🚀 go-pugleaf

I've just installed gcc-2.3.1.  I'm having difficulties on two fronts:

	1. The generated code does not seem to be able to be debugged with dbx.

	2. Assigning to a double that is a member of a dynamically allocated
       structure is problematic.  Doubles seem to have different alignment
       requirements than with the compiler shipped with the OS.

The particulars:

I'm running on a Sun sparc with SunOS 4.1.1.  The "other" compiler is the
Sun C compiler shipped with the system (We got these machines prior to
unbundling).  Gcc is configured as

	#!/bin/sh
	# GCC was configured as follows:
	./configure --target=sparc-sun-sunos4.1
	echo host=sparc-sun-sunos4.1 target=sparc-sun-sunos4.1 build=sparc-sun-sunos4.1

With -g my unserstanding is gcc generates dbx debugging info with gdb
extensions.  Debugging with gdb works fine.  Dbx crashes.  I recompiled and
replaced -g with -gdbx.  The compiler complains that -gdbx is not supported.
A quick perusal through the compiler source reveals that -gstabs enables dbx
debugging output without the gdb extensions.  I recompile and try again.  No
success.  My program has a section that is dynamically linked (i.e. compiled
with -fpic).  Could this have anything to do with the problem?

The problem with doubles is a little more subtle.  Assigning to a double
that is the member of a structure causes a bus error.

I'm using an in-house memory allocation package that piggybacks off of
malloc and provides memory analysis and debugging capabilities.  This
packages has been in use for several years in a number of projects.  It's
very stable and I have good confidence in it.

The memory package guarantees that all memory blocks returned are aligned on
4 byte boundaries.  Everything works fine with Sun C.  With GNUC bus errors
ensue unless I change the memory package to align to 8 byte boundaries.
What's the problem here?  Is Sun C "smarter" about padding the strucutre
properly so that sizeof does the right thing and the proper alignments are
enforced?  Maybe the real question is: why did it operate proplery under Sun
C?  Since doubles require 8 byte alignments it shouldn't have worked.

Are there any known bugs relating to these issues?  Am I doing something
wrong?  

Thank you in advance for any light you can shed on these problems


Mark Glasser
------------------------------------------------------------------------
markg@nextwave.com                            Nextwave Design Automation
                                                450 San Antonio Road #27
                                                     Palo Alto, CA 94306
                                                    (voice) 415.855.9795
                                                      (fax) 415.855.9792
------------------------------------------------------------------------