Hello, I don't see any difference what we use here because we used -1 in ip_vs_fill_iph_skb. If not validated once, with -1 we can get a NEXTHDR_NONE result when ipv6_find_hdr returns without error.
Hello I'll do that. We will save some cycles here, very few actually.. I'm not sure about the mobility header if it can break this. Have not read the RFC :-) The -1 is OK for me right now Regards Han
Hello, Thanks for the confirmation. Then may be Hans can post a fix for this problem after checking the callers of ipv6_find_hdr. Yes, it is not expected, the protocol was already validated. May be w
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Thu, 20 Feb 2014 19:56:19 +0100
Hi, I ack the working of that change for my specific case: passing pmtud's correctly: Feb 20 18:58:59 c43236 kernel: [ 721.473388] IPVS: Enter: ip_vs_icmp_xmit_v6, net/netfilter/ipvs/ip_vs_xmit.c lin
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Thu, 20 Feb 2014 14:10:44 +0100
Hi, It will be on my next desktop reboot test round :-). I think I will test 2 hours before EOD, that's around 18:00+0100 -- To unsubscribe from this list: send the line "unsubscribe lvs-devel" in th
Can you try Julians patch insted ? mine was not so good :-) i.e. - if (target < 0) + if (target < 0 || found) break; Regards Hans Attachment: smime.p7s Description: S/MIME cryptographic signature
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Thu, 20 Feb 2014 09:51:55 +0100
Hi, Ah, f**k me :-(. I was so caught up in reloading ipvs that I forgot to reload wherever that function is in, but I guess that's static :-(. So tests like that will take me some longer to perform.
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Thu, 20 Feb 2014 09:40:09 +0100
Hi, Yes :-). I figured that after I added debugging to pinpoint it ;-). I always saw that message in the log but I thought that's some new warning about other devices bugs. The message as it is right
Hi Julian As usual I'm to quick to send the mail... It will work for hmark and it looks like it will work for others, with -1 Maybe Patrick have another opinion... Regards Hans Attachment: smime.p7s
Hi Ansis & Patrick I saw that too in openvswitch/actions.c, i.e it will break your patch But if you want to find a specific header ex. NEXTHDR_GRE, that is not in ipv6_ext_hdr() ipv6_find_hdr() will
Hello, This is against the goal of the above commit. May be above check should be: if (target < 0 || found) break; We have to check some callers with -1, may be some need check for NEXTHDR_NONE, for
Hello, You mean "IPv6 header not found" ? I guess it happens for the second protocol = ipv6_find_hdr(skb, &offs, -1, &fragoffs, NULL); when icmp_offset was 0. Message is from ipv6_find_hdr. Regards -
My changes to this function were necessary for the Open vSwitch set_ipv6() action implementation so that checksums would be correctly recalculated. I introduced IP6_FH_F_SKIP_RH flag that skips all R
Hello This one bothers me, Why does it returns -ENOENT ??? When enter ipv6_find_hdr() the initial nexthdr should be 0x3a ICMPv6 u8 nexthdr = ipv6_hdr(skb)->nexthdr; and target also 0x3a i.e. found is
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Wed, 19 Feb 2014 19:21:04 +0100
Hans, I want to keep the patch as is, but change the description: Fix regression introduced in 3.8 with commit 9195bb8e381d81 ("ipv6: improve ipv6_find_hdr() to skip empty routing headers") which bro
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Wed, 19 Feb 2014 19:02:32 +0100
Hi, Found it: if ((!ipv6_ext_hdr(nexthdr)) || nexthdr == NEXTHDR_NONE) So, it keeps traversing until no more ext headers are found (i.e. the protocol header), or no headers at all. Then it tests if y
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Wed, 19 Feb 2014 18:53:36 +0100
If I look at the ipv6_find_hdr code, it will actually loop until all headers have been visited, because no header will equal -1. Eventually it will stop because there are no next headers anymore. So
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Wed, 19 Feb 2014 18:05:05 +0100
Hi Hans, Ah bah, I thought they move every extra header after the payload. I will take a hard look into that one :-). If I patch my kernel: ard@freeze8dev:/mnt/source/kernels/build-hp-ws/l-3.13.3$ di
Hi Ard and Ansis The problem is if icmp6 is not the first header it will not work... i.e. it can be other headers before icmp and if you have -1 you will not always get the icmp header. The patch tha
Author: Art -kwaak- van Breemen <ard@xxxxxxxxxxxxxxx>
Date: Wed, 19 Feb 2014 11:27:52 +0100
Thanks. I was a reading it, but something with impatience and lazy, and worst of all, I still don't know an easy way to blame without cloning the repository :-( <snipped proof of lazy incompetence> ^