netdev - Re: [PATCH] loopback: Force LOOPBACK

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: <f79075f1-3c95-8bb3-6600-c6dd2ce7e1ce@gmail.com>
Date:   Wed, 21 Jun 2017 12:35:12 +0300
From:   Serhey Popovych <serhe.popovych@...il.com>
To:     David Miller <davem@...emloft.net>
Cc:     netdev@...r.kernel.org
Subject: Re: [PATCH] loopback: Force LOOPBACK_IFINDEX for registration


> 
>> From: Serhey Popovych <serhe.popovych@...il.com>
>> Date: Fri, 16 Jun 2017 15:10:03 +0300
>>
>>> Now with commit 9c7dafb (net: Allow to create links with
>>> given ifindex) support registration of network devices
>>> with specific ifindex is added.
>>>
>>> We can force loopback network device index before call to
>>> register_netdev() to ensure we always configure it with
>>> LOOPBACK_IFINDEX.
>>>
>>> Kill BUG_ON() since system can continue without network
>>> namespace failed in loopback init path, unless it is
>>> init_net namespace where we panic() anyway.
>>>
>>> Signed-off-by: Serhey Popovych <serhe.popovych@...il.com>
>>
>> Is the BUG_ON() triggering, if so why?
>>
>> It looks to me that unless there is a bug, this assignment
>> is unnecessary.
>>
> Okay, get your position, thanks!
> 

Sorry for raising again, but my objectives following:

BUG_ON() might be compiled out when CONFIG_BUG=n.
That's open possibility in buggy setups to have
"lo" with ->ifindex other than LOOPBACK_IFINDEX.

By forcing register_netdevice() via setting
ifindex = LOOPBACK_IFINDEX we address CONFIG_BUG=n
case.

We can leave BUG_ON(), but add ifindex = LOOPBACK_IFINDEX
to ensure register_netdevice() operates properly.

What do you think?

-- 
Thanks, Serhey