[11268] in Athena Bugs
Re: inetd
daemon@ATHENA.MIT.EDU (jweiss@MIT.EDU)
Sun Oct 24 17:07:21 1993
From: jweiss@MIT.EDU
Date: Sun, 24 Oct 1993 17:07:06 -0400
To: bugs@MIT.EDU, bug-dialup@MIT.EDU
Cc:
In-Reply-To: [10715]
I'm re-submitting this, because we're still getting bitten by it, and
it isn't in netprob. I believe there is a version of inetd that is
running on the PO's which *may* have this bug fixed.
Jonathon
[10715] daemon@ATHENA.MIT.EDU (John Carr) Athena Bugs 07/29/93 10:20 (21 lines)
Date: Thu, 29 Jul 1993 10:20:17 EDT
From: John Carr <jfc@Athena.MIT.EDU>
On a DECstation running 7.4 or 7.6, if an inetd service is unswitched and
inetd shuts down another service due to too many requests, then the next
request for the first service will cause inetd to dump core. This appears
to be caused by use of a negative argument to one of the FD_* select macros.
inetd does not crash if the service is switched and enabled with access_on.
This is the cause of the recent inetd crash on primavera.
A bug was fixed in the athena inetd that is likely related:
revision 1.21
date: 93/02/13 00:08:30; author: probe; state: Exp; lines added/del: 3/1
If the fd is not active (-1), don't bother with FD_ISSET, since that may
reference a negative array element of the fd_set structure... (oops).
--[10715]--
