[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <632d79bf-b87c-b011-35ce-b38d418ba1b3@mellanox.com>
Date: Thu, 18 Oct 2018 08:40:30 +0000
From: Eran Ben Elisha <eranbe@...lanox.com>
To: Cong Wang <xiyou.wangcong@...il.com>
CC: Dave Jones <davej@...emonkey.org.uk>,
Linux Kernel Network Developers <netdev@...r.kernel.org>,
Tariq Toukan <tariqt@...lanox.com>,
Saeed Mahameed <saeedm@...lanox.com>
Subject: Re: bond: take rcu lock in netpoll_send_skb_on_dev
On 10/18/2018 8:46 AM, Cong Wang wrote:
> On Mon, Oct 15, 2018 at 4:36 AM Eran Ben Elisha <eranbe@...lanox.com> wrote:
>> Hi,
>>
>> This suggested fix introduced a regression while using netconsole module
>> with mlx5_core module loaded.
>
> It is already reported here:
> https://marc.info/?l=linux-kernel&m=153917359528669&w=2
>
>
>>
>> During irq handling, we hit a warning that this rcu_read_lock_bh cannot
>> be taken inside an IRQ.
>
> Yes, I mentioned the same even before this patch was sent out:
> https://marc.info/?l=linux-netdev&m=153816136624679&w=2
Thanks Cong.
From the discussion, I understand that the solution shouldn't be
touching netpoll_send_skb_on_dev. Some modules/drivers logging will now
trigger traces while netconsole is loaded.
DaveJ,
Can you please submit a proper fix or at least revert the current one in
the meanwhile.
Thanks,
Eran
>
> Thanks.
>
Powered by blists - more mailing lists