netdev - Re: [PATCH net] be2net: Fix NULL pointer dereference in be_tx

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

Message-ID: <d2aebf72-d3d7-7d8e-d570-c4a487a50623@redhat.com>
Date:   Wed, 28 Nov 2018 20:29:44 +0100
From:   Ivan Vecera <ivecera@...hat.com>
To:     David Miller <davem@...emloft.net>
Cc:     poros@...hat.com, netdev@...r.kernel.org
Subject: Re: [PATCH net] be2net: Fix NULL pointer dereference in
 be_tx_timeout()

On 28. 11. 18 20:00, David Miller wrote:
> From: Ivan Vecera <ivecera@...hat.com>
> Date: Wed, 28 Nov 2018 11:12:22 +0100
> 
>> On 27. 11. 18 23:51, David Miller wrote:
>>> From: Petr Oros <poros@...hat.com>
>>> Date: Thu, 22 Nov 2018 15:24:07 +0100
>>>
>>>> @@ -4700,8 +4700,11 @@ int be_update_queues(struct be_adapter
>>>> *adapter)
>>>>    	struct net_device *netdev = adapter->netdev;
>>>>    	int status;
>>>>    -	if (netif_running(netdev))
>>>> +	if (netif_running(netdev)) {
>>>> +		/* prevent netdev watchdog during tx queue destroy */
>>>> +		netif_carrier_off(netdev);
>>>>    		be_close(netdev);
>>>> +	}
>>> This will make userspace networking daemons will think that the link
>>> went down.
>>> This absolutely should not be a side effect of making a simple
>>> TX queue configuration change via ethtool.
>>>
>>
>> Yes, you're right Dave. But the same thing (netif_carrier_off()) is
>> done by number of other drivers (igb, tg3, ixgbe...) during
>> .set_channels() callback. The patch that Petr sent does the
>> practically the same thing like this one:
> 
> Bug exist in other drivers, thanks for reporting that.
> 
> It doesn't mean you should add the same bug here as well.
> 
OK.
And the right way? netif_device_detach() that does not fire linkwatch events?

Thx,
Ivan