public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] Optimiza aggregate a = b = c = {} (PR c/78408)
@ 2016-11-25 19:32 Jakub Jelinek
  2016-11-28  9:50 ` Richard Biener
  0 siblings, 1 reply; 11+ messages in thread
From: Jakub Jelinek @ 2016-11-25 19:32 UTC (permalink / raw)
  To: Richard Biener; +Cc: gcc-patches

Hi!

This patch optimizes a = {}; b = a; into a = {}; b = {};
and similarly for memset instead of the first stmt and/or memcpy
instead of the second one.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2016-11-25  Jakub Jelinek  <jakub@redhat.com>

	PR c/78408
	* tree-ssa-ccp.c: Include tree-dfa.h.
	(optimize_memcpy): New function.
	(pass_fold_builtins::execute): Use it.  Remove useless conditional
	break after BUILT_IN_VA_*.

	* gcc.dg/pr78408.c: New test.

--- gcc/tree-ssa-ccp.c.jj	2016-11-18 20:04:27.000000000 +0100
+++ gcc/tree-ssa-ccp.c	2016-11-25 17:54:26.862166658 +0100
@@ -143,6 +143,7 @@ along with GCC; see the file COPYING3.
 #include "stor-layout.h"
 #include "optabs-query.h"
 #include "tree-ssa-ccp.h"
+#include "tree-dfa.h"
 
 /* Possible lattice values.  */
 typedef enum
@@ -2928,6 +2929,113 @@ optimize_atomic_bit_test_and (gimple_stm
   release_ssa_name (lhs);
 }
 
