From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [regression] 2.6.37+ commit 0363466866d9.... breaks tcp ipv6 Date: Tue, 18 Jan 2011 23:03:58 +0100 Message-ID: <1295388238.8449.10.camel@edumazet-laptop> References: <4D335417.80704@xmsnet.nl> <4D35EF64.1040906@xmsnet.nl> <4D36092B.1090406@xmsnet.nl> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Jesse Gross , LKML , netdev To: Hans de Bruin Return-path: In-Reply-To: <4D36092B.1090406@xmsnet.nl> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Le mardi 18 janvier 2011 =C3=A0 22:42 +0100, Hans de Bruin a =C3=A9crit= : > On 01/18/2011 09:06 PM, Jesse Gross wrote: > > On Tue, Jan 18, 2011 at 11:52 AM, Hans de Bruin wrote: > >> On 01/16/2011 09:24 PM, Hans de Bruin wrote: > >>> > >>> After last nights compile i lost the possibility to connect to ss= h and > >>> http over ipv6. The connection stops at syn_sent. connections to = my > >>> machine end in syn_recv. ping6 still works. > >>> > >> > >> The bisect ended in: > >> > >> 0363466866d901fbc658f4e63dd61e7cc93dd0af is the first bad commit > >> commit 0363466866d901fbc658f4e63dd61e7cc93dd0af > >> Author: Jesse Gross > >> Date: Sun Jan 9 06:23:35 2011 +0000 > >> > >> net offloading: Convert checksums to use centrally computed fe= atures. > >> > >> In order to compute the features for other offloads (primarily > >> scatter/gather), we need to first check the ability of the NIC= to > >> offload the checksum for the packet. Since we have already co= mputed > >> this, we can directly use the result instead of figuring it ou= t > >> again. > >> > >> Signed-off-by: Jesse Gross > >> Signed-off-by: David S. Miller > >> > >> > >> ssh ::1 still works. And since dns still works I guess udp is not= affected. > >> My nic is a: > >> > >> 09:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM575= 2 Gigabit > >> Ethernet PCI Express (rev 02) > > > > Are you using vlans? If so, can you please test this patch? > > http://patchwork.ozlabs.org/patch/79264/ > > >=20 > No I am not using vlans. The option is even not set in the config fil= e.=20 > Except for the disk less bit is a straightforward setup: >=20 > bash-4.1# ip addr show > 1: lo: mtu 16436 qdisc noqueue state UNKNOWN > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 > inet 127.0.0.1/8 scope host lo > inet6 ::1/128 scope host > valid_lft forever preferred_lft forever > 2: eth0: mtu 1500 qdisc mq state UP= =20 > qlen 1000 > link/ether 00:1c:23:2d:73:87 brd ff:ff:ff:ff:ff:ff > inet 10.10.0.6/16 brd 10.10.255.255 scope global eth0 > inet6 2001:610:76e:0:21c:23ff:fe2d:7387/64 scope global dynamic > valid_lft 86399sec preferred_lft 14399sec > inet6 fe80::21c:23ff:fe2d:7387/64 scope link > valid_lft forever preferred_lft forever > 3: sit0: mtu 1480 qdisc noop state DOWN > link/sit 0.0.0.0 brd 0.0.0.0 > bash-4.1# ip route show > 10.10.0.0/16 dev eth0 proto kernel scope link src 10.10.0.6 > default via 10.10.0.1 dev eth0 > bash-4.1# ip -6 route show > 2001:610:76e::/64 dev eth0 proto kernel metric 256 expires 86404se= c > fe80::/64 dev eth0 proto kernel metric 256 > ff00::/8 dev eth0 metric 256 > default via fe80::230:18ff:feae:75d8 dev eth0 proto kernel metric 1= 024=20 > expires 29sec hoplimit 64 > bash-4.1# iptables -L > Chain INPUT (policy ACCEPT) > target prot opt source destination >=20 > Chain FORWARD (policy ACCEPT) > target prot opt source destination >=20 > Chain OUTPUT (policy ACCEPT) > target prot opt source destination > bash-4.1# ip6tables -L > Chain INPUT (policy ACCEPT) > target prot opt source destination >=20 > Chain FORWARD (policy ACCEPT) > target prot opt source destination >=20 > Chain OUTPUT (policy ACCEPT) > target prot opt source destination > bash-4.1# brctl show > bridge name bridge id STP enabled interfaces > bash-4.1# cat /proc/cmdline > root=3D/dev/nfs nfsroot=3D10.10.0.2:/nfs/root/psion/,v3,tcp ro ip=3D:= :: ::dhcp=20 > BOOT_IMAGE=3Dnightlybuild-psion >=20 You could try "tcpdump -i eth0 ip6 -v" I guess you receive frames with bad checksums You can ask other system to not offload tx checksums ethtool -K eth0 tx off Please give result of (on both machines) lspci -v