From owner-freebsd-hackers  Tue Jan 31 09:42:55 1995
Return-Path: hackers-owner
Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id JAA22032 for hackers-outgoing; Tue, 31 Jan 1995 09:42:55 -0800
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.34]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id JAA22026 for <freebsd-hackers@freefall.cdrom.com>; Tue, 31 Jan 1995 09:42:46 -0800
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id EAA27602; Wed, 1 Feb 1995 04:39:30 +1100
Date: Wed, 1 Feb 1995 04:39:30 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199501311739.EAA27602@godzilla.zeta.org.au>
To: freebsd-hackers@freefall.cdrom.com, kargl@troutmask.apl.washington.edu
Subject: Re: tar dumps core on 950112-SNAP
Sender: hackers-owner@FreeBSD.org
Precedence: bulk

>Simple question: should tar dump core if it tries to
>access /proc where /proc is, of course, mounted as a
>procfs?

No.  I think /proc is still buggy.  Regular files in it
are irregular.  Some are irregular enough to confuse tar.

BTW, for `cd /proc; wc */*', wc gives up immediately
after not being able to read the first file (0/ctl).

>I am going to start to sup the -current tree, so I 
>wanted to backup my entire system.  I cd to / and
>simply issued `tar cvf /dev/rst0 .i >& tar.log &'.  The system
>barfed on /proc. If /proc is unmounted the tar executes
>as expected.  The following error message is printed
>to the console.

You have to skip /proc somehow.  I use a regexp to weed out
junk in "/" (most mount points, but not /usr).

>One minor problem: Some programs include the following lines:

>#include <stdio.h>
>int foo() {
>   extern char *sys_errlist[];
>....
>}

>FreeBSD-2.[01].x declares

>extern __const char *__const sys_errlist[];
                      ^^it doesn't have this bit probably should

>in stdio.h.  gcc complains about the redeclaration, then halts
>compilation. I modified the sources to read

>#include <stdio.h>
>#include <sys/param.h>
>int foo() {
>#if BSD < 199303
>   extern char *sys_errlist[];
>#endif
>....
>}

>Is this the proper way to handle the problem?

Not really.  It works for one system, but for 100 systems that have
sys_errlist it generalizes to a 100-line ifdef, and then there are
systems that don't have sys_errlist but have strerror()...

Bruce