
      if (i->Xin.FpUnary.op != Xfp_MOV)
         return False;
      *src = i->Xin.FpUnary.src;
      *dst = i->Xin.FpUnary.dst;
      return True;
   }
   if (i->tag == Xin_SseReRg) {
      if (i->Xin.SseReRg.op != Xsse_MOV)
         return False;
      *src = i->Xin.SseReRg.src;
      *dst = i->Xin.SseReRg.dst;
      return True;
   }
   return False;
}


/* Generate x86 spill/reload instructions under the direction of the
   register allocator.  Note it's critical these don't write the
   condition codes. */

X86Instr* genSpill_X86 ( HReg rreg, Int offsetB, Bool mode64 )
{
   X86AMode* am;
   vassert(offsetB >= 0);
   vassert(!hregIsVirtual(rreg));
   vassert(mode64 == False);
   am = X86AMode_IR(offsetB, hregX86_EBP());

   switch (hregClass(rreg)) {
      case HRcInt32:
         return X86Instr_Alu32M ( Xalu_MOV, X86RI_Reg(rreg), am );
      case HRcFlt64:
         return X86Instr_FpLdSt ( False/*store*/, 8, rreg, am );
      case HRcVec128:
         return X86Instr_SseLdSt ( False/*store*/, rreg, am );
      default: 
         ppHRegClass(hregClass(rreg));
         vpanic("genSpill_X86: unimplemented regclass");
   }
}

X86Instr* genReload_X86 ( HReg rreg, Int offsetB, Bool mode64 )
{
   X86AMode* am;
   vassert(offsetB >= 0);
   vassert(!hregIsVirtual(rreg));
   vassert(mode64 == False);
   am = X86AMode_IR(offsetB, hregX86_EBP());
   switch (hregClass(rreg)) {
      case HRcInt32:
         return X86Instr_Alu32R ( Xalu_MOV, X86RMI_Mem(am), rreg );
      case HRcFlt64:
         return X86Instr_FpLdSt ( True/*load*/, 8, rreg, am );
      case HRcVec128:
         return X86Instr_SseLdSt ( True/*load*/, rreg, am );
      default: 
         ppHRegClass(hregClass(rreg));
         vpanic("genReload_X86: unimplemented regclass");
   }
}


/* --------- The x86 assembler (bleh.) --------- */

static UChar iregNo ( HReg r )
{
   UInt n;
   vassert(hregClass(r) == HRcInt32);
   vassert(!hregIsVirtual(r));
   n = hregNumber(r);
   vassert(n <= 7);
   return toUChar(n);
}

static UInt fregNo ( HReg r )
{
   UInt n;
   vassert(hregClass(r) == HRcFlt64);
   vassert(!hregIsVirtual(r));
   n = hregNumber(r);
   vassert(n <= 5);
   return n;
}

static UInt vregNo ( HReg r )
{
   UInt n;
   vassert(hregClass(r) == HRcVec128);
   vassert(!hregIsVirtual(r));
   n = hregNumber(r);
   vassert(n <= 7);
   return n;
}

static UChar mkModRegRM ( UChar mod, UChar reg, UChar regmem )
{
   return toUChar( ((mod & 3) << 6) 
                   | ((reg & 7) << 3) 
                   | (regmem & 7) );
}

static UChar mkSIB ( Int shift, Int regindex, Int regbase )
{
   return toUChar( ((shift & 3) << 6) 
                   | ((regindex & 7) << 3) 
                   | (regbase & 7) );
}

static UChar* emit32 ( UChar* p, UInt w32 )
{
   *p++ = toUChar( w32        & 0x000000FF);
   *p++ = toUChar((w32 >>  8) & 0x000000FF);
   *p++ = toUChar((w32 >> 16) & 0x000000FF);
   *p++ = toUChar((w32 >> 24) & 0x000000FF);
   return p;
}

/* Does a sign-extend of the lowest 8 bits give 
   the original number? */
static Bool fits8bits ( UInt w32 )
{
   Int i32 = (Int)w32;
   return toBool(i32 == ((i32 << 24) >> 24));
}


/* Forming mod-reg-rm bytes and scale-index-base bytes.

     greg,  0(ereg)    |  ereg != ESP && ereg != EBP
                       =  00 greg ereg

     greg,  d8(ereg)   |  ereg != ESP
                       =  01 greg ereg, d8

     greg,  d32(ereg)  |  ereg != ESP
                       =  10 greg ereg, d32

     greg,  d8(%esp)   =  01 greg 100, 0x24, d8

     -----------------------------------------------

     greg,  d8(base,index,scale)  
               |  index != ESP
               =  01 greg 100, scale index base, d8

     greg,  d32(base,index,scale)
               |  index != ESP
               =  10 greg 100, scale index base, d32
*/
static UChar* doAMode_M ( UChar* p, HReg greg, X86AMode* am ) 
{
   if (am->tag == Xam_IR) {
      if (am->Xam.IR.imm == 0 
          && am->Xam.IR.reg != hregX86_ESP()
          && am->Xam.IR.reg != hregX86_EBP() ) {
         *p++ = mkModRegRM(0, iregNo(greg), iregNo(am->Xam.IR.reg));
         return p;
      }
      if (fits8bits(am->Xam.IR.imm)
          && am->Xam.IR.reg != hregX86_ESP()) {
         *p++ = mkModRegRM(1, iregNo(greg), iregNo(am->Xam.IR.reg));
         *p++ = toUChar(am->Xam.IR.imm & 0xFF);
         return p;
      }
      if (am->Xam.IR.reg != hregX86_ESP()) {
         *p++ = mkModRegRM(2, iregNo(greg), iregNo(am->Xam.IR.reg));
         p = emit32(p, am->Xam.IR.imm);
         return p;
      }
      if (am->Xam.IR.reg == hregX86_ESP()
          && fits8bits(am->Xam.IR.imm)) {
 	 *p++ = mkModRegRM(1, iregNo(greg), 4);
         *p++ = 0x24;
         *p++ = toUChar(am->Xam.IR.imm & 0xFF);
         return p;
      }
      ppX86AMode(am);
      vpanic("doAMode_M: can't emit amode IR");
      /*NOTREACHED*/
   }
   if (am->tag == Xam_IRRS) {
      if (fits8bits(am->Xam.IRRS.imm)
          && am->Xam.IRRS.index != hregX86_ESP()) {
         *p++ = mkModRegRM(1, iregNo(greg), 4);
         *p++ = mkSIB(am->Xam.IRRS.shift, am->Xam.IRRS.index, 
                                          am->Xam.IRRS.base);
         *p++ = toUChar(am->Xam.IRRS.imm & 0xFF);
         return p;
      }
      if (am->Xam.IRRS.index != hregX86_ESP()) {
         *p++ = mkModRegRM(2, iregNo(greg), 4);
         *p++ = mkSIB(am->Xam.IRRS.shift, am->Xam.IRRS.index,
                                          am->Xam.IRRS.base);
         p = emit32(p, am->Xam.IRRS.imm);
         return p;
      }
      ppX86AMode(am);
      vpanic("doAMode_M: can't emit amode IRRS");
      /*NOTREACHED*/
   }
   vpanic("doAMode_M: unknown amode");
   /*NOTREACHED*/
}


/* Emit a mod-reg-rm byte when the rm bit denotes a reg. */
static UChar* doAMode_R ( UChar* p, HReg greg, HReg ereg ) 
{
   *p++ = mkModRegRM(3, iregNo(greg), iregNo(ereg));
   return p;
}


