mode-switching.c   [plain text]


/* CPU mode switching
   Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005
   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 2, 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 COPYING.  If not, write to the Free
Software Foundation, 51 Franklin Street, Fifth Floor, Boston, MA
02110-1301, USA.  */

#include "config.h"
#include "system.h"
#include "coretypes.h"
#include "tm.h"
#include "rtl.h"
#include "regs.h"
#include "hard-reg-set.h"
#include "flags.h"
#include "real.h"
#include "insn-config.h"
#include "recog.h"
#include "basic-block.h"
#include "output.h"
#include "tm_p.h"
#include "function.h"
#include "tree-pass.h"
#include "timevar.h"

/* We want target macros for the mode switching code to be able to refer
   to instruction attribute values.  */
#include "insn-attr.h"

#ifdef OPTIMIZE_MODE_SWITCHING

/* The algorithm for setting the modes consists of scanning the insn list
   and finding all the insns which require a specific mode.  Each insn gets
   a unique struct seginfo element.  These structures are inserted into a list
   for each basic block.  For each entity, there is an array of bb_info over
   the flow graph basic blocks (local var 'bb_info'), and contains a list
   of all insns within that basic block, in the order they are encountered.

   For each entity, any basic block WITHOUT any insns requiring a specific
   mode are given a single entry, without a mode.  (Each basic block
   in the flow graph must have at least one entry in the segment table.)

   The LCM algorithm is then run over the flow graph to determine where to
   place the sets to the highest-priority value in respect of first the first
   insn in any one block.  Any adjustments required to the transparency
   vectors are made, then the next iteration starts for the next-lower
   priority mode, till for each entity all modes are exhausted.

   More details are located in the code for optimize_mode_switching().  */

/* This structure contains the information for each insn which requires
   either single or double mode to be set.
   MODE is the mode this insn must be executed in.
   INSN_PTR is the insn to be executed (may be the note that marks the
   beginning of a basic block).
   BBNUM is the flow graph basic block this insn occurs in.
   NEXT is the next insn in the same basic block.  */
struct seginfo
{
  int mode;
  rtx insn_ptr;
  int bbnum;
  struct seginfo *next;
  HARD_REG_SET regs_live;
};

struct bb_info
{
  struct seginfo *seginfo;
  int computing;
};

/* These bitmaps are used for the LCM algorithm.  */

static sbitmap *antic;
static sbitmap *transp;
static sbitmap *comp;

static struct seginfo * new_seginfo (int, rtx, int, HARD_REG_SET);
static void add_seginfo (struct bb_info *, struct seginfo *);
static void reg_dies (rtx, HARD_REG_SET);
static void reg_becomes_live (rtx, rtx, void *);
static void make_preds_opaque (basic_block, int);


/* This function will allocate a new BBINFO structure, initialized
   with the MODE, INSN, and basic block BB parameters.  */

static struct seginfo *
new_seginfo (int mode, rtx insn, int bb, HARD_REG_SET regs_live)
{
  struct seginfo *ptr;
  ptr = XNEW (struct seginfo);
  ptr->mode = mode;
  ptr->insn_ptr = insn;
  ptr->bbnum = bb;
  ptr->next = NULL;
  COPY_HARD_REG_SET (ptr->regs_live, regs_live);
  return ptr;
}

/* Add a seginfo element to the end of a list.
   HEAD is a pointer to the list beginning.
   INFO is the structure to be linked in.  */

static void
add_seginfo (struct bb_info *head, struct seginfo *info)
{
  struct seginfo *ptr;

  if (head->seginfo == NULL)
    head->seginfo = info;
  else
    {
      ptr = head->seginfo;
      while (ptr->next != NULL)
	ptr = ptr->next;
      ptr->next = info;
    }
}

/* Make all predecessors of basic block B opaque, recursively, till we hit
   some that are already non-transparent, or an edge where aux is set; that
   denotes that a mode set is to be done on that edge.
   J is the bit number in the bitmaps that corresponds to the entity that
   we are currently handling mode-switching for.  */

static void
make_preds_opaque (basic_block b, int j)
{
  edge e;
  edge_iterator ei;

