[thirdparty/binutils-gdb.git] / sim / mcore / interp.c

/* Simulator for Motorola's MCore processor
   Copyright (C) 1999-2015 Free Software Foundation, Inc.
   Contributed by Cygnus Solutions.

This file is part of GDB, the GNU debugger.

This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program.  If not, see <http://www.gnu.org/licenses/>.  */

#include "config.h"
#include <signal.h>
#include <stdlib.h>
#include <string.h>
#include <sys/times.h>
#include <sys/param.h>
#include <unistd.h>
#include "bfd.h"
#include "gdb/callback.h"
#include "libiberty.h"
#include "gdb/remote-sim.h"

#include "sim-main.h"
#include "sim-base.h"
#include "sim-options.h"

#define target_big_endian (CURRENT_TARGET_BYTE_ORDER == BIG_ENDIAN)


static unsigned long
mcore_extract_unsigned_integer (unsigned char *addr, int len)
{
  unsigned long retval;
  unsigned char * p;
  unsigned char * startaddr = (unsigned char *)addr;
  unsigned char * endaddr = startaddr + len;

  if (len > (int) sizeof (unsigned long))
    printf ("That operation is not available on integers of more than %zu bytes.",
	    sizeof (unsigned long));

  /* Start at the most significant end of the integer, and work towards
     the least significant.  */
  retval = 0;

  if (! target_big_endian)
    {
      for (p = endaddr; p > startaddr;)
	retval = (retval << 8) | * -- p;
    }
  else
    {
      for (p = startaddr; p < endaddr;)
	retval = (retval << 8) | * p ++;
    }

  return retval;
}

static void
mcore_store_unsigned_integer (unsigned char *addr, int len, unsigned long val)
{
  unsigned char * p;
  unsigned char * startaddr = (unsigned char *)addr;
  unsigned char * endaddr = startaddr + len;

  if (! target_big_endian)
    {
      for (p = startaddr; p < endaddr;)
	{
	  * p ++ = val & 0xff;
	  val >>= 8;
	}
    }
  else
    {
      for (p = endaddr; p > startaddr;)
	{
	  * -- p = val & 0xff;
	  val >>= 8;
	}
    }
}

/* The machine state.
   This state is maintained in host byte order.  The
   fetch/store register functions must translate between host
   byte order and the target processor byte order.
   Keeping this data in target byte order simplifies the register
   read/write functions.  Keeping this data in native order improves
   the performance of the simulator.  Simulation speed is deemed more
   important.  */
/* TODO: Should be moved to sim-main.h:sim_cpu.  */

/* The ordering of the mcore_regset structure is matched in the
   gdb/config/mcore/tm-mcore.h file in the REGISTER_NAMES macro.  */
struct mcore_regset
{
  word	          gregs [16];		/* primary registers */
  word	          alt_gregs [16];	/* alt register file */
  word	          cregs [32];		/* control registers */
  int		  ticks;
  int		  stalls;
  int		  cycles;
  int		  insts;
  int		  exception;
  unsigned long   msize;
  unsigned char * memory;
  word *          active_gregs;
};

union
{
  struct mcore_regset asregs;
  word asints [1];		/* but accessed larger... */
} cpu;

#define LAST_VALID_CREG	32		/* only 0..12 implemented */
#define	NUM_MCORE_REGS	(16 + 16 + LAST_VALID_CREG + 1)

static int memcycles = 1;

static int issue_messages = 0;

#define gr	asregs.active_gregs
#define cr	asregs.cregs
#define sr	asregs.cregs[0]
#define	vbr	asregs.cregs[1]
#define	esr	asregs.cregs[2]
#define	fsr	asregs.cregs[3]
#define	epc	asregs.cregs[4]
#define	fpc	asregs.cregs[5]
#define	ss0	asregs.cregs[6]
#define	ss1	asregs.cregs[7]
#define	ss2	asregs.cregs[8]
#define	ss3	asregs.cregs[9]
#define	ss4	asregs.cregs[10]
#define	gcr	asregs.cregs[11]
#define	gsr	asregs.cregs[12]
#define mem	asregs.memory

/* maniuplate the carry bit */
#define	C_ON()	 (cpu.sr & 1)
#define	C_VALUE() (cpu.sr & 1)
#define	C_OFF()	 ((cpu.sr & 1) == 0)
#define	SET_C()	 {cpu.sr |= 1;}
#define	CLR_C()	 {cpu.sr &= 0xfffffffe;}
#define	NEW_C(v) {CLR_C(); cpu.sr |= ((v) & 1);}

#define	SR_AF() ((cpu.sr >> 1) & 1)

#define	TRAPCODE	1	/* r1 holds which function we want */
#define	PARM1	2		/* first parameter  */
#define	PARM2	3
#define	PARM3	4
#define	PARM4	5
#define	RET1	2		/* register for return values. */

static void
wbat (word x, word v)
{
  if (((uword)x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "byte write to 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
    }
  else
    {
      unsigned char *p = cpu.mem + x;
      p[0] = v;
    }
}

static void
wlat (word x, word v)
{
  if (((uword)x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "word write to 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
    }
  else
    {
      if ((x & 3) != 0)
	{
	  if (issue_messages)
	    fprintf (stderr, "word write to unaligned memory address: 0x%x\n", x);

	  cpu.asregs.exception = SIGBUS;
	}
      else if (! target_big_endian)
	{
	  unsigned char * p = cpu.mem + x;
	  p[3] = v >> 24;
	  p[2] = v >> 16;
	  p[1] = v >> 8;
	  p[0] = v;
	}
      else
	{
	  unsigned char * p = cpu.mem + x;
	  p[0] = v >> 24;
	  p[1] = v >> 16;
	  p[2] = v >> 8;
	  p[3] = v;
	}
    }
}

static void
what (word x, word v)
{
  if (((uword)x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "short write to 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
    }
  else
    {
      if ((x & 1) != 0)
	{
	  if (issue_messages)
	    fprintf (stderr, "short write to unaligned memory address: 0x%x\n",
		     x);

	  cpu.asregs.exception = SIGBUS;
	}
      else if (! target_big_endian)
	{
	  unsigned char * p = cpu.mem + x;
	  p[1] = v >> 8;
	  p[0] = v;
	}
      else
	{
	  unsigned char * p = cpu.mem + x;
	  p[0] = v >> 8;
	  p[1] = v;
	}
    }
}

/* Read functions.  */
static int
rbat (word x)
{
  if (((uword)x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "byte read from 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
      return 0;
    }
  else
    {
      unsigned char * p = cpu.mem + x;
      return p[0];
    }
}

static int
rlat (word x)
{
  if (((uword) x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "word read from 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
      return 0;
    }
  else
    {
      if ((x & 3) != 0)
	{
	  if (issue_messages)
	    fprintf (stderr, "word read from unaligned address: 0x%x\n", x);

	  cpu.asregs.exception = SIGBUS;
	  return 0;
	}
      else if (! target_big_endian)
	{
	  unsigned char * p = cpu.mem + x;
	  return (p[3] << 24) | (p[2] << 16) | (p[1] << 8) | p[0];
	}
      else
	{
	  unsigned char * p = cpu.mem + x;
	  return (p[0] << 24) | (p[1] << 16) | (p[2] << 8) | p[3];
	}
    }
}

static int
rhat (word x)
{
  if (((uword)x) >= cpu.asregs.msize)
    {
      if (issue_messages)
	fprintf (stderr, "short read from 0x%x outside memory range\n", x);

      cpu.asregs.exception = SIGSEGV;
      return 0;
    }
  else
    {
      if ((x & 1) != 0)
	{
	  if (issue_messages)
	    fprintf (stderr, "short read from unaligned address: 0x%x\n", x);

	  cpu.asregs.exception = SIGBUS;
	  return 0;
	}
      else if (! target_big_endian)
	{
	  unsigned char * p = cpu.mem + x;
	  return (p[1] << 8) | p[0];
	}
      else
	{
	  unsigned char * p = cpu.mem + x;
	  return (p[0] << 8) | p[1];
	}
    }
}


/* Default to a 8 Mbyte (== 2^23) memory space.  */
/* TODO: Delete all this custom memory logic and move to common sim helpers.  */
static int sim_memory_size = 23;

#define	MEM_SIZE_FLOOR	64
static void
sim_size (int power)
{
  sim_memory_size = power;
  cpu.asregs.msize = 1 << sim_memory_size;

  if (cpu.mem)
    free (cpu.mem);

  /* Watch out for the '0 count' problem. There's probably a better
     way.. e.g., why do we use 64 here?  */
  if (cpu.asregs.msize < 64)	/* Ensure a boundary.  */
    cpu.mem = (unsigned char *) calloc (64, (64 + cpu.asregs.msize) / 64);
  else
    cpu.mem = (unsigned char *) calloc (64, cpu.asregs.msize / 64);

  if (!cpu.mem)
    {
      if (issue_messages)
	fprintf (stderr,
		 "Not enough VM for simulation of %lu bytes of RAM\n",
		 cpu.asregs.msize);

      cpu.asregs.msize = 1;
      cpu.mem = (unsigned char *) calloc (1, 1);
    }
}

static void
init_pointers (void)
{
  if (cpu.asregs.msize != (1 << sim_memory_size))
    sim_size (sim_memory_size);
}

static void
set_initial_gprs (SIM_CPU *scpu)
{
  int i;
  long space;
  unsigned long memsize;

  init_pointers ();

  /* Set up machine just out of reset.  */
  CPU_PC_SET (scpu, 0);
  cpu.sr = 0;

  memsize = cpu.asregs.msize / (1024 * 1024);

  if (issue_messages > 1)
    fprintf (stderr, "Simulated memory of %lu Mbytes (0x0 .. 0x%08lx)\n",
	     memsize, cpu.asregs.msize - 1);

  /* Clean out the GPRs and alternate GPRs.  */
  for (i = 0; i < 16; i++)
    {
      cpu.asregs.gregs[i] = 0;
      cpu.asregs.alt_gregs[i] = 0;
    }

  /* Make our register set point to the right place.  */
  if (SR_AF())
    cpu.asregs.active_gregs = &cpu.asregs.alt_gregs[0];
  else
    cpu.asregs.active_gregs = &cpu.asregs.gregs[0];

  /* ABI specifies initial values for these registers.  */
  cpu.gr[0] = cpu.asregs.msize - 4;

  /* dac fix, the stack address must be 8-byte aligned! */
  cpu.gr[0] = cpu.gr[0] - cpu.gr[0] % 8;
  cpu.gr[PARM1] = 0;
  cpu.gr[PARM2] = 0;
  cpu.gr[PARM3] = 0;
  cpu.gr[PARM4] = cpu.gr[0];
}

/* Read/write functions for system call interface.  */

static int
syscall_read_mem (host_callback *cb, struct cb_syscall *sc,
		  unsigned long taddr, char *buf, int bytes)
{
  memcpy (buf, cpu.mem + taddr, bytes);
  return bytes;
}

static int
syscall_write_mem (host_callback *cb, struct cb_syscall *sc,
		  unsigned long taddr, const char *buf, int bytes)
{
  memcpy (cpu.mem + taddr, buf, bytes);
  return bytes;
}

/* Simulate a monitor trap.  */

static void
handle_trap1 (SIM_DESC sd)
{
  host_callback *cb = STATE_CALLBACK (sd);
  CB_SYSCALL sc;

  CB_SYSCALL_INIT (&sc);

  sc.func = cpu.gr[TRAPCODE];
  sc.arg1 = cpu.gr[PARM1];
  sc.arg2 = cpu.gr[PARM2];
  sc.arg3 = cpu.gr[PARM3];
  sc.arg4 = cpu.gr[PARM4];

  sc.p1 = (PTR) sd;
  sc.p2 = (PTR) STATE_CPU (sd, 0);
  sc.read_mem = syscall_read_mem;
  sc.write_mem = syscall_write_mem;

  cb_syscall (cb, &sc);

  /* XXX: We don't pass back the actual errno value.  */
  cpu.gr[RET1] = sc.result;
}

static void
process_stub (SIM_DESC sd, int what)
{
  /* These values should match those in libgloss/mcore/syscalls.s.  */
  switch (what)
    {
    case 3:  /* _read */
    case 4:  /* _write */
    case 5:  /* _open */
    case 6:  /* _close */
    case 10: /* _unlink */
    case 19: /* _lseek */
    case 43: /* _times */
      cpu.gr [TRAPCODE] = what;
      handle_trap1 (sd);
      break;

    default:
      if (issue_messages)
	fprintf (stderr, "Unhandled stub opcode: %d\n", what);
      break;
    }
}

static void
util (SIM_DESC sd, unsigned what)
{
  switch (what)
    {
    case 0:	/* exit */
      cpu.asregs.exception = SIGQUIT;
      break;

    case 1:	/* printf */
      {
	unsigned long a[6];
	unsigned char *s;
	int i;

	a[0] = (unsigned long)(cpu.mem + cpu.gr[PARM1]);

	for (s = (unsigned char *)a[0], i = 1 ; *s && i < 6 ; s++)
	  {
	    if (*s == '%')
	      {
		if (*++s == 's')
		  a[i] = (unsigned long)(cpu.mem + cpu.gr[PARM1+i]);
		else
		  a[i] = cpu.gr[i+PARM1];
		i++;
	      }
	  }

	cpu.gr[RET1] = printf ((char *)a[0], a[1], a[2], a[3], a[4], a[5]);
      }
      break;

    case 2:	/* scanf */
      if (issue_messages)
	fprintf (stderr, "WARNING: scanf unimplemented\n");
      break;

    case 3:	/* utime */
      cpu.gr[RET1] = cpu.asregs.insts;
      break;

    case 0xFF:
      process_stub (sd, cpu.gr[1]);
      break;

    default:
      if (issue_messages)
	fprintf (stderr, "Unhandled util code: %x\n", what);
      break;
    }
}

/* For figuring out whether we carried; addc/subc use this. */
static int
iu_carry (unsigned long a, unsigned long b, int cin)
{
  unsigned long	x;

  x = (a & 0xffff) + (b & 0xffff) + cin;
  x = (x >> 16) + (a >> 16) + (b >> 16);
  x >>= 16;

  return (x != 0);
}

/* TODO: Convert to common watchpoints.  */
#undef WATCHFUNCTIONS
#ifdef WATCHFUNCTIONS

#define MAXWL 80
word WL[MAXWL];
char * WLstr[MAXWL];

int ENDWL=0;
int WLincyc;
int WLcyc[MAXWL];
int WLcnts[MAXWL];
int WLmax[MAXWL];
int WLmin[MAXWL];
word WLendpc;
int WLbcyc;
int WLW;
#endif

#define RD	(inst        & 0xF)
#define RS	((inst >> 4) & 0xF)
#define RX	((inst >> 8) & 0xF)
#define IMM5	((inst >> 4) & 0x1F)
#define IMM4	((inst) & 0xF)

static int tracing = 0;

void
sim_resume (SIM_DESC sd, int step, int siggnal)
{
  SIM_CPU *scpu = STATE_CPU (sd, 0);
  int needfetch;
  word ibuf;
  word pc;
  unsigned short inst;
  int memops;
  int bonus_cycles;
  int insts;
  int w;
  int cycs;
#ifdef WATCHFUNCTIONS
  word WLhash;
#endif

  cpu.asregs.exception = step ? SIGTRAP: 0;
  pc = CPU_PC_GET (scpu);

  /* Fetch the initial instructions that we'll decode. */
  ibuf = rlat (pc & 0xFFFFFFFC);
  needfetch = 0;

  memops = 0;
  bonus_cycles = 0;
  insts = 0;

  /* make our register set point to the right place */
  if (SR_AF ())
    cpu.asregs.active_gregs = & cpu.asregs.alt_gregs[0];
  else
    cpu.asregs.active_gregs = & cpu.asregs.gregs[0];

#ifdef WATCHFUNCTIONS
  /* make a hash to speed exec loop, hope it's nonzero */
  WLhash = 0xFFFFFFFF;

  for (w = 1; w <= ENDWL; w++)
    WLhash = WLhash & WL[w];
#endif

  do
    {
      word oldpc;

      insts ++;

      if (pc & 02)
	{
	  if (! target_big_endian)
	    inst = ibuf >> 16;
	  else
	    inst = ibuf & 0xFFFF;
	  needfetch = 1;
	}
      else
	{
	  if (! target_big_endian)
	    inst = ibuf & 0xFFFF;
	  else
	    inst = ibuf >> 16;
	}

#ifdef WATCHFUNCTIONS
      /* now scan list of watch addresses, if match, count it and
	 note return address and count cycles until pc=return address */

      if ((WLincyc == 1) && (pc == WLendpc))
	{
	  cycs = (cpu.asregs.cycles + (insts + bonus_cycles +
				       (memops * memcycles)) - WLbcyc);

	  if (WLcnts[WLW] == 1)
	    {
	      WLmax[WLW] = cycs;
	      WLmin[WLW] = cycs;
	      WLcyc[WLW] = 0;
	    }

	  if (cycs > WLmax[WLW])
	    {
	      WLmax[WLW] = cycs;
	    }

	  if (cycs < WLmin[WLW])
	    {
	      WLmin[WLW] = cycs;
	    }

	  WLcyc[WLW] += cycs;
	  WLincyc = 0;
	  WLendpc = 0;
	}

      /* Optimize with a hash to speed loop.  */
      if (WLincyc == 0)
	{
          if ((WLhash == 0) || ((WLhash & pc) != 0))
	    {
	      for (w=1; w <= ENDWL; w++)
		{
		  if (pc == WL[w])
		    {
		      WLcnts[w]++;
		      WLbcyc = cpu.asregs.cycles + insts
			+ bonus_cycles + (memops * memcycles);
		      WLendpc = cpu.gr[15];
		      WLincyc = 1;
		      WLW = w;
		      break;
		    }
		}
	    }
	}
#endif

      if (tracing)
	fprintf (stderr, "%.4x: inst = %.4x ", pc, inst);

      oldpc = pc;

      pc += 2;

      switch (inst >> 8)
	{
	case 0x00:
	  switch RS
	    {
	    case 0x0:
	      switch RD
		{
		case 0x0:				/* bkpt */
		  cpu.asregs.exception = SIGTRAP;
		  pc -= 2;
		  break;

		case 0x1:				/* sync */
		  break;

		case 0x2:				/* rte */
		  pc = cpu.epc;
		  cpu.sr = cpu.esr;
		  needfetch = 1;

		  if (SR_AF ())
		    cpu.asregs.active_gregs = & cpu.asregs.alt_gregs[0];
		  else
		    cpu.asregs.active_gregs = & cpu.asregs.gregs[0];
		  break;

		case 0x3:				/* rfi */
		  pc = cpu.fpc;
		  cpu.sr = cpu.fsr;
		  needfetch = 1;

		  if (SR_AF ())
		    cpu.asregs.active_gregs = &cpu.asregs.alt_gregs[0];
		  else
		    cpu.asregs.active_gregs = &cpu.asregs.gregs[0];
		  break;

		case 0x4:				/* stop */
		  if (issue_messages)
		    fprintf (stderr, "WARNING: stop unimplemented\n");
		  break;

		case 0x5:				/* wait */
		  if (issue_messages)
		    fprintf (stderr, "WARNING: wait unimplemented\n");
		  break;

		case 0x6:				/* doze */
		  if (issue_messages)
		    fprintf (stderr, "WARNING: doze unimplemented\n");
		  break;

		case 0x7:
		  cpu.asregs.exception = SIGILL;	/* illegal */
		  break;

		case 0x8:				/* trap 0 */
		case 0xA:				/* trap 2 */
		case 0xB:				/* trap 3 */
		  cpu.asregs.exception = SIGTRAP;
		  break;

		case 0xC:				/* trap 4 */
		case 0xD:				/* trap 5 */
		case 0xE:				/* trap 6 */
		  cpu.asregs.exception = SIGILL;	/* illegal */
		  break;

		case 0xF: 				/* trap 7 */
		  cpu.asregs.exception = SIGTRAP;	/* integer div-by-0 */
		  break;

		case 0x9:				/* trap 1 */
		  handle_trap1 (sd);
		  break;
		}
	      break;

	    case 0x1:
	      cpu.asregs.exception = SIGILL;		/* illegal */
	      break;

	    case 0x2:					/* mvc */
	      cpu.gr[RD] = C_VALUE();
	      break;
	    case 0x3:					/* mvcv */
	      cpu.gr[RD] = C_OFF();
	      break;
	    case 0x4:					/* ldq */
	      {
		word addr = cpu.gr[RD];
		int regno = 4;			/* always r4-r7 */

		bonus_cycles++;
		memops += 4;
		do
		  {
		    cpu.gr[regno] = rlat(addr);
		    addr += 4;
		    regno++;
		  }
		while ((regno&0x3) != 0);
	      }
	      break;
	    case 0x5:					/* stq */
	      {
		word addr = cpu.gr[RD];
		int regno = 4;			/* always r4-r7 */

		memops += 4;
		bonus_cycles++;
		do
		  {
		    wlat(addr, cpu.gr[regno]);
		    addr += 4;
		    regno++;
		  }
		while ((regno & 0x3) != 0);
	      }
	      break;
	    case 0x6:					/* ldm */
	      {
		word addr = cpu.gr[0];
		int regno = RD;