/* Emit ffree %st(7) */
static UChar* do_ffree_st7 ( UChar* p )
{
   *p++ = 0xDD;
   *p++ = 0xC7;
   return p;
}

/* Emit fstp %st(i), 1 <= i <= 7 */
static UChar* do_fstp_st ( UChar* p, Int i )
{
   vassert(1 <= i && i <= 7);
   *p++ = 0xDD;
   *p++ = toUChar(0xD8+i);
   return p;
}

/* Emit fld %st(i), 0 <= i <= 6 */
static UChar* do_fld_st ( UChar* p, Int i )
{
   vassert(0 <= i && i <= 6);
   *p++ = 0xD9;
   *p++ = toUChar(0xC0+i);
   return p;
}

/* Emit f<op> %st(0) */
static UChar* do_fop1_st ( UChar* p, X86FpOp op )
{
   switch (op) {
      case Xfp_NEG:    *p++ = 0xD9; *p++ = 0xE0; break;
      case Xfp_ABS:    *p++ = 0xD9; *p++ = 0xE1; break;
      case Xfp_SQRT:   *p++ = 0xD9; *p++ = 0xFA; break;
      case Xfp_ROUND:  *p++ = 0xD9; *p++ = 0xFC; break;
      case Xfp_SIN:    *p++ = 0xD9; *p++ = 0xFE; break;
      case Xfp_COS:    *p++ = 0xD9; *p++ = 0xFF; break;
      case Xfp_2XM1:   *p++ = 0xD9; *p++ = 0xF0; break;
      case Xfp_MOV:    break;
      case Xfp_TAN:    p = do_ffree_st7(p); /* since fptan pushes 1.0 */
                       *p++ = 0xD9; *p++ = 0xF2; /* fptan */
                       *p++ = 0xD9; *p++ = 0xF7; /* fincstp */
                       break;
      default: vpanic("do_fop1_st: unknown op");
   }
   return p;
}

/* Emit f<op> %st(i), 1 <= i <= 5 */
static UChar* do_fop2_st ( UChar* p, X86FpOp op, Int i )
{
#  define fake(_n) mkHReg((_n), HRcInt32, False)
   Int subopc;
   switch (op) {
      case Xfp_ADD: subopc = 0; break;
      case Xfp_SUB: subopc = 4; break;
      case Xfp_MUL: subopc = 1; break;
      case Xfp_DIV: subopc = 6; break;
      default: vpanic("do_fop2_st: unknown op");
   }
   *p++ = 0xD8;
   p    = doAMode_R(p, fake(subopc), fake(i));
   return p;
#  undef fake
}

/* Push a 32-bit word on the stack.  The word depends on tags[3:0];
each byte is either 0x00 or 0xFF depending on the corresponding bit in tags[].
*/
static UChar* push_word_from_tags ( UChar* p, UShort tags )
{
   UInt w;
   vassert(0 == (tags & ~0xF));
   if (tags == 0) {
      /* pushl $0x00000000 */
      *p++ = 0x6A;
      *p++ = 0x00;
   }
   else 
   /* pushl $0xFFFFFFFF */
   if (tags == 0xF) {
      *p++ = 0x6A;
      *p++ = 0xFF;
   } else {
      vassert(0); /* awaiting test case */
      w = 0;
      if (tags & 1) w |= 0x000000FF;
      if (tags & 2) w |= 0x0000FF00;
      if (tags & 4) w |= 0x00FF0000;
      if (tags & 8) w |= 0xFF000000;
      *p++ = 0x68;
      p = emit32(p, w);
   }
   return p;
}

/* Emit an instruction into buf and return the number of bytes used.
   Note that buf is not the insn's final place, and therefore it is
   imperative to emit position-independent code. */

Int emit_X86Instr ( UChar* buf, Int nbuf, X86Instr* i, 
                    Bool mode64, void* dispatch )
{
   UInt irno, opc, opc_rr, subopc_imm, opc_imma, opc_cl, opc_imm, subopc;

   UInt   xtra;
   UChar* p = &buf[0];
   UChar* ptmp;
   vassert(nbuf >= 32);
   vassert(mode64 == False);

   /* Wrap an integer as a int register, for use assembling
      GrpN insns, in which the greg field is used as a sub-opcode
      and does not really contain a register. */
#  define fake(_n) mkHReg((_n), HRcInt32, False)

   /* vex_printf("asm  ");ppX86Instr(i, mode64); vex_printf("\n"); */

   switch (i->tag) {

   case Xin_Alu32R:
      /* Deal specially with MOV */
      if (i->Xin.Alu32R.op == Xalu_MOV) {
         switch (i->Xin.Alu32R.src->tag) {
            case Xrmi_Imm:
               *p++ = toUChar(0xB8 + iregNo(i->Xin.Alu32R.dst));
               p = emit32(p, i->Xin.Alu32R.src->Xrmi.Imm.imm32);
               goto done;
            case Xrmi_Reg:
               *p++ = 0x89;
               p = doAMode_R(p, i->Xin.Alu32R.src->Xrmi.Reg.reg,
                                i->Xin.Alu32R.dst);
               goto done;
            case Xrmi_Mem:
               *p++ = 0x8B;
               p = doAMode_M(p, i->Xin.Alu32R.dst, 
                                i->Xin.Alu32R.src->Xrmi.Mem.am);
               goto done;
            default:
               goto bad;
         }
      }
      /* MUL */
      if (i->Xin.Alu32R.op == Xalu_MUL) {
         switch (i->Xin.Alu32R.src->tag) {
            case Xrmi_Reg:
               *p++ = 0x0F;
               *p++ = 0xAF;
               p = doAMode_R(p, i->Xin.Alu32R.dst,
                                i->Xin.Alu32R.src->Xrmi.Reg.reg);
               goto done;
            case Xrmi_Mem:
               *p++ = 0x0F;
               *p++ = 0xAF;
               p = doAMode_M(p, i->Xin.Alu32R.dst,
                                i->Xin.Alu32R.src->Xrmi.Mem.am);
               goto done;
            case Xrmi_Imm:
               if (fits8bits(i->Xin.Alu32R.src->Xrmi.Imm.imm32)) {
                  *p++ = 0x6B;
                  p = doAMode_R(p, i->Xin.Alu32R.dst, i->Xin.Alu32R.dst);
                  *p++ = toUChar(0xFF & i->Xin.Alu32R.src->Xrmi.Imm.imm32);
               } else {
                  *p++ = 0x69;
                  p = doAMode_R(p, i->Xin.Alu32R.dst, i->Xin.Alu32R.dst);
                  p = emit32(p, i->Xin.Alu32R.src->Xrmi.Imm.imm32);
               }
               goto done;
            default:
               goto bad;
         }
      }
      /* ADD/SUB/ADC/SBB/AND/OR/XOR/CMP */
      opc = opc_rr = subopc_imm = opc_imma = 0;
      switch (i->Xin.Alu32R.op) {
         case Xalu_ADC: opc = 0x13; opc_rr = 0x11; 
                        subopc_imm = 2; opc_imma = 0x15; break;
         case Xalu_ADD: opc = 0x03; opc_rr = 0x01; 
                        subopc_imm = 0; opc_imma = 0x05; break;
         case Xalu_SUB: opc = 0x2B; opc_rr = 0x29; 
                        subopc_imm = 5; opc_imma = 0x2D; break;
         case Xalu_SBB: opc = 0x1B; opc_rr = 0x19; 
                        subopc_imm = 3; opc_imma = 0x1D; break;
         case Xalu_AND: opc = 0x23; opc_rr = 0x21; 
                        subopc_imm = 4; opc_imma = 0x25; break;
         case Xalu_XOR: opc = 0x33; opc_rr = 0x31; 
                        subopc_imm = 6; opc_imma = 0x35; break;
         case Xalu_OR:  opc = 0x0B; opc_rr = 0x09; 
                        subopc_imm = 1; opc_imma = 0x0D; break;
         case Xalu_CMP: opc = 0x3B; opc_rr = 0x39; 
                        subopc_imm = 7; opc_imma = 0x3D; break;
         default: goto bad;
      }
      switch (i->Xin.Alu32R.src->tag) {
         case Xrmi_Imm:
            if (i->Xin.Alu32R.dst == hregX86_EAX()
                && !fits8bits(i->Xin.Alu32R.src->Xrmi.Imm.imm32)) {
               *p++ = toUChar(opc_imma);
               p = emit32(p, i->Xin.Alu32R.src->Xrmi.Imm.imm32);
            } else
            if (fits8bits(i->Xin.Alu32R.src->Xrmi.Imm.imm32)) {
               *p++ = 0x83; 
               p    = doAMode_R(p, fake(subopc_imm), i->Xin.Alu32R.dst);
               *p++ = toUChar(0xFF & i->Xin.Alu32R.src->Xrmi.Imm.imm32);
            } else {
               *p++ = 0x81; 
               p    = doAMode_R(p, fake(subopc_imm), i->Xin.Alu32R.dst);
               p    = emit32(p, i->Xin.Alu32R.src->Xrmi.Imm.imm32);
            }
            goto done;
         case Xrmi_Reg:
            *p++ = toUChar(opc_rr);
            p = doAMode_R(p, i->Xin.Alu32R.src->Xrmi.Reg.reg,
                             i->Xin.Alu32R.dst);
            goto done;
         case Xrmi_Mem:
            *p++ = toUChar(opc);
            p = doAMode_M(p, i->Xin.Alu32R.dst,
                             i->Xin.Alu32R.src->Xrmi.Mem.am);
            goto done;
         default: 
            goto bad;
      }
      break;

