Introduction
File Name Conventions
Object files produced by the
assembler are, by
convention, appended with a suffix of
.preobj to distinguish them
from other file types.
Object File Format
The
assembler produces relocatable object files
which may be
linked with other object
files to produce a executable
binary
image which can be executed by the simulation
machine. The object file is split into a number of sections as displayed
in the diagram below :
Header
The header of the object file mainly contains version details and other
properties about the file such as what the source file was called. It also
contains placeholder values for base offsets used in relocation. The
header can mainly be regarded as non-vital to the process of compilation
but may be extended in the future.
Export Dictionary
The export dictionary contains a number of entries which explain which
symbols in the object are availible for other object to use. Each entry
contains the symbol name and the address of the symbol in the instruction
code. This mechanism allows other object files to declare a symbol external
for the linker to resolve against symbols declared as exported in other
object files.
Relocation Dictionary
In order for the instruction code to be relocatable, the object file needs
to know which instructions require relocation and by which method. Currently
instructions which makde references to internal symbols are marked as needing
relative relocation and those which reference external symbols are marked
as needing absolute relocation. Partition relocation entries inform the
linker which partition name an instruction uses so that the final partition
number can be patched.
Partition Dictionary
The partition dictionary holds information about the partitions that should
be created to service this object file. Partition parameters such as size
and stride are store alongside the symbolic name of the partition. This
is used in conjunction with relocation entries to patch instructions and
make them use the correct partition numbers.
Instruction Code
The instruction code is a block of
encoded instructions which represent the code for the object file.
These instructions are in a relocateable form and not suitable for direct
execution on the simulation
machine.
In order to create executable code, the instruction code block needs to
have relocation performed on the instructions so that they address the
right locations in memory and use the right partition numbers.
Constant Pool
The constant pool is a block of constant string and numeric data as
specified in the constant segment of the
assembly
language program.
