Path: utzoo!telly!ddsw1!spl1!lll-winken!uunet!tut.cis.ohio-state.edu!ukma!gatech!rebel!didsgn!allan
From: allan@didsgn.UUCP (didsgn)
Newsgroups: gnu.gcc
Subject: Re: gas with COFF output
Summary: I've got a coff gas
Message-ID: <507@didsgn.UUCP>
Date: 12 Jan 89 18:41:33 GMT
References: <462@infovax.lan.informatik.tu-muenchen.dbp.de> <4055@omepd.UUCP>
Organization: Digital Design Inc., Norcross GA
Lines: 46

In article <4055@omepd.UUCP>, chrisb@mipon2.intel.com (Chris Benenati) writes:
> In article <462@infovax.lan.informatik.tu-muenchen.dbp.de> schneck@lan.informatik.tu-muenchen.dbp.de (Bernhard Schneck) writes:
> 
> >Is a version of (or a patch for) gas available which produces COFF output ?
> 
> I'd also be interested in coff output from gas -- and, of course, in a
> version of 'ld' that understands coff.
> 
> And we're probably not the only two interested parties.
> 
> I'd certainly appreciate it if any responses to Bernard's query would
> be posted here rather than just being sent to him.

Well, ok. I've created a module for gas that produces COFF style code. The
module is based upon the "write.c" of the standard gas package. There are
a few changes necessary in one of the include files and a couple of modules
as there are some name conflicts between my systems COFF include files and
the gas (4.3?) files. However, they are not too major.

However, things are not so simple. The system I have is a SYSV system from
Charles River Data Systems. While they are fairly compatible with the AT&T
COFF definition, they differ in a few points. However, these differences
would not be too difficult for someone willing to sort through them. I am
willing to help whoever does the AT&T coff modification to my "write.coff.c"
module, but I cannot test it. I can send a list of differences that CRDS
says exists.

I've been meaning to make up a package and send it to GNU central, so I guess
I will (to gnu central: do you have a UUCP address you wish me to send this?). 

BTW, the version of gas that I have is 1.25.

The COFF output module handles COFF entities such as line numbers, but does
not yet handle all the symbolic data type definition symbols to support
a symbolic debugger, but someday I'll get that part done.

Any questions, please feel free to call.

-Allan G. Schrum

...!gatech!rebel!didsgn!allan

Digital Design, Inc.
3060 Business Park Drive, Ste E
Norcross, GA 30071
(404) 447-0274