New vectorizer messages; message format change.

[official-gcc.git] / gcc / function.c

/* Expands front end tree to back end RTL for GCC.
   Copyright (C) 1987-2013 Free Software Foundation, Inc.

This file is part of GCC.

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

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

You should have received a copy of the GNU General Public License
along with GCC; see the file COPYING3.  If not see
<http://www.gnu.org/licenses/>.  */

/* This file handles the generation of rtl code from tree structure
   at the level of the function as a whole.
   It creates the rtl expressions for parameters and auto variables
   and has full responsibility for allocating stack slots.

   `expand_function_start' is called at the beginning of a function,
   before the function body is parsed, and `expand_function_end' is
   called after parsing the body.

   Call `assign_stack_local' to allocate a stack slot for a local variable.
   This is usually done during the RTL generation for the function body,
   but it can also be done in the reload pass when a pseudo-register does
   not get a hard register.  */

#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "tm.h"
#include "rtl-error.h"
#include "tree.h"
#include "flags.h"
#include "except.h"
#include "function.h"
#include "expr.h"
#include "optabs.h"
#include "libfuncs.h"
#include "regs.h"
#include "hard-reg-set.h"
#include "insn-config.h"
#include "recog.h"
#include "output.h"
#include "basic-block.h"
#include "hashtab.h"
#include "ggc.h"
#include "tm_p.h"
#include "langhooks.h"
#include "target.h"
#include "common/common-target.h"
#include "gimple.h"
#include "tree-pass.h"
#include "predict.h"
#include "df.h"
#include "params.h"
#include "bb-reorder.h"

/* So we can assign to cfun in this file.  */
#undef cfun

#ifndef STACK_ALIGNMENT_NEEDED
#define STACK_ALIGNMENT_NEEDED 1
#endif

#define STACK_BYTES (STACK_BOUNDARY / BITS_PER_UNIT)

/* Some systems use __main in a way incompatible with its use in gcc, in these
   cases use the macros NAME__MAIN to give a quoted symbol and SYMBOL__MAIN to
   give the same symbol without quotes for an alternative entry point.  You
   must define both, or neither.  */
#ifndef NAME__MAIN
#define NAME__MAIN "__main"
#endif

/* Round a value to the lowest integer less than it that is a multiple of
   the required alignment.  Avoid using division in case the value is
   negative.  Assume the alignment is a power of two.  */
#define FLOOR_ROUND(VALUE,ALIGN) ((VALUE) & ~((ALIGN) - 1))

/* Similar, but round to the next highest integer that meets the
   alignment.  */
#define CEIL_ROUND(VALUE,ALIGN) (((VALUE) + (ALIGN) - 1) & ~((ALIGN)- 1))

/* Nonzero once virtual register instantiation has been done.
   assign_stack_local uses frame_pointer_rtx when this is nonzero.
   calls.c:emit_library_call_value_1 uses it to set up
   post-instantiation libcalls.  */
int virtuals_instantiated;

/* Assign unique numbers to labels generated for profiling, debugging, etc.  */
static GTY(()) int funcdef_no;

/* These variables hold pointers to functions to create and destroy
   target specific, per-function data structures.  */
struct machine_function * (*init_machine_status) (void);

/* The currently compiled function.  */
struct function *cfun = 0;

/* These hashes record the prologue and epilogue insns.  */
static GTY((if_marked ("ggc_marked_p"), param_is (struct rtx_def)))
  htab_t prologue_insn_hash;
static GTY((if_marked ("ggc_marked_p"), param_is (struct rtx_def)))
  htab_t epilogue_insn_hash;
\f

htab_t types_used_by_vars_hash = NULL;
vec<tree, va_gc> *types_used_by_cur_var_decl;

/* Forward declarations.  */

static struct temp_slot *find_temp_slot_from_address (rtx);
static void pad_to_arg_alignment (struct args_size *, int, struct args_size *);
static void pad_below (struct args_size *, enum machine_mode, tree);
static void reorder_blocks_1 (rtx, tree, vec<tree> *);
static int all_blocks (tree, tree *);
static tree *get_block_vector (tree, int *);
extern tree debug_find_var_in_block_tree (tree, tree);
/* We always define `record_insns' even if it's not used so that we
   can always export `prologue_epilogue_contains'.  */
static void record_insns (rtx, rtx, htab_t *) ATTRIBUTE_UNUSED;
static bool contains (const_rtx, htab_t);
static void prepare_function_start (void);
static void do_clobber_return_reg (rtx, void *);
static void do_use_return_reg (rtx, void *);
static void set_insn_locations (rtx, int) ATTRIBUTE_UNUSED;
\f
/* Stack of nested functions.  */
/* Keep track of the cfun stack.  */

typedef struct function *function_p;

static vec<function_p> function_context_stack;

/* Save the current context for compilation of a nested function.
   This is called from language-specific code.  */

void
push_function_context (void)
{
  if (cfun == 0)
    allocate_struct_function (NULL, false);

  function_context_stack.safe_push (cfun);
  set_cfun (NULL);
}

/* Restore the last saved context, at the end of a nested function.
   This function is called from language-specific code.  */

void
pop_function_context (void)
{
  struct function *p = function_context_stack.pop ();
  set_cfun (p);
  current_function_decl = p->decl;

  /* Reset variables that have known state during rtx generation.  */
  virtuals_instantiated = 0;
  generating_concat_p = 1;
}

/* Clear out all parts of the state in F that can safely be discarded
   after the function has been parsed, but not compiled, to let
   garbage collection reclaim the memory.  */

void
free_after_parsing (struct function *f)
{
  f->language = 0;
}

/* Clear out all parts of the state in F that can safely be discarded
   after the function has been compiled, to let garbage collection
   reclaim the memory.  */

void
free_after_compilation (struct function *f)
{
  prologue_insn_hash = NULL;
  epilogue_insn_hash = NULL;

  free (crtl->emit.regno_pointer_align);

  memset (crtl, 0, sizeof (struct rtl_data));
  f->eh = NULL;
  f->machine = NULL;
  f->cfg = NULL;

  regno_reg_rtx = NULL;
}
\f
/* Return size needed for stack frame based on slots so far allocated.
   This size counts from zero.  It is not rounded to PREFERRED_STACK_BOUNDARY;
   the caller may have to do that.  */

HOST_WIDE_INT
get_frame_size (void)
{
  if (FRAME_GROWS_DOWNWARD)
    return -frame_offset;
  else
    return frame_offset;
}

/* Issue an error message and return TRUE if frame OFFSET overflows in
   the signed target pointer arithmetics for function FUNC.  Otherwise
   return FALSE.  */

bool
frame_offset_overflow (HOST_WIDE_INT offset, tree func)
{
  unsigned HOST_WIDE_INT size = FRAME_GROWS_DOWNWARD ? -offset : offset;

  if (size > ((unsigned HOST_WIDE_INT) 1 << (GET_MODE_BITSIZE (Pmode) - 1))
               /* Leave room for the fixed part of the frame.  */
               - 64 * UNITS_PER_WORD)
    {
      error_at (DECL_SOURCE_LOCATION (func),
                "total size of local objects too large");
      return TRUE;
    }

  return FALSE;
}

/* Return stack slot alignment in bits for TYPE and MODE.  */

static unsigned int
get_stack_local_alignment (tree type, enum machine_mode mode)
{
  unsigned int alignment;

  if (mode == BLKmode)
    alignment = BIGGEST_ALIGNMENT;
  else
    alignment = GET_MODE_ALIGNMENT (mode);

  /* Allow the frond-end to (possibly) increase the alignment of this
     stack slot.  */
  if (! type)
    type = lang_hooks.types.type_for_mode (mode, 0);

  return STACK_SLOT_ALIGNMENT (type, mode, alignment);
}

/* Determine whether it is possible to fit a stack slot of size SIZE and
   alignment ALIGNMENT into an area in the stack frame that starts at
   frame offset START and has a length of LENGTH.  If so, store the frame
   offset to be used for the stack slot in *POFFSET and return true;
   return false otherwise.  This function will extend the frame size when
   given a start/length pair that lies at the end of the frame.  */

static bool
try_fit_stack_local (HOST_WIDE_INT start, HOST_WIDE_INT length,
                     HOST_WIDE_INT size, unsigned int alignment,
                     HOST_WIDE_INT *poffset)
{
  HOST_WIDE_INT this_frame_offset;
  int frame_off, frame_alignment, frame_phase;

  /* Calculate how many bytes the start of local variables is off from
     stack alignment.  */
  frame_alignment = PREFERRED_STACK_BOUNDARY / BITS_PER_UNIT;
  frame_off = STARTING_FRAME_OFFSET % frame_alignment;
  frame_phase = frame_off ? frame_alignment - frame_off : 0;

  /* Round the frame offset to the specified alignment.  */

  /*  We must be careful here, since FRAME_OFFSET might be negative and
      division with a negative dividend isn't as well defined as we might
      like.  So we instead assume that ALIGNMENT is a power of two and
      use logical operations which are unambiguous.  */
  if (FRAME_GROWS_DOWNWARD)
    this_frame_offset
      = (FLOOR_ROUND (start + length - size - frame_phase,
                      (unsigned HOST_WIDE_INT) alignment)
         + frame_phase);
  else
    this_frame_offset
      = (CEIL_ROUND (start - frame_phase,
                     (unsigned HOST_WIDE_INT) alignment)
         + frame_phase);

  /* See if it fits.  If this space is at the edge of the frame,
     consider extending the frame to make it fit.  Our caller relies on
     this when allocating a new slot.  */
  if (frame_offset == start && this_frame_offset < frame_offset)
    frame_offset = this_frame_offset;
  else if (this_frame_offset < start)
    return false;
  else if (start + length == frame_offset
           && this_frame_offset + size > start + length)
    frame_offset = this_frame_offset + size;
  else if (this_frame_offset + size > start + length)
    return false;

  *poffset = this_frame_offset;
  return true;
}

/* Create a new frame_space structure describing free space in the stack
   frame beginning at START and ending at END, and chain it into the
   function's frame_space_list.  */

static void
add_frame_space (HOST_WIDE_INT start, HOST_WIDE_INT end)
{
  struct frame_space *space = ggc_alloc_frame_space ();
  space->next = crtl->frame_space_list;
  crtl->frame_space_list = space;
  space->start = start;
  space->length = end - start;
}

/* Allocate a stack slot of SIZE bytes and return a MEM rtx for it
   with machine mode MODE.

   ALIGN controls the amount of alignment for the address of the slot:
   0 means according to MODE,
   -1 means use BIGGEST_ALIGNMENT and round size to multiple of that,
   -2 means use BITS_PER_UNIT,
   positive specifies alignment boundary in bits.

   KIND has ASLK_REDUCE_ALIGN bit set if it is OK to reduce
   alignment and ASLK_RECORD_PAD bit set if we should remember
   extra space we allocated for alignment purposes.  When we are
   called from assign_stack_temp_for_type, it is not set so we don't
   track the same stack slot in two independent lists.

   We do not round to stack_boundary here.  */

rtx
assign_stack_local_1 (enum machine_mode mode, HOST_WIDE_INT size,
                      int align, int kind)
{
  rtx x, addr;
  int bigend_correction = 0;
  HOST_WIDE_INT slot_offset = 0, old_frame_offset;
  unsigned int alignment, alignment_in_bits;

  if (align == 0)
    {
      alignment = get_stack_local_alignment (NULL, mode);
      alignment /= BITS_PER_UNIT;
    }
  else if (align == -1)
    {
      alignment = BIGGEST_ALIGNMENT / BITS_PER_UNIT;
      size = CEIL_ROUND (size, alignment);
    }
  else if (align == -2)
    alignment = 1; /* BITS_PER_UNIT / BITS_PER_UNIT */
  else
    alignment = align / BITS_PER_UNIT;

  alignment_in_bits = alignment * BITS_PER_UNIT;

  /* Ignore alignment if it exceeds MAX_SUPPORTED_STACK_ALIGNMENT.  */
  if (alignment_in_bits > MAX_SUPPORTED_STACK_ALIGNMENT)
    {
      alignment_in_bits = MAX_SUPPORTED_STACK_ALIGNMENT;
      alignment = alignment_in_bits / BITS_PER_UNIT;
    }

  if (SUPPORTS_STACK_ALIGNMENT)
    {
      if (crtl->stack_alignment_estimated < alignment_in_bits)
        {
          if (!crtl->stack_realign_processed)
            crtl->stack_alignment_estimated = alignment_in_bits;
          else
            {
              /* If stack is realigned and stack alignment value
                 hasn't been finalized, it is OK not to increase
                 stack_alignment_estimated.  The bigger alignment
                 requirement is recorded in stack_alignment_needed
                 below.  */
              gcc_assert (!crtl->stack_realign_finalized);
              if (!crtl->stack_realign_needed)
                {
                  /* It is OK to reduce the alignment as long as the
                     requested size is 0 or the estimated stack
                     alignment >= mode alignment.  */
                  gcc_assert ((kind & ASLK_REDUCE_ALIGN)
                              || size == 0
                              || (crtl->stack_alignment_estimated
                                  >= GET_MODE_ALIGNMENT (mode)));
                  alignment_in_bits = crtl->stack_alignment_estimated;
                  alignment = alignment_in_bits / BITS_PER_UNIT;
                }
            }
        }
    }

  if (crtl->stack_alignment_needed < alignment_in_bits)
    crtl->stack_alignment_needed = alignment_in_bits;
  if (crtl->max_used_stack_slot_alignment < alignment_in_bits)
    crtl->max_used_stack_slot_alignment = alignment_in_bits;

  if (mode != BLKmode || size != 0)
    {
      if (kind & ASLK_RECORD_PAD)
        {
          struct frame_space **psp;

          for (psp = &crtl->frame_space_list; *psp; psp = &(*psp)->next)
            {
              struct frame_space *space = *psp;
              if (!try_fit_stack_local (space->start, space->length, size,
                                        alignment, &slot_offset))
                continue;
              *psp = space->next;
              if (slot_offset > space->start)
                add_frame_space (space->start, slot_offset);
              if (slot_offset + size < space->start + space->length)
                add_frame_space (slot_offset + size,
                                 space->start + space->length);
              goto found_space;
            }
        }
    }
  else if (!STACK_ALIGNMENT_NEEDED)
    {
      slot_offset = frame_offset;
      goto found_space;
    }

  old_frame_offset = frame_offset;

  if (FRAME_GROWS_DOWNWARD)
    {
      frame_offset -= size;
      try_fit_stack_local (frame_offset, size, size, alignment, &slot_offset);

      if (kind & ASLK_RECORD_PAD)
        {
          if (slot_offset > frame_offset)
            add_frame_space (frame_offset, slot_offset);
          if (slot_offset + size < old_frame_offset)
            add_frame_space (slot_offset + size, old_frame_offset);
        }
    }
  else
    {
      frame_offset += size;
      try_fit_stack_local (old_frame_offset, size, size, alignment, &slot_offset);

      if (kind & ASLK_RECORD_PAD)
        {
          if (slot_offset > old_frame_offset)
            add_frame_space (old_frame_offset, slot_offset);
          if (slot_offset + size < frame_offset)
            add_frame_space (slot_offset + size, frame_offset);
        }
    }

 found_space:
  /* On a big-endian machine, if we are allocating more space than we will use,
     use the least significant bytes of those that are allocated.  */
  if (BYTES_BIG_ENDIAN && mode != BLKmode && GET_MODE_SIZE (mode) < size)
    bigend_correction = size - GET_MODE_SIZE (mode);

  /* If we have already instantiated virtual registers, return the actual
     address relative to the frame pointer.  */
  if (virtuals_instantiated)
    addr = plus_constant (Pmode, frame_pointer_rtx,
                          trunc_int_for_mode
                          (slot_offset + bigend_correction
                           + STARTING_FRAME_OFFSET, Pmode));
  else
    addr = plus_constant (Pmode, virtual_stack_vars_rtx,
                          trunc_int_for_mode
                          (slot_offset + bigend_correction,
                           Pmode));

  x = gen_rtx_MEM (mode, addr);
  set_mem_align (x, alignment_in_bits);
  MEM_NOTRAP_P (x) = 1;

  stack_slot_list
    = gen_rtx_EXPR_LIST (VOIDmode, x, stack_slot_list);

  if (frame_offset_overflow (frame_offset, current_function_decl))
    frame_offset = 0;

  return x;
}

/* Wrap up assign_stack_local_1 with last parameter as false.  */

rtx
assign_stack_local (enum machine_mode mode, HOST_WIDE_INT size, int align)
{
  return assign_stack_local_1 (mode, size, align, ASLK_RECORD_PAD);
}
\f
/* In order to evaluate some expressions, such as function calls returning
   structures in memory, we need to temporarily allocate stack locations.
   We record each allocated temporary in the following structure.

   Associated with each temporary slot is a nesting level.  When we pop up
   one level, all temporaries associated with the previous level are freed.
   Normally, all temporaries are freed after the execution of the statement
   in which they were created.  However, if we are inside a ({...}) grouping,
   the result may be in a temporary and hence must be preserved.  If the
   result could be in a temporary, we preserve it if we can determine which
   one it is in.  If we cannot determine which temporary may contain the
   result, all temporaries are preserved.  A temporary is preserved by
   pretending it was allocated at the previous nesting level.  */

struct GTY(()) temp_slot {
  /* Points to next temporary slot.  */
  struct temp_slot *next;
  /* Points to previous temporary slot.  */
  struct temp_slot *prev;
  /* The rtx to used to reference the slot.  */
  rtx slot;
  /* The size, in units, of the slot.  */
  HOST_WIDE_INT size;
  /* The type of the object in the slot, or zero if it doesn't correspond
     to a type.  We use this to determine whether a slot can be reused.
     It can be reused if objects of the type of the new slot will always
     conflict with objects of the type of the old slot.  */
  tree type;
  /* The alignment (in bits) of the slot.  */
  unsigned int align;
  /* Nonzero if this temporary is currently in use.  */
  char in_use;
  /* Nesting level at which this slot is being used.  */
  int level;
  /* The offset of the slot from the frame_pointer, including extra space
     for alignment.  This info is for combine_temp_slots.  */
  HOST_WIDE_INT base_offset;
  /* The size of the slot, including extra space for alignment.  This
     info is for combine_temp_slots.  */
  HOST_WIDE_INT full_size;
};

/* A table of addresses that represent a stack slot.  The table is a mapping
   from address RTXen to a temp slot.  */
static GTY((param_is(struct temp_slot_address_entry))) htab_t temp_slot_address_table;
static size_t n_temp_slots_in_use;

/* Entry for the above hash table.  */
struct GTY(()) temp_slot_address_entry {
  hashval_t hash;
  rtx address;
  struct temp_slot *temp_slot;
};

/* Removes temporary slot TEMP from LIST.  */

static void
cut_slot_from_list (struct temp_slot *temp, struct temp_slot **list)
{
  if (temp->next)
    temp->next->prev = temp->prev;
  if (temp->prev)
    temp->prev->next = temp->next;
  else
    *list = temp->next;

  temp->prev = temp->next = NULL;
}

/* Inserts temporary slot TEMP to LIST.  */

static void
insert_slot_to_list (struct temp_slot *temp, struct temp_slot **list)
{
  temp->next = *list;
  if (*list)
    (*list)->prev = temp;
  temp->prev = NULL;
  *list = temp;
}

/* Returns the list of used temp slots at LEVEL.  */

static struct temp_slot **
temp_slots_at_level (int level)
{
  if (level >= (int) vec_safe_length (used_temp_slots))
    vec_safe_grow_cleared (used_temp_slots, level + 1);

  return &(*used_temp_slots)[level];
}

/* Returns the maximal temporary slot level.  */

static int
max_slot_level (void)
{
  if (!used_temp_slots)
    return -1;

  return used_temp_slots->length () - 1;
}

/* Moves temporary slot TEMP to LEVEL.  */

static void
move_slot_to_level (struct temp_slot *temp, int level)
{
  cut_slot_from_list (temp, temp_slots_at_level (temp->level));
  insert_slot_to_list (temp, temp_slots_at_level (level));
  temp->level = level;
}

/* Make temporary slot TEMP available.  */

static void
make_slot_available (struct temp_slot *temp)
{
  cut_slot_from_list (temp, temp_slots_at_level (temp->level));
  insert_slot_to_list (temp, &avail_temp_slots);
  temp->in_use = 0;
  temp->level = -1;
  n_temp_slots_in_use--;
}

/* Compute the hash value for an address -> temp slot mapping.
   The value is cached on the mapping entry.  */
static hashval_t
temp_slot_address_compute_hash (struct temp_slot_address_entry *t)
{
  int do_not_record = 0;
  return hash_rtx (t->address, GET_MODE (t->address),
                   &do_not_record, NULL, false);
}

/* Return the hash value for an address -> temp slot mapping.  */
static hashval_t
temp_slot_address_hash (const void *p)
{
  const struct temp_slot_address_entry *t;
  t = (const struct temp_slot_address_entry *) p;
  return t->hash;
}

/* Compare two address -> temp slot mapping entries.  */
static int
temp_slot_address_eq (const void *p1, const void *p2)
{
  const struct temp_slot_address_entry *t1, *t2;
  t1 = (const struct temp_slot_address_entry *) p1;
  t2 = (const struct temp_slot_address_entry *) p2;
  return exp_equiv_p (t1->address, t2->address, 0, true);
}

/* Add ADDRESS as an alias of TEMP_SLOT to the addess -> temp slot mapping.  */
static void
insert_temp_slot_address (rtx address, struct temp_slot *temp_slot)
{
  void **slot;
  struct temp_slot_address_entry *t = ggc_alloc_temp_slot_address_entry ();
  t->address = address;
  t->temp_slot = temp_slot;
  t->hash = temp_slot_address_compute_hash (t);
  slot = htab_find_slot_with_hash (temp_slot_address_table, t, t->hash, INSERT);
  *slot = t;
}

/* Remove an address -> temp slot mapping entry if the temp slot is
   not in use anymore.  Callback for remove_unused_temp_slot_addresses.  */
static int
remove_unused_temp_slot_addresses_1 (void **slot, void *data ATTRIBUTE_UNUSED)
{
  const struct temp_slot_address_entry *t;
  t = (const struct temp_slot_address_entry *) *slot;
  if (! t->temp_slot->in_use)
    htab_clear_slot (temp_slot_address_table, slot);
  return 1;
}

/* Remove all mappings of addresses to unused temp slots.  */
static void
remove_unused_temp_slot_addresses (void)
{
  /* Use quicker clearing if there aren't any active temp slots.  */
  if (n_temp_slots_in_use)
    htab_traverse (temp_slot_address_table,
                   remove_unused_temp_slot_addresses_1,
                   NULL);
  else
    htab_empty (temp_slot_address_table);
}

/* Find the temp slot corresponding to the object at address X.  */

static struct temp_slot *
find_temp_slot_from_address (rtx x)
{
  struct temp_slot *p;
  struct temp_slot_address_entry tmp, *t;

  /* First try the easy way:
     See if X exists in the address -> temp slot mapping.  */
  tmp.address = x;
  tmp.temp_slot = NULL;
  tmp.hash = temp_slot_address_compute_hash (&tmp);
  t = (struct temp_slot_address_entry *)
    htab_find_with_hash (temp_slot_address_table, &tmp, tmp.hash);
  if (t)
    return t->temp_slot;

  /* If we have a sum involving a register, see if it points to a temp
     slot.  */
  if (GET_CODE (x) == PLUS && REG_P (XEXP (x, 0))
      && (p = find_temp_slot_from_address (XEXP (x, 0))) != 0)
    return p;
  else if (GET_CODE (x) == PLUS && REG_P (XEXP (x, 1))
           && (p = find_temp_slot_from_address (XEXP (x, 1))) != 0)
    return p;

  /* Last resort: Address is a virtual stack var address.  */
  if (GET_CODE (x) == PLUS
      && XEXP (x, 0) == virtual_stack_vars_rtx
      && CONST_INT_P (XEXP (x, 1)))
    {
      int i;
      for (i = max_slot_level (); i >= 0; i--)
        for (p = *temp_slots_at_level (i); p; p = p->next)
          {
            if (INTVAL (XEXP (x, 1)) >= p->base_offset
                && INTVAL (XEXP (x, 1)) < p->base_offset + p->full_size)
              return p;
          }
    }

  return NULL;
}
\f
/* Allocate a temporary stack slot and record it for possible later
   reuse.

   MODE is the machine mode to be given to the returned rtx.

   SIZE is the size in units of the space required.  We do no rounding here
   since assign_stack_local will do any required rounding.

   TYPE is the type that will be used for the stack slot.  */

rtx
assign_stack_temp_for_type (enum machine_mode mode, HOST_WIDE_INT size,
                            tree type)
{
  unsigned int align;
  struct temp_slot *p, *best_p = 0, *selected = NULL, **pp;
  rtx slot;

  /* If SIZE is -1 it means that somebody tried to allocate a temporary
     of a variable size.  */
  gcc_assert (size != -1);

  align = get_stack_local_alignment (type, mode);

  /* Try to find an available, already-allocated temporary of the proper
     mode which meets the size and alignment requirements.  Choose the
     smallest one with the closest alignment.

     If assign_stack_temp is called outside of the tree->rtl expansion,
     we cannot reuse the stack slots (that may still refer to
     VIRTUAL_STACK_VARS_REGNUM).  */
  if (!virtuals_instantiated)
    {
      for (p = avail_temp_slots; p; p = p->next)
        {
          if (p->align >= align && p->size >= size
              && GET_MODE (p->slot) == mode
              && objects_must_conflict_p (p->type, type)
              && (best_p == 0 || best_p->size > p->size
                  || (best_p->size == p->size && best_p->align > p->align)))
            {
              if (p->align == align && p->size == size)
                {
                  selected = p;
                  cut_slot_from_list (selected, &avail_temp_slots);
                  best_p = 0;
                  break;
                }
              best_p = p;
            }
        }
    }

  /* Make our best, if any, the one to use.  */
  if (best_p)
    {
      selected = best_p;
      cut_slot_from_list (selected, &avail_temp_slots);

      /* If there are enough aligned bytes left over, make them into a new
         temp_slot so that the extra bytes don't get wasted.  Do this only
         for BLKmode slots, so that we can be sure of the alignment.  */
      if (GET_MODE (best_p->slot) == BLKmode)
        {
          int alignment = best_p->align / BITS_PER_UNIT;
          HOST_WIDE_INT rounded_size = CEIL_ROUND (size, alignment);

          if (best_p->size - rounded_size >= alignment)
            {
              p = ggc_alloc_temp_slot ();
              p->in_use = 0;
              p->size = best_p->size - rounded_size;
              p->base_offset = best_p->base_offset + rounded_size;
              p->full_size = best_p->full_size - rounded_size;
              p->slot = adjust_address_nv (best_p->slot, BLKmode, rounded_size);
              p->align = best_p->align;
              p->type = best_p->type;
              insert_slot_to_list (p, &avail_temp_slots);

              stack_slot_list = gen_rtx_EXPR_LIST (VOIDmode, p->slot,
                                                   stack_slot_list);

              best_p->size = rounded_size;
              best_p->full_size = rounded_size;
            }
        }
    }

  /* If we still didn't find one, make a new temporary.  */
  if (selected == 0)
    {
      HOST_WIDE_INT frame_offset_old = frame_offset;

      p = ggc_alloc_temp_slot ();

      /* We are passing an explicit alignment request to assign_stack_local.
         One side effect of that is assign_stack_local will not round SIZE
         to ensure the frame offset remains suitably aligned.

         So for requests which depended on the rounding of SIZE, we go ahead
         and round it now.  We also make sure ALIGNMENT is at least
         BIGGEST_ALIGNMENT.  */
      gcc_assert (mode != BLKmode || align == BIGGEST_ALIGNMENT);
      p->slot = assign_stack_local_1 (mode,
                                      (mode == BLKmode
                                       ? CEIL_ROUND (size,
                                                     (int) align
                                                     / BITS_PER_UNIT)
                                       : size),
                                      align, 0);

      p->align = align;

      /* The following slot size computation is necessary because we don't
         know the actual size of the temporary slot until assign_stack_local
         has performed all the frame alignment and size rounding for the
         requested temporary.  Note that extra space added for alignment
         can be either above or below this stack slot depending on which
         way the frame grows.  We include the extra space if and only if it
         is above this slot.  */
      if (FRAME_GROWS_DOWNWARD)
        p->size = frame_offset_old - frame_offset;
      else
        p->size = size;

      /* Now define the fields used by combine_temp_slots.  */
      if (FRAME_GROWS_DOWNWARD)
        {
          p->base_offset = frame_offset;
          p->full_size = frame_offset_old - frame_offset;
        }
      else
        {
          p->base_offset = frame_offset_old;
          p->full_size = frame_offset - frame_offset_old;
        }

      selected = p;
    }

  p = selected;
  p->in_use = 1;
  p->type = type;
  p->level = temp_slot_level;
  n_temp_slots_in_use++;

  pp = temp_slots_at_level (p->level);
  insert_slot_to_list (p, pp);
  insert_temp_slot_address (XEXP (p->slot, 0), p);

  /* Create a new MEM rtx to avoid clobbering MEM flags of old slots.  */
  slot = gen_rtx_MEM (mode, XEXP (p->slot, 0));
  stack_slot_list = gen_rtx_EXPR_LIST (VOIDmode, slot, stack_slot_list);

  /* If we know the alias set for the memory that will be used, use
     it.  If there's no TYPE, then we don't know anything about the
     alias set for the memory.  */
  set_mem_alias_set (slot, type ? get_alias_set (type) : 0);
  set_mem_align (slot, align);

  /* If a type is specified, set the relevant flags.  */
  if (type != 0)
    MEM_VOLATILE_P (slot) = TYPE_VOLATILE (type);
  MEM_NOTRAP_P (slot) = 1;

  return slot;
}

/* Allocate a temporary stack slot and record it for possible later
   reuse.  First two arguments are same as in preceding function.  */

rtx
assign_stack_temp (enum machine_mode mode, HOST_WIDE_INT size)
{
  return assign_stack_temp_for_type (mode, size, NULL_TREE);
}
\f
/* Assign a temporary.
   If TYPE_OR_DECL is a decl, then we are doing it on behalf of the decl
   and so that should be used in error messages.  In either case, we
   allocate of the given type.
   MEMORY_REQUIRED is 1 if the result must be addressable stack memory;
   it is 0 if a register is OK.
   DONT_PROMOTE is 1 if we should not promote values in register
   to wider modes.  */

rtx
assign_temp (tree type_or_decl, int memory_required,
             int dont_promote ATTRIBUTE_UNUSED)
{
  tree type, decl;
  enum machine_mode mode;
#ifdef PROMOTE_MODE
  int unsignedp;
#endif

  if (DECL_P (type_or_decl))
    decl = type_or_decl, type = TREE_TYPE (decl);
  else
    decl = NULL, type = type_or_decl;

  mode = TYPE_MODE (type);
#ifdef PROMOTE_MODE
  unsignedp = TYPE_UNSIGNED (type);
#endif

  if (mode == BLKmode || memory_required)
    {
      HOST_WIDE_INT size = int_size_in_bytes (type);
      rtx tmp;

      /* Zero sized arrays are GNU C extension.  Set size to 1 to avoid
         problems with allocating the stack space.  */
      if (size == 0)
        size = 1;

      /* Unfortunately, we don't yet know how to allocate variable-sized
         temporaries.  However, sometimes we can find a fixed upper limit on
         the size, so try that instead.  */
      else if (size == -1)
        size = max_int_size_in_bytes (type);

      /* The size of the temporary may be too large to fit into an integer.  */
      /* ??? Not sure this should happen except for user silliness, so limit
         this to things that aren't compiler-generated temporaries.  The
         rest of the time we'll die in assign_stack_temp_for_type.  */
      if (decl && size == -1
          && TREE_CODE (TYPE_SIZE_UNIT (type)) == INTEGER_CST)
        {
          error ("size of variable %q+D is too large", decl);
          size = 1;
        }

      tmp = assign_stack_temp_for_type (mode, size, type);
      return tmp;
    }

#ifdef PROMOTE_MODE
  if (! dont_promote)
    mode = promote_mode (type, mode, &unsignedp);
#endif

  return gen_reg_rtx (mode);
}
\f
/* Combine temporary stack slots which are adjacent on the stack.

