netdev - Re: [PATCH net] bonding: fix packets_per

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]

Date:	Thu, 05 Dec 2013 13:42:08 +0100
From:	Nikolay Aleksandrov <nikolay@...hat.com>
To:	Eric Dumazet <eric.dumazet@...il.com>,
	Veaceslav Falico <vfalico@...hat.com>
CC:	netdev@...r.kernel.org, Andy Gospodarek <andy@...yhouse.net>,
	Jay Vosburgh <fubar@...ibm.com>,
	"David S. Miller" <davem@...emloft.net>
Subject: Re: [PATCH net] bonding: fix packets_per_slave showing

On 12/05/2013 01:33 PM, Eric Dumazet wrote:
> On Thu, 2013-12-05 at 12:08 +0100, Veaceslav Falico wrote:
>> On Thu, Dec 05, 2013 at 11:36:58AM +0100, Nikolay Aleksandrov wrote:
>>> There's an issue when showing the value of packets_per_slave due to
>>> using signed integer. The value may be < 0 and thus not put through
>>> reciprocal_value() before showing. This patch makes it use unsigned
>>> integer when showing it.
>>
>> I was already checking my basic algebra knowledge here,
>> reciprocal_value(reciprocal_value(0..USHRT_MAX)) can become negative?!? :)
>>
>> If anyone's also wondering...
>>
>> packets_per_slave is reciprocal_value(0..USHRT_MAX), and thus can indeed be
>> negative, and then the code
>>
>> 	if (packets_per_slave > 1)
>> 		packets_per_slave = reciprocal_value(packets_per_slave);
>>
>> would fail to recognise that it's a reciprocal_divide() value, and not a
>> standard 0/1 option (in bond_rr_gen_slave_id() we verify it via a switch,
>> so we're safe there) - and thus output nonsense.
>>
>> Acked-by: Veaceslav Falico <vfalico@...hat.com>
> 
> This code is very confusing.
> 
> Please rename bond->params.packets_per_slave
> to bond->params.reciprocal_packets_per_slave 
> 
> To make clear that its the reciprocal value.
> 
> Also the module parameter is named packets_per_slave, it would be nice
> if same name was not reused as local variable in bond_rr_gen_slave_id()
> 
> bond_check_params() reads the sys value several times.
> 
> This is racy with /sys access.
> 
IIRC bond_check_params() runs before sysfs is initialized for the bond device.

> You should use ACCESS_ONCE() to make sure nothing bad happens.
> 
Actually I think ACCESS_ONCE() should be added to bond_show_packets_per_slave
and to bond_rr_gen_slave_id() just as a precaution. It'll also serve the purpose
to show what's intended.
What do you think ?
> 
> 
Thanks for the feedback, I'll address the ACCESS_ONCE() in a separate patch for
net and leave the renaming to my net-next patch which will take care of the
types as well.

Cheers,
 Nik


--
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