[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4ECA6742.9080906@trash.net>
Date: Mon, 21 Nov 2011 15:59:14 +0100
From: Patrick McHardy <kaber@...sh.net>
To: Eric Dumazet <eric.dumazet@...il.com>
CC: netfilter-devel@...r.kernel.org, netdev@...r.kernel.org,
ulrich.weber@...hos.com
Subject: Re: [PATCH 02/17] netfilter: nat: remove module reference counting
from NAT protocols
On 11/21/2011 08:36 AM, Eric Dumazet wrote:
> Le lundi 21 novembre 2011 à 06:45 +0100, kaber@...sh.net a écrit :
>> From: Patrick McHardy<kaber@...sh.net>
>>
>> The only remaining user of NAT protocol module reference counting is NAT
>> ctnetlink support. Since this is a fairly short sequence of code, convert
>> over to use RCU and remove module reference counting.
>>
>> Signed-off-by: Patrick McHardy<kaber@...sh.net>
>> @@ -612,16 +592,18 @@ static int nfnetlink_parse_nat_proto(struct nlattr *attr,
>> if (err< 0)
>> return err;
>>
>> - npt = nf_nat_proto_find_get(nf_ct_protonum(ct));
>> + rcu_read_lock();
>> + npt = __nf_nat_proto_find(nf_ct_protonum(ct));
>> if (npt->nlattr_to_range)
>> err = npt->nlattr_to_range(tb, range);
>> - nf_nat_proto_put(npt);
>> + rcu_read_unlock();
>> return err;
>> }
> Seems good, but you should explain in changelog why its safe.
>
> (Using rcu also means special action must be taken in the write side,
> here the module removal)
I'll update the changelog for the next submission.
>>
>> static const struct nla_policy nat_nla_policy[CTA_NAT_MAX+1] = {
>> [CTA_NAT_MINIP] = { .type = NLA_U32 },
>> [CTA_NAT_MAXIP] = { .type = NLA_U32 },
>> + [CTA_NAT_PROTO] = { .type = NLA_NESTED },
>> };
> I just dont understand this part :-
Yeah, it doesn't belong into this patch.
--
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
Powered by blists - more mailing lists