[PATCH] Set the correct RTNL family for multicast netconf messages

[Sven-Thorsten Dietrich]

There may be other cases that require a special case, so I chose the switch approach.

Please note, that all other messages from ipmr[6].c correctly set the family, while these do not.

Thanks

Sven



Subject: Set correct RTNL family for multicast netconf messages
From: Sven-Thorsten Dietrich sv...@vyatta.com Thu Jun 27 16:40:17 2013 -0700
Date: Thu Jun 27 16:40:17 2013 -0700:


diff --git a/net/ipv4/devinet.c b/net/ipv4/devinet.c
index dfc39d4..695858b 100644
--- a/net/ipv4/devinet.c
+++ b/net/ipv4/devinet.c
@@ -1705,7 +1705,16 @@ static int inet_netconf_fill_devconf(struct sk_buff *skb, int ifindex,
return -EMSGSIZE;

ncm = nlmsg_data(nlh);
- ncm->ncm_family = AF_INET;
+
+ switch (type) {
+ case NETCONFA_MC_FORWARDING:
+ ncm->ncm_family = RTNL_FAMILY_IPMR;
+ break;
+
+ default:
+ ncm->ncm_family = AF_INET;
+ break;
+ }

if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0)
goto nla_put_failure;
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 4ab4c38..a177da4 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -492,7 +492,16 @@ static int inet6_netconf_fill_devconf(struct sk_buff *skb, int ifindex,
return -EMSGSIZE;

ncm = nlmsg_data(nlh);
- ncm->ncm_family = AF_INET6;
+
+ switch (type) {
+ case NETCONFA_MC_FORWARDING:
+ ncm->ncm_family = RTNL_FAMILY_IP6MR;
+ break;
+
+ default:
+ ncm->ncm_family = AF_INET6;
+ break;
+ }

if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0)
goto nla_put_failure;

[Hannes Frederic Sowa]

Hm, are you sure? NETCONFA_MC_FORWARDING is of type RTM_NEWNETCONF
and expects ncm_family to be either AF_INET or AF_INET6 (at least in
iproute2/ipmonitor.c).

Greetings,

Hannes

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majo...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

[Stephen Hemminger]

I agree with Sven on this, looks like the recent addition of netconf
configuration to netlink didn't embrace how multicast is handled in kernel.

Multicast forwarding is a routing related configuration value.
All the multicast routing events come in as special family RTNL_FAMILY_IPMR
(see net/ipv4/ipmr.c function ipmr_fill_route). I would expect that multicast
routing daemons would like to be able to use special family to listen for
all multicast related changes (and not see non-multicast events).


Minor nit: the patch is formatted incorrectly (case should line up with switch).

[Hannes Frederic Sowa]

On Thu, Jun 27, 2013 at 06:33:42PM -0700, Stephen Hemminger wrote:
> On Fri, 28 Jun 2013 03:23:07 +0200
> Hannes Frederic Sowa <han...@stressinduktion.org> wrote:
> > Hm, are you sure? NETCONFA_MC_FORWARDING is of type RTM_NEWNETCONF
> > and expects ncm_family to be either AF_INET or AF_INET6 (at least in
> > iproute2/ipmonitor.c).
> >
>
> I agree with Sven on this, looks like the recent addition of netconf
> configuration to netlink didn't embrace how multicast is handled in kernel.
>
> Multicast forwarding is a routing related configuration value.
> All the multicast routing events come in as special family RTNL_FAMILY_IPMR
> (see net/ipv4/ipmr.c function ipmr_fill_route). I would expect that multicast
> routing daemons would like to be able to use special family to listen for
> all multicast related changes (and not see non-multicast events).
>
>
> Minor nit: the patch is formatted incorrectly (case should line up with switch).

Yes, this seems reasonable but would need a small update to ipnetconf.c, too.

Thanks,

Hannes

[Nicolas Dichtel]

Le 28/06/2013 03:51, Hannes Frederic Sowa a écrit :
> On Thu, Jun 27, 2013 at 06:33:42PM -0700, Stephen Hemminger wrote:
>> On Fri, 28 Jun 2013 03:23:07 +0200
>> Hannes Frederic Sowa <han...@stressinduktion.org> wrote:
>>> Hm, are you sure? NETCONFA_MC_FORWARDING is of type RTM_NEWNETCONF
>>> and expects ncm_family to be either AF_INET or AF_INET6 (at least in
>>> iproute2/ipmonitor.c).
>>>
>>
>> I agree with Sven on this, looks like the recent addition of netconf
>> configuration to netlink didn't embrace how multicast is handled in kernel.
>>
>> Multicast forwarding is a routing related configuration value.
>> All the multicast routing events come in as special family RTNL_FAMILY_IPMR
>> (see net/ipv4/ipmr.c function ipmr_fill_route). I would expect that multicast
>> routing daemons would like to be able to use special family to listen for
>> all multicast related changes (and not see non-multicast events).
>>
>>
>> Minor nit: the patch is formatted incorrectly (case should line up with switch).
>
> Yes, this seems reasonable but would need a small update to ipnetconf.c, too.

I also agree with Sven and Stephen.

Note also that the Signed-off-by line is missing in the commit log.

Sven, can you take care of the iproute2 patch? If not, let me know so I can do it.


Thank you,
Nicolas

[Stephen Hemminger]

Found another issue that needs some thought.
If type == ALL, in that case the family comes up as AF_INET.
That means that if application is doing a request to get netconf
it will receive a different answer than if it is montoring for netconf
changes.

One way to solve would be to split fill_devconf into two parts, one
for unicast, and one for multicast.

[Nicolas Dichtel]

If I understand well, to get all conf variables for IPv4, you will need to make
two dump?

Note that the initial idea of netconf was to be able to dump via netlink the
content of /proc/sys/net/ipv4/conf/ or /proc/sys/net/ipv6/conf/, hence
AF_INET[6] was used to specify 'ipv4' or 'ipv6'.

[Stephen Hemminger]

Revised version of Sven's patch. The idea is that multicast forwarding
should be under the multicast address family like other multicast netlink
messages.

This version generates each family under separate headers when
doing dump all.

Compile tested only, this is to show some of the issues that need
to be covered.

Signed-off-by: Stephen Hemminger <ste...@networkplumber.org>


--- a/net/ipv4/devinet.c 2013-06-11 09:50:21.550918636 -0700
+++ b/net/ipv4/devinet.c 2013-06-28 09:07:18.147829543 -0700
@@ -1685,6 +1685,11 @@ static int inet_netconf_msgsize_devconf(
size += nla_total_size(4);
if (type == -1 || type == NETCONFA_RP_FILTER)
size += nla_total_size(4);
+
+ /* additional header for MC family */
+ if (type == -1)
+ size += NLMSG_ALIGN(sizeof(struct netconfmsg))
+ + nla_total_size(4);
if (type == -1 || type == NETCONFA_MC_FORWARDING)
size += nla_total_size(4);

@@ -1694,7 +1699,7 @@ static int inet_netconf_msgsize_devconf(

static int inet_netconf_fill_devconf(struct sk_buff *skb, int ifindex,

struct ipv4_devconf *devconf, u32 portid,
u32 seq, int event, unsigned int flags,
- int type)
+ u8 family, int type)
{
struct nlmsghdr *nlh;
struct netconfmsg *ncm;
@@ -1705,21 +1710,24 @@ static int inet_netconf_fill_devconf(str

return -EMSGSIZE;

ncm = nlmsg_data(nlh);
- ncm->ncm_family = AF_INET;

+ ncm->ncm_family = family;

if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0)
goto nla_put_failure;

/* type -1 is used for ALL */
- if ((type == -1 || type == NETCONFA_FORWARDING) &&
+ if (((type == -1 && family == AF_INET) ||
+ type == NETCONFA_FORWARDING) &&
nla_put_s32(skb, NETCONFA_FORWARDING,
IPV4_DEVCONF(*devconf, FORWARDING)) < 0)
goto nla_put_failure;
- if ((type == -1 || type == NETCONFA_RP_FILTER) &&
+ if (((type == -1 && family == AF_INET) ||
+ type == NETCONFA_RP_FILTER) &&
nla_put_s32(skb, NETCONFA_RP_FILTER,
IPV4_DEVCONF(*devconf, RP_FILTER)) < 0)
goto nla_put_failure;
- if ((type == -1 || type == NETCONFA_MC_FORWARDING) &&
+ if (((type == -1 || family == RTNL_FAMILY_IPMR) ||
+ type == NETCONFA_MC_FORWARDING) &&
nla_put_s32(skb, NETCONFA_MC_FORWARDING,
IPV4_DEVCONF(*devconf, MC_FORWARDING)) < 0)
goto nla_put_failure;
@@ -1737,12 +1745,17 @@ void inet_netconf_notify_devconf(struct
struct sk_buff *skb;
int err = -ENOBUFS;

