re PR target/18916 (mis-aligned vector code with copy memory (-maltivec))

PR target/18916
	* builtins.c (std_gimplify_va_arg_expr): Adjust alignment of *ap.
	* expr.h (struct locate_and_pad_arg_data): Add "boundary".
	* function.c (locate_and_pad_parm): Set new field.
	(assign_parm_find_stack_rtl): Use it instead of FUNCTION_ARG_BOUNDARY.
	Tweak where_pad test to include "none".  Always set mem align for
	stack_parm.
	(assign_parm_adjust_stack_rtl): Discard stack_parm if alignment
	not sufficient for type.
	(assign_parm_setup_block): If stack_parm is zero on entry, always
	make a new stack local.  Block move old stack parm if necessary
	to new aligned stack local.
	(assign_parm_setup_stack): Use a block move to handle
	potentially misaligned entry_parm.
	(assign_parms_unsplit_complex): Specify required alignment when
	creating stack local.
	* calls.c (compute_argument_addresses): Override alignment of stack
	arg calculated from its type with the alignment given by
	FUNCTION_ARG_BOUNDARY.
	(store_one_arg): Likewise.

From-SVN: r93179

gcc/ChangeLog

@@ -1,4 +1,28 @@
+2005-01-11  Alan Modra  <amodra@bigpond.net.au>
+
+	PR target/18916
+	* builtins.c (std_gimplify_va_arg_expr): Adjust alignment of *ap.
+	* expr.h (struct locate_and_pad_arg_data): Add "boundary".
+	* function.c (locate_and_pad_parm): Set new field.
+	(assign_parm_find_stack_rtl): Use it instead of FUNCTION_ARG_BOUNDARY.
+	Tweak where_pad test to include "none".  Always set mem align for
+	stack_parm.
+	(assign_parm_adjust_stack_rtl): Discard stack_parm if alignment
+	not sufficient for type.
+	(assign_parm_setup_block): If stack_parm is zero on entry, always
+	make a new stack local.  Block move old stack parm if necessary
+	to new aligned stack local.
+	(assign_parm_setup_stack): Use a block move to handle
+	potentially misaligned entry_parm.
+	(assign_parms_unsplit_complex): Specify required alignment when
+	creating stack local.
+	* calls.c (compute_argument_addresses): Override alignment of stack
+	arg calculated from its type with the alignment given by
+	FUNCTION_ARG_BOUNDARY.
+	(store_one_arg): Likewise.
+
 2005-01-11  Jan Beulich  <jbeulich@novell.com>
+
 	* config/ia64/ia64.md (zero_extendsidi2): Replace zxt4 by addp4.
 	Change respective itanium_class attribute to ialu.
 	(shladdp4_internal): New.

gcc/builtins.c

@@ -1,6 +1,6 @@
 /* Expand builtin functions.
    Copyright (C) 1988, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
-   2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
+   2000, 2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
 
 This file is part of GCC.
 
@@ -3894,6 +3894,18 @@ std_gimplify_va_arg_expr (tree valist, tree type, tree *pre_p, tree *post_p)
 		  build2 (BIT_AND_EXPR, TREE_TYPE (valist), valist_tmp, t));
       gimplify_and_add (t, pre_p);
     }
+  else
+    boundary = align;
+
+  /* If the actual alignment is less than the alignment of the type,
+     adjust the type accordingly so that we don't assume strict alignment
+     when deferencing the pointer.  */
+  boundary *= BITS_PER_UNIT;
+  if (boundary < TYPE_ALIGN (type))
+    {
+      type = build_variant_type_copy (type);
+      TYPE_ALIGN (type) = boundary;
+    }
 
   /* Compute the rounded size of the type.  */
   type_size = size_in_bytes (type);

gcc/calls.c

@@ -1,6 +1,7 @@
 /* Convert function calls to rtl insns, for GNU C compiler.
    Copyright (C) 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
-   1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
+   1999, 2000, 2001, 2002, 2003, 2004, 2005
+   Free Software Foundation, Inc.
 
 This file is part of GCC.
 
@@ -1357,6 +1358,7 @@ compute_argument_addresses (struct arg_data *args, rtx argblock, int num_actuals
 	  rtx offset = ARGS_SIZE_RTX (args[i].locate.offset);
 	  rtx slot_offset = ARGS_SIZE_RTX (args[i].locate.slot_offset);
 	  rtx addr;
+	  unsigned int align, boundary;
 
