netdev - Re: [PATCH net 1/3] net: mvneta: Fix spinlock usage

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: <87wppc2iqz.fsf@free-electrons.com>
Date:	Wed, 09 Mar 2016 08:49:40 +0100
From:	Gregory CLEMENT <gregory.clement@...e-electrons.com>
To:	Jisheng Zhang <jszhang@...vell.com>
Cc:	"David S. Miller" <davem@...emloft.net>,
	Thomas Petazzoni <thomas.petazzoni@...e-electrons.com>,
	<linux-kernel@...r.kernel.org>, <netdev@...r.kernel.org>,
	Lior Amsalem <alior@...vell.com>, Andrew Lunn <andrew@...n.ch>,
	Jason Cooper <jason@...edaemon.net>,
	Ofer Heifetz <oferh@...vell.com>,
	Nadav Haklai <nadavh@...vell.com>,
	Patrick Uiterwijk <patrick@...terwijk.org>,
	"Marcin Wojtas" <mw@...ihalf.com>,
	Dimitri Epshtein <dima@...vell.com>,
	<linux-arm-kernel@...ts.infradead.org>,
	Sebastian Hesselbarth <sebastian.hesselbarth@...il.com>
Subject: Re: [PATCH net 1/3] net: mvneta: Fix spinlock usage

Hi Jisheng,
 
 On mer., mars 09 2016, Jisheng Zhang <jszhang@...vell.com> wrote:

> Dear Gregory,
>
> On Tue, 8 Mar 2016 13:57:04 +0100 Gregory CLEMENT wrote:
>
>> In the previous patch, the spinlock was not initialized. While it didn't
>> cause any trouble yet it could be a problem to use it uninitialized.
>> 
>> The most annoying part was the critical section protected by the spinlock
>> in mvneta_stop(). Some of the functions could sleep as pointed when
>> activated CONFIG_DEBUG_ATOMIC_SLEEP. Actually, in mvneta_stop() we only
>> need to protect the is_stopped flagged, indeed the code of the notifier
>> for CPU online is protected by the same spinlock, so when we get the
>> lock, the notifer work is done.
>> 
>> Reported-by: Patrick Uiterwijk <patrick@...terwijk.org>
>> Signed-off-by: Gregory CLEMENT <gregory.clement@...e-electrons.com>
>> ---
>>  drivers/net/ethernet/marvell/mvneta.c | 11 ++++++-----
>>  1 file changed, 6 insertions(+), 5 deletions(-)
>> 
>> diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
>> index b0ae69f84493..8dc7df2edff6 100644
>> --- a/drivers/net/ethernet/marvell/mvneta.c
>> +++ b/drivers/net/ethernet/marvell/mvneta.c
>> @@ -3070,17 +3070,17 @@ static int mvneta_stop(struct net_device *dev)
>>  	struct mvneta_port *pp = netdev_priv(dev);
>>  
>>  	/* Inform that we are stopping so we don't want to setup the
>> -	 * driver for new CPUs in the notifiers
>> +	 * driver for new CPUs in the notifiers. The code of the
>> +	 * notifier for CPU online is protected by the same spinlock,
>> +	 * so when we get the lock, the notifer work is done.
>>  	 */
>>  	spin_lock(&pp->lock);
>>  	pp->is_stopped = true;
>> +	spin_unlock(&pp->lock);
>
> This fix sleep in atomic issue. But
> I see race here. Let's assume is_stopped is false.

You forgot that the lock was hold in the mvneta_percpu_notifier so your
scenario can't happen.

>
> cpu0:                           	cpu1:
> mvneta_percpu_notifier():		mvneta_stop():
>

spin_lock(&pp->lock);

> if (pp->is_stopped) {
> 	spin_unlock(&pp->lock);
> 	break;
> }
>

                                      the lock is hold in
                                      mvneta_percpu_notifier(), so as
                                      said in the comment this cpu is
                                      waiting for on the following
                                      line:
                                      spin_lock(&pp->lock);

                                      This code will be executed only
                                      when the lock will be released
> 					pp->is_stopped = true;
> 					spin_unlock(&pp->lock);
>
>
> netif_tx_stop_all_queues(pp->dev);
> for_each_online_cpu(other_cpu) {
> ....
>
So what will happen is:
cpu0:                           	cpu1:
mvneta_percpu_notifier():		mvneta_stop():

spin_lock(&pp->lock);
if (pp->is_stopped) {
	spin_unlock(&pp->lock);
	break;
}
                                        spin_lock(&pp->lock);

netif_tx_stop_all_queues(pp->dev);
for_each_online_cpu(other_cpu) {
....
spin_unlock(&pp->lock);
					pp->is_stopped = true;
					spin_unlock(&pp->lock);


Gregory

-- 
Gregory Clement, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com