invoke.texi (align-threshold, [...]): Document.

	* invoke.texi (align-threshold, align-loop-iterations): Document.
	* final.c: Include cfgloop.h, params.h
	(compute_alignments): Dump decisions and compare them with loop
	structure; honor given parameters.
	(pass_compute_alignments): New dump file.
	* params.def (PARAM_ALIGN_THRESHOLD, PARAM_ALIGN_LOOP_ITERATIONS): New.
	* Makefile.in (final.o): Add dependency on cfgloop.h and params.h

From-SVN: r129162

5 files changed, 81 insertions, 7 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 03230d1..4717ca8 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,13 @@
+2007-10-09  Jan Hubicka  <jh@suse.cz>
+
+	* invoke.texi (align-threshold, align-loop-iterations): Document.
+	* final.c: Include cfgloop.h, params.h
+	(compute_alignments): Dump decisions and compare them with loop
+	structure; honor given parameters.
+	(pass_compute_alignments): New dump file.
+	* params.def (PARAM_ALIGN_THRESHOLD, PARAM_ALIGN_LOOP_ITERATIONS): New.
+	* Makefile.in (final.o): Add dependency on cfgloop.h and params.h
+
 2007-10-09  James E. Wilson  <wilson@specifix.com>
 
 	PR tree-optimization/33655
diff --git a/gcc/Makefile.in b/gcc/Makefile.in
index c31b259..5aee30e 100644
--- a/gcc/Makefile.in
+++ b/gcc/Makefile.in
@@ -2803,7 +2803,8 @@ final.o : final.c $(CONFIG_H) $(SYSTEM_H) coretypes.h $(TM_H) $(RTL_H) \
    insn-config.h $(INSN_ATTR_H) $(FUNCTION_H) output.h hard-reg-set.h \
    except.h debug.h xcoffout.h toplev.h reload.h dwarf2out.h tree-pass.h \
    $(BASIC_BLOCK_H) $(TM_P_H) $(TARGET_H) $(EXPR_H) $(CFGLAYOUT_H) dbxout.h \
-   $(TIMEVAR_H) $(CGRAPH_H) $(COVERAGE_H) $(REAL_H) $(DF_H) vecprim.h $(GGC_H)
+   $(TIMEVAR_H) $(CGRAPH_H) $(COVERAGE_H) $(REAL_H) $(DF_H) vecprim.h $(GGC_H) \
+   $(CFGLOOP_H) $(PARAMS_H)
 recog.o : recog.c $(CONFIG_H) $(SYSTEM_H) coretypes.h $(TM_H) $(RTL_H) \
    $(FUNCTION_H) $(BASIC_BLOCK_H) $(REGS_H) $(RECOG_H) $(EXPR_H) \
    $(FLAGS_H) insn-config.h $(INSN_ATTR_H) toplev.h output.h reload.h \
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index ab7da0f..e5318eb 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -6873,6 +6873,16 @@ with unknown.  We predict the known number of iterations correctly, while
 the unknown number of iterations average to roughly 10.  This means that the
 loop without bounds would appear artificially cold relative to the other one.
 
+@item align-threshold
+
+Select fraction of the maximal frequency of executions of basic block in
+function given basic block will get aligned.
+
+@item align-loop-iterations
+
+A loop expected to iterate at lest the selected number of iterations will get
+aligned.
+
 @item tracer-dynamic-coverage
 @itemx tracer-dynamic-coverage-feedback
 
diff --git a/gcc/final.c b/gcc/final.c
index b59a222..654f847 100644
--- a/gcc/final.c
+++ b/gcc/final.c
@@ -78,6 +78,8 @@ along with GCC; see the file COPYING3.  If not see
 #include "df.h"
 #include "vecprim.h"
 #include "ggc.h"
+#include "cfgloop.h"
+#include "params.h"
 
 #ifdef XCOFF_DEBUGGING_INFO
 #include "xcoffout.h"		/* Needed for external data
@@ -673,6 +675,8 @@ compute_alignments (void)
 {
   int log, max_skip, max_log;
   basic_block bb;
+  int freq_max = 0;
+  int freq_threshold = 0;
 
   if (label_align)
     {
@@ -688,6 +692,19 @@ compute_alignments (void)
   if (! optimize || optimize_size)
     return 0;
 
+  if (dump_file)
+    {
+      dump_flow_info (dump_file, TDF_DETAILS);
+      flow_loops_dump (dump_file, NULL, 1);
+      loop_optimizer_init (AVOID_CFG_MODIFICATIONS);
+    }
+  FOR_EACH_BB (bb)
+    if (bb->frequency > freq_max)
+      freq_max = bb->frequency;
+  freq_threshold = freq_max / PARAM_VALUE (PARAM_ALIGN_THRESHOLD);
+
+  if (dump_file)
+    fprintf(dump_file, "freq_max: %i\n",freq_max);
   FOR_EACH_BB (bb)
     {
       rtx label = BB_HEAD (bb);
@@ -697,7 +714,12 @@ compute_alignments (void)
 
       if (!LABEL_P (label)
 	  || probably_never_executed_bb_p (bb))
-	continue;
+	{
+	  if (dump_file)
+	    fprintf(dump_file, "BB %4i freq %4i loop %2i loop_depth %2i skipped.\n",
+		    bb->index, bb->frequency, bb->loop_father->num, bb->loop_depth);
+	  continue;
+	}
       max_log = LABEL_ALIGN (label);
       max_skip = LABEL_ALIGN_MAX_SKIP;
 