 	  /* Skip this parm if it will not be passed on the stack.  */
 	  if (! args[i].pass_on_stack && args[i].reg != 0)
@@ -1369,9 +1371,18 @@ compute_argument_addresses (struct arg_data *args, rtx argblock, int num_actuals
 
 	  addr = plus_constant (addr, arg_offset);
 	  args[i].stack = gen_rtx_MEM (args[i].mode, addr);
-	  set_mem_align (args[i].stack, PARM_BOUNDARY);
 	  set_mem_attributes (args[i].stack,
 			      TREE_TYPE (args[i].tree_value), 1);
+	  align = BITS_PER_UNIT;
+	  boundary = args[i].locate.boundary;
+	  if (args[i].locate.where_pad != downward)
+	    align = boundary;
+	  else if (GET_CODE (offset) == CONST_INT)
+	    {
+	      align = INTVAL (offset) * BITS_PER_UNIT | boundary;
+	      align = align & -align;
+	    }
+	  set_mem_align (args[i].stack, align);
 
 	  if (GET_CODE (slot_offset) == CONST_INT)
 	    addr = plus_constant (arg_reg, INTVAL (slot_offset));
@@ -1380,9 +1391,9 @@ compute_argument_addresses (struct arg_data *args, rtx argblock, int num_actuals
 
 	  addr = plus_constant (addr, arg_offset);
 	  args[i].stack_slot = gen_rtx_MEM (args[i].mode, addr);
-	  set_mem_align (args[i].stack_slot, PARM_BOUNDARY);
 	  set_mem_attributes (args[i].stack_slot,
 			      TREE_TYPE (args[i].tree_value), 1);
+	  set_mem_align (args[i].stack_slot, args[i].locate.boundary);
 
 	  /* Function incoming arguments may overlap with sibling call
 	     outgoing arguments and we cannot allow reordering of reads
@@ -4119,9 +4130,7 @@ store_one_arg (struct arg_data *arg, rtx argblock, int flags,
 				  NULL_RTX, TYPE_MODE (sizetype), 0);
 	}
 
-      /* Some types will require stricter alignment, which will be
+      parm_align = arg->locate.boundary;
-	 provided for elsewhere in argument layout.  */
-      parm_align = MAX (PARM_BOUNDARY, TYPE_ALIGN (TREE_TYPE (pval)));
 
       /* When an argument is padded down, the block is aligned to
 	 PARM_BOUNDARY, but the actual argument isn't.  */

gcc/expr.h

@@ -1,6 +1,7 @@
 /* Definitions for code generation pass of GNU compiler.
    Copyright (C) 1987, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998,
-   1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
+   1999, 2000, 2001, 2002, 2003, 2004, 2005
+   Free Software Foundation, Inc.
 
 This file is part of GCC.
 
@@ -116,6 +117,8 @@ struct locate_and_pad_arg_data
   struct args_size alignment_pad;
   /* Which way we should pad this arg.  */
   enum direction where_pad;
+  /* slot_offset is at least this aligned.  */
+  unsigned int boundary;
 };
 
 /* Add the value of the tree INC to the `struct args_size' TO.  */

gcc/function.c

@@ -1,6 +1,7 @@
 /* Expands front end tree to back end RTL for GCC.
    Copyright (C) 1987, 1988, 1989, 1991, 1992, 1993, 1994, 1995, 1996, 1997,
-   1998, 1999, 2000, 2001, 2002, 2003, 2004 Free Software Foundation, Inc.
+   1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005
+   Free Software Foundation, Inc.
 
 This file is part of GCC.
 
@@ -2405,22 +2406,21 @@ assign_parm_find_stack_rtl (tree parm, struct assign_parm_data_one *data)
 
   set_mem_attributes (stack_parm, parm, 1);
 
-  boundary = FUNCTION_ARG_BOUNDARY (data->promoted_mode, data->passed_type);
+  boundary = data->locate.boundary;
-  align = 0;
+  align = BITS_PER_UNIT;
 
   /* If we're padding upward, we know that the alignment of the slot
      is 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 == upward || data->entry_parm)
+  if (data->locate.where_pad != downward || data->entry_parm)
     align = boundary;
   else if (GET_CODE (offset_rtx) == CONST_INT)
     {
       align = INTVAL (offset_rtx) * BITS_PER_UNIT | boundary;
       align = align & -align;
     }
-  if (align > 0)
+  set_mem_align (stack_parm, align);
-    set_mem_align (stack_parm, align);
 
   if (data->entry_parm)
     set_reg_attrs_for_parm (data->entry_parm, stack_parm);
@@ -2492,7 +2492,6 @@ assign_parm_adjust_entry_rtl (struct assign_parm_data_one *data)
 /* 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)
 {
@@ -2501,8 +2500,12 @@ assign_parm_adjust_stack_rtl (struct assign_parm_data_one *data)
   /* 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 (STRICT_ALIGNMENT && stack_parm
+  if (stack_parm
-      && GET_MODE_ALIGNMENT (data->nominal_mode) > MEM_ALIGN (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,
@@ -2548,6 +2551,8 @@ assign_parm_setup_block (struct assign_parm_data_all *all,
 {
   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);
@@ -2593,30 +2598,34 @@ assign_parm_setup_block (struct assign_parm_data_all *all,
       return;
     }
 
