Report information
The Basics
Id:
41757
Status:
resolved
Estimated:
8 hours (480 minutes)
Worked:
6 hours (360 minutes)
Users:
marcin: 6 hours (360 minutes)
Priority:
Low/Low
Queue:

BugTracker
Version Fixed:
4.4.0 4.3.6 4.1-ESV-R15
Version Found:
(no value)
Versions Affected:
(no value)
Versions Planned:
4.4.0 4.3.6 4.1-ESV-R15
Priority:
P2 Normal
Severity:
S2 Normal
CVSS Score:
(no value)
CVE ID:
(no value)
Component:
DHCP Common
Area:
bug

Dates
Created:Fri, 19 Feb 2016 13:04:21 -0500
Updated:Wed, 28 Jun 2017 14:09:42 -0400
Closed:Wed, 28 Jun 2017 13:24:11 -0400



This bug tracker is no longer active.

Please go to our Gitlab to submit issues (both feature requests and bug reports) for active projects maintained by Internet Systems Consortium (ISC).

Due to security and confidentiality requirements, full access is limited to the primary maintainers.

Subject: "bad udp checksums in packets" in isc-dhcpd-4.3.3-P1 @ FreeBSD
Date: Fri, 19 Feb 2016 19:04:09 +0100
To: dhcp-bugs@isc.org
From: "Mark Martinec" <Mark.Martinec@ijs.si>
Using isc-dhcpd-4.3.3-P1 from FreeBSD ports, on an amd64 host (Intel Xeon), running FreeBSD 10.3-BETA2. The ethernet interface driver is igb, GbE NIC is Intel PRO/1000 Network Connection version - 2.4.0. The dhcpd server (in IPv4 mode) is running on a physical machine, there is *no* virtualization involved, nor a jail or VIMAGE a host firewall. While it seems that most clients successfully communicate with a dhcpd server, I can see a steady trickle of log entries like: Feb 19 14:26:06 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:06 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:07 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:08 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:08 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:14 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:14 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:14 dhcp2 dhcpd: 4 bad udp checksums in 5 packets Feb 19 14:26:14 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:18 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:19 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:34 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 4 bad udp checksums in 7 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:36 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:39 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:40 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Feb 19 14:26:40 dhcp2 dhcpd: 3 bad udp checksums in 5 packets Other services on this host have no problem with communication, netstat -I igb0 does not show any I/O errors on the interface. Note that this dhcpd server sees both the proxied as well as link-local client requests. Disabling checksum offloading on the NIC avoids the problem! (ifconfig igb0 -rxcsum). I can see that a recent release of isc-dhcpd-4.3 received a bunch of workarounds for just this kind of a problem, although these all concentrated on Linux or on virtualizes network interfaces. Seems the same problem still persists on a FreeBSD platform. Mark
Hello Mark: We have done some digging into this. What you are actually seeing is the server detecting "bad" checksums on its own outbound packets. Under FreeBSD, our code uses the Berkely Packet Filter to read packets, hence we end up seeing our own outbound messages. With CSUM offloading, virtually all outbound packets will not have CSUMs at the point the filter sees them. While the log messages are probably annoying, there are no communication issues or packets being lost. We are close to our next releases, 4.3.4 and 4.1-ESV-R13, due out in March 2016. Given their proximity, we'll have to consider any potential solutions for our next release, 4.4.0. If you have any further thoughts on the matter, please let us know. Thank you for taking the time to report the issue and for your interest in our software. Sincerely, Thomas Markwalder ISC Software Engineering
Hello Mark, We have addressed the issue of excessive 'bad checksum' logging by lowering the logging level for this message to 'debug' from 'info'. This fix will be available in our next releases of ISC DHCP: 4.3.6 (July 2017), 4.1-ESV-R15 (July 2017) and 4.4.0 (around Q4'17). We like to give our contributors credit for their submissions by citing them in the release notes. If you would like to be recognized in this manner, please respond with how you wish to be identified. Typically it is by name and/or organization. Thank you for interest and for taking the time and effort to report the issue. We welcome any future submissions you choose to make. Marcin Siodelski ISC Engineering