netdev - Re: [PATCH net] net: bridge: mcast: don't delete permanent entries when fast leave is enabled

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <651caecf-00dc-cb1a-1908-cb606f263843@cumulusnetworks.com>
Date:   Tue, 30 Jul 2019 20:23:29 +0300
From:   Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
To:     David Miller <davem@...emloft.net>
Cc:     netdev@...r.kernel.org, roopa@...ulusnetworks.com,
        bridge@...ts.linux-foundation.org
Subject: Re: [PATCH net] net: bridge: mcast: don't delete permanent entries
 when fast leave is enabled

On 30/07/2019 20:21, Nikolay Aleksandrov wrote:
> On 30/07/2019 20:18, David Miller wrote:
>> From: Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
>> Date: Tue, 30 Jul 2019 14:21:00 +0300
>>
>>> diff --git a/net/bridge/br_multicast.c b/net/bridge/br_multicast.c
>>> index 3d8deac2353d..f8cac3702712 100644
>>> --- a/net/bridge/br_multicast.c
>>> +++ b/net/bridge/br_multicast.c
>>> @@ -1388,6 +1388,9 @@ br_multicast_leave_group(struct net_bridge *br,
>>>  			if (!br_port_group_equal(p, port, src))
>>>  				continue;
>>>  
>>> +			if (p->flags & MDB_PG_FLAGS_PERMANENT)
>>> +				break;
>>> +
>>
>> Like David, I also don't understand why this can be a break.  Is it because
>> permanent entries are always the last on the list?  Why will there be no
>> other entries that might need to be processed on the list?
>>
> 
> The reason is that only one port can match. See the first clause of br_port_group_equal,
> that port can participate only once. We could easily add a break statement in the end
> when a match is found and it will be correct. Even in the presence of MULTICAST_TO_UNICAST
> flag, the port must match and can be added only once.
> 

Like I wrote in the patch I plan to re-work that code in net-next to remove the
duplication and make it more understandable to avoid such confusions. This code
will be functionally equivalent if I put continue there, we'll just walk over
all of them even after a match or permanent are found. There can only be one
match though as I said, so walking the rest of the ports is a waste.