   This allows for better use of already allocated stack space.  This is only
   done for BLKmode slots because we can be sure that we won't have alignment
   problems in this case.  */

static void
combine_temp_slots (void)
{
  struct temp_slot *p, *q, *next, *next_q;
  int num_slots;

  /* We can't combine slots, because the information about which slot
     is in which alias set will be lost.  */
  if (flag_strict_aliasing)
    return;

  /* If there are a lot of temp slots, don't do anything unless
     high levels of optimization.  */
  if (! flag_expensive_optimizations)
    for (p = avail_temp_slots, num_slots = 0; p; p = p->next, num_slots++)
      if (num_slots > 100 || (num_slots > 10 && optimize == 0))
        return;

  for (p = avail_temp_slots; p; p = next)
    {
      int delete_p = 0;

      next = p->next;

      if (GET_MODE (p->slot) != BLKmode)
        continue;

      for (q = p->next; q; q = next_q)
        {
          int delete_q = 0;

          next_q = q->next;

          if (GET_MODE (q->slot) != BLKmode)
            continue;

          if (p->base_offset + p->full_size == q->base_offset)
            {
              /* Q comes after P; combine Q into P.  */
              p->size += q->size;
              p->full_size += q->full_size;
              delete_q = 1;
            }
          else if (q->base_offset + q->full_size == p->base_offset)
            {
              /* P comes after Q; combine P into Q.  */
              q->size += p->size;
              q->full_size += p->full_size;
              delete_p = 1;
              break;
            }
          if (delete_q)
            cut_slot_from_list (q, &avail_temp_slots);
        }

      /* Either delete P or advance past it.  */
      if (delete_p)
        cut_slot_from_list (p, &avail_temp_slots);
    }
}
\f
/* Indicate that NEW_RTX is an alternate way of referring to the temp
   slot that previously was known by OLD_RTX.  */

void
update_temp_slot_address (rtx old_rtx, rtx new_rtx)
{
  struct temp_slot *p;

  if (rtx_equal_p (old_rtx, new_rtx))
    return;

  p = find_temp_slot_from_address (old_rtx);

  /* If we didn't find one, see if both OLD_RTX is a PLUS.  If so, and
     NEW_RTX is a register, see if one operand of the PLUS is a
     temporary location.  If so, NEW_RTX points into it.  Otherwise,
     if both OLD_RTX and NEW_RTX are a PLUS and if there is a register
     in common between them.  If so, try a recursive call on those
     values.  */
  if (p == 0)
    {
      if (GET_CODE (old_rtx) != PLUS)
        return;

      if (REG_P (new_rtx))
        {
          update_temp_slot_address (XEXP (old_rtx, 0), new_rtx);
          update_temp_slot_address (XEXP (old_rtx, 1), new_rtx);
          return;
        }
      else if (GET_CODE (new_rtx) != PLUS)
        return;

      if (rtx_equal_p (XEXP (old_rtx, 0), XEXP (new_rtx, 0)))
        update_temp_slot_address (XEXP (old_rtx, 1), XEXP (new_rtx, 1));
      else if (rtx_equal_p (XEXP (old_rtx, 1), XEXP (new_rtx, 0)))
        update_temp_slot_address (XEXP (old_rtx, 0), XEXP (new_rtx, 1));
      else if (rtx_equal_p (XEXP (old_rtx, 0), XEXP (new_rtx, 1)))
        update_temp_slot_address (XEXP (old_rtx, 1), XEXP (new_rtx, 0));
      else if (rtx_equal_p (XEXP (old_rtx, 1), XEXP (new_rtx, 1)))
        update_temp_slot_address (XEXP (old_rtx, 0), XEXP (new_rtx, 0));

      return;
    }

  /* Otherwise add an alias for the temp's address.  */
  insert_temp_slot_address (new_rtx, p);
}

/* If X could be a reference to a temporary slot, mark that slot as
   belonging to the to one level higher than the current level.  If X
   matched one of our slots, just mark that one.  Otherwise, we can't
   easily predict which it is, so upgrade all of them.

   This is called when an ({...}) construct occurs and a statement
   returns a value in memory.  */

void
preserve_temp_slots (rtx x)
{
  struct temp_slot *p = 0, *next;

  if (x == 0)
    return;

  /* If X is a register that is being used as a pointer, see if we have
     a temporary slot we know it points to.  */
  if (REG_P (x) && REG_POINTER (x))
    p = find_temp_slot_from_address (x);

  /* If X is not in memory or is at a constant address, it cannot be in
     a temporary slot.  */
  if (p == 0 && (!MEM_P (x) || CONSTANT_P (XEXP (x, 0))))
    return;

  /* First see if we can find a match.  */
  if (p == 0)
    p = find_temp_slot_from_address (XEXP (x, 0));

  if (p != 0)
    {
      if (p->level == temp_slot_level)
        move_slot_to_level (p, temp_slot_level - 1);
      return;
    }

  /* Otherwise, preserve all non-kept slots at this level.  */
  for (p = *temp_slots_at_level (temp_slot_level); p; p = next)
    {
      next = p->next;
      move_slot_to_level (p, temp_slot_level - 1);
    }
}

/* Free all temporaries used so far.  This is normally called at the
   end of generating code for a statement.  */

void
free_temp_slots (void)
{
  struct temp_slot *p, *next;
  bool some_available = false;

  for (p = *temp_slots_at_level (temp_slot_level); p; p = next)
    {
      next = p->next;
      make_slot_available (p);
      some_available = true;
    }

  if (some_available)
    {
      remove_unused_temp_slot_addresses ();
      combine_temp_slots ();
    }
}

/* Push deeper into the nesting level for stack temporaries.  */

void
push_temp_slots (void)
{
  temp_slot_level++;
}

/* Pop a temporary nesting level.  All slots in use in the current level
   are freed.  */

void
pop_temp_slots (void)
{
  free_temp_slots ();
  temp_slot_level--;
}

/* Initialize temporary slots.  */

void
init_temp_slots (void)
{
  /* We have not allocated any temporaries yet.  */
  avail_temp_slots = 0;
  vec_alloc (used_temp_slots, 0);
  temp_slot_level = 0;
  n_temp_slots_in_use = 0;

  /* Set up the table to map addresses to temp slots.  */
  if (! temp_slot_address_table)
    temp_slot_address_table = htab_create_ggc (32,
                                               temp_slot_address_hash,
                                               temp_slot_address_eq,
                                               NULL);
  else
    htab_empty (temp_slot_address_table);
}
\f
/* Functions and data structures to keep track of the values hard regs
   had at the start of the function.  */

/* Private type used by get_hard_reg_initial_reg, get_hard_reg_initial_val,
   and has_hard_reg_initial_val..  */
typedef struct GTY(()) initial_value_pair {
  rtx hard_reg;
  rtx pseudo;
} initial_value_pair;
/* ???  This could be a VEC but there is currently no way to define an
   opaque VEC type.  This could be worked around by defining struct
   initial_value_pair in function.h.  */
typedef struct GTY(()) initial_value_struct {
  int num_entries;
  int max_entries;
  initial_value_pair * GTY ((length ("%h.num_entries"))) entries;
} initial_value_struct;

/* If a pseudo represents an initial hard reg (or expression), return
   it, else return NULL_RTX.  */

rtx
get_hard_reg_initial_reg (rtx reg)
{
  struct initial_value_struct *ivs = crtl->hard_reg_initial_vals;
  int i;

  if (ivs == 0)
    return NULL_RTX;

  for (i = 0; i < ivs->num_entries; i++)
    if (rtx_equal_p (ivs->entries[i].pseudo, reg))
      return ivs->entries[i].hard_reg;

  return NULL_RTX;
}

/* Make sure that there's a pseudo register of mode MODE that stores the
   initial value of hard register REGNO.  Return an rtx for such a pseudo.  */

rtx
get_hard_reg_initial_val (enum machine_mode mode, unsigned int regno)
{
  struct initial_value_struct *ivs;
  rtx rv;

  rv = has_hard_reg_initial_val (mode, regno);
  if (rv)
    return rv;

  ivs = crtl->hard_reg_initial_vals;
  if (ivs == 0)
    {
      ivs = ggc_alloc_initial_value_struct ();
      ivs->num_entries = 0;
      ivs->max_entries = 5;
      ivs->entries = ggc_alloc_vec_initial_value_pair (5);
      crtl->hard_reg_initial_vals = ivs;
    }

  if (ivs->num_entries >= ivs->max_entries)
    {
      ivs->max_entries += 5;
      ivs->entries = GGC_RESIZEVEC (initial_value_pair, ivs->entries,
                                    ivs->max_entries);
    }

  ivs->entries[ivs->num_entries].hard_reg = gen_rtx_REG (mode, regno);
  ivs->entries[ivs->num_entries].pseudo = gen_reg_rtx (mode);

  return ivs->entries[ivs->num_entries++].pseudo;
}

/* See if get_hard_reg_initial_val has been used to create a pseudo
   for the initial value of hard register REGNO in mode MODE.  Return
   the associated pseudo if so, otherwise return NULL.  */

rtx
has_hard_reg_initial_val (enum machine_mode mode, unsigned int regno)
{
  struct initial_value_struct *ivs;
  int i;

  ivs = crtl->hard_reg_initial_vals;
  if (ivs != 0)
    for (i = 0; i < ivs->num_entries; i++)
      if (GET_MODE (ivs->entries[i].hard_reg) == mode
          && REGNO (ivs->entries[i].hard_reg) == regno)
        return ivs->entries[i].pseudo;

  return NULL_RTX;
}

unsigned int
emit_initial_value_sets (void)
{
  struct initial_value_struct *ivs = crtl->hard_reg_initial_vals;
  int i;
  rtx seq;

  if (ivs == 0)
    return 0;

  start_sequence ();
  for (i = 0; i < ivs->num_entries; i++)
    emit_move_insn (ivs->entries[i].pseudo, ivs->entries[i].hard_reg);
  seq = get_insns ();
  end_sequence ();

  emit_insn_at_entry (seq);
  return 0;
}

/* Return the hardreg-pseudoreg initial values pair entry I and
   TRUE if I is a valid entry, or FALSE if I is not a valid entry.  */
bool
initial_value_entry (int i, rtx *hreg, rtx *preg)
{
  struct initial_value_struct *ivs = crtl->hard_reg_initial_vals;
  if (!ivs || i >= ivs->num_entries)
    return false;

  *hreg = ivs->entries[i].hard_reg;
  *preg = ivs->entries[i].pseudo;
  return true;
}
\f
/* These routines are responsible for converting virtual register references
   to the actual hard register references once RTL generation is complete.

   The following four variables are used for communication between the
   routines.  They contain the offsets of the virtual registers from their
   respective hard registers.  */

static int in_arg_offset;
static int var_offset;
static int dynamic_offset;
static int out_arg_offset;
static int cfa_offset;

/* In most machines, the stack pointer register is equivalent to the bottom
   of the stack.  */

#ifndef STACK_POINTER_OFFSET
#define STACK_POINTER_OFFSET    0
#endif

/* If not defined, pick an appropriate default for the offset of dynamically
   allocated memory depending on the value of ACCUMULATE_OUTGOING_ARGS,
   REG_PARM_STACK_SPACE, and OUTGOING_REG_PARM_STACK_SPACE.  */

#ifndef STACK_DYNAMIC_OFFSET

/* The bottom of the stack points to the actual arguments.  If
   REG_PARM_STACK_SPACE is defined, this includes the space for the register
   parameters.  However, if OUTGOING_REG_PARM_STACK space is not defined,
   stack space for register parameters is not pushed by the caller, but
   rather part of the fixed stack areas and hence not included in
   `crtl->outgoing_args_size'.  Nevertheless, we must allow
   for it when allocating stack dynamic objects.  */

#if defined(REG_PARM_STACK_SPACE)
#define STACK_DYNAMIC_OFFSET(FNDECL)    \
((ACCUMULATE_OUTGOING_ARGS                                                    \
  ? (crtl->outgoing_args_size                                 \
     + (OUTGOING_REG_PARM_STACK_SPACE ((!(FNDECL) ? NULL_TREE : TREE_TYPE (FNDECL))) ? 0 \
                                               : REG_PARM_STACK_SPACE (FNDECL))) \
  : 0) + (STACK_POINTER_OFFSET))
#else
#define STACK_DYNAMIC_OFFSET(FNDECL)    \
((ACCUMULATE_OUTGOING_ARGS ? crtl->outgoing_args_size : 0)            \
 + (STACK_POINTER_OFFSET))
#endif
#endif

\f
/* Given a piece of RTX and a pointer to a HOST_WIDE_INT, if the RTX
   is a virtual register, return the equivalent hard register and set the
   offset indirectly through the pointer.  Otherwise, return 0.  */

static rtx
instantiate_new_reg (rtx x, HOST_WIDE_INT *poffset)
{
  rtx new_rtx;
  HOST_WIDE_INT offset;

  if (x == virtual_incoming_args_rtx)
    {
      if (stack_realign_drap)
        {
          /* Replace virtual_incoming_args_rtx with internal arg
             pointer if DRAP is used to realign stack.  */
          new_rtx = crtl->args.internal_arg_pointer;
          offset = 0;
        }
      else
        new_rtx = arg_pointer_rtx, offset = in_arg_offset;
    }
  else if (x == virtual_stack_vars_rtx)
    new_rtx = frame_pointer_rtx, offset = var_offset;
  else if (x == virtual_stack_dynamic_rtx)
    new_rtx = stack_pointer_rtx, offset = dynamic_offset;
  else if (x == virtual_outgoing_args_rtx)
    new_rtx = stack_pointer_rtx, offset = out_arg_offset;
  else if (x == virtual_cfa_rtx)
    {
#ifdef FRAME_POINTER_CFA_OFFSET
      new_rtx = frame_pointer_rtx;
#else
      new_rtx = arg_pointer_rtx;
#endif
      offset = cfa_offset;
    }
  else if (x == virtual_preferred_stack_boundary_rtx)
    {
      new_rtx = GEN_INT (crtl->preferred_stack_boundary / BITS_PER_UNIT);
      offset = 0;
    }
  else
    return NULL_RTX;

  *poffset = offset;
  return new_rtx;
}

/* A subroutine of instantiate_virtual_regs, called via for_each_rtx.
   Instantiate any virtual registers present inside of *LOC.  The expression
   is simplified, as much as possible, but is not to be considered "valid"
   in any sense implied by the target.  If any change is made, set CHANGED
   to true.  */

static int
instantiate_virtual_regs_in_rtx (rtx *loc, void *data)
{
  HOST_WIDE_INT offset;
  bool *changed = (bool *) data;
  rtx x, new_rtx;

  x = *loc;
  if (x == 0)
    return 0;

  switch (GET_CODE (x))
    {
    case REG:
      new_rtx = instantiate_new_reg (x, &offset);
      if (new_rtx)
        {
          *loc = plus_constant (GET_MODE (x), new_rtx, offset);
          if (changed)
            *changed = true;
        }
      return -1;

    case PLUS:
      new_rtx = instantiate_new_reg (XEXP (x, 0), &offset);
      if (new_rtx)
        {
          new_rtx = plus_constant (GET_MODE (x), new_rtx, offset);
          *loc = simplify_gen_binary (PLUS, GET_MODE (x), new_rtx, XEXP (x, 1));
          if (changed)
            *changed = true;
          return -1;
        }

      /* FIXME -- from old code */
          /* If we have (plus (subreg (virtual-reg)) (const_int)), we know
             we can commute the PLUS and SUBREG because pointers into the
             frame are well-behaved.  */
      break;

    default:
      break;
    }

  return 0;
}

/* A subroutine of instantiate_virtual_regs_in_insn.  Return true if X
   matches the predicate for insn CODE operand OPERAND.  */

static int
safe_insn_predicate (int code, int operand, rtx x)
{
  return code < 0 || insn_operand_matches ((enum insn_code) code, operand, x);
}

/* A subroutine of instantiate_virtual_regs.  Instantiate any virtual
   registers present inside of insn.  The result will be a valid insn.  */

static void
instantiate_virtual_regs_in_insn (rtx insn)
{
  HOST_WIDE_INT offset;
  int insn_code, i;
  bool any_change = false;
  rtx set, new_rtx, x, seq;

  /* There are some special cases to be handled first.  */
  set = single_set (insn);
  if (set)
    {
      /* We're allowed to assign to a virtual register.  This is interpreted
         to mean that the underlying register gets assigned the inverse
         transformation.  This is used, for example, in the handling of
         non-local gotos.  */
      new_rtx = instantiate_new_reg (SET_DEST (set), &offset);
      if (new_rtx)
        {
          start_sequence ();

          for_each_rtx (&SET_SRC (set), instantiate_virtual_regs_in_rtx, NULL);
          x = simplify_gen_binary (PLUS, GET_MODE (new_rtx), SET_SRC (set),
                                   GEN_INT (-offset));
          x = force_operand (x, new_rtx);
          if (x != new_rtx)
            emit_move_insn (new_rtx, x);

          seq = get_insns ();
          end_sequence ();

          emit_insn_before (seq, insn);
          delete_insn (insn);
          return;
        }

      /* Handle a straight copy from a virtual register by generating a
         new add insn.  The difference between this and falling through
         to the generic case is avoiding a new pseudo and eliminating a
         move insn in the initial rtl stream.  */
      new_rtx = instantiate_new_reg (SET_SRC (set), &offset);
      if (new_rtx && offset != 0
          && REG_P (SET_DEST (set))
          && REGNO (SET_DEST (set)) > LAST_VIRTUAL_REGISTER)
        {
          start_sequence ();

          x = expand_simple_binop (GET_MODE (SET_DEST (set)), PLUS,
                                   new_rtx, GEN_INT (offset), SET_DEST (set),
                                   1, OPTAB_LIB_WIDEN);
          if (x != SET_DEST (set))
            emit_move_insn (SET_DEST (set), x);

          seq = get_insns ();
          end_sequence ();

          emit_insn_before (seq, insn);
          delete_insn (insn);
          return;
        }

      extract_insn (insn);
      insn_code = INSN_CODE (insn);

      /* Handle a plus involving a virtual register by determining if the
         operands remain valid if they're modified in place.  */
      if (GET_CODE (SET_SRC (set)) == PLUS
          && recog_data.n_operands >= 3
          && recog_data.operand_loc[1] == &XEXP (SET_SRC (set), 0)
          && recog_data.operand_loc[2] == &XEXP (SET_SRC (set), 1)
          && CONST_INT_P (recog_data.operand[2])
          && (new_rtx = instantiate_new_reg (recog_data.operand[1], &offset)))
        {
          offset += INTVAL (recog_data.operand[2]);

          /* If the sum is zero, then replace with a plain move.  */
          if (offset == 0
              && REG_P (SET_DEST (set))
              && REGNO (SET_DEST (set)) > LAST_VIRTUAL_REGISTER)
            {
              start_sequence ();
              emit_move_insn (SET_DEST (set), new_rtx);
              seq = get_insns ();
              end_sequence ();

              emit_insn_before (seq, insn);
              delete_insn (insn);
              return;
            }

          x = gen_int_mode (offset, recog_data.operand_mode[2]);

          /* Using validate_change and apply_change_group here leaves
             recog_data in an invalid state.  Since we know exactly what
             we want to check, do those two by hand.  */
          if (safe_insn_predicate (insn_code, 1, new_rtx)
              && safe_insn_predicate (insn_code, 2, x))
            {
              *recog_data.operand_loc[1] = recog_data.operand[1] = new_rtx;
              *recog_data.operand_loc[2] = recog_data.operand[2] = x;
              any_change = true;

              /* Fall through into the regular operand fixup loop in
                 order to take care of operands other than 1 and 2.  */
            }
        }
    }
  else
    {
      extract_insn (insn);
      insn_code = INSN_CODE (insn);
    }

  /* In the general case, we expect virtual registers to appear only in
     operands, and then only as either bare registers or inside memories.  */
  for (i = 0; i < recog_data.n_operands; ++i)
    {
      x = recog_data.operand[i];
      switch (GET_CODE (x))
        {
        case MEM:
          {
            rtx addr = XEXP (x, 0);
            bool changed = false;

            for_each_rtx (&addr, instantiate_virtual_regs_in_rtx, &changed);
            if (!changed)
              continue;

            start_sequence ();
            x = replace_equiv_address (x, addr);
            /* It may happen that the address with the virtual reg
               was valid (e.g. based on the virtual stack reg, which might
               be acceptable to the predicates with all offsets), whereas
               the address now isn't anymore, for instance when the address
               is still offsetted, but the base reg isn't virtual-stack-reg
               anymore.  Below we would do a force_reg on the whole operand,
               but this insn might actually only accept memory.  Hence,
               before doing that last resort, try to reload the address into
               a register, so this operand stays a MEM.  */
            if (!safe_insn_predicate (insn_code, i, x))
              {
                addr = force_reg (GET_MODE (addr), addr);
                x = replace_equiv_address (x, addr);
              }
            seq = get_insns ();
            end_sequence ();
            if (seq)
              emit_insn_before (seq, insn);
          }
          break;

        case REG:
          new_rtx = instantiate_new_reg (x, &offset);
          if (new_rtx == NULL)
            continue;
          if (offset == 0)
            x = new_rtx;
          else
            {
              start_sequence ();

              /* Careful, special mode predicates may have stuff in
                 insn_data[insn_code].operand[i].mode that isn't useful
                 to us for computing a new value.  */
              /* ??? Recognize address_operand and/or "p" constraints
                 to see if (plus new offset) is a valid before we put
                 this through expand_simple_binop.  */
              x = expand_simple_binop (GET_MODE (x), PLUS, new_rtx,
                                       GEN_INT (offset), NULL_RTX,
                                       1, OPTAB_LIB_WIDEN);
              seq = get_insns ();
              end_sequence ();
              emit_insn_before (seq, insn);
            }
          break;

        case SUBREG:
          new_rtx = instantiate_new_reg (SUBREG_REG (x), &offset);
          if (new_rtx == NULL)
            continue;
          if (offset != 0)
            {
              start_sequence ();
              new_rtx = expand_simple_binop (GET_MODE (new_rtx), PLUS, new_rtx,
                                         GEN_INT (offset), NULL_RTX,
                                         1, OPTAB_LIB_WIDEN);
              seq = get_insns ();
              end_sequence ();
              emit_insn_before (seq, insn);
            }
          x = simplify_gen_subreg (recog_data.operand_mode[i], new_rtx,
                                   GET_MODE (new_rtx), SUBREG_BYTE (x));
          gcc_assert (x);
          break;

        default:
          continue;
        }

      /* At this point, X contains the new value for the operand.
         Validate the new value vs the insn predicate.  Note that
         asm insns will have insn_code -1 here.  */
      if (!safe_insn_predicate (insn_code, i, x))
        {
          start_sequence ();
          if (REG_P (x))
            {
              gcc_assert (REGNO (x) <= LAST_VIRTUAL_REGISTER);
              x = copy_to_reg (x);
            }
          else
            x = force_reg (insn_data[insn_code].operand[i].mode, x);
          seq = get_insns ();
          end_sequence ();
          if (seq)
            emit_insn_before (seq, insn);
        }

      *recog_data.operand_loc[i] = recog_data.operand[i] = x;
      any_change = true;
    }

  if (any_change)
    {
      /* Propagate operand changes into the duplicates.  */
      for (i = 0; i < recog_data.n_dups; ++i)
        *recog_data.dup_loc[i]
          = copy_rtx (recog_data.operand[(unsigned)recog_data.dup_num[i]]);

      /* Force re-recognition of the instruction for validation.  */
      INSN_CODE (insn) = -1;
    }

  if (asm_noperands (PATTERN (insn)) >= 0)
    {
      if (!check_asm_operands (PATTERN (insn)))
        {
          error_for_asm (insn, "impossible constraint in %<asm%>");
          /* For asm goto, instead of fixing up all the edges
             just clear the template and clear input operands
             (asm goto doesn't have any output operands).  */
          if (JUMP_P (insn))
            {
              rtx asm_op = extract_asm_operands (PATTERN (insn));
              ASM_OPERANDS_TEMPLATE (asm_op) = ggc_strdup ("");
              ASM_OPERANDS_INPUT_VEC (asm_op) = rtvec_alloc (0);
              ASM_OPERANDS_INPUT_CONSTRAINT_VEC (asm_op) = rtvec_alloc (0);
            }
          else
            delete_insn (insn);
        }
    }
  else
    {
      if (recog_memoized (insn) < 0)
        fatal_insn_not_found (insn);
    }
}

/* Subroutine of instantiate_decls.  Given RTL representing a decl,
   do any instantiation required.  */

void
instantiate_decl_rtl (rtx x)
{
  rtx addr;

  if (x == 0)
    return;

  /* If this is a CONCAT, recurse for the pieces.  */
  if (GET_CODE (x) == CONCAT)
    {
      instantiate_decl_rtl (XEXP (x, 0));
      instantiate_decl_rtl (XEXP (x, 1));
      return;
    }

  /* If this is not a MEM, no need to do anything.  Similarly if the
     address is a constant or a register that is not a virtual register.  */
  if (!MEM_P (x))
    return;

  addr = XEXP (x, 0);
  if (CONSTANT_P (addr)
      || (REG_P (addr)
          && (REGNO (addr) < FIRST_VIRTUAL_REGISTER
              || REGNO (addr) > LAST_VIRTUAL_REGISTER)))
    return;

  for_each_rtx (&XEXP (x, 0), instantiate_virtual_regs_in_rtx, NULL);
}

/* Helper for instantiate_decls called via walk_tree: Process all decls
   in the given DECL_VALUE_EXPR.  */

static tree
instantiate_expr (tree *tp, int *walk_subtrees, void *data ATTRIBUTE_UNUSED)
{
  tree t = *tp;
  if (! EXPR_P (t))
    {
      *walk_subtrees = 0;
      if (DECL_P (t))
        {
          if (DECL_RTL_SET_P (t))
            instantiate_decl_rtl (DECL_RTL (t));
          if (TREE_CODE (t) == PARM_DECL && DECL_NAMELESS (t)
              && DECL_INCOMING_RTL (t))
            instantiate_decl_rtl (DECL_INCOMING_RTL (t));
          if ((TREE_CODE (t) == VAR_DECL
               || TREE_CODE (t) == RESULT_DECL)
              && DECL_HAS_VALUE_EXPR_P (t))
            {
              tree v = DECL_VALUE_EXPR (t);
              walk_tree (&v, instantiate_expr, NULL, NULL);
            }
        }
    }
  return NULL;
}

/* Subroutine of instantiate_decls: Process all decls in the given
   BLOCK node and all its subblocks.  */

static void
instantiate_decls_1 (tree let)
{
  tree t;

  for (t = BLOCK_VARS (let); t; t = DECL_CHAIN (t))
    {
      if (DECL_RTL_SET_P (t))
        instantiate_decl_rtl (DECL_RTL (t));
      if (TREE_CODE (t) == VAR_DECL && DECL_HAS_VALUE_EXPR_P (t))
        {
          tree v = DECL_VALUE_EXPR (t);
          walk_tree (&v, instantiate_expr, NULL, NULL);
        }
    }

  /* Process all subblocks.  */
  for (t = BLOCK_SUBBLOCKS (let); t; t = BLOCK_CHAIN (t))
    instantiate_decls_1 (t);
}

/* Scan all decls in FNDECL (both variables and parameters) and instantiate
   all virtual registers in their DECL_RTL's.  */

static void
instantiate_decls (tree fndecl)
{
  tree decl;
  unsigned ix;

  /* Process all parameters of the function.  */
  for (decl = DECL_ARGUMENTS (fndecl); decl; decl = DECL_CHAIN (decl))
    {
      instantiate_decl_rtl (DECL_RTL (decl));
      instantiate_decl_rtl (DECL_INCOMING_RTL (decl));
      if (DECL_HAS_VALUE_EXPR_P (decl))
        {
          tree v = DECL_VALUE_EXPR (decl);
          walk_tree (&v, instantiate_expr, NULL, NULL);
        }
    }

  if ((decl = DECL_RESULT (fndecl))
      && TREE_CODE (decl) == RESULT_DECL)
    {
      if (DECL_RTL_SET_P (decl))
        instantiate_decl_rtl (DECL_RTL (decl));
      if (DECL_HAS_VALUE_EXPR_P (decl))
        {
          tree v = DECL_VALUE_EXPR (decl);
          walk_tree (&v, instantiate_expr, NULL, NULL);
        }
    }

  /* Now process all variables defined in the function or its subblocks.  */
  instantiate_decls_1 (DECL_INITIAL (fndecl));

  FOR_EACH_LOCAL_DECL (cfun, ix, decl)
    if (DECL_RTL_SET_P (decl))
      instantiate_decl_rtl (DECL_RTL (decl));
  vec_free (cfun->local_decls);
}

/* Pass through the INSNS of function FNDECL and convert virtual register
   references to hard register references.  */

static unsigned int
instantiate_virtual_regs (void)
{
  rtx insn;