+ BUG_ON(type == -1); /* ALL is not valid for notification */
+
skb = nlmsg_new(inet_netconf_msgsize_devconf(type), GFP_ATOMIC);
if (skb == NULL)
goto errout;

err = inet_netconf_fill_devconf(skb, ifindex, devconf, 0, 0,
- RTM_NEWNETCONF, 0, type);
+ RTM_NEWNETCONF, 0,
+ (type == NETCONFA_MC_FORWARDING) ?
+ RTNL_FAMILY_IPMR : AF_INET,
+ type);
if (err < 0) {
/* -EMSGSIZE implies BUG in inet_netconf_msgsize_devconf() */
WARN_ON(err == -EMSGSIZE);
@@ -1811,13 +1824,23 @@ static int inet_netconf_get_devconf(stru
err = inet_netconf_fill_devconf(skb, ifindex, devconf,
NETLINK_CB(in_skb).portid,
nlh->nlmsg_seq, RTM_NEWNETCONF, 0,
- -1);
+ AF_INET, -1);
if (err < 0) {
/* -EMSGSIZE implies BUG in inet_netconf_msgsize_devconf() */
WARN_ON(err == -EMSGSIZE);
kfree_skb(skb);
goto errout;
}
+
+ err = inet_netconf_fill_devconf(skb, ifindex, devconf,
+ NETLINK_CB(in_skb).portid,
+ nlh->nlmsg_seq, RTM_NEWNETCONF, 0,
+ RTNL_FAMILY_IPMR, -1);
+ if (err < 0) {
+ WARN_ON(err == -EMSGSIZE);
+ kfree_skb(skb);
+ goto errout;
+ }
err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
errout:
return err;
@@ -1855,10 +1878,23 @@ static int inet_netconf_dump_devconf(str
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF,
NLM_F_MULTI,
+ AF_INET,
+ -1) <= 0) {
+ rcu_read_unlock();
+ goto done;
+ }
+ if (inet_netconf_fill_devconf(skb, dev->ifindex,
+ &in_dev->cnf,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF,
+ NLM_F_MULTI,
+ RTNL_FAMILY_IPMR,
-1) <= 0) {
rcu_read_unlock();
goto done;
}
+
nl_dump_check_consistent(cb, nlmsg_hdr(skb));
cont:
idx++;
@@ -1871,10 +1907,18 @@ cont:
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF, NLM_F_MULTI,
- -1) <= 0)
+ AF_INET, -1) <= 0)
+ goto done;
+
+ if (inet_netconf_fill_devconf(skb, NETCONFA_IFINDEX_ALL,
+ net->ipv4.devconf_all,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF, NLM_F_MULTI,
+ RTNL_FAMILY_IPMR, -1) <= 0)
goto done;
- else
- h++;
+
+ h++;
}
if (h == NETDEV_HASHENTRIES + 1) {
if (inet_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT,
@@ -1882,10 +1926,18 @@ cont:
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF, NLM_F_MULTI,
- -1) <= 0)
+ AF_INET, -1) <= 0)
goto done;
- else
- h++;
+
+ if (inet_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT,
+ net->ipv4.devconf_dflt,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF, NLM_F_MULTI,
+ RTNL_FAMILY_IPMR, -1) <= 0)
+ goto done;
+
+ h++;
}
done:
cb->args[0] = h;
--- a/net/ipv6/addrconf.c 2013-06-28 08:17:16.424664740 -0700
+++ b/net/ipv6/addrconf.c 2013-06-28 09:10:32.336959806 -0700
@@ -471,6 +471,9 @@ static int inet6_netconf_msgsize_devconf
if (type == -1 || type == NETCONFA_FORWARDING)
size += nla_total_size(4);
#ifdef CONFIG_IPV6_MROUTE
+ if (type == -1)
+ size += NLMSG_ALIGN(sizeof(struct netconfmsg))
+ + nla_total_size(4);
if (type == -1 || type == NETCONFA_MC_FORWARDING)
size += nla_total_size(4);
#endif
@@ -481,7 +484,7 @@ static int inet6_netconf_msgsize_devconf

static int inet6_netconf_fill_devconf(struct sk_buff *skb, int ifindex,

struct ipv6_devconf *devconf, u32 portid,
u32 seq, int event, unsigned int flags,
- int type)
+ u8 family, int type)
{
struct nlmsghdr *nlh;
struct netconfmsg *ncm;
@@ -492,17 +495,19 @@ static int inet6_netconf_fill_devconf(st

return -EMSGSIZE;

ncm = nlmsg_data(nlh);
- ncm->ncm_family = AF_INET6;

+ ncm->ncm_family = family;

if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0)
goto nla_put_failure;

/* type -1 is used for ALL */
- if ((type == -1 || type == NETCONFA_FORWARDING) &&
+ if (((type == -1 && family == AF_INET6) ||
+ type == NETCONFA_FORWARDING) &&
nla_put_s32(skb, NETCONFA_FORWARDING, devconf->forwarding) < 0)
goto nla_put_failure;
#ifdef CONFIG_IPV6_MROUTE
- if ((type == -1 || type == NETCONFA_MC_FORWARDING) &&
+ if (((type == -1 && family == RTNL_FAMILY_IP6MR) ||
+ type == NETCONFA_MC_FORWARDING) &&
nla_put_s32(skb, NETCONFA_MC_FORWARDING,
devconf->mc_forwarding) < 0)
goto nla_put_failure;
@@ -520,12 +525,23 @@ void inet6_netconf_notify_devconf(struct
struct sk_buff *skb;
int err = -ENOBUFS;