   case Xin_Alu32M:
      /* Deal specially with MOV */
      if (i->Xin.Alu32M.op == Xalu_MOV) {
         switch (i->Xin.Alu32M.src->tag) {
            case Xri_Reg:
               *p++ = 0x89;
               p = doAMode_M(p, i->Xin.Alu32M.src->Xri.Reg.reg,
                                i->Xin.Alu32M.dst);
               goto done;
            case Xri_Imm:
               *p++ = 0xC7;
               p = doAMode_M(p, fake(0), i->Xin.Alu32M.dst);
               p = emit32(p, i->Xin.Alu32M.src->Xri.Imm.imm32);
               goto done;
            default: 
               goto bad;
         }
      }
      /* ADD/SUB/ADC/SBB/AND/OR/XOR/CMP.  MUL is not
         allowed here. */
      opc = subopc_imm = opc_imma = 0;
      switch (i->Xin.Alu32M.op) {
         case Xalu_ADD: opc = 0x01; subopc_imm = 0; break;
         case Xalu_SUB: opc = 0x29; subopc_imm = 5; break;
         default: goto bad;
      }
      switch (i->Xin.Alu32M.src->tag) {
         case Xri_Reg:
            *p++ = toUChar(opc);
            p = doAMode_M(p, i->Xin.Alu32M.src->Xri.Reg.reg,
                             i->Xin.Alu32M.dst);
            goto done;
         case Xri_Imm:
            if (fits8bits(i->Xin.Alu32M.src->Xri.Imm.imm32)) {
               *p++ = 0x83;
               p    = doAMode_M(p, fake(subopc_imm), i->Xin.Alu32M.dst);
               *p++ = toUChar(0xFF & i->Xin.Alu32M.src->Xri.Imm.imm32);
               goto done;
            } else {
               *p++ = 0x81;
               p    = doAMode_M(p, fake(subopc_imm), i->Xin.Alu32M.dst);
               p    = emit32(p, i->Xin.Alu32M.src->Xri.Imm.imm32);
               goto done;
            }
         default: 
            goto bad;
      }
      break;

   case Xin_Sh32:
      opc_cl = opc_imm = subopc = 0;
      switch (i->Xin.Sh32.op) {
         case Xsh_SHR: opc_cl = 0xD3; opc_imm = 0xC1; subopc = 5; break;
         case Xsh_SAR: opc_cl = 0xD3; opc_imm = 0xC1; subopc = 7; break;
         case Xsh_SHL: opc_cl = 0xD3; opc_imm = 0xC1; subopc = 4; break;
         default: goto bad;
      }
      if (i->Xin.Sh32.src == 0) {
         *p++ = toUChar(opc_cl);
         p = doAMode_R(p, fake(subopc), i->Xin.Sh32.dst);
      } else {
         *p++ = toUChar(opc_imm);
         p = doAMode_R(p, fake(subopc), i->Xin.Sh32.dst);
         *p++ = (UChar)(i->Xin.Sh32.src);
      }
      goto done;

   case Xin_Test32:
      /* testl $imm32, %reg */
      *p++ = 0xF7;
      p = doAMode_R(p, fake(0), i->Xin.Test32.dst);
      p = emit32(p, i->Xin.Test32.imm32);
      goto done;

   case Xin_Unary32:
      if (i->Xin.Unary32.op == Xun_NOT) {
         *p++ = 0xF7;
         p = doAMode_R(p, fake(2), i->Xin.Unary32.dst);
         goto done;
      }
      if (i->Xin.Unary32.op == Xun_NEG) {
         *p++ = 0xF7;
         p = doAMode_R(p, fake(3), i->Xin.Unary32.dst);
         goto done;
      }
      break;

   case Xin_MulL:
      subopc = i->Xin.MulL.syned ? 5 : 4;
      *p++ = 0xF7;
      switch (i->Xin.MulL.src->tag)  {
         case Xrm_Mem:
            p = doAMode_M(p, fake(subopc),
                             i->Xin.MulL.src->Xrm.Mem.am);
            goto done;
         case Xrm_Reg:
            p = doAMode_R(p, fake(subopc), 
                             i->Xin.MulL.src->Xrm.Reg.reg);
            goto done;
         default:
            goto bad;
      }
      break;

   case Xin_Div:
      subopc = i->Xin.Div.syned ? 7 : 6;
      *p++ = 0xF7;
      switch (i->Xin.Div.src->tag)  {
         case Xrm_Mem:
            p = doAMode_M(p, fake(subopc),
                             i->Xin.Div.src->Xrm.Mem.am);
            goto done;
         case Xrm_Reg:
            p = doAMode_R(p, fake(subopc), 
                             i->Xin.Div.src->Xrm.Reg.reg);
            goto done;
         default:
            goto bad;
      }
      break;

   case Xin_Sh3232:
      vassert(i->Xin.Sh3232.op == Xsh_SHL || i->Xin.Sh3232.op == Xsh_SHR);
      if (i->Xin.Sh3232.amt == 0) {
         /* shldl/shrdl by %cl */
         *p++ = 0x0F;
         if (i->Xin.Sh3232.op == Xsh_SHL) {
            *p++ = 0xA5;
         } else {
            *p++ = 0xAD;
         }
         p = doAMode_R(p, i->Xin.Sh3232.src, i->Xin.Sh3232.dst);
         goto done;
      }
      break;

