cfgloop.h (fix_loop_placement, [...]): Declare.

	* cfgloop.h (fix_loop_placement, can_duplicate_loop_p,
	duplicate_loop_to_header_edge, loopify, remove_path, split_loop_bb):
	Declare.
	(DLTHE_FLAG_UPDATE_FREQ): New.
	* cfgloopmanip.c (duplicate_loop, duplicate_subloops, copy_loops_to,
	loop_redirect_edge, loop_delete_branch_edge, copy_bbs, remove_bbs,
	rpe_enum_p, find_branch, alp_enum_p, add_loop, fix_loop_placements,
	fix_bb_placement, fix_bb_placements, place_new_loop,
	scale_loop_frequencies, scale_bbs_frequencies, record_exit_edges):
	New static functions.
	(fix_loop_placement, can_duplicate_loop_p,
	duplicate_loop_to_header_edge, loopify, remove_path, split_loop_bb):
	New functions.

	* cfgloop.h (loop_optimizer_init, loop_optimizer_finalize,
	unswitch_loops): Declare.
	* loop-init.c: New file.
	* loop-unswitch.c: New file.
	* Makefile.in (loop-init.o, loop-unswitch.o): New.
	* params.def (PARAM_MAX_UNSWITCH_INSNS, PARAM_MAX_UNSWITCH_LEVEL): New.
	* toplev.c (DFI_loop2): New dump.
	(flag_unswitch_loops): New.
	(lang_independent_options): Add it.
	(rest_of_compilation): Call new loop optimizer.
	(parse_options_and_default_flags): Turn flag_unswitch_loops on with -O3.

From-SVN: r62578

gcc/ChangeLog

+2003-02-08  Zdenek Dvorak  <rakdver@atrey.karlin.mff.cuni.cz>
+
+	* cfgloop.h (fix_loop_placement, can_duplicate_loop_p,
+	duplicate_loop_to_header_edge, loopify, remove_path, split_loop_bb):
+	Declare.
+	(DLTHE_FLAG_UPDATE_FREQ): New.
+	* cfgloopmanip.c (duplicate_loop, duplicate_subloops, copy_loops_to,
+	loop_redirect_edge, loop_delete_branch_edge, copy_bbs, remove_bbs,
+	rpe_enum_p, find_branch, alp_enum_p, add_loop, fix_loop_placements,
+	fix_bb_placement, fix_bb_placements, place_new_loop,
+	scale_loop_frequencies, scale_bbs_frequencies, record_exit_edges):
+	New static functions.
+	(fix_loop_placement, can_duplicate_loop_p,
+	duplicate_loop_to_header_edge, loopify, remove_path, split_loop_bb):
+	New functions.
+
+	* cfgloop.h (loop_optimizer_init, loop_optimizer_finalize,
+	unswitch_loops): Declare.
+	* loop-init.c: New file.
+	* loop-unswitch.c: New file.
+	* Makefile.in (loop-init.o, loop-unswitch.o): New.
+	* params.def (PARAM_MAX_UNSWITCH_INSNS, PARAM_MAX_UNSWITCH_LEVEL): New.
+	* toplev.c (DFI_loop2): New dump.
+	(flag_unswitch_loops): New.
+	(lang_independent_options): Add it.
+	(rest_of_compilation): Call new loop optimizer.
+	(parse_options_and_default_flags): Turn flag_unswitch_loops on with -O3.
+
 2003-02-08  Kazu Hirata  <kazu@cs.umass.edu>
 
 	* config/h8300/clzsi2.c: New.

gcc/Makefile.in

@@ -768,7 +768,7 @@ C_OBJS = c-parse.o c-lang.o c-pretty-print.o $(C_AND_OBJC_OBJS)
 
 OBJS = alias.o bb-reorder.o bitmap.o builtins.o caller-save.o calls.o	   \
  cfg.o cfganal.o cfgbuild.o cfgcleanup.o cfglayout.o cfgloop.o		   \
- cfgloopanal.o cfgloopmanip.o 						   \
+ cfgloopanal.o cfgloopmanip.o loop-init.o loop-unswitch.o		   \
  cfgrtl.o combine.o conflict.o convert.o cse.o cselib.o dbxout.o	   \
  debug.o df.o diagnostic.o doloop.o dominance.o		                   \
  dwarf2asm.o dwarf2out.o dwarfout.o emit-rtl.o except.o explow.o	   \
@@ -1600,6 +1600,12 @@ cfgloopanal.o : cfgloopanal.c $(CONFIG_H) $(SYSTEM_H) $(RTL_H) \
    $(BASIC_BLOCK_H) hard-reg-set.h cfgloop.h $(EXPR_H) coretypes.h $(TM_H)
 cfgloopmanip.o : cfgloopmanip.c $(CONFIG_H) $(SYSTEM_H) $(RTL_H) \
    $(BASIC_BLOCK_H) hard-reg-set.h cfgloop.h cfglayout.h output.h coretypes.h $(TM_H)
+loop-init.o : loop-init.c $(CONFIG_H) $(SYSTEM_H) $(RTL_H) gcov-io.h \
+   gcov-iov.h $(BASIC_BLOCK_H) hard-reg-set.h cfgloop.h cfglayout.h profile.h \
+   coretypes.h $(TM_H)
+loop-unswitch.o : loop-unswitch.c $(CONFIG_H) $(SYSTEM_H) $(RTL_H) $(TM_H) \
+   $(BASIC_BLOCK_H) hard-reg-set.h cfgloop.h cfglayout.h params.h \
+   output.h $(EXPR_H) coretypes.h $(TM_H)
 dominance.o : dominance.c $(CONFIG_H) $(SYSTEM_H) coretypes.h $(TM_H) $(RTL_H) \
    hard-reg-set.h $(BASIC_BLOCK_H) et-forest.h
 et-forest.o : et-forest.c $(CONFIG_H) $(SYSTEM_H) coretypes.h $(TM_H) et-forest.h alloc-pool.h