+/* Optimize
+   a = {};
+   b = a;
+   into
+   a = {};
+   b = {};
+   Similarly for memset (&a, ..., sizeof (a)); instead of a = {};
+   and/or memcpy (&b, &a, sizeof (a)); instead of b = a;  */
+
+static void
+optimize_memcpy (gimple_stmt_iterator *gsip, tree dest, tree src)
+{
+  gimple *stmt = gsi_stmt (*gsip);
+  if (gimple_has_volatile_ops (stmt)
+      || TREE_THIS_VOLATILE (dest)
+      || TREE_THIS_VOLATILE (src))
+    return;
+
+  tree vuse = gimple_vuse (stmt);
+  if (vuse == NULL)
+    return;
+
+  gimple *defstmt = SSA_NAME_DEF_STMT (vuse);
+  tree src2 = NULL_TREE;
+  tree val = integer_zero_node;
+  if (gimple_store_p (defstmt)
+      && gimple_assign_single_p (defstmt)
+      && TREE_CODE (gimple_assign_rhs1 (defstmt)) == CONSTRUCTOR
+      && CONSTRUCTOR_NELTS (gimple_assign_rhs1 (defstmt)) == 0
+      && !gimple_clobber_p (defstmt))
+    src2 = gimple_assign_lhs (defstmt);
+  else if (gimple_call_builtin_p (defstmt, BUILT_IN_MEMSET)
+	   && TREE_CODE (gimple_call_arg (defstmt, 0)) == ADDR_EXPR
+	   && TREE_CODE (gimple_call_arg (defstmt, 1)) == INTEGER_CST
+	   && TREE_CODE (gimple_call_arg (defstmt, 2)) == INTEGER_CST)
+    {
+      HOST_WIDE_INT ssize, max_size, off;
+      bool reverse;
+      src2 = TREE_OPERAND (gimple_call_arg (defstmt, 0), 0);
+      get_ref_base_and_extent (src2, &off, &ssize, &max_size, &reverse);
+      if (ssize != max_size
+	  || (ssize % BITS_PER_UNIT) != 0
+	  || !wi::eq_p (gimple_call_arg (defstmt, 2), ssize / BITS_PER_UNIT))
+	src2 = NULL_TREE;
+      else
+	{
+	  val = gimple_call_arg (defstmt, 1);
+	  if (!integer_zerop (val) && is_gimple_assign (stmt))
+	    src2 = NULL_TREE;
+	}
+    }
+
+  if (src2 == NULL_TREE)
+    return;
+
+  if (!operand_equal_p (src, src2, 0))
+    {
+      /* Handle also
+	 a = {};
+	 MEM[(char * {ref-all})&b] = MEM[(char * {ref-all})&a];  */
+      if (is_gimple_assign (stmt)
+	  && TREE_CODE (src) == MEM_REF
+	  && integer_zerop (TREE_OPERAND (src, 1))
+	  && TREE_CODE (TREE_OPERAND (src, 0)) == ADDR_EXPR
+	  && DECL_P (src2)
+	  && operand_equal_p (TREE_OPERAND (TREE_OPERAND (src, 0), 0),
+			      src2, 0)
+	  && tree_int_cst_equal (TYPE_SIZE (TREE_TYPE (src)),
+				 DECL_SIZE (src2)))
+	src = TREE_OPERAND (TREE_OPERAND (src, 0), 0);
+      else
+	return;
+    }
+  if (refs_may_alias_p (dest, src))
+    return;
+
+  if (dump_file && (dump_flags & TDF_DETAILS))
+    {
+      fprintf (dump_file, "Simplified\n  ");
+      print_gimple_stmt (dump_file, stmt, 0, dump_flags);
+      fprintf (dump_file, "after previous\n  ");
+      print_gimple_stmt (dump_file, defstmt, 0, dump_flags);
+    }
+
+  if (is_gimple_assign (stmt))
+    {
+      tree ctor = build_constructor (TREE_TYPE (dest), NULL);
+      gimple_assign_set_rhs_from_tree (gsip, ctor);
+      update_stmt (stmt);
+    }
+  else
+    {
+      gcall *call = as_a <gcall *> (stmt);
+      tree fndecl = builtin_decl_implicit (BUILT_IN_MEMSET);
+      gimple_call_set_fndecl (call, fndecl);
+      gimple_call_set_fntype (call, TREE_TYPE (fndecl));
+      gimple_call_set_arg (call, 1, val);
+      update_stmt (stmt);
+    }
+
+  if (dump_file && (dump_flags & TDF_DETAILS))
+    {
+      fprintf (dump_file, "into\n  ");
+      print_gimple_stmt (dump_file, stmt, 0, dump_flags);
+    }
+}
+
 /* A simple pass that attempts to fold all builtin functions.  This pass
    is run after we've propagated as many constants as we can.  */
 
@@ -2994,6 +3102,9 @@ pass_fold_builtins::execute (function *f
 		      continue;
 		    }
 		}
+	      else if (gimple_assign_load_p (stmt) && gimple_store_p (stmt))
+		optimize_memcpy (&i, gimple_assign_lhs (stmt),
+				 gimple_assign_rhs1 (stmt));
 	      gsi_next (&i);
 	      continue;
 	    }
@@ -3109,14 +3220,39 @@ pass_fold_builtins::execute (function *f
 						false, false);
 		  break;
 
+		case BUILT_IN_MEMCPY:
+		  if (gimple_call_builtin_p (stmt, BUILT_IN_NORMAL)
+		      && TREE_CODE (gimple_call_arg (stmt, 0)) == ADDR_EXPR
+		      && TREE_CODE (gimple_call_arg (stmt, 1)) == ADDR_EXPR
+		      && TREE_CODE (gimple_call_arg (stmt, 2)) == INTEGER_CST)
+		    {
+		      tree dest = TREE_OPERAND (gimple_call_arg (stmt, 0), 0);
+		      tree src = TREE_OPERAND (gimple_call_arg (stmt, 1), 0);
+		      HOST_WIDE_INT dsize, ssize, max_size, off;
+		      bool reverse;
+		      get_ref_base_and_extent (dest, &off, &dsize, &max_size,
+					       &reverse);
+		      if (dsize != max_size)
+			break;
+		      get_ref_base_and_extent (src, &off, &ssize, &max_size,
+					       &reverse);
+		      if (ssize != max_size
+			  || ssize != dsize
+			  || (ssize % BITS_PER_UNIT) != 0)
+			break;
+		      if (!wi::eq_p (gimple_call_arg (stmt, 2),
+				     ssize / BITS_PER_UNIT))
+			break;
+		      optimize_memcpy (&i, dest, src);
+		    }
+		  break;
+
 		case BUILT_IN_VA_START:
 		case BUILT_IN_VA_END:
 		case BUILT_IN_VA_COPY:
 		  /* These shouldn't be folded before pass_stdarg.  */
 		  result = optimize_stdarg_builtin (stmt);
-		  if (result)
-		    break;
-		  /* FALLTHRU */
+		  break;
 
