From DShambroom@gte.com Tue Jan 30 18:12:23 2001 Received: from MIT.EDU (SOUTH-STATION-ANNEX.MIT.EDU [18.72.1.2]) by rt-11.mit.edu (8.9.3/8.9.3) with SMTP id SAA24585 for ; Tue, 30 Jan 2001 18:12:22 -0500 (EST) Received: from [132.197.8.26] by MIT.EDU with SMTP id AA13708; Tue, 30 Jan 01 18:10:54 EST Received: from gte.com (coltrane.gte.com [132.197.114.11]) by newman.gte.com (8.9.1/8.9.1) with ESMTP id SAA26826; Tue, 30 Jan 2001 18:12:20 -0500 (EST) Message-Id: <3A7749C2.53EBEC5C@gte.com> Date: Tue, 30 Jan 2001 18:09:54 -0500 From: "W. David Shambroom" Sender: wds1@gte.com To: krb5-bugs@MIT.EDU Subject: bug in kdc/network.c >Number: 919 >Category: krb5-kdc >Synopsis: KDC fails to initialize >Confidential: no >Severity: critical >Priority: high >Responsible: raeburn >State: open >Class: sw-bug >Submitter-Id: unknown >Arrival-Date: Tue Jan 30 18:13:00 EST 2001 >Last-Modified: Thu Feb 01 22:11:00 EST 2001 >Originator: W. David Shambroom >Organization: Verizon Technology >Release: krb5-1.2.1 >Environment: System: AIX uno1 3 4 000327294C00 >Description: krb5kdc fails to initialize, reporting: krb5kdc: no sockets set up? >How-To-Repeat: Run krb5kdc. >Fix: Problem is in line 159 of kdc/network.c. System can return 5 ifreq structures of size 144,32,40,144,32 bytes respectively, corresponding to loopback link level, loopback IPv4, loopback IPv6, ethernet link level, and ethernet IPv4. Last 2 are not retrieved, since (320-144+32+40<32+40) is false. Increasing constant (40) to a larger value (say 128) fixes. Note that some versions of AIX have a bug that causes a buffer overflow if the ioctl() call is made with too small a buffer. A workaround for this is to increase the initial buffer size, e.g., to 1024 B. >Audit-Trail: Responsible-Changed-From-To: krb5-unassigned->raeburn Responsible-Changed-By: raeburn Responsible-Changed-When: Thu Feb 1 21:48:42 2001 Responsible-Changed-Why: Local address lookups are my area... From: Ken Raeburn To: DShambroom@gte.com Cc: krb5-bugs@MIT.EDU Subject: Re: krb5-kdc/919: KDC fails to initialize Date: Thu, 1 Feb 2001 22:06:15 -0500 (EST) Thanks, I'll raise the fudge factor from 40 to 128 or so. Do you have more details on the buffer overflow issue with SIOCGIFCONF? A quick search through Google got me some other interesting info on SIOCGIFCONF issues, but not the AIX bug. Ken >Unformatted: