/* Intel 386 target-dependent stuff.

   Copyright (C) 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997,
   1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
   Free Software Foundation, Inc.

   This file is part of GDB.

   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 "defs.h"
#include "arch-utils.h"
#include "command.h"
#include "dummy-frame.h"
#include "dwarf2-frame.h"
#include "doublest.h"
#include "frame.h"
#include "frame-base.h"
#include "frame-unwind.h"
#include "inferior.h"
#include "gdbcmd.h"
#include "gdbcore.h"
#include "gdbtypes.h"
#include "objfiles.h"
#include "osabi.h"
#include "regcache.h"
#include "reggroups.h"
#include "regset.h"
#include "symfile.h"
#include "symtab.h"
#include "target.h"
#include "value.h"
#include "dis-asm.h"

#include "gdb_assert.h"
#include "gdb_string.h"

#include "i386-tdep.h"
#include "i387-tdep.h"

/* Register names.  */

static char *i386_register_names[] =
{
  "eax",   "ecx",    "edx",   "ebx",
  "esp",   "ebp",    "esi",   "edi",
  "eip",   "eflags", "cs",    "ss",
  "ds",    "es",     "fs",    "gs",
  "st0",   "st1",    "st2",   "st3",
  "st4",   "st5",    "st6",   "st7",
  "fctrl", "fstat",  "ftag",  "fiseg",
  "fioff", "foseg",  "fooff", "fop",
  "xmm0",  "xmm1",   "xmm2",  "xmm3",
  "xmm4",  "xmm5",   "xmm6",  "xmm7",
  "mxcsr"
};

static const int i386_num_register_names = ARRAY_SIZE (i386_register_names);

/* Register names for MMX pseudo-registers.  */

static char *i386_mmx_names[] =
{
  "mm0", "mm1", "mm2", "mm3",
  "mm4", "mm5", "mm6", "mm7"
};

static const int i386_num_mmx_regs = ARRAY_SIZE (i386_mmx_names);

static int
i386_mmx_regnum_p (struct gdbarch *gdbarch, int regnum)
{
  int mm0_regnum = gdbarch_tdep (gdbarch)->mm0_regnum;

  if (mm0_regnum < 0)
    return 0;

  return (regnum >= mm0_regnum && regnum < mm0_regnum + i386_num_mmx_regs);
}

/* SSE register?  */

static int
i386_sse_regnum_p (struct gdbarch *gdbarch, int regnum)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (I387_NUM_XMM_REGS (tdep) == 0)
    return 0;

  return (I387_XMM0_REGNUM (tdep) <= regnum
	  && regnum < I387_MXCSR_REGNUM (tdep));
}

static int
i386_mxcsr_regnum_p (struct gdbarch *gdbarch, int regnum)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (I387_NUM_XMM_REGS (tdep) == 0)
    return 0;

  return (regnum == I387_MXCSR_REGNUM (tdep));
}

/* FP register?  */

int
i386_fp_regnum_p (struct gdbarch *gdbarch, int regnum)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (I387_ST0_REGNUM (tdep) < 0)
    return 0;

  return (I387_ST0_REGNUM (tdep) <= regnum
	  && regnum < I387_FCTRL_REGNUM (tdep));
}

int
i386_fpc_regnum_p (struct gdbarch *gdbarch, int regnum)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (I387_ST0_REGNUM (tdep) < 0)
    return 0;

  return (I387_FCTRL_REGNUM (tdep) <= regnum 
	  && regnum < I387_XMM0_REGNUM (tdep));
}

/* Return the name of register REGNUM.  */

const char *
i386_register_name (struct gdbarch *gdbarch, int regnum)
{
  if (i386_mmx_regnum_p (gdbarch, regnum))
    return i386_mmx_names[regnum - I387_MM0_REGNUM (gdbarch_tdep (gdbarch))];

  if (regnum >= 0 && regnum < i386_num_register_names)
    return i386_register_names[regnum];

  return NULL;
}

/* Convert a dbx register number REG to the appropriate register
   number used by GDB.  */

static int
i386_dbx_reg_to_regnum (struct gdbarch *gdbarch, int reg)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  /* This implements what GCC calls the "default" register map
     (dbx_register_map[]).  */

  if (reg >= 0 && reg <= 7)
    {
      /* General-purpose registers.  The debug info calls %ebp
         register 4, and %esp register 5.  */
      if (reg == 4)
        return 5;
      else if (reg == 5)
        return 4;
      else return reg;
    }
  else if (reg >= 12 && reg <= 19)
    {
      /* Floating-point registers.  */
      return reg - 12 + I387_ST0_REGNUM (tdep);
    }
  else if (reg >= 21 && reg <= 28)
    {
      /* SSE registers.  */
      return reg - 21 + I387_XMM0_REGNUM (tdep);
    }
  else if (reg >= 29 && reg <= 36)
    {
      /* MMX registers.  */
      return reg - 29 + I387_MM0_REGNUM (tdep);
    }

  /* This will hopefully provoke a warning.  */
  return gdbarch_num_regs (gdbarch) + gdbarch_num_pseudo_regs (gdbarch);
}

/* Convert SVR4 register number REG to the appropriate register number
   used by GDB.  */

static int
i386_svr4_reg_to_regnum (struct gdbarch *gdbarch, int reg)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  /* This implements the GCC register map that tries to be compatible
     with the SVR4 C compiler for DWARF (svr4_dbx_register_map[]).  */

  /* The SVR4 register numbering includes %eip and %eflags, and
     numbers the floating point registers differently.  */
  if (reg >= 0 && reg <= 9)
    {
      /* General-purpose registers.  */
      return reg;
    }
  else if (reg >= 11 && reg <= 18)
    {
      /* Floating-point registers.  */
      return reg - 11 + I387_ST0_REGNUM (tdep);
    }
  else if (reg >= 21 && reg <= 36)
    {
      /* The SSE and MMX registers have the same numbers as with dbx.  */
      return i386_dbx_reg_to_regnum (gdbarch, reg);
    }

  switch (reg)
    {
    case 37: return I387_FCTRL_REGNUM (tdep);
    case 38: return I387_FSTAT_REGNUM (tdep);
    case 39: return I387_MXCSR_REGNUM (tdep);
    case 40: return I386_ES_REGNUM;
    case 41: return I386_CS_REGNUM;
    case 42: return I386_SS_REGNUM;
    case 43: return I386_DS_REGNUM;
    case 44: return I386_FS_REGNUM;
    case 45: return I386_GS_REGNUM;
    }

  /* This will hopefully provoke a warning.  */
  return gdbarch_num_regs (gdbarch) + gdbarch_num_pseudo_regs (gdbarch);
}


/* This is the variable that is set with "set disassembly-flavor", and
   its legitimate values.  */
static const char att_flavor[] = "att";
static const char intel_flavor[] = "intel";
static const char *valid_flavors[] =
{
  att_flavor,
  intel_flavor,
  NULL
};
static const char *disassembly_flavor = att_flavor;


/* Use the program counter to determine the contents and size of a
   breakpoint instruction.  Return a pointer to a string of bytes that
   encode a breakpoint instruction, store the length of the string in
   *LEN and optionally adjust *PC to point to the correct memory
   location for inserting the breakpoint.

   On the i386 we have a single breakpoint that fits in a single byte
   and can be inserted anywhere.

   This function is 64-bit safe.  */

static const gdb_byte *
i386_breakpoint_from_pc (struct gdbarch *gdbarch, CORE_ADDR *pc, int *len)
{
  static gdb_byte break_insn[] = { 0xcc }; /* int 3 */

  *len = sizeof (break_insn);
  return break_insn;
}

/* Displaced instruction handling.  */


static int
i386_absolute_jmp_p (gdb_byte *insn)
{
  /* jmp far (absolute address in operand) */
  if (insn[0] == 0xea)
    return 1;

  if (insn[0] == 0xff)
    {
      /* jump near, absolute indirect (/4) */
      if ((insn[1] & 0x38) == 0x20)
        return 1;

      /* jump far, absolute indirect (/5) */
      if ((insn[1] & 0x38) == 0x28)
        return 1;
    }

  return 0;
}

static int
i386_absolute_call_p (gdb_byte *insn)
{
  /* call far, absolute */
  if (insn[0] == 0x9a)
    return 1;

  if (insn[0] == 0xff)
    {
      /* Call near, absolute indirect (/2) */
      if ((insn[1] & 0x38) == 0x10)
        return 1;

      /* Call far, absolute indirect (/3) */
      if ((insn[1] & 0x38) == 0x18)
        return 1;
    }

  return 0;
}

static int
i386_ret_p (gdb_byte *insn)
{
  switch (insn[0])
    {
    case 0xc2: /* ret near, pop N bytes */
    case 0xc3: /* ret near */
    case 0xca: /* ret far, pop N bytes */
    case 0xcb: /* ret far */
    case 0xcf: /* iret */
      return 1;

    default:
      return 0;
    }
}

static int
i386_call_p (gdb_byte *insn)
{
  if (i386_absolute_call_p (insn))
    return 1;

  /* call near, relative */
  if (insn[0] == 0xe8)
    return 1;

  return 0;
}

static int
i386_breakpoint_p (gdb_byte *insn)
{
  return insn[0] == 0xcc;       /* int 3 */
}

/* Return non-zero if INSN is a system call, and set *LENGTHP to its
   length in bytes.  Otherwise, return zero.  */
static int
i386_syscall_p (gdb_byte *insn, ULONGEST *lengthp)
{
  if (insn[0] == 0xcd)
    {
      *lengthp = 2;
      return 1;
    }

  return 0;
}

/* Fix up the state of registers and memory after having single-stepped
   a displaced instruction.  */
void
i386_displaced_step_fixup (struct gdbarch *gdbarch,
                           struct displaced_step_closure *closure,
                           CORE_ADDR from, CORE_ADDR to,
                           struct regcache *regs)
{
  /* The offset we applied to the instruction's address.
     This could well be negative (when viewed as a signed 32-bit
     value), but ULONGEST won't reflect that, so take care when
     applying it.  */
  ULONGEST insn_offset = to - from;

  /* Since we use simple_displaced_step_copy_insn, our closure is a
     copy of the instruction.  */
  gdb_byte *insn = (gdb_byte *) closure;

  if (debug_displaced)
    fprintf_unfiltered (gdb_stdlog,
                        "displaced: fixup (0x%s, 0x%s), "
                        "insn = 0x%02x 0x%02x ...\n",
                        paddr_nz (from), paddr_nz (to), insn[0], insn[1]);

  /* The list of issues to contend with here is taken from
     resume_execution in arch/i386/kernel/kprobes.c, Linux 2.6.20.
     Yay for Free Software!  */

  /* Relocate the %eip, if necessary.  */

  /* Except in the case of absolute or indirect jump or call
     instructions, or a return instruction, the new eip is relative to
     the displaced instruction; make it relative.  Well, signal
     handler returns don't need relocation either, but we use the
     value of %eip to recognize those; see below.  */
  if (! i386_absolute_jmp_p (insn)
      && ! i386_absolute_call_p (insn)
      && ! i386_ret_p (insn))
    {
      ULONGEST orig_eip;
      ULONGEST insn_len;

      regcache_cooked_read_unsigned (regs, I386_EIP_REGNUM, &orig_eip);

      /* A signal trampoline system call changes the %eip, resuming
         execution of the main program after the signal handler has
         returned.  That makes them like 'return' instructions; we
         shouldn't relocate %eip.

         But most system calls don't, and we do need to relocate %eip.

         Our heuristic for distinguishing these cases: if stepping
         over the system call instruction left control directly after
         the instruction, the we relocate --- control almost certainly
         doesn't belong in the displaced copy.  Otherwise, we assume
         the instruction has put control where it belongs, and leave
         it unrelocated.  Goodness help us if there are PC-relative
         system calls.  */
      if (i386_syscall_p (insn, &insn_len)
          && orig_eip != to + insn_len)
        {
          if (debug_displaced)
            fprintf_unfiltered (gdb_stdlog,
                                "displaced: syscall changed %%eip; "
                                "not relocating\n");
        }
      else
        {
          ULONGEST eip = (orig_eip - insn_offset) & 0xffffffffUL;

          /* If we have stepped over a breakpoint, set the %eip to
             point at the breakpoint instruction itself.

