Date: Thu, 21 Mar 1996 10:59:02 -0800 From: John Polstra <jdp@polstra.com> To: freebsd-hackers@freebsd.org Subject: "nm -og": binutils vs. FreeBSD vs. SunOS vs. Posix Message-ID: <199603211859.KAA16365@austin.polstra.com>
next in thread | raw e-mail | index | archive | help
I need some advice about whether something in GNU binutils-2.6 is a bug
or not. It has to do with the output of "nm -og *.o".
The "nm" from binutils-2.6 offers three different output formats, "bsd",
"sysv", and "posix". Both the "bsd" and "posix" formats produce output
like this:
bar.o:00000004 T _barproc
bar.o: U _printf
foo.o:00000004 T _fooproc
foo.o: U _printf
But FreeBSD's "nm" produces output like this:
bar.o:
bar.o:00000004 T _barproc
bar.o: U _printf
foo.o:
foo.o:00000004 T _fooproc
foo.o: U _printf
The blank lines don't matter, but the "bar.o:" lines are crucial -- our
"lorder" relies on them. I checked the bsd4.4-lite2 sources, and their
"lorder" is identical to ours, so I assume that their "nm" has the same
behavior as ours.
So, I would say that GNU's "nm" is not giving correct "bsd" output.
Except ... I also tried it on a SunOS-4.1 system, and their "nm" agrees
with GNU's. So I don't know whether to report a bug to GNU or not.
Two questions: What do other BSD-derived systems do? And, what does
Posix say about it, if anything?
If SunOS is the odd man out, then I could create a fourth format
"sunos", fix the "bsd" behavior, and probably get GNU to accept the
changes.
Opinions, please.
--
John Polstra jdp@polstra.com
John D. Polstra & Co., Inc. Seattle, Washington USA
"Self-knowledge is always bad news." -- John Barth
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199603211859.KAA16365>