lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Message-ID: <6.2.0.14.0.20050719180915.04374ea8@pop.frh.utn.edu.ar> Date: Tue Jul 19 23:10:33 2005 From: fernando at frh.utn.edu.ar (Fernando Gont) Subject: (ICMP attacks against TCP) (was Re: HPSBUX01137 SSRT5954 rev.4 - HP-UX TCP/IP Remote Denial of Service (DoS)) At 08:35 a.m. 19/07/2005, Security Alert wrote: > Discussion of ip_pmtu_strategy > ---------------------------------- > >The default value for ip_pmtu_strategy is 1. This allows for PMTU >discovery. Once the issue of this Security Bulletin has been >resolved via patches the ip_pmtu_strategy value of 1 will again be >the preferred setting for most situations. > >The ip_pmtu_strategy values of 0 and 3 set the PMTU to a fixed >size for destinations which are not on the local network. > >The ip_pmtu_strategy value of 0 sets the PMTU to 576 bytes. >Routers are required to handle packets of at least this size. > >The ip_pmtu_strategy value of 3 sets the PMTU to 1500 bytes. This >will generally result in more efficient transmission than the 576 >byte PMTU. If it is known that the routers involved can handle a >1500 byte MTU the ip_pmtu_strategy value of 3 is preferred. These assumptions are completely wrong. Please read http://www.gont.com.ar/drafts/icmp-attacks-against-tcp.html The IPv4 minimum MTU is 68, and not 576. If you blindly send packets larger than 68 with the DF bit set, in the case there's an intermmediate with an MTU lower that 576, the connection will stall. 576 is the minimum reassembly buffer size. That is the minimum packet size every *end-system* should be able to reassemble, and NOT the minimum packet size that can get to destination without fragmentation. Kindest regards, -- Fernando Gont e-mail: fernando@...t.com.ar || fgont@....org
Powered by blists - more mailing lists