             (gdbarch_decr_pc_after_break was never something the core
             of GDB should have been concerned with; arch-specific
             code should be making PC values consistent before
             presenting them to GDB.)  */
          if (i386_breakpoint_p (insn))
            {
              if (debug_displaced)
                fprintf_unfiltered (gdb_stdlog,
                                    "displaced: stepped breakpoint\n");
              eip--;
            }

          regcache_cooked_write_unsigned (regs, I386_EIP_REGNUM, eip);

          if (debug_displaced)
            fprintf_unfiltered (gdb_stdlog,
                                "displaced: "
                                "relocated %%eip from 0x%s to 0x%s\n",
                                paddr_nz (orig_eip), paddr_nz (eip));
        }
    }

  /* If the instruction was PUSHFL, then the TF bit will be set in the
     pushed value, and should be cleared.  We'll leave this for later,
     since GDB already messes up the TF flag when stepping over a
     pushfl.  */

  /* If the instruction was a call, the return address now atop the
     stack is the address following the copied instruction.  We need
     to make it the address following the original instruction.  */
  if (i386_call_p (insn))
    {
      ULONGEST esp;
      ULONGEST retaddr;
      const ULONGEST retaddr_len = 4;

      regcache_cooked_read_unsigned (regs, I386_ESP_REGNUM, &esp);
      retaddr = read_memory_unsigned_integer (esp, retaddr_len);
      retaddr = (retaddr - insn_offset) & 0xffffffffUL;
      write_memory_unsigned_integer (esp, retaddr_len, retaddr);

      if (debug_displaced)
        fprintf_unfiltered (gdb_stdlog,
                            "displaced: relocated return addr at 0x%s "
                            "to 0x%s\n",
                            paddr_nz (esp),
                            paddr_nz (retaddr));
    }
}


#ifdef I386_REGNO_TO_SYMMETRY
#error "The Sequent Symmetry is no longer supported."
#endif

/* According to the System V ABI, the registers %ebp, %ebx, %edi, %esi
   and %esp "belong" to the calling function.  Therefore these
   registers should be saved if they're going to be modified.  */

/* The maximum number of saved registers.  This should include all
   registers mentioned above, and %eip.  */
#define I386_NUM_SAVED_REGS	I386_NUM_GREGS

struct i386_frame_cache
{
  /* Base address.  */
  CORE_ADDR base;
  LONGEST sp_offset;
  CORE_ADDR pc;

  /* Saved registers.  */
  CORE_ADDR saved_regs[I386_NUM_SAVED_REGS];
  CORE_ADDR saved_sp;
  int saved_sp_reg;
  int pc_in_eax;

  /* Stack space reserved for local variables.  */
  long locals;
};

/* Allocate and initialize a frame cache.  */

static struct i386_frame_cache *
i386_alloc_frame_cache (void)
{
  struct i386_frame_cache *cache;
  int i;

  cache = FRAME_OBSTACK_ZALLOC (struct i386_frame_cache);

  /* Base address.  */
  cache->base = 0;
  cache->sp_offset = -4;
  cache->pc = 0;

  /* Saved registers.  We initialize these to -1 since zero is a valid
     offset (that's where %ebp is supposed to be stored).  */
  for (i = 0; i < I386_NUM_SAVED_REGS; i++)
    cache->saved_regs[i] = -1;
  cache->saved_sp = 0;
  cache->saved_sp_reg = -1;
  cache->pc_in_eax = 0;

  /* Frameless until proven otherwise.  */
  cache->locals = -1;

  return cache;
}

/* If the instruction at PC is a jump, return the address of its
   target.  Otherwise, return PC.  */

static CORE_ADDR
i386_follow_jump (CORE_ADDR pc)
{
  gdb_byte op;
  long delta = 0;
  int data16 = 0;

  target_read_memory (pc, &op, 1);
  if (op == 0x66)
    {
      data16 = 1;
      op = read_memory_unsigned_integer (pc + 1, 1);
    }

  switch (op)
    {
    case 0xe9:
      /* Relative jump: if data16 == 0, disp32, else disp16.  */
      if (data16)
	{
	  delta = read_memory_integer (pc + 2, 2);

	  /* Include the size of the jmp instruction (including the
             0x66 prefix).  */
	  delta += 4;
	}
      else
	{
	  delta = read_memory_integer (pc + 1, 4);

	  /* Include the size of the jmp instruction.  */
	  delta += 5;
	}
      break;
    case 0xeb:
      /* Relative jump, disp8 (ignore data16).  */
      delta = read_memory_integer (pc + data16 + 1, 1);

      delta += data16 + 2;
      break;
    }

  return pc + delta;
}

/* Check whether PC points at a prologue for a function returning a
   structure or union.  If so, it updates CACHE and returns the
   address of the first instruction after the code sequence that
   removes the "hidden" argument from the stack or CURRENT_PC,
   whichever is smaller.  Otherwise, return PC.  */

static CORE_ADDR
i386_analyze_struct_return (CORE_ADDR pc, CORE_ADDR current_pc,
			    struct i386_frame_cache *cache)
{
  /* Functions that return a structure or union start with:

        popl %eax             0x58
        xchgl %eax, (%esp)    0x87 0x04 0x24
     or xchgl %eax, 0(%esp)   0x87 0x44 0x24 0x00

     (the System V compiler puts out the second `xchg' instruction,
     and the assembler doesn't try to optimize it, so the 'sib' form
     gets generated).  This sequence is used to get the address of the
     return buffer for a function that returns a structure.  */
  static gdb_byte proto1[3] = { 0x87, 0x04, 0x24 };
  static gdb_byte proto2[4] = { 0x87, 0x44, 0x24, 0x00 };
  gdb_byte buf[4];
  gdb_byte op;

  if (current_pc <= pc)
    return pc;

  target_read_memory (pc, &op, 1);

  if (op != 0x58)		/* popl %eax */
    return pc;

  target_read_memory (pc + 1, buf, 4);
  if (memcmp (buf, proto1, 3) != 0 && memcmp (buf, proto2, 4) != 0)
    return pc;

  if (current_pc == pc)
    {
      cache->sp_offset += 4;
      return current_pc;
    }

  if (current_pc == pc + 1)
    {
      cache->pc_in_eax = 1;
      return current_pc;
    }
  
  if (buf[1] == proto1[1])
    return pc + 4;
  else
    return pc + 5;
}

static CORE_ADDR
i386_skip_probe (CORE_ADDR pc)
{
  /* A function may start with

        pushl constant
        call _probe
	addl $4, %esp
	   
     followed by

        pushl %ebp

     etc.  */
  gdb_byte buf[8];
  gdb_byte op;

  target_read_memory (pc, &op, 1);

  if (op == 0x68 || op == 0x6a)
    {
      int delta;

      /* Skip past the `pushl' instruction; it has either a one-byte or a
	 four-byte operand, depending on the opcode.  */
      if (op == 0x68)
	delta = 5;
      else
	delta = 2;

      /* Read the following 8 bytes, which should be `call _probe' (6
	 bytes) followed by `addl $4,%esp' (2 bytes).  */
      read_memory (pc + delta, buf, sizeof (buf));
      if (buf[0] == 0xe8 && buf[6] == 0xc4 && buf[7] == 0x4)
	pc += delta + sizeof (buf);
    }

  return pc;
}

/* GCC 4.1 and later, can put code in the prologue to realign the
   stack pointer.  Check whether PC points to such code, and update
   CACHE accordingly.  Return the first instruction after the code
   sequence or CURRENT_PC, whichever is smaller.  If we don't
   recognize the code, return PC.  */

static CORE_ADDR
i386_analyze_stack_align (CORE_ADDR pc, CORE_ADDR current_pc,
			  struct i386_frame_cache *cache)
{
  /* There are 2 code sequences to re-align stack before the frame
     gets set up:

	1. Use a caller-saved saved register:

		leal  4(%esp), %reg
		andl  $-XXX, %esp
		pushl -4(%reg)

	2. Use a callee-saved saved register:

		pushl %reg
		leal  8(%esp), %reg
		andl  $-XXX, %esp
		pushl -4(%reg)

     "andl $-XXX, %esp" can be either 3 bytes or 6 bytes:
     
     	0x83 0xe4 0xf0			andl $-16, %esp
     	0x81 0xe4 0x00 0xff 0xff 0xff	andl $-256, %esp
   */

  gdb_byte buf[14];
  int reg;
  int offset, offset_and;
  static int regnums[8] = {
    I386_EAX_REGNUM,		/* %eax */
    I386_ECX_REGNUM,		/* %ecx */
    I386_EDX_REGNUM,		/* %edx */
    I386_EBX_REGNUM,		/* %ebx */
    I386_ESP_REGNUM,		/* %esp */
    I386_EBP_REGNUM,		/* %ebp */
    I386_ESI_REGNUM,		/* %esi */
    I386_EDI_REGNUM		/* %edi */
  };

  if (target_read_memory (pc, buf, sizeof buf))
    return pc;

  /* Check caller-saved saved register.  The first instruction has
     to be "leal 4(%esp), %reg".  */
  if (buf[0] == 0x8d && buf[2] == 0x24 && buf[3] == 0x4)
    {
      /* MOD must be binary 10 and R/M must be binary 100.  */
      if ((buf[1] & 0xc7) != 0x44)
	return pc;

      /* REG has register number.  */
      reg = (buf[1] >> 3) & 7;
      offset = 4;
    }
  else
    {
      /* Check callee-saved saved register.  The first instruction
	 has to be "pushl %reg".  */
      if ((buf[0] & 0xf8) != 0x50)
	return pc;

      /* Get register.  */
      reg = buf[0] & 0x7;

      /* The next instruction has to be "leal 8(%esp), %reg".  */
      if (buf[1] != 0x8d || buf[3] != 0x24 || buf[4] != 0x8)
	return pc;

      /* MOD must be binary 10 and R/M must be binary 100.  */
      if ((buf[2] & 0xc7) != 0x44)
	return pc;
      
      /* REG has register number.  Registers in pushl and leal have to
	 be the same.  */
      if (reg != ((buf[2] >> 3) & 7))
	return pc;

      offset = 5;
    }

  /* Rigister can't be %esp nor %ebp.  */
  if (reg == 4 || reg == 5)
    return pc;

  /* The next instruction has to be "andl $-XXX, %esp".  */
  if (buf[offset + 1] != 0xe4
      || (buf[offset] != 0x81 && buf[offset] != 0x83))
    return pc;

  offset_and = offset;
  offset += buf[offset] == 0x81 ? 6 : 3;

  /* The next instruction has to be "pushl -4(%reg)".  8bit -4 is
     0xfc.  REG must be binary 110 and MOD must be binary 01.  */
  if (buf[offset] != 0xff
      || buf[offset + 2] != 0xfc
      || (buf[offset + 1] & 0xf8) != 0x70)
    return pc;

  /* R/M has register.  Registers in leal and pushl have to be the
     same.  */
  if (reg != (buf[offset + 1] & 7))
    return pc;

  if (current_pc > pc + offset_and)
    cache->saved_sp_reg = regnums[reg];

  return min (pc + offset + 3, current_pc);
}

/* Maximum instruction length we need to handle.  */
#define I386_MAX_MATCHED_INSN_LEN	6

/* Instruction description.  */
struct i386_insn
{
  size_t len;
  gdb_byte insn[I386_MAX_MATCHED_INSN_LEN];
  gdb_byte mask[I386_MAX_MATCHED_INSN_LEN];
};

/* Search for the instruction at PC in the list SKIP_INSNS.  Return
   the first instruction description that matches.  Otherwise, return
   NULL.  */

static struct i386_insn *
i386_match_insn (CORE_ADDR pc, struct i386_insn *skip_insns)
{
  struct i386_insn *insn;
  gdb_byte op;

  target_read_memory (pc, &op, 1);

  for (insn = skip_insns; insn->len > 0; insn++)
    {
      if ((op & insn->mask[0]) == insn->insn[0])
	{
	  gdb_byte buf[I386_MAX_MATCHED_INSN_LEN - 1];
	  int insn_matched = 1;
	  size_t i;

	  gdb_assert (insn->len > 1);
	  gdb_assert (insn->len <= I386_MAX_MATCHED_INSN_LEN);

	  target_read_memory (pc + 1, buf, insn->len - 1);
	  for (i = 1; i < insn->len; i++)
	    {
	      if ((buf[i - 1] & insn->mask[i]) != insn->insn[i])
		insn_matched = 0;
	    }

	  if (insn_matched)
	    return insn;
	}
    }

  return NULL;
}

/* Some special instructions that might be migrated by GCC into the
   part of the prologue that sets up the new stack frame.  Because the
   stack frame hasn't been setup yet, no registers have been saved
   yet, and only the scratch registers %eax, %ecx and %edx can be
   touched.  */

struct i386_insn i386_frame_setup_skip_insns[] =
{
  /* Check for `movb imm8, r' and `movl imm32, r'. 
    
