Echelon Neuron - Manual (Page 62)

Arrays
The Neuron Chip and Smart Transceiver hardware has no data alignment
requirement, beyond the natural byte alignment. Thus, arrays are consecutive
elements without intervening padding. The first element (the element at index
0) is at the lowest address of the array. The remaining elements appear
consecutively at addresses that increase as multiples of the array-element size.
Arrays are therefore packed objects with no unused space between the elements.
Array sizes are always exactly the size of an individual element multiplied by the
number of elements in the array.
Structures and Unions
A structure is a group of fields packed into consecutive byte locations. The first
field is at the lowest address of the structure. The remaining fields appear
consecutively at addresses that increase by adding the size of the previous fields
to the beginning address of the structure. The size of a structure is always
exactly the sum of the sizes of each individual element in the structure.
A union is a group of overlaid fields. Each field starts at the beginning of the
union. The size of a union is always exactly the maximum of the sizes of each
individual field in the union.
Bitfields
Bitfields in a Neuron C structure or union are always packed within one or more
bytes. A single bitfield cannot cross a byte boundary; thus, the maximum bitfield
size is eight bits. The first bitfield in a byte uses the most significant bits, and
subsequent bitfields within the byte use the next most significant bits until all
bits are used.
Anonymous bitfields of non-zero size are supported as placeholders, allowing for
exact positioning of the named bitfields within their container byte. Anonymous
bitfields of size zero always complete the current container.
A new byte is begun when a bitfield is encountered that is too large for the
remaining bits, or when an anonymous bitfield is encountered with size 0, or
when a non-bitfield object is encountered.
Calling a Neuron C Function from Assembly Code
When a Neuron assembly function calls a Neuron C function, the main task of
the assembly function is to prepare the stack for the function by pushing the
function arguments in the correct order, and to call the function. Because the
called function clears the stack as part of its operation, the calling assembly
function need only retrieve the function’s result data (if any) from the stack.
The general approach to calling Neuron C functions from assembly code includes
performing the following steps:
1. Import the symbol for the function.
2. Preserve all general-purpose registers in use.
3. Push the function arguments onto the data stack, from right to left.
52 Interfacing with a Neuron C Application