RISC-V: Handle vector type alignment.

For riscv64-linux target, first half of fix for
FAIL: gdb.base/gnu_vector.exp: call add_various_floatvecs

GCC gives vectors natural aligment based on total size, not element size,
bounded by the maximum supported type alignment.

	gdb/
	* riscv-tdep.c (BIGGEST_ALIGNMENT): New.
	(riscv_type_alignment) <TYPE_CODE_ARRAY>: If TYPE_VECTOR, return min
	of TYPE_LENGTH and BIGGEST_ALIGNMENT.

gdb/ChangeLog

@@ -1,5 +1,9 @@
 2018-11-14  Jim Wilson  <jimw@sifive.com>
 
+	* riscv-tdep.c (BIGGEST_ALIGNMENT): New.
+	(riscv_type_alignment) <TYPE_CODE_ARRAY>: If TYPE_VECTOR, return min
+	of TYPE_LENGTH and BIGGEST_ALIGNMENT.
+
 	* riscv-tdep.c (riscv_call_arg_scalar_int): Use std::min when
 	setting len.  New local align, set to max of arg align and xlen,
 	and pass to first riscv_assign_stack_location call.

gdb/riscv-tdep.c

@@ -59,6 +59,9 @@
 /* The stack must be 16-byte aligned.  */
 #define SP_ALIGNMENT 16
 
+/* The biggest alignment that the target supports.  */
+#define BIGGEST_ALIGNMENT 16
+
 /* Forward declarations.  */
 static bool riscv_has_feature (struct gdbarch *gdbarch, char feature);
 
@@ -1642,6 +1645,10 @@ riscv_type_alignment (struct type *t)
       return TYPE_LENGTH (t);
 
     case TYPE_CODE_ARRAY:
+      if (TYPE_VECTOR (t))
+	return std::min (TYPE_LENGTH (t), (unsigned) BIGGEST_ALIGNMENT);
+      /* FALLTHROUGH */
+
     case TYPE_CODE_COMPLEX:
       return riscv_type_alignment (TYPE_TARGET_TYPE (t));