netdev - Re: [PATCH] via-rhine: Fix 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: <51EC0D03.4070508@nod.at>
Date:	Sun, 21 Jul 2013 18:32:03 +0200
From:	Richard Weinberger <richard@....at>
To:	Ben Hutchings <bhutchings@...arflare.com>
CC:	rl@...lgate.ch, netdev@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] via-rhine: Fix tx_timeout handling

Am 21.07.2013 18:18, schrieb Ben Hutchings:
> On Fri, 2013-07-19 at 23:30 +0200, Richard Weinberger wrote:
>> rhine_reset_task() misses to call netif_stop_queue(),
>> this can lead to a crash if work is still scheduled while
>> we're resetting the tx queue.
>>
>> Fixes:
>> [   93.591707] BUG: unable to handle kernel NULL pointer dereference at 0000004c
>> [   93.595514] IP: [<c119d10d>] rhine_napipoll+0x491/0x6e
>>
>> Signed-off-by: Richard Weinberger <richard@....at>
>> ---
>>  drivers/net/ethernet/via/via-rhine.c | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/net/ethernet/via/via-rhine.c b/drivers/net/ethernet/via/via-rhine.c
>> index b75eb9e..57e1b40 100644
>> --- a/drivers/net/ethernet/via/via-rhine.c
>> +++ b/drivers/net/ethernet/via/via-rhine.c
>> @@ -1615,6 +1615,7 @@ static void rhine_reset_task(struct work_struct *work)
>>  		goto out_unlock;
>>  
>>  	napi_disable(&rp->napi);
>> +	netif_stop_queue(dev);
> 
> This is not really fixing the bug because there is no synchronisation
> with the TX scheduler.  You can call netif_tx_disable() instead to do
> that.

I guess other drivers suffer from that too.
A quick grep showed that not many drivers are using netif_tx_disable().

> (I also think that it is preferable to use
> netif_device_{detach,attach}() to stop the queue during reconfiguration,
> as this is independent of TX completions and the watchdog.)

So the correct down sequence is napi_disable() -> netif_tx_disable() -> netif_device_detach()?

Thanks,
//richard
--
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