Skip to contentSkip to conversation

ipv6: fixes for the multicast routes

PATCH Applied Matteo Croce 4 participants 10