     ??? Should we handle 16-bit operand-sizes here?  */

  /* `movb imm8, %al' and `movb imm8, %ah' */
  /* `movb imm8, %cl' and `movb imm8, %ch' */
  { 2, { 0xb0, 0x00 }, { 0xfa, 0x00 } },
  /* `movb imm8, %dl' and `movb imm8, %dh' */
  { 2, { 0xb2, 0x00 }, { 0xfb, 0x00 } },
  /* `movl imm32, %eax' and `movl imm32, %ecx' */
  { 5, { 0xb8 }, { 0xfe } },
  /* `movl imm32, %edx' */
  { 5, { 0xba }, { 0xff } },

  /* Check for `mov imm32, r32'.  Note that there is an alternative
     encoding for `mov m32, %eax'.

     ??? Should we handle SIB adressing here?
     ??? Should we handle 16-bit operand-sizes here?  */

  /* `movl m32, %eax' */
  { 5, { 0xa1 }, { 0xff } },
  /* `movl m32, %eax' and `mov; m32, %ecx' */
  { 6, { 0x89, 0x05 }, {0xff, 0xf7 } },
  /* `movl m32, %edx' */
  { 6, { 0x89, 0x15 }, {0xff, 0xff } },

  /* Check for `xorl r32, r32' and the equivalent `subl r32, r32'.
     Because of the symmetry, there are actually two ways to encode
     these instructions; opcode bytes 0x29 and 0x2b for `subl' and
     opcode bytes 0x31 and 0x33 for `xorl'.  */

  /* `subl %eax, %eax' */
  { 2, { 0x29, 0xc0 }, { 0xfd, 0xff } },
  /* `subl %ecx, %ecx' */
  { 2, { 0x29, 0xc9 }, { 0xfd, 0xff } },
  /* `subl %edx, %edx' */
  { 2, { 0x29, 0xd2 }, { 0xfd, 0xff } },
  /* `xorl %eax, %eax' */
  { 2, { 0x31, 0xc0 }, { 0xfd, 0xff } },
  /* `xorl %ecx, %ecx' */
  { 2, { 0x31, 0xc9 }, { 0xfd, 0xff } },
  /* `xorl %edx, %edx' */
  { 2, { 0x31, 0xd2 }, { 0xfd, 0xff } },
  { 0 }
};


/* Check whether PC points to a no-op instruction.  */
static CORE_ADDR
i386_skip_noop (CORE_ADDR pc)
{
  gdb_byte op;
  int check = 1;

  target_read_memory (pc, &op, 1);

  while (check) 
    {
      check = 0;
      /* Ignore `nop' instruction.  */
      if (op == 0x90) 
	{
	  pc += 1;
	  target_read_memory (pc, &op, 1);
	  check = 1;
	}
      /* Ignore no-op instruction `mov %edi, %edi'.
	 Microsoft system dlls often start with
	 a `mov %edi,%edi' instruction.
	 The 5 bytes before the function start are
	 filled with `nop' instructions.
	 This pattern can be used for hot-patching:
	 The `mov %edi, %edi' instruction can be replaced by a
	 near jump to the location of the 5 `nop' instructions
	 which can be replaced by a 32-bit jump to anywhere
	 in the 32-bit address space.  */

      else if (op == 0x8b)
	{
	  target_read_memory (pc + 1, &op, 1);
	  if (op == 0xff)
	    {
	      pc += 2;
	      target_read_memory (pc, &op, 1);
	      check = 1;
	    }
	}
    }
  return pc; 
}

/* Check whether PC points at a code that sets up a new stack frame.
   If so, it updates CACHE and returns the address of the first
   instruction after the sequence that sets up the frame or LIMIT,
   whichever is smaller.  If we don't recognize the code, return PC.  */

static CORE_ADDR
i386_analyze_frame_setup (CORE_ADDR pc, CORE_ADDR limit,
			  struct i386_frame_cache *cache)
{
  struct i386_insn *insn;
  gdb_byte op;
  int skip = 0;

  if (limit <= pc)
    return limit;

  target_read_memory (pc, &op, 1);

  if (op == 0x55)		/* pushl %ebp */
    {
      /* Take into account that we've executed the `pushl %ebp' that
	 starts this instruction sequence.  */
      cache->saved_regs[I386_EBP_REGNUM] = 0;
      cache->sp_offset += 4;
      pc++;

      /* If that's all, return now.  */
      if (limit <= pc)
	return limit;

      /* Check for some special instructions that might be migrated by
	 GCC into the prologue and skip them.  At this point in the
	 prologue, code should only touch the scratch registers %eax,
	 %ecx and %edx, so while the number of posibilities is sheer,
	 it is limited.

	 Make sure we only skip these instructions if we later see the
	 `movl %esp, %ebp' that actually sets up the frame.  */
      while (pc + skip < limit)
	{
	  insn = i386_match_insn (pc + skip, i386_frame_setup_skip_insns);
	  if (insn == NULL)
	    break;

	  skip += insn->len;
	}

      /* If that's all, return now.  */
      if (limit <= pc + skip)
	return limit;

      target_read_memory (pc + skip, &op, 1);

      /* Check for `movl %esp, %ebp' -- can be written in two ways.  */
      switch (op)
	{
	case 0x8b:
	  if (read_memory_unsigned_integer (pc + skip + 1, 1) != 0xec)
	    return pc;
	  break;
	case 0x89:
	  if (read_memory_unsigned_integer (pc + skip + 1, 1) != 0xe5)
	    return pc;
	  break;
	default:
	  return pc;
	}

      /* OK, we actually have a frame.  We just don't know how large
	 it is yet.  Set its size to zero.  We'll adjust it if
	 necessary.  We also now commit to skipping the special
	 instructions mentioned before.  */
      cache->locals = 0;
      pc += (skip + 2);

      /* If that's all, return now.  */
      if (limit <= pc)
	return limit;

      /* Check for stack adjustment 

	    subl $XXX, %esp

	 NOTE: You can't subtract a 16-bit immediate from a 32-bit
	 reg, so we don't have to worry about a data16 prefix.  */
      target_read_memory (pc, &op, 1);
      if (op == 0x83)
	{
	  /* `subl' with 8-bit immediate.  */
	  if (read_memory_unsigned_integer (pc + 1, 1) != 0xec)
	    /* Some instruction starting with 0x83 other than `subl'.  */
	    return pc;

	  /* `subl' with signed 8-bit immediate (though it wouldn't
	     make sense to be negative).  */
	  cache->locals = read_memory_integer (pc + 2, 1);
	  return pc + 3;
	}
      else if (op == 0x81)
	{
	  /* Maybe it is `subl' with a 32-bit immediate.  */
	  if (read_memory_unsigned_integer (pc + 1, 1) != 0xec)
	    /* Some instruction starting with 0x81 other than `subl'.  */
	    return pc;

	  /* It is `subl' with a 32-bit immediate.  */
	  cache->locals = read_memory_integer (pc + 2, 4);
	  return pc + 6;
	}
      else
	{
	  /* Some instruction other than `subl'.  */
	  return pc;
	}
    }
  else if (op == 0xc8)		/* enter */
    {
      cache->locals = read_memory_unsigned_integer (pc + 1, 2);
      return pc + 4;
    }

  return pc;
}

/* Check whether PC points at code that saves registers on the stack.
   If so, it updates CACHE and returns the address of the first
   instruction after the register saves or CURRENT_PC, whichever is
   smaller.  Otherwise, return PC.  */

static CORE_ADDR
i386_analyze_register_saves (CORE_ADDR pc, CORE_ADDR current_pc,
			     struct i386_frame_cache *cache)
{
  CORE_ADDR offset = 0;
  gdb_byte op;
  int i;

  if (cache->locals > 0)
    offset -= cache->locals;
  for (i = 0; i < 8 && pc < current_pc; i++)
    {
      target_read_memory (pc, &op, 1);
      if (op < 0x50 || op > 0x57)
	break;

      offset -= 4;
      cache->saved_regs[op - 0x50] = offset;
      cache->sp_offset += 4;
      pc++;
    }

  return pc;
}

/* Do a full analysis of the prologue at PC and update CACHE
   accordingly.  Bail out early if CURRENT_PC is reached.  Return the
   address where the analysis stopped.

   We handle these cases:

   The startup sequence can be at the start of the function, or the
   function can start with a branch to startup code at the end.

   %ebp can be set up with either the 'enter' instruction, or "pushl
   %ebp, movl %esp, %ebp" (`enter' is too slow to be useful, but was
   once used in the System V compiler).

   Local space is allocated just below the saved %ebp by either the
   'enter' instruction, or by "subl $<size>, %esp".  'enter' has a
   16-bit unsigned argument for space to allocate, and the 'addl'
   instruction could have either a signed byte, or 32-bit immediate.

   Next, the registers used by this function are pushed.  With the
   System V compiler they will always be in the order: %edi, %esi,
   %ebx (and sometimes a harmless bug causes it to also save but not
   restore %eax); however, the code below is willing to see the pushes
   in any order, and will handle up to 8 of them.
 