  /* Compute the offsets to use for this function.  */
  in_arg_offset = FIRST_PARM_OFFSET (current_function_decl);
  var_offset = STARTING_FRAME_OFFSET;
  dynamic_offset = STACK_DYNAMIC_OFFSET (current_function_decl);
  out_arg_offset = STACK_POINTER_OFFSET;
#ifdef FRAME_POINTER_CFA_OFFSET
  cfa_offset = FRAME_POINTER_CFA_OFFSET (current_function_decl);
#else
  cfa_offset = ARG_POINTER_CFA_OFFSET (current_function_decl);
#endif

  /* Initialize recognition, indicating that volatile is OK.  */
  init_recog ();

  /* Scan through all the insns, instantiating every virtual register still
     present.  */
  for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
    if (INSN_P (insn))
      {
        /* These patterns in the instruction stream can never be recognized.
           Fortunately, they shouldn't contain virtual registers either.  */
        if (GET_CODE (PATTERN (insn)) == USE
            || GET_CODE (PATTERN (insn)) == CLOBBER
            || GET_CODE (PATTERN (insn)) == ASM_INPUT)
          continue;
        else if (DEBUG_INSN_P (insn))
          for_each_rtx (&INSN_VAR_LOCATION (insn),
                        instantiate_virtual_regs_in_rtx, NULL);
        else
          instantiate_virtual_regs_in_insn (insn);

        if (INSN_DELETED_P (insn))
          continue;

        for_each_rtx (&REG_NOTES (insn), instantiate_virtual_regs_in_rtx, NULL);

        /* Instantiate any virtual registers in CALL_INSN_FUNCTION_USAGE.  */
        if (CALL_P (insn))
          for_each_rtx (&CALL_INSN_FUNCTION_USAGE (insn),
                        instantiate_virtual_regs_in_rtx, NULL);
      }

  /* Instantiate the virtual registers in the DECLs for debugging purposes.  */
  instantiate_decls (current_function_decl);

  targetm.instantiate_decls ();

  /* Indicate that, from now on, assign_stack_local should use
     frame_pointer_rtx.  */
  virtuals_instantiated = 1;

  return 0;
}

namespace {

const pass_data pass_data_instantiate_virtual_regs =
{
  RTL_PASS, /* type */
  "vregs", /* name */
  OPTGROUP_NONE, /* optinfo_flags */
  false, /* has_gate */
  true, /* has_execute */
  TV_NONE, /* tv_id */
  0, /* properties_required */
  0, /* properties_provided */
  0, /* properties_destroyed */
  0, /* todo_flags_start */
  0, /* todo_flags_finish */
};

class pass_instantiate_virtual_regs : public rtl_opt_pass
{
public:
  pass_instantiate_virtual_regs(gcc::context *ctxt)
    : rtl_opt_pass(pass_data_instantiate_virtual_regs, ctxt)
  {}

  /* opt_pass methods: */
  unsigned int execute () { return instantiate_virtual_regs (); }

}; // class pass_instantiate_virtual_regs

} // anon namespace

rtl_opt_pass *
make_pass_instantiate_virtual_regs (gcc::context *ctxt)
{
  return new pass_instantiate_virtual_regs (ctxt);
}

\f
/* Return 1 if EXP is an aggregate type (or a value with aggregate type).
   This means a type for which function calls must pass an address to the
   function or get an address back from the function.
   EXP may be a type node or an expression (whose type is tested).  */

int
aggregate_value_p (const_tree exp, const_tree fntype)
{
  const_tree type = (TYPE_P (exp)) ? exp : TREE_TYPE (exp);
  int i, regno, nregs;
  rtx reg;

  if (fntype)
    switch (TREE_CODE (fntype))
      {
      case CALL_EXPR:
        {
          tree fndecl = get_callee_fndecl (fntype);
          fntype = (fndecl
                    ? TREE_TYPE (fndecl)
                    : TREE_TYPE (TREE_TYPE (CALL_EXPR_FN (fntype))));
        }
        break;
      case FUNCTION_DECL:
        fntype = TREE_TYPE (fntype);
        break;
      case FUNCTION_TYPE:
      case METHOD_TYPE:
        break;
      case IDENTIFIER_NODE:
        fntype = NULL_TREE;
        break;
      default:
        /* We don't expect other tree types here.  */
        gcc_unreachable ();
      }

  if (VOID_TYPE_P (type))
    return 0;

  /* If a record should be passed the same as its first (and only) member
     don't pass it as an aggregate.  */
  if (TREE_CODE (type) == RECORD_TYPE && TYPE_TRANSPARENT_AGGR (type))
    return aggregate_value_p (first_field (type), fntype);

  /* If the front end has decided that this needs to be passed by
     reference, do so.  */
  if ((TREE_CODE (exp) == PARM_DECL || TREE_CODE (exp) == RESULT_DECL)
      && DECL_BY_REFERENCE (exp))
    return 1;

  /* Function types that are TREE_ADDRESSABLE force return in memory.  */
  if (fntype && TREE_ADDRESSABLE (fntype))
    return 1;

  /* Types that are TREE_ADDRESSABLE must be constructed in memory,
     and thus can't be returned in registers.  */
  if (TREE_ADDRESSABLE (type))
    return 1;

  if (flag_pcc_struct_return && AGGREGATE_TYPE_P (type))
    return 1;

  if (targetm.calls.return_in_memory (type, fntype))
    return 1;

  /* Make sure we have suitable call-clobbered regs to return
     the value in; if not, we must return it in memory.  */
  reg = hard_function_value (type, 0, fntype, 0);

  /* If we have something other than a REG (e.g. a PARALLEL), then assume
     it is OK.  */
  if (!REG_P (reg))
    return 0;

  regno = REGNO (reg);
  nregs = hard_regno_nregs[regno][TYPE_MODE (type)];
  for (i = 0; i < nregs; i++)
    if (! call_used_regs[regno + i])
      return 1;

  return 0;
}
\f
/* Return true if we should assign DECL a pseudo register; false if it
   should live on the local stack.  */

bool
use_register_for_decl (const_tree decl)
{
  if (!targetm.calls.allocate_stack_slots_for_args())
    return true;

  /* Honor volatile.  */
  if (TREE_SIDE_EFFECTS (decl))
    return false;

  /* Honor addressability.  */
  if (TREE_ADDRESSABLE (decl))
    return false;

  /* Only register-like things go in registers.  */
  if (DECL_MODE (decl) == BLKmode)
    return false;

  /* If -ffloat-store specified, don't put explicit float variables
     into registers.  */
  /* ??? This should be checked after DECL_ARTIFICIAL, but tree-ssa
     propagates values across these stores, and it probably shouldn't.  */
  if (flag_float_store && FLOAT_TYPE_P (TREE_TYPE (decl)))
    return false;

  /* If we're not interested in tracking debugging information for
     this decl, then we can certainly put it in a register.  */
  if (DECL_IGNORED_P (decl))
    return true;

  if (optimize)
    return true;

  if (!DECL_REGISTER (decl))
    return false;

  switch (TREE_CODE (TREE_TYPE (decl)))
    {
    case RECORD_TYPE:
    case UNION_TYPE:
    case QUAL_UNION_TYPE:
      /* When not optimizing, disregard register keyword for variables with
         types containing methods, otherwise the methods won't be callable
         from the debugger.  */
      if (TYPE_METHODS (TREE_TYPE (decl)))
        return false;
      break;
    default:
      break;
    }

  return true;
}

/* Return true if TYPE should be passed by invisible reference.  */

bool
pass_by_reference (CUMULATIVE_ARGS *ca, enum machine_mode mode,
                   tree type, bool named_arg)
{
  if (type)
    {
      /* If this type contains non-trivial constructors, then it is
         forbidden for the middle-end to create any new copies.  */
      if (TREE_ADDRESSABLE (type))
        return true;

      /* GCC post 3.4 passes *all* variable sized types by reference.  */
      if (!TYPE_SIZE (type) || TREE_CODE (TYPE_SIZE (type)) != INTEGER_CST)
        return true;

      /* If a record type should be passed the same as its first (and only)
         member, use the type and mode of that member.  */
      if (TREE_CODE (type) == RECORD_TYPE && TYPE_TRANSPARENT_AGGR (type))
        {
          type = TREE_TYPE (first_field (type));
          mode = TYPE_MODE (type);
        }
    }

  return targetm.calls.pass_by_reference (pack_cumulative_args (ca), mode,
                                          type, named_arg);
}

/* Return true if TYPE, which is passed by reference, should be callee
   copied instead of caller copied.  */

bool
reference_callee_copied (CUMULATIVE_ARGS *ca, enum machine_mode mode,
                         tree type, bool named_arg)
{
  if (type && TREE_ADDRESSABLE (type))
    return false;
  return targetm.calls.callee_copies (pack_cumulative_args (ca), mode, type,
                                      named_arg);
}

/* Structures to communicate between the subroutines of assign_parms.
   The first holds data persistent across all parameters, the second
   is cleared out for each parameter.  */

struct assign_parm_data_all
{
  /* When INIT_CUMULATIVE_ARGS gets revamped, allocating CUMULATIVE_ARGS
     should become a job of the target or otherwise encapsulated.  */
  CUMULATIVE_ARGS args_so_far_v;
  cumulative_args_t args_so_far;
  struct args_size stack_args_size;
  tree function_result_decl;
  tree orig_fnargs;
  rtx first_conversion_insn;
  rtx last_conversion_insn;
  HOST_WIDE_INT pretend_args_size;
  HOST_WIDE_INT extra_pretend_bytes;
  int reg_parm_stack_space;
};

struct assign_parm_data_one
{
  tree nominal_type;
  tree passed_type;
  rtx entry_parm;
  rtx stack_parm;
  enum machine_mode nominal_mode;
  enum machine_mode passed_mode;
  enum machine_mode promoted_mode;
  struct locate_and_pad_arg_data locate;
  int partial;
  BOOL_BITFIELD named_arg : 1;
  BOOL_BITFIELD passed_pointer : 1;
  BOOL_BITFIELD on_stack : 1;
  BOOL_BITFIELD loaded_in_reg : 1;
};

/* A subroutine of assign_parms.  Initialize ALL.  */

static void
assign_parms_initialize_all (struct assign_parm_data_all *all)
{
  tree fntype ATTRIBUTE_UNUSED;

  memset (all, 0, sizeof (*all));

  fntype = TREE_TYPE (current_function_decl);

#ifdef INIT_CUMULATIVE_INCOMING_ARGS
  INIT_CUMULATIVE_INCOMING_ARGS (all->args_so_far_v, fntype, NULL_RTX);
#else
  INIT_CUMULATIVE_ARGS (all->args_so_far_v, fntype, NULL_RTX,
                        current_function_decl, -1);
#endif
  all->args_so_far = pack_cumulative_args (&all->args_so_far_v);

#ifdef REG_PARM_STACK_SPACE
  all->reg_parm_stack_space = REG_PARM_STACK_SPACE (current_function_decl);
#endif
}

/* If ARGS contains entries with complex types, split the entry into two
   entries of the component type.  Return a new list of substitutions are
   needed, else the old list.  */

static void
split_complex_args (vec<tree> *args)
{
  unsigned i;
  tree p;

  FOR_EACH_VEC_ELT (*args, i, p)
    {
      tree type = TREE_TYPE (p);
      if (TREE_CODE (type) == COMPLEX_TYPE
          && targetm.calls.split_complex_arg (type))
        {
          tree decl;
          tree subtype = TREE_TYPE (type);
          bool addressable = TREE_ADDRESSABLE (p);

          /* Rewrite the PARM_DECL's type with its component.  */
          p = copy_node (p);
          TREE_TYPE (p) = subtype;
          DECL_ARG_TYPE (p) = TREE_TYPE (DECL_ARG_TYPE (p));
          DECL_MODE (p) = VOIDmode;
          DECL_SIZE (p) = NULL;
          DECL_SIZE_UNIT (p) = NULL;
          /* If this arg must go in memory, put it in a pseudo here.
             We can't allow it to go in memory as per normal parms,
             because the usual place might not have the imag part
             adjacent to the real part.  */
          DECL_ARTIFICIAL (p) = addressable;
          DECL_IGNORED_P (p) = addressable;
          TREE_ADDRESSABLE (p) = 0;
          layout_decl (p, 0);
          (*args)[i] = p;

          /* Build a second synthetic decl.  */
          decl = build_decl (EXPR_LOCATION (p),
                             PARM_DECL, NULL_TREE, subtype);
          DECL_ARG_TYPE (decl) = DECL_ARG_TYPE (p);
          DECL_ARTIFICIAL (decl) = addressable;
          DECL_IGNORED_P (decl) = addressable;
          layout_decl (decl, 0);
          args->safe_insert (++i, decl);
        }
    }
}

/* A subroutine of assign_parms.  Adjust the parameter list to incorporate
   the hidden struct return argument, and (abi willing) complex args.
   Return the new parameter list.  */

static vec<tree> 
assign_parms_augmented_arg_list (struct assign_parm_data_all *all)
{
  tree fndecl = current_function_decl;
  tree fntype = TREE_TYPE (fndecl);
  vec<tree> fnargs = vNULL;
  tree arg;

  for (arg = DECL_ARGUMENTS (fndecl); arg; arg = DECL_CHAIN (arg))
    fnargs.safe_push (arg);

  all->orig_fnargs = DECL_ARGUMENTS (fndecl);

  /* If struct value address is treated as the first argument, make it so.  */
  if (aggregate_value_p (DECL_RESULT (fndecl), fndecl)
      && ! cfun->returns_pcc_struct
      && targetm.calls.struct_value_rtx (TREE_TYPE (fndecl), 1) == 0)
    {
      tree type = build_pointer_type (TREE_TYPE (fntype));
      tree decl;

      decl = build_decl (DECL_SOURCE_LOCATION (fndecl),
                         PARM_DECL, get_identifier (".result_ptr"), type);
      DECL_ARG_TYPE (decl) = type;
      DECL_ARTIFICIAL (decl) = 1;
      DECL_NAMELESS (decl) = 1;
      TREE_CONSTANT (decl) = 1;

      DECL_CHAIN (decl) = all->orig_fnargs;
      all->orig_fnargs = decl;
      fnargs.safe_insert (0, decl);

      all->function_result_decl = decl;
    }

  /* If the target wants to split complex arguments into scalars, do so.  */
  if (targetm.calls.split_complex_arg)
    split_complex_args (&fnargs);

  return fnargs;
}

/* A subroutine of assign_parms.  Examine PARM and pull out type and mode
   data for the parameter.  Incorporate ABI specifics such as pass-by-
   reference and type promotion.  */

static void
assign_parm_find_data_types (struct assign_parm_data_all *all, tree parm,
                             struct assign_parm_data_one *data)
{
  tree nominal_type, passed_type;
  enum machine_mode nominal_mode, passed_mode, promoted_mode;
  int unsignedp;

  memset (data, 0, sizeof (*data));

  /* NAMED_ARG is a misnomer.  We really mean 'non-variadic'. */
  if (!cfun->stdarg)
    data->named_arg = 1;  /* No variadic parms.  */
  else if (DECL_CHAIN (parm))
    data->named_arg = 1;  /* Not the last non-variadic parm. */
  else if (targetm.calls.strict_argument_naming (all->args_so_far))
    data->named_arg = 1;  /* Only variadic ones are unnamed.  */
  else
    data->named_arg = 0;  /* Treat as variadic.  */

  nominal_type = TREE_TYPE (parm);
  passed_type = DECL_ARG_TYPE (parm);

  /* Look out for errors propagating this far.  Also, if the parameter's
     type is void then its value doesn't matter.  */
  if (TREE_TYPE (parm) == error_mark_node
      /* This can happen after weird syntax errors
         or if an enum type is defined among the parms.  */
      || TREE_CODE (parm) != PARM_DECL
      || passed_type == NULL
      || VOID_TYPE_P (nominal_type))
    {
      nominal_type = passed_type = void_type_node;
      nominal_mode = passed_mode = promoted_mode = VOIDmode;
      goto egress;
    }

  /* Find mode of arg as it is passed, and mode of arg as it should be
     during execution of this function.  */
  passed_mode = TYPE_MODE (passed_type);
  nominal_mode = TYPE_MODE (nominal_type);

  /* If the parm is to be passed as a transparent union or record, use the
     type of the first field for the tests below.  We have already verified
     that the modes are the same.  */
  if ((TREE_CODE (passed_type) == UNION_TYPE
       || TREE_CODE (passed_type) == RECORD_TYPE)
      && TYPE_TRANSPARENT_AGGR (passed_type))
    passed_type = TREE_TYPE (first_field (passed_type));

  /* See if this arg was passed by invisible reference.  */
  if (pass_by_reference (&all->args_so_far_v, passed_mode,
                         passed_type, data->named_arg))
    {
      passed_type = nominal_type = build_pointer_type (passed_type);
      data->passed_pointer = true;
      passed_mode = nominal_mode = TYPE_MODE (nominal_type);
    }

  /* Find mode as it is passed by the ABI.  */
  unsignedp = TYPE_UNSIGNED (passed_type);
  promoted_mode = promote_function_mode (passed_type, passed_mode, &unsignedp,
                                         TREE_TYPE (current_function_decl), 0);

 egress:
  data->nominal_type = nominal_type;
  data->passed_type = passed_type;
  data->nominal_mode = nominal_mode;
  data->passed_mode = passed_mode;
  data->promoted_mode = promoted_mode;
}

/* A subroutine of assign_parms.  Invoke setup_incoming_varargs.  */

static void
assign_parms_setup_varargs (struct assign_parm_data_all *all,
                            struct assign_parm_data_one *data, bool no_rtl)
{
  int varargs_pretend_bytes = 0;

  targetm.calls.setup_incoming_varargs (all->args_so_far,
                                        data->promoted_mode,
                                        data->passed_type,
                                        &varargs_pretend_bytes, no_rtl);

  /* If the back-end has requested extra stack space, record how much is
     needed.  Do not change pretend_args_size otherwise since it may be
     nonzero from an earlier partial argument.  */
  if (varargs_pretend_bytes > 0)
    all->pretend_args_size = varargs_pretend_bytes;
}

/* A subroutine of assign_parms.  Set DATA->ENTRY_PARM corresponding to
   the incoming location of the current parameter.  */

static void
assign_parm_find_entry_rtl (struct assign_parm_data_all *all,
                            struct assign_parm_data_one *data)
{
  HOST_WIDE_INT pretend_bytes = 0;
  rtx entry_parm;
  bool in_regs;

  if (data->promoted_mode == VOIDmode)
    {
      data->entry_parm = data->stack_parm = const0_rtx;
      return;
    }

  entry_parm = targetm.calls.function_incoming_arg (all->args_so_far,
                                                    data->promoted_mode,
                                                    data->passed_type,
                                                    data->named_arg);

  if (entry_parm == 0)
    data->promoted_mode = data->passed_mode;

  /* Determine parm's home in the stack, in case it arrives in the stack
     or we should pretend it did.  Compute the stack position and rtx where
     the argument arrives and its size.

     There is one complexity here:  If this was a parameter that would
     have been passed in registers, but wasn't only because it is
     __builtin_va_alist, we want locate_and_pad_parm to treat it as if
     it came in a register so that REG_PARM_STACK_SPACE isn't skipped.
     In this case, we call FUNCTION_ARG with NAMED set to 1 instead of 0
     as it was the previous time.  */
  in_regs = entry_parm != 0;
#ifdef STACK_PARMS_IN_REG_PARM_AREA
  in_regs = true;
#endif
  if (!in_regs && !data->named_arg)
    {
      if (targetm.calls.pretend_outgoing_varargs_named (all->args_so_far))
        {
          rtx tem;
          tem = targetm.calls.function_incoming_arg (all->args_so_far,
                                                     data->promoted_mode,
                                                     data->passed_type, true);
          in_regs = tem != NULL;
        }
    }

  /* If this parameter was passed both in registers and in the stack, use
     the copy on the stack.  */
  if (targetm.calls.must_pass_in_stack (data->promoted_mode,
                                        data->passed_type))
    entry_parm = 0;

  if (entry_parm)
    {
      int partial;

      partial = targetm.calls.arg_partial_bytes (all->args_so_far,
                                                 data->promoted_mode,
                                                 data->passed_type,
                                                 data->named_arg);
      data->partial = partial;

      /* The caller might already have allocated stack space for the
         register parameters.  */
      if (partial != 0 && all->reg_parm_stack_space == 0)
        {
          /* Part of this argument is passed in registers and part
             is passed on the stack.  Ask the prologue code to extend
             the stack part so that we can recreate the full value.

             PRETEND_BYTES is the size of the registers we need to store.
             CURRENT_FUNCTION_PRETEND_ARGS_SIZE is the amount of extra
             stack space that the prologue should allocate.

             Internally, gcc assumes that the argument pointer is aligned
             to STACK_BOUNDARY bits.  This is used both for alignment
             optimizations (see init_emit) and to locate arguments that are
             aligned to more than PARM_BOUNDARY bits.  We must preserve this
             invariant by rounding CURRENT_FUNCTION_PRETEND_ARGS_SIZE up to
             a stack boundary.  */

          /* We assume at most one partial arg, and it must be the first
             argument on the stack.  */
          gcc_assert (!all->extra_pretend_bytes && !all->pretend_args_size);

          pretend_bytes = partial;
          all->pretend_args_size = CEIL_ROUND (pretend_bytes, STACK_BYTES);

          /* We want to align relative to the actual stack pointer, so
             don't include this in the stack size until later.  */
          all->extra_pretend_bytes = all->pretend_args_size;
        }
    }

  locate_and_pad_parm (data->promoted_mode, data->passed_type, in_regs,
                       entry_parm ? data->partial : 0, current_function_decl,
                       &all->stack_args_size, &data->locate);

  /* Update parm_stack_boundary if this parameter is passed in the
     stack.  */
  if (!in_regs && crtl->parm_stack_boundary < data->locate.boundary)
    crtl->parm_stack_boundary = data->locate.boundary;

  /* Adjust offsets to include the pretend args.  */
  pretend_bytes = all->extra_pretend_bytes - pretend_bytes;
  data->locate.slot_offset.constant += pretend_bytes;
  data->locate.offset.constant += pretend_bytes;

  data->entry_parm = entry_parm;
}

/* A subroutine of assign_parms.  If there is actually space on the stack
   for this parm, count it in stack_args_size and return true.  */

static bool
assign_parm_is_stack_parm (struct assign_parm_data_all *all,
                           struct assign_parm_data_one *data)
{
  /* Trivially true if we've no incoming register.  */
  if (data->entry_parm == NULL)
    ;
  /* Also true if we're partially in registers and partially not,
     since we've arranged to drop the entire argument on the stack.  */
  else if (data->partial != 0)
    ;
  /* Also true if the target says that it's passed in both registers
     and on the stack.  */
  else if (GET_CODE (data->entry_parm) == PARALLEL
           && XEXP (XVECEXP (data->entry_parm, 0, 0), 0) == NULL_RTX)
    ;
  /* Also true if the target says that there's stack allocated for
     all register parameters.  */
  else if (all->reg_parm_stack_space > 0)
    ;
  /* Otherwise, no, this parameter has no ABI defined stack slot.  */
  else
    return false;

  all->stack_args_size.constant += data->locate.size.constant;
  if (data->locate.size.var)
    ADD_PARM_SIZE (all->stack_args_size, data->locate.size.var);

  return true;
}

/* A subroutine of assign_parms.  Given that this parameter is allocated
   stack space by the ABI, find it.  */

static void
assign_parm_find_stack_rtl (tree parm, struct assign_parm_data_one *data)
{
  rtx offset_rtx, stack_parm;
  unsigned int align, boundary;

  /* If we're passing this arg using a reg, make its stack home the
     aligned stack slot.  */
  if (data->entry_parm)
    offset_rtx = ARGS_SIZE_RTX (data->locate.slot_offset);
  else
    offset_rtx = ARGS_SIZE_RTX (data->locate.offset);

  stack_parm = crtl->args.internal_arg_pointer;
  if (offset_rtx != const0_rtx)
    stack_parm = gen_rtx_PLUS (Pmode, stack_parm, offset_rtx);
  stack_parm = gen_rtx_MEM (data->promoted_mode, stack_parm);

  if (!data->passed_pointer)
    {
      set_mem_attributes (stack_parm, parm, 1);
      /* set_mem_attributes could set MEM_SIZE to the passed mode's size,
         while promoted mode's size is needed.  */
      if (data->promoted_mode != BLKmode
          && data->promoted_mode != DECL_MODE (parm))
        {
          set_mem_size (stack_parm, GET_MODE_SIZE (data->promoted_mode));
          if (MEM_EXPR (stack_parm) && MEM_OFFSET_KNOWN_P (stack_parm))
            {
              int offset = subreg_lowpart_offset (DECL_MODE (parm),
                                                  data->promoted_mode);
              if (offset)
                set_mem_offset (stack_parm, MEM_OFFSET (stack_parm) - offset);
            }
        }
    }

  boundary = data->locate.boundary;
  align = BITS_PER_UNIT;

  /* If we're padding upward, we know that the alignment of the slot
     is TARGET_FUNCTION_ARG_BOUNDARY.  If we're using slot_offset, we're
     intentionally forcing upward padding.  Otherwise we have to come
     up with a guess at the alignment based on OFFSET_RTX.  */
  if (data->locate.where_pad != downward || data->entry_parm)
    align = boundary;
  else if (CONST_INT_P (offset_rtx))
    {
      align = INTVAL (offset_rtx) * BITS_PER_UNIT | boundary;
      align = align & -align;
    }
  set_mem_align (stack_parm, align);

  if (data->entry_parm)
    set_reg_attrs_for_parm (data->entry_parm, stack_parm);

  data->stack_parm = stack_parm;
}

/* A subroutine of assign_parms.  Adjust DATA->ENTRY_RTL such that it's
   always valid and contiguous.  */

static void
assign_parm_adjust_entry_rtl (struct assign_parm_data_one *data)
{
  rtx entry_parm = data->entry_parm;
  rtx stack_parm = data->stack_parm;

  /* If this parm was passed part in regs and part in memory, pretend it
     arrived entirely in memory by pushing the register-part onto the stack.
     In the special case of a DImode or DFmode that is split, we could put
     it together in a pseudoreg directly, but for now that's not worth
     bothering with.  */
  if (data->partial != 0)
    {
      /* Handle calls that pass values in multiple non-contiguous
         locations.  The Irix 6 ABI has examples of this.  */
      if (GET_CODE (entry_parm) == PARALLEL)
        emit_group_store (validize_mem (stack_parm), entry_parm,
                          data->passed_type,
                          int_size_in_bytes (data->passed_type));
      else
        {
          gcc_assert (data->partial % UNITS_PER_WORD == 0);
          move_block_from_reg (REGNO (entry_parm), validize_mem (stack_parm),
                               data->partial / UNITS_PER_WORD);
        }

      entry_parm = stack_parm;
    }

  /* If we didn't decide this parm came in a register, by default it came
     on the stack.  */
  else if (entry_parm == NULL)
    entry_parm = stack_parm;

  /* When an argument is passed in multiple locations, we can't make use
     of this information, but we can save some copying if the whole argument
     is passed in a single register.  */
  else if (GET_CODE (entry_parm) == PARALLEL
           && data->nominal_mode != BLKmode
           && data->passed_mode != BLKmode)
    {
      size_t i, len = XVECLEN (entry_parm, 0);

      for (i = 0; i < len; i++)
        if (XEXP (XVECEXP (entry_parm, 0, i), 0) != NULL_RTX
            && REG_P (XEXP (XVECEXP (entry_parm, 0, i), 0))
            && (GET_MODE (XEXP (XVECEXP (entry_parm, 0, i), 0))
                == data->passed_mode)
            && INTVAL (XEXP (XVECEXP (entry_parm, 0, i), 1)) == 0)
          {
            entry_parm = XEXP (XVECEXP (entry_parm, 0, i), 0);
            break;
          }
    }

  data->entry_parm = entry_parm;
}

/* A subroutine of assign_parms.  Reconstitute any values which were
   passed in multiple registers and would fit in a single register.  */

static void
assign_parm_remove_parallels (struct assign_parm_data_one *data)
{
  rtx entry_parm = data->entry_parm;

  /* Convert the PARALLEL to a REG of the same mode as the parallel.
     This can be done with register operations rather than on the
     stack, even if we will store the reconstituted parameter on the
     stack later.  */
  if (GET_CODE (entry_parm) == PARALLEL && GET_MODE (entry_parm) != BLKmode)
    {
      rtx parmreg = gen_reg_rtx (GET_MODE (entry_parm));
      emit_group_store (parmreg, entry_parm, data->passed_type,
                        GET_MODE_SIZE (GET_MODE (entry_parm)));
      entry_parm = parmreg;
    }

  data->entry_parm = entry_parm;
}

/* A subroutine of assign_parms.  Adjust DATA->STACK_RTL such that it's
   always valid and properly aligned.  */

static void
assign_parm_adjust_stack_rtl (struct assign_parm_data_one *data)
{
  rtx stack_parm = data->stack_parm;

  /* If we can't trust the parm stack slot to be aligned enough for its
     ultimate type, don't use that slot after entry.  We'll make another
     stack slot, if we need one.  */
  if (stack_parm
      && ((STRICT_ALIGNMENT
           && GET_MODE_ALIGNMENT (data->nominal_mode) > MEM_ALIGN (stack_parm))
          || (data->nominal_type
              && TYPE_ALIGN (data->nominal_type) > MEM_ALIGN (stack_parm)
              && MEM_ALIGN (stack_parm) < PREFERRED_STACK_BOUNDARY)))
    stack_parm = NULL;

  /* If parm was passed in memory, and we need to convert it on entry,
     don't store it back in that same slot.  */
  else if (data->entry_parm == stack_parm
           && data->nominal_mode != BLKmode
           && data->nominal_mode != data->passed_mode)
    stack_parm = NULL;

  /* If stack protection is in effect for this function, don't leave any
     pointers in their passed stack slots.  */
  else if (crtl->stack_protect_guard
           && (flag_stack_protect == 2
               || data->passed_pointer
               || POINTER_TYPE_P (data->nominal_type)))
    stack_parm = NULL;

  data->stack_parm = stack_parm;
}

/* A subroutine of assign_parms.  Return true if the current parameter
   should be stored as a BLKmode in the current frame.  */

static bool
assign_parm_setup_block_p (struct assign_parm_data_one *data)
{
  if (data->nominal_mode == BLKmode)
    return true;
  if (GET_MODE (data->entry_parm) == BLKmode)
    return true;

