This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
re: i386-netbsdelf description
- To: Richard Henderson <rth at redhat dot com>
- subject: re: i386-netbsdelf description
- From: matthew green <mrg at cygnus dot com>
- Date: Wed, 18 Jul 2001 02:41:20 +1000
- Cc: "Joseph S. Myers" <jsm28 at cam dot ac dot uk>, gcc-patches at gcc dot gnu dot org
- organisation: Red Hat, Inc.
On Tue, Jul 17, 2001 at 11:23:17PM +1000, matthew green wrote:
>
> On Sun, Jul 15, 2001 at 11:43:38AM +1000, matthew green wrote:
> > #ifdef HAVE_KPRINTF_ATTRIBUTE
> > #define CPP_PREDEFINES "\
> > -Di386 -D__NetBSD__ -D__ELF__ -D__KPRINTF_ATTRIBUTE__ \
> > -Asystem(unix) -Asystem(NetBSD)"
>
> Where is HAVE_KPRINTF_ATTRIBUTE supposed to get defined?
> What does it mean?
>
> it's an extension used by the printf() in the NetBSD kernel. this
> change can be kept local to our tree, or not.
That answers the question of whether it is used, but not how.
You don't have a separate compiler for building the kernel, surely.
Therefore the question: how does HAVE_KPRINTF_ATTRIBUTE get defined?
Besides -- it seems inordinately pointless to do this. Why wouldn't
the kernel sources define __KPRINTF_ATTRIBUTE__? Why would you hard
code this into the compiler?
the code that actually implements this is only in the netbsd source
tree. i'll leave this part out of the patch.
> hmm, so exceptions should just work?
Yes.
> they didn't in egcs 1.1.2.
You were using a.out then, weren't you? Plus that amazingly
broken GAS 1.something.
no, for about 2 years now, the i386 port has been using ELF,
with binutils 2.9.1 & egcs 1.1* or so... hmm.
> however, all these symbols are added by collect2!
Hum... looks like you're not building constructors properly.
Do me and yourself a favour and rearrange things to use elfos.h.
OK.