   If the setup sequence is at the end of the function, then the next
   instruction will be a branch back to the start.  */

static CORE_ADDR
i386_analyze_prologue (CORE_ADDR pc, CORE_ADDR current_pc,
		       struct i386_frame_cache *cache)
{
  pc = i386_skip_noop (pc);
  pc = i386_follow_jump (pc);
  pc = i386_analyze_struct_return (pc, current_pc, cache);
  pc = i386_skip_probe (pc);
  pc = i386_analyze_stack_align (pc, current_pc, cache);
  pc = i386_analyze_frame_setup (pc, current_pc, cache);
  return i386_analyze_register_saves (pc, current_pc, cache);
}

/* Return PC of first real instruction.  */

static CORE_ADDR
i386_skip_prologue (struct gdbarch *gdbarch, CORE_ADDR start_pc)
{
  static gdb_byte pic_pat[6] =
  {
    0xe8, 0, 0, 0, 0,		/* call 0x0 */
    0x5b,			/* popl %ebx */
  };
  struct i386_frame_cache cache;
  CORE_ADDR pc;
  gdb_byte op;
  int i;

  cache.locals = -1;
  pc = i386_analyze_prologue (start_pc, 0xffffffff, &cache);
  if (cache.locals < 0)
    return start_pc;

  /* Found valid frame setup.  */

  /* The native cc on SVR4 in -K PIC mode inserts the following code
     to get the address of the global offset table (GOT) into register
     %ebx:

        call	0x0
	popl    %ebx
        movl    %ebx,x(%ebp)    (optional)
        addl    y,%ebx

     This code is with the rest of the prologue (at the end of the
     function), so we have to skip it to get to the first real
     instruction at the start of the function.  */

  for (i = 0; i < 6; i++)
    {
      target_read_memory (pc + i, &op, 1);
      if (pic_pat[i] != op)
	break;
    }
  if (i == 6)
    {
      int delta = 6;

      target_read_memory (pc + delta, &op, 1);

      if (op == 0x89)		/* movl %ebx, x(%ebp) */
	{
	  op = read_memory_unsigned_integer (pc + delta + 1, 1);

	  if (op == 0x5d)	/* One byte offset from %ebp.  */
	    delta += 3;
	  else if (op == 0x9d)	/* Four byte offset from %ebp.  */
	    delta += 6;
	  else			/* Unexpected instruction.  */
	    delta = 0;

          target_read_memory (pc + delta, &op, 1);
	}

      /* addl y,%ebx */
      if (delta > 0 && op == 0x81
	  && read_memory_unsigned_integer (pc + delta + 1, 1) == 0xc3)
	{
	  pc += delta + 6;
	}
    }

  /* If the function starts with a branch (to startup code at the end)
     the last instruction should bring us back to the first
     instruction of the real code.  */
  if (i386_follow_jump (start_pc) != start_pc)
    pc = i386_follow_jump (pc);

  return pc;
}

/* Check that the code pointed to by PC corresponds to a call to
   __main, skip it if so.  Return PC otherwise.  */

CORE_ADDR
i386_skip_main_prologue (struct gdbarch *gdbarch, CORE_ADDR pc)
{
  gdb_byte op;

  target_read_memory (pc, &op, 1);
  if (op == 0xe8)
    {
      gdb_byte buf[4];

      if (target_read_memory (pc + 1, buf, sizeof buf) == 0)
 	{
	  /* Make sure address is computed correctly as a 32bit
	     integer even if CORE_ADDR is 64 bit wide.  */
 	  struct minimal_symbol *s;
 	  CORE_ADDR call_dest = pc + 5 + extract_signed_integer (buf, 4);

	  call_dest = call_dest & 0xffffffffU;
 	  s = lookup_minimal_symbol_by_pc (call_dest);
 	  if (s != NULL
 	      && SYMBOL_LINKAGE_NAME (s) != NULL
 	      && strcmp (SYMBOL_LINKAGE_NAME (s), "__main") == 0)
 	    pc += 5;
 	}
    }

  return pc;
}

/* This function is 64-bit safe.  */

static CORE_ADDR
i386_unwind_pc (struct gdbarch *gdbarch, struct frame_info *next_frame)
{
  gdb_byte buf[8];

  frame_unwind_register (next_frame, gdbarch_pc_regnum (gdbarch), buf);
  return extract_typed_address (buf, builtin_type (gdbarch)->builtin_func_ptr);
}


/* Normal frames.  */

static struct i386_frame_cache *
i386_frame_cache (struct frame_info *this_frame, void **this_cache)
{
  struct i386_frame_cache *cache;
  gdb_byte buf[4];
  int i;

  if (*this_cache)
    return *this_cache;

  cache = i386_alloc_frame_cache ();
  *this_cache = cache;

  /* In principle, for normal frames, %ebp holds the frame pointer,
     which holds the base address for the current stack frame.
     However, for functions that don't need it, the frame pointer is
     optional.  For these "frameless" functions the frame pointer is
     actually the frame pointer of the calling frame.  Signal
     trampolines are just a special case of a "frameless" function.
     They (usually) share their frame pointer with the frame that was
     in progress when the signal occurred.  */

  get_frame_register (this_frame, I386_EBP_REGNUM, buf);
  cache->base = extract_unsigned_integer (buf, 4);
  if (cache->base == 0)
    return cache;

  /* For normal frames, %eip is stored at 4(%ebp).  */
  cache->saved_regs[I386_EIP_REGNUM] = 4;

  cache->pc = get_frame_func (this_frame);
  if (cache->pc != 0)
    i386_analyze_prologue (cache->pc, get_frame_pc (this_frame), cache);

  if (cache->saved_sp_reg != -1)
    {
      /* Saved stack pointer has been saved.  */
      get_frame_register (this_frame, cache->saved_sp_reg, buf);
      cache->saved_sp = extract_unsigned_integer(buf, 4);
    }

  if (cache->locals < 0)
    {
      /* We didn't find a valid frame, which means that CACHE->base
	 currently holds the frame pointer for our calling frame.  If
	 we're at the start of a function, or somewhere half-way its
	 prologue, the function's frame probably hasn't been fully
	 setup yet.  Try to reconstruct the base address for the stack
	 frame by looking at the stack pointer.  For truly "frameless"
	 functions this might work too.  */

      if (cache->saved_sp_reg != -1)
	{
	  /* We're halfway aligning the stack.  */
	  cache->base = ((cache->saved_sp - 4) & 0xfffffff0) - 4;
	  cache->saved_regs[I386_EIP_REGNUM] = cache->saved_sp - 4;

	  /* This will be added back below.  */
	  cache->saved_regs[I386_EIP_REGNUM] -= cache->base;
	}
      else
	{
	  get_frame_register (this_frame, I386_ESP_REGNUM, buf);
	  cache->base = extract_unsigned_integer (buf, 4) + cache->sp_offset;
	}
    }

  /* Now that we have the base address for the stack frame we can
     calculate the value of %esp in the calling frame.  */
  if (cache->saved_sp == 0)
    cache->saved_sp = cache->base + 8;

  /* Adjust all the saved registers such that they contain addresses
     instead of offsets.  */
  for (i = 0; i < I386_NUM_SAVED_REGS; i++)
    if (cache->saved_regs[i] != -1)
      cache->saved_regs[i] += cache->base;

  return cache;
}

static void
i386_frame_this_id (struct frame_info *this_frame, void **this_cache,
		    struct frame_id *this_id)
{
  struct i386_frame_cache *cache = i386_frame_cache (this_frame, this_cache);

  /* This marks the outermost frame.  */
  if (cache->base == 0)
    return;

  /* See the end of i386_push_dummy_call.  */
  (*this_id) = frame_id_build (cache->base + 8, cache->pc);
}

static struct value *
i386_frame_prev_register (struct frame_info *this_frame, void **this_cache,
			  int regnum)
{
  struct i386_frame_cache *cache = i386_frame_cache (this_frame, this_cache);

  gdb_assert (regnum >= 0);

  /* The System V ABI says that:

     "The flags register contains the system flags, such as the
     direction flag and the carry flag.  The direction flag must be
     set to the forward (that is, zero) direction before entry and
     upon exit from a function.  Other user flags have no specified
     role in the standard calling sequence and are not preserved."

     To guarantee the "upon exit" part of that statement we fake a
     saved flags register that has its direction flag cleared.

     Note that GCC doesn't seem to rely on the fact that the direction
     flag is cleared after a function return; it always explicitly
     clears the flag before operations where it matters.

     FIXME: kettenis/20030316: I'm not quite sure whether this is the
     right thing to do.  The way we fake the flags register here makes
     it impossible to change it.  */

  if (regnum == I386_EFLAGS_REGNUM)
    {
      ULONGEST val;

      val = get_frame_register_unsigned (this_frame, regnum);
      val &= ~(1 << 10);
      return frame_unwind_got_constant (this_frame, regnum, val);
    }

  if (regnum == I386_EIP_REGNUM && cache->pc_in_eax)
    return frame_unwind_got_register (this_frame, regnum, I386_EAX_REGNUM);

  if (regnum == I386_ESP_REGNUM && cache->saved_sp)
    return frame_unwind_got_constant (this_frame, regnum, cache->saved_sp);

  if (regnum < I386_NUM_SAVED_REGS && cache->saved_regs[regnum] != -1)
    return frame_unwind_got_memory (this_frame, regnum,
				    cache->saved_regs[regnum]);

  return frame_unwind_got_register (this_frame, regnum, regnum);
}

static const struct frame_unwind i386_frame_unwind =
{
  NORMAL_FRAME,
  i386_frame_this_id,
  i386_frame_prev_register,
  NULL,
  default_frame_sniffer
};


/* Signal trampolines.  */

static struct i386_frame_cache *
i386_sigtramp_frame_cache (struct frame_info *this_frame, void **this_cache)
{
  struct i386_frame_cache *cache;
  struct gdbarch_tdep *tdep = gdbarch_tdep (get_frame_arch (this_frame));
  CORE_ADDR addr;
  gdb_byte buf[4];

  if (*this_cache)
    return *this_cache;

  cache = i386_alloc_frame_cache ();

  get_frame_register (this_frame, I386_ESP_REGNUM, buf);
  cache->base = extract_unsigned_integer (buf, 4) - 4;

  addr = tdep->sigcontext_addr (this_frame);
  if (tdep->sc_reg_offset)
    {
      int i;

      gdb_assert (tdep->sc_num_regs <= I386_NUM_SAVED_REGS);

      for (i = 0; i < tdep->sc_num_regs; i++)
	if (tdep->sc_reg_offset[i] != -1)
	  cache->saved_regs[i] = addr + tdep->sc_reg_offset[i];
    }
  else
    {
      cache->saved_regs[I386_EIP_REGNUM] = addr + tdep->sc_pc_offset;
      cache->saved_regs[I386_ESP_REGNUM] = addr + tdep->sc_sp_offset;
    }

  *this_cache = cache;
  return cache;
}

static void
i386_sigtramp_frame_this_id (struct frame_info *this_frame, void **this_cache,
			     struct frame_id *this_id)
{
  struct i386_frame_cache *cache =
    i386_sigtramp_frame_cache (this_frame, this_cache);

  /* See the end of i386_push_dummy_call.  */
  (*this_id) = frame_id_build (cache->base + 8, get_frame_pc (this_frame));
}

static struct value *
i386_sigtramp_frame_prev_register (struct frame_info *this_frame,
				   void **this_cache, int regnum)
{
  /* Make sure we've initialized the cache.  */
  i386_sigtramp_frame_cache (this_frame, this_cache);

  return i386_frame_prev_register (this_frame, this_cache, regnum);
}

static int
i386_sigtramp_frame_sniffer (const struct frame_unwind *self,
			     struct frame_info *this_frame,
			     void **this_prologue_cache)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (get_frame_arch (this_frame));

  /* We shouldn't even bother if we don't have a sigcontext_addr
     handler.  */
  if (tdep->sigcontext_addr == NULL)
    return 0;

  if (tdep->sigtramp_p != NULL)
    {
      if (tdep->sigtramp_p (this_frame))
	return 1;
    }

  if (tdep->sigtramp_start != 0)
    {
      CORE_ADDR pc = get_frame_pc (this_frame);

      gdb_assert (tdep->sigtramp_end != 0);
      if (pc >= tdep->sigtramp_start && pc < tdep->sigtramp_end)
	return 1;
    }

  return 0;
}

static const struct frame_unwind i386_sigtramp_frame_unwind =
{
  SIGTRAMP_FRAME,
  i386_sigtramp_frame_this_id,
  i386_sigtramp_frame_prev_register,
  NULL,
  i386_sigtramp_frame_sniffer
};


static CORE_ADDR
i386_frame_base_address (struct frame_info *this_frame, void **this_cache)
{
  struct i386_frame_cache *cache = i386_frame_cache (this_frame, this_cache);

  return cache->base;
}

static const struct frame_base i386_frame_base =
{
  &i386_frame_unwind,
  i386_frame_base_address,
  i386_frame_base_address,
  i386_frame_base_address
};

static struct frame_id
i386_dummy_id (struct gdbarch *gdbarch, struct frame_info *this_frame)
{
  CORE_ADDR fp;

  fp = get_frame_register_unsigned (this_frame, I386_EBP_REGNUM);

  /* See the end of i386_push_dummy_call.  */
  return frame_id_build (fp + 8, get_frame_pc (this_frame));
}