   case Xin_Push:
      switch (i->Xin.Push.src->tag) {
         case Xrmi_Mem: 
            *p++ = 0xFF;
            p = doAMode_M(p, fake(6), i->Xin.Push.src->Xrmi.Mem.am);
            goto done;
         case Xrmi_Imm:
            *p++ = 0x68;
            p = emit32(p, i->Xin.Push.src->Xrmi.Imm.imm32);
            goto done;
         case Xrmi_Reg:
            *p++ = toUChar(0x50 + iregNo(i->Xin.Push.src->Xrmi.Reg.reg));
            goto done;
        default: 
            goto bad;
      }

   case Xin_Call:
      /* See detailed comment for Xin_Call in getRegUsage_X86Instr above
         for explanation of this. */
      switch (i->Xin.Call.regparms) {
         case 0: irno = iregNo(hregX86_EAX()); break;
         case 1: irno = iregNo(hregX86_EDX()); break;
         case 2: irno = iregNo(hregX86_ECX()); break;
         case 3: irno = iregNo(hregX86_EDI()); break;
         default: vpanic(" emit_X86Instr:call:regparms");
      }
      /* jump over the following two insns if the condition does not
         hold */
      if (i->Xin.Call.cond != Xcc_ALWAYS) {
         *p++ = toUChar(0x70 + (0xF & (i->Xin.Call.cond ^ 1)));
         *p++ = 0x07; /* 7 bytes in the next two insns */
      }
      /* movl $target, %tmp */
      *p++ = toUChar(0xB8 + irno);
      p = emit32(p, i->Xin.Call.target);
      /* call *%tmp */
      *p++ = 0xFF;
      *p++ = toUChar(0xD0 + irno);
      goto done;

   case Xin_Goto:
      /* Use ptmp for backpatching conditional jumps. */
      ptmp = NULL;

      /* First off, if this is conditional, create a conditional
	 jump over the rest of it. */
      if (i->Xin.Goto.cond != Xcc_ALWAYS) {
         /* jmp fwds if !condition */
         *p++ = toUChar(0x70 + (0xF & (i->Xin.Goto.cond ^ 1)));
         ptmp = p; /* fill in this bit later */
         *p++ = 0; /* # of bytes to jump over; don't know how many yet. */
      }

      /* If a non-boring, set %ebp (the guest state pointer)
         appropriately. */
      /* movl $magic_number, %ebp */
      switch (i->Xin.Goto.jk) {
         case Ijk_ClientReq: 
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_CLIENTREQ); break;
         case Ijk_Sys_int128:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_SYS_INT128); break;
         case Ijk_Yield: 
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_YIELD); break;
         case Ijk_EmWarn:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_EMWARN); break;
         case Ijk_MapFail:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_MAPFAIL); break;
         case Ijk_NoDecode:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_NODECODE); break;
         case Ijk_TInval:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_TINVAL); break;
         case Ijk_Sys_sysenter:
            *p++ = 0xBD;
            p = emit32(p, VEX_TRC_JMP_SYS_SYSENTER); break;
         case Ijk_Ret:
	 case Ijk_Call:
         case Ijk_Boring:
            break;
         default: 
            ppIRJumpKind(i->Xin.Goto.jk);
            vpanic("emit_X86Instr.Xin_Goto: unknown jump kind");
      }

      /* Get the destination address into %eax */
      if (i->Xin.Goto.dst->tag == Xri_Imm) {
         /* movl $immediate, %eax */
         *p++ = 0xB8;
         p = emit32(p, i->Xin.Goto.dst->Xri.Imm.imm32);
      } else {
         vassert(i->Xin.Goto.dst->tag == Xri_Reg);
         /* movl %reg, %eax */
         if (i->Xin.Goto.dst->Xri.Reg.reg != hregX86_EAX()) {
            *p++ = 0x89;
            p = doAMode_R(p, i->Xin.Goto.dst->Xri.Reg.reg, hregX86_EAX());
         }
      }

      /* Get the dispatcher address into %edx.  This has to happen
         after the load of %eax since %edx might be carrying the value
         destined for %eax immediately prior to this Xin_Goto. */
      vassert(sizeof(UInt) == sizeof(void*));
      vassert(dispatch != NULL);
      /* movl $imm32, %edx */
      *p++ = 0xBA;
      p = emit32(p, (UInt)dispatch);

      /* jmp *%edx */
      *p++ = 0xFF;
      *p++ = 0xE2;

      /* Fix up the conditional jump, if there was one. */
      if (i->Xin.Goto.cond != Xcc_ALWAYS) {
         Int delta = p - ptmp;
	 vassert(delta > 0 && delta < 20);
         *ptmp = toUChar(delta-1);
      }
      goto done;

   case Xin_CMov32:
      vassert(i->Xin.CMov32.cond != Xcc_ALWAYS);

      /* This generates cmov, which is illegal on P54/P55. */
      /*
      *p++ = 0x0F;
      *p++ = toUChar(0x40 + (0xF & i->Xin.CMov32.cond));
      if (i->Xin.CMov32.src->tag == Xrm_Reg) {
         p = doAMode_R(p, i->Xin.CMov32.dst, i->Xin.CMov32.src->Xrm.Reg.reg);
         goto done;
      }
      if (i->Xin.CMov32.src->tag == Xrm_Mem) {
         p = doAMode_M(p, i->Xin.CMov32.dst, i->Xin.CMov32.src->Xrm.Mem.am);
         goto done;
      }
      */

      /* Alternative version which works on any x86 variant. */
      /* jmp fwds if !condition */
      *p++ = toUChar(0x70 + (i->Xin.CMov32.cond ^ 1));
      *p++ = 0; /* # of bytes in the next bit, which we don't know yet */
      ptmp = p;

      switch (i->Xin.CMov32.src->tag) {
         case Xrm_Reg:
            /* Big sigh.  This is movl E -> G ... */
            *p++ = 0x89;
            p = doAMode_R(p, i->Xin.CMov32.src->Xrm.Reg.reg,
                             i->Xin.CMov32.dst);

            break;
         case Xrm_Mem:
            /* ... whereas this is movl G -> E.  That's why the args
               to doAMode_R appear to be the wrong way round in the
               Xrm_Reg case. */
            *p++ = 0x8B;
            p = doAMode_M(p, i->Xin.CMov32.dst,
                             i->Xin.CMov32.src->Xrm.Mem.am);
            break;
         default:
            goto bad;
      }
      /* Fill in the jump offset. */
      *(ptmp-1) = toUChar(p - ptmp);
      goto done;

      break;

   case Xin_LoadEX:
      if (i->Xin.LoadEX.szSmall == 1 && !i->Xin.LoadEX.syned) {
         /* movzbl */
         *p++ = 0x0F;
         *p++ = 0xB6;
         p = doAMode_M(p, i->Xin.LoadEX.dst, i->Xin.LoadEX.src); 
         goto done;
      }
      if (i->Xin.LoadEX.szSmall == 2 && !i->Xin.LoadEX.syned) {
         /* movzwl */
         *p++ = 0x0F;
         *p++ = 0xB7;
         p = doAMode_M(p, i->Xin.LoadEX.dst, i->Xin.LoadEX.src); 
         goto done;
      }
      break;

   case Xin_Set32:
      /* Make the destination register be 1 or 0, depending on whether
         the relevant condition holds.  We have to dodge and weave
         when the destination is %esi or %edi as we cannot directly
         emit the native 'setb %reg' for those.  Further complication:
         the top 24 bits of the destination should be forced to zero,
         but doing 'xor %r,%r' kills the flag(s) we are about to read.
         Sigh.  So start off my moving $0 into the dest. */

