netdev - Re: latest netdev net-next kernel panic

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

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <8ae3175c-272c-decf-3b6e-014d67638b23@itcare.pl>
Date:   Fri, 8 Sep 2017 00:38:50 +0200
From:   Paweł Staszewski <pstaszewski@...are.pl>
To:     Eric Dumazet <eric.dumazet@...il.com>
Cc:     Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: latest netdev net-next kernel panic

What is weird - changed in bios and enabled p-states and no panic from 
30 minutes

w
  00:38:07 up 33 min,  2 users,  load average: 0.10, 0.13, 0.17
USER     TTY        LOGIN@   IDLE   JCPU   PCPU WHAT

So the problem can be related to some cpu weird stuff with c/p states or 
it is random.



W dniu 2017-09-08 o 00:08, Paweł Staszewski pisze:
> Also this panic occured about 3 minutes after traffic rise to 40G
>
> before host was working without trafic for about 30minutes - after 
> traffic start to flow this bug start.
>
> Now i have this host without traffic and it is working for almost 1 
> hour :)
>
>
>
> W dniu 2017-09-07 o 23:47, Paweł Staszewski pisze:
>> Hi Eric
>>
>>
>> today upgraded some host from 4.13.0-rc7+ to latest net-next from git
>>
>> and run some tests simulate clients traffic 40Gbit / 5Mpps mixed tcp 
>> + udp
>>
>> There is also bgpd running that feeds FIB with full bgp feed - about 
>> 700k prefixes (so routing tables is about 700k routes with about 300 
>> nexthops)
>>
>>
>> There is also simulated about 1000 ip neigh hosts (arp) with about 
>> 300 ip interfaces assigned on vlans
>>
>>
>>
>> With kernel 4.13.0-rc7+ (git from 05.09.2017) all tests are good 
>> (besides that there is panic after about 10 hours - every time same 
>> like in this link:
>>
>> https://bugzilla.kernel.org/attachment.cgi?id=258257
>>
>> But looks like acpi related or c/p states with new intel cpus... dono 
>> now.
>>
>>
>>
>>
>> W dniu 2017-09-07 o 23:39, Eric Dumazet pisze:
>>> On Thu, 2017-09-07 at 23:30 +0200, Paweł Staszewski wrote:
>>>> I want to mention also here cause last trace for RIP was dst_dev_put
>>>>
>>>> https://bugzilla.kernel.org/attachment.cgi?id=258259
>>> Can you provide a bit more context ?
>>>
>>> Thanks !
>>>
>>>
>>>
>>
>>
>
>