netdev - Re: [PATCH] inet_diag: fix panic when unload inet

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: <50615B30.80901@cn.fujitsu.com>
Date:	Tue, 25 Sep 2012 15:20:16 +0800
From:	Gao feng <gaofeng@...fujitsu.com>
To:	Eric Dumazet <eric.dumazet@...il.com>
CC:	davem@...emloft.net, stephen.hemminger@...tta.com, jengelh@...i.de,
	kuznet@....inr.ac.ru, netdev@...r.kernel.org
Subject: Re: [PATCH] inet_diag: fix panic when unload inet_diag

于 2012年09月25日 14:36, Eric Dumazet 写道:
> On Tue, 2012-09-25 at 10:48 +0800, Gao feng wrote:
>> when inet_diag being compiled as module, inet_diag_handler_dump
>> set netlink_dump_control.dump to inet_diag_dump,so if module
>> inet_diag is unloaded,netlink will still try to call this function
>> in netlink_dump. this will cause kernel panic.
>>
>> fix this by adding a reference of inet_diag module before
>> setting netlink_callback, and release this reference in
>> netlink_callback.done.
>>
>> Thanks for all help from Stephen,Jan and Eric.
> ...
> 
>>  
>> @@ -1001,8 +1025,26 @@ static int inet_diag_handler_dump(struct sk_buff *skb, struct nlmsghdr *h)
>>  		{
>>  			struct netlink_dump_control c = {
>>  				.dump = inet_diag_dump,
>> +				.done = inet_diag_done,
>>  			};
>> -			return netlink_dump_start(net->diag_nlsk, skb, h, &c);
>> +			int err;
>> +			/*
>> +			 * netlink_dump will call inet_diag_dump,
>> +			 * so we need a reference of THIS_MODULE.
>> +			 */
>> +			if (!try_module_get(THIS_MODULE))
>> +				return -EPROTONOSUPPORT;
>> +
>> +			err = netlink_dump_start(net->diag_nlsk, skb, h, &c);
>> +
>> +			if ((err != -EINTR) && (err != -ENOBUFS)) {
>> +				/*
>> +				 * netlink_callback set failed, release the
>> +				 * referenct of THIS_MODULE.
>> +				 */
>> +				module_put(THIS_MODULE);
>> +			}
>> +			return err;
>>  		}
>>  	}
>>  
> 
> Hmm... this seems error prone...
> 
> In the future, netlink_dump_start() could be changed to return other
> errors than EINTR or ENOBUFS that need the module_put()
> 

EINTR and ENOBUFS is returned by netlink_dump, netlink_dump is called by
netlink_dump_start after netlink_callback being set successfully.
so this checking of EINTR and ENOBUFS here is to determinate if we set
netlink_callback successfully.

I think in order to reduce error prone,we have to change netlink_dump_start
to determinate if we set netlink_callback successfully.

> I would change netlink_dump_start() to __netlink_dump_start() and add a
> module param to it, so that this module stuff is centralized in
> __netlink_dump_start()
> 
> Then, instead of calling (from inet_diag)
> 
> netlink_dump_start(net->diag_nlsk, skb, nlh, &c);
> 
> you would use :
> 
> __netlink_dump_start(net->diag_nlsk, skb, nlh, &c, THIS_MODULE);
> 
> I wonder if this fix is not needed elsewhere eventually
> (net/unix/af_unix.c for example ?)
> 

do you mean net/unix/unix_diag.c ?
I test nfnetlink module,it has the same problem.

It's need to modify netlink_dump_start not only wrap netlink_dump_start.
--
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