Date: Fri, 03 Aug 2001 20:33:09 +0300
From: "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender: halo1 AT zahav DOT net DOT il
To: "Mark E." <snowball3 AT bigfoot DOT com>
Message-Id: <6137-Fri03Aug2001203307+0300-eliz@is.elta.co.il>
X-Mailer: Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.9
CC: djgpp-workers AT delorie DOT com
In-reply-to: <3B6A9705.32294.192A26@localhost> (snowball3@bigfoot.com)
Subject: Re: Test binaries of gcc-3.0.1 20010802 (prerelease)
References: <3B6AD5E9 DOT 23174 DOT 98E16A AT localhost> (pavenis AT lanet DOT lv) <3B6A9705 DOT 32294 DOT 192A26 AT localhost>
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> From: "Mark E." <snowball3 AT bigfoot DOT com>
> Date: Fri, 3 Aug 2001 12:20:21 -0400
> 
> Wasn't the register-naming scheme for all our debug formats broken until 
> recently? If so, couldn't we use the dwarf2 scheme (the svr4_* array in 
> gcc) for all debug formats.

What was broken was the way GDB interpreted the register numbers
recorded in the debug info.  It was broken in _all_ x86 ports of GDB,
because GDB would simply use the same numbers as in its internal
numbering scheme.  These two schemes are similar, but not identical.

This is now fixed in GDB's CVS (and the fixed mapping will be in the
upcoming release of GDB 5.1).  The way it is fixed assumes the
dbx_register_map[n] mapping for COFF and stabs, and
svr4_dbx_register_map[n] for DWARF2.  It is possible for the DJGPP
port to override this default mapping in its own configuration files,
but I don't see why we should use a map that is different from the
vast majority of other x86 ports.