      /* Do we need to swap in %eax? */
      if (iregNo(i->Xin.Set32.dst) >= 4) {
         /* xchg %eax, %dst */
         *p++ = toUChar(0x90 + iregNo(i->Xin.Set32.dst));
         /* movl $0, %eax */
         *p++ =toUChar(0xB8 + iregNo(hregX86_EAX()));
         p = emit32(p, 0);
         /* setb lo8(%eax) */
         *p++ = 0x0F; 
         *p++ = toUChar(0x90 + (0xF & i->Xin.Set32.cond));
         p = doAMode_R(p, fake(0), hregX86_EAX());
         /* xchg %eax, %dst */
         *p++ = toUChar(0x90 + iregNo(i->Xin.Set32.dst));
      } else {
         /* movl $0, %dst */
         *p++ = toUChar(0xB8 + iregNo(i->Xin.Set32.dst));
         p = emit32(p, 0);
         /* setb lo8(%dst) */
         *p++ = 0x0F; 
         *p++ = toUChar(0x90 + (0xF & i->Xin.Set32.cond));
         p = doAMode_R(p, fake(0), i->Xin.Set32.dst);
      }
      goto done;

   case Xin_Bsfr32:
      *p++ = 0x0F;
      if (i->Xin.Bsfr32.isFwds) {
         *p++ = 0xBC;
      } else {
         *p++ = 0xBD;
      }
      p = doAMode_R(p, i->Xin.Bsfr32.dst, i->Xin.Bsfr32.src);
      goto done;

   case Xin_MFence:
      /* see comment in hdefs.h re this insn */
      if (0) vex_printf("EMIT FENCE\n");
      switch (i->Xin.MFence.subarch) {
         case VexSubArchX86_sse0:
            /* lock addl $0,0(%esp) */
            *p++ = 0xF0; *p++ = 0x83; *p++ = 0x44; 
            *p++ = 0x24; *p++ = 0x00; *p++ = 0x00;
            goto done;
         case VexSubArchX86_sse1:
            /* sfence */
            *p++ = 0x0F; *p++ = 0xAE; *p++ = 0xF8;
            /* lock addl $0,0(%esp) */
            *p++ = 0xF0; *p++ = 0x83; *p++ = 0x44; 
            *p++ = 0x24; *p++ = 0x00; *p++ = 0x00;
            goto done;
         case VexSubArchX86_sse2:
            /* mfence */
            *p++ = 0x0F; *p++ = 0xAE; *p++ = 0xF0;
            goto done;
         default: 
            vpanic("emit_X86Instr:mfence:subarch");
      }
      break;

   case Xin_Store:
      if (i->Xin.Store.sz == 2) {
         /* This case, at least, is simple, given that we can
            reference the low 16 bits of any integer register. */
         *p++ = 0x66;
         *p++ = 0x89;
         p = doAMode_M(p, i->Xin.Store.src, i->Xin.Store.dst);
         goto done;
      }

      if (i->Xin.Store.sz == 1) {
         /* We have to do complex dodging and weaving if src is not
            the low 8 bits of %eax/%ebx/%ecx/%edx. */
         if (iregNo(i->Xin.Store.src) < 4) {
            /* we're OK, can do it directly */
            *p++ = 0x88;
            p = doAMode_M(p, i->Xin.Store.src, i->Xin.Store.dst);
           goto done;
         } else {
            /* Bleh.  This means the source is %edi or %esi.  Since
               the address mode can only mention three registers, at
               least one of %eax/%ebx/%ecx/%edx must be available to
               temporarily swap the source into, so the store can
               happen.  So we have to look at the regs mentioned
               in the amode. */
            HReg swap = INVALID_HREG;
            HReg  eax = hregX86_EAX(), ebx = hregX86_EBX(), 
                  ecx = hregX86_ECX(), edx = hregX86_EDX();
            Bool a_ok = True, b_ok = True, c_ok = True, d_ok = True;
            HRegUsage u;
            Int j;
            initHRegUsage(&u);
            addRegUsage_X86AMode(&u,  i->Xin.Store.dst);
            for (j = 0; j < u.n_used; j++) {
               HReg r = u.hreg[j];
               if (r == eax) a_ok = False;
               if (r == ebx) b_ok = False;
               if (r == ecx) c_ok = False;
               if (r == edx) d_ok = False;
            }
            if (a_ok) swap = eax;
            if (b_ok) swap = ebx;
            if (c_ok) swap = ecx;
            if (d_ok) swap = edx;
            vassert(swap != INVALID_HREG);
            /* xchgl %source, %swap. Could do better if swap is %eax. */
            *p++ = 0x87;
            p = doAMode_R(p, i->Xin.Store.src, swap);
            /* movb lo8{%swap}, (dst) */
            *p++ = 0x88;
            p = doAMode_M(p, swap, i->Xin.Store.dst);
            /* xchgl %source, %swap. Could do better if swap is %eax. */
            *p++ = 0x87;
            p = doAMode_R(p, i->Xin.Store.src, swap);
            goto done;
         }
      } /* if (i->Xin.Store.sz == 1) */
      break;

   case Xin_FpUnary:
      /* gop %src, %dst
         --> ffree %st7 ; fld %st(src) ; fop %st(0) ; fstp %st(1+dst)
      */
      p = do_ffree_st7(p);
      p = do_fld_st(p, 0+hregNumber(i->Xin.FpUnary.src));
      p = do_fop1_st(p, i->Xin.FpUnary.op);
      p = do_fstp_st(p, 1+hregNumber(i->Xin.FpUnary.dst));
      goto done;