@@ -708,6 +730,18 @@ compute_alignments (void)
 	  else
 	    branch_frequency += EDGE_FREQUENCY (e);
 	}
+      if (dump_file)
+	{
+	  fprintf(dump_file, "BB %4i freq %4i loop %2i loop_depth %2i fall %4i branch %4i",
+		  bb->index, bb->frequency, bb->loop_father->num,
+		  bb->loop_depth,
+		  fallthru_frequency, branch_frequency);
+	  if (!bb->loop_father->inner && bb->loop_father->num)
+	    fprintf (dump_file, " inner_loop");
+	  if (bb->loop_father->header == bb)
+	    fprintf (dump_file, " loop_header");
+	  fprintf (dump_file, "\n");
+	}
 
       /* There are two purposes to align block with no fallthru incoming edge:
 	 1) to avoid fetch stalls when branch destination is near cache boundary
@@ -720,12 +754,14 @@ compute_alignments (void)
 	 when function is called.  */
 
       if (!has_fallthru
-	  && (branch_frequency > BB_FREQ_MAX / 10
+	  && (branch_frequency > freq_threshold
 	      || (bb->frequency > bb->prev_bb->frequency * 10
 		  && (bb->prev_bb->frequency
 		      <= ENTRY_BLOCK_PTR->frequency / 2))))
 	{
 	  log = JUMP_ALIGN (label);
+	  if (dump_file)
+	    fprintf(dump_file, "  jump alignment added.\n");
 	  if (max_log < log)
 	    {
 	      max_log = log;
@@ -736,10 +772,13 @@ compute_alignments (void)
 	 align it.  It is most likely a first block of loop.  */
       if (has_fallthru
 	  && maybe_hot_bb_p (bb)
-	  && branch_frequency + fallthru_frequency > BB_FREQ_MAX / 10
-	  && branch_frequency > fallthru_frequency * 2)
+	  && branch_frequency + fallthru_frequency > freq_threshold
+	  && (branch_frequency
+	      > fallthru_frequency * PARAM_VALUE (PARAM_ALIGN_LOOP_ITERATIONS)))
 	{
 	  log = LOOP_ALIGN (label);
+	  if (dump_file)
+	    fprintf(dump_file, "  internal loop alignment added.\n");
 	  if (max_log < log)
 	    {
 	      max_log = log;
@@ -749,12 +788,15 @@ compute_alignments (void)
       LABEL_TO_ALIGNMENT (label) = max_log;
       LABEL_TO_MAX_SKIP (label) = max_skip;
     }
+
+  if (dump_file)
+    loop_optimizer_finalize ();
   return 0;
 }
 
 struct tree_opt_pass pass_compute_alignments =
 {
-  NULL,                                 /* name */
+  "alignments",                         /* name */
   NULL,                                 /* gate */
   compute_alignments,                   /* execute */
   NULL,                                 /* sub */
@@ -765,7 +807,8 @@ struct tree_opt_pass pass_compute_alignments =
   0,                                    /* properties_provided */
   0,                                    /* properties_destroyed */
   0,                                    /* todo_flags_start */
-  0,                                    /* todo_flags_finish */
+  TODO_dump_func | TODO_verify_rtl_sharing
+  | TODO_ggc_collect,                   /* todo_flags_finish */
   0                                     /* letter */
 };
 
diff --git a/gcc/params.def b/gcc/params.def
index c9ae99b..5cb1f6d 100644
--- a/gcc/params.def
+++ b/gcc/params.def
@@ -334,6 +334,16 @@ DEFPARAM(HOT_BB_FREQUENCY_FRACTION,
 	 "Select fraction of the maximal frequency of executions of basic block in function given basic block needs to have to be considered hot",
 	 1000, 0, 0)
 
+DEFPARAM (PARAM_ALIGN_THRESHOLD,
+	  "align-threshold",
+	  "Select fraction of the maximal frequency of executions of basic block in function given basic block get alignment",
+	  100, 0, 0)
+
+DEFPARAM (PARAM_ALIGN_LOOP_ITERATIONS,
+	  "align-loop-iterations",
+	  "Loops iterating at least selected number of iterations will get loop alignement.",
+	  4, 0, 0)
+
 /* For guessed profiles, the loops having unknown number of iterations
    are predicted to iterate relatively few (10) times at average.
    For functions containing one loop with large known number of iterations