  FOR_EACH_EDGE (e, ei, b->preds)
    {
      basic_block pb = e->src;

      if (e->aux || ! TEST_BIT (transp[pb->index], j))
	continue;

      RESET_BIT (transp[pb->index], j);
      make_preds_opaque (pb, j);
    }
}

/* Record in LIVE that register REG died.  */

static void
reg_dies (rtx reg, HARD_REG_SET live)
{
  int regno, nregs;

  if (!REG_P (reg))
    return;

  regno = REGNO (reg);
  if (regno < FIRST_PSEUDO_REGISTER)
    for (nregs = hard_regno_nregs[regno][GET_MODE (reg)] - 1; nregs >= 0;
	 nregs--)
      CLEAR_HARD_REG_BIT (live, regno + nregs);
}

/* Record in LIVE that register REG became live.
   This is called via note_stores.  */

static void
reg_becomes_live (rtx reg, rtx setter ATTRIBUTE_UNUSED, void *live)
{
  int regno, nregs;

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

  if (!REG_P (reg))
    return;

  regno = REGNO (reg);
  if (regno < FIRST_PSEUDO_REGISTER)
    for (nregs = hard_regno_nregs[regno][GET_MODE (reg)] - 1; nregs >= 0;
	 nregs--)
      SET_HARD_REG_BIT (* (HARD_REG_SET *) live, regno + nregs);
}

/* Make sure if MODE_ENTRY is defined the MODE_EXIT is defined
   and vice versa.  */
#if defined (MODE_ENTRY) != defined (MODE_EXIT)
 #error "Both MODE_ENTRY and MODE_EXIT must be defined"
#endif

#if defined (MODE_ENTRY) && defined (MODE_EXIT)
/* Split the fallthrough edge to the exit block, so that we can note
   that there NORMAL_MODE is required.  Return the new block if it's
   inserted before the exit block.  Otherwise return null.  */

static basic_block
create_pre_exit (int n_entities, int *entity_map, const int *num_modes)
{
  edge eg;
  edge_iterator ei;
  basic_block pre_exit;

  /* The only non-call predecessor at this stage is a block with a
     fallthrough edge; there can be at most one, but there could be
     none at all, e.g. when exit is called.  */
  pre_exit = 0;
  FOR_EACH_EDGE (eg, ei, EXIT_BLOCK_PTR->preds)
    if (eg->flags & EDGE_FALLTHRU)
      {
	basic_block src_bb = eg->src;
	regset live_at_end = src_bb->il.rtl->global_live_at_end;
	rtx last_insn, ret_reg;

	gcc_assert (!pre_exit);
	/* If this function returns a value at the end, we have to
	   insert the final mode switch before the return value copy
	   to its hard register.  */
	if (EDGE_COUNT (EXIT_BLOCK_PTR->preds) == 1
	    && NONJUMP_INSN_P ((last_insn = BB_END (src_bb)))
	    && GET_CODE (PATTERN (last_insn)) == USE
	    && GET_CODE ((ret_reg = XEXP (PATTERN (last_insn), 0))) == REG)
	  {
	    int ret_start = REGNO (ret_reg);
	    int nregs = hard_regno_nregs[ret_start][GET_MODE (ret_reg)];
	    int ret_end = ret_start + nregs;
	    int short_block = 0;
	    int maybe_builtin_apply = 0;
	    int forced_late_switch = 0;
	    rtx before_return_copy;

	    do
	      {
		rtx return_copy = PREV_INSN (last_insn);
		rtx return_copy_pat, copy_reg;
		int copy_start, copy_num;
		int j;