   case Xin_FpBinary:
      if (i->Xin.FpBinary.op == Xfp_YL2X
          || i->Xin.FpBinary.op == Xfp_YL2XP1) {
         /* Have to do this specially. */
         /* ffree %st7 ; fld %st(srcL) ; 
            ffree %st7 ; fld %st(srcR+1) ; fyl2x{p1} ; fstp(1+dst) */
         p = do_ffree_st7(p);
         p = do_fld_st(p, 0+hregNumber(i->Xin.FpBinary.srcL));
         p = do_ffree_st7(p);
         p = do_fld_st(p, 1+hregNumber(i->Xin.FpBinary.srcR));
         *p++ = 0xD9; 
         *p++ = toUChar(i->Xin.FpBinary.op==Xfp_YL2X ? 0xF1 : 0xF9);
         p = do_fstp_st(p, 1+hregNumber(i->Xin.FpBinary.dst));
         goto done;
      }
      if (i->Xin.FpBinary.op == Xfp_ATAN) {
         /* Have to do this specially. */
         /* ffree %st7 ; fld %st(srcL) ; 
            ffree %st7 ; fld %st(srcR+1) ; fpatan ; fstp(1+dst) */
         p = do_ffree_st7(p);
         p = do_fld_st(p, 0+hregNumber(i->Xin.FpBinary.srcL));
         p = do_ffree_st7(p);
         p = do_fld_st(p, 1+hregNumber(i->Xin.FpBinary.srcR));
         *p++ = 0xD9; *p++ = 0xF3;
         p = do_fstp_st(p, 1+hregNumber(i->Xin.FpBinary.dst));
         goto done;
      }
      if (i->Xin.FpBinary.op == Xfp_PREM
          || i->Xin.FpBinary.op == Xfp_PREM1
          || i->Xin.FpBinary.op == Xfp_SCALE) {
         /* Have to do this specially. */
         /* ffree %st7 ; fld %st(srcR) ; 
            ffree %st7 ; fld %st(srcL+1) ; fprem/fprem1/fscale ; fstp(2+dst) ; 
            fincstp ; ffree %st7 */
         p = do_ffree_st7(p);
         p = do_fld_st(p, 0+hregNumber(i->Xin.FpBinary.srcR));
         p = do_ffree_st7(p);
         p = do_fld_st(p, 1+hregNumber(i->Xin.FpBinary.srcL));
         *p++ = 0xD9;
         switch (i->Xin.FpBinary.op) {
            case Xfp_PREM: *p++ = 0xF8; break;
            case Xfp_PREM1: *p++ = 0xF5; break;
            case Xfp_SCALE: *p++ =  0xFD; break;
            default: vpanic("emitX86Instr(FpBinary,PREM/PREM1/SCALE)");
         }
         p = do_fstp_st(p, 2+hregNumber(i->Xin.FpBinary.dst));
         *p++ = 0xD9; *p++ = 0xF7;
         p = do_ffree_st7(p);
         goto done;
      }
      /* General case */
      /* gop %srcL, %srcR, %dst
         --> ffree %st7 ; fld %st(srcL) ; fop %st(1+srcR) ; fstp %st(1+dst)
      */
      p = do_ffree_st7(p);
      p = do_fld_st(p, 0+hregNumber(i->Xin.FpBinary.srcL));
      p = do_fop2_st(p, i->Xin.FpBinary.op, 
                        1+hregNumber(i->Xin.FpBinary.srcR));
      p = do_fstp_st(p, 1+hregNumber(i->Xin.FpBinary.dst));
      goto done;

   case Xin_FpLdSt:
      vassert(i->Xin.FpLdSt.sz == 4 || i->Xin.FpLdSt.sz == 8);
      if (i->Xin.FpLdSt.isLoad) {
         /* Load from memory into %fakeN.  
            --> ffree %st(7) ; fld{s/l} amode ; fstp st(N+1) 
         */
         p = do_ffree_st7(p);
         *p++ = toUChar(i->Xin.FpLdSt.sz==4 ? 0xD9 : 0xDD);
	 p = doAMode_M(p, fake(0)/*subopcode*/, i->Xin.FpLdSt.addr);
         p = do_fstp_st(p, 1+hregNumber(i->Xin.FpLdSt.reg));
         goto done;
      } else {
         /* Store from %fakeN into memory.
            --> ffree %st(7) ; fld st(N) ; fstp{l|s} amode
	 */
         p = do_ffree_st7(p);
         p = do_fld_st(p, 0+hregNumber(i->Xin.FpLdSt.reg));
         *p++ = toUChar(i->Xin.FpLdSt.sz==4 ? 0xD9 : 0xDD);
         p = doAMode_M(p, fake(3)/*subopcode*/, i->Xin.FpLdSt.addr);
         goto done;
      }
      break;

   case Xin_FpLdStI:
      if (i->Xin.FpLdStI.isLoad) {
         /* Load from memory into %fakeN, converting from an int.  
            --> ffree %st(7) ; fild{w/l/ll} amode ; fstp st(N+1) 
         */
         switch (i->Xin.FpLdStI.sz) {
            case 8:  opc = 0xDF; subopc_imm = 5; break;
            case 4:  opc = 0xDB; subopc_imm = 0; break;
            case 2:  vassert(0); opc = 0xDF; subopc_imm = 0; break;
            default: vpanic("emitX86Instr(Xin_FpLdStI-load)");
         }
         p = do_ffree_st7(p);
         *p++ = toUChar(opc);
         p = doAMode_M(p, fake(subopc_imm)/*subopcode*/, i->Xin.FpLdStI.addr);
         p = do_fstp_st(p, 1+hregNumber(i->Xin.FpLdStI.reg));
         goto done;
      } else {
         /* Store from %fakeN into memory, converting to an int.
            --> ffree %st(7) ; fld st(N) ; fistp{w/l/ll} amode
	 */
         switch (i->Xin.FpLdStI.sz) {
            case 8:  opc = 0xDF; subopc_imm = 7; break;
            case 4:  opc = 0xDB; subopc_imm = 3; break;
            case 2:  opc = 0xDF; subopc_imm = 3; break;
            default: vpanic("emitX86Instr(Xin_FpLdStI-store)");
         }
         p = do_ffree_st7(p);
         p = do_fld_st(p, 0+hregNumber(i->Xin.FpLdStI.reg));
         *p++ = toUChar(opc);
         p = doAMode_M(p, fake(subopc_imm)/*subopcode*/, i->Xin.FpLdStI.addr);
         goto done;
      }
      break;

   case Xin_Fp64to32:
      /* ffree %st7 ; fld %st(src) */
      p = do_ffree_st7(p);
      p = do_fld_st(p, 0+fregNo(i->Xin.Fp64to32.src));
      /* subl $4, %esp */
      *p++ = 0x83; *p++ = 0xEC; *p++ = 0x04;
      /* fstps (%esp) */
      *p++ = 0xD9; *p++ = 0x1C; *p++ = 0x24;
      /* flds (%esp) */
      *p++ = 0xD9; *p++ = 0x04; *p++ = 0x24;
      /* addl $4, %esp */
      *p++ = 0x83; *p++ = 0xC4; *p++ = 0x04;
      /* fstp %st(1+dst) */
      p = do_fstp_st(p, 1+fregNo(i->Xin.Fp64to32.dst));
      goto done;

   case Xin_FpCMov:
      /* jmp fwds if !condition */
      *p++ = toUChar(0x70 + (i->Xin.FpCMov.cond ^ 1));
      *p++ = 0; /* # of bytes in the next bit, which we don't know yet */
      ptmp = p;

      /* ffree %st7 ; fld %st(src) ; fstp %st(1+dst) */
      p = do_ffree_st7(p);
      p = do_fld_st(p, 0+fregNo(i->Xin.FpCMov.src));
      p = do_fstp_st(p, 1+fregNo(i->Xin.FpCMov.dst));

      /* Fill in the jump offset. */
      *(ptmp-1) = toUChar(p - ptmp);
      goto done;

   case Xin_FpLdCW:
      *p++ = 0xD9;
      p = doAMode_M(p, fake(5)/*subopcode*/, i->Xin.FpLdCW.addr);
      goto done;

   case Xin_FpStSW_AX:
      /* note, this emits fnstsw %ax, not fstsw %ax */
      *p++ = 0xDF;
      *p++ = 0xE0;
      goto done;

   case Xin_FpCmp:
      /* gcmp %fL, %fR, %dst
         -> ffree %st7; fpush %fL ; fucomp %(fR+1) ; 
            fnstsw %ax ; movl %eax, %dst 
      */
      /* ffree %st7 */
      p = do_ffree_st7(p);
      /* fpush %fL */
      p = do_fld_st(p, 0+fregNo(i->Xin.FpCmp.srcL));
      /* fucomp %(fR+1) */
      *p++ = 0xDD;
      *p++ = toUChar(0xE8 + (7 & (1+fregNo(i->Xin.FpCmp.srcR))));
      /* fnstsw %ax */
      *p++ = 0xDF;
      *p++ = 0xE0;
      /*  movl %eax, %dst */
      *p++ = 0x89;
      p = doAMode_R(p, hregX86_EAX(), i->Xin.FpCmp.dst);
      goto done;

   case Xin_SseConst: {
      UShort con = i->Xin.SseConst.con;
      p = push_word_from_tags(p, toUShort((con >> 12) & 0xF));
      p = push_word_from_tags(p, toUShort((con >> 8) & 0xF));
      p = push_word_from_tags(p, toUShort((con >> 4) & 0xF));
      p = push_word_from_tags(p, toUShort(con & 0xF));
      /* movl (%esp), %xmm-dst */
      *p++ = 0x0F;
      *p++ = 0x10;
      *p++ = toUChar(0x04 + 8 * (7 & vregNo(i->Xin.SseConst.dst)));
      *p++ = 0x24;
      /* addl $16, %esp */
      *p++ = 0x83;
      *p++ = 0xC4;
      *p++ = 0x10;
      goto done;
   }

   case Xin_SseLdSt:
      *p++ = 0x0F; 
      *p++ = toUChar(i->Xin.SseLdSt.isLoad ? 0x10 : 0x11);
      p = doAMode_M(p, fake(vregNo(i->Xin.SseLdSt.reg)), i->Xin.SseLdSt.addr);
      goto done;

   case Xin_SseLdzLO:
      vassert(i->Xin.SseLdzLO.sz == 4 || i->Xin.SseLdzLO.sz == 8);
      /* movs[sd] amode, %xmm-dst */
      *p++ = toUChar(i->Xin.SseLdzLO.sz==4 ? 0xF3 : 0xF2);
      *p++ = 0x0F; 
      *p++ = 0x10; 
      p = doAMode_M(p, fake(vregNo(i->Xin.SseLdzLO.reg)), 
                       i->Xin.SseLdzLO.addr);
      goto done;

   case Xin_Sse32Fx4:
      xtra = 0;
      *p++ = 0x0F;
      switch (i->Xin.Sse32Fx4.op) {
         case Xsse_ADDF:   *p++ = 0x58; break;
         case Xsse_DIVF:   *p++ = 0x5E; break;
         case Xsse_MAXF:   *p++ = 0x5F; break;
         case Xsse_MINF:   *p++ = 0x5D; break;
         case Xsse_MULF:   *p++ = 0x59; break;
         case Xsse_RCPF:   *p++ = 0x53; break;
         case Xsse_RSQRTF: *p++ = 0x52; break;
         case Xsse_SQRTF:  *p++ = 0x51; break;
         case Xsse_SUBF:   *p++ = 0x5C; break;
         case Xsse_CMPEQF: *p++ = 0xC2; xtra = 0x100; break;
         case Xsse_CMPLTF: *p++ = 0xC2; xtra = 0x101; break;
         case Xsse_CMPLEF: *p++ = 0xC2; xtra = 0x102; break;
         case Xsse_CMPUNF: *p++ = 0xC2; xtra = 0x103; break;
         default: goto bad;
      }
      p = doAMode_R(p, fake(vregNo(i->Xin.Sse32Fx4.dst)),
                       fake(vregNo(i->Xin.Sse32Fx4.src)) );
      if (xtra & 0x100)
         *p++ = toUChar(xtra & 0xFF);
      goto done;

   case Xin_Sse64Fx2:
      xtra = 0;
      *p++ = 0x66;
      *p++ = 0x0F;
      switch (i->Xin.Sse64Fx2.op) {
         case Xsse_ADDF:   *p++ = 0x58; break;
         case Xsse_DIVF:   *p++ = 0x5E; break;
         case Xsse_MAXF:   *p++ = 0x5F; break;
         case Xsse_MINF:   *p++ = 0x5D; break;
         case Xsse_MULF:   *p++ = 0x59; break;
         case Xsse_RCPF:   *p++ = 0x53; break;
         case Xsse_RSQRTF: *p++ = 0x52; break;
         case Xsse_SQRTF:  *p++ = 0x51; break;
         case Xsse_SUBF:   *p++ = 0x5C; break;
         case Xsse_CMPEQF: *p++ = 0xC2; xtra = 0x100; break;
         case Xsse_CMPLTF: *p++ = 0xC2; xtra = 0x101; break;
         case Xsse_CMPLEF: *p++ = 0xC2; xtra = 0x102; break;
         case Xsse_CMPUNF: *p++ = 0xC2; xtra = 0x103; break;
         default: goto bad;
      }
      p = doAMode_R(p, fake(vregNo(i->Xin.Sse64Fx2.dst)),
                       fake(vregNo(i->Xin.Sse64Fx2.src)) );
      if (xtra & 0x100)
         *p++ = toUChar(xtra & 0xFF);
      goto done;

   case Xin_Sse32FLo:
      xtra = 0;
      *p++ = 0xF3;
      *p++ = 0x0F;
      switch (i->Xin.Sse32FLo.op) {
         case Xsse_ADDF:   *p++ = 0x58; break;
         case Xsse_DIVF:   *p++ = 0x5E; break;
         case Xsse_MAXF:   *p++ = 0x5F; break;
         case Xsse_MINF:   *p++ = 0x5D; break;
         case Xsse_MULF:   *p++ = 0x59; break;
         case Xsse_RCPF:   *p++ = 0x53; break;
         case Xsse_RSQRTF: *p++ = 0x52; break;
         case Xsse_SQRTF:  *p++ = 0x51; break;
         case Xsse_SUBF:   *p++ = 0x5C; break;
         case Xsse_CMPEQF: *p++ = 0xC2; xtra = 0x100; break;
         case Xsse_CMPLTF: *p++ = 0xC2; xtra = 0x101; break;
         case Xsse_CMPLEF: *p++ = 0xC2; xtra = 0x102; break;
         case Xsse_CMPUNF: *p++ = 0xC2; xtra = 0x103; break;
         default: goto bad;
      }
      p = doAMode_R(p, fake(vregNo(i->Xin.Sse32FLo.dst)),
                       fake(vregNo(i->Xin.Sse32FLo.src)) );
      if (xtra & 0x100)
         *p++ = toUChar(xtra & 0xFF);
      goto done;

   case Xin_Sse64FLo:
      xtra = 0;
      *p++ = 0xF2;
      *p++ = 0x0F;
      switch (i->Xin.Sse64FLo.op) {
         case Xsse_ADDF:   *p++ = 0x58; break;
         case Xsse_DIVF:   *p++ = 0x5E; break;
         case Xsse_MAXF:   *p++ = 0x5F; break;
         case Xsse_MINF:   *p++ = 0x5D; break;
         case Xsse_MULF:   *p++ = 0x59; break;
         case Xsse_RCPF:   *p++ = 0x53; break;
         case Xsse_RSQRTF: *p++ = 0x52; break;
         case Xsse_SQRTF:  *p++ = 0x51; break;
         case Xsse_SUBF:   *p++ = 0x5C; break;
         case Xsse_CMPEQF: *p++ = 0xC2; xtra = 0x100; break;
         case Xsse_CMPLTF: *p++ = 0xC2; xtra = 0x101; break;
         case Xsse_CMPLEF: *p++ = 0xC2; xtra = 0x102; break;
         case Xsse_CMPUNF: *p++ = 0xC2; xtra = 0x103; break;
         default: goto bad;
      }
      p = doAMode_R(p, fake(vregNo(i->Xin.Sse64FLo.dst)),
                       fake(vregNo(i->Xin.Sse64FLo.src)) );
      if (xtra & 0x100)
         *p++ = toUChar(xtra & 0xFF);
      goto done;