#ifdef BLOCK_REG_PADDING
  /* Only assign_parm_setup_block knows how to deal with register arguments
     that are padded at the least significant end.  */
  if (REG_P (data->entry_parm)
      && GET_MODE_SIZE (data->promoted_mode) < UNITS_PER_WORD
      && (BLOCK_REG_PADDING (data->passed_mode, data->passed_type, 1)
          == (BYTES_BIG_ENDIAN ? upward : downward)))
    return true;
#endif

  return false;
}

/* A subroutine of assign_parms.  Arrange for the parameter to be
   present and valid in DATA->STACK_RTL.  */

static void
assign_parm_setup_block (struct assign_parm_data_all *all,
                         tree parm, struct assign_parm_data_one *data)
{
  rtx entry_parm = data->entry_parm;
  rtx stack_parm = data->stack_parm;
  HOST_WIDE_INT size;
  HOST_WIDE_INT size_stored;

  if (GET_CODE (entry_parm) == PARALLEL)
    entry_parm = emit_group_move_into_temps (entry_parm);

  size = int_size_in_bytes (data->passed_type);
  size_stored = CEIL_ROUND (size, UNITS_PER_WORD);
  if (stack_parm == 0)
    {
      DECL_ALIGN (parm) = MAX (DECL_ALIGN (parm), BITS_PER_WORD);
      stack_parm = assign_stack_local (BLKmode, size_stored,
                                       DECL_ALIGN (parm));
      if (GET_MODE_SIZE (GET_MODE (entry_parm)) == size)
        PUT_MODE (stack_parm, GET_MODE (entry_parm));
      set_mem_attributes (stack_parm, parm, 1);
    }

  /* If a BLKmode arrives in registers, copy it to a stack slot.  Handle
     calls that pass values in multiple non-contiguous locations.  */
  if (REG_P (entry_parm) || GET_CODE (entry_parm) == PARALLEL)
    {
      rtx mem;

      /* Note that we will be storing an integral number of words.
         So we have to be careful to ensure that we allocate an
         integral number of words.  We do this above when we call
         assign_stack_local if space was not allocated in the argument
         list.  If it was, this will not work if PARM_BOUNDARY is not
         a multiple of BITS_PER_WORD.  It isn't clear how to fix this
         if it becomes a problem.  Exception is when BLKmode arrives
         with arguments not conforming to word_mode.  */

      if (data->stack_parm == 0)
        ;
      else if (GET_CODE (entry_parm) == PARALLEL)
        ;
      else
        gcc_assert (!size || !(PARM_BOUNDARY % BITS_PER_WORD));

      mem = validize_mem (stack_parm);

      /* Handle values in multiple non-contiguous locations.  */
      if (GET_CODE (entry_parm) == PARALLEL)
        {
          push_to_sequence2 (all->first_conversion_insn,
                             all->last_conversion_insn);
          emit_group_store (mem, entry_parm, data->passed_type, size);
          all->first_conversion_insn = get_insns ();
          all->last_conversion_insn = get_last_insn ();
          end_sequence ();
        }

      else if (size == 0)
        ;

      /* If SIZE is that of a mode no bigger than a word, just use
         that mode's store operation.  */
      else if (size <= UNITS_PER_WORD)
        {
          enum machine_mode mode
            = mode_for_size (size * BITS_PER_UNIT, MODE_INT, 0);

          if (mode != BLKmode
#ifdef BLOCK_REG_PADDING
              && (size == UNITS_PER_WORD
                  || (BLOCK_REG_PADDING (mode, data->passed_type, 1)
                      != (BYTES_BIG_ENDIAN ? upward : downward)))
#endif
              )
            {
              rtx reg;

              /* We are really truncating a word_mode value containing
                 SIZE bytes into a value of mode MODE.  If such an
                 operation requires no actual instructions, we can refer
                 to the value directly in mode MODE, otherwise we must
                 start with the register in word_mode and explicitly
                 convert it.  */
              if (TRULY_NOOP_TRUNCATION (size * BITS_PER_UNIT, BITS_PER_WORD))
                reg = gen_rtx_REG (mode, REGNO (entry_parm));
              else
                {
                  reg = gen_rtx_REG (word_mode, REGNO (entry_parm));
                  reg = convert_to_mode (mode, copy_to_reg (reg), 1);
                }
              emit_move_insn (change_address (mem, mode, 0), reg);
            }

          /* Blocks smaller than a word on a BYTES_BIG_ENDIAN
             machine must be aligned to the left before storing
             to memory.  Note that the previous test doesn't
             handle all cases (e.g. SIZE == 3).  */
          else if (size != UNITS_PER_WORD
#ifdef BLOCK_REG_PADDING
                   && (BLOCK_REG_PADDING (mode, data->passed_type, 1)
                       == downward)
#else
                   && BYTES_BIG_ENDIAN
#endif
                   )
            {
              rtx tem, x;
              int by = (UNITS_PER_WORD - size) * BITS_PER_UNIT;
              rtx reg = gen_rtx_REG (word_mode, REGNO (entry_parm));

              x = expand_shift (LSHIFT_EXPR, word_mode, reg, by, NULL_RTX, 1);
              tem = change_address (mem, word_mode, 0);
              emit_move_insn (tem, x);
            }
          else
            move_block_from_reg (REGNO (entry_parm), mem,
                                 size_stored / UNITS_PER_WORD);
        }
      else
        move_block_from_reg (REGNO (entry_parm), mem,
                             size_stored / UNITS_PER_WORD);
    }
  else if (data->stack_parm == 0)
    {
      push_to_sequence2 (all->first_conversion_insn, all->last_conversion_insn);
      emit_block_move (stack_parm, data->entry_parm, GEN_INT (size),
                       BLOCK_OP_NORMAL);
      all->first_conversion_insn = get_insns ();
      all->last_conversion_insn = get_last_insn ();
      end_sequence ();
    }

  data->stack_parm = stack_parm;
  SET_DECL_RTL (parm, stack_parm);
}

/* A subroutine of assign_parms.  Allocate a pseudo to hold the current
   parameter.  Get it there.  Perform all ABI specified conversions.  */

static void
assign_parm_setup_reg (struct assign_parm_data_all *all, tree parm,
                       struct assign_parm_data_one *data)
{
  rtx parmreg, validated_mem;
  rtx equiv_stack_parm;
  enum machine_mode promoted_nominal_mode;
  int unsignedp = TYPE_UNSIGNED (TREE_TYPE (parm));
  bool did_conversion = false;
  bool need_conversion, moved;

  /* Store the parm in a pseudoregister during the function, but we may
     need to do it in a wider mode.  Using 2 here makes the result
     consistent with promote_decl_mode and thus expand_expr_real_1.  */
  promoted_nominal_mode
    = promote_function_mode (data->nominal_type, data->nominal_mode, &unsignedp,
                             TREE_TYPE (current_function_decl), 2);

  parmreg = gen_reg_rtx (promoted_nominal_mode);

  if (!DECL_ARTIFICIAL (parm))
    mark_user_reg (parmreg);

  /* If this was an item that we received a pointer to,
     set DECL_RTL appropriately.  */
  if (data->passed_pointer)
    {
      rtx x = gen_rtx_MEM (TYPE_MODE (TREE_TYPE (data->passed_type)), parmreg);
      set_mem_attributes (x, parm, 1);
      SET_DECL_RTL (parm, x);
    }
  else
    SET_DECL_RTL (parm, parmreg);

  assign_parm_remove_parallels (data);

  /* Copy the value into the register, thus bridging between
     assign_parm_find_data_types and expand_expr_real_1.  */

  equiv_stack_parm = data->stack_parm;
  validated_mem = validize_mem (data->entry_parm);

  need_conversion = (data->nominal_mode != data->passed_mode
                     || promoted_nominal_mode != data->promoted_mode);
  moved = false;

  if (need_conversion
      && GET_MODE_CLASS (data->nominal_mode) == MODE_INT
      && data->nominal_mode == data->passed_mode
      && data->nominal_mode == GET_MODE (data->entry_parm))
    {
      /* ENTRY_PARM has been converted to PROMOTED_MODE, its
         mode, by the caller.  We now have to convert it to
         NOMINAL_MODE, if different.  However, PARMREG may be in
         a different mode than NOMINAL_MODE if it is being stored
         promoted.

         If ENTRY_PARM is a hard register, it might be in a register
         not valid for operating in its mode (e.g., an odd-numbered
         register for a DFmode).  In that case, moves are the only
         thing valid, so we can't do a convert from there.  This
         occurs when the calling sequence allow such misaligned
         usages.

         In addition, the conversion may involve a call, which could
         clobber parameters which haven't been copied to pseudo
         registers yet.

         First, we try to emit an insn which performs the necessary
         conversion.  We verify that this insn does not clobber any
         hard registers.  */

      enum insn_code icode;
      rtx op0, op1;

      icode = can_extend_p (promoted_nominal_mode, data->passed_mode,
                            unsignedp);

      op0 = parmreg;
      op1 = validated_mem;
      if (icode != CODE_FOR_nothing
          && insn_operand_matches (icode, 0, op0)
          && insn_operand_matches (icode, 1, op1))
        {
          enum rtx_code code = unsignedp ? ZERO_EXTEND : SIGN_EXTEND;
          rtx insn, insns, t = op1;
          HARD_REG_SET hardregs;

          start_sequence ();
          /* If op1 is a hard register that is likely spilled, first
             force it into a pseudo, otherwise combiner might extend
             its lifetime too much.  */
          if (GET_CODE (t) == SUBREG)
            t = SUBREG_REG (t);
          if (REG_P (t)
              && HARD_REGISTER_P (t)
              && ! TEST_HARD_REG_BIT (fixed_reg_set, REGNO (t))
              && targetm.class_likely_spilled_p (REGNO_REG_CLASS (REGNO (t))))
            {
              t = gen_reg_rtx (GET_MODE (op1));
              emit_move_insn (t, op1);
            }
          else
            t = op1;
          insn = gen_extend_insn (op0, t, promoted_nominal_mode,
                                  data->passed_mode, unsignedp);
          emit_insn (insn);
          insns = get_insns ();

          moved = true;
          CLEAR_HARD_REG_SET (hardregs);
          for (insn = insns; insn && moved; insn = NEXT_INSN (insn))
            {
              if (INSN_P (insn))
                note_stores (PATTERN (insn), record_hard_reg_sets,
                             &hardregs);
              if (!hard_reg_set_empty_p (hardregs))
                moved = false;
            }

          end_sequence ();

          if (moved)
            {
              emit_insn (insns);
              if (equiv_stack_parm != NULL_RTX)
                equiv_stack_parm = gen_rtx_fmt_e (code, GET_MODE (parmreg),
                                                  equiv_stack_parm);
            }
        }
    }

  if (moved)
    /* Nothing to do.  */
    ;
  else if (need_conversion)
    {
      /* We did not have an insn to convert directly, or the sequence
         generated appeared unsafe.  We must first copy the parm to a
         pseudo reg, and save the conversion until after all
         parameters have been moved.  */

      int save_tree_used;
      rtx tempreg = gen_reg_rtx (GET_MODE (data->entry_parm));

      emit_move_insn (tempreg, validated_mem);

      push_to_sequence2 (all->first_conversion_insn, all->last_conversion_insn);
      tempreg = convert_to_mode (data->nominal_mode, tempreg, unsignedp);

      if (GET_CODE (tempreg) == SUBREG
          && GET_MODE (tempreg) == data->nominal_mode
          && REG_P (SUBREG_REG (tempreg))
          && data->nominal_mode == data->passed_mode
          && GET_MODE (SUBREG_REG (tempreg)) == GET_MODE (data->entry_parm)
          && GET_MODE_SIZE (GET_MODE (tempreg))
             < GET_MODE_SIZE (GET_MODE (data->entry_parm)))
        {
          /* The argument is already sign/zero extended, so note it
             into the subreg.  */
          SUBREG_PROMOTED_VAR_P (tempreg) = 1;
          SUBREG_PROMOTED_UNSIGNED_SET (tempreg, unsignedp);
        }

      /* TREE_USED gets set erroneously during expand_assignment.  */
      save_tree_used = TREE_USED (parm);
      expand_assignment (parm, make_tree (data->nominal_type, tempreg), false);
      TREE_USED (parm) = save_tree_used;
      all->first_conversion_insn = get_insns ();
      all->last_conversion_insn = get_last_insn ();
      end_sequence ();

      did_conversion = true;
    }
  else
    emit_move_insn (parmreg, validated_mem);

  /* If we were passed a pointer but the actual value can safely live
     in a register, put it in one.  */
  if (data->passed_pointer
      && TYPE_MODE (TREE_TYPE (parm)) != BLKmode
      /* If by-reference argument was promoted, demote it.  */
      && (TYPE_MODE (TREE_TYPE (parm)) != GET_MODE (DECL_RTL (parm))
          || use_register_for_decl (parm)))
    {
      /* We can't use nominal_mode, because it will have been set to
         Pmode above.  We must use the actual mode of the parm.  */
      parmreg = gen_reg_rtx (TYPE_MODE (TREE_TYPE (parm)));
      mark_user_reg (parmreg);

      if (GET_MODE (parmreg) != GET_MODE (DECL_RTL (parm)))
        {
          rtx tempreg = gen_reg_rtx (GET_MODE (DECL_RTL (parm)));
          int unsigned_p = TYPE_UNSIGNED (TREE_TYPE (parm));

          push_to_sequence2 (all->first_conversion_insn,
                             all->last_conversion_insn);
          emit_move_insn (tempreg, DECL_RTL (parm));
          tempreg = convert_to_mode (GET_MODE (parmreg), tempreg, unsigned_p);
          emit_move_insn (parmreg, tempreg);
          all->first_conversion_insn = get_insns ();
          all->last_conversion_insn = get_last_insn ();
          end_sequence ();

          did_conversion = true;
        }
      else
        emit_move_insn (parmreg, DECL_RTL (parm));

      SET_DECL_RTL (parm, parmreg);

      /* STACK_PARM is the pointer, not the parm, and PARMREG is
         now the parm.  */
      data->stack_parm = NULL;
    }

  /* Mark the register as eliminable if we did no conversion and it was
     copied from memory at a fixed offset, and the arg pointer was not
     copied to a pseudo-reg.  If the arg pointer is a pseudo reg or the
     offset formed an invalid address, such memory-equivalences as we
     make here would screw up life analysis for it.  */
  if (data->nominal_mode == data->passed_mode
      && !did_conversion
      && data->stack_parm != 0
      && MEM_P (data->stack_parm)
      && data->locate.offset.var == 0
      && reg_mentioned_p (virtual_incoming_args_rtx,
                          XEXP (data->stack_parm, 0)))
    {
      rtx linsn = get_last_insn ();
      rtx sinsn, set;

      /* Mark complex types separately.  */
      if (GET_CODE (parmreg) == CONCAT)
        {
          enum machine_mode submode
            = GET_MODE_INNER (GET_MODE (parmreg));
          int regnor = REGNO (XEXP (parmreg, 0));
          int regnoi = REGNO (XEXP (parmreg, 1));
          rtx stackr = adjust_address_nv (data->stack_parm, submode, 0);
          rtx stacki = adjust_address_nv (data->stack_parm, submode,
                                          GET_MODE_SIZE (submode));

          /* Scan backwards for the set of the real and
             imaginary parts.  */
          for (sinsn = linsn; sinsn != 0;
               sinsn = prev_nonnote_insn (sinsn))
            {
              set = single_set (sinsn);
              if (set == 0)
                continue;

              if (SET_DEST (set) == regno_reg_rtx [regnoi])
                set_unique_reg_note (sinsn, REG_EQUIV, stacki);
              else if (SET_DEST (set) == regno_reg_rtx [regnor])
                set_unique_reg_note (sinsn, REG_EQUIV, stackr);
            }
        }
      else 
        set_dst_reg_note (linsn, REG_EQUIV, equiv_stack_parm, parmreg);
    }

  /* For pointer data type, suggest pointer register.  */
  if (POINTER_TYPE_P (TREE_TYPE (parm)))
    mark_reg_pointer (parmreg,
                      TYPE_ALIGN (TREE_TYPE (TREE_TYPE (parm))));
}

/* A subroutine of assign_parms.  Allocate stack space to hold the current
   parameter.  Get it there.  Perform all ABI specified conversions.  */

static void
assign_parm_setup_stack (struct assign_parm_data_all *all, tree parm,
                         struct assign_parm_data_one *data)
{
  /* Value must be stored in the stack slot STACK_PARM during function
     execution.  */
  bool to_conversion = false;

  assign_parm_remove_parallels (data);

  if (data->promoted_mode != data->nominal_mode)
    {
      /* Conversion is required.  */
      rtx tempreg = gen_reg_rtx (GET_MODE (data->entry_parm));

      emit_move_insn (tempreg, validize_mem (data->entry_parm));

      push_to_sequence2 (all->first_conversion_insn, all->last_conversion_insn);
      to_conversion = true;

      data->entry_parm = convert_to_mode (data->nominal_mode, tempreg,
                                          TYPE_UNSIGNED (TREE_TYPE (parm)));

      if (data->stack_parm)
        {
          int offset = subreg_lowpart_offset (data->nominal_mode,
                                              GET_MODE (data->stack_parm));
          /* ??? This may need a big-endian conversion on sparc64.  */
          data->stack_parm
            = adjust_address (data->stack_parm, data->nominal_mode, 0);
          if (offset && MEM_OFFSET_KNOWN_P (data->stack_parm))
            set_mem_offset (data->stack_parm,
                            MEM_OFFSET (data->stack_parm) + offset);
        }
    }

  if (data->entry_parm != data->stack_parm)
    {
      rtx src, dest;

      if (data->stack_parm == 0)
        {
          int align = STACK_SLOT_ALIGNMENT (data->passed_type,
                                            GET_MODE (data->entry_parm),
                                            TYPE_ALIGN (data->passed_type));
          data->stack_parm
            = assign_stack_local (GET_MODE (data->entry_parm),
                                  GET_MODE_SIZE (GET_MODE (data->entry_parm)),
                                  align);
          set_mem_attributes (data->stack_parm, parm, 1);
        }

      dest = validize_mem (data->stack_parm);
      src = validize_mem (data->entry_parm);

      if (MEM_P (src))
        {
          /* Use a block move to handle potentially misaligned entry_parm.  */
          if (!to_conversion)
            push_to_sequence2 (all->first_conversion_insn,
                               all->last_conversion_insn);
          to_conversion = true;

          emit_block_move (dest, src,
                           GEN_INT (int_size_in_bytes (data->passed_type)),
                           BLOCK_OP_NORMAL);
        }
      else
        emit_move_insn (dest, src);
    }

  if (to_conversion)
    {
      all->first_conversion_insn = get_insns ();
      all->last_conversion_insn = get_last_insn ();
      end_sequence ();
    }

  SET_DECL_RTL (parm, data->stack_parm);
}

/* A subroutine of assign_parms.  If the ABI splits complex arguments, then
   undo the frobbing that we did in assign_parms_augmented_arg_list.  */

static void
assign_parms_unsplit_complex (struct assign_parm_data_all *all,
                              vec<tree> fnargs)
{
  tree parm;
  tree orig_fnargs = all->orig_fnargs;
  unsigned i = 0;

  for (parm = orig_fnargs; parm; parm = TREE_CHAIN (parm), ++i)
    {
      if (TREE_CODE (TREE_TYPE (parm)) == COMPLEX_TYPE
          && targetm.calls.split_complex_arg (TREE_TYPE (parm)))
        {
          rtx tmp, real, imag;
          enum machine_mode inner = GET_MODE_INNER (DECL_MODE (parm));

          real = DECL_RTL (fnargs[i]);
          imag = DECL_RTL (fnargs[i + 1]);
          if (inner != GET_MODE (real))
            {
              real = gen_lowpart_SUBREG (inner, real);
              imag = gen_lowpart_SUBREG (inner, imag);
            }

          if (TREE_ADDRESSABLE (parm))
            {
              rtx rmem, imem;
              HOST_WIDE_INT size = int_size_in_bytes (TREE_TYPE (parm));
              int align = STACK_SLOT_ALIGNMENT (TREE_TYPE (parm),
                                                DECL_MODE (parm),
                                                TYPE_ALIGN (TREE_TYPE (parm)));

              /* split_complex_arg put the real and imag parts in
                 pseudos.  Move them to memory.  */
              tmp = assign_stack_local (DECL_MODE (parm), size, align);
              set_mem_attributes (tmp, parm, 1);
              rmem = adjust_address_nv (tmp, inner, 0);
              imem = adjust_address_nv (tmp, inner, GET_MODE_SIZE (inner));
              push_to_sequence2 (all->first_conversion_insn,
                                 all->last_conversion_insn);
              emit_move_insn (rmem, real);
              emit_move_insn (imem, imag);
              all->first_conversion_insn = get_insns ();
              all->last_conversion_insn = get_last_insn ();
              end_sequence ();
            }
          else
            tmp = gen_rtx_CONCAT (DECL_MODE (parm), real, imag);
          SET_DECL_RTL (parm, tmp);

          real = DECL_INCOMING_RTL (fnargs[i]);
          imag = DECL_INCOMING_RTL (fnargs[i + 1]);
          if (inner != GET_MODE (real))
            {
              real = gen_lowpart_SUBREG (inner, real);
              imag = gen_lowpart_SUBREG (inner, imag);
            }
          tmp = gen_rtx_CONCAT (DECL_MODE (parm), real, imag);
          set_decl_incoming_rtl (parm, tmp, false);
          i++;
        }
    }
}

/* Assign RTL expressions to the function's parameters.  This may involve
   copying them into registers and using those registers as the DECL_RTL.  */

static void
assign_parms (tree fndecl)
{
  struct assign_parm_data_all all;
  tree parm;
  vec<tree> fnargs;
  unsigned i;

  crtl->args.internal_arg_pointer
    = targetm.calls.internal_arg_pointer ();

  assign_parms_initialize_all (&all);
  fnargs = assign_parms_augmented_arg_list (&all);

  FOR_EACH_VEC_ELT (fnargs, i, parm)
    {
      struct assign_parm_data_one data;

      /* Extract the type of PARM; adjust it according to ABI.  */
      assign_parm_find_data_types (&all, parm, &data);

      /* Early out for errors and void parameters.  */
      if (data.passed_mode == VOIDmode)
        {
          SET_DECL_RTL (parm, const0_rtx);
          DECL_INCOMING_RTL (parm) = DECL_RTL (parm);
          continue;
        }

      /* Estimate stack alignment from parameter alignment.  */
      if (SUPPORTS_STACK_ALIGNMENT)
        {
          unsigned int align
            = targetm.calls.function_arg_boundary (data.promoted_mode,
                                                   data.passed_type);
          align = MINIMUM_ALIGNMENT (data.passed_type, data.promoted_mode,
                                     align);
          if (TYPE_ALIGN (data.nominal_type) > align)
            align = MINIMUM_ALIGNMENT (data.nominal_type,
                                       TYPE_MODE (data.nominal_type),
                                       TYPE_ALIGN (data.nominal_type));
          if (crtl->stack_alignment_estimated < align)
            {
              gcc_assert (!crtl->stack_realign_processed);
              crtl->stack_alignment_estimated = align;
            }
        }

      if (cfun->stdarg && !DECL_CHAIN (parm))
        assign_parms_setup_varargs (&all, &data, false);

      /* Find out where the parameter arrives in this function.  */
      assign_parm_find_entry_rtl (&all, &data);

      /* Find out where stack space for this parameter might be.  */
      if (assign_parm_is_stack_parm (&all, &data))
        {
          assign_parm_find_stack_rtl (parm, &data);
          assign_parm_adjust_entry_rtl (&data);
        }

      /* Record permanently how this parm was passed.  */
      if (data.passed_pointer)
        {
          rtx incoming_rtl
            = gen_rtx_MEM (TYPE_MODE (TREE_TYPE (data.passed_type)),
                           data.entry_parm);
          set_decl_incoming_rtl (parm, incoming_rtl, true);
        }
      else
        set_decl_incoming_rtl (parm, data.entry_parm, false);

      /* Update info on where next arg arrives in registers.  */
      targetm.calls.function_arg_advance (all.args_so_far, data.promoted_mode,
                                          data.passed_type, data.named_arg);

      assign_parm_adjust_stack_rtl (&data);

      if (assign_parm_setup_block_p (&data))
        assign_parm_setup_block (&all, parm, &data);
      else if (data.passed_pointer || use_register_for_decl (parm))
        assign_parm_setup_reg (&all, parm, &data);
      else
        assign_parm_setup_stack (&all, parm, &data);
    }

  if (targetm.calls.split_complex_arg)
    assign_parms_unsplit_complex (&all, fnargs);

  fnargs.release ();

  /* Output all parameter conversion instructions (possibly including calls)
     now that all parameters have been copied out of hard registers.  */
  emit_insn (all.first_conversion_insn);

  /* Estimate reload stack alignment from scalar return mode.  */
  if (SUPPORTS_STACK_ALIGNMENT)
    {
      if (DECL_RESULT (fndecl))
        {
          tree type = TREE_TYPE (DECL_RESULT (fndecl));
          enum machine_mode mode = TYPE_MODE (type);

          if (mode != BLKmode
              && mode != VOIDmode
              && !AGGREGATE_TYPE_P (type))
            {
              unsigned int align = GET_MODE_ALIGNMENT (mode);
              if (crtl->stack_alignment_estimated < align)
                {
                  gcc_assert (!crtl->stack_realign_processed);
                  crtl->stack_alignment_estimated = align;
                }
            }
        }
    }

  /* If we are receiving a struct value address as the first argument, set up
     the RTL for the function result. As this might require code to convert
     the transmitted address to Pmode, we do this here to ensure that possible
     preliminary conversions of the address have been emitted already.  */
  if (all.function_result_decl)
    {
      tree result = DECL_RESULT (current_function_decl);
      rtx addr = DECL_RTL (all.function_result_decl);
      rtx x;

      if (DECL_BY_REFERENCE (result))
        {
          SET_DECL_VALUE_EXPR (result, all.function_result_decl);
          x = addr;
        }
      else
        {
          SET_DECL_VALUE_EXPR (result,
                               build1 (INDIRECT_REF, TREE_TYPE (result),
                                       all.function_result_decl));
          addr = convert_memory_address (Pmode, addr);
          x = gen_rtx_MEM (DECL_MODE (result), addr);
          set_mem_attributes (x, result, 1);
        }

      DECL_HAS_VALUE_EXPR_P (result) = 1;

      SET_DECL_RTL (result, x);
    }

  /* We have aligned all the args, so add space for the pretend args.  */
  crtl->args.pretend_args_size = all.pretend_args_size;
  all.stack_args_size.constant += all.extra_pretend_bytes;
  crtl->args.size = all.stack_args_size.constant;

  /* Adjust function incoming argument size for alignment and
     minimum length.  */

#ifdef REG_PARM_STACK_SPACE
  crtl->args.size = MAX (crtl->args.size,
                                    REG_PARM_STACK_SPACE (fndecl));
#endif

  crtl->args.size = CEIL_ROUND (crtl->args.size,
                                           PARM_BOUNDARY / BITS_PER_UNIT);

#ifdef ARGS_GROW_DOWNWARD
  crtl->args.arg_offset_rtx
    = (all.stack_args_size.var == 0 ? GEN_INT (-all.stack_args_size.constant)
       : expand_expr (size_diffop (all.stack_args_size.var,
                                   size_int (-all.stack_args_size.constant)),
                      NULL_RTX, VOIDmode, EXPAND_NORMAL));
#else
  crtl->args.arg_offset_rtx = ARGS_SIZE_RTX (all.stack_args_size);
#endif

  /* See how many bytes, if any, of its args a function should try to pop
     on return.  */

  crtl->args.pops_args = targetm.calls.return_pops_args (fndecl,
                                                         TREE_TYPE (fndecl),
                                                         crtl->args.size);

  /* For stdarg.h function, save info about
     regs and stack space used by the named args.  */

  crtl->args.info = all.args_so_far_v;

  /* Set the rtx used for the function return value.  Put this in its
     own variable so any optimizers that need this information don't have
     to include tree.h.  Do this here so it gets done when an inlined
     function gets output.  */

  crtl->return_rtx
    = (DECL_RTL_SET_P (DECL_RESULT (fndecl))
       ? DECL_RTL (DECL_RESULT (fndecl)) : NULL_RTX);

  /* If scalar return value was computed in a pseudo-reg, or was a named
     return value that got dumped to the stack, copy that to the hard
     return register.  */
  if (DECL_RTL_SET_P (DECL_RESULT (fndecl)))
    {
      tree decl_result = DECL_RESULT (fndecl);
      rtx decl_rtl = DECL_RTL (decl_result);

      if (REG_P (decl_rtl)
          ? REGNO (decl_rtl) >= FIRST_PSEUDO_REGISTER
          : DECL_REGISTER (decl_result))
        {
          rtx real_decl_rtl;

          real_decl_rtl = targetm.calls.function_value (TREE_TYPE (decl_result),
                                                        fndecl, true);
          REG_FUNCTION_VALUE_P (real_decl_rtl) = 1;
          /* The delay slot scheduler assumes that crtl->return_rtx
             holds the hard register containing the return value, not a
             temporary pseudo.  */
          crtl->return_rtx = real_decl_rtl;
        }
    }
}

/* A subroutine of gimplify_parameters, invoked via walk_tree.
   For all seen types, gimplify their sizes.  */

static tree
gimplify_parm_type (tree *tp, int *walk_subtrees, void *data)
{
  tree t = *tp;

  *walk_subtrees = 0;
  if (TYPE_P (t))
    {
      if (POINTER_TYPE_P (t))
        *walk_subtrees = 1;
      else if (TYPE_SIZE (t) && !TREE_CONSTANT (TYPE_SIZE (t))
               && !TYPE_SIZES_GIMPLIFIED (t))
        {
          gimplify_type_sizes (t, (gimple_seq *) data);
          *walk_subtrees = 1;
        }
    }

  return NULL;
}

/* Gimplify the parameter list for current_function_decl.  This involves
   evaluating SAVE_EXPRs of variable sized parameters and generating code
   to implement callee-copies reference parameters.  Returns a sequence of
   statements to add to the beginning of the function.  */

gimple_seq
gimplify_parameters (void)
{
  struct assign_parm_data_all all;
  tree parm;
  gimple_seq stmts = NULL;
  vec<tree> fnargs;
  unsigned i;

  assign_parms_initialize_all (&all);
  fnargs = assign_parms_augmented_arg_list (&all);

  FOR_EACH_VEC_ELT (fnargs, i, parm)
    {
      struct assign_parm_data_one data;

      /* Extract the type of PARM; adjust it according to ABI.  */
      assign_parm_find_data_types (&all, parm, &data);

      /* Early out for errors and void parameters.  */
      if (data.passed_mode == VOIDmode || DECL_SIZE (parm) == NULL)
        continue;

      /* Update info on where next arg arrives in registers.  */
      targetm.calls.function_arg_advance (all.args_so_far, data.promoted_mode,
                                          data.passed_type, data.named_arg);

      /* ??? Once upon a time variable_size stuffed parameter list
         SAVE_EXPRs (amongst others) onto a pending sizes list.  This
         turned out to be less than manageable in the gimple world.
         Now we have to hunt them down ourselves.  */
      walk_tree_without_duplicates (&data.passed_type,
                                    gimplify_parm_type, &stmts);

      if (TREE_CODE (DECL_SIZE_UNIT (parm)) != INTEGER_CST)
        {
          gimplify_one_sizepos (&DECL_SIZE (parm), &stmts);
          gimplify_one_sizepos (&DECL_SIZE_UNIT (parm), &stmts);
        }

      if (data.passed_pointer)
        {
          tree type = TREE_TYPE (data.passed_type);
          if (reference_callee_copied (&all.args_so_far_v, TYPE_MODE (type),
                                       type, data.named_arg))
            {
              tree local, t;

              /* For constant-sized objects, this is trivial; for
                 variable-sized objects, we have to play games.  */
              if (TREE_CODE (DECL_SIZE_UNIT (parm)) == INTEGER_CST
                  && !(flag_stack_check == GENERIC_STACK_CHECK
                       && compare_tree_int (DECL_SIZE_UNIT (parm),
                                            STACK_CHECK_MAX_VAR_SIZE) > 0))
                {
                  local = create_tmp_var (type, get_name (parm));
                  DECL_IGNORED_P (local) = 0;
                  /* If PARM was addressable, move that flag over
                     to the local copy, as its address will be taken,
                     not the PARMs.  Keep the parms address taken
                     as we'll query that flag during gimplification.  */
                  if (TREE_ADDRESSABLE (parm))
                    TREE_ADDRESSABLE (local) = 1;
                  else if (TREE_CODE (type) == COMPLEX_TYPE
                           || TREE_CODE (type) == VECTOR_TYPE)
                    DECL_GIMPLE_REG_P (local) = 1;
                }
              else
                {
                  tree ptr_type, addr;

                  ptr_type = build_pointer_type (type);
                  addr = create_tmp_reg (ptr_type, get_name (parm));
                  DECL_IGNORED_P (addr) = 0;
                  local = build_fold_indirect_ref (addr);

                  t = builtin_decl_explicit (BUILT_IN_ALLOCA_WITH_ALIGN);
                  t = build_call_expr (t, 2, DECL_SIZE_UNIT (parm),
                                       size_int (DECL_ALIGN (parm)));

                  /* The call has been built for a variable-sized object.  */
                  CALL_ALLOCA_FOR_VAR_P (t) = 1;
                  t = fold_convert (ptr_type, t);
                  t = build2 (MODIFY_EXPR, TREE_TYPE (addr), addr, t);
                  gimplify_and_add (t, &stmts);
                }

              gimplify_assign (local, parm, &stmts);

              SET_DECL_VALUE_EXPR (parm, local);
              DECL_HAS_VALUE_EXPR_P (parm) = 1;
            }
        }
    }

  fnargs.release ();

  return stmts;
}
\f
/* Compute the size and offset from the start of the stacked arguments for a
   parm passed in mode PASSED_MODE and with type TYPE.

   INITIAL_OFFSET_PTR points to the current offset into the stacked
   arguments.

   The starting offset and size for this parm are returned in
   LOCATE->OFFSET and LOCATE->SIZE, respectively.  When IN_REGS is
   nonzero, the offset is that of stack slot, which is returned in
   LOCATE->SLOT_OFFSET.  LOCATE->ALIGNMENT_PAD is the amount of
   padding required from the initial offset ptr to the stack slot.

   IN_REGS is nonzero if the argument will be passed in registers.  It will
   never be set if REG_PARM_STACK_SPACE is not defined.

   FNDECL is the function in which the argument was defined.

   There are two types of rounding that are done.  The first, controlled by
   TARGET_FUNCTION_ARG_BOUNDARY, forces the offset from the start of the
   argument list to be aligned to the specific boundary (in bits).  This
   rounding affects the initial and starting offsets, but not the argument
   size.

   The second, controlled by FUNCTION_ARG_PADDING and PARM_BOUNDARY,
   optionally rounds the size of the parm to PARM_BOUNDARY.  The
   initial offset is not affected by this rounding, while the size always
   is and the starting offset may be.  */

/*  LOCATE->OFFSET will be negative for ARGS_GROW_DOWNWARD case;
    INITIAL_OFFSET_PTR is positive because locate_and_pad_parm's
    callers pass in the total size of args so far as
    INITIAL_OFFSET_PTR.  LOCATE->SIZE is always positive.  */

void
locate_and_pad_parm (enum machine_mode passed_mode, tree type, int in_regs,
                     int partial, tree fndecl ATTRIBUTE_UNUSED,
                     struct args_size *initial_offset_ptr,
                     struct locate_and_pad_arg_data *locate)
{
  tree sizetree;
  enum direction where_pad;
  unsigned int boundary, round_boundary;
  int reg_parm_stack_space = 0;
  int part_size_in_regs;

#ifdef REG_PARM_STACK_SPACE
  reg_parm_stack_space = REG_PARM_STACK_SPACE (fndecl);

  /* If we have found a stack parm before we reach the end of the
     area reserved for registers, skip that area.  */
  if (! in_regs)
    {
      if (reg_parm_stack_space > 0)
        {
          if (initial_offset_ptr->var)
            {
              initial_offset_ptr->var
                = size_binop (MAX_EXPR, ARGS_SIZE_TREE (*initial_offset_ptr),
                              ssize_int (reg_parm_stack_space));
              initial_offset_ptr->constant = 0;
            }
          else if (initial_offset_ptr->constant < reg_parm_stack_space)
            initial_offset_ptr->constant = reg_parm_stack_space;
        }
    }
#endif /* REG_PARM_STACK_SPACE */

  part_size_in_regs = (reg_parm_stack_space == 0 ? partial : 0);

  sizetree
    = type ? size_in_bytes (type) : size_int (GET_MODE_SIZE (passed_mode));
  where_pad = FUNCTION_ARG_PADDING (passed_mode, type);
  boundary = targetm.calls.function_arg_boundary (passed_mode, type);
  round_boundary = targetm.calls.function_arg_round_boundary (passed_mode,
                                                              type);
  locate->where_pad = where_pad;

  /* Alignment can't exceed MAX_SUPPORTED_STACK_ALIGNMENT.  */
  if (boundary > MAX_SUPPORTED_STACK_ALIGNMENT)
    boundary = MAX_SUPPORTED_STACK_ALIGNMENT;

  locate->boundary = boundary;

  if (SUPPORTS_STACK_ALIGNMENT)
    {
      /* stack_alignment_estimated can't change after stack has been
         realigned.  */
      if (crtl->stack_alignment_estimated < boundary)
        {
          if (!crtl->stack_realign_processed)
            crtl->stack_alignment_estimated = boundary;
          else
            {
              /* If stack is realigned and stack alignment value
                 hasn't been finalized, it is OK not to increase
                 stack_alignment_estimated.  The bigger alignment
                 requirement is recorded in stack_alignment_needed
                 below.  */
              gcc_assert (!crtl->stack_realign_finalized
                          && crtl->stack_realign_needed);
            }
        }
    }

  /* Remember if the outgoing parameter requires extra alignment on the
     calling function side.  */
  if (crtl->stack_alignment_needed < boundary)
    crtl->stack_alignment_needed = boundary;
  if (crtl->preferred_stack_boundary < boundary)
    crtl->preferred_stack_boundary = boundary;

#ifdef ARGS_GROW_DOWNWARD
  locate->slot_offset.constant = -initial_offset_ptr->constant;
  if (initial_offset_ptr->var)
    locate->slot_offset.var = size_binop (MINUS_EXPR, ssize_int (0),
                                          initial_offset_ptr->var);

  {
    tree s2 = sizetree;
    if (where_pad != none
        && (!host_integerp (sizetree, 1)
            || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % round_boundary))
      s2 = round_up (s2, round_boundary / BITS_PER_UNIT);
    SUB_PARM_SIZE (locate->slot_offset, s2);
  }

  locate->slot_offset.constant += part_size_in_regs;

  if (!in_regs
#ifdef REG_PARM_STACK_SPACE
      || REG_PARM_STACK_SPACE (fndecl) > 0
#endif
     )
    pad_to_arg_alignment (&locate->slot_offset, boundary,
                          &locate->alignment_pad);

  locate->size.constant = (-initial_offset_ptr->constant
                           - locate->slot_offset.constant);
  if (initial_offset_ptr->var)
    locate->size.var = size_binop (MINUS_EXPR,
                                   size_binop (MINUS_EXPR,
                                               ssize_int (0),
                                               initial_offset_ptr->var),
                                   locate->slot_offset.var);

  /* Pad_below needs the pre-rounded size to know how much to pad
     below.  */
  locate->offset = locate->slot_offset;
  if (where_pad == downward)
    pad_below (&locate->offset, passed_mode, sizetree);

#else /* !ARGS_GROW_DOWNWARD */
  if (!in_regs
#ifdef REG_PARM_STACK_SPACE
      || REG_PARM_STACK_SPACE (fndecl) > 0
#endif
      )
    pad_to_arg_alignment (initial_offset_ptr, boundary,
                          &locate->alignment_pad);
  locate->slot_offset = *initial_offset_ptr;

#ifdef PUSH_ROUNDING
  if (passed_mode != BLKmode)
    sizetree = size_int (PUSH_ROUNDING (TREE_INT_CST_LOW (sizetree)));
#endif

  /* Pad_below needs the pre-rounded size to know how much to pad below
     so this must be done before rounding up.  */
  locate->offset = locate->slot_offset;
  if (where_pad == downward)
    pad_below (&locate->offset, passed_mode, sizetree);

  if (where_pad != none
      && (!host_integerp (sizetree, 1)
          || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % round_boundary))
    sizetree = round_up (sizetree, round_boundary / BITS_PER_UNIT);

  ADD_PARM_SIZE (locate->size, sizetree);

  locate->size.constant -= part_size_in_regs;
#endif /* ARGS_GROW_DOWNWARD */

#ifdef FUNCTION_ARG_OFFSET
  locate->offset.constant += FUNCTION_ARG_OFFSET (passed_mode, type);
#endif
}

/* Round the stack offset in *OFFSET_PTR up to a multiple of BOUNDARY.
   BOUNDARY is measured in bits, but must be a multiple of a storage unit.  */

static void
pad_to_arg_alignment (struct args_size *offset_ptr, int boundary,
                      struct args_size *alignment_pad)
{
  tree save_var = NULL_TREE;
  HOST_WIDE_INT save_constant = 0;
  int boundary_in_bytes = boundary / BITS_PER_UNIT;
  HOST_WIDE_INT sp_offset = STACK_POINTER_OFFSET;

#ifdef SPARC_STACK_BOUNDARY_HACK
  /* ??? The SPARC port may claim a STACK_BOUNDARY higher than
     the real alignment of %sp.  However, when it does this, the
     alignment of %sp+STACK_POINTER_OFFSET is STACK_BOUNDARY.  */
  if (SPARC_STACK_BOUNDARY_HACK)
    sp_offset = 0;
#endif

  if (boundary > PARM_BOUNDARY)
    {
      save_var = offset_ptr->var;
      save_constant = offset_ptr->constant;
    }

  alignment_pad->var = NULL_TREE;
  alignment_pad->constant = 0;

  if (boundary > BITS_PER_UNIT)
    {
      if (offset_ptr->var)
        {
          tree sp_offset_tree = ssize_int (sp_offset);
          tree offset = size_binop (PLUS_EXPR,
                                    ARGS_SIZE_TREE (*offset_ptr),
                                    sp_offset_tree);
#ifdef ARGS_GROW_DOWNWARD
          tree rounded = round_down (offset, boundary / BITS_PER_UNIT);
#else
          tree rounded = round_up   (offset, boundary / BITS_PER_UNIT);
#endif

          offset_ptr->var = size_binop (MINUS_EXPR, rounded, sp_offset_tree);
          /* ARGS_SIZE_TREE includes constant term.  */
          offset_ptr->constant = 0;
          if (boundary > PARM_BOUNDARY)
            alignment_pad->var = size_binop (MINUS_EXPR, offset_ptr->var,
                                             save_var);
        }
      else
        {
          offset_ptr->constant = -sp_offset +
#ifdef ARGS_GROW_DOWNWARD
            FLOOR_ROUND (offset_ptr->constant + sp_offset, boundary_in_bytes);
#else
            CEIL_ROUND (offset_ptr->constant + sp_offset, boundary_in_bytes);
#endif
            if (boundary > PARM_BOUNDARY)
              alignment_pad->constant = offset_ptr->constant - save_constant;
        }
    }
}

static void
pad_below (struct args_size *offset_ptr, enum machine_mode passed_mode, tree sizetree)
{
  if (passed_mode != BLKmode)
    {
      if (GET_MODE_BITSIZE (passed_mode) % PARM_BOUNDARY)
        offset_ptr->constant
          += (((GET_MODE_BITSIZE (passed_mode) + PARM_BOUNDARY - 1)
               / PARM_BOUNDARY * PARM_BOUNDARY / BITS_PER_UNIT)
              - GET_MODE_SIZE (passed_mode));
    }
  else
    {
      if (TREE_CODE (sizetree) != INTEGER_CST
          || (TREE_INT_CST_LOW (sizetree) * BITS_PER_UNIT) % PARM_BOUNDARY)
        {
          /* Round the size up to multiple of PARM_BOUNDARY bits.  */
          tree s2 = round_up (sizetree, PARM_BOUNDARY / BITS_PER_UNIT);
          /* Add it in.  */
          ADD_PARM_SIZE (*offset_ptr, s2);
          SUB_PARM_SIZE (*offset_ptr, sizetree);
        }
    }
}
\f

/* True if register REGNO was alive at a place where `setjmp' was
   called and was set more than once or is an argument.  Such regs may
   be clobbered by `longjmp'.  */

static bool
regno_clobbered_at_setjmp (bitmap setjmp_crosses, int regno)
{
  /* There appear to be cases where some local vars never reach the
     backend but have bogus regnos.  */
  if (regno >= max_reg_num ())
    return false;

  return ((REG_N_SETS (regno) > 1
           || REGNO_REG_SET_P (df_get_live_out (ENTRY_BLOCK_PTR), regno))
          && REGNO_REG_SET_P (setjmp_crosses, regno));
}

/* Walk the tree of blocks describing the binding levels within a
   function and warn about variables the might be killed by setjmp or
   vfork.  This is done after calling flow_analysis before register
   allocation since that will clobber the pseudo-regs to hard
   regs.  */

static void
setjmp_vars_warning (bitmap setjmp_crosses, tree block)
{
  tree decl, sub;

  for (decl = BLOCK_VARS (block); decl; decl = DECL_CHAIN (decl))
    {
      if (TREE_CODE (decl) == VAR_DECL
          && DECL_RTL_SET_P (decl)
          && REG_P (DECL_RTL (decl))
          && regno_clobbered_at_setjmp (setjmp_crosses, REGNO (DECL_RTL (decl))))
        warning (OPT_Wclobbered, "variable %q+D might be clobbered by"
                 " %<longjmp%> or %<vfork%>", decl);
    }

  for (sub = BLOCK_SUBBLOCKS (block); sub; sub = BLOCK_CHAIN (sub))
    setjmp_vars_warning (setjmp_crosses, sub);
}

/* Do the appropriate part of setjmp_vars_warning
   but for arguments instead of local variables.  */

static void
setjmp_args_warning (bitmap setjmp_crosses)
{
  tree decl;
  for (decl = DECL_ARGUMENTS (current_function_decl);
       decl; decl = DECL_CHAIN (decl))
    if (DECL_RTL (decl) != 0
        && REG_P (DECL_RTL (decl))
        && regno_clobbered_at_setjmp (setjmp_crosses, REGNO (DECL_RTL (decl))))
      warning (OPT_Wclobbered,
               "argument %q+D might be clobbered by %<longjmp%> or %<vfork%>",
               decl);
}

/* Generate warning messages for variables live across setjmp.  */

void
generate_setjmp_warnings (void)
{
  bitmap setjmp_crosses = regstat_get_setjmp_crosses ();

  if (n_basic_blocks == NUM_FIXED_BLOCKS
      || bitmap_empty_p (setjmp_crosses))
    return;

  setjmp_vars_warning (setjmp_crosses, DECL_INITIAL (current_function_decl));
  setjmp_args_warning (setjmp_crosses);
}

\f
/* Reverse the order of elements in the fragment chain T of blocks,
   and return the new head of the chain (old last element).
   In addition to that clear BLOCK_SAME_RANGE flags when needed
   and adjust BLOCK_SUPERCONTEXT from the super fragment to
   its super fragment origin.  */

static tree
block_fragments_nreverse (tree t)
{
  tree prev = 0, block, next, prev_super = 0;
  tree super = BLOCK_SUPERCONTEXT (t);
  if (BLOCK_FRAGMENT_ORIGIN (super))
    super = BLOCK_FRAGMENT_ORIGIN (super);
  for (block = t; block; block = next)
    {
      next = BLOCK_FRAGMENT_CHAIN (block);
      BLOCK_FRAGMENT_CHAIN (block) = prev;
      if ((prev && !BLOCK_SAME_RANGE (prev))
          || (BLOCK_FRAGMENT_CHAIN (BLOCK_SUPERCONTEXT (block))
              != prev_super))
        BLOCK_SAME_RANGE (block) = 0;
      prev_super = BLOCK_SUPERCONTEXT (block);
      BLOCK_SUPERCONTEXT (block) = super;
      prev = block;
    }
  t = BLOCK_FRAGMENT_ORIGIN (t);
  if (BLOCK_FRAGMENT_CHAIN (BLOCK_SUPERCONTEXT (t))
      != prev_super)
    BLOCK_SAME_RANGE (t) = 0;
  BLOCK_SUPERCONTEXT (t) = super;
  return prev;
}

/* Reverse the order of elements in the chain T of blocks,
   and return the new head of the chain (old last element).
   Also do the same on subblocks and reverse the order of elements
   in BLOCK_FRAGMENT_CHAIN as well.  */

static tree
blocks_nreverse_all (tree t)
{
  tree prev = 0, block, next;
  for (block = t; block; block = next)
    {
      next = BLOCK_CHAIN (block);
      BLOCK_CHAIN (block) = prev;
      if (BLOCK_FRAGMENT_CHAIN (block)
          && BLOCK_FRAGMENT_ORIGIN (block) == NULL_TREE)
        {
          BLOCK_FRAGMENT_CHAIN (block)
            = block_fragments_nreverse (BLOCK_FRAGMENT_CHAIN (block));
          if (!BLOCK_SAME_RANGE (BLOCK_FRAGMENT_CHAIN (block)))
            BLOCK_SAME_RANGE (block) = 0;
        }
      BLOCK_SUBBLOCKS (block) = blocks_nreverse_all (BLOCK_SUBBLOCKS (block));
      prev = block;
    }
  return prev;
}


/* Identify BLOCKs referenced by more than one NOTE_INSN_BLOCK_{BEG,END},
   and create duplicate blocks.  */
/* ??? Need an option to either create block fragments or to create
   abstract origin duplicates of a source block.  It really depends
   on what optimization has been performed.  */

void
reorder_blocks (void)
{
  tree block = DECL_INITIAL (current_function_decl);
  vec<tree> block_stack;

  if (block == NULL_TREE)
    return;

  block_stack.create (10);

  /* Reset the TREE_ASM_WRITTEN bit for all blocks.  */
  clear_block_marks (block);

  /* Prune the old trees away, so that they don't get in the way.  */
  BLOCK_SUBBLOCKS (block) = NULL_TREE;
  BLOCK_CHAIN (block) = NULL_TREE;

  /* Recreate the block tree from the note nesting.  */
  reorder_blocks_1 (get_insns (), block, &block_stack);
  BLOCK_SUBBLOCKS (block) = blocks_nreverse_all (BLOCK_SUBBLOCKS (block));

  block_stack.release ();
}

/* Helper function for reorder_blocks.  Reset TREE_ASM_WRITTEN.  */

void
clear_block_marks (tree block)
{
  while (block)
    {
      TREE_ASM_WRITTEN (block) = 0;
      clear_block_marks (BLOCK_SUBBLOCKS (block));
      block = BLOCK_CHAIN (block);
    }
}

static void
reorder_blocks_1 (rtx insns, tree current_block, vec<tree> *p_block_stack)
{
  rtx insn;
  tree prev_beg = NULL_TREE, prev_end = NULL_TREE;

  for (insn = insns; insn; insn = NEXT_INSN (insn))
    {
      if (NOTE_P (insn))
        {
          if (NOTE_KIND (insn) == NOTE_INSN_BLOCK_BEG)
            {
              tree block = NOTE_BLOCK (insn);
              tree origin;

              gcc_assert (BLOCK_FRAGMENT_ORIGIN (block) == NULL_TREE);
              origin = block;

              if (prev_end)
                BLOCK_SAME_RANGE (prev_end) = 0;
              prev_end = NULL_TREE;

              /* If we have seen this block before, that means it now
                 spans multiple address regions.  Create a new fragment.  */
              if (TREE_ASM_WRITTEN (block))
                {
                  tree new_block = copy_node (block);

                  BLOCK_SAME_RANGE (new_block) = 0;
                  BLOCK_FRAGMENT_ORIGIN (new_block) = origin;
                  BLOCK_FRAGMENT_CHAIN (new_block)
                    = BLOCK_FRAGMENT_CHAIN (origin);
                  BLOCK_FRAGMENT_CHAIN (origin) = new_block;

                  NOTE_BLOCK (insn) = new_block;
                  block = new_block;
                }

              if (prev_beg == current_block && prev_beg)
                BLOCK_SAME_RANGE (block) = 1;

              prev_beg = origin;

              BLOCK_SUBBLOCKS (block) = 0;
              TREE_ASM_WRITTEN (block) = 1;
              /* When there's only one block for the entire function,
                 current_block == block and we mustn't do this, it
                 will cause infinite recursion.  */
              if (block != current_block)
                {
                  tree super;
                  if (block != origin)
                    gcc_assert (BLOCK_SUPERCONTEXT (origin) == current_block
                                || BLOCK_FRAGMENT_ORIGIN (BLOCK_SUPERCONTEXT
                                                                      (origin))
                                   == current_block);
                  if (p_block_stack->is_empty ())
                    super = current_block;
                  else
                    {
                      super = p_block_stack->last ();
                      gcc_assert (super == current_block
                                  || BLOCK_FRAGMENT_ORIGIN (super)
                                     == current_block);
                    }
                  BLOCK_SUPERCONTEXT (block) = super;
                  BLOCK_CHAIN (block) = BLOCK_SUBBLOCKS (current_block);
                  BLOCK_SUBBLOCKS (current_block) = block;
                  current_block = origin;
                }
              p_block_stack->safe_push (block);
            }
          else if (NOTE_KIND (insn) == NOTE_INSN_BLOCK_END)
            {
              NOTE_BLOCK (insn) = p_block_stack->pop ();
              current_block = BLOCK_SUPERCONTEXT (current_block);
              if (BLOCK_FRAGMENT_ORIGIN (current_block))
                current_block = BLOCK_FRAGMENT_ORIGIN (current_block);
              prev_beg = NULL_TREE;
              prev_end = BLOCK_SAME_RANGE (NOTE_BLOCK (insn))
                         ? NOTE_BLOCK (insn) : NULL_TREE;
            }
        }
      else
        {
          prev_beg = NULL_TREE;
          if (prev_end)
            BLOCK_SAME_RANGE (prev_end) = 0;
          prev_end = NULL_TREE;
        }
    }
}

/* Reverse the order of elements in the chain T of blocks,
   and return the new head of the chain (old last element).  */

tree
blocks_nreverse (tree t)
{
  tree prev = 0, block, next;
  for (block = t; block; block = next)
    {
      next = BLOCK_CHAIN (block);
      BLOCK_CHAIN (block) = prev;
      prev = block;
    }
  return prev;
}

/* Concatenate two chains of blocks (chained through BLOCK_CHAIN)
   by modifying the last node in chain 1 to point to chain 2.  */

tree
block_chainon (tree op1, tree op2)
{
  tree t1;

  if (!op1)
    return op2;
  if (!op2)
    return op1;

  for (t1 = op1; BLOCK_CHAIN (t1); t1 = BLOCK_CHAIN (t1))
    continue;
  BLOCK_CHAIN (t1) = op2;

#ifdef ENABLE_TREE_CHECKING
  {
    tree t2;
    for (t2 = op2; t2; t2 = BLOCK_CHAIN (t2))
      gcc_assert (t2 != t1);
  }
#endif

  return op1;
}

/* Count the subblocks of the list starting with BLOCK.  If VECTOR is
   non-NULL, list them all into VECTOR, in a depth-first preorder
   traversal of the block tree.  Also clear TREE_ASM_WRITTEN in all
   blocks.  */

static int
all_blocks (tree block, tree *vector)
{
  int n_blocks = 0;

  while (block)
    {
      TREE_ASM_WRITTEN (block) = 0;

      /* Record this block.  */
      if (vector)
        vector[n_blocks] = block;

      ++n_blocks;

      /* Record the subblocks, and their subblocks...  */
      n_blocks += all_blocks (BLOCK_SUBBLOCKS (block),
                              vector ? vector + n_blocks : 0);
      block = BLOCK_CHAIN (block);
    }

  return n_blocks;
}

/* Return a vector containing all the blocks rooted at BLOCK.  The
   number of elements in the vector is stored in N_BLOCKS_P.  The
   vector is dynamically allocated; it is the caller's responsibility
   to call `free' on the pointer returned.  */

static tree *
get_block_vector (tree block, int *n_blocks_p)
{
  tree *block_vector;

  *n_blocks_p = all_blocks (block, NULL);
  block_vector = XNEWVEC (tree, *n_blocks_p);
  all_blocks (block, block_vector);

  return block_vector;
}

static GTY(()) int next_block_index = 2;

/* Set BLOCK_NUMBER for all the blocks in FN.  */

void
number_blocks (tree fn)
{
  int i;
  int n_blocks;
  tree *block_vector;

  /* For SDB and XCOFF debugging output, we start numbering the blocks
     from 1 within each function, rather than keeping a running
     count.  */
#if defined (SDB_DEBUGGING_INFO) || defined (XCOFF_DEBUGGING_INFO)
  if (write_symbols == SDB_DEBUG || write_symbols == XCOFF_DEBUG)
    next_block_index = 1;
#endif

  block_vector = get_block_vector (DECL_INITIAL (fn), &n_blocks);

  /* The top-level BLOCK isn't numbered at all.  */
  for (i = 1; i < n_blocks; ++i)
    /* We number the blocks from two.  */
    BLOCK_NUMBER (block_vector[i]) = next_block_index++;

  free (block_vector);

  return;
}

/* If VAR is present in a subblock of BLOCK, return the subblock.  */

DEBUG_FUNCTION tree
debug_find_var_in_block_tree (tree var, tree block)
{
  tree t;

  for (t = BLOCK_VARS (block); t; t = TREE_CHAIN (t))
    if (t == var)
      return block;

  for (t = BLOCK_SUBBLOCKS (block); t; t = TREE_CHAIN (t))
    {
      tree ret = debug_find_var_in_block_tree (var, t);
      if (ret)
        return ret;
    }

  return NULL_TREE;
}
\f
/* Keep track of whether we're in a dummy function context.  If we are,
   we don't want to invoke the set_current_function hook, because we'll
   get into trouble if the hook calls target_reinit () recursively or
   when the initial initialization is not yet complete.  */

static bool in_dummy_function;

/* Invoke the target hook when setting cfun.  Update the optimization options
   if the function uses different options than the default.  */

static void
invoke_set_current_function_hook (tree fndecl)
{
  if (!in_dummy_function)
    {
      tree opts = ((fndecl)
                   ? DECL_FUNCTION_SPECIFIC_OPTIMIZATION (fndecl)
                   : optimization_default_node);

      if (!opts)
        opts = optimization_default_node;

      /* Change optimization options if needed.  */
      if (optimization_current_node != opts)
        {
          optimization_current_node = opts;
          cl_optimization_restore (&global_options, TREE_OPTIMIZATION (opts));
        }

      targetm.set_current_function (fndecl);
      this_fn_optabs = this_target_optabs;

      if (opts != optimization_default_node)
        {
          init_tree_optimization_optabs (opts);
          if (TREE_OPTIMIZATION_OPTABS (opts))
            this_fn_optabs = (struct target_optabs *)
              TREE_OPTIMIZATION_OPTABS (opts);
        }
    }
}

/* cfun should never be set directly; use this function.  */

void
set_cfun (struct function *new_cfun)
{
  if (cfun != new_cfun)
    {
      cfun = new_cfun;
      invoke_set_current_function_hook (new_cfun ? new_cfun->decl : NULL_TREE);
    }
}

/* Initialized with NOGC, making this poisonous to the garbage collector.  */

static vec<function_p> cfun_stack;

/* Push the current cfun onto the stack, and set cfun to new_cfun.  Also set
   current_function_decl accordingly.  */

void
push_cfun (struct function *new_cfun)
{
  gcc_assert ((!cfun && !current_function_decl)
              || (cfun && current_function_decl == cfun->decl));
  cfun_stack.safe_push (cfun);
  current_function_decl = new_cfun ? new_cfun->decl : NULL_TREE;
  set_cfun (new_cfun);
}

/* Pop cfun from the stack.  Also set current_function_decl accordingly.  */

void
pop_cfun (void)
{
  struct function *new_cfun = cfun_stack.pop ();
  /* When in_dummy_function, we do have a cfun but current_function_decl is
     NULL.  We also allow pushing NULL cfun and subsequently changing
     current_function_decl to something else and have both restored by
     pop_cfun.  */
  gcc_checking_assert (in_dummy_function
                       || !cfun
                       || current_function_decl == cfun->decl);
  set_cfun (new_cfun);
  current_function_decl = new_cfun ? new_cfun->decl : NULL_TREE;
}

/* Return value of funcdef and increase it.  */
int
get_next_funcdef_no (void)
{
  return funcdef_no++;
}

/* Return value of funcdef.  */
int
get_last_funcdef_no (void)
{
  return funcdef_no;
}

/* Allocate a function structure for FNDECL and set its contents
   to the defaults.  Set cfun to the newly-allocated object.
   Some of the helper functions invoked during initialization assume
   that cfun has already been set.  Therefore, assign the new object
   directly into cfun and invoke the back end hook explicitly at the
   very end, rather than initializing a temporary and calling set_cfun
   on it.