gcc/cfgloop.h

@@ -293,6 +293,7 @@ extern void cancel_loop			PARAMS ((struct loops *, struct loop *));
 extern void cancel_loop_tree		PARAMS ((struct loops *, struct loop *));
 
 extern basic_block loop_split_edge_with PARAMS ((edge, rtx, struct loops *));
+extern int fix_loop_placement		PARAMS ((struct loop *));
 
 enum
 {
@@ -312,3 +313,27 @@ extern rtx count_loop_iterations	PARAMS ((struct loop_desc *, rtx, rtx));
 extern bool just_once_each_iteration_p	PARAMS ((struct loops *,struct loop *,
 						 basic_block));
 extern unsigned expected_loop_iterations PARAMS ((const struct loop *));
+
+/* Loop manipulation.  */
+extern bool can_duplicate_loop_p	PARAMS ((struct loop *loop));
+
+#define DLTHE_FLAG_UPDATE_FREQ	1	/* Update frequencies in
+					   duplicate_loop_to_header_edge.  */
+
+extern int duplicate_loop_to_header_edge PARAMS ((struct loop *, edge,
+						struct loops *, unsigned,
+						sbitmap, edge, edge *,
+						unsigned *, int));
+extern struct loop *loopify		PARAMS ((struct loops *, edge,
+						edge, basic_block));
+extern bool remove_path			PARAMS ((struct loops *, edge));
+extern edge split_loop_bb		PARAMS ((struct loops *, basic_block,
+						rtx));
+
+/* Loop optimizer initialization.  */
+extern struct loops *loop_optimizer_init PARAMS ((FILE *));
+extern void loop_optimizer_finalize	PARAMS ((struct loops *, FILE *));
+
+/* Optimization passes.  */
+extern void unswitch_loops		PARAMS ((struct loops *));
+

gcc/loop-init.c

+/* Loop optimizer initialization routines.
+   Copyright (C) 2002, 2003 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, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.  */
+
+#include "config.h"
+#include "system.h"
+#include "coretypes.h"
+#include "tm.h"
+#include "rtl.h"
+#include "hard-reg-set.h"
+#include "basic-block.h"
+#include "cfgloop.h"
+#include "cfglayout.h"
+#include "gcov-io.h"
+#include "profile.h"
+
+/* Initialize loop optimizer.  */
+
+struct loops *
+loop_optimizer_init (dumpfile)
+     FILE *dumpfile;
+{
+  struct loops *loops = xcalloc (1, sizeof (struct loops));
+  edge e;
+
+  /* Avoid annoying special cases of edges going to exit
+     block.  */
+  for (e = EXIT_BLOCK_PTR->pred; e; e = e->pred_next)
+    if ((e->flags & EDGE_FALLTHRU) && e->src->succ->succ_next)
+      split_edge (e);
+
+  /* Find the loops.  */
+
+  if (flow_loops_find (loops, LOOP_TREE) <= 1)
+    {
+      /* No loops.  */
+      flow_loops_free (loops);
+      free (loops);
+      return NULL;
+    }
+
+  /* Not going to update these.  */
+  free (loops->cfg.rc_order);
+  loops->cfg.rc_order = NULL;
+  free (loops->cfg.dfs_order);
+  loops->cfg.dfs_order = NULL;
+
+  /* Initialize structures for layout changes.  */
+  cfg_layout_initialize (loops);
+
+  /* Create pre-headers.  */
+  create_preheaders (loops, CP_SIMPLE_PREHEADERS | CP_INSIDE_CFGLAYOUT);
+
+  /* Force all latches to have only single successor.  */
+  force_single_succ_latches (loops);
+
+  /* Mark irreducible loops.  */
+  mark_irreducible_loops (loops);
+
+  /* Dump loops.  */
+  flow_loops_dump (loops, dumpfile, NULL, 1);
+
+#ifdef ENABLE_CHECKING
+  verify_dominators (loops->cfg.dom);
+  verify_loop_structure (loops);
+#endif
+
+  return loops;
+}
+
+/* Finalize loop optimizer.  */
+void
+loop_optimizer_finalize (loops, dumpfile)
+     struct loops *loops;
+     FILE *dumpfile;
+{
+  basic_block bb;
+
+  /* Finalize layout changes.  */
+  /* Make chain.  */
+  FOR_EACH_BB (bb)
+    if (bb->next_bb != EXIT_BLOCK_PTR)
+      RBI (bb)->next = bb->next_bb;
+
+  /* Another dump.  */
+  flow_loops_dump (loops, dumpfile, NULL, 1);
+
+  /* Clean up.  */
+  flow_loops_free (loops);
+  free (loops);
+ 
+  /* Finalize changes.  */
+  cfg_layout_finalize ();
+
+  /* Checking.  */
+#ifdef ENABLE_CHECKING
+  verify_flow_info ();
+#endif
+}
+

