From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ian Lance Taylor <ian@cygnus.com>
To: manfred@s.direktnet.de
Cc: bfd@cygnus.com, gas2@cygnus.com, Manfred.Hollstein@ks.sel.alcatel.de
Subject: Re: Please send in binutils patches
Date: Thu, 12 Feb 1998 09:02:00 -0000
Message-id: <199802121702.MAA01772@subrogation.cygnus.com>
References: <9802120754.AA20936@lts.sel.alcatel.de>
X-SW-Source: 1998/msg00011.html

   Date: 	Thu, 12 Feb 98 08:54:10 +0100
   From: Manfred Hollstein <manfred@s.direktnet.de>

   But on    m88k-motorola-sysv3  I  wasn't  able  to    compile the  new
   `gas/ehopt.c' without my appended patch. The next machine in the queue
   then is the slowest one: m68k-motorola-sysv; but as  it uses COFF like
   the m88k I don't expect any more problems.

Thanks--I have already checked in a different patch.

The ehopt.c file is new.  It is intended to optimize the exception
frame information now generated by gcc 2.8 and egcs.  If there are any
g++ users on these lists, it might be interesting to try it and see
whether it makes a significant different in the size of your .o files
and your final executables, and, more importantly, whether it works
correctly.  The exception frame optimization should happen
automatically, although you can disable it with the
--traditional-format assembler option.

Ian

Index: ehopt.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gas/ehopt.c,v
retrieving revision 1.3
retrieving revision 1.4
diff -u -r1.3 -r1.4
--- ehopt.c	1998/02/06 23:14:42	1.3
+++ ehopt.c	1998/02/10 22:59:36	1.4
@@ -109,7 +109,11 @@
   current_subseg = now_subseg;
   subseg_new (".eh_frame", 0);
   f = seg_info (now_seg)->frchainP->frch_root;
+#ifdef BFD_ASSEMBLER
   fix = seg_info (now_seg)->frchainP->fix_root;
+#else
+  fix = *seg_fix_rootP;
+#endif
   subseg_set (current_seg, current_subseg);
 
   /* Look through the frags of the section to find the code alignment.  */