 		default:;
 		}
--- gcc/testsuite/gcc.dg/pr78408.c.jj	2016-11-25 18:02:47.344752199 +0100
+++ gcc/testsuite/gcc.dg/pr78408.c	2016-11-25 18:02:26.000000000 +0100
@@ -0,0 +1,78 @@
+/* PR c/78408 */
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-fab1-details" } */
+/* { dg-final { scan-tree-dump-times "after previous" 16 "fab1" } } */
+
+struct S { char a[32]; };
+struct T { char a[65536]; };
+void bar (int, struct S *, struct S *, struct T *, struct T *);
+
+void
+f1 (void)
+{
+  struct S a, b;
+  struct T c, d;
+  a = b = (struct S) {};
+  c = d = (struct T) {};
+  bar (1, &a, &b, &c, &d);
+}
+
+void
+f2 (void)
+{
+  struct S a, b;
+  struct T c, d;
+  b = (struct S) {};
+  a = b;
+  d = (struct T) {};
+  c = d;
+  bar (2, &a, &b, &c, &d);
+}
+
+void
+f3 (void)
+{
+  struct S a, b;
+  struct T c, d;
+  __builtin_memset (&b, 0, sizeof (b));
+  a = b;
+  __builtin_memset (&d, 0, sizeof (d));
+  c = d;
+  bar (3, &a, &b, &c, &d);
+}
+
+
+void
+f4 (void)
+{
+  struct S a, b;
+  struct T c, d;
+  b = (struct S) {};
+  __builtin_memcpy (&a, &b, sizeof (b));
+  d = (struct T) {};
+  __builtin_memcpy (&c, &d, sizeof (d));
+  bar (4, &a, &b, &c, &d);
+}
+
+void
+f5 (void)
+{
+  struct S a, b;
+  struct T c, d;
+  __builtin_memset (&b, 0, sizeof (b));
+  __builtin_memcpy (&a, &b, sizeof (b));
+  __builtin_memset (&d, 0, sizeof (d));
+  __builtin_memcpy (&c, &d, sizeof (d));
+  bar (5, &a, &b, &c, &d);
+}
+
+void
+f6 (void)
+{
+  struct S a, b, e, g;
+  struct T c, d, f, h;
+  g = e = a = b = (struct S) {};
+  h = f = c = d = (struct T) {};
+  bar (6, &a, &b, &c, &d);
+  bar (6, &e, &g, &f, &h);
+}

	Jakub

^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2016-12-19 20:17 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2016-11-25 19:32 [PATCH] Optimiza aggregate a = b = c = {} (PR c/78408) Jakub Jelinek
2016-11-28  9:50 ` Richard Biener
2016-12-13 11:36   ` Jakub Jelinek
2016-12-14 12:29     ` Richard Biener
2016-12-15 16:44       ` [PATCH] Optimiza aggregate a = b = c = {} (PR c/78408, take 2) Jakub Jelinek
2016-12-16 12:53         ` Richard Biener
2016-12-16 13:57           ` Jakub Jelinek
2016-12-16 13:17             ` Richard Biener
2016-12-16 14:02               ` Jakub Jelinek
2016-12-16 14:15                 ` Richard Biener
2016-12-19 20:24           ` Jeff Law

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).