short_bit. Number of bits in a C/C++ short variable. Default is 2*TARGET_CHAR_BIT. TARGET_CHAR_BIT is a defined constant, which if not set explicitly defaults to 8.

int_bit, long_bit, long_long_bit, float_bit, double_bit, long_double_bit. These are analogous to short and are the number of bits in a C/C++ variable of the corresponding time. Defaults are 4*TARGET_CHAR_BIT for int, long and float and 4*TARGET_CHAR_BIT for long long, double and long double.

ptr_bit. Number of bits in a C/C++ pointer. Default is 4*TARGET_CHAR_BIT.

addr_bit. Number of bits in a C/C++ address. Almost always this is the same as the number of bits in a pointer, but there are a small number of architectures for which pointers cannot reach all addresses. Default is 4*TARGET_CHAR_BIT.

float_format, double_format and long_double_format. These point to an array of C structs (one for each endianism), defining the format for each of the floating point types. A number of these arrays are predefined. They in turn are built on top of a set of standard types defined by the library libiberty.

char_signed. 1 if char to be treated as signed, 0 if char is to be treated as unsigned. The default is -1 (undefined), so this should always be set.