gcc/params.def

@@ -151,6 +151,17 @@ DEFPARAM(PARAM_MAX_UNROLLED_INSNS,
 	 "The maximum number of instructions to consider to unroll in a loop",
 	 100)
 
+/* The maximum number of insns of an unswitched loop.  */
+DEFPARAM(PARAM_MAX_UNSWITCH_INSNS,
+	"max-unswitch-insns",
+	"The maximum number of insns of an unswitched loop",
+	50)
+/* The maximum level of recursion in unswitch_single_loop.  */
+DEFPARAM(PARAM_MAX_UNSWITCH_LEVEL,
+	"max-unswitch-level",
+	"The maximum number of unswitchings in a single loop",
+	3)
+
 DEFPARAM(HOT_BB_COUNT_FRACTION,
 	 "hot-bb-count-fraction",
 	 "Select fraction of the maximal count of repetitions of basic block in \

gcc/toplev.c

@@ -239,6 +239,7 @@ enum dump_file_index
   DFI_bp,
   DFI_ce1,
   DFI_tracer,
+  DFI_loop2,
   DFI_cse2,
   DFI_life,
   DFI_combine,
@@ -289,6 +290,7 @@ static struct dump_file_info dump_file[DFI_MAX] =
   { "bp",	'b', 1, 0, 0 },
   { "ce1",	'C', 1, 0, 0 },
   { "tracer",	'T', 1, 0, 0 },
+  { "loop2",	'L', 1, 0, 0 },
   { "cse2",	't', 1, 0, 0 },
   { "life",	'f', 1, 0, 0 },	/* Yes, duplicate enable switch.  */
   { "combine",	'c', 1, 0, 0 },
@@ -519,6 +521,9 @@ int flag_unroll_loops;
 
 int flag_unroll_all_loops;
 
+/* Nonzero enables loop unswitching.  */
+int flag_unswitch_loops;
+
 /* Nonzero enables prefetch optimizations for arrays in loops.  */
 
 int flag_prefetch_loop_arrays;
@@ -997,6 +1002,8 @@ static const lang_independent_options f_options[] =
    N_("Perform loop unrolling when iteration count is known") },
   {"unroll-all-loops", &flag_unroll_all_loops, 1,
    N_("Perform loop unrolling for all loops") },
+  {"unswitch-loops", &flag_unswitch_loops, 1,
+   N_("Perform loop unswitching") },
   {"prefetch-loop-arrays", &flag_prefetch_loop_arrays, 1,
    N_("Generate prefetch instructions, if available, for arrays in loops") },
   {"move-all-movables", &flag_move_all_movables, 1,
@@ -3057,6 +3064,38 @@ rest_of_compilation (decl)
       timevar_pop (TV_TRACER);
     }
 
+  /* Perform loop optimalizations.  It might be better to do them a bit
+     sooner, but we want the profile feedback to work more efficiently.  */
+  if (optimize > 0
+      && flag_unswitch_loops)
+    {
+      struct loops *loops;
+      timevar_push (TV_LOOP);
+      open_dump_file (DFI_loop2, decl);
+      if (rtl_dump_file)
+	dump_flow_info (rtl_dump_file);
+
+      loops = loop_optimizer_init (rtl_dump_file);
+
+      if (loops)
+	{
+	  /* The optimalizations:  */
+	  if (flag_unswitch_loops)
+	    unswitch_loops (loops);
+
+	  loop_optimizer_finalize (loops, rtl_dump_file);
+	}
+
+      cleanup_cfg (CLEANUP_EXPENSIVE);
+      delete_trivially_dead_insns (insns, max_reg_num ());
+      reg_scan (insns, max_reg_num (), 0);
+      if (rtl_dump_file)
+	dump_flow_info (rtl_dump_file);
+      close_dump_file (DFI_loop2, print_rtl_with_bb, get_insns ());
+      timevar_pop (TV_LOOP);
+      ggc_collect ();
+    }
+
   if (flag_rerun_cse_after_loop)
     {
       timevar_push (TV_CSE2);
@@ -4884,6 +4923,7 @@ parse_options_and_default_flags (argc, argv)
     {
       flag_inline_functions = 1;
       flag_rename_registers = 1;
+      flag_unswitch_loops = 1;
     }
 
   if (optimize < 2 || optimize_size)