/* Figure out where the longjmp will land.  Slurp the args out of the
   stack.  We expect the first arg to be a pointer to the jmp_buf
   structure from which we extract the address that we will land at.
   This address is copied into PC.  This routine returns non-zero on
   success.  */

static int
i386_get_longjmp_target (struct frame_info *frame, CORE_ADDR *pc)
{
  gdb_byte buf[4];
  CORE_ADDR sp, jb_addr;
  struct gdbarch *gdbarch = get_frame_arch (frame);
  int jb_pc_offset = gdbarch_tdep (gdbarch)->jb_pc_offset;

  /* If JB_PC_OFFSET is -1, we have no way to find out where the
     longjmp will land.  */
  if (jb_pc_offset == -1)
    return 0;

  get_frame_register (frame, I386_ESP_REGNUM, buf);
  sp = extract_unsigned_integer (buf, 4);
  if (target_read_memory (sp + 4, buf, 4))
    return 0;

  jb_addr = extract_unsigned_integer (buf, 4);
  if (target_read_memory (jb_addr + jb_pc_offset, buf, 4))
    return 0;

  *pc = extract_unsigned_integer (buf, 4);
  return 1;
}


/* Check whether TYPE must be 16-byte-aligned when passed as a
   function argument.  16-byte vectors, _Decimal128 and structures or
   unions containing such types must be 16-byte-aligned; other
   arguments are 4-byte-aligned.  */

static int
i386_16_byte_align_p (struct type *type)
{
  type = check_typedef (type);
  if ((TYPE_CODE (type) == TYPE_CODE_DECFLOAT
       || (TYPE_CODE (type) == TYPE_CODE_ARRAY && TYPE_VECTOR (type)))
      && TYPE_LENGTH (type) == 16)
    return 1;
  if (TYPE_CODE (type) == TYPE_CODE_ARRAY)
    return i386_16_byte_align_p (TYPE_TARGET_TYPE (type));
  if (TYPE_CODE (type) == TYPE_CODE_STRUCT
      || TYPE_CODE (type) == TYPE_CODE_UNION)
    {
      int i;
      for (i = 0; i < TYPE_NFIELDS (type); i++)
	{
	  if (i386_16_byte_align_p (TYPE_FIELD_TYPE (type, i)))
	    return 1;
	}
    }
  return 0;
}

static CORE_ADDR
i386_push_dummy_call (struct gdbarch *gdbarch, struct value *function,
		      struct regcache *regcache, CORE_ADDR bp_addr, int nargs,
		      struct value **args, CORE_ADDR sp, int struct_return,
		      CORE_ADDR struct_addr)
{
  gdb_byte buf[4];
  int i;
  int write_pass;
  int args_space = 0;

  /* Determine the total space required for arguments and struct
     return address in a first pass (allowing for 16-byte-aligned
     arguments), then push arguments in a second pass.  */

  for (write_pass = 0; write_pass < 2; write_pass++)
    {
      int args_space_used = 0;
      int have_16_byte_aligned_arg = 0;

      if (struct_return)
	{
	  if (write_pass)
	    {
	      /* Push value address.  */
	      store_unsigned_integer (buf, 4, struct_addr);
	      write_memory (sp, buf, 4);
	      args_space_used += 4;
	    }
	  else
	    args_space += 4;
	}

      for (i = 0; i < nargs; i++)
	{
	  int len = TYPE_LENGTH (value_enclosing_type (args[i]));

	  if (write_pass)
	    {
	      if (i386_16_byte_align_p (value_enclosing_type (args[i])))
		args_space_used = align_up (args_space_used, 16);

	      write_memory (sp + args_space_used,
			    value_contents_all (args[i]), len);
	      /* The System V ABI says that:

	      "An argument's size is increased, if necessary, to make it a
	      multiple of [32-bit] words.  This may require tail padding,
	      depending on the size of the argument."

	      This makes sure the stack stays word-aligned.  */
	      args_space_used += align_up (len, 4);
	    }
	  else
	    {
	      if (i386_16_byte_align_p (value_enclosing_type (args[i])))
		{
		  args_space = align_up (args_space, 16);
		  have_16_byte_aligned_arg = 1;
		}
	      args_space += align_up (len, 4);
	    }
	}

      if (!write_pass)
	{
	  if (have_16_byte_aligned_arg)
	    args_space = align_up (args_space, 16);
	  sp -= args_space;
	}
    }

  /* Store return address.  */
  sp -= 4;
  store_unsigned_integer (buf, 4, bp_addr);
  write_memory (sp, buf, 4);

  /* Finally, update the stack pointer...  */
  store_unsigned_integer (buf, 4, sp);
  regcache_cooked_write (regcache, I386_ESP_REGNUM, buf);

  /* ...and fake a frame pointer.  */
  regcache_cooked_write (regcache, I386_EBP_REGNUM, buf);

  /* MarkK wrote: This "+ 8" is all over the place:
     (i386_frame_this_id, i386_sigtramp_frame_this_id,
     i386_dummy_id).  It's there, since all frame unwinders for
     a given target have to agree (within a certain margin) on the
     definition of the stack address of a frame.  Otherwise frame id
     comparison might not work correctly.  Since DWARF2/GCC uses the
     stack address *before* the function call as a frame's CFA.  On
     the i386, when %ebp is used as a frame pointer, the offset
     between the contents %ebp and the CFA as defined by GCC.  */
  return sp + 8;
}

/* These registers are used for returning integers (and on some
   targets also for returning `struct' and `union' values when their
   size and alignment match an integer type).  */
#define LOW_RETURN_REGNUM	I386_EAX_REGNUM /* %eax */
#define HIGH_RETURN_REGNUM	I386_EDX_REGNUM /* %edx */

/* Read, for architecture GDBARCH, a function return value of TYPE
   from REGCACHE, and copy that into VALBUF.  */

static void
i386_extract_return_value (struct gdbarch *gdbarch, struct type *type,
			   struct regcache *regcache, gdb_byte *valbuf)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
  int len = TYPE_LENGTH (type);
  gdb_byte buf[I386_MAX_REGISTER_SIZE];

  if (TYPE_CODE (type) == TYPE_CODE_FLT)
    {
      if (tdep->st0_regnum < 0)
	{
	  warning (_("Cannot find floating-point return value."));
	  memset (valbuf, 0, len);
	  return;
	}

      /* Floating-point return values can be found in %st(0).  Convert
	 its contents to the desired type.  This is probably not
	 exactly how it would happen on the target itself, but it is
	 the best we can do.  */
      regcache_raw_read (regcache, I386_ST0_REGNUM, buf);
      convert_typed_floating (buf, builtin_type_i387_ext, valbuf, type);
    }
  else
    {
      int low_size = register_size (gdbarch, LOW_RETURN_REGNUM);
      int high_size = register_size (gdbarch, HIGH_RETURN_REGNUM);

      if (len <= low_size)
	{
	  regcache_raw_read (regcache, LOW_RETURN_REGNUM, buf);
	  memcpy (valbuf, buf, len);
	}
      else if (len <= (low_size + high_size))
	{
	  regcache_raw_read (regcache, LOW_RETURN_REGNUM, buf);
	  memcpy (valbuf, buf, low_size);
	  regcache_raw_read (regcache, HIGH_RETURN_REGNUM, buf);
	  memcpy (valbuf + low_size, buf, len - low_size);
	}
      else
	internal_error (__FILE__, __LINE__,
			_("Cannot extract return value of %d bytes long."), len);
    }
}

/* Write, for architecture GDBARCH, a function return value of TYPE
   from VALBUF into REGCACHE.  */

static void
i386_store_return_value (struct gdbarch *gdbarch, struct type *type,
			 struct regcache *regcache, const gdb_byte *valbuf)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
  int len = TYPE_LENGTH (type);

  if (TYPE_CODE (type) == TYPE_CODE_FLT)
    {
      ULONGEST fstat;
      gdb_byte buf[I386_MAX_REGISTER_SIZE];

      if (tdep->st0_regnum < 0)
	{
	  warning (_("Cannot set floating-point return value."));
	  return;
	}

      /* Returning floating-point values is a bit tricky.  Apart from
         storing the return value in %st(0), we have to simulate the
         state of the FPU at function return point.  */

      /* Convert the value found in VALBUF to the extended
	 floating-point format used by the FPU.  This is probably
	 not exactly how it would happen on the target itself, but
	 it is the best we can do.  */
      convert_typed_floating (valbuf, type, buf, builtin_type_i387_ext);
      regcache_raw_write (regcache, I386_ST0_REGNUM, buf);

      /* Set the top of the floating-point register stack to 7.  The
         actual value doesn't really matter, but 7 is what a normal
         function return would end up with if the program started out
         with a freshly initialized FPU.  */
      regcache_raw_read_unsigned (regcache, I387_FSTAT_REGNUM (tdep), &fstat);
      fstat |= (7 << 11);
      regcache_raw_write_unsigned (regcache, I387_FSTAT_REGNUM (tdep), fstat);

      /* Mark %st(1) through %st(7) as empty.  Since we set the top of
         the floating-point register stack to 7, the appropriate value
         for the tag word is 0x3fff.  */
      regcache_raw_write_unsigned (regcache, I387_FTAG_REGNUM (tdep), 0x3fff);
    }
  else
    {
      int low_size = register_size (gdbarch, LOW_RETURN_REGNUM);
      int high_size = register_size (gdbarch, HIGH_RETURN_REGNUM);

      if (len <= low_size)
	regcache_raw_write_part (regcache, LOW_RETURN_REGNUM, 0, len, valbuf);
      else if (len <= (low_size + high_size))
	{
	  regcache_raw_write (regcache, LOW_RETURN_REGNUM, valbuf);
	  regcache_raw_write_part (regcache, HIGH_RETURN_REGNUM, 0,
				   len - low_size, valbuf + low_size);
	}
      else
	internal_error (__FILE__, __LINE__,
			_("Cannot store return value of %d bytes long."), len);
    }
}


/* This is the variable that is set with "set struct-convention", and
   its legitimate values.  */
static const char default_struct_convention[] = "default";
static const char pcc_struct_convention[] = "pcc";
static const char reg_struct_convention[] = "reg";
static const char *valid_conventions[] =
{
  default_struct_convention,
  pcc_struct_convention,
  reg_struct_convention,
  NULL
};
static const char *struct_convention = default_struct_convention;

/* Return non-zero if TYPE, which is assumed to be a structure,
   a union type, or an array type, should be returned in registers
   for architecture GDBARCH.  */

static int
i386_reg_struct_return_p (struct gdbarch *gdbarch, struct type *type)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);
  enum type_code code = TYPE_CODE (type);
  int len = TYPE_LENGTH (type);

  gdb_assert (code == TYPE_CODE_STRUCT
              || code == TYPE_CODE_UNION
              || code == TYPE_CODE_ARRAY);

  if (struct_convention == pcc_struct_convention
      || (struct_convention == default_struct_convention
	  && tdep->struct_return == pcc_struct_return))
    return 0;

  /* Structures consisting of a single `float', `double' or 'long
     double' member are returned in %st(0).  */
  if (code == TYPE_CODE_STRUCT && TYPE_NFIELDS (type) == 1)
    {
      type = check_typedef (TYPE_FIELD_TYPE (type, 0));
      if (TYPE_CODE (type) == TYPE_CODE_FLT)
	return (len == 4 || len == 8 || len == 12);
    }

  return (len == 1 || len == 2 || len == 4 || len == 8);
}

/* Determine, for architecture GDBARCH, how a return value of TYPE
   should be returned.  If it is supposed to be returned in registers,
   and READBUF is non-zero, read the appropriate value from REGCACHE,
   and copy it into READBUF.  If WRITEBUF is non-zero, write the value
   from WRITEBUF into REGCACHE.  */

static enum return_value_convention
i386_return_value (struct gdbarch *gdbarch, struct type *func_type,
		   struct type *type, struct regcache *regcache,
		   gdb_byte *readbuf, const gdb_byte *writebuf)
{
  enum type_code code = TYPE_CODE (type);

  if (((code == TYPE_CODE_STRUCT
	|| code == TYPE_CODE_UNION
	|| code == TYPE_CODE_ARRAY)
       && !i386_reg_struct_return_p (gdbarch, type))
      /* 128-bit decimal float uses the struct return convention.  */
      || (code == TYPE_CODE_DECFLOAT && TYPE_LENGTH (type) == 16))
    {
      /* The System V ABI says that:

	 "A function that returns a structure or union also sets %eax
	 to the value of the original address of the caller's area
	 before it returns.  Thus when the caller receives control
	 again, the address of the returned object resides in register
	 %eax and can be used to access the object."