		/* bonus cycle is really only needed if
		   the next insn shifts the last reg loaded.

		   bonus_cycles++;
		*/
		memops += 16-regno;
		while (regno <= 0xF)
		  {
		    cpu.gr[regno] = rlat(addr);
		    addr += 4;
		    regno++;
		  }
	      }
	      break;
	    case 0x7:					/* stm */
	      {
		word addr = cpu.gr[0];
		int regno = RD;

		/* this should be removed! */
		/*  bonus_cycles ++; */

		memops += 16 - regno;
		while (regno <= 0xF)
		  {
		    wlat(addr, cpu.gr[regno]);
		    addr += 4;
		    regno++;
		  }
	      }
	      break;

	    case 0x8:					/* dect */
	      cpu.gr[RD] -= C_VALUE();
	      break;
	    case 0x9:					/* decf */
	      cpu.gr[RD] -= C_OFF();
	      break;
	    case 0xA:					/* inct */
	      cpu.gr[RD] += C_VALUE();
	      break;
	    case 0xB:					/* incf */
	      cpu.gr[RD] += C_OFF();
	      break;
	    case 0xC:					/* jmp */
	      pc = cpu.gr[RD];
	      if (tracing && RD == 15)
		fprintf (stderr, "Func return, r2 = %x, r3 = %x\n",
			 cpu.gr[2], cpu.gr[3]);
	      bonus_cycles++;
	      needfetch = 1;
	      break;
	    case 0xD:					/* jsr */
	      cpu.gr[15] = pc;
	      pc = cpu.gr[RD];
	      bonus_cycles++;
	      needfetch = 1;
	      break;
	    case 0xE:					/* ff1 */
	      {
		word tmp, i;
		tmp = cpu.gr[RD];
		for (i = 0; !(tmp & 0x80000000) && i < 32; i++)
		  tmp <<= 1;
		cpu.gr[RD] = i;
	      }
	      break;
	    case 0xF:					/* brev */
	      {
		word tmp;
		tmp = cpu.gr[RD];
		tmp = ((tmp & 0xaaaaaaaa) >>  1) | ((tmp & 0x55555555) <<  1);
		tmp = ((tmp & 0xcccccccc) >>  2) | ((tmp & 0x33333333) <<  2);
		tmp = ((tmp & 0xf0f0f0f0) >>  4) | ((tmp & 0x0f0f0f0f) <<  4);
		tmp = ((tmp & 0xff00ff00) >>  8) | ((tmp & 0x00ff00ff) <<  8);
		cpu.gr[RD] = ((tmp & 0xffff0000) >> 16) | ((tmp & 0x0000ffff) << 16);
	      }
	      break;
	    }
	  break;
	case 0x01:
	  switch RS
	    {
	    case 0x0:					/* xtrb3 */
	      cpu.gr[1] = (cpu.gr[RD]) & 0xFF;
	      NEW_C (cpu.gr[RD] != 0);
	      break;
	    case 0x1:					/* xtrb2 */
	      cpu.gr[1] = (cpu.gr[RD]>>8) & 0xFF;
	      NEW_C (cpu.gr[RD] != 0);
	      break;
	    case 0x2:					/* xtrb1 */
	      cpu.gr[1] = (cpu.gr[RD]>>16) & 0xFF;
	      NEW_C (cpu.gr[RD] != 0);
	      break;
	    case 0x3:					/* xtrb0 */
	      cpu.gr[1] = (cpu.gr[RD]>>24) & 0xFF;
	      NEW_C (cpu.gr[RD] != 0);
	      break;
	    case 0x4:					/* zextb */
	      cpu.gr[RD] &= 0x000000FF;
	      break;
	    case 0x5:					/* sextb */
	      {
		long tmp;
		tmp = cpu.gr[RD];
		tmp <<= 24;
		tmp >>= 24;
		cpu.gr[RD] = tmp;
	      }
	      break;
	    case 0x6:					/* zexth */
	      cpu.gr[RD] &= 0x0000FFFF;
	      break;
	    case 0x7:					/* sexth */
	      {
		long tmp;
		tmp = cpu.gr[RD];
		tmp <<= 16;
		tmp >>= 16;
		cpu.gr[RD] = tmp;
	      }
	      break;
	    case 0x8:					/* declt */
	      --cpu.gr[RD];
	      NEW_C ((long)cpu.gr[RD] < 0);
	      break;
	    case 0x9:					/* tstnbz */
	      {
		word tmp = cpu.gr[RD];
		NEW_C ((tmp & 0xFF000000) != 0 &&
		       (tmp & 0x00FF0000) != 0 && (tmp & 0x0000FF00) != 0 &&
		       (tmp & 0x000000FF) != 0);
	      }
	      break;
	    case 0xA:					/* decgt */
	      --cpu.gr[RD];
	      NEW_C ((long)cpu.gr[RD] > 0);
	      break;
	    case 0xB:					/* decne */
	      --cpu.gr[RD];
	      NEW_C ((long)cpu.gr[RD] != 0);
	      break;
	    case 0xC:					/* clrt */
	      if (C_ON())
		cpu.gr[RD] = 0;
	      break;
	    case 0xD:					/* clrf */
	      if (C_OFF())
		cpu.gr[RD] = 0;
	      break;
	    case 0xE:					/* abs */
	      if (cpu.gr[RD] & 0x80000000)
		cpu.gr[RD] = ~cpu.gr[RD] + 1;
	      break;
	    case 0xF:					/* not */
	      cpu.gr[RD] = ~cpu.gr[RD];
	      break;
	    }
	  break;
	case 0x02:					/* movt */
	  if (C_ON())
	    cpu.gr[RD] = cpu.gr[RS];
	  break;
	case 0x03:					/* mult */
	  /* consume 2 bits per cycle from rs, until rs is 0 */
	  {
	    unsigned int t = cpu.gr[RS];
	    int ticks;
	    for (ticks = 0; t != 0 ; t >>= 2)
	      ticks++;
	    bonus_cycles += ticks;
	  }
	  bonus_cycles += 2;  /* min. is 3, so add 2, plus ticks above */
	  if (tracing)
	    fprintf (stderr, "  mult %x by %x to give %x",
		     cpu.gr[RD], cpu.gr[RS], cpu.gr[RD] * cpu.gr[RS]);
	  cpu.gr[RD] = cpu.gr[RD] * cpu.gr[RS];
	  break;
	case 0x04:					/* loopt */
	  if (C_ON())
	    {
	      pc += (IMM4 << 1) - 32;
	      bonus_cycles ++;
	      needfetch = 1;
	    }
	  --cpu.gr[RS];				/* not RD! */
	  NEW_C (((long)cpu.gr[RS]) > 0);
	  break;
	case 0x05:					/* subu */
	  cpu.gr[RD] -= cpu.gr[RS];
	  break;
	case 0x06:					/* addc */
	  {
	    unsigned long tmp, a, b;
	    a = cpu.gr[RD];
	    b = cpu.gr[RS];
	    cpu.gr[RD] = a + b + C_VALUE ();
	    tmp = iu_carry (a, b, C_VALUE ());
	    NEW_C (tmp);
	  }
	  break;
	case 0x07:					/* subc */
	  {
	    unsigned long tmp, a, b;
	    a = cpu.gr[RD];
	    b = cpu.gr[RS];
	    cpu.gr[RD] = a - b + C_VALUE () - 1;
	    tmp = iu_carry (a,~b, C_VALUE ());
	    NEW_C (tmp);
	  }
	  break;
	case 0x08:					/* illegal */
	case 0x09:					/* illegal*/
	  cpu.asregs.exception = SIGILL;
	  break;
	case 0x0A:					/* movf */
	  if (C_OFF())
	    cpu.gr[RD] = cpu.gr[RS];
	  break;
	case 0x0B:					/* lsr */
	  {
	    unsigned long dst, src;
	    dst = cpu.gr[RD];
	    src = cpu.gr[RS];
	    /* We must not rely solely upon the native shift operations, since they
	       may not match the M*Core's behaviour on boundary conditions.  */
	    dst = src > 31 ? 0 : dst >> src;
	    cpu.gr[RD] = dst;
	  }
	  break;
	case 0x0C:					/* cmphs */
	  NEW_C ((unsigned long )cpu.gr[RD] >=
		 (unsigned long)cpu.gr[RS]);
	  break;
	case 0x0D:					/* cmplt */
	  NEW_C ((long)cpu.gr[RD] < (long)cpu.gr[RS]);
	  break;
	case 0x0E:					/* tst */
	  NEW_C ((cpu.gr[RD] & cpu.gr[RS]) != 0);
	  break;
	case 0x0F:					/* cmpne */
	  NEW_C (cpu.gr[RD] != cpu.gr[RS]);
	  break;
	case 0x10: case 0x11:				/* mfcr */
	  {
	    unsigned r;
	    r = IMM5;
	    if (r <= LAST_VALID_CREG)
	      cpu.gr[RD] = cpu.cr[r];
	    else
	      cpu.asregs.exception = SIGILL;
	  }
	  break;

	case 0x12:					/* mov */
	  cpu.gr[RD] = cpu.gr[RS];
	  if (tracing)
	    fprintf (stderr, "MOV %x into reg %d", cpu.gr[RD], RD);
	  break;

	case 0x13:					/* bgenr */
	  if (cpu.gr[RS] & 0x20)
	    cpu.gr[RD] = 0;
	  else
	    cpu.gr[RD] = 1 << (cpu.gr[RS] & 0x1F);
	  break;

	case 0x14:					/* rsub */
	  cpu.gr[RD] = cpu.gr[RS] - cpu.gr[RD];
	  break;

	case 0x15:					/* ixw */
	  cpu.gr[RD] += cpu.gr[RS]<<2;
	  break;

	case 0x16:					/* and */
	  cpu.gr[RD] &= cpu.gr[RS];
	  break;

	case 0x17:					/* xor */
	  cpu.gr[RD] ^= cpu.gr[RS];
	  break;

	case 0x18: case 0x19:				/* mtcr */
	  {
	    unsigned r;
	    r = IMM5;
	    if (r <= LAST_VALID_CREG)
	      cpu.cr[r] = cpu.gr[RD];
	    else
	      cpu.asregs.exception = SIGILL;

	    /* we might have changed register sets... */
	    if (SR_AF ())
	      cpu.asregs.active_gregs = & cpu.asregs.alt_gregs[0];
	    else
	      cpu.asregs.active_gregs = & cpu.asregs.gregs[0];
	  }
	  break;

	case 0x1A:					/* asr */
	  /* We must not rely solely upon the native shift operations, since they
	     may not match the M*Core's behaviour on boundary conditions.  */
	  if (cpu.gr[RS] > 30)
	    cpu.gr[RD] = ((long) cpu.gr[RD]) < 0 ? -1 : 0;
	  else
	    cpu.gr[RD] = (long) cpu.gr[RD] >> cpu.gr[RS];
	  break;

	case 0x1B:					/* lsl */
	  /* We must not rely solely upon the native shift operations, since they
	     may not match the M*Core's behaviour on boundary conditions.  */
	  cpu.gr[RD] = cpu.gr[RS] > 31 ? 0 : cpu.gr[RD] << cpu.gr[RS];
	  break;

	case 0x1C:					/* addu */
	  cpu.gr[RD] += cpu.gr[RS];
	  break;

	case 0x1D:					/* ixh */
	  cpu.gr[RD] += cpu.gr[RS] << 1;
	  break;

	case 0x1E:					/* or */
	  cpu.gr[RD] |= cpu.gr[RS];
	  break;

	case 0x1F:					/* andn */
	  cpu.gr[RD] &= ~cpu.gr[RS];
	  break;
	case 0x20: case 0x21:				/* addi */
	  cpu.gr[RD] =
	    cpu.gr[RD] + (IMM5 + 1);
	  break;
	case 0x22: case 0x23:				/* cmplti */
	  {
	    int tmp = (IMM5 + 1);
	    if (cpu.gr[RD] < tmp)
	      {
	        SET_C();
	      }
	    else
	      {
	        CLR_C();
	      }
	  }
	  break;
	case 0x24: case 0x25:				/* subi */
	  cpu.gr[RD] =
	    cpu.gr[RD] - (IMM5 + 1);
	  break;
	case 0x26: case 0x27:				/* illegal */
	  cpu.asregs.exception = SIGILL;
	  break;
	case 0x28: case 0x29:				/* rsubi */
	  cpu.gr[RD] =
	    IMM5 - cpu.gr[RD];
	  break;
	case 0x2A: case 0x2B:				/* cmpnei */
	  if (cpu.gr[RD] != IMM5)
	    {
	      SET_C();
	    }
	  else
	    {
	      CLR_C();
	    }
	  break;

	case 0x2C: case 0x2D:				/* bmaski, divu */
	  {
	    unsigned imm = IMM5;

	    if (imm == 1)
	      {
		int exe;
		int rxnlz, r1nlz;
		unsigned int rx, r1;

		rx = cpu.gr[RD];
		r1 = cpu.gr[1];
		exe = 0;

		/* unsigned divide */
		cpu.gr[RD] = (word) ((unsigned int) cpu.gr[RD] / (unsigned int)cpu.gr[1] );

		/* compute bonus_cycles for divu */
		for (r1nlz = 0; ((r1 & 0x80000000) == 0) && (r1nlz < 32); r1nlz ++)
		  r1 = r1 << 1;

		for (rxnlz = 0; ((rx & 0x80000000) == 0) && (rxnlz < 32); rxnlz ++)
		  rx = rx << 1;

		if (r1nlz < rxnlz)
		  exe += 4;
		else
		  exe += 5 + r1nlz - rxnlz;

		if (exe >= (2 * memcycles - 1))
		  {
		    bonus_cycles += exe - (2 * memcycles) + 1;
		  }
	      }
	    else if (imm == 0 || imm >= 8)
	      {
		/* bmaski */
		if (imm == 0)
		  cpu.gr[RD] = -1;
		else
		  cpu.gr[RD] = (1 << imm) - 1;
	      }
	    else
	      {
		/* illegal */
		cpu.asregs.exception = SIGILL;
	      }
	  }
	  break;
	case 0x2E: case 0x2F:				/* andi */
	  cpu.gr[RD] = cpu.gr[RD] & IMM5;
	  break;
	case 0x30: case 0x31:				/* bclri */
	  cpu.gr[RD] = cpu.gr[RD] & ~(1<<IMM5);
	  break;
	case 0x32: case 0x33:				/* bgeni, divs */
	  {
	    unsigned imm = IMM5;
	    if (imm == 1)
	      {
		int exe,sc;
		int rxnlz, r1nlz;
		signed int rx, r1;

		/* compute bonus_cycles for divu */
		rx = cpu.gr[RD];
		r1 = cpu.gr[1];
		exe = 0;

		if (((rx < 0) && (r1 > 0)) || ((rx >= 0) && (r1 < 0)))
		  sc = 1;
		else
		  sc = 0;

		rx = abs (rx);
		r1 = abs (r1);

		/* signed divide, general registers are of type int, so / op is OK */
		cpu.gr[RD] = cpu.gr[RD] / cpu.gr[1];

		for (r1nlz = 0; ((r1 & 0x80000000) == 0) && (r1nlz < 32) ; r1nlz ++ )
		  r1 = r1 << 1;

		for (rxnlz = 0; ((rx & 0x80000000) == 0) && (rxnlz < 32) ; rxnlz ++ )
		  rx = rx << 1;

		if (r1nlz < rxnlz)
		  exe += 5;
		else
		  exe += 6 + r1nlz - rxnlz + sc;

		if (exe >= (2 * memcycles - 1))
		  {
		    bonus_cycles += exe - (2 * memcycles) + 1;
		  }
	      }
	    else if (imm >= 7)
	      {
		/* bgeni */
		cpu.gr[RD] = (1 << IMM5);
	      }
	    else
	      {
		/* illegal */
		cpu.asregs.exception = SIGILL;
	      }
	    break;
	  }
	case 0x34: case 0x35:				/* bseti */
	  cpu.gr[RD] = cpu.gr[RD] | (1 << IMM5);
	  break;
	case 0x36: case 0x37:				/* btsti */
	  NEW_C (cpu.gr[RD] >> IMM5);
	  break;
	case 0x38: case 0x39:				/* xsr, rotli */
	  {
	    unsigned imm = IMM5;
	    unsigned long tmp = cpu.gr[RD];
	    if (imm == 0)
	      {
		word cbit;
		cbit = C_VALUE();
		NEW_C (tmp);
		cpu.gr[RD] = (cbit << 31) | (tmp >> 1);
	      }
	    else
	      cpu.gr[RD] = (tmp << imm) | (tmp >> (32 - imm));
	  }
	  break;
	case 0x3A: case 0x3B:				/* asrc, asri */
	  {
	    unsigned imm = IMM5;
	    long tmp = cpu.gr[RD];
	    if (imm == 0)
	      {
		NEW_C (tmp);
		cpu.gr[RD] = tmp >> 1;
	      }
	    else
	      cpu.gr[RD] = tmp >> imm;
	  }
	  break;
	case 0x3C: case 0x3D:				/* lslc, lsli */
	  {
	    unsigned imm = IMM5;
	    unsigned long tmp = cpu.gr[RD];
	    if (imm == 0)
	      {
		NEW_C (tmp >> 31);
		cpu.gr[RD] = tmp << 1;
	      }
	    else
	      cpu.gr[RD] = tmp << imm;
	  }
	  break;
	case 0x3E: case 0x3F:				/* lsrc, lsri */
	  {
	    unsigned imm = IMM5;
	    unsigned long tmp = cpu.gr[RD];
	    if (imm == 0)
	      {
		NEW_C (tmp);
		cpu.gr[RD] = tmp >> 1;
	      }
	    else
	      cpu.gr[RD] = tmp >> imm;
	  }
	  break;
	case 0x40: case 0x41: case 0x42: case 0x43:
	case 0x44: case 0x45: case 0x46: case 0x47:
	case 0x48: case 0x49: case 0x4A: case 0x4B:
	case 0x4C: case 0x4D: case 0x4E: case 0x4F:
	  cpu.asregs.exception = SIGILL;
	  break;
	case 0x50:
	  util (sd, inst & 0xFF);
	  break;
	case 0x51: case 0x52: case 0x53:
	case 0x54: case 0x55: case 0x56: case 0x57:
	case 0x58: case 0x59: case 0x5A: case 0x5B:
	case 0x5C: case 0x5D: case 0x5E: case 0x5F:
	  cpu.asregs.exception = SIGILL;
	  break;
	case 0x60: case 0x61: case 0x62: case 0x63:	/* movi  */
	case 0x64: case 0x65: case 0x66: case 0x67:
	  cpu.gr[RD] = (inst >> 4) & 0x7F;
	  break;
	case 0x68: case 0x69: case 0x6A: case 0x6B:
	case 0x6C: case 0x6D: case 0x6E: case 0x6F:	/* illegal */
	  cpu.asregs.exception = SIGILL;
	  break;
	case 0x71: case 0x72: case 0x73:
	case 0x74: case 0x75: case 0x76: case 0x77:
	case 0x78: case 0x79: case 0x7A: case 0x7B:
	case 0x7C: case 0x7D: case 0x7E:		/* lrw */
	  cpu.gr[RX] =  rlat ((pc + ((inst & 0xFF) << 2)) & 0xFFFFFFFC);
	  if (tracing)
	    fprintf (stderr, "LRW of 0x%x from 0x%x to reg %d",
		     rlat ((pc + ((inst & 0xFF) << 2)) & 0xFFFFFFFC),
		     (pc + ((inst & 0xFF) << 2)) & 0xFFFFFFFC, RX);
	  memops++;
	  break;
	case 0x7F:					/* jsri */
	  cpu.gr[15] = pc;
	  if (tracing)
	    fprintf (stderr, "func call: r2 = %x r3 = %x r4 = %x r5 = %x r6 = %x r7 = %x\n",
		     cpu.gr[2], cpu.gr[3], cpu.gr[4], cpu.gr[5], cpu.gr[6], cpu.gr[7]);
	case 0x70:					/* jmpi */
	  pc = rlat ((pc + ((inst & 0xFF) << 2)) & 0xFFFFFFFC);
	  memops++;
	  bonus_cycles++;
	  needfetch = 1;
	  break;

