nasm/rdoff/README

RDOFF Utils v0.3
================

The files contained in this directory are the C source code of a set
of tools (and general purpose library files) for the manipulation of
RDOFF version 2 object files. Note that these programs (with the
exception of 'rdfdump') will NOT work with version 1 object files. See
the subdirectory v1 for programs that perform that task. 

Note: If you do not have a v1 subdirectory, you may have unpacked the
ZIP file without specifying the 'restore directory structure' option -
delete these files, and run your ZIP extracter again with this option 
turned on ('-d' for PKUNZIP).

RDOFF version 1 is no longer really supported, you should be using
v2 instead now.

There is also a 'Changes' file, which documents the differences between
RDOFF 1 and 2, and an 'rdoff2.txt' file, with complete documentation for the
new format.

Here is a brief summary of the programs' usage:

rdfdump
=======

This tool prints a list of the header records in an RDOFF object in 
human-readable form, and optionally prints a hex dump of the contents
of the segments.

Usage:
	rdfdump [-v] filename

The -v flag specifies that the hex dump (see above) should be printed.

Changes from previous versions:

* rdfdump supports both version 1 and 2 of RDOFF.
* rdfdump now gives warnings if the RDOFF2 format is violated (it
  looks for incorrect lengths for header records, and checks the
  overall length count at the start of the file)

ldrdf
=====

This tool is a version of unix 'ld' (or DOS 'link') for use with RDOFF
files. It is capable of linking RDOFF objects, and libraries produced
with the 'rdlib' utility discussed below.

In normal usage, its command line takes the form:

	ldrdf [-o output-file] object files [-llibrary ...]

Libraries must be specified with their path as no search is performed.
Modules in libraries are not linked to the program unless they are
referred to.

Most of its options are not implemented, but those that are are listed here:

   -v	increase verbosity level. Currently 4 verbosity levels are 
	available: default (which only prints error information), normal
	(which prints information about the produced object, -v), medium
	(which prints information about what the program is doing, -v -v)
	and high (which prints all available information, -v -v -v).

   -p	change alignment value to which multiple segments combigned into
	a single segment should be aligned (must be either 1, 2, 4, 8,
	16, 32 or 256. Default is 16).

The default output filename is 'aout.rdx'.

rdx
===

This program simply loads and executes an RDOFF object, by calling
'_main', which it expects to be a C-style function, which will accept
two parameters, argc and argv in normal C style.

rdflib
======

This program creates a library file for use with ldrdf.

It is supplied with a shell script 'makelib' which should probably be used
to create libraries.

Usage:
	rdflib command library [optional arguments]

Valid commands are:

	c	Create (or truncate) the library
	a	Add a module (requires a filename and a name to give the
		module, ie 'rdflib a libc.rdl strcpy.rdf strcpy' puts the
		file 'strcpy.rdf' into 'libc.rdl', and calls it 'strcpy'.
	x	Extract (arguments are the opposite to the 'a' command,
		ie you'd do 'rdflib x libc.rdl strcpy strcpy.rdf to get
		a copy of strcpy.rdf back out again...)
        t       List modules in the library

A remove command will be added soon (it is already documented
as existing, but I haven't had time to implement it... if anyone
else wants to do this, they're welcome to. The file format should be
amply documented in the source code... look at 'rdflib.c' and 'rdlib.c',
and the relevant sections of 'ldrdf.c' to see how libraries can be
handled).

Library functions
=================

The files 'rdoff.c', 'rdoff.h', 'rdfload.c' and 'rdfload.h' contain
code which you may find useful. They retain the same interface as
the previous version, so any code that used them previously should
still work OK (maybe). 'rdoff.c' contains at the top a line:

#define STRICT_ERRORS

Comment this line out if you wish to use record types other than the
7 predefined types; it will then not report such records as an error,
but accept them gracefully, and read them byte for byte into
a 'generic record' (see the definition of GenericRec in 'rdoff.h').

If you are using these functions to write RDF modules (rather than
just reading them), then please note the existance of a new function
'rdfaddsegment(rdf_headerbuf,long)'. This must be called once for
each segment in your object, to tell the header writing functions
how long the segment is.

BUGS
====

This product has recently undergone a major revision, and as such there 
are probably several bugs left over from the testing phase (although the
previous version had quite a few that have now been fixed!). Could you
please report any bugs to me at the address below, including the following
information:

  - A description of the bug
  - What you think the program should be doing
  - Which programs you are using
  - Which operating system you are using, and which C compiler was used to
    compile the programs (or state that the pre-compiled versions were used).
  - If appropriate, any of the following:
    * source code (preferably cut down to a minimum that will still assemble
      and show the bug)
    * the output of rdfdump on produced modules (or send the module if the
      problem is in code generated)
    * exact descriptions of error messages/symptoms/etc

TODO
====

There are still various things unimplemented that I would like to add.
If you want to find out what these are, search near the top of each *.c
file for a comment containing the word 'TODO'. A brief list is given here:

- Improve the performace of ldrdf (there are several enhancements I can think
  of that wouldn't be too hard to add)
- Stop assuming that we're on a little endian machine
- Make everything work with both formats (?)
- Add extra functions to ldrdf (strip symbols/keep symbol list)
- Check for more bugs

One last thing I have to say: good luck! Whatever it is that you want to use
RDOFF for, I hope its a success.  People out there are using it for many
diverse applications, from operating system boot-loaders to loadable modules
in games.  Whatever your application is, I hope that it works, and that you
have a good time writing it.



Julian Hall <jules@earthcorp.com>