	 So the ABI guarantees that we can always find the return
	 value just after the function has returned.  */

      /* Note that the ABI doesn't mention functions returning arrays,
         which is something possible in certain languages such as Ada.
         In this case, the value is returned as if it was wrapped in
         a record, so the convention applied to records also applies
         to arrays.  */

      if (readbuf)
	{
	  ULONGEST addr;

	  regcache_raw_read_unsigned (regcache, I386_EAX_REGNUM, &addr);
	  read_memory (addr, readbuf, TYPE_LENGTH (type));
	}

      return RETURN_VALUE_ABI_RETURNS_ADDRESS;
    }

  /* This special case is for structures consisting of a single
     `float', `double' or 'long double' member.  These structures are
     returned in %st(0).  For these structures, we call ourselves
     recursively, changing TYPE into the type of the first member of
     the structure.  Since that should work for all structures that
     have only one member, we don't bother to check the member's type
     here.  */
  if (code == TYPE_CODE_STRUCT && TYPE_NFIELDS (type) == 1)
    {
      type = check_typedef (TYPE_FIELD_TYPE (type, 0));
      return i386_return_value (gdbarch, func_type, type, regcache,
				readbuf, writebuf);
    }

  if (readbuf)
    i386_extract_return_value (gdbarch, type, regcache, readbuf);
  if (writebuf)
    i386_store_return_value (gdbarch, type, regcache, writebuf);

  return RETURN_VALUE_REGISTER_CONVENTION;
}


/* Type for %eflags.  */
struct type *i386_eflags_type;

/* Type for %mxcsr.  */
struct type *i386_mxcsr_type;

/* Construct types for ISA-specific registers.  */
static void
i386_init_types (void)
{
  struct type *type;

  type = init_flags_type ("builtin_type_i386_eflags", 4);
  append_flags_type_flag (type, 0, "CF");
  append_flags_type_flag (type, 1, NULL);
  append_flags_type_flag (type, 2, "PF");
  append_flags_type_flag (type, 4, "AF");
  append_flags_type_flag (type, 6, "ZF");
  append_flags_type_flag (type, 7, "SF");
  append_flags_type_flag (type, 8, "TF");
  append_flags_type_flag (type, 9, "IF");
  append_flags_type_flag (type, 10, "DF");
  append_flags_type_flag (type, 11, "OF");
  append_flags_type_flag (type, 14, "NT");
  append_flags_type_flag (type, 16, "RF");
  append_flags_type_flag (type, 17, "VM");
  append_flags_type_flag (type, 18, "AC");
  append_flags_type_flag (type, 19, "VIF");
  append_flags_type_flag (type, 20, "VIP");
  append_flags_type_flag (type, 21, "ID");
  i386_eflags_type = type;

  type = init_flags_type ("builtin_type_i386_mxcsr", 4);
  append_flags_type_flag (type, 0, "IE");
  append_flags_type_flag (type, 1, "DE");
  append_flags_type_flag (type, 2, "ZE");
  append_flags_type_flag (type, 3, "OE");
  append_flags_type_flag (type, 4, "UE");
  append_flags_type_flag (type, 5, "PE");
  append_flags_type_flag (type, 6, "DAZ");
  append_flags_type_flag (type, 7, "IM");
  append_flags_type_flag (type, 8, "DM");
  append_flags_type_flag (type, 9, "ZM");
  append_flags_type_flag (type, 10, "OM");
  append_flags_type_flag (type, 11, "UM");
  append_flags_type_flag (type, 12, "PM");
  append_flags_type_flag (type, 15, "FZ");
  i386_mxcsr_type = type;
}

/* Construct vector type for MMX registers.  */
struct type *
i386_mmx_type (struct gdbarch *gdbarch)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (!tdep->i386_mmx_type)
    {
      /* The type we're building is this: */
#if 0
      union __gdb_builtin_type_vec64i
      {
        int64_t uint64;
        int32_t v2_int32[2];
        int16_t v4_int16[4];
        int8_t v8_int8[8];
      };
#endif

      struct type *t;

      t = init_composite_type ("__gdb_builtin_type_vec64i", TYPE_CODE_UNION);
      append_composite_type_field (t, "uint64", builtin_type_int64);
      append_composite_type_field (t, "v2_int32",
				   init_vector_type (builtin_type_int32, 2));
      append_composite_type_field (t, "v4_int16",
				   init_vector_type (builtin_type_int16, 4));
      append_composite_type_field (t, "v8_int8",
				   init_vector_type (builtin_type_int8, 8));

      TYPE_VECTOR (t) = 1;
      TYPE_NAME (t) = "builtin_type_vec64i";
      tdep->i386_mmx_type = t;
    }

  return tdep->i386_mmx_type;
}

struct type *
i386_sse_type (struct gdbarch *gdbarch)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (!tdep->i386_sse_type)
    {
      /* The type we're building is this: */
#if 0
      union __gdb_builtin_type_vec128i
      {
        int128_t uint128;
        int64_t v2_int64[2];
        int32_t v4_int32[4];
        int16_t v8_int16[8];
        int8_t v16_int8[16];
        double v2_double[2];
        float v4_float[4];
      };
#endif

      struct type *t;

      t = init_composite_type ("__gdb_builtin_type_vec128i", TYPE_CODE_UNION);
      append_composite_type_field (t, "v4_float",
				   init_vector_type (builtin_type (gdbarch)
						     ->builtin_float, 4));
      append_composite_type_field (t, "v2_double",
				   init_vector_type (builtin_type (gdbarch)
						     ->builtin_double, 2));
      append_composite_type_field (t, "v16_int8",
				   init_vector_type (builtin_type_int8, 16));
      append_composite_type_field (t, "v8_int16",
				   init_vector_type (builtin_type_int16, 8));
      append_composite_type_field (t, "v4_int32",
				   init_vector_type (builtin_type_int32, 4));
      append_composite_type_field (t, "v2_int64",
				   init_vector_type (builtin_type_int64, 2));
      append_composite_type_field (t, "uint128", builtin_type_int128);

      TYPE_VECTOR (t) = 1;
      TYPE_NAME (t) = "builtin_type_vec128i";
      tdep->i386_sse_type = t;
    }

  return tdep->i386_sse_type;
}

/* Return the GDB type object for the "standard" data type of data in
   register REGNUM.  Perhaps %esi and %edi should go here, but
   potentially they could be used for things other than address.  */

static struct type *
i386_register_type (struct gdbarch *gdbarch, int regnum)
{
  if (regnum == I386_EIP_REGNUM)
    return builtin_type (gdbarch)->builtin_func_ptr;

  if (regnum == I386_EFLAGS_REGNUM)
    return i386_eflags_type;

  if (regnum == I386_EBP_REGNUM || regnum == I386_ESP_REGNUM)
    return builtin_type (gdbarch)->builtin_data_ptr;

  if (i386_fp_regnum_p (gdbarch, regnum))
    return builtin_type_i387_ext;

  if (i386_mmx_regnum_p (gdbarch, regnum))
    return i386_mmx_type (gdbarch);

  if (i386_sse_regnum_p (gdbarch, regnum))
    return i386_sse_type (gdbarch);

  if (regnum == I387_MXCSR_REGNUM (gdbarch_tdep (gdbarch)))
    return i386_mxcsr_type;

  return builtin_type (gdbarch)->builtin_int;
}

/* Map a cooked register onto a raw register or memory.  For the i386,
   the MMX registers need to be mapped onto floating point registers.  */

static int
i386_mmx_regnum_to_fp_regnum (struct regcache *regcache, int regnum)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (get_regcache_arch (regcache));
  int mmxreg, fpreg;
  ULONGEST fstat;
  int tos;

  mmxreg = regnum - tdep->mm0_regnum;
  regcache_raw_read_unsigned (regcache, I387_FSTAT_REGNUM (tdep), &fstat);
  tos = (fstat >> 11) & 0x7;
  fpreg = (mmxreg + tos) % 8;

  return (I387_ST0_REGNUM (tdep) + fpreg);
}

static void
i386_pseudo_register_read (struct gdbarch *gdbarch, struct regcache *regcache,
			   int regnum, gdb_byte *buf)
{
  if (i386_mmx_regnum_p (gdbarch, regnum))
    {
      gdb_byte mmx_buf[MAX_REGISTER_SIZE];
      int fpnum = i386_mmx_regnum_to_fp_regnum (regcache, regnum);

      /* Extract (always little endian).  */
      regcache_raw_read (regcache, fpnum, mmx_buf);
      memcpy (buf, mmx_buf, register_size (gdbarch, regnum));
    }
  else
    regcache_raw_read (regcache, regnum, buf);
}

static void
i386_pseudo_register_write (struct gdbarch *gdbarch, struct regcache *regcache,
			    int regnum, const gdb_byte *buf)
{
  if (i386_mmx_regnum_p (gdbarch, regnum))
    {
      gdb_byte mmx_buf[MAX_REGISTER_SIZE];
      int fpnum = i386_mmx_regnum_to_fp_regnum (regcache, regnum);

      /* Read ...  */
      regcache_raw_read (regcache, fpnum, mmx_buf);
      /* ... Modify ... (always little endian).  */
      memcpy (mmx_buf, buf, register_size (gdbarch, regnum));
      /* ... Write.  */
      regcache_raw_write (regcache, fpnum, mmx_buf);
    }
  else
    regcache_raw_write (regcache, regnum, buf);
}


/* Return the register number of the register allocated by GCC after
   REGNUM, or -1 if there is no such register.  */

static int
i386_next_regnum (int regnum)
{
  /* GCC allocates the registers in the order:

     %eax, %edx, %ecx, %ebx, %esi, %edi, %ebp, %esp, ...

