C6X port 8/11: A new FUNCTION_ARG macro

Index: doc/tm.texi
===================================================================
--- doc/tm.texi.orig
+++ doc/tm.texi
@@ -4163,9 +4163,10 @@  to pad out an argument with extra space.
 @code{enum direction}: either @code{upward} to pad above the argument,
 @code{downward} to pad below, or @code{none} to inhibit padding.
 
-The @emph{amount} of padding is always just enough to reach the next
-multiple of @code{TARGET_FUNCTION_ARG_BOUNDARY}; this macro does not
-control it.
+The @emph{amount} of padding is not controlled by this macro.  It is
+always just enough to reach the next multiple of the alignment boundary,
+which is usually @code{PARM_BOUNDARY}, or @code{FUNCTION_ARG_BOUNDARY}
+if @code{FUNCTION_ARG_ROUND_TO_ARG_BOUNDARY} is defined.
 
 This macro has a default definition which is right for most systems.
 For little-endian machines, the default is to pad upward.  For
@@ -4198,6 +4199,12 @@  with the specified mode and type.  The d
 @code{PARM_BOUNDARY} for all arguments.
 @end deftypefn
 
+@defmac FUNCTION_ARG_ROUND_TO_ARG_BOUNDARY
+Normally, the size of an argument is rounded up to @code{PARM_BOUNDARY}.
+Define this macro if you want the value of @code{FUNCTION_ARG_BOUNDARY}
+to be used for this rounding instead.
+@end defmac
+
 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
 A C expression that is nonzero if @var{regno} is the number of a hard
 register in which function arguments are sometimes passed.  This does
Index: doc/tm.texi.in
===================================================================
--- doc/tm.texi.in.orig
+++ doc/tm.texi.in
@@ -4151,9 +4151,10 @@  to pad out an argument with extra space.
 @code{enum direction}: either @code{upward} to pad above the argument,
 @code{downward} to pad below, or @code{none} to inhibit padding.
 
-The @emph{amount} of padding is always just enough to reach the next
-multiple of @code{TARGET_FUNCTION_ARG_BOUNDARY}; this macro does not
-control it.
+The @emph{amount} of padding is not controlled by this macro.  It is
+always just enough to reach the next multiple of the alignment boundary,
+which is usually @code{PARM_BOUNDARY}, or @code{FUNCTION_ARG_BOUNDARY}
+if @code{FUNCTION_ARG_ROUND_TO_ARG_BOUNDARY} is defined.
 
 This macro has a default definition which is right for most systems.
 For little-endian machines, the default is to pad upward.  For
@@ -4186,6 +4187,12 @@  with the specified mode and type.  The d
 @code{PARM_BOUNDARY} for all arguments.
 @end deftypefn
 
+@defmac FUNCTION_ARG_ROUND_TO_ARG_BOUNDARY
+Normally, the size of an argument is rounded up to @code{PARM_BOUNDARY}.
+Define this macro if you want the value of @code{FUNCTION_ARG_BOUNDARY}
+to be used for this rounding instead.
+@end defmac
+
 @defmac FUNCTION_ARG_REGNO_P (@var{regno})
 A C expression that is nonzero if @var{regno} is the number of a hard
 register in which function arguments are sometimes passed.  This does
Index: function.c
===================================================================
--- function.c.orig
+++ function.c
@@ -3709,7 +3709,7 @@  locate_and_pad_parm (enum machine_mode p
 {
   tree sizetree;
   enum direction where_pad;
-  unsigned int boundary;
+  unsigned int boundary, round_boundary;
   int reg_parm_stack_space = 0;
   int part_size_in_regs;
 
@@ -3741,6 +3741,11 @@  locate_and_pad_parm (enum machine_mode p
     = type ? size_in_bytes (type) : size_int (GET_MODE_SIZE (passed_mode));
   where_pad = FUNCTION_ARG_PADDING (passed_mode, type);
   boundary = targetm.calls.function_arg_boundary (passed_mode, type);
+#ifdef FUNCTION_ARG_ROUND_TO_ARG_BOUNDARY
+  round_boundary = boundary;
+#else
+  round_boundary = PARM_BOUNDARY;
+#endif
   locate->where_pad = where_pad;
 
   /* Alignment can't exceed MAX_SUPPORTED_STACK_ALIGNMENT.  */
@@ -3787,8 +3792,8 @@  locate_and_pad_parm (enum machine_mode p
     tree s2 = sizetree;
     if (where_pad != none
 	&& (!host_integerp (sizetree, 1)
-	    || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % PARM_BOUNDARY))
-      s2 = round_up (s2, PARM_BOUNDARY / BITS_PER_UNIT);
+	    || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % round_boundary))
+      s2 = round_up (s2, round_boundary / BITS_PER_UNIT);
     SUB_PARM_SIZE (locate->slot_offset, s2);
   }
 
@@ -3840,8 +3845,8 @@  locate_and_pad_parm (enum machine_mode p
 
   if (where_pad != none
       && (!host_integerp (sizetree, 1)
-	  || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % PARM_BOUNDARY))
-    sizetree = round_up (sizetree, PARM_BOUNDARY / BITS_PER_UNIT);
+	  || (tree_low_cst (sizetree, 1) * BITS_PER_UNIT) % round_boundary))
+    sizetree = round_up (sizetree, round_boundary / BITS_PER_UNIT);
 
   ADD_PARM_SIZE (locate->size, sizetree);