2.0.6.beta.14 argus startup problems

[Carter Bullard]

Hey Andreas,
   You are absolutely correct, but what I was trying to do
with the first getopt() was to process a subset of the
commandline arguments, no matter where they were in the
commandline, and then after that, process them again in
another pass.  Putting the '+' doesn't solve the problem,
as it can stop in the middle of the processing.

   I just removed the first getopt() in my fix and processed
the string myself.  Total avoidance fix ;)

Thanks for the bug hunt.  I'm interested in the fact that
it was on your RH9 machine.  I've been running it on RH9
for a while, and didn't notice any problems, but I almost
always get the options from /etc/argus.conf

Carter



-----Original Message-----
From: owner-argus-info at lists.andrew.cmu.edu
[mailto:owner-argus-info at lists.andrew.cmu.edu] On Behalf Of Andreas Östling
Sent: Friday, December 05, 2003 4:29 AM
To: argus-info at lists.andrew.cmu.edu
Subject: Re: 2.0.6.beta.14 argus startup problems



I was looking into the getopt stuff some more and a simple printf statement
showed that the first call to getopt messed things up (only on RH9).
Here is what it lookes like right after the second getopt call:

# bin/argus_linux  -i eth1 -w /tmp/foo -i eth2
optind=3, opterr=0, optarg=-w, op=i
optind=5, opterr=0, optarg=eth1, op=i
argus_linux[9578]: ArgusInitSource: pcap_open_live ioctl: No such device

When commenting out the first getopt block, the result is instead:
# bin/argus_linux  -i eth1 -w /tmp/foo -i eth2
optind=3, opterr=0, optarg=eth1, op=i
optind=5, opterr=0, optarg=/tmp/foo, op=w
optind=7, opterr=0, optarg=eth2, op=i

The getopt man page says "If the first character of optstring is  `+'  or
the
environment  variable  POSIXLY_CORRECT  is  set, then option processing
stops as soon as a non-option argument is encountered.", which acually seems

to do the trick.

This gigantic diff (against 2.0.6 rc) fixes it on my RH9 at least:

--- argus.c.org 2003-12-05 10:08:17.000000000 +0100
+++ argus.c     2003-12-05 10:08:25.000000000 +0100
@@ -231,7 +231,7 @@
    setArgusMarReportInterval (ARGUS_MARSTATUSTIMER);

    opterr = 0; optind = 1;
-   while ((op = getopt (argc, argv, "F:Xh")) != EOF) {
+   while ((op = getopt (argc, argv, "+F:Xh")) != EOF) {
       switch (op) {
          case 'F':
             ArgusParseResourceFile (optarg);

I don't know if this is the best way of fixing it or if it has other
implications on other OSes, but maybe it's a start.

/Andreas


On Friday 05 December 2003 06.26, Peter Van Epp wrote:
> It probably got introduced back about beta.12 or so. There was a
> problem with the BSDs and getopt (that didn't occur on linux at least not
> on RedHat 7.3) that Carter fixed. I don't remember if I tried the fix on
> RedHat 9 at that point, and I'm pretty sure Carter is still running 7.3.
It
> may be profitable to try the 2.0.6 release candidate (although I have a
few
> BSD fixes for it already, none of them are around the getopt stuff).