	case 0x80: case 0x81: case 0x82: case 0x83:
	case 0x84: case 0x85: case 0x86: case 0x87:
	case 0x88: case 0x89: case 0x8A: case 0x8B:
	case 0x8C: case 0x8D: case 0x8E: case 0x8F:	/* ld */
	  cpu.gr[RX] = rlat (cpu.gr[RD] + ((inst >> 2) & 0x003C));
	  if (tracing)
	    fprintf (stderr, "load reg %d from 0x%x with 0x%x",
		     RX,
		     cpu.gr[RD] + ((inst >> 2) & 0x003C), cpu.gr[RX]);
	  memops++;
	  break;
	case 0x90: case 0x91: case 0x92: case 0x93:
	case 0x94: case 0x95: case 0x96: case 0x97:
	case 0x98: case 0x99: case 0x9A: case 0x9B:
	case 0x9C: case 0x9D: case 0x9E: case 0x9F:	/* st */
	  wlat (cpu.gr[RD] + ((inst >> 2) & 0x003C), cpu.gr[RX]);
	  if (tracing)
	    fprintf (stderr, "store reg %d (containing 0x%x) to 0x%x",
		     RX, cpu.gr[RX],
		     cpu.gr[RD] + ((inst >> 2) & 0x003C));
	  memops++;
	  break;
	case 0xA0: case 0xA1: case 0xA2: case 0xA3:
	case 0xA4: case 0xA5: case 0xA6: case 0xA7:
	case 0xA8: case 0xA9: case 0xAA: case 0xAB:
	case 0xAC: case 0xAD: case 0xAE: case 0xAF:	/* ld.b */
	  cpu.gr[RX] = rbat (cpu.gr[RD] + RS);
	  memops++;
	  break;
	case 0xB0: case 0xB1: case 0xB2: case 0xB3:
	case 0xB4: case 0xB5: case 0xB6: case 0xB7:
	case 0xB8: case 0xB9: case 0xBA: case 0xBB:
	case 0xBC: case 0xBD: case 0xBE: case 0xBF:	/* st.b */
	  wbat (cpu.gr[RD] + RS, cpu.gr[RX]);
	  memops++;
	  break;
	case 0xC0: case 0xC1: case 0xC2: case 0xC3:
	case 0xC4: case 0xC5: case 0xC6: case 0xC7:
	case 0xC8: case 0xC9: case 0xCA: case 0xCB:
	case 0xCC: case 0xCD: case 0xCE: case 0xCF:	/* ld.h */
	  cpu.gr[RX] = rhat (cpu.gr[RD] + ((inst >> 3) & 0x001E));
	  memops++;
	  break;
	case 0xD0: case 0xD1: case 0xD2: case 0xD3:
	case 0xD4: case 0xD5: case 0xD6: case 0xD7:
	case 0xD8: case 0xD9: case 0xDA: case 0xDB:
	case 0xDC: case 0xDD: case 0xDE: case 0xDF:	/* st.h */
	  what (cpu.gr[RD] + ((inst >> 3) & 0x001E), cpu.gr[RX]);
	  memops++;
	  break;
	case 0xE8: case 0xE9: case 0xEA: case 0xEB:
	case 0xEC: case 0xED: case 0xEE: case 0xEF:	/* bf */
	  if (C_OFF())
	    {
	      int disp;
	      disp = inst & 0x03FF;
	      if (inst & 0x0400)
		disp |= 0xFFFFFC00;
	      pc += disp<<1;
	      bonus_cycles++;
	      needfetch = 1;
	    }
	  break;
	case 0xE0: case 0xE1: case 0xE2: case 0xE3:
	case 0xE4: case 0xE5: case 0xE6: case 0xE7:	/* bt */
	  if (C_ON())
	    {
	      int disp;
	      disp = inst & 0x03FF;
	      if (inst & 0x0400)
		disp |= 0xFFFFFC00;
	      pc += disp<<1;
	      bonus_cycles++;
	      needfetch = 1;
	    }
	  break;

	case 0xF8: case 0xF9: case 0xFA: case 0xFB:
	case 0xFC: case 0xFD: case 0xFE: case 0xFF:	/* bsr */
	  cpu.gr[15] = pc;
	case 0xF0: case 0xF1: case 0xF2: case 0xF3:
	case 0xF4: case 0xF5: case 0xF6: case 0xF7:	/* br */
	  {
	    int disp;
	    disp = inst & 0x03FF;
	    if (inst & 0x0400)
	      disp |= 0xFFFFFC00;
	    pc += disp<<1;
	    bonus_cycles++;
	    needfetch = 1;
	  }
	  break;

	}

      if (tracing)
	fprintf (stderr, "\n");

      if (needfetch)
	{
	  /* Do not let him fetch from a bad address! */
	  if (((uword)pc) >= cpu.asregs.msize)
	    {
	      if (issue_messages)
		fprintf (stderr, "PC loaded at 0x%x is outside of available memory! (0x%x)\n", oldpc, pc);

	      cpu.asregs.exception = SIGSEGV;
	    }
	  else
	    {
	      ibuf = rlat (pc & 0xFFFFFFFC);
	      needfetch = 0;
	    }
	}
    }
  while (!cpu.asregs.exception);

  /* Hide away the things we've cached while executing.  */
  CPU_PC_SET (scpu, pc);
  cpu.asregs.insts += insts;		/* instructions done ... */
  cpu.asregs.cycles += insts;		/* and each takes a cycle */
  cpu.asregs.cycles += bonus_cycles;	/* and extra cycles for branches */
  cpu.asregs.cycles += memops * memcycles;	/* and memop cycle delays */
}


int
sim_write (SIM_DESC sd, SIM_ADDR addr, const unsigned char *buffer, int size)
{
  int i;
  init_pointers ();

  memcpy (& cpu.mem[addr], buffer, size);

  return size;
}

int
sim_read (SIM_DESC sd, SIM_ADDR addr, unsigned char *buffer, int size)
{
  int i;
  init_pointers ();

  memcpy (buffer, & cpu.mem[addr], size);

  return size;
}


int
sim_store_register (SIM_DESC sd, int rn, unsigned char *memory, int length)
{
  init_pointers ();

  if (rn < NUM_MCORE_REGS && rn >= 0)
    {
      if (length == 4)
	{
	  long ival;

	  /* misalignment safe */
	  ival = mcore_extract_unsigned_integer (memory, 4);
	  cpu.asints[rn] = ival;
	}

      return 4;
    }
  else
    return 0;
}

int
sim_fetch_register (SIM_DESC sd, int rn, unsigned char *memory, int length)
{
  init_pointers ();

  if (rn < NUM_MCORE_REGS && rn >= 0)
    {
      if (length == 4)
	{
	  long ival = cpu.asints[rn];

	  /* misalignment-safe */
	  mcore_store_unsigned_integer (memory, 4, ival);
	}

      return 4;
    }
  else
    return 0;
}

void
sim_stop_reason (SIM_DESC sd, enum sim_stop *reason, int *sigrc)
{
  if (cpu.asregs.exception == SIGQUIT)
    {
      * reason = sim_exited;
      * sigrc = cpu.gr[PARM1];
    }
  else
    {
      * reason = sim_stopped;
      * sigrc = cpu.asregs.exception;
    }
}

void
sim_info (SIM_DESC sd, int verbose)
{
#ifdef WATCHFUNCTIONS
  int w, wcyc;
#endif
  double virttime = cpu.asregs.cycles / 36.0e6;
  host_callback *callback = STATE_CALLBACK (sd);

  callback->printf_filtered (callback, "\n\n# instructions executed  %10d\n",
			     cpu.asregs.insts);
  callback->printf_filtered (callback, "# cycles                 %10d\n",
			     cpu.asregs.cycles);
  callback->printf_filtered (callback, "# pipeline stalls        %10d\n",
			     cpu.asregs.stalls);
  callback->printf_filtered (callback, "# virtual time taken     %10.4f\n",
			     virttime);

