From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stephen Hemminger Subject: Fw: [Bug 107551] New: IPV4 Reassembling Problem Date: Mon, 9 Nov 2015 10:40:01 -0800 Message-ID: <20151109104001.09d86030@xeon-e3> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit To: netdev@vger.kernel.org Return-path: Received: from mail-pa0-f45.google.com ([209.85.220.45]:33585 "EHLO mail-pa0-f45.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751691AbbKISjv (ORCPT ); Mon, 9 Nov 2015 13:39:51 -0500 Received: by pabfh17 with SMTP id fh17so207070821pab.0 for ; Mon, 09 Nov 2015 10:39:50 -0800 (PST) Received: from xeon-e3 (static-50-53-82-155.bvtn.or.frontiernet.net. [50.53.82.155]) by smtp.gmail.com with ESMTPSA id w11sm10287548pbs.87.2015.11.09.10.39.50 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 09 Nov 2015 10:39:50 -0800 (PST) Sender: netdev-owner@vger.kernel.org List-ID: Begin forwarded message: Date: Mon, 9 Nov 2015 11:04:29 +0000 From: "bugzilla-daemon@bugzilla.kernel.org" To: "shemminger@linux-foundation.org" Subject: [Bug 107551] New: IPV4 Reassembling Problem https://bugzilla.kernel.org/show_bug.cgi?id=107551 Bug ID: 107551 Summary: IPV4 Reassembling Problem Product: Networking Version: 2.5 Kernel Version: 4.2.3 Hardware: Other OS: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: IPV4 Assignee: shemminger@linux-foundation.org Reporter: bernd.weiberg@siemens.com Regression: No I have an ipv4 reassembling issue using the latest Kernel (4.2.3) in an nios2 environment. I first noticed that issue when sending UDP-Pakets from an Host PC to my device. Each packet has a payload of 4096 Byte. Due to the MTU of 1500 Byte each packet will be split into three ip-packets. Approximately 1 of 200 Packets were not correctly reassembled by the Kernel, so they cannot be received within user-space. The command nstat says, that there are reassembling errors due to timeout, but at least in wireshark all packets are correctly displayed. Sometime the kernel issues a icmp-messages, saying that there was a timeout during ip reassembling. I fir9st thought that it might be a performance problem, so I increased the time between the packets to about one packet per 500 ms but without success. Then I decided to reduce the payload to 1024 Byte, because it might be a buffer (or TSE FIFO) problem. Now all packets are send into one ipv4 packet and there is nothing lost on the receiver side, but when I reduced the MTU of the host from 1500 to 800 byte (each datagram will be split into two packets now) the problem accurse again! So this might be a reassembling issue. -- You are receiving this mail because: You are the assignee for the bug.