   case Xin_SseReRg:
#     define XX(_n) *p++ = (_n)
      switch (i->Xin.SseReRg.op) {
         case Xsse_MOV:     /*movups*/ XX(0x0F); XX(0x10); break;
         case Xsse_OR:                 XX(0x0F); XX(0x56); break;
         case Xsse_XOR:                XX(0x0F); XX(0x57); break;
         case Xsse_AND:                XX(0x0F); XX(0x54); break;
         case Xsse_PACKSSD:  XX(0x66); XX(0x0F); XX(0x6B); break;
         case Xsse_PACKSSW:  XX(0x66); XX(0x0F); XX(0x63); break;
         case Xsse_PACKUSW:  XX(0x66); XX(0x0F); XX(0x67); break;
         case Xsse_ADD8:     XX(0x66); XX(0x0F); XX(0xFC); break;
         case Xsse_ADD16:    XX(0x66); XX(0x0F); XX(0xFD); break;
         case Xsse_ADD32:    XX(0x66); XX(0x0F); XX(0xFE); break;
         case Xsse_ADD64:    XX(0x66); XX(0x0F); XX(0xD4); break;
         case Xsse_QADD8S:   XX(0x66); XX(0x0F); XX(0xEC); break;
         case Xsse_QADD16S:  XX(0x66); XX(0x0F); XX(0xED); break;
         case Xsse_QADD8U:   XX(0x66); XX(0x0F); XX(0xDC); break;
         case Xsse_QADD16U:  XX(0x66); XX(0x0F); XX(0xDD); break;
         case Xsse_AVG8U:    XX(0x66); XX(0x0F); XX(0xE0); break;
         case Xsse_AVG16U:   XX(0x66); XX(0x0F); XX(0xE3); break;
         case Xsse_CMPEQ8:   XX(0x66); XX(0x0F); XX(0x74); break;
         case Xsse_CMPEQ16:  XX(0x66); XX(0x0F); XX(0x75); break;
         case Xsse_CMPEQ32:  XX(0x66); XX(0x0F); XX(0x76); break;
         case Xsse_CMPGT8S:  XX(0x66); XX(0x0F); XX(0x64); break;
         case Xsse_CMPGT16S: XX(0x66); XX(0x0F); XX(0x65); break;
         case Xsse_CMPGT32S: XX(0x66); XX(0x0F); XX(0x66); break;
         case Xsse_MAX16S:   XX(0x66); XX(0x0F); XX(0xEE); break;
         case Xsse_MAX8U:    XX(0x66); XX(0x0F); XX(0xDE); break;
         case Xsse_MIN16S:   XX(0x66); XX(0x0F); XX(0xEA); break;
         case Xsse_MIN8U:    XX(0x66); XX(0x0F); XX(0xDA); break;
         case Xsse_MULHI16U: XX(0x66); XX(0x0F); XX(0xE4); break;
         case Xsse_MULHI16S: XX(0x66); XX(0x0F); XX(0xE5); break;
         case Xsse_MUL16:    XX(0x66); XX(0x0F); XX(0xD5); break;
         case Xsse_SHL16:    XX(0x66); XX(0x0F); XX(0xF1); break;
         case Xsse_SHL32:    XX(0x66); XX(0x0F); XX(0xF2); break;
         case Xsse_SHL64:    XX(0x66); XX(0x0F); XX(0xF3); break;
         case Xsse_SAR16:    XX(0x66); XX(0x0F); XX(0xE1); break;
         case Xsse_SAR32:    XX(0x66); XX(0x0F); XX(0xE2); break;
         case Xsse_SHR16:    XX(0x66); XX(0x0F); XX(0xD1); break;
         case Xsse_SHR32:    XX(0x66); XX(0x0F); XX(0xD2); break;
         case Xsse_SHR64:    XX(0x66); XX(0x0F); XX(0xD3); break;
         case Xsse_SUB8:     XX(0x66); XX(0x0F); XX(0xF8); break;
         case Xsse_SUB16:    XX(0x66); XX(0x0F); XX(0xF9); break;
         case Xsse_SUB32:    XX(0x66); XX(0x0F); XX(0xFA); break;
         case Xsse_SUB64:    XX(0x66); XX(0x0F); XX(0xFB); break;
         case Xsse_QSUB8S:   XX(0x66); XX(0x0F); XX(0xE8); break;
         case Xsse_QSUB16S:  XX(0x66); XX(0x0F); XX(0xE9); break;
         case Xsse_QSUB8U:   XX(0x66); XX(0x0F); XX(0xD8); break;
         case Xsse_QSUB16U:  XX(0x66); XX(0x0F); XX(0xD9); break;
         case Xsse_UNPCKHB:  XX(0x66); XX(0x0F); XX(0x68); break;
         case Xsse_UNPCKHW:  XX(0x66); XX(0x0F); XX(0x69); break;
         case Xsse_UNPCKHD:  XX(0x66); XX(0x0F); XX(0x6A); break;
         case Xsse_UNPCKHQ:  XX(0x66); XX(0x0F); XX(0x6D); break;
         case Xsse_UNPCKLB:  XX(0x66); XX(0x0F); XX(0x60); break;
         case Xsse_UNPCKLW:  XX(0x66); XX(0x0F); XX(0x61); break;
         case Xsse_UNPCKLD:  XX(0x66); XX(0x0F); XX(0x62); break;
         case Xsse_UNPCKLQ:  XX(0x66); XX(0x0F); XX(0x6C); break;
         default: goto bad;
      }
      p = doAMode_R(p, fake(vregNo(i->Xin.SseReRg.dst)),
                       fake(vregNo(i->Xin.SseReRg.src)) );
#     undef XX
      goto done;

   case Xin_SseCMov:
      /* jmp fwds if !condition */
      *p++ = toUChar(0x70 + (i->Xin.SseCMov.cond ^ 1));
      *p++ = 0; /* # of bytes in the next bit, which we don't know yet */
      ptmp = p;

      /* movaps %src, %dst */
      *p++ = 0x0F; 
      *p++ = 0x28; 
      p = doAMode_R(p, fake(vregNo(i->Xin.SseCMov.dst)),
                       fake(vregNo(i->Xin.SseCMov.src)) );

      /* Fill in the jump offset. */
      *(ptmp-1) = toUChar(p - ptmp);
      goto done;

   case Xin_SseShuf:
      *p++ = 0x66; 
      *p++ = 0x0F; 
      *p++ = 0x70; 
      p = doAMode_R(p, fake(vregNo(i->Xin.SseShuf.dst)),
                       fake(vregNo(i->Xin.SseShuf.src)) );
      *p++ = (UChar)(i->Xin.SseShuf.order);
      goto done;

   default: 
      goto bad;
   }

  bad:
   ppX86Instr(i, mode64);
   vpanic("emit_X86Instr");
   /*NOTREACHED*/
   
  done:
   vassert(p - &buf[0] <= 32);
   return p - &buf[0];

#  undef fake
}

/*---------------------------------------------------------------*/
/*--- end                                    host-x86/hdefs.c ---*/
/*---------------------------------------------------------------*/