+  size = int_size_in_bytes (data->passed_type);
+  size_stored = CEIL_ROUND (size, UNITS_PER_WORD);
+  if (stack_parm == 0)
+    {
+      stack_parm = assign_stack_local (BLKmode, size_stored,
+				       TYPE_ALIGN (data->passed_type));
+      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)
     {
-      HOST_WIDE_INT size = int_size_in_bytes (data->passed_type);
-      HOST_WIDE_INT size_stored = CEIL_ROUND (size, UNITS_PER_WORD);
       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 below in the
+	 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 (stack_parm == 0)
+      if (data->stack_parm == 0)
-	{
+	;
-	  stack_parm = assign_stack_local (BLKmode, size_stored, 0);
-	  data->stack_parm = stack_parm;
-	  PUT_MODE (stack_parm, GET_MODE (entry_parm));
-	  set_mem_attributes (stack_parm, parm, 1);
-	}
       else if (GET_CODE (entry_parm) == PARALLEL)
 	;
       else
@@ -2686,7 +2695,16 @@ assign_parm_setup_block (struct assign_parm_data_all *all,
 	move_block_from_reg (REGNO (entry_parm), mem,
 			     size_stored / UNITS_PER_WORD);
     }
+  else if (data->stack_parm == 0)
+    {
+      push_to_sequence (all->conversion_insns);
+      emit_block_move (stack_parm, data->entry_parm, GEN_INT (size),
+		       BLOCK_OP_NORMAL);
+      all->conversion_insns = get_insns ();
+      end_sequence ();
+    }
 
+  data->stack_parm = stack_parm;
   SET_DECL_RTL (parm, stack_parm);
 }
 
@@ -2887,6 +2905,7 @@ assign_parm_setup_stack (struct assign_parm_data_all *all, tree parm,
 {
   /* Value must be stored in the stack slot STACK_PARM during function
      execution.  */
+  bool to_conversion = false;
 
   if (data->promoted_mode != data->nominal_mode)
     {
@@ -2896,6 +2915,8 @@ assign_parm_setup_stack (struct assign_parm_data_all *all, tree parm,
       emit_move_insn (tempreg, validize_mem (data->entry_parm));
 
       push_to_sequence (all->conversion_insns);
+      to_conversion = true;
+
       data->entry_parm = convert_to_mode (data->nominal_mode, tempreg,
 					  TYPE_UNSIGNED (TREE_TYPE (parm)));
 
@@ -2903,33 +2924,43 @@ assign_parm_setup_stack (struct assign_parm_data_all *all, tree parm,
 	/* ??? This may need a big-endian conversion on sparc64.  */
 	data->stack_parm
 	  = adjust_address (data->stack_parm, data->nominal_mode, 0);
-
-      all->conversion_insns = get_insns ();
-      end_sequence ();
     }
 
   if (data->entry_parm != data->stack_parm)
     {
+      rtx src, dest;
+
       if (data->stack_parm == 0)
 	{
 	  data->stack_parm
 	    = assign_stack_local (GET_MODE (data->entry_parm),
 				  GET_MODE_SIZE (GET_MODE (data->entry_parm)),
-				  0);
+				  TYPE_ALIGN (data->passed_type));
 	  set_mem_attributes (data->stack_parm, parm, 1);
 	}
 
-      if (data->promoted_mode != data->nominal_mode)
+      dest = validize_mem (data->stack_parm);
+      src = validize_mem (data->entry_parm);
+
+      if (MEM_P (src))
 	{
-	  push_to_sequence (all->conversion_insns);
+	  /* Use a block move to handle potentially misaligned entry_parm.  */
-	  emit_move_insn (validize_mem (data->stack_parm),
+	  if (!to_conversion)
-			  validize_mem (data->entry_parm));
+	    push_to_sequence (all->conversion_insns);
-	  all->conversion_insns = get_insns ();
+	  to_conversion = true;
-	  end_sequence ();
+
+	  emit_block_move (dest, src,
+			   GEN_INT (int_size_in_bytes (data->passed_type)),
+			   BLOCK_OP_NORMAL);
 	}
       else
-	emit_move_insn (validize_mem (data->stack_parm),
+	emit_move_insn (dest, src);
-			validize_mem (data->entry_parm));
+    }
+
+  if (to_conversion)
+    {
+      all->conversion_insns = get_insns ();
+      end_sequence ();
     }
 
   SET_DECL_RTL (parm, data->stack_parm);
@@ -2967,7 +2998,8 @@ assign_parms_unsplit_complex (struct assign_parm_data_all *all, tree fnargs)
 
 	      /* split_complex_arg put the real and imag parts in
 		 pseudos.  Move them to memory.  */
-	      tmp = assign_stack_local (DECL_MODE (parm), size, 0);
+	      tmp = assign_stack_local (DECL_MODE (parm), size,
+					TYPE_ALIGN (TREE_TYPE (parm)));
 	      set_mem_attributes (tmp, parm, 1);
 	      rmem = adjust_address_nv (tmp, inner, 0);
 	      imem = adjust_address_nv (tmp, inner, GET_MODE_SIZE (inner));
@@ -3411,6 +3443,7 @@ locate_and_pad_parm (enum machine_mode passed_mode, tree type, int in_regs,
   where_pad = FUNCTION_ARG_PADDING (passed_mode, type);
   boundary = FUNCTION_ARG_BOUNDARY (passed_mode, type);
   locate->where_pad = where_pad;
+  locate->boundary = boundary;
 
 #ifdef ARGS_GROW_DOWNWARD
   locate->slot_offset.constant = -initial_offset_ptr->constant;