netdev - Re: [Bug 68011] New: spin_unlock is missed in function (netpoll_send_skb_on_dev) in file (linux-3.12/net/core/netpoll.c)

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]

Date:	Fri, 3 Jan 2014 04:50:07 +0000 (UTC)
From:	Cong Wang <xiyou.wangcong@...il.com>
To:	netdev@...r.kernel.org
Subject: Re: [Bug 68011] New: spin_unlock is missed in function
 (netpoll_send_skb_on_dev) in file (linux-3.12/net/core/netpoll.c)

On Fri, 03 Jan 2014 at 00:51 GMT, David Miller <davem@...emloft.net> wrote:
> From: Stephen Hemminger <stephen@...workplumber.org>
> Date: Wed, 1 Jan 2014 21:52:31 -0800
>
>> In function (netpoll_send_skb_on_dev) in file (linux-3.12/net/core/netpoll.c):
>> 
>> The structure (txq->_xmit_lock) gets successfully locked at line (383) by
>> (__netif_tx_trylock(txq)) and unlocked by (__netif_tx_unlock(txq)) at line
>> (398).
>> 
>> The problem occurs when the loop breaks at line (390) and the structure
>> (txq->_xmit_lock) still locked. In that case, the structure (txq->_xmit_lock)
>> never gets unlocked.
>> 
>> A possible solution is to call (__netif_tx_unlock(txq)) before the break at
>> line (390)
>
> This code path has another problem, if __vlan_put_tag() returns NULL then
> the function exit path is going to try and insert that NULL skb back onto
> the npinfo->txq, resulting in a crash.
>
> We effectively, therefore, don't have the packet any more and as
> things stand the one and only thing we could do is just unlock and
> return immediately.  But this means we won't retry sending later, and
> will thus lose the frame.
>
> netpoll should try to be as drop free as possible, that's why it has all
> of this retry logic.
>
> There really should be a version of __vlan_put_tag() that allows the caller
> to unwind and try again somehow.
>
> Here's what I'll apply and queue up for -stable, thanks.
>
>====================
> [PATCH] netpoll: Fix missing TXQ unlock and and OOPS.
>
> The VLAN tag handling code in netpoll_send_skb_on_dev() has two problems.
>
> 1) It exits without unlocking the TXQ.
>
> 2) It then tries to queue a NULL skb to npinfo->txq.
>
> Reported-by: Ahmed Tamrawi <atamrawi@...tate.edu>
> Signed-off-by: David S. Miller <davem@...emloft.net>

Reviewed-by: Cong Wang <xiyou.wangcong@...il.com>

Thanks!

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html