+ BUG_ON(type == -1);
+
skb = nlmsg_new(inet6_netconf_msgsize_devconf(type), GFP_ATOMIC);
if (skb == NULL)
goto errout;

+#ifdef CONFIG_IPV6_MROUTE
+ if (type == NETCONFA_MC_FORWARDING)
+ err = inet6_netconf_fill_devconf(skb, ifindex, devconf, 0, 0,
+ RTM_NEWNETCONF, 0,
+ RTNL_FAMILY_IP6MR, type);
+ else
+#endif
err = inet6_netconf_fill_devconf(skb, ifindex, devconf, 0, 0,
- RTM_NEWNETCONF, 0, type);
+ RTM_NEWNETCONF, 0,
+ AF_INET6, type);
+#endif
if (err < 0) {
/* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
WARN_ON(err == -EMSGSIZE);
@@ -592,13 +608,25 @@ static int inet6_netconf_get_devconf(str
err = inet6_netconf_fill_devconf(skb, ifindex, devconf,
NETLINK_CB(in_skb).portid,
nlh->nlmsg_seq, RTM_NEWNETCONF, 0,
- -1);
+ AF_INET6, -1);
+ if (err < 0) {
+ /* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
+ WARN_ON(err == -EMSGSIZE);
+ kfree_skb(skb);
+ goto errout;
+ }
+#ifdef CONFIG_IPV6_MROUTE
+ err = inet6_netconf_fill_devconf(skb, ifindex, devconf,
+ NETLINK_CB(in_skb).portid,
+ nlh->nlmsg_seq, RTM_NEWNETCONF, 0,
+ RTNL_FAMILY_IP6MR, -1);
if (err < 0) {
/* -EMSGSIZE implies BUG in inet6_netconf_msgsize_devconf() */
WARN_ON(err == -EMSGSIZE);
kfree_skb(skb);
goto errout;
}
+#endif
err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
errout:
return err;
@@ -636,10 +664,23 @@ static int inet6_netconf_dump_devconf(st
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF,
NLM_F_MULTI,
+ AF_INET6, -1) <= 0) {
+ rcu_read_unlock();
+ goto done;
+ }
+#ifdef CONFIG_IPV6_MROUTE
+ if (inet6_netconf_fill_devconf(skb, dev->ifindex,
+ &idev->cnf,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF,
+ NLM_F_MULTI,
+ RTNL_FAMILY_IP6MR,
-1) <= 0) {
rcu_read_unlock();
goto done;
}
+#endif
nl_dump_check_consistent(cb, nlmsg_hdr(skb));
cont:
idx++;
@@ -652,10 +693,18 @@ cont:
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF, NLM_F_MULTI,
- -1) <= 0)
+ AF_INET6, -1) <= 0)
goto done;
- else
- h++;
+#ifdef CONFIG_IPV6_MROUTE
+ if (inet6_netconf_fill_devconf(skb, NETCONFA_IFINDEX_ALL,
+ net->ipv6.devconf_all,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF, NLM_F_MULTI,
+ RTNL_FAMILY_IP6MR, -1) <= 0)
+ goto done;
+#endif
+ h++;
}
if (h == NETDEV_HASHENTRIES + 1) {
if (inet6_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT,
@@ -663,9 +712,17 @@ cont:
NETLINK_CB(cb->skb).portid,
cb->nlh->nlmsg_seq,
RTM_NEWNETCONF, NLM_F_MULTI,
- -1) <= 0)
+ AF_INET6, -1) <= 0)
+ goto done;
+#ifdef CONFIG_IPV6_MROUTE
+ if (inet6_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT,
+ net->ipv6.devconf_dflt,
+ NETLINK_CB(cb->skb).portid,
+ cb->nlh->nlmsg_seq,
+ RTM_NEWNETCONF, NLM_F_MULTI,
+ RTNL_FAMILY_IP6MR, -1) <= 0)
goto done;
- else
+#endif
h++;
}
done:

[Nicolas Dichtel]

Le 28/06/2013 18:13, Stephen Hemminger a �crit :

> Revised version of Sven's patch. The idea is that multicast forwarding
> should be under the multicast address family like other multicast netlink
> messages.
>
> This version generates each family under separate headers when
> doing dump all.
>
> Compile tested only, this is to show some of the issues that need
> to be covered.

I'm ok with the principle, just some comment below.

should be '&&'

inet_netconf_fill_devconf() is called twice, so to be homogeneous, we can call
inet_netconf_msgsize_devconf() twice too and add a argument 'family' to this
function. Same for IPv6.

This line should be reindent too.