     Since storing a variable in %esp doesn't make any sense we return
     -1 for %ebp and for %esp itself.  */
  static int next_regnum[] =
  {
    I386_EDX_REGNUM,		/* Slot for %eax.  */
    I386_EBX_REGNUM,		/* Slot for %ecx.  */
    I386_ECX_REGNUM,		/* Slot for %edx.  */
    I386_ESI_REGNUM,		/* Slot for %ebx.  */
    -1, -1,			/* Slots for %esp and %ebp.  */
    I386_EDI_REGNUM,		/* Slot for %esi.  */
    I386_EBP_REGNUM		/* Slot for %edi.  */
  };

  if (regnum >= 0 && regnum < sizeof (next_regnum) / sizeof (next_regnum[0]))
    return next_regnum[regnum];

  return -1;
}

/* Return nonzero if a value of type TYPE stored in register REGNUM
   needs any special handling.  */

static int
i386_convert_register_p (struct gdbarch *gdbarch, int regnum, struct type *type)
{
  int len = TYPE_LENGTH (type);

  /* Values may be spread across multiple registers.  Most debugging
     formats aren't expressive enough to specify the locations, so
     some heuristics is involved.  Right now we only handle types that
     have a length that is a multiple of the word size, since GCC
     doesn't seem to put any other types into registers.  */
  if (len > 4 && len % 4 == 0)
    {
      int last_regnum = regnum;

      while (len > 4)
	{
	  last_regnum = i386_next_regnum (last_regnum);
	  len -= 4;
	}

      if (last_regnum != -1)
	return 1;
    }

  return i387_convert_register_p (gdbarch, regnum, type);
}

/* Read a value of type TYPE from register REGNUM in frame FRAME, and
   return its contents in TO.  */

static void
i386_register_to_value (struct frame_info *frame, int regnum,
			struct type *type, gdb_byte *to)
{
  struct gdbarch *gdbarch = get_frame_arch (frame);
  int len = TYPE_LENGTH (type);

  /* FIXME: kettenis/20030609: What should we do if REGNUM isn't
     available in FRAME (i.e. if it wasn't saved)?  */

  if (i386_fp_regnum_p (gdbarch, regnum))
    {
      i387_register_to_value (frame, regnum, type, to);
      return;
    }

  /* Read a value spread across multiple registers.  */

  gdb_assert (len > 4 && len % 4 == 0);

  while (len > 0)
    {
      gdb_assert (regnum != -1);
      gdb_assert (register_size (gdbarch, regnum) == 4);

      get_frame_register (frame, regnum, to);
      regnum = i386_next_regnum (regnum);
      len -= 4;
      to += 4;
    }
}

/* Write the contents FROM of a value of type TYPE into register
   REGNUM in frame FRAME.  */

static void
i386_value_to_register (struct frame_info *frame, int regnum,
			struct type *type, const gdb_byte *from)
{
  int len = TYPE_LENGTH (type);

  if (i386_fp_regnum_p (get_frame_arch (frame), regnum))
    {
      i387_value_to_register (frame, regnum, type, from);
      return;
    }

  /* Write a value spread across multiple registers.  */

  gdb_assert (len > 4 && len % 4 == 0);

  while (len > 0)
    {
      gdb_assert (regnum != -1);
      gdb_assert (register_size (get_frame_arch (frame), regnum) == 4);

      put_frame_register (frame, regnum, from);
      regnum = i386_next_regnum (regnum);
      len -= 4;
      from += 4;
    }
}

/* Supply register REGNUM from the buffer specified by GREGS and LEN
   in the general-purpose register set REGSET to register cache
   REGCACHE.  If REGNUM is -1, do this for all registers in REGSET.  */

void
i386_supply_gregset (const struct regset *regset, struct regcache *regcache,
		     int regnum, const void *gregs, size_t len)
{
  const struct gdbarch_tdep *tdep = gdbarch_tdep (regset->arch);
  const gdb_byte *regs = gregs;
  int i;

  gdb_assert (len == tdep->sizeof_gregset);

  for (i = 0; i < tdep->gregset_num_regs; i++)
    {
      if ((regnum == i || regnum == -1)
	  && tdep->gregset_reg_offset[i] != -1)
	regcache_raw_supply (regcache, i, regs + tdep->gregset_reg_offset[i]);
    }
}

/* Collect register REGNUM from the register cache REGCACHE and store
   it in the buffer specified by GREGS and LEN as described by the
   general-purpose register set REGSET.  If REGNUM is -1, do this for
   all registers in REGSET.  */

void
i386_collect_gregset (const struct regset *regset,
		      const struct regcache *regcache,
		      int regnum, void *gregs, size_t len)
{
  const struct gdbarch_tdep *tdep = gdbarch_tdep (regset->arch);
  gdb_byte *regs = gregs;
  int i;

  gdb_assert (len == tdep->sizeof_gregset);

  for (i = 0; i < tdep->gregset_num_regs; i++)
    {
      if ((regnum == i || regnum == -1)
	  && tdep->gregset_reg_offset[i] != -1)
	regcache_raw_collect (regcache, i, regs + tdep->gregset_reg_offset[i]);
    }
}

/* Supply register REGNUM from the buffer specified by FPREGS and LEN
   in the floating-point register set REGSET to register cache
   REGCACHE.  If REGNUM is -1, do this for all registers in REGSET.  */

static void
i386_supply_fpregset (const struct regset *regset, struct regcache *regcache,
		      int regnum, const void *fpregs, size_t len)
{
  const struct gdbarch_tdep *tdep = gdbarch_tdep (regset->arch);

  if (len == I387_SIZEOF_FXSAVE)
    {
      i387_supply_fxsave (regcache, regnum, fpregs);
      return;
    }

  gdb_assert (len == tdep->sizeof_fpregset);
  i387_supply_fsave (regcache, regnum, fpregs);
}

/* Collect register REGNUM from the register cache REGCACHE and store
   it in the buffer specified by FPREGS and LEN as described by the
   floating-point register set REGSET.  If REGNUM is -1, do this for
   all registers in REGSET.  */

static void
i386_collect_fpregset (const struct regset *regset,
		       const struct regcache *regcache,
		       int regnum, void *fpregs, size_t len)
{
  const struct gdbarch_tdep *tdep = gdbarch_tdep (regset->arch);

  if (len == I387_SIZEOF_FXSAVE)
    {
      i387_collect_fxsave (regcache, regnum, fpregs);
      return;
    }

  gdb_assert (len == tdep->sizeof_fpregset);
  i387_collect_fsave (regcache, regnum, fpregs);
}

/* Return the appropriate register set for the core section identified
   by SECT_NAME and SECT_SIZE.  */

const struct regset *
i386_regset_from_core_section (struct gdbarch *gdbarch,
			       const char *sect_name, size_t sect_size)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  if (strcmp (sect_name, ".reg") == 0 && sect_size == tdep->sizeof_gregset)
    {
      if (tdep->gregset == NULL)
	tdep->gregset = regset_alloc (gdbarch, i386_supply_gregset,
				      i386_collect_gregset);
      return tdep->gregset;
    }

  if ((strcmp (sect_name, ".reg2") == 0 && sect_size == tdep->sizeof_fpregset)
      || (strcmp (sect_name, ".reg-xfp") == 0
	  && sect_size == I387_SIZEOF_FXSAVE))
    {
      if (tdep->fpregset == NULL)
	tdep->fpregset = regset_alloc (gdbarch, i386_supply_fpregset,
				       i386_collect_fpregset);
      return tdep->fpregset;
    }

  return NULL;
}


/* Stuff for WIN32 PE style DLL's but is pretty generic really.  */

CORE_ADDR
i386_pe_skip_trampoline_code (CORE_ADDR pc, char *name)
{
  if (pc && read_memory_unsigned_integer (pc, 2) == 0x25ff) /* jmp *(dest) */
    {
      unsigned long indirect = read_memory_unsigned_integer (pc + 2, 4);
      struct minimal_symbol *indsym =
	indirect ? lookup_minimal_symbol_by_pc (indirect) : 0;
      char *symname = indsym ? SYMBOL_LINKAGE_NAME (indsym) : 0;

      if (symname)
	{
	  if (strncmp (symname, "__imp_", 6) == 0
	      || strncmp (symname, "_imp_", 5) == 0)
	    return name ? 1 : read_memory_unsigned_integer (indirect, 4);
	}
    }
  return 0;			/* Not a trampoline.  */
}


/* Return whether the THIS_FRAME corresponds to a sigtramp
   routine.  */

static int
i386_sigtramp_p (struct frame_info *this_frame)
{
  CORE_ADDR pc = get_frame_pc (this_frame);
  char *name;

  find_pc_partial_function (pc, &name, NULL, NULL);
  return (name && strcmp ("_sigtramp", name) == 0);
}


/* We have two flavours of disassembly.  The machinery on this page
   deals with switching between those.  */

static int
i386_print_insn (bfd_vma pc, struct disassemble_info *info)
{
  gdb_assert (disassembly_flavor == att_flavor
	      || disassembly_flavor == intel_flavor);

  /* FIXME: kettenis/20020915: Until disassembler_options is properly
     constified, cast to prevent a compiler warning.  */
  info->disassembler_options = (char *) disassembly_flavor;

  return print_insn_i386 (pc, info);
}


/* There are a few i386 architecture variants that differ only
   slightly from the generic i386 target.  For now, we don't give them
   their own source file, but include them here.  As a consequence,
   they'll always be included.  */

/* System V Release 4 (SVR4).  */

/* Return whether THIS_FRAME corresponds to a SVR4 sigtramp
   routine.  */

static int
i386_svr4_sigtramp_p (struct frame_info *this_frame)
{
  CORE_ADDR pc = get_frame_pc (this_frame);
  char *name;

  /* UnixWare uses _sigacthandler.  The origin of the other symbols is
     currently unknown.  */
  find_pc_partial_function (pc, &name, NULL, NULL);
  return (name && (strcmp ("_sigreturn", name) == 0
		   || strcmp ("_sigacthandler", name) == 0
		   || strcmp ("sigvechandler", name) == 0));
}

/* Assuming THIS_FRAME is for a SVR4 sigtramp routine, return the
   address of the associated sigcontext (ucontext) structure.  */

static CORE_ADDR
i386_svr4_sigcontext_addr (struct frame_info *this_frame)
{
  gdb_byte buf[4];
  CORE_ADDR sp;

  get_frame_register (this_frame, I386_ESP_REGNUM, buf);
  sp = extract_unsigned_integer (buf, 4);

  return read_memory_unsigned_integer (sp + 8, 4);
}


/* Generic ELF.  */

void
i386_elf_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
{
  /* We typically use stabs-in-ELF with the SVR4 register numbering.  */
  set_gdbarch_stab_reg_to_regnum (gdbarch, i386_svr4_reg_to_regnum);
}

/* System V Release 4 (SVR4).  */

void
i386_svr4_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  /* System V Release 4 uses ELF.  */
  i386_elf_init_abi (info, gdbarch);

  /* System V Release 4 has shared libraries.  */
  set_gdbarch_skip_trampoline_code (gdbarch, find_solib_trampoline_target);

  tdep->sigtramp_p = i386_svr4_sigtramp_p;
  tdep->sigcontext_addr = i386_svr4_sigcontext_addr;
  tdep->sc_pc_offset = 36 + 14 * 4;
  tdep->sc_sp_offset = 36 + 17 * 4;

  tdep->jb_pc_offset = 20;
}

/* DJGPP.  */

static void
i386_go32_init_abi (struct gdbarch_info info, struct gdbarch *gdbarch)
{
  struct gdbarch_tdep *tdep = gdbarch_tdep (gdbarch);

  /* DJGPP doesn't have any special frames for signal handlers.  */
  tdep->sigtramp_p = NULL;

  tdep->jb_pc_offset = 36;
}


/* i386 register groups.  In addition to the normal groups, add "mmx"
   and "sse".  */

static struct reggroup *i386_sse_reggroup;
static struct reggroup *i386_mmx_reggroup;

static void
i386_init_reggroups (void)
{
  i386_sse_reggroup = reggroup_new ("sse", USER_REGGROUP);
  i386_mmx_reggroup = reggroup_new ("mmx", USER_REGGROUP);
}

static void
i386_add_reggroups (struct gdbarch *gdbarch)
{
  reggroup_add (gdbarch, i386_sse_reggroup);
  reggroup_add (gdbarch, i386_mmx_reggroup);
  reggroup_add (gdbarch, general_reggroup);
  reggroup_add (gdbarch, float_reggroup);
  reggroup_add (gdbarch, all_reggroup);
  reggroup_add (gdbarch, save_reggroup);
  reggroup_add (gdbarch, restore_reggroup);
  reggroup_add (gdbarch, vector_reggroup);
  reggroup_add (gdbarch, system_reggroup);
}

int
i386_register_reggroup_p (struct gdbarch *gdbarch, int regnum,
			  struct reggroup *group)
{
  int sse_regnum_p = (i386_sse_regnum_p (gdbarch, regnum)
		      || i386_mxcsr_regnum_p (gdbarch, regnum));
  int fp_regnum_p = (i386_fp_regnum_p (gdbarch, regnum)
		     || i386_fpc_regnum_p (gdbarch, regnum));
  int mmx_regnum_p = (i386_mmx_regnum_p (gdbarch, regnum));

  if (group == i386_mmx_reggroup)
    return mmx_regnum_p;
  if (group == i386_sse_reggroup)
    return sse_regnum_p;
  if (group == vector_reggroup)
    return (mmx_regnum_p || sse_regnum_p);
  if (group == float_reggroup)
    return fp_regnum_p;
  if (group == general_reggroup)
    return (!fp_regnum_p && !mmx_regnum_p && !sse_regnum_p);

  return default_register_reggroup_p (gdbarch, regnum, group);
}


/* Get the ARGIth function argument for the current function.  */

static CORE_ADDR
i386_fetch_pointer_argument (struct frame_info *frame, int argi, 
			     struct type *type)
{
  CORE_ADDR sp = get_frame_register_unsigned  (frame, I386_ESP_REGNUM);
  return read_memory_unsigned_integer (sp + (4 * (argi + 1)), 4);
}

static void
i386_skip_permanent_breakpoint (struct regcache *regcache)
{
  CORE_ADDR current_pc = regcache_read_pc (regcache);