   ABSTRACT_P is true if this is a function that will never be seen by
   the middle-end.  Such functions are front-end concepts (like C++
   function templates) that do not correspond directly to functions
   placed in object files.  */

void
allocate_struct_function (tree fndecl, bool abstract_p)
{
  tree fntype = fndecl ? TREE_TYPE (fndecl) : NULL_TREE;

  cfun = ggc_alloc_cleared_function ();

  init_eh_for_function ();

  if (init_machine_status)
    cfun->machine = (*init_machine_status) ();

#ifdef OVERRIDE_ABI_FORMAT
  OVERRIDE_ABI_FORMAT (fndecl);
#endif

  if (fndecl != NULL_TREE)
    {
      DECL_STRUCT_FUNCTION (fndecl) = cfun;
      cfun->decl = fndecl;
      current_function_funcdef_no = get_next_funcdef_no ();
    }

  invoke_set_current_function_hook (fndecl);

  if (fndecl != NULL_TREE)
    {
      tree result = DECL_RESULT (fndecl);
      if (!abstract_p && aggregate_value_p (result, fndecl))
        {
#ifdef PCC_STATIC_STRUCT_RETURN
          cfun->returns_pcc_struct = 1;
#endif
          cfun->returns_struct = 1;
        }

      cfun->stdarg = stdarg_p (fntype);

      /* Assume all registers in stdarg functions need to be saved.  */
      cfun->va_list_gpr_size = VA_LIST_MAX_GPR_SIZE;
      cfun->va_list_fpr_size = VA_LIST_MAX_FPR_SIZE;

      /* ??? This could be set on a per-function basis by the front-end
         but is this worth the hassle?  */
      cfun->can_throw_non_call_exceptions = flag_non_call_exceptions;
    }
}

/* This is like allocate_struct_function, but pushes a new cfun for FNDECL
   instead of just setting it.  */

void
push_struct_function (tree fndecl)
{
  /* When in_dummy_function we might be in the middle of a pop_cfun and
     current_function_decl and cfun may not match.  */
  gcc_assert (in_dummy_function
              || (!cfun && !current_function_decl)
              || (cfun && current_function_decl == cfun->decl));
  cfun_stack.safe_push (cfun);
  current_function_decl = fndecl;
  allocate_struct_function (fndecl, false);
}

/* Reset crtl and other non-struct-function variables to defaults as
   appropriate for emitting rtl at the start of a function.  */

static void
prepare_function_start (void)
{
  gcc_assert (!crtl->emit.x_last_insn);
  init_temp_slots ();
  init_emit ();
  init_varasm_status ();
  init_expr ();
  default_rtl_profile ();

  if (flag_stack_usage_info)
    {
      cfun->su = ggc_alloc_cleared_stack_usage ();
      cfun->su->static_stack_size = -1;
    }

  cse_not_expected = ! optimize;

  /* Caller save not needed yet.  */
  caller_save_needed = 0;

  /* We haven't done register allocation yet.  */
  reg_renumber = 0;

  /* Indicate that we have not instantiated virtual registers yet.  */
  virtuals_instantiated = 0;

  /* Indicate that we want CONCATs now.  */
  generating_concat_p = 1;

  /* Indicate we have no need of a frame pointer yet.  */
  frame_pointer_needed = 0;
}

/* Initialize the rtl expansion mechanism so that we can do simple things
   like generate sequences.  This is used to provide a context during global
   initialization of some passes.  You must call expand_dummy_function_end
   to exit this context.  */

void
init_dummy_function_start (void)
{
  gcc_assert (!in_dummy_function);
  in_dummy_function = true;
  push_struct_function (NULL_TREE);
  prepare_function_start ();
}

/* Generate RTL for the start of the function SUBR (a FUNCTION_DECL tree node)
   and initialize static variables for generating RTL for the statements
   of the function.  */

void
init_function_start (tree subr)
{
  if (subr && DECL_STRUCT_FUNCTION (subr))
    set_cfun (DECL_STRUCT_FUNCTION (subr));
  else
    allocate_struct_function (subr, false);
  prepare_function_start ();
  decide_function_section (subr);

  /* Warn if this value is an aggregate type,
     regardless of which calling convention we are using for it.  */
  if (AGGREGATE_TYPE_P (TREE_TYPE (DECL_RESULT (subr))))
    warning (OPT_Waggregate_return, "function returns an aggregate");
}


void
expand_main_function (void)
{
#if (defined(INVOKE__main)                              \
     || (!defined(HAS_INIT_SECTION)                     \
         && !defined(INIT_SECTION_ASM_OP)               \
         && !defined(INIT_ARRAY_SECTION_ASM_OP)))
  emit_library_call (init_one_libfunc (NAME__MAIN), LCT_NORMAL, VOIDmode, 0);
#endif
}
\f
/* Expand code to initialize the stack_protect_guard.  This is invoked at
   the beginning of a function to be protected.  */

#ifndef HAVE_stack_protect_set
# define HAVE_stack_protect_set         0
# define gen_stack_protect_set(x,y)     (gcc_unreachable (), NULL_RTX)
#endif

void
stack_protect_prologue (void)
{
  tree guard_decl = targetm.stack_protect_guard ();
  rtx x, y;

  x = expand_normal (crtl->stack_protect_guard);
  y = expand_normal (guard_decl);

  /* Allow the target to copy from Y to X without leaking Y into a
     register.  */
  if (HAVE_stack_protect_set)
    {
      rtx insn = gen_stack_protect_set (x, y);
      if (insn)
        {
          emit_insn (insn);
          return;
        }
    }

  /* Otherwise do a straight move.  */
  emit_move_insn (x, y);
}

/* Expand code to verify the stack_protect_guard.  This is invoked at
   the end of a function to be protected.  */

#ifndef HAVE_stack_protect_test
# define HAVE_stack_protect_test                0
# define gen_stack_protect_test(x, y, z)        (gcc_unreachable (), NULL_RTX)
#endif

void
stack_protect_epilogue (void)
{
  tree guard_decl = targetm.stack_protect_guard ();
  rtx label = gen_label_rtx ();
  rtx x, y, tmp;

  x = expand_normal (crtl->stack_protect_guard);
  y = expand_normal (guard_decl);

  /* Allow the target to compare Y with X without leaking either into
     a register.  */
  switch (HAVE_stack_protect_test != 0)
    {
    case 1:
      tmp = gen_stack_protect_test (x, y, label);
      if (tmp)
        {
          emit_insn (tmp);
          break;
        }
      /* FALLTHRU */

    default:
      emit_cmp_and_jump_insns (x, y, EQ, NULL_RTX, ptr_mode, 1, label);
      break;
    }

  /* The noreturn predictor has been moved to the tree level.  The rtl-level
     predictors estimate this branch about 20%, which isn't enough to get
     things moved out of line.  Since this is the only extant case of adding
     a noreturn function at the rtl level, it doesn't seem worth doing ought
     except adding the prediction by hand.  */
  tmp = get_last_insn ();
  if (JUMP_P (tmp))
    predict_insn_def (tmp, PRED_NORETURN, TAKEN);

  expand_call (targetm.stack_protect_fail (), NULL_RTX, /*ignore=*/true);
  free_temp_slots ();
  emit_label (label);
}
\f
/* Start the RTL for a new function, and set variables used for
   emitting RTL.
   SUBR is the FUNCTION_DECL node.
   PARMS_HAVE_CLEANUPS is nonzero if there are cleanups associated with
   the function's parameters, which must be run at any return statement.  */

void
expand_function_start (tree subr)
{
  /* Make sure volatile mem refs aren't considered
     valid operands of arithmetic insns.  */
  init_recog_no_volatile ();

  crtl->profile
    = (profile_flag
       && ! DECL_NO_INSTRUMENT_FUNCTION_ENTRY_EXIT (subr));

  crtl->limit_stack
    = (stack_limit_rtx != NULL_RTX && ! DECL_NO_LIMIT_STACK (subr));

  /* Make the label for return statements to jump to.  Do not special
     case machines with special return instructions -- they will be
     handled later during jump, ifcvt, or epilogue creation.  */
  return_label = gen_label_rtx ();

  /* Initialize rtx used to return the value.  */
  /* Do this before assign_parms so that we copy the struct value address
     before any library calls that assign parms might generate.  */

  /* Decide whether to return the value in memory or in a register.  */
  if (aggregate_value_p (DECL_RESULT (subr), subr))
    {
      /* Returning something that won't go in a register.  */
      rtx value_address = 0;

#ifdef PCC_STATIC_STRUCT_RETURN
      if (cfun->returns_pcc_struct)
        {
          int size = int_size_in_bytes (TREE_TYPE (DECL_RESULT (subr)));
          value_address = assemble_static_space (size);
        }
      else
#endif
        {
          rtx sv = targetm.calls.struct_value_rtx (TREE_TYPE (subr), 2);
          /* Expect to be passed the address of a place to store the value.
             If it is passed as an argument, assign_parms will take care of
             it.  */
          if (sv)
            {
              value_address = gen_reg_rtx (Pmode);
              emit_move_insn (value_address, sv);
            }
        }
      if (value_address)
        {
          rtx x = value_address;
          if (!DECL_BY_REFERENCE (DECL_RESULT (subr)))
            {
              x = gen_rtx_MEM (DECL_MODE (DECL_RESULT (subr)), x);
              set_mem_attributes (x, DECL_RESULT (subr), 1);
            }
          SET_DECL_RTL (DECL_RESULT (subr), x);
        }
    }
  else if (DECL_MODE (DECL_RESULT (subr)) == VOIDmode)
    /* If return mode is void, this decl rtl should not be used.  */
    SET_DECL_RTL (DECL_RESULT (subr), NULL_RTX);
  else
    {
      /* Compute the return values into a pseudo reg, which we will copy
         into the true return register after the cleanups are done.  */
      tree return_type = TREE_TYPE (DECL_RESULT (subr));
      if (TYPE_MODE (return_type) != BLKmode
          && targetm.calls.return_in_msb (return_type))
        /* expand_function_end will insert the appropriate padding in
           this case.  Use the return value's natural (unpadded) mode
           within the function proper.  */
        SET_DECL_RTL (DECL_RESULT (subr),
                      gen_reg_rtx (TYPE_MODE (return_type)));
      else
        {
          /* In order to figure out what mode to use for the pseudo, we
             figure out what the mode of the eventual return register will
             actually be, and use that.  */
          rtx hard_reg = hard_function_value (return_type, subr, 0, 1);

          /* Structures that are returned in registers are not
             aggregate_value_p, so we may see a PARALLEL or a REG.  */
          if (REG_P (hard_reg))
            SET_DECL_RTL (DECL_RESULT (subr),
                          gen_reg_rtx (GET_MODE (hard_reg)));
          else
            {
              gcc_assert (GET_CODE (hard_reg) == PARALLEL);
              SET_DECL_RTL (DECL_RESULT (subr), gen_group_rtx (hard_reg));
            }
        }

      /* Set DECL_REGISTER flag so that expand_function_end will copy the
         result to the real return register(s).  */
      DECL_REGISTER (DECL_RESULT (subr)) = 1;
    }

  /* Initialize rtx for parameters and local variables.
     In some cases this requires emitting insns.  */
  assign_parms (subr);

  /* If function gets a static chain arg, store it.  */
  if (cfun->static_chain_decl)
    {
      tree parm = cfun->static_chain_decl;
      rtx local, chain, insn;

      local = gen_reg_rtx (Pmode);
      chain = targetm.calls.static_chain (current_function_decl, true);

      set_decl_incoming_rtl (parm, chain, false);
      SET_DECL_RTL (parm, local);
      mark_reg_pointer (local, TYPE_ALIGN (TREE_TYPE (TREE_TYPE (parm))));

      insn = emit_move_insn (local, chain);

      /* Mark the register as eliminable, similar to parameters.  */
      if (MEM_P (chain)
          && reg_mentioned_p (arg_pointer_rtx, XEXP (chain, 0)))
        set_dst_reg_note (insn, REG_EQUIV, chain, local);
    }

  /* If the function receives a non-local goto, then store the
     bits we need to restore the frame pointer.  */
  if (cfun->nonlocal_goto_save_area)
    {
      tree t_save;
      rtx r_save;

      tree var = TREE_OPERAND (cfun->nonlocal_goto_save_area, 0);
      gcc_assert (DECL_RTL_SET_P (var));

      t_save = build4 (ARRAY_REF,
                       TREE_TYPE (TREE_TYPE (cfun->nonlocal_goto_save_area)),
                       cfun->nonlocal_goto_save_area,
                       integer_zero_node, NULL_TREE, NULL_TREE);
      r_save = expand_expr (t_save, NULL_RTX, VOIDmode, EXPAND_WRITE);
      gcc_assert (GET_MODE (r_save) == Pmode);

      emit_move_insn (r_save, targetm.builtin_setjmp_frame_value ());
      update_nonlocal_goto_save_area ();
    }

  /* The following was moved from init_function_start.
     The move is supposed to make sdb output more accurate.  */
  /* Indicate the beginning of the function body,
     as opposed to parm setup.  */
  emit_note (NOTE_INSN_FUNCTION_BEG);

  gcc_assert (NOTE_P (get_last_insn ()));

  parm_birth_insn = get_last_insn ();

  if (crtl->profile)
    {
#ifdef PROFILE_HOOK
      PROFILE_HOOK (current_function_funcdef_no);
#endif
    }

  /* If we are doing generic stack checking, the probe should go here.  */
  if (flag_stack_check == GENERIC_STACK_CHECK)
    stack_check_probe_note = emit_note (NOTE_INSN_DELETED);
}
\f
/* Undo the effects of init_dummy_function_start.  */
void
expand_dummy_function_end (void)
{
  gcc_assert (in_dummy_function);

  /* End any sequences that failed to be closed due to syntax errors.  */
  while (in_sequence_p ())
    end_sequence ();

  /* Outside function body, can't compute type's actual size
     until next function's body starts.  */

  free_after_parsing (cfun);
  free_after_compilation (cfun);
  pop_cfun ();
  in_dummy_function = false;
}

/* Call DOIT for each hard register used as a return value from
   the current function.  */

void
diddle_return_value (void (*doit) (rtx, void *), void *arg)
{
  rtx outgoing = crtl->return_rtx;

  if (! outgoing)
    return;

  if (REG_P (outgoing))
    (*doit) (outgoing, arg);
  else if (GET_CODE (outgoing) == PARALLEL)
    {
      int i;

      for (i = 0; i < XVECLEN (outgoing, 0); i++)
        {
          rtx x = XEXP (XVECEXP (outgoing, 0, i), 0);

          if (REG_P (x) && REGNO (x) < FIRST_PSEUDO_REGISTER)
            (*doit) (x, arg);
        }
    }
}

static void
do_clobber_return_reg (rtx reg, void *arg ATTRIBUTE_UNUSED)
{
  emit_clobber (reg);
}

void
clobber_return_register (void)
{
  diddle_return_value (do_clobber_return_reg, NULL);

  /* In case we do use pseudo to return value, clobber it too.  */
  if (DECL_RTL_SET_P (DECL_RESULT (current_function_decl)))
    {
      tree decl_result = DECL_RESULT (current_function_decl);
      rtx decl_rtl = DECL_RTL (decl_result);
      if (REG_P (decl_rtl) && REGNO (decl_rtl) >= FIRST_PSEUDO_REGISTER)
        {
          do_clobber_return_reg (decl_rtl, NULL);
        }
    }
}

static void
do_use_return_reg (rtx reg, void *arg ATTRIBUTE_UNUSED)
{
  emit_use (reg);
}

static void
use_return_register (void)
{
  diddle_return_value (do_use_return_reg, NULL);
}

/* Possibly warn about unused parameters.  */
void
do_warn_unused_parameter (tree fn)
{
  tree decl;

  for (decl = DECL_ARGUMENTS (fn);
       decl; decl = DECL_CHAIN (decl))
    if (!TREE_USED (decl) && TREE_CODE (decl) == PARM_DECL
        && DECL_NAME (decl) && !DECL_ARTIFICIAL (decl)
        && !TREE_NO_WARNING (decl))
      warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
}

/* Generate RTL for the end of the current function.  */

void
expand_function_end (void)
{
  rtx clobber_after;

  /* If arg_pointer_save_area was referenced only from a nested
     function, we will not have initialized it yet.  Do that now.  */
  if (arg_pointer_save_area && ! crtl->arg_pointer_save_area_init)
    get_arg_pointer_save_area ();

  /* If we are doing generic stack checking and this function makes calls,
     do a stack probe at the start of the function to ensure we have enough
     space for another stack frame.  */
  if (flag_stack_check == GENERIC_STACK_CHECK)
    {
      rtx insn, seq;

      for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
        if (CALL_P (insn))
          {
            rtx max_frame_size = GEN_INT (STACK_CHECK_MAX_FRAME_SIZE);
            start_sequence ();
            if (STACK_CHECK_MOVING_SP)
              anti_adjust_stack_and_probe (max_frame_size, true);
            else
              probe_stack_range (STACK_OLD_CHECK_PROTECT, max_frame_size);
            seq = get_insns ();
            end_sequence ();
            set_insn_locations (seq, prologue_location);
            emit_insn_before (seq, stack_check_probe_note);
            break;
          }
    }

  /* End any sequences that failed to be closed due to syntax errors.  */
  while (in_sequence_p ())
    end_sequence ();

  clear_pending_stack_adjust ();
  do_pending_stack_adjust ();

  /* Output a linenumber for the end of the function.
     SDB depends on this.  */
  set_curr_insn_location (input_location);

  /* Before the return label (if any), clobber the return
     registers so that they are not propagated live to the rest of
     the function.  This can only happen with functions that drop
     through; if there had been a return statement, there would
     have either been a return rtx, or a jump to the return label.

     We delay actual code generation after the current_function_value_rtx
     is computed.  */
  clobber_after = get_last_insn ();

  /* Output the label for the actual return from the function.  */
  emit_label (return_label);

  if (targetm_common.except_unwind_info (&global_options) == UI_SJLJ)
    {
      /* Let except.c know where it should emit the call to unregister
         the function context for sjlj exceptions.  */
      if (flag_exceptions)
        sjlj_emit_function_exit_after (get_last_insn ());
    }
  else
    {
      /* We want to ensure that instructions that may trap are not
         moved into the epilogue by scheduling, because we don't
         always emit unwind information for the epilogue.  */
      if (cfun->can_throw_non_call_exceptions)
        emit_insn (gen_blockage ());
    }

  /* If this is an implementation of throw, do what's necessary to
     communicate between __builtin_eh_return and the epilogue.  */
  expand_eh_return ();

  /* If scalar return value was computed in a pseudo-reg, or was a named
     return value that got dumped to the stack, copy that to the hard
     return register.  */
  if (DECL_RTL_SET_P (DECL_RESULT (current_function_decl)))
    {
      tree decl_result = DECL_RESULT (current_function_decl);
      rtx decl_rtl = DECL_RTL (decl_result);

      if (REG_P (decl_rtl)
          ? REGNO (decl_rtl) >= FIRST_PSEUDO_REGISTER
          : DECL_REGISTER (decl_result))
        {
          rtx real_decl_rtl = crtl->return_rtx;

          /* This should be set in assign_parms.  */
          gcc_assert (REG_FUNCTION_VALUE_P (real_decl_rtl));

          /* If this is a BLKmode structure being returned in registers,
             then use the mode computed in expand_return.  Note that if
             decl_rtl is memory, then its mode may have been changed,
             but that crtl->return_rtx has not.  */
          if (GET_MODE (real_decl_rtl) == BLKmode)
            PUT_MODE (real_decl_rtl, GET_MODE (decl_rtl));

          /* If a non-BLKmode return value should be padded at the least
             significant end of the register, shift it left by the appropriate
             amount.  BLKmode results are handled using the group load/store
             machinery.  */
          if (TYPE_MODE (TREE_TYPE (decl_result)) != BLKmode
              && REG_P (real_decl_rtl)
              && targetm.calls.return_in_msb (TREE_TYPE (decl_result)))
            {
              emit_move_insn (gen_rtx_REG (GET_MODE (decl_rtl),
                                           REGNO (real_decl_rtl)),
                              decl_rtl);
              shift_return_value (GET_MODE (decl_rtl), true, real_decl_rtl);
            }
          /* If a named return value dumped decl_return to memory, then
             we may need to re-do the PROMOTE_MODE signed/unsigned
             extension.  */
          else if (GET_MODE (real_decl_rtl) != GET_MODE (decl_rtl))
            {
              int unsignedp = TYPE_UNSIGNED (TREE_TYPE (decl_result));
              promote_function_mode (TREE_TYPE (decl_result),
                                     GET_MODE (decl_rtl), &unsignedp,
                                     TREE_TYPE (current_function_decl), 1);

              convert_move (real_decl_rtl, decl_rtl, unsignedp);
            }
          else if (GET_CODE (real_decl_rtl) == PARALLEL)
            {
              /* If expand_function_start has created a PARALLEL for decl_rtl,
                 move the result to the real return registers.  Otherwise, do
                 a group load from decl_rtl for a named return.  */
              if (GET_CODE (decl_rtl) == PARALLEL)
                emit_group_move (real_decl_rtl, decl_rtl);
              else
                emit_group_load (real_decl_rtl, decl_rtl,
                                 TREE_TYPE (decl_result),
                                 int_size_in_bytes (TREE_TYPE (decl_result)));
            }
          /* In the case of complex integer modes smaller than a word, we'll
             need to generate some non-trivial bitfield insertions.  Do that
             on a pseudo and not the hard register.  */
          else if (GET_CODE (decl_rtl) == CONCAT
                   && GET_MODE_CLASS (GET_MODE (decl_rtl)) == MODE_COMPLEX_INT
                   && GET_MODE_BITSIZE (GET_MODE (decl_rtl)) <= BITS_PER_WORD)
            {
              int old_generating_concat_p;
              rtx tmp;

              old_generating_concat_p = generating_concat_p;
              generating_concat_p = 0;
              tmp = gen_reg_rtx (GET_MODE (decl_rtl));
              generating_concat_p = old_generating_concat_p;

              emit_move_insn (tmp, decl_rtl);
              emit_move_insn (real_decl_rtl, tmp);
            }
          else
            emit_move_insn (real_decl_rtl, decl_rtl);
        }
    }

  /* If returning a structure, arrange to return the address of the value
     in a place where debuggers expect to find it.

     If returning a structure PCC style,
     the caller also depends on this value.
     And cfun->returns_pcc_struct is not necessarily set.  */
  if (cfun->returns_struct
      || cfun->returns_pcc_struct)
    {
      rtx value_address = DECL_RTL (DECL_RESULT (current_function_decl));
      tree type = TREE_TYPE (DECL_RESULT (current_function_decl));
      rtx outgoing;

      if (DECL_BY_REFERENCE (DECL_RESULT (current_function_decl)))
        type = TREE_TYPE (type);
      else
        value_address = XEXP (value_address, 0);

      outgoing = targetm.calls.function_value (build_pointer_type (type),
                                               current_function_decl, true);

      /* Mark this as a function return value so integrate will delete the
         assignment and USE below when inlining this function.  */
      REG_FUNCTION_VALUE_P (outgoing) = 1;

      /* The address may be ptr_mode and OUTGOING may be Pmode.  */
      value_address = convert_memory_address (GET_MODE (outgoing),
                                              value_address);

      emit_move_insn (outgoing, value_address);

      /* Show return register used to hold result (in this case the address
         of the result.  */
      crtl->return_rtx = outgoing;
    }

  /* Emit the actual code to clobber return register.  */
  {
    rtx seq;

    start_sequence ();
    clobber_return_register ();
    seq = get_insns ();
    end_sequence ();

    emit_insn_after (seq, clobber_after);
  }

  /* Output the label for the naked return from the function.  */
  if (naked_return_label)
    emit_label (naked_return_label);

  /* @@@ This is a kludge.  We want to ensure that instructions that
     may trap are not moved into the epilogue by scheduling, because
     we don't always emit unwind information for the epilogue.  */
  if (cfun->can_throw_non_call_exceptions
      && targetm_common.except_unwind_info (&global_options) != UI_SJLJ)
    emit_insn (gen_blockage ());

  /* If stack protection is enabled for this function, check the guard.  */
  if (crtl->stack_protect_guard)
    stack_protect_epilogue ();

  /* If we had calls to alloca, and this machine needs
     an accurate stack pointer to exit the function,
     insert some code to save and restore the stack pointer.  */
  if (! EXIT_IGNORE_STACK
      && cfun->calls_alloca)
    {
      rtx tem = 0, seq;

      start_sequence ();
      emit_stack_save (SAVE_FUNCTION, &tem);
      seq = get_insns ();
      end_sequence ();
      emit_insn_before (seq, parm_birth_insn);

      emit_stack_restore (SAVE_FUNCTION, tem);
    }

  /* ??? This should no longer be necessary since stupid is no longer with
     us, but there are some parts of the compiler (eg reload_combine, and
     sh mach_dep_reorg) that still try and compute their own lifetime info
     instead of using the general framework.  */
  use_return_register ();
}

rtx
get_arg_pointer_save_area (void)
{
  rtx ret = arg_pointer_save_area;

  if (! ret)
    {
      ret = assign_stack_local (Pmode, GET_MODE_SIZE (Pmode), 0);
      arg_pointer_save_area = ret;
    }

  if (! crtl->arg_pointer_save_area_init)
    {
      rtx seq;

      /* Save the arg pointer at the beginning of the function.  The
         generated stack slot may not be a valid memory address, so we
         have to check it and fix it if necessary.  */
      start_sequence ();
      emit_move_insn (validize_mem (ret),
                      crtl->args.internal_arg_pointer);
      seq = get_insns ();
      end_sequence ();

      push_topmost_sequence ();
      emit_insn_after (seq, entry_of_function ());
      pop_topmost_sequence ();

      crtl->arg_pointer_save_area_init = true;
    }

  return ret;
}
\f
/* Add a list of INSNS to the hash HASHP, possibly allocating HASHP
   for the first time.  */

static void
record_insns (rtx insns, rtx end, htab_t *hashp)
{
  rtx tmp;
  htab_t hash = *hashp;

  if (hash == NULL)
    *hashp = hash
      = htab_create_ggc (17, htab_hash_pointer, htab_eq_pointer, NULL);

  for (tmp = insns; tmp != end; tmp = NEXT_INSN (tmp))
    {
      void **slot = htab_find_slot (hash, tmp, INSERT);
      gcc_assert (*slot == NULL);
      *slot = tmp;
    }
}

/* INSN has been duplicated or replaced by as COPY, perhaps by duplicating a
   basic block, splitting or peepholes.  If INSN is a prologue or epilogue
   insn, then record COPY as well.  */

void
maybe_copy_prologue_epilogue_insn (rtx insn, rtx copy)
{
  htab_t hash;
  void **slot;

  hash = epilogue_insn_hash;
  if (!hash || !htab_find (hash, insn))
    {
      hash = prologue_insn_hash;
      if (!hash || !htab_find (hash, insn))
        return;
    }

  slot = htab_find_slot (hash, copy, INSERT);
  gcc_assert (*slot == NULL);
  *slot = copy;
}

/* Set the location of the insn chain starting at INSN to LOC.  */
static void
set_insn_locations (rtx insn, int loc)
{
  while (insn != NULL_RTX)
    {
      if (INSN_P (insn))
        INSN_LOCATION (insn) = loc;
      insn = NEXT_INSN (insn);
    }
}

/* Determine if any INSNs in HASH are, or are part of, INSN.  Because
   we can be running after reorg, SEQUENCE rtl is possible.  */

static bool
contains (const_rtx insn, htab_t hash)
{
  if (hash == NULL)
    return false;

  if (NONJUMP_INSN_P (insn) && GET_CODE (PATTERN (insn)) == SEQUENCE)
    {
      int i;
      for (i = XVECLEN (PATTERN (insn), 0) - 1; i >= 0; i--)
        if (htab_find (hash, XVECEXP (PATTERN (insn), 0, i)))
          return true;
      return false;
    }

  return htab_find (hash, insn) != NULL;
}

int
prologue_epilogue_contains (const_rtx insn)
{
  if (contains (insn, prologue_insn_hash))
    return 1;
  if (contains (insn, epilogue_insn_hash))
    return 1;
  return 0;
}

#ifdef HAVE_simple_return

/* Return true if INSN requires the stack frame to be set up.
   PROLOGUE_USED contains the hard registers used in the function
   prologue.  SET_UP_BY_PROLOGUE is the set of registers we expect the
   prologue to set up for the function.  */
bool
requires_stack_frame_p (rtx insn, HARD_REG_SET prologue_used,
                        HARD_REG_SET set_up_by_prologue)
{
  df_ref *df_rec;
  HARD_REG_SET hardregs;
  unsigned regno;

  if (CALL_P (insn))
    return !SIBLING_CALL_P (insn);

  /* We need a frame to get the unique CFA expected by the unwinder.  */
  if (cfun->can_throw_non_call_exceptions && can_throw_internal (insn))
    return true;

  CLEAR_HARD_REG_SET (hardregs);
  for (df_rec = DF_INSN_DEFS (insn); *df_rec; df_rec++)
    {
      rtx dreg = DF_REF_REG (*df_rec);

      if (!REG_P (dreg))
        continue;

      add_to_hard_reg_set (&hardregs, GET_MODE (dreg),
                           REGNO (dreg));
    }
  if (hard_reg_set_intersect_p (hardregs, prologue_used))
    return true;
  AND_COMPL_HARD_REG_SET (hardregs, call_used_reg_set);
  for (regno = 0; regno < FIRST_PSEUDO_REGISTER; regno++)
    if (TEST_HARD_REG_BIT (hardregs, regno)
        && df_regs_ever_live_p (regno))
      return true;

  for (df_rec = DF_INSN_USES (insn); *df_rec; df_rec++)
    {
      rtx reg = DF_REF_REG (*df_rec);

      if (!REG_P (reg))
        continue;

      add_to_hard_reg_set (&hardregs, GET_MODE (reg),
                           REGNO (reg));
    }
  if (hard_reg_set_intersect_p (hardregs, set_up_by_prologue))
    return true;

  return false;
}

/* See whether BB has a single successor that uses [REGNO, END_REGNO),
   and if BB is its only predecessor.  Return that block if so,
   otherwise return null.  */

static basic_block
next_block_for_reg (basic_block bb, int regno, int end_regno)
{
  edge e, live_edge;
  edge_iterator ei;
  bitmap live;
  int i;

  live_edge = NULL;
  FOR_EACH_EDGE (e, ei, bb->succs)
    {
      live = df_get_live_in (e->dest);
      for (i = regno; i < end_regno; i++)
        if (REGNO_REG_SET_P (live, i))
          {
            if (live_edge && live_edge != e)
              return NULL;
            live_edge = e;
          }
    }

  /* We can sometimes encounter dead code.  Don't try to move it
     into the exit block.  */
  if (!live_edge || live_edge->dest == EXIT_BLOCK_PTR)
    return NULL;

  /* Reject targets of abnormal edges.  This is needed for correctness
     on ports like Alpha and MIPS, whose pic_offset_table_rtx can die on
     exception edges even though it is generally treated as call-saved
     for the majority of the compilation.  Moving across abnormal edges
     isn't going to be interesting for shrink-wrap usage anyway.  */
  if (live_edge->flags & EDGE_ABNORMAL)
    return NULL;

  if (EDGE_COUNT (live_edge->dest->preds) > 1)
    return NULL;

  return live_edge->dest;
}

/* Try to move INSN from BB to a successor.  Return true on success.
   USES and DEFS are the set of registers that are used and defined
   after INSN in BB.  */

static bool
move_insn_for_shrink_wrap (basic_block bb, rtx insn,
                           const HARD_REG_SET uses,
                           const HARD_REG_SET defs)
{
  rtx set, src, dest;
  bitmap live_out, live_in, bb_uses, bb_defs;
  unsigned int i, dregno, end_dregno, sregno, end_sregno;
  basic_block next_block;

  /* Look for a simple register copy.  */
  set = single_set (insn);
  if (!set)
    return false;
  src = SET_SRC (set);
  dest = SET_DEST (set);
  if (!REG_P (dest) || !REG_P (src))
    return false;

  /* Make sure that the source register isn't defined later in BB.  */
  sregno = REGNO (src);
  end_sregno = END_REGNO (src);
  if (overlaps_hard_reg_set_p (defs, GET_MODE (src), sregno))
    return false;

  /* Make sure that the destination register isn't referenced later in BB.  */
  dregno = REGNO (dest);
  end_dregno = END_REGNO (dest);
  if (overlaps_hard_reg_set_p (uses, GET_MODE (dest), dregno)
      || overlaps_hard_reg_set_p (defs, GET_MODE (dest), dregno))
    return false;

  /* See whether there is a successor block to which we could move INSN.  */
  next_block = next_block_for_reg (bb, dregno, end_dregno);
  if (!next_block)
    return false;

  /* At this point we are committed to moving INSN, but let's try to
     move it as far as we can.  */
  do
    {
      live_out = df_get_live_out (bb);
      live_in = df_get_live_in (next_block);
      bb = next_block;

      /* Check whether BB uses DEST or clobbers DEST.  We need to add
         INSN to BB if so.  Either way, DEST is no longer live on entry,
         except for any part that overlaps SRC (next loop).  */
      bb_uses = &DF_LR_BB_INFO (bb)->use;
      bb_defs = &DF_LR_BB_INFO (bb)->def;
      if (df_live)
        {
          for (i = dregno; i < end_dregno; i++)
            {
              if (REGNO_REG_SET_P (bb_uses, i) || REGNO_REG_SET_P (bb_defs, i)
                  || REGNO_REG_SET_P (&DF_LIVE_BB_INFO (bb)->gen, i))
                next_block = NULL;
              CLEAR_REGNO_REG_SET (live_out, i);
              CLEAR_REGNO_REG_SET (live_in, i);
            }

          /* Check whether BB clobbers SRC.  We need to add INSN to BB if so.
             Either way, SRC is now live on entry.  */
          for (i = sregno; i < end_sregno; i++)
            {
              if (REGNO_REG_SET_P (bb_defs, i)
                  || REGNO_REG_SET_P (&DF_LIVE_BB_INFO (bb)->gen, i))
                next_block = NULL;
              SET_REGNO_REG_SET (live_out, i);
              SET_REGNO_REG_SET (live_in, i);
            }
        }
      else
        {
          /* DF_LR_BB_INFO (bb)->def does not comprise the DF_REF_PARTIAL and
             DF_REF_CONDITIONAL defs.  So if DF_LIVE doesn't exist, i.e.
             at -O1, just give up searching NEXT_BLOCK.  */
          next_block = NULL;
          for (i = dregno; i < end_dregno; i++)
            {
              CLEAR_REGNO_REG_SET (live_out, i);
              CLEAR_REGNO_REG_SET (live_in, i);
            }

          for (i = sregno; i < end_sregno; i++)
            {
              SET_REGNO_REG_SET (live_out, i);
              SET_REGNO_REG_SET (live_in, i);
            }
        }

      /* If we don't need to add the move to BB, look for a single
         successor block.  */
      if (next_block)
        next_block = next_block_for_reg (next_block, dregno, end_dregno);
    }
  while (next_block);

  /* BB now defines DEST.  It only uses the parts of DEST that overlap SRC
     (next loop).  */
  for (i = dregno; i < end_dregno; i++)
    {
      CLEAR_REGNO_REG_SET (bb_uses, i);
      SET_REGNO_REG_SET (bb_defs, i);
    }

  /* BB now uses SRC.  */
  for (i = sregno; i < end_sregno; i++)
    SET_REGNO_REG_SET (bb_uses, i);

  emit_insn_after (PATTERN (insn), bb_note (bb));
  delete_insn (insn);
  return true;
}

/* Look for register copies in the first block of the function, and move
   them down into successor blocks if the register is used only on one
   path.  This exposes more opportunities for shrink-wrapping.  These
   kinds of sets often occur when incoming argument registers are moved
   to call-saved registers because their values are live across one or
   more calls during the function.  */

static void
prepare_shrink_wrap (basic_block entry_block)
{
  rtx insn, curr, x;
  HARD_REG_SET uses, defs;
  df_ref *ref;

  CLEAR_HARD_REG_SET (uses);
  CLEAR_HARD_REG_SET (defs);
  FOR_BB_INSNS_REVERSE_SAFE (entry_block, insn, curr)
    if (NONDEBUG_INSN_P (insn)
        && !move_insn_for_shrink_wrap (entry_block, insn, uses, defs))
      {
        /* Add all defined registers to DEFs.  */
        for (ref = DF_INSN_DEFS (insn); *ref; ref++)
          {
            x = DF_REF_REG (*ref);
            if (REG_P (x) && HARD_REGISTER_P (x))
              SET_HARD_REG_BIT (defs, REGNO (x));
          }

        /* Add all used registers to USESs.  */
        for (ref = DF_INSN_USES (insn); *ref; ref++)
          {
            x = DF_REF_REG (*ref);
            if (REG_P (x) && HARD_REGISTER_P (x))
              SET_HARD_REG_BIT (uses, REGNO (x));
          }
      }
}

#endif

#ifdef HAVE_return
/* Insert use of return register before the end of BB.  */

static void
emit_use_return_register_into_block (basic_block bb)
{
  rtx seq, insn;
  start_sequence ();
  use_return_register ();
  seq = get_insns ();
  end_sequence ();
  insn = BB_END (bb);
#ifdef HAVE_cc0
  if (reg_mentioned_p (cc0_rtx, PATTERN (insn)))
    insn = prev_cc0_setter (insn);
#endif
  emit_insn_before (seq, insn);
}


/* Create a return pattern, either simple_return or return, depending on
   simple_p.  */

static rtx
gen_return_pattern (bool simple_p)
{
#ifdef HAVE_simple_return
  return simple_p ? gen_simple_return () : gen_return ();
#else
  gcc_assert (!simple_p);
  return gen_return ();
#endif
}

/* Insert an appropriate return pattern at the end of block BB.  This
   also means updating block_for_insn appropriately.  SIMPLE_P is
   the same as in gen_return_pattern and passed to it.  */

static void
emit_return_into_block (bool simple_p, basic_block bb)
{
  rtx jump, pat;
  jump = emit_jump_insn_after (gen_return_pattern (simple_p), BB_END (bb));
  pat = PATTERN (jump);
  if (GET_CODE (pat) == PARALLEL)
    pat = XVECEXP (pat, 0, 0);
  gcc_assert (ANY_RETURN_P (pat));
  JUMP_LABEL (jump) = pat;
}
#endif

/* Set JUMP_LABEL for a return insn.  */

void
set_return_jump_label (rtx returnjump)
{
  rtx pat = PATTERN (returnjump);
  if (GET_CODE (pat) == PARALLEL)
    pat = XVECEXP (pat, 0, 0);
  if (ANY_RETURN_P (pat))
    JUMP_LABEL (returnjump) = pat;
  else
    JUMP_LABEL (returnjump) = ret_rtx;
}

#ifdef HAVE_simple_return
/* Create a copy of BB instructions and insert at BEFORE.  Redirect
   preds of BB to COPY_BB if they don't appear in NEED_PROLOGUE.  */
static void
dup_block_and_redirect (basic_block bb, basic_block copy_bb, rtx before,
                        bitmap_head *need_prologue)
{
  edge_iterator ei;
  edge e;
  rtx insn = BB_END (bb);

  /* We know BB has a single successor, so there is no need to copy a
     simple jump at the end of BB.  */
  if (simplejump_p (insn))
    insn = PREV_INSN (insn);

  start_sequence ();
  duplicate_insn_chain (BB_HEAD (bb), insn);
  if (dump_file)
    {
      unsigned count = 0;
      for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
        if (active_insn_p (insn))
          ++count;
      fprintf (dump_file, "Duplicating bb %d to bb %d, %u active insns.\n",
               bb->index, copy_bb->index, count);
    }
  insn = get_insns ();
  end_sequence ();
  emit_insn_before (insn, before);

  /* Redirect all the paths that need no prologue into copy_bb.  */
  for (ei = ei_start (bb->preds); (e = ei_safe_edge (ei)); )
    if (!bitmap_bit_p (need_prologue, e->src->index))
      {
        int freq = EDGE_FREQUENCY (e);
        copy_bb->count += e->count;
        copy_bb->frequency += EDGE_FREQUENCY (e);
        e->dest->count -= e->count;
        if (e->dest->count < 0)
          e->dest->count = 0;
        e->dest->frequency -= freq;
        if (e->dest->frequency < 0)
          e->dest->frequency = 0;
        redirect_edge_and_branch_force (e, copy_bb);
        continue;
      }
    else
      ei_next (&ei);
}
#endif

#if defined (HAVE_return) || defined (HAVE_simple_return)
/* Return true if there are any active insns between HEAD and TAIL.  */
static bool
active_insn_between (rtx head, rtx tail)
{
  while (tail)
    {
      if (active_insn_p (tail))
        return true;
      if (tail == head)
        return false;
      tail = PREV_INSN (tail);
    }
  return false;
}

/* LAST_BB is a block that exits, and empty of active instructions.
   Examine its predecessors for jumps that can be converted to
   (conditional) returns.  */
static vec<edge> 
convert_jumps_to_returns (basic_block last_bb, bool simple_p,
                          vec<edge> unconverted ATTRIBUTE_UNUSED)
{
  int i;
  basic_block bb;
  rtx label;
  edge_iterator ei;
  edge e;
  vec<basic_block> src_bbs;

  src_bbs.create (EDGE_COUNT (last_bb->preds));
  FOR_EACH_EDGE (e, ei, last_bb->preds)
    if (e->src != ENTRY_BLOCK_PTR)
      src_bbs.quick_push (e->src);

  label = BB_HEAD (last_bb);

  FOR_EACH_VEC_ELT (src_bbs, i, bb)
    {
      rtx jump = BB_END (bb);

      if (!JUMP_P (jump) || JUMP_LABEL (jump) != label)
        continue;

      e = find_edge (bb, last_bb);

      /* If we have an unconditional jump, we can replace that
         with a simple return instruction.  */
      if (simplejump_p (jump))
        {
          /* The use of the return register might be present in the exit
             fallthru block.  Either:
             - removing the use is safe, and we should remove the use in
             the exit fallthru block, or
             - removing the use is not safe, and we should add it here.
             For now, we conservatively choose the latter.  Either of the
             2 helps in crossjumping.  */
          emit_use_return_register_into_block (bb);

          emit_return_into_block (simple_p, bb);
          delete_insn (jump);
        }

      /* If we have a conditional jump branching to the last
         block, we can try to replace that with a conditional
         return instruction.  */
      else if (condjump_p (jump))
        {
          rtx dest;

          if (simple_p)
            dest = simple_return_rtx;
          else
            dest = ret_rtx;
          if (!redirect_jump (jump, dest, 0))
            {
#ifdef HAVE_simple_return
              if (simple_p)
                {
                  if (dump_file)
                    fprintf (dump_file,
                             "Failed to redirect bb %d branch.\n", bb->index);
                  unconverted.safe_push (e);
                }
#endif
              continue;
            }

          /* See comment in simplejump_p case above.  */
          emit_use_return_register_into_block (bb);

          /* If this block has only one successor, it both jumps
             and falls through to the fallthru block, so we can't
             delete the edge.  */
          if (single_succ_p (bb))
            continue;
        }
      else
        {
#ifdef HAVE_simple_return
          if (simple_p)
            {
              if (dump_file)
                fprintf (dump_file,
                         "Failed to redirect bb %d branch.\n", bb->index);
              unconverted.safe_push (e);
            }
#endif
          continue;
        }

      /* Fix up the CFG for the successful change we just made.  */
      redirect_edge_succ (e, EXIT_BLOCK_PTR);
      e->flags &= ~EDGE_CROSSING;
    }
  src_bbs.release ();
  return unconverted;
}

/* Emit a return insn for the exit fallthru block.  */
static basic_block
emit_return_for_exit (edge exit_fallthru_edge, bool simple_p)
{
  basic_block last_bb = exit_fallthru_edge->src;

  if (JUMP_P (BB_END (last_bb)))
    {
      last_bb = split_edge (exit_fallthru_edge);
      exit_fallthru_edge = single_succ_edge (last_bb);
    }
  emit_barrier_after (BB_END (last_bb));
  emit_return_into_block (simple_p, last_bb);
  exit_fallthru_edge->flags &= ~EDGE_FALLTHRU;
  return last_bb;
}
#endif


/* Generate the prologue and epilogue RTL if the machine supports it.  Thread
   this into place with notes indicating where the prologue ends and where
   the epilogue begins.  Update the basic block information when possible.

   Notes on epilogue placement:
   There are several kinds of edges to the exit block:
   * a single fallthru edge from LAST_BB
   * possibly, edges from blocks containing sibcalls
   * possibly, fake edges from infinite loops

   The epilogue is always emitted on the fallthru edge from the last basic
   block in the function, LAST_BB, into the exit block.

   If LAST_BB is empty except for a label, it is the target of every
   other basic block in the function that ends in a return.  If a
   target has a return or simple_return pattern (possibly with
   conditional variants), these basic blocks can be changed so that a
   return insn is emitted into them, and their target is adjusted to
   the real exit block.

   Notes on shrink wrapping: We implement a fairly conservative
   version of shrink-wrapping rather than the textbook one.  We only
   generate a single prologue and a single epilogue.  This is
   sufficient to catch a number of interesting cases involving early
   exits.

   First, we identify the blocks that require the prologue to occur before
   them.  These are the ones that modify a call-saved register, or reference
   any of the stack or frame pointer registers.  To simplify things, we then
   mark everything reachable from these blocks as also requiring a prologue.
   This takes care of loops automatically, and avoids the need to examine
   whether MEMs reference the frame, since it is sufficient to check for
   occurrences of the stack or frame pointer.

   We then compute the set of blocks for which the need for a prologue
   is anticipatable (borrowing terminology from the shrink-wrapping
   description in Muchnick's book).  These are the blocks which either
   require a prologue themselves, or those that have only successors
   where the prologue is anticipatable.  The prologue needs to be
   inserted on all edges from BB1->BB2 where BB2 is in ANTIC and BB1
   is not.  For the moment, we ensure that only one such edge exists.

   The epilogue is placed as described above, but we make a
   distinction between inserting return and simple_return patterns
   when modifying other blocks that end in a return.  Blocks that end
   in a sibcall omit the sibcall_epilogue if the block is not in
   ANTIC.  */

static void
thread_prologue_and_epilogue_insns (void)
{
  bool inserted;
#ifdef HAVE_simple_return
  vec<edge> unconverted_simple_returns = vNULL;
  bool nonempty_prologue;
  bitmap_head bb_flags;
  unsigned max_grow_size;
#endif
  rtx returnjump;
  rtx seq ATTRIBUTE_UNUSED, epilogue_end ATTRIBUTE_UNUSED;
  rtx prologue_seq ATTRIBUTE_UNUSED, split_prologue_seq ATTRIBUTE_UNUSED;
  edge e, entry_edge, orig_entry_edge, exit_fallthru_edge;
  edge_iterator ei;

  df_analyze ();

  rtl_profile_for_bb (ENTRY_BLOCK_PTR);

  inserted = false;
  seq = NULL_RTX;
  epilogue_end = NULL_RTX;
  returnjump = NULL_RTX;

  /* Can't deal with multiple successors of the entry block at the
     moment.  Function should always have at least one entry
     point.  */
  gcc_assert (single_succ_p (ENTRY_BLOCK_PTR));
  entry_edge = single_succ_edge (ENTRY_BLOCK_PTR);
  orig_entry_edge = entry_edge;

  split_prologue_seq = NULL_RTX;
  if (flag_split_stack
      && (lookup_attribute ("no_split_stack", DECL_ATTRIBUTES (cfun->decl))
          == NULL))
    {
#ifndef HAVE_split_stack_prologue
      gcc_unreachable ();
#else
      gcc_assert (HAVE_split_stack_prologue);

      start_sequence ();
      emit_insn (gen_split_stack_prologue ());
      split_prologue_seq = get_insns ();
      end_sequence ();

      record_insns (split_prologue_seq, NULL, &prologue_insn_hash);
      set_insn_locations (split_prologue_seq, prologue_location);
#endif
    }

  prologue_seq = NULL_RTX;
#ifdef HAVE_prologue
  if (HAVE_prologue)
    {
      start_sequence ();
      seq = gen_prologue ();
      emit_insn (seq);

      /* Insert an explicit USE for the frame pointer
         if the profiling is on and the frame pointer is required.  */
      if (crtl->profile && frame_pointer_needed)
        emit_use (hard_frame_pointer_rtx);

      /* Retain a map of the prologue insns.  */
      record_insns (seq, NULL, &prologue_insn_hash);
      emit_note (NOTE_INSN_PROLOGUE_END);

      /* Ensure that instructions are not moved into the prologue when
         profiling is on.  The call to the profiling routine can be
         emitted within the live range of a call-clobbered register.  */
      if (!targetm.profile_before_prologue () && crtl->profile)
        emit_insn (gen_blockage ());

      prologue_seq = get_insns ();
      end_sequence ();
      set_insn_locations (prologue_seq, prologue_location);
    }
#endif

#ifdef HAVE_simple_return
  bitmap_initialize (&bb_flags, &bitmap_default_obstack);

  /* Try to perform a kind of shrink-wrapping, making sure the
     prologue/epilogue is emitted only around those parts of the
     function that require it.  */

  nonempty_prologue = false;
  for (seq = prologue_seq; seq; seq = NEXT_INSN (seq))
    if (!NOTE_P (seq) || NOTE_KIND (seq) != NOTE_INSN_PROLOGUE_END)
      {
        nonempty_prologue = true;
        break;
      }
      
  if (flag_shrink_wrap && HAVE_simple_return
      && (targetm.profile_before_prologue () || !crtl->profile)
      && nonempty_prologue && !crtl->calls_eh_return)
    {
      HARD_REG_SET prologue_clobbered, prologue_used, live_on_edge;
      struct hard_reg_set_container set_up_by_prologue;
      rtx p_insn;
      vec<basic_block> vec;
      basic_block bb;
      bitmap_head bb_antic_flags;
      bitmap_head bb_on_list;
      bitmap_head bb_tail;

      if (dump_file)
        fprintf (dump_file, "Attempting shrink-wrapping optimization.\n");

      /* Compute the registers set and used in the prologue.  */
      CLEAR_HARD_REG_SET (prologue_clobbered);
      CLEAR_HARD_REG_SET (prologue_used);
      for (p_insn = prologue_seq; p_insn; p_insn = NEXT_INSN (p_insn))
        {
          HARD_REG_SET this_used;
          if (!NONDEBUG_INSN_P (p_insn))
            continue;

          CLEAR_HARD_REG_SET (this_used);
          note_uses (&PATTERN (p_insn), record_hard_reg_uses,
                     &this_used);
          AND_COMPL_HARD_REG_SET (this_used, prologue_clobbered);
          IOR_HARD_REG_SET (prologue_used, this_used);
          note_stores (PATTERN (p_insn), record_hard_reg_sets,
                       &prologue_clobbered);
        }

      prepare_shrink_wrap (entry_edge->dest);

      bitmap_initialize (&bb_antic_flags, &bitmap_default_obstack);
      bitmap_initialize (&bb_on_list, &bitmap_default_obstack);
      bitmap_initialize (&bb_tail, &bitmap_default_obstack);

      /* Find the set of basic blocks that require a stack frame,
         and blocks that are too big to be duplicated.  */

      vec.create (n_basic_blocks);

      CLEAR_HARD_REG_SET (set_up_by_prologue.set);
      add_to_hard_reg_set (&set_up_by_prologue.set, Pmode,
                           STACK_POINTER_REGNUM);
      add_to_hard_reg_set (&set_up_by_prologue.set, Pmode, ARG_POINTER_REGNUM);
      if (frame_pointer_needed)
        add_to_hard_reg_set (&set_up_by_prologue.set, Pmode,
                             HARD_FRAME_POINTER_REGNUM);
      if (pic_offset_table_rtx)
        add_to_hard_reg_set (&set_up_by_prologue.set, Pmode,
                             PIC_OFFSET_TABLE_REGNUM);
      if (crtl->drap_reg)
        add_to_hard_reg_set (&set_up_by_prologue.set,
                             GET_MODE (crtl->drap_reg),
                             REGNO (crtl->drap_reg));
      if (targetm.set_up_by_prologue)
        targetm.set_up_by_prologue (&set_up_by_prologue);

      /* We don't use a different max size depending on
         optimize_bb_for_speed_p because increasing shrink-wrapping
         opportunities by duplicating tail blocks can actually result
         in an overall decrease in code size.  */
      max_grow_size = get_uncond_jump_length ();
      max_grow_size *= PARAM_VALUE (PARAM_MAX_GROW_COPY_BB_INSNS);

      FOR_EACH_BB (bb)
        {
          rtx insn;
          unsigned size = 0;

          FOR_BB_INSNS (bb, insn)
            if (NONDEBUG_INSN_P (insn))
              {
                if (requires_stack_frame_p (insn, prologue_used,
                                            set_up_by_prologue.set))
                  {
                    if (bb == entry_edge->dest)
                      goto fail_shrinkwrap;
                    bitmap_set_bit (&bb_flags, bb->index);
                    vec.quick_push (bb);
                    break;
                  }
                else if (size <= max_grow_size)
                  {
                    size += get_attr_min_length (insn);
                    if (size > max_grow_size)
                      bitmap_set_bit (&bb_on_list, bb->index);
                  }
              }
        }

      /* Blocks that really need a prologue, or are too big for tails.  */
      bitmap_ior_into (&bb_on_list, &bb_flags);

      /* For every basic block that needs a prologue, mark all blocks
         reachable from it, so as to ensure they are also seen as
         requiring a prologue.  */
      while (!vec.is_empty ())
        {
          basic_block tmp_bb = vec.pop ();

          FOR_EACH_EDGE (e, ei, tmp_bb->succs)
            if (e->dest != EXIT_BLOCK_PTR
                && bitmap_set_bit (&bb_flags, e->dest->index))
              vec.quick_push (e->dest);
        }

      /* Find the set of basic blocks that need no prologue, have a
         single successor, can be duplicated, meet a max size
         requirement, and go to the exit via like blocks.  */
      vec.quick_push (EXIT_BLOCK_PTR);
      while (!vec.is_empty ())
        {
          basic_block tmp_bb = vec.pop ();

          FOR_EACH_EDGE (e, ei, tmp_bb->preds)
            if (single_succ_p (e->src)
                && !bitmap_bit_p (&bb_on_list, e->src->index)
                && can_duplicate_block_p (e->src))
              {
                edge pe;
                edge_iterator pei;

                /* If there is predecessor of e->src which doesn't
                   need prologue and the edge is complex,
                   we might not be able to redirect the branch
                   to a copy of e->src.  */
                FOR_EACH_EDGE (pe, pei, e->src->preds)
                  if ((pe->flags & EDGE_COMPLEX) != 0
                      && !bitmap_bit_p (&bb_flags, pe->src->index))
                    break;
                if (pe == NULL && bitmap_set_bit (&bb_tail, e->src->index))
                  vec.quick_push (e->src);
              }
        }

      /* Now walk backwards from every block that is marked as needing
         a prologue to compute the bb_antic_flags bitmap.  Exclude
         tail blocks; They can be duplicated to be used on paths not
         needing a prologue.  */
      bitmap_clear (&bb_on_list);
      bitmap_and_compl (&bb_antic_flags, &bb_flags, &bb_tail);
      FOR_EACH_BB (bb)
        {
          if (!bitmap_bit_p (&bb_antic_flags, bb->index))
            continue;
          FOR_EACH_EDGE (e, ei, bb->preds)
            if (!bitmap_bit_p (&bb_antic_flags, e->src->index)
                && bitmap_set_bit (&bb_on_list, e->src->index))
              vec.quick_push (e->src);
        }
      while (!vec.is_empty ())
        {
          basic_block tmp_bb = vec.pop ();
          bool all_set = true;

          bitmap_clear_bit (&bb_on_list, tmp_bb->index);
          FOR_EACH_EDGE (e, ei, tmp_bb->succs)
            if (!bitmap_bit_p (&bb_antic_flags, e->dest->index))
              {
                all_set = false;
                break;
              }

          if (all_set)
            {
              bitmap_set_bit (&bb_antic_flags, tmp_bb->index);
              FOR_EACH_EDGE (e, ei, tmp_bb->preds)
                if (!bitmap_bit_p (&bb_antic_flags, e->src->index)
                    && bitmap_set_bit (&bb_on_list, e->src->index))
                  vec.quick_push (e->src);
            }
        }
      /* Find exactly one edge that leads to a block in ANTIC from
         a block that isn't.  */
      if (!bitmap_bit_p (&bb_antic_flags, entry_edge->dest->index))
        FOR_EACH_BB (bb)
          {
            if (!bitmap_bit_p (&bb_antic_flags, bb->index))
              continue;
            FOR_EACH_EDGE (e, ei, bb->preds)
              if (!bitmap_bit_p (&bb_antic_flags, e->src->index))
                {
                  if (entry_edge != orig_entry_edge)
                    {
                      entry_edge = orig_entry_edge;
                      if (dump_file)
                        fprintf (dump_file, "More than one candidate edge.\n");
                      goto fail_shrinkwrap;
                    }
                  if (dump_file)
                    fprintf (dump_file, "Found candidate edge for "
                             "shrink-wrapping, %d->%d.\n", e->src->index,
                             e->dest->index);
                  entry_edge = e;
                }
          }

      if (entry_edge != orig_entry_edge)
        {
          /* Test whether the prologue is known to clobber any register
             (other than FP or SP) which are live on the edge.  */
          CLEAR_HARD_REG_BIT (prologue_clobbered, STACK_POINTER_REGNUM);
          if (frame_pointer_needed)
            CLEAR_HARD_REG_BIT (prologue_clobbered, HARD_FRAME_POINTER_REGNUM);
          REG_SET_TO_HARD_REG_SET (live_on_edge,
                                   df_get_live_in (entry_edge->dest));
          if (hard_reg_set_intersect_p (live_on_edge, prologue_clobbered))
            {
              entry_edge = orig_entry_edge;
              if (dump_file)
                fprintf (dump_file,
                         "Shrink-wrapping aborted due to clobber.\n");
            }
        }
      if (entry_edge != orig_entry_edge)
        {
          crtl->shrink_wrapped = true;
          if (dump_file)
            fprintf (dump_file, "Performing shrink-wrapping.\n");

          /* Find tail blocks reachable from both blocks needing a
             prologue and blocks not needing a prologue.  */
          if (!bitmap_empty_p (&bb_tail))
            FOR_EACH_BB (bb)
              {
                bool some_pro, some_no_pro;
                if (!bitmap_bit_p (&bb_tail, bb->index))
                  continue;
                some_pro = some_no_pro = false;
                FOR_EACH_EDGE (e, ei, bb->preds)
                  {
                    if (bitmap_bit_p (&bb_flags, e->src->index))
                      some_pro = true;
                    else
                      some_no_pro = true;
                  }
                if (some_pro && some_no_pro)
                  vec.quick_push (bb);
                else
                  bitmap_clear_bit (&bb_tail, bb->index);
              }
          /* Find the head of each tail.  */
          while (!vec.is_empty ())
            {
              basic_block tbb = vec.pop ();

              if (!bitmap_bit_p (&bb_tail, tbb->index))
                continue;

              while (single_succ_p (tbb))
                {
                  tbb = single_succ (tbb);
                  bitmap_clear_bit (&bb_tail, tbb->index);
                }
            }
          /* Now duplicate the tails.  */
          if (!bitmap_empty_p (&bb_tail))
            FOR_EACH_BB_REVERSE (bb)
              {
                basic_block copy_bb, tbb;
                rtx insert_point;
                int eflags;

                if (!bitmap_clear_bit (&bb_tail, bb->index))
                  continue;

                /* Create a copy of BB, instructions and all, for
                   use on paths that don't need a prologue.
                   Ideal placement of the copy is on a fall-thru edge
                   or after a block that would jump to the copy.  */ 
                FOR_EACH_EDGE (e, ei, bb->preds)
                  if (!bitmap_bit_p (&bb_flags, e->src->index)
                      && single_succ_p (e->src))
                    break;
                if (e)
                  {
                    /* Make sure we insert after any barriers.  */
                    rtx end = get_last_bb_insn (e->src);
                    copy_bb = create_basic_block (NEXT_INSN (end),
                                                  NULL_RTX, e->src);
                    BB_COPY_PARTITION (copy_bb, e->src);
                  }
                else
                  {
                    /* Otherwise put the copy at the end of the function.  */
                    copy_bb = create_basic_block (NULL_RTX, NULL_RTX,
                                                  EXIT_BLOCK_PTR->prev_bb);
                    BB_COPY_PARTITION (copy_bb, bb);
                  }

                insert_point = emit_note_after (NOTE_INSN_DELETED,
                                                BB_END (copy_bb));
                emit_barrier_after (BB_END (copy_bb));

                tbb = bb;
                while (1)
                  {
                    dup_block_and_redirect (tbb, copy_bb, insert_point,
                                            &bb_flags);
                    tbb = single_succ (tbb);
                    if (tbb == EXIT_BLOCK_PTR)
                      break;
                    e = split_block (copy_bb, PREV_INSN (insert_point));
                    copy_bb = e->dest;
                  }

                /* Quiet verify_flow_info by (ab)using EDGE_FAKE.
                   We have yet to add a simple_return to the tails,
                   as we'd like to first convert_jumps_to_returns in
                   case the block is no longer used after that.  */
                eflags = EDGE_FAKE;
                if (CALL_P (PREV_INSN (insert_point))
                    && SIBLING_CALL_P (PREV_INSN (insert_point)))
                  eflags = EDGE_SIBCALL | EDGE_ABNORMAL;
                make_single_succ_edge (copy_bb, EXIT_BLOCK_PTR, eflags);

                /* verify_flow_info doesn't like a note after a
                   sibling call.  */
                delete_insn (insert_point);
                if (bitmap_empty_p (&bb_tail))
                  break;
              }
        }

    fail_shrinkwrap:
      bitmap_clear (&bb_tail);
      bitmap_clear (&bb_antic_flags);
      bitmap_clear (&bb_on_list);
      vec.release ();
    }
#endif

  if (split_prologue_seq != NULL_RTX)
    {
      insert_insn_on_edge (split_prologue_seq, orig_entry_edge);
      inserted = true;
    }
  if (prologue_seq != NULL_RTX)
    {
      insert_insn_on_edge (prologue_seq, entry_edge);
      inserted = true;
    }

  /* If the exit block has no non-fake predecessors, we don't need
     an epilogue.  */
  FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
    if ((e->flags & EDGE_FAKE) == 0)
      break;
  if (e == NULL)
    goto epilogue_done;

  rtl_profile_for_bb (EXIT_BLOCK_PTR);

  exit_fallthru_edge = find_fallthru_edge (EXIT_BLOCK_PTR->preds);

  /* If we're allowed to generate a simple return instruction, then by
     definition we don't need a full epilogue.  If the last basic
     block before the exit block does not contain active instructions,
     examine its predecessors and try to emit (conditional) return
     instructions.  */
#ifdef HAVE_simple_return
  if (entry_edge != orig_entry_edge)
    {
      if (optimize)
        {
          unsigned i, last;

          /* convert_jumps_to_returns may add to EXIT_BLOCK_PTR->preds
             (but won't remove).  Stop at end of current preds.  */
          last = EDGE_COUNT (EXIT_BLOCK_PTR->preds);
          for (i = 0; i < last; i++)
            {
              e = EDGE_I (EXIT_BLOCK_PTR->preds, i);
              if (LABEL_P (BB_HEAD (e->src))
                  && !bitmap_bit_p (&bb_flags, e->src->index)
                  && !active_insn_between (BB_HEAD (e->src), BB_END (e->src)))
                unconverted_simple_returns
                  = convert_jumps_to_returns (e->src, true,
                                              unconverted_simple_returns);
            }
        }

      if (exit_fallthru_edge != NULL
          && EDGE_COUNT (exit_fallthru_edge->src->preds) != 0
          && !bitmap_bit_p (&bb_flags, exit_fallthru_edge->src->index))
        {
          basic_block last_bb;

          last_bb = emit_return_for_exit (exit_fallthru_edge, true);
          returnjump = BB_END (last_bb);
          exit_fallthru_edge = NULL;
        }
    }
#endif
#ifdef HAVE_return
  if (HAVE_return)
    {
      if (exit_fallthru_edge == NULL)
        goto epilogue_done;

      if (optimize)
        {
          basic_block last_bb = exit_fallthru_edge->src;

          if (LABEL_P (BB_HEAD (last_bb))
              && !active_insn_between (BB_HEAD (last_bb), BB_END (last_bb)))
            convert_jumps_to_returns (last_bb, false, vNULL);

          if (EDGE_COUNT (last_bb->preds) != 0
              && single_succ_p (last_bb))
            {
              last_bb = emit_return_for_exit (exit_fallthru_edge, false);
              epilogue_end = returnjump = BB_END (last_bb);
#ifdef HAVE_simple_return
              /* Emitting the return may add a basic block.
                 Fix bb_flags for the added block.  */
              if (last_bb != exit_fallthru_edge->src)
                bitmap_set_bit (&bb_flags, last_bb->index);
#endif
              goto epilogue_done;
            }
        }
    }
#endif

  /* A small fib -- epilogue is not yet completed, but we wish to re-use
     this marker for the splits of EH_RETURN patterns, and nothing else
     uses the flag in the meantime.  */
  epilogue_completed = 1;

#ifdef HAVE_eh_return
  /* Find non-fallthru edges that end with EH_RETURN instructions.  On
     some targets, these get split to a special version of the epilogue
     code.  In order to be able to properly annotate these with unwind
     info, try to split them now.  If we get a valid split, drop an
     EPILOGUE_BEG note and mark the insns as epilogue insns.  */
  FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
    {
      rtx prev, last, trial;

      if (e->flags & EDGE_FALLTHRU)
        continue;
      last = BB_END (e->src);
      if (!eh_returnjump_p (last))
        continue;

      prev = PREV_INSN (last);
      trial = try_split (PATTERN (last), last, 1);
      if (trial == last)
        continue;

      record_insns (NEXT_INSN (prev), NEXT_INSN (trial), &epilogue_insn_hash);
      emit_note_after (NOTE_INSN_EPILOGUE_BEG, prev);
    }
#endif

  /* If nothing falls through into the exit block, we don't need an
     epilogue.  */

  if (exit_fallthru_edge == NULL)
    goto epilogue_done;

#ifdef HAVE_epilogue
  if (HAVE_epilogue)
    {
      start_sequence ();
      epilogue_end = emit_note (NOTE_INSN_EPILOGUE_BEG);
      seq = gen_epilogue ();
      if (seq)
        emit_jump_insn (seq);

      /* Retain a map of the epilogue insns.  */
      record_insns (seq, NULL, &epilogue_insn_hash);
      set_insn_locations (seq, epilogue_location);

      seq = get_insns ();
      returnjump = get_last_insn ();
      end_sequence ();

      insert_insn_on_edge (seq, exit_fallthru_edge);
      inserted = true;

      if (JUMP_P (returnjump))
        set_return_jump_label (returnjump);
    }
  else
#endif
    {
      basic_block cur_bb;

      if (! next_active_insn (BB_END (exit_fallthru_edge->src)))
        goto epilogue_done;
      /* We have a fall-through edge to the exit block, the source is not
         at the end of the function, and there will be an assembler epilogue
         at the end of the function.
         We can't use force_nonfallthru here, because that would try to
         use return.  Inserting a jump 'by hand' is extremely messy, so
         we take advantage of cfg_layout_finalize using
         fixup_fallthru_exit_predecessor.  */
      cfg_layout_initialize (0);
      FOR_EACH_BB (cur_bb)
        if (cur_bb->index >= NUM_FIXED_BLOCKS
            && cur_bb->next_bb->index >= NUM_FIXED_BLOCKS)
          cur_bb->aux = cur_bb->next_bb;
      cfg_layout_finalize ();
    }

epilogue_done:

  default_rtl_profile ();

  if (inserted)
    {
      sbitmap blocks;

      commit_edge_insertions ();

      /* Look for basic blocks within the prologue insns.  */
      blocks = sbitmap_alloc (last_basic_block);
      bitmap_clear (blocks);
      bitmap_set_bit (blocks, entry_edge->dest->index);
      bitmap_set_bit (blocks, orig_entry_edge->dest->index);
      find_many_sub_basic_blocks (blocks);
      sbitmap_free (blocks);

      /* The epilogue insns we inserted may cause the exit edge to no longer
         be fallthru.  */
      FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
        {
          if (((e->flags & EDGE_FALLTHRU) != 0)
              && returnjump_p (BB_END (e->src)))
            e->flags &= ~EDGE_FALLTHRU;
        }
    }

#ifdef HAVE_simple_return
  /* If there were branches to an empty LAST_BB which we tried to
     convert to conditional simple_returns, but couldn't for some
     reason, create a block to hold a simple_return insn and redirect
     those remaining edges.  */
  if (!unconverted_simple_returns.is_empty ())
    {
      basic_block simple_return_block_hot = NULL;
      basic_block simple_return_block_cold = NULL;
      edge pending_edge_hot = NULL;
      edge pending_edge_cold = NULL;
      basic_block exit_pred;
      int i;

      gcc_assert (entry_edge != orig_entry_edge);

      /* See if we can reuse the last insn that was emitted for the
         epilogue.  */
      if (returnjump != NULL_RTX
          && JUMP_LABEL (returnjump) == simple_return_rtx)
        {
          e = split_block (BLOCK_FOR_INSN (returnjump), PREV_INSN (returnjump));
          if (BB_PARTITION (e->src) == BB_HOT_PARTITION)
            simple_return_block_hot = e->dest;
          else
            simple_return_block_cold = e->dest;
        }

      /* Also check returns we might need to add to tail blocks.  */
      FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
        if (EDGE_COUNT (e->src->preds) != 0
            && (e->flags & EDGE_FAKE) != 0
            && !bitmap_bit_p (&bb_flags, e->src->index))
          {
            if (BB_PARTITION (e->src) == BB_HOT_PARTITION)
              pending_edge_hot = e;
            else
              pending_edge_cold = e;
          }
      
      /* Save a pointer to the exit's predecessor BB for use in
         inserting new BBs at the end of the function. Do this
         after the call to split_block above which may split
         the original exit pred.  */
      exit_pred = EXIT_BLOCK_PTR->prev_bb;

      FOR_EACH_VEC_ELT (unconverted_simple_returns, i, e)
        {
          basic_block *pdest_bb;
          edge pending;

          if (BB_PARTITION (e->src) == BB_HOT_PARTITION)
            {
              pdest_bb = &simple_return_block_hot;
              pending = pending_edge_hot;
            }
          else
            {
              pdest_bb = &simple_return_block_cold;
              pending = pending_edge_cold;
            }

          if (*pdest_bb == NULL && pending != NULL)
            {
              emit_return_into_block (true, pending->src);
              pending->flags &= ~(EDGE_FALLTHRU | EDGE_FAKE);
              *pdest_bb = pending->src;
            }
          else if (*pdest_bb == NULL)
            {
              basic_block bb;
              rtx start;

              bb = create_basic_block (NULL, NULL, exit_pred);
              BB_COPY_PARTITION (bb, e->src);
              start = emit_jump_insn_after (gen_simple_return (),
                                            BB_END (bb));
              JUMP_LABEL (start) = simple_return_rtx;
              emit_barrier_after (start);

              *pdest_bb = bb;
              make_edge (bb, EXIT_BLOCK_PTR, 0);
            }
          redirect_edge_and_branch_force (e, *pdest_bb);
        }
      unconverted_simple_returns.release ();
    }

  if (entry_edge != orig_entry_edge)
    {
      FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
        if (EDGE_COUNT (e->src->preds) != 0
            && (e->flags & EDGE_FAKE) != 0
            && !bitmap_bit_p (&bb_flags, e->src->index))
          {
            emit_return_into_block (true, e->src);
            e->flags &= ~(EDGE_FALLTHRU | EDGE_FAKE);
          }
    }
#endif

#ifdef HAVE_sibcall_epilogue
  /* Emit sibling epilogues before any sibling call sites.  */
  for (ei = ei_start (EXIT_BLOCK_PTR->preds); (e = ei_safe_edge (ei)); )
    {
      basic_block bb = e->src;
      rtx insn = BB_END (bb);
      rtx ep_seq;

      if (!CALL_P (insn)
          || ! SIBLING_CALL_P (insn)
#ifdef HAVE_simple_return
          || (entry_edge != orig_entry_edge
              && !bitmap_bit_p (&bb_flags, bb->index))
#endif
          )
        {
          ei_next (&ei);
          continue;
        }

      ep_seq = gen_sibcall_epilogue ();
      if (ep_seq)
        {
          start_sequence ();
          emit_note (NOTE_INSN_EPILOGUE_BEG);
          emit_insn (ep_seq);
          seq = get_insns ();
          end_sequence ();

          /* Retain a map of the epilogue insns.  Used in life analysis to
             avoid getting rid of sibcall epilogue insns.  Do this before we
             actually emit the sequence.  */
          record_insns (seq, NULL, &epilogue_insn_hash);
          set_insn_locations (seq, epilogue_location);

          emit_insn_before (seq, insn);
        }
      ei_next (&ei);
    }
#endif

#ifdef HAVE_epilogue
  if (epilogue_end)
    {
      rtx insn, next;

      /* Similarly, move any line notes that appear after the epilogue.
         There is no need, however, to be quite so anal about the existence
         of such a note.  Also possibly move
         NOTE_INSN_FUNCTION_BEG notes, as those can be relevant for debug
         info generation.  */
      for (insn = epilogue_end; insn; insn = next)
        {
          next = NEXT_INSN (insn);
          if (NOTE_P (insn)
              && (NOTE_KIND (insn) == NOTE_INSN_FUNCTION_BEG))
            reorder_insns (insn, insn, PREV_INSN (epilogue_end));
        }
    }
#endif

#ifdef HAVE_simple_return
  bitmap_clear (&bb_flags);
#endif

  /* Threading the prologue and epilogue changes the artificial refs
     in the entry and exit blocks.  */
  epilogue_completed = 1;
  df_update_entry_exit_and_calls ();
}

/* Reposition the prologue-end and epilogue-begin notes after
   instruction scheduling.  */

void
reposition_prologue_and_epilogue_notes (void)
{
#if defined (HAVE_prologue) || defined (HAVE_epilogue) \
    || defined (HAVE_sibcall_epilogue)
  /* Since the hash table is created on demand, the fact that it is
     non-null is a signal that it is non-empty.  */
  if (prologue_insn_hash != NULL)
    {
      size_t len = htab_elements (prologue_insn_hash);
      rtx insn, last = NULL, note = NULL;

      /* Scan from the beginning until we reach the last prologue insn.  */
      /* ??? While we do have the CFG intact, there are two problems:
         (1) The prologue can contain loops (typically probing the stack),
             which means that the end of the prologue isn't in the first bb.
         (2) Sometimes the PROLOGUE_END note gets pushed into the next bb.  */
      for (insn = get_insns (); insn; insn = NEXT_INSN (insn))
        {
          if (NOTE_P (insn))
            {
              if (NOTE_KIND (insn) == NOTE_INSN_PROLOGUE_END)
                note = insn;
            }
          else if (contains (insn, prologue_insn_hash))
            {
              last = insn;
              if (--len == 0)
                break;
            }
        }

      if (last)
        {
          if (note == NULL)
            {
              /* Scan forward looking for the PROLOGUE_END note.  It should
                 be right at the beginning of the block, possibly with other
                 insn notes that got moved there.  */
              for (note = NEXT_INSN (last); ; note = NEXT_INSN (note))
                {
                  if (NOTE_P (note)
                      && NOTE_KIND (note) == NOTE_INSN_PROLOGUE_END)
                    break;
                }
            }

          /* Avoid placing note between CODE_LABEL and BASIC_BLOCK note.  */
          if (LABEL_P (last))
            last = NEXT_INSN (last);
          reorder_insns (note, note, last);
        }
    }

  if (epilogue_insn_hash != NULL)
    {
      edge_iterator ei;
      edge e;

      FOR_EACH_EDGE (e, ei, EXIT_BLOCK_PTR->preds)
        {
          rtx insn, first = NULL, note = NULL;
          basic_block bb = e->src;

          /* Scan from the beginning until we reach the first epilogue insn. */
          FOR_BB_INSNS (bb, insn)
            {
              if (NOTE_P (insn))
                {
                  if (NOTE_KIND (insn) == NOTE_INSN_EPILOGUE_BEG)
                    {
                      note = insn;
                      if (first != NULL)
                        break;
                    }
                }
              else if (first == NULL && contains (insn, epilogue_insn_hash))
                {
                  first = insn;
                  if (note != NULL)
                    break;
                }
            }

          if (note)
            {
              /* If the function has a single basic block, and no real
                 epilogue insns (e.g. sibcall with no cleanup), the
                 epilogue note can get scheduled before the prologue
                 note.  If we have frame related prologue insns, having
                 them scanned during the epilogue will result in a crash.
                 In this case re-order the epilogue note to just before
                 the last insn in the block.  */
              if (first == NULL)
                first = BB_END (bb);

              if (PREV_INSN (first) != note)
                reorder_insns (note, note, PREV_INSN (first));
            }
        }
    }
#endif /* HAVE_prologue or HAVE_epilogue */
}

/* Returns the name of function declared by FNDECL.  */
const char *
fndecl_name (tree fndecl)
{
  if (fndecl == NULL)
    return "(nofn)";
  return lang_hooks.decl_printable_name (fndecl, 2);
}

/* Returns the name of function FN.  */
const char *
function_name (struct function *fn)
{
  tree fndecl = (fn == NULL) ? NULL : fn->decl;
  return fndecl_name (fndecl);
}

/* Returns the name of the current function.  */
const char *
current_function_name (void)
{
  return function_name (cfun);
}
\f

static unsigned int
rest_of_handle_check_leaf_regs (void)
{
#ifdef LEAF_REGISTERS
  crtl->uses_only_leaf_regs
    = optimize > 0 && only_leaf_regs_used () && leaf_function_p ();
#endif
  return 0;
}

/* Insert a TYPE into the used types hash table of CFUN.  */

static void
used_types_insert_helper (tree type, struct function *func)
{
  if (type != NULL && func != NULL)
    {
      void **slot;

      if (func->used_types_hash == NULL)
        func->used_types_hash = htab_create_ggc (37, htab_hash_pointer,
                                                 htab_eq_pointer, NULL);
      slot = htab_find_slot (func->used_types_hash, type, INSERT);
      if (*slot == NULL)
        *slot = type;
    }
}

/* Given a type, insert it into the used hash table in cfun.  */
void
used_types_insert (tree t)
{
  while (POINTER_TYPE_P (t) || TREE_CODE (t) == ARRAY_TYPE)
    if (TYPE_NAME (t))
      break;
    else
      t = TREE_TYPE (t);
  if (TREE_CODE (t) == ERROR_MARK)
    return;
  if (TYPE_NAME (t) == NULL_TREE
      || TYPE_NAME (t) == TYPE_NAME (TYPE_MAIN_VARIANT (t)))
    t = TYPE_MAIN_VARIANT (t);
  if (debug_info_level > DINFO_LEVEL_NONE)
    {
      if (cfun)
        used_types_insert_helper (t, cfun);
      else
        {
          /* So this might be a type referenced by a global variable.
             Record that type so that we can later decide to emit its
             debug information.  */
          vec_safe_push (types_used_by_cur_var_decl, t);
        }
    }
}

/* Helper to Hash a struct types_used_by_vars_entry.  */

static hashval_t
hash_types_used_by_vars_entry (const struct types_used_by_vars_entry *entry)
{
  gcc_assert (entry && entry->var_decl && entry->type);

  return iterative_hash_object (entry->type,
                                iterative_hash_object (entry->var_decl, 0));
}

/* Hash function of the types_used_by_vars_entry hash table.  */

hashval_t
types_used_by_vars_do_hash (const void *x)
{
  const struct types_used_by_vars_entry *entry =
    (const struct types_used_by_vars_entry *) x;

  return hash_types_used_by_vars_entry (entry);
}

/*Equality function of the types_used_by_vars_entry hash table.  */

int
types_used_by_vars_eq (const void *x1, const void *x2)
{
  const struct types_used_by_vars_entry *e1 =
    (const struct types_used_by_vars_entry *) x1;
  const struct types_used_by_vars_entry *e2 =
    (const struct types_used_by_vars_entry *)x2;

  return (e1->var_decl == e2->var_decl && e1->type == e2->type);
}

/* Inserts an entry into the types_used_by_vars_hash hash table. */

void
types_used_by_var_decl_insert (tree type, tree var_decl)
{
  if (type != NULL && var_decl != NULL)
    {
      void **slot;
      struct types_used_by_vars_entry e;
      e.var_decl = var_decl;
      e.type = type;
      if (types_used_by_vars_hash == NULL)
        types_used_by_vars_hash =
          htab_create_ggc (37, types_used_by_vars_do_hash,
                           types_used_by_vars_eq, NULL);
      slot = htab_find_slot_with_hash (types_used_by_vars_hash, &e,
                                       hash_types_used_by_vars_entry (&e), INSERT);
      if (*slot == NULL)
        {
          struct types_used_by_vars_entry *entry;
          entry = ggc_alloc_types_used_by_vars_entry ();
          entry->type = type;
          entry->var_decl = var_decl;
          *slot = entry;
        }
    }
}

namespace {

const pass_data pass_data_leaf_regs =
{
  RTL_PASS, /* type */
  "*leaf_regs", /* name */
  OPTGROUP_NONE, /* optinfo_flags */
  false, /* has_gate */
  true, /* has_execute */
  TV_NONE, /* tv_id */
  0, /* properties_required */
  0, /* properties_provided */
  0, /* properties_destroyed */
  0, /* todo_flags_start */
  0, /* todo_flags_finish */
};

class pass_leaf_regs : public rtl_opt_pass
{
public:
  pass_leaf_regs(gcc::context *ctxt)
    : rtl_opt_pass(pass_data_leaf_regs, ctxt)
  {}

  /* opt_pass methods: */
  unsigned int execute () { return rest_of_handle_check_leaf_regs (); }

}; // class pass_leaf_regs

} // anon namespace

rtl_opt_pass *
make_pass_leaf_regs (gcc::context *ctxt)
{
  return new pass_leaf_regs (ctxt);
}

static unsigned int
rest_of_handle_thread_prologue_and_epilogue (void)
{
  if (optimize)
    cleanup_cfg (CLEANUP_EXPENSIVE);

  /* On some machines, the prologue and epilogue code, or parts thereof,
     can be represented as RTL.  Doing so lets us schedule insns between
     it and the rest of the code and also allows delayed branch
     scheduling to operate in the epilogue.  */
  thread_prologue_and_epilogue_insns ();

  /* The stack usage info is finalized during prologue expansion.  */
  if (flag_stack_usage_info)
    output_stack_usage ();

  return 0;
}

namespace {

const pass_data pass_data_thread_prologue_and_epilogue =
{
  RTL_PASS, /* type */
  "pro_and_epilogue", /* name */
  OPTGROUP_NONE, /* optinfo_flags */
  false, /* has_gate */
  true, /* has_execute */
  TV_THREAD_PROLOGUE_AND_EPILOGUE, /* tv_id */
  0, /* properties_required */
  0, /* properties_provided */
  0, /* properties_destroyed */
  TODO_verify_flow, /* todo_flags_start */
  ( TODO_df_verify | TODO_df_finish
    | TODO_verify_rtl_sharing ), /* todo_flags_finish */
};

class pass_thread_prologue_and_epilogue : public rtl_opt_pass
{
public:
  pass_thread_prologue_and_epilogue(gcc::context *ctxt)
    : rtl_opt_pass(pass_data_thread_prologue_and_epilogue, ctxt)
  {}

  /* opt_pass methods: */
  unsigned int execute () {
    return rest_of_handle_thread_prologue_and_epilogue ();
  }

}; // class pass_thread_prologue_and_epilogue

} // anon namespace

rtl_opt_pass *
make_pass_thread_prologue_and_epilogue (gcc::context *ctxt)
{
  return new pass_thread_prologue_and_epilogue (ctxt);
}
\f

/* This mini-pass fixes fall-out from SSA in asm statements that have
   in-out constraints.  Say you start with

     orig = inout;
     asm ("": "+mr" (inout));
     use (orig);

   which is transformed very early to use explicit output and match operands:

     orig = inout;
     asm ("": "=mr" (inout) : "0" (inout));
     use (orig);

   Or, after SSA and copyprop,

     asm ("": "=mr" (inout_2) : "0" (inout_1));
     use (inout_1);

   Clearly inout_2 and inout_1 can't be coalesced easily anymore, as
   they represent two separate values, so they will get different pseudo
   registers during expansion.  Then, since the two operands need to match
   per the constraints, but use different pseudo registers, reload can
   only register a reload for these operands.  But reloads can only be
   satisfied by hardregs, not by memory, so we need a register for this
   reload, just because we are presented with non-matching operands.
   So, even though we allow memory for this operand, no memory can be
   used for it, just because the two operands don't match.  This can
   cause reload failures on register-starved targets.

   So it's a symptom of reload not being able to use memory for reloads
   or, alternatively it's also a symptom of both operands not coming into
   reload as matching (in which case the pseudo could go to memory just
   fine, as the alternative allows it, and no reload would be necessary).
   We fix the latter problem here, by transforming

     asm ("": "=mr" (inout_2) : "0" (inout_1));

   back to

     inout_2 = inout_1;
     asm ("": "=mr" (inout_2) : "0" (inout_2));  */

static void
match_asm_constraints_1 (rtx insn, rtx *p_sets, int noutputs)
{
  int i;
  bool changed = false;
  rtx op = SET_SRC (p_sets[0]);
  int ninputs = ASM_OPERANDS_INPUT_LENGTH (op);
  rtvec inputs = ASM_OPERANDS_INPUT_VEC (op);
  bool *output_matched = XALLOCAVEC (bool, noutputs);

  memset (output_matched, 0, noutputs * sizeof (bool));
  for (i = 0; i < ninputs; i++)
    {
      rtx input, output, insns;
      const char *constraint = ASM_OPERANDS_INPUT_CONSTRAINT (op, i);
      char *end;
      int match, j;

      if (*constraint == '%')
        constraint++;

      match = strtoul (constraint, &end, 10);
      if (end == constraint)
        continue;

      gcc_assert (match < noutputs);
      output = SET_DEST (p_sets[match]);
      input = RTVEC_ELT (inputs, i);
      /* Only do the transformation for pseudos.  */
      if (! REG_P (output)
          || rtx_equal_p (output, input)
          || (GET_MODE (input) != VOIDmode
              && GET_MODE (input) != GET_MODE (output)))
        continue;

      /* We can't do anything if the output is also used as input,
         as we're going to overwrite it.  */
      for (j = 0; j < ninputs; j++)
        if (reg_overlap_mentioned_p (output, RTVEC_ELT (inputs, j)))
          break;
      if (j != ninputs)
        continue;

      /* Avoid changing the same input several times.  For
         asm ("" : "=mr" (out1), "=mr" (out2) : "0" (in), "1" (in));
         only change in once (to out1), rather than changing it
         first to out1 and afterwards to out2.  */
      if (i > 0)
        {
          for (j = 0; j < noutputs; j++)
            if (output_matched[j] && input == SET_DEST (p_sets[j]))
              break;
          if (j != noutputs)
            continue;
        }
      output_matched[match] = true;

      start_sequence ();
      emit_move_insn (output, input);
      insns = get_insns ();
      end_sequence ();
      emit_insn_before (insns, insn);

      /* Now replace all mentions of the input with output.  We can't
         just replace the occurrence in inputs[i], as the register might
         also be used in some other input (or even in an address of an
         output), which would mean possibly increasing the number of
         inputs by one (namely 'output' in addition), which might pose
         a too complicated problem for reload to solve.  E.g. this situation:

           asm ("" : "=r" (output), "=m" (input) : "0" (input))

         Here 'input' is used in two occurrences as input (once for the
         input operand, once for the address in the second output operand).
         If we would replace only the occurrence of the input operand (to
         make the matching) we would be left with this:

           output = input
           asm ("" : "=r" (output), "=m" (input) : "0" (output))

         Now we suddenly have two different input values (containing the same
         value, but different pseudos) where we formerly had only one.
         With more complicated asms this might lead to reload failures
         which wouldn't have happen without this pass.  So, iterate over
         all operands and replace all occurrences of the register used.  */
      for (j = 0; j < noutputs; j++)
        if (!rtx_equal_p (SET_DEST (p_sets[j]), input)
            && reg_overlap_mentioned_p (input, SET_DEST (p_sets[j])))
          SET_DEST (p_sets[j]) = replace_rtx (SET_DEST (p_sets[j]),
                                              input, output);
      for (j = 0; j < ninputs; j++)
        if (reg_overlap_mentioned_p (input, RTVEC_ELT (inputs, j)))
          RTVEC_ELT (inputs, j) = replace_rtx (RTVEC_ELT (inputs, j),
                                               input, output);

      changed = true;
    }

  if (changed)
    df_insn_rescan (insn);
}

static unsigned
rest_of_match_asm_constraints (void)
{
  basic_block bb;
  rtx insn, pat, *p_sets;
  int noutputs;

  if (!crtl->has_asm_statement)
    return 0;

  df_set_flags (DF_DEFER_INSN_RESCAN);
  FOR_EACH_BB (bb)
    {
      FOR_BB_INSNS (bb, insn)
        {
          if (!INSN_P (insn))
            continue;

          pat = PATTERN (insn);
          if (GET_CODE (pat) == PARALLEL)
            p_sets = &XVECEXP (pat, 0, 0), noutputs = XVECLEN (pat, 0);
          else if (GET_CODE (pat) == SET)
            p_sets = &PATTERN (insn), noutputs = 1;
          else
            continue;

          if (GET_CODE (*p_sets) == SET
              && GET_CODE (SET_SRC (*p_sets)) == ASM_OPERANDS)
            match_asm_constraints_1 (insn, p_sets, noutputs);
         }
    }

  return TODO_df_finish;
}

namespace {

const pass_data pass_data_match_asm_constraints =
{
  RTL_PASS, /* type */
  "asmcons", /* name */
  OPTGROUP_NONE, /* optinfo_flags */
  false, /* has_gate */
  true, /* has_execute */
  TV_NONE, /* tv_id */
  0, /* properties_required */
  0, /* properties_provided */
  0, /* properties_destroyed */
  0, /* todo_flags_start */
  0, /* todo_flags_finish */
};

class pass_match_asm_constraints : public rtl_opt_pass
{
public:
  pass_match_asm_constraints(gcc::context *ctxt)
    : rtl_opt_pass(pass_data_match_asm_constraints, ctxt)
  {}

  /* opt_pass methods: */
  unsigned int execute () { return rest_of_match_asm_constraints (); }

}; // class pass_match_asm_constraints

} // anon namespace

rtl_opt_pass *
make_pass_match_asm_constraints (gcc::context *ctxt)
{
  return new pass_match_asm_constraints (ctxt);
}


#include "gt-function.h"