|
|||||||||||
|
linux-ipsec: a byteorder bug in spi.c
From: Josef Ungerman <ungerman(at)students.zcu.cz>
Date: Fri Jan 15 1999 - 04:40:05 EST
1.
$Id: spi.c,v 1.16 1998/08/28 03:14:12 rgb Exp $
line 726
case 'w':
/* replay_window = htonl(strtoul(optarg, &endptr, 0)); // flaw */
/*** 'int replay_window' had the NET format now ...*/
replay_window = strtoul(optarg, &endptr, 0); /* added fix*/
if(!(endptr == optarg + strlen(optarg))) {
fprintf(stderr, "%s: Invalid character in replay_window parameter: %s\n",
program, optarg);
exit (1);
Simply, the htonl() conversion is made AFTER the range checks.
other notes:
at line 389 doesn't work (while writing into '/dev/ipsec'): xd->ame_replayp = xd->ame_ooowin = 0; ?
3.
FreeS/WAN runs fine on intels, i have tried out all modes. I have written some overview on IPSec and ISAKMP and a description of caught IPSec packets. All is it on my homepage, all in Czech language ;-) , it will be part of my diploma thesis (VPN, firewall management). Pluto works too, but doesn't want to negotiate keys between two machines on the local eth segment in our lab I didn't succeed in running IPSec on localhost at home (packets are sent but don't came back). The Linux crashed during experiments with an aliasing+ipsec on loopback. looking forward to FreeSWAN-1.0
josef ungerman
This archive was generated by hypermail 2.1.8 : Wed Aug 23 2006 - 12:59:29 EDT |
||||||||||
|
|||||||||||