 /* On i386, breakpoint is exactly 1 byte long, so we just
    adjust the PC in the regcache.  */
  current_pc += 1;
  regcache_write_pc (regcache, current_pc);
}


static struct gdbarch *
i386_gdbarch_init (struct gdbarch_info info, struct gdbarch_list *arches)
{
  struct gdbarch_tdep *tdep;
  struct gdbarch *gdbarch;

  /* If there is already a candidate, use it.  */
  arches = gdbarch_list_lookup_by_info (arches, &info);
  if (arches != NULL)
    return arches->gdbarch;

  /* Allocate space for the new architecture.  */
  tdep = XCALLOC (1, struct gdbarch_tdep);
  gdbarch = gdbarch_alloc (&info, tdep);

  /* General-purpose registers.  */
  tdep->gregset = NULL;
  tdep->gregset_reg_offset = NULL;
  tdep->gregset_num_regs = I386_NUM_GREGS;
  tdep->sizeof_gregset = 0;

  /* Floating-point registers.  */
  tdep->fpregset = NULL;
  tdep->sizeof_fpregset = I387_SIZEOF_FSAVE;

  /* The default settings include the FPU registers, the MMX registers
     and the SSE registers.  This can be overridden for a specific ABI
     by adjusting the members `st0_regnum', `mm0_regnum' and
     `num_xmm_regs' of `struct gdbarch_tdep', otherwise the registers
     will show up in the output of "info all-registers".  Ideally we
     should try to autodetect whether they are available, such that we
     can prevent "info all-registers" from displaying registers that
     aren't available.

     NOTE: kevinb/2003-07-13: ... if it's a choice between printing
     [the SSE registers] always (even when they don't exist) or never
     showing them to the user (even when they do exist), I prefer the
     former over the latter.  */

  tdep->st0_regnum = I386_ST0_REGNUM;

  /* The MMX registers are implemented as pseudo-registers.  Put off
     calculating the register number for %mm0 until we know the number
     of raw registers.  */
  tdep->mm0_regnum = 0;

  /* I386_NUM_XREGS includes %mxcsr, so substract one.  */
  tdep->num_xmm_regs = I386_NUM_XREGS - 1;

  tdep->jb_pc_offset = -1;
  tdep->struct_return = pcc_struct_return;
  tdep->sigtramp_start = 0;
  tdep->sigtramp_end = 0;
  tdep->sigtramp_p = i386_sigtramp_p;
  tdep->sigcontext_addr = NULL;
  tdep->sc_reg_offset = NULL;
  tdep->sc_pc_offset = -1;
  tdep->sc_sp_offset = -1;

  /* The format used for `long double' on almost all i386 targets is
     the i387 extended floating-point format.  In fact, of all targets
     in the GCC 2.95 tree, only OSF/1 does it different, and insists
     on having a `long double' that's not `long' at all.  */
  set_gdbarch_long_double_format (gdbarch, floatformats_i387_ext);

  /* Although the i387 extended floating-point has only 80 significant
     bits, a `long double' actually takes up 96, probably to enforce
     alignment.  */
  set_gdbarch_long_double_bit (gdbarch, 96);

  /* The default ABI includes general-purpose registers, 
     floating-point registers, and the SSE registers.  */
  set_gdbarch_num_regs (gdbarch, I386_SSE_NUM_REGS);
  set_gdbarch_register_name (gdbarch, i386_register_name);
  set_gdbarch_register_type (gdbarch, i386_register_type);

  /* Register numbers of various important registers.  */
  set_gdbarch_sp_regnum (gdbarch, I386_ESP_REGNUM); /* %esp */
  set_gdbarch_pc_regnum (gdbarch, I386_EIP_REGNUM); /* %eip */
  set_gdbarch_ps_regnum (gdbarch, I386_EFLAGS_REGNUM); /* %eflags */
  set_gdbarch_fp0_regnum (gdbarch, I386_ST0_REGNUM); /* %st(0) */

  /* NOTE: kettenis/20040418: GCC does have two possible register
     numbering schemes on the i386: dbx and SVR4.  These schemes
     differ in how they number %ebp, %esp, %eflags, and the
     floating-point registers, and are implemented by the arrays
     dbx_register_map[] and svr4_dbx_register_map in
     gcc/config/i386.c.  GCC also defines a third numbering scheme in
     gcc/config/i386.c, which it designates as the "default" register
     map used in 64bit mode.  This last register numbering scheme is
     implemented in dbx64_register_map, and is used for AMD64; see
     amd64-tdep.c.

     Currently, each GCC i386 target always uses the same register
     numbering scheme across all its supported debugging formats
     i.e. SDB (COFF), stabs and DWARF 2.  This is because
     gcc/sdbout.c, gcc/dbxout.c and gcc/dwarf2out.c all use the
     DBX_REGISTER_NUMBER macro which is defined by each target's
     respective config header in a manner independent of the requested
     output debugging format.

     This does not match the arrangement below, which presumes that
     the SDB and stabs numbering schemes differ from the DWARF and
     DWARF 2 ones.  The reason for this arrangement is that it is
     likely to get the numbering scheme for the target's
     default/native debug format right.  For targets where GCC is the
     native compiler (FreeBSD, NetBSD, OpenBSD, GNU/Linux) or for
     targets where the native toolchain uses a different numbering
     scheme for a particular debug format (stabs-in-ELF on Solaris)
     the defaults below will have to be overridden, like
     i386_elf_init_abi() does.  */

  /* Use the dbx register numbering scheme for stabs and COFF.  */
  set_gdbarch_stab_reg_to_regnum (gdbarch, i386_dbx_reg_to_regnum);
  set_gdbarch_sdb_reg_to_regnum (gdbarch, i386_dbx_reg_to_regnum);

  /* Use the SVR4 register numbering scheme for DWARF 2.  */
  set_gdbarch_dwarf2_reg_to_regnum (gdbarch, i386_svr4_reg_to_regnum);

  /* We don't set gdbarch_stab_reg_to_regnum, since ECOFF doesn't seem to
     be in use on any of the supported i386 targets.  */

  set_gdbarch_print_float_info (gdbarch, i387_print_float_info);

  set_gdbarch_get_longjmp_target (gdbarch, i386_get_longjmp_target);

  /* Call dummy code.  */
  set_gdbarch_push_dummy_call (gdbarch, i386_push_dummy_call);

  set_gdbarch_convert_register_p (gdbarch, i386_convert_register_p);
  set_gdbarch_register_to_value (gdbarch,  i386_register_to_value);
  set_gdbarch_value_to_register (gdbarch, i386_value_to_register);

  set_gdbarch_return_value (gdbarch, i386_return_value);

  set_gdbarch_skip_prologue (gdbarch, i386_skip_prologue);

  /* Stack grows downward.  */
  set_gdbarch_inner_than (gdbarch, core_addr_lessthan);

  set_gdbarch_breakpoint_from_pc (gdbarch, i386_breakpoint_from_pc);
  set_gdbarch_decr_pc_after_break (gdbarch, 1);
  set_gdbarch_max_insn_length (gdbarch, I386_MAX_INSN_LEN);

  set_gdbarch_frame_args_skip (gdbarch, 8);

  /* Wire in the MMX registers.  */
  set_gdbarch_num_pseudo_regs (gdbarch, i386_num_mmx_regs);
  set_gdbarch_pseudo_register_read (gdbarch, i386_pseudo_register_read);
  set_gdbarch_pseudo_register_write (gdbarch, i386_pseudo_register_write);

  set_gdbarch_print_insn (gdbarch, i386_print_insn);

  set_gdbarch_dummy_id (gdbarch, i386_dummy_id);

  set_gdbarch_unwind_pc (gdbarch, i386_unwind_pc);

  /* Add the i386 register groups.  */
  i386_add_reggroups (gdbarch);
  set_gdbarch_register_reggroup_p (gdbarch, i386_register_reggroup_p);

  /* Helper for function argument information.  */
  set_gdbarch_fetch_pointer_argument (gdbarch, i386_fetch_pointer_argument);

  /* Hook in the DWARF CFI frame unwinder.  */
  dwarf2_append_unwinders (gdbarch);

  frame_base_set_default (gdbarch, &i386_frame_base);

  /* Hook in ABI-specific overrides, if they have been registered.  */
  gdbarch_init_osabi (info, gdbarch);

  frame_unwind_append_unwinder (gdbarch, &i386_sigtramp_frame_unwind);
  frame_unwind_append_unwinder (gdbarch, &i386_frame_unwind);

  /* If we have a register mapping, enable the generic core file
     support, unless it has already been enabled.  */
  if (tdep->gregset_reg_offset
      && !gdbarch_regset_from_core_section_p (gdbarch))
    set_gdbarch_regset_from_core_section (gdbarch,
					  i386_regset_from_core_section);

  /* Unless support for MMX has been disabled, make %mm0 the first
     pseudo-register.  */
  if (tdep->mm0_regnum == 0)
    tdep->mm0_regnum = gdbarch_num_regs (gdbarch);

  set_gdbarch_skip_permanent_breakpoint (gdbarch,
					 i386_skip_permanent_breakpoint);

  return gdbarch;
}

static enum gdb_osabi
i386_coff_osabi_sniffer (bfd *abfd)
{
  if (strcmp (bfd_get_target (abfd), "coff-go32-exe") == 0
      || strcmp (bfd_get_target (abfd), "coff-go32") == 0)
    return GDB_OSABI_GO32;

  return GDB_OSABI_UNKNOWN;
}


/* Provide a prototype to silence -Wmissing-prototypes.  */
void _initialize_i386_tdep (void);

void
_initialize_i386_tdep (void)
{
  register_gdbarch_init (bfd_arch_i386, i386_gdbarch_init);

  /* Add the variable that controls the disassembly flavor.  */
  add_setshow_enum_cmd ("disassembly-flavor", no_class, valid_flavors,
			&disassembly_flavor, _("\
Set the disassembly flavor."), _("\
Show the disassembly flavor."), _("\
The valid values are \"att\" and \"intel\", and the default value is \"att\"."),
			NULL,
			NULL, /* FIXME: i18n: */
			&setlist, &showlist);

  /* Add the variable that controls the convention for returning
     structs.  */
  add_setshow_enum_cmd ("struct-convention", no_class, valid_conventions,
			&struct_convention, _("\
Set the convention for returning small structs."), _("\
Show the convention for returning small structs."), _("\
Valid values are \"default\", \"pcc\" and \"reg\", and the default value\n\
is \"default\"."),
			NULL,
			NULL, /* FIXME: i18n: */
			&setlist, &showlist);

  gdbarch_register_osabi_sniffer (bfd_arch_i386, bfd_target_coff_flavour,
				  i386_coff_osabi_sniffer);

  gdbarch_register_osabi (bfd_arch_i386, 0, GDB_OSABI_SVR4,
			  i386_svr4_init_abi);
  gdbarch_register_osabi (bfd_arch_i386, 0, GDB_OSABI_GO32,
			  i386_go32_init_abi);

  /* Initialize the i386-specific register groups & types.  */
  i386_init_reggroups ();
  i386_init_types();
}