#ifdef WATCHFUNCTIONS
  callback->printf_filtered (callback, "\nNumber of watched functions: %d\n",
			     ENDWL);

  wcyc = 0;

  for (w = 1; w <= ENDWL; w++)
    {
      callback->printf_filtered (callback, "WL = %s %8x\n",WLstr[w],WL[w]);
      callback->printf_filtered (callback, "  calls = %d, cycles = %d\n",
				 WLcnts[w],WLcyc[w]);

      if (WLcnts[w] != 0)
	callback->printf_filtered (callback,
				   "  maxcpc = %d, mincpc = %d, avecpc = %d\n",
				   WLmax[w],WLmin[w],WLcyc[w]/WLcnts[w]);
      wcyc += WLcyc[w];
    }

  callback->printf_filtered (callback,
			     "Total cycles for watched functions: %d\n",wcyc);
#endif
}

static sim_cia
mcore_pc_get (sim_cpu *cpu)
{
  return cpu->pc;
}

static void
mcore_pc_set (sim_cpu *cpu, sim_cia pc)
{
  cpu->pc = pc;
}

static void
free_state (SIM_DESC sd)
{
  if (STATE_MODULES (sd) != NULL)
    sim_module_uninstall (sd);
  sim_cpu_free_all (sd);
  sim_state_free (sd);
}

SIM_DESC
sim_open (SIM_OPEN_KIND kind, host_callback *cb, struct bfd *abfd, char **argv)
{
  SIM_DESC sd = sim_state_alloc (kind, cb);
  int i, osize;
  SIM_ASSERT (STATE_MAGIC (sd) == SIM_MAGIC_NUMBER);

  /* The cpu data is kept in a separately allocated chunk of memory.  */
  if (sim_cpu_alloc_all (sd, 1, /*cgen_cpu_max_extra_bytes ()*/0) != SIM_RC_OK)
    {
      free_state (sd);
      return 0;
    }

  if (sim_pre_argv_init (sd, argv[0]) != SIM_RC_OK)
    {
      free_state (sd);
      return 0;
    }

  /* getopt will print the error message so we just have to exit if this fails.
     FIXME: Hmmm...  in the case of gdb we need getopt to call
     print_filtered.  */
  if (sim_parse_args (sd, argv) != SIM_RC_OK)
    {
      free_state (sd);
      return 0;
    }

  /* Check for/establish the a reference program image.  */
  if (sim_analyze_program (sd,
			   (STATE_PROG_ARGV (sd) != NULL
			    ? *STATE_PROG_ARGV (sd)
			    : NULL), abfd) != SIM_RC_OK)
    {
      free_state (sd);
      return 0;
    }

  /* Configure/verify the target byte order and other runtime
     configuration options.  */
  if (sim_config (sd) != SIM_RC_OK)
    {
      sim_module_uninstall (sd);
      return 0;
    }

  if (sim_post_argv_init (sd) != SIM_RC_OK)
    {
      /* Uninstall the modules to avoid memory leaks,
	 file descriptor leaks, etc.  */
      sim_module_uninstall (sd);
      return 0;
    }

  osize = sim_memory_size;

  if (kind == SIM_OPEN_STANDALONE)
    issue_messages = 1;

  /* Discard and reacquire memory -- start with a clean slate.  */
  sim_size (1);		/* small */
  sim_size (osize);	/* and back again */

  /* CPU specific initialization.  */
  for (i = 0; i < MAX_NR_PROCESSORS; ++i)
    {
      SIM_CPU *cpu = STATE_CPU (sd, i);

      CPU_PC_FETCH (cpu) = mcore_pc_get;
      CPU_PC_STORE (cpu) = mcore_pc_set;

      set_initial_gprs (cpu);	/* Reset the GPR registers.  */
    }

  return sd;
}

void
sim_close (SIM_DESC sd, int quitting)
{
  /* nothing to do */
}

SIM_RC
sim_create_inferior (SIM_DESC sd, struct bfd *prog_bfd, char **argv, char **env)
{
  SIM_CPU *scpu = STATE_CPU (sd, 0);
  char ** avp;
  int nargs = 0;
  int nenv = 0;
  int s_length;
  int l;
  unsigned long strings;
  unsigned long pointers;
  unsigned long hi_stack;


  /* Set the initial register set.  */
  l = issue_messages;
  issue_messages = 0;
  set_initial_gprs (scpu);
  issue_messages = l;

  hi_stack = cpu.asregs.msize - 4;
  CPU_PC_SET (scpu, bfd_get_start_address (prog_bfd));

  /* Calculate the argument and environment strings.  */
  s_length = 0;
  nargs = 0;
  avp = argv;
  while (avp && *avp)
    {
      l = strlen (*avp) + 1;	/* include the null */
      s_length += (l + 3) & ~3;	/* make it a 4 byte boundary */
      nargs++; avp++;
    }

  nenv = 0;
  avp = env;
  while (avp && *avp)
    {
      l = strlen (*avp) + 1;	/* include the null */
      s_length += (l + 3) & ~ 3;/* make it a 4 byte boundary */
      nenv++; avp++;
    }

  /* Claim some memory for the pointers and strings. */
  pointers = hi_stack - sizeof(word) * (nenv+1+nargs+1);
  pointers &= ~3;		/* must be 4-byte aligned */
  cpu.gr[0] = pointers;

  strings = cpu.gr[0] - s_length;
  strings &= ~3;		/* want to make it 4-byte aligned */
  cpu.gr[0] = strings;
  /* dac fix, the stack address must be 8-byte aligned! */
  cpu.gr[0] = cpu.gr[0] - cpu.gr[0] % 8;

  /* Loop through the arguments and fill them in.  */
  cpu.gr[PARM1] = nargs;
  if (nargs == 0)
    {
      /* No strings to fill in.  */
      cpu.gr[PARM2] = 0;
    }
  else
    {
      cpu.gr[PARM2] = pointers;
      avp = argv;
      while (avp && *avp)
	{
	  /* Save where we're putting it.  */
	  wlat (pointers, strings);

	  /* Copy the string.  */
	  l = strlen (* avp) + 1;
	  strcpy ((char *)(cpu.mem + strings), *avp);

	  /* Bump the pointers.  */
	  avp++;
	  pointers += 4;
	  strings += l+1;
	}

      /* A null to finish the list.  */
      wlat (pointers, 0);
      pointers += 4;
    }

  /* Now do the environment pointers.  */
  if (nenv == 0)
    {
      /* No strings to fill in.  */
      cpu.gr[PARM3] = 0;
    }
  else
    {
      cpu.gr[PARM3] = pointers;
      avp = env;

      while (avp && *avp)
	{
	  /* Save where we're putting it.  */
	  wlat (pointers, strings);

	  /* Copy the string.  */
	  l = strlen (* avp) + 1;
	  strcpy ((char *)(cpu.mem + strings), *avp);

	  /* Bump the pointers.  */
	  avp++;
	  pointers += 4;
	  strings += l+1;
	}

      /* A null to finish the list.  */
      wlat (pointers, 0);
      pointers += 4;
    }

  return SIM_RC_OK;
}