		if (INSN_P (return_copy))
		  {
		    if (GET_CODE (PATTERN (return_copy)) == USE
			&& GET_CODE (XEXP (PATTERN (return_copy), 0)) == REG
			&& (FUNCTION_VALUE_REGNO_P
			    (REGNO (XEXP (PATTERN (return_copy), 0)))))
		      {
			maybe_builtin_apply = 1;
			last_insn = return_copy;
			continue;
		      }
		    /* If the return register is not (in its entirety)
		       likely spilled, the return copy might be
		       partially or completely optimized away.  */
		    return_copy_pat = single_set (return_copy);
		    if (!return_copy_pat)
		      {
			return_copy_pat = PATTERN (return_copy);
			if (GET_CODE (return_copy_pat) != CLOBBER)
			  break;
		      }
		    copy_reg = SET_DEST (return_copy_pat);
		    if (GET_CODE (copy_reg) == REG)
		      copy_start = REGNO (copy_reg);
		    else if (GET_CODE (copy_reg) == SUBREG
			     && GET_CODE (SUBREG_REG (copy_reg)) == REG)
		      copy_start = REGNO (SUBREG_REG (copy_reg));
		    else
		      break;
		    if (copy_start >= FIRST_PSEUDO_REGISTER)
		      break;
		    copy_num
		      = hard_regno_nregs[copy_start][GET_MODE (copy_reg)];

		    /* If the return register is not likely spilled, - as is
		       the case for floating point on SH4 - then it might
		       be set by an arithmetic operation that needs a
		       different mode than the exit block.  */
		    for (j = n_entities - 1; j >= 0; j--)
		      {
			int e = entity_map[j];
			int mode = MODE_NEEDED (e, return_copy);

			if (mode != num_modes[e] && mode != MODE_EXIT (e))
			  break;
		      }
		    if (j >= 0)
		      {
			/* For the SH4, floating point loads depend on fpscr,
			   thus we might need to put the final mode switch
			   after the return value copy.  That is still OK,
			   because a floating point return value does not
			   conflict with address reloads.  */
			if (copy_start >= ret_start
			    && copy_start + copy_num <= ret_end
			    && OBJECT_P (SET_SRC (return_copy_pat)))
			  forced_late_switch = 1;
			break;
		      }

		    if (copy_start >= ret_start
			&& copy_start + copy_num <= ret_end)
		      nregs -= copy_num;
		    else if (!maybe_builtin_apply
			     || !FUNCTION_VALUE_REGNO_P (copy_start))
		      break;
		    last_insn = return_copy;
		  }
		/* ??? Exception handling can lead to the return value
		   copy being already separated from the return value use,
		   as in  unwind-dw2.c .
		   Similarly, conditionally returning without a value,
		   and conditionally using builtin_return can lead to an
		   isolated use.  */
		if (return_copy == BB_HEAD (src_bb))
		  {
		    short_block = 1;
		    break;
		  }
		last_insn = return_copy;
	      }
	    while (nregs);
	    
	    /* If we didn't see a full return value copy, verify that there
	       is a plausible reason for this.  If some, but not all of the
	       return register is likely spilled, we can expect that there
	       is a copy for the likely spilled part.  */
	    gcc_assert (!nregs
			|| forced_late_switch
			|| short_block
			|| !(CLASS_LIKELY_SPILLED_P
			     (REGNO_REG_CLASS (ret_start)))
			|| (nregs
			    != hard_regno_nregs[ret_start][GET_MODE (ret_reg)])
			/* For multi-hard-register floating point
		   	   values, sometimes the likely-spilled part
		   	   is ordinarily copied first, then the other
		   	   part is set with an arithmetic operation.
		   	   This doesn't actually cause reload
		   	   failures, so let it pass.  */
			|| (GET_MODE_CLASS (GET_MODE (ret_reg)) != MODE_INT
			    && nregs != 1));
	    
	    if (INSN_P (last_insn))
	      {
		before_return_copy
		  = emit_note_before (NOTE_INSN_DELETED, last_insn);
		/* Instructions preceding LAST_INSN in the same block might
		   require a different mode than MODE_EXIT, so if we might
		   have such instructions, keep them in a separate block
		   from pre_exit.  */
		if (last_insn != BB_HEAD (src_bb))
		  src_bb = split_block (src_bb,
					PREV_INSN (before_return_copy))->dest;
	      }
	    else
	      before_return_copy = last_insn;
	    pre_exit = split_block (src_bb, before_return_copy)->src;
	  }
	else
	  {
	    pre_exit = split_edge (eg);
	    COPY_REG_SET (pre_exit->il.rtl->global_live_at_start, live_at_end);
	    COPY_REG_SET (pre_exit->il.rtl->global_live_at_end, live_at_end);
	  }
      }

