This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: dwarf2 eh for alpha, mark 2
- To: rth at cygnus dot com
- Subject: Re: dwarf2 eh for alpha, mark 2
- From: Jim Wilson <wilson at cygnus dot com>
- Date: Wed, 9 Sep 1998 09:42:21 -0700
- Cc: egcs-patches at cygnus dot com, g++-local at cygnus dot com
- Newsgroups: cygnus.local.g++
- Organization: Cygnus Solutions, CA
In article <19980909012313.C18516.cygnus.local.g++@dot.cygnus.com> you write:
>
>--G4iJoqBmSsgzjUCe
>Content-Type: text/plain; charset=us-ascii
>
>I see that Irix defaults to sjlj eh, but not other MIPSen, so I must presume
>it does.
Irix is ambiguous. Irix6 was in fact the first target to use dwarf2
eh, so the code certainly does work under Irix, and has always worked under
Irix.
SGI's Irix5 assembler is really an ECOFF assembler pretending to be an ELF
assembler, and hence is unable to support DWARF2, so this was disabled for
irix5 unless you configure with GNU as. In practice, everyone uses GNU as
because -g won't work otherwise, and hence everyone using Irix5 is using
the dwarf2 eh code.
I see now however that dwarf2 eh was accidentally disabled for irix6 when
it was disabled for irix5. Argh. This happened in February, and I am
even the bozo who broke it. I checked in the following patch to fix it.
Wed Sep 9 09:36:51 1998 Jim Wilson <wilson@cygnus.com>
* iris6.h (DWARF2_UNWIND_INFO): Undef.
Index: iris6.h
===================================================================
RCS file: /cvs/cvsfiles/devo/gcc/config/mips/iris6.h,v
retrieving revision 1.38
diff -p -r1.38 iris6.h
*** iris6.h 1998/08/19 22:26:28 1.38
--- iris6.h 1998/09/09 16:36:03
*************** Boston, MA 02111-1307, USA. */
*** 30,35 ****
--- 30,39 ----
#include "mips/iris5.h"
#include "mips/abi64.h"
+ /* Irix6 assembler does handle DWARF2 directives. Override setting in
+ irix5.h file. */
+ #undef DWARF2_UNWIND_INFO
+
/* For Irix 6, -mabi=64 implies TARGET_LONG64. */
/* This is handled in override_options. */