[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <2249f4d8-b962-a953-70f6-33a5dc4ca5f2@mellanox.com>
Date: Mon, 9 Oct 2017 13:04:27 +0300
From: Tariq Toukan <tariqt@...lanox.com>
To: Florian Westphal <fw@...len.de>, Tariq Toukan <tariqt@...lanox.com>
Cc: Pablo Neira Ayuso <pablo@...filter.org>,
netfilter-devel@...r.kernel.org, Aaron Conole <aconole@...heb.org>,
davem@...emloft.net, netdev@...r.kernel.org
Subject: Re: [PATCH 41/47] netfilter: convert hook list to an array
On 09/10/2017 12:31 PM, Florian Westphal wrote:
> Tariq Toukan <tariqt@...lanox.com> wrote:
>> On 04/09/2017 1:42 AM, Pablo Neira Ayuso wrote:
>>> From: Aaron Conole <aconole@...heb.org>
>>>
>>> This converts the storage and layout of netfilter hook entries from a
>>> linked list to an array. After this commit, hook entries will be
>>> stored adjacent in memory. The next pointer is no longer required.
>>>
>>> The ops pointers are stored at the end of the array as they are only
>>> used in the register/unregister path and in the legacy br_netfilter code.
>>>
>>> nf_unregister_net_hooks() is slower than needed as it just calls
>>> nf_unregister_net_hook in a loop (i.e. at least n synchronize_net()
>>> calls), this will be addressed in followup patch.
>>>
>>> Test setup:
>>> - ixgbe 10gbit
>>> - netperf UDP_STREAM, 64 byte packets
>>> - 5 hooks: (raw + mangle prerouting, mangle+filter input, inet filter):
>>> empty mangle and raw prerouting, mangle and filter input hooks:
>>> 353.9
>>> this patch:
>>> 364.2
>>>
>>> Signed-off-by: Aaron Conole <aconole@...heb.org>
>>> Signed-off-by: Florian Westphal <fw@...len.de>
>>> Signed-off-by: Pablo Neira Ayuso <pablo@...filter.org>
>>> ---
>>
>> Hi,
>>
>> We experience a regression in server with iommu enabled.
>> After installing kernel and rebooting the server, it crashes during boot.
>> Please see trace below.
>>
>> Bisecting points to this patch.
>
> Hmm, strange because
>
>> [ 25.907811] BUG: unable to handle kernel NULL pointer dereference at
>> 000000000000003c
>> [ 25.907828] IP: _raw_read_lock_bh+0x15/0x40
>
> ... this says that ebt_table is NULL (0x3c is the offset of the rwlock).
>
> If you don't have that fix already, does
No, didn't have it.
> https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf.git/commit/?id=e6b72ee88a56bcfe63f72e9c30766484c45bec72
>
> netfilter: ebtables: fix race condition in frame_filter_net_init()
>
> resolve this bug for you?
>
Now I applied the fix and bug is resolved.
Many thanks!
Tariq
Powered by blists - more mailing lists