  return pre_exit;
}
#endif

/* Find all insns that need a particular mode setting, and insert the
   necessary mode switches.  Return true if we did work.  */

static int
optimize_mode_switching (void)
{
  rtx insn;
  int e;
  basic_block bb;
  int need_commit = 0;
  sbitmap *kill;
  struct edge_list *edge_list;
  static const int num_modes[] = NUM_MODES_FOR_MODE_SWITCHING;
#define N_ENTITIES ARRAY_SIZE (num_modes)
  int entity_map[N_ENTITIES];
  struct bb_info *bb_info[N_ENTITIES];
  int i, j;
  int n_entities;
  int max_num_modes = 0;
  bool emited = false;
  basic_block post_entry ATTRIBUTE_UNUSED, pre_exit ATTRIBUTE_UNUSED;

  clear_bb_flags ();

  for (e = N_ENTITIES - 1, n_entities = 0; e >= 0; e--)
    if (OPTIMIZE_MODE_SWITCHING (e))
      {
	int entry_exit_extra = 0;

	/* Create the list of segments within each basic block.
	   If NORMAL_MODE is defined, allow for two extra
	   blocks split from the entry and exit block.  */
#if defined (MODE_ENTRY) && defined (MODE_EXIT)
	entry_exit_extra = 3;
#endif
	bb_info[n_entities]
	  = XCNEWVEC (struct bb_info, last_basic_block + entry_exit_extra);
	entity_map[n_entities++] = e;
	if (num_modes[e] > max_num_modes)
	  max_num_modes = num_modes[e];
      }

  if (! n_entities)
    return 0;

#if defined (MODE_ENTRY) && defined (MODE_EXIT)
  /* Split the edge from the entry block, so that we can note that
     there NORMAL_MODE is supplied.  */
  post_entry = split_edge (single_succ_edge (ENTRY_BLOCK_PTR));
  pre_exit = create_pre_exit (n_entities, entity_map, num_modes);
#endif

  /* Create the bitmap vectors.  */

  antic = sbitmap_vector_alloc (last_basic_block, n_entities);
  transp = sbitmap_vector_alloc (last_basic_block, n_entities);
  comp = sbitmap_vector_alloc (last_basic_block, n_entities);

  sbitmap_vector_ones (transp, last_basic_block);

  for (j = n_entities - 1; j >= 0; j--)
    {
      int e = entity_map[j];
      int no_mode = num_modes[e];
      struct bb_info *info = bb_info[j];

      /* Determine what the first use (if any) need for a mode of entity E is.
	 This will be the mode that is anticipatable for this block.
	 Also compute the initial transparency settings.  */
      FOR_EACH_BB (bb)
	{
	  struct seginfo *ptr;
	  int last_mode = no_mode;
	  HARD_REG_SET live_now;

	  REG_SET_TO_HARD_REG_SET (live_now,
				   bb->il.rtl->global_live_at_start);

	  /* Pretend the mode is clobbered across abnormal edges.  */
	  {
	    edge_iterator ei;
	    edge e;
	    FOR_EACH_EDGE (e, ei, bb->preds)
	      if (e->flags & EDGE_COMPLEX)
		break;
	    if (e)
	      {
		ptr = new_seginfo (no_mode, BB_HEAD (bb), bb->index, live_now);
		add_seginfo (info + bb->index, ptr);
		RESET_BIT (transp[bb->index], j);
	      }
	  }

	  for (insn = BB_HEAD (bb);
	       insn != NULL && insn != NEXT_INSN (BB_END (bb));
	       insn = NEXT_INSN (insn))
	    {
	      if (INSN_P (insn))
		{
		  int mode = MODE_NEEDED (e, insn);
		  rtx link;

		  if (mode != no_mode && mode != last_mode)
		    {
		      last_mode = mode;
		      ptr = new_seginfo (mode, insn, bb->index, live_now);
		      add_seginfo (info + bb->index, ptr);
		      RESET_BIT (transp[bb->index], j);
		    }
#ifdef MODE_AFTER
		  last_mode = MODE_AFTER (last_mode, insn);
#endif
		  /* Update LIVE_NOW.  */
		  for (link = REG_NOTES (insn); link; link = XEXP (link, 1))
		    if (REG_NOTE_KIND (link) == REG_DEAD)
		      reg_dies (XEXP (link, 0), live_now);

		  note_stores (PATTERN (insn), reg_becomes_live, &live_now);
		  for (link = REG_NOTES (insn); link; link = XEXP (link, 1))
		    if (REG_NOTE_KIND (link) == REG_UNUSED)
		      reg_dies (XEXP (link, 0), live_now);
		}
	    }

	  info[bb->index].computing = last_mode;
	  /* Check for blocks without ANY mode requirements.  */
	  if (last_mode == no_mode)
	    {
	      ptr = new_seginfo (no_mode, BB_END (bb), bb->index, live_now);
	      add_seginfo (info + bb->index, ptr);
	    }
	}
#if defined (MODE_ENTRY) && defined (MODE_EXIT)
      {
	int mode = MODE_ENTRY (e);

	if (mode != no_mode)
	  {
	    bb = post_entry;

	    /* By always making this nontransparent, we save
	       an extra check in make_preds_opaque.  We also
	       need this to avoid confusing pre_edge_lcm when
	       antic is cleared but transp and comp are set.  */
	    RESET_BIT (transp[bb->index], j);

	    /* Insert a fake computing definition of MODE into entry
	       blocks which compute no mode. This represents the mode on
	       entry.  */
	    info[bb->index].computing = mode;

	    if (pre_exit)
	      info[pre_exit->index].seginfo->mode = MODE_EXIT (e);
	  }
      }
#endif /* NORMAL_MODE */
    }

  kill = sbitmap_vector_alloc (last_basic_block, n_entities);
  for (i = 0; i < max_num_modes; i++)
    {
      int current_mode[N_ENTITIES];
      sbitmap *delete;
      sbitmap *insert;

      /* Set the anticipatable and computing arrays.  */
      sbitmap_vector_zero (antic, last_basic_block);
      sbitmap_vector_zero (comp, last_basic_block);
      for (j = n_entities - 1; j >= 0; j--)
	{
	  int m = current_mode[j] = MODE_PRIORITY_TO_MODE (entity_map[j], i);
	  struct bb_info *info = bb_info[j];

	  FOR_EACH_BB (bb)
	    {
	      if (info[bb->index].seginfo->mode == m)
		SET_BIT (antic[bb->index], j);

	      if (info[bb->index].computing == m)
		SET_BIT (comp[bb->index], j);
	    }
	}

      /* Calculate the optimal locations for the
	 placement mode switches to modes with priority I.  */

      FOR_EACH_BB (bb)
	sbitmap_not (kill[bb->index], transp[bb->index]);
      edge_list = pre_edge_lcm (n_entities, transp, comp, antic,
				kill, &insert, &delete);

      for (j = n_entities - 1; j >= 0; j--)
	{
	  /* Insert all mode sets that have been inserted by lcm.  */
	  int no_mode = num_modes[entity_map[j]];

	  /* Wherever we have moved a mode setting upwards in the flow graph,
	     the blocks between the new setting site and the now redundant
	     computation ceases to be transparent for any lower-priority
	     mode of the same entity.  First set the aux field of each
	     insertion site edge non-transparent, then propagate the new
	     non-transparency from the redundant computation upwards till
	     we hit an insertion site or an already non-transparent block.  */
	  for (e = NUM_EDGES (edge_list) - 1; e >= 0; e--)
	    {
	      edge eg = INDEX_EDGE (edge_list, e);
	      int mode;
	      basic_block src_bb;
	      HARD_REG_SET live_at_edge;
	      rtx mode_set;

	      eg->aux = 0;

	      if (! TEST_BIT (insert[e], j))
		continue;

	      eg->aux = (void *)1;

	      mode = current_mode[j];
	      src_bb = eg->src;

	      REG_SET_TO_HARD_REG_SET (live_at_edge,
				       src_bb->il.rtl->global_live_at_end);

	      start_sequence ();
	      EMIT_MODE_SET (entity_map[j], mode, live_at_edge);
	      mode_set = get_insns ();
	      end_sequence ();

	      /* Do not bother to insert empty sequence.  */
	      if (mode_set == NULL_RTX)
		continue;

	      /* We should not get an abnormal edge here.  */
	      gcc_assert (! (eg->flags & EDGE_ABNORMAL));

	      need_commit = 1;
	      insert_insn_on_edge (mode_set, eg);
	    }

	  FOR_EACH_BB_REVERSE (bb)
	    if (TEST_BIT (delete[bb->index], j))
	      {
		make_preds_opaque (bb, j);
		/* Cancel the 'deleted' mode set.  */
		bb_info[j][bb->index].seginfo->mode = no_mode;
	      }
	}

      sbitmap_vector_free (delete);
      sbitmap_vector_free (insert);
      clear_aux_for_edges ();
      free_edge_list (edge_list);
    }

  /* Now output the remaining mode sets in all the segments.  */
  for (j = n_entities - 1; j >= 0; j--)
    {
      int no_mode = num_modes[entity_map[j]];

      FOR_EACH_BB_REVERSE (bb)
	{
	  struct seginfo *ptr, *next;
	  for (ptr = bb_info[j][bb->index].seginfo; ptr; ptr = next)
	    {
	      next = ptr->next;
	      if (ptr->mode != no_mode)
		{
		  rtx mode_set;

		  start_sequence ();
		  EMIT_MODE_SET (entity_map[j], ptr->mode, ptr->regs_live);
		  mode_set = get_insns ();
		  end_sequence ();

		  /* Insert MODE_SET only if it is nonempty.  */
		  if (mode_set != NULL_RTX)
		    {
		      emited = true;
		      if (NOTE_P (ptr->insn_ptr)
			  && (NOTE_LINE_NUMBER (ptr->insn_ptr)
			      == NOTE_INSN_BASIC_BLOCK))
			emit_insn_after (mode_set, ptr->insn_ptr);
		      else
			emit_insn_before (mode_set, ptr->insn_ptr);
		    }
		}

	      free (ptr);
	    }
	}

      free (bb_info[j]);
    }

  /* Finished. Free up all the things we've allocated.  */

  sbitmap_vector_free (kill);
  sbitmap_vector_free (antic);
  sbitmap_vector_free (transp);
  sbitmap_vector_free (comp);

  if (need_commit)
    commit_edge_insertions ();

#if defined (MODE_ENTRY) && defined (MODE_EXIT)
  cleanup_cfg (CLEANUP_NO_INSN_DEL);
#else
  if (!need_commit && !emited)
    return 0;
#endif

  max_regno = max_reg_num ();
  allocate_reg_info (max_regno, FALSE, FALSE);
  update_life_info_in_dirty_blocks (UPDATE_LIFE_GLOBAL_RM_NOTES,
				    (PROP_DEATH_NOTES | PROP_KILL_DEAD_CODE
				     | PROP_SCAN_DEAD_CODE));

  return 1;
}

#endif /* OPTIMIZE_MODE_SWITCHING */

static bool
gate_mode_switching (void)
{
#ifdef OPTIMIZE_MODE_SWITCHING
  return true;
#else
  return false;
#endif
}

static unsigned int
rest_of_handle_mode_switching (void)
{
#ifdef OPTIMIZE_MODE_SWITCHING
  no_new_pseudos = 0;
  optimize_mode_switching ();
  no_new_pseudos = 1;
#endif /* OPTIMIZE_MODE_SWITCHING */
  return 0;
}


struct tree_opt_pass pass_mode_switching =
{
  "mode-sw",                            /* name */
  gate_mode_switching,                  /* gate */
  rest_of_handle_mode_switching,        /* execute */
  NULL,                                 /* sub */
  NULL,                                 /* next */
  0,                                    /* static_pass_number */
  TV_MODE_SWITCH,                       /* tv_id */
  0,                                    /* properties_required */
  0,                                    /* properties_provided */
  0,                                    /* properties_destroyed */
  0,                                    /* todo_flags_start */
  TODO_dump_func,                       /* todo_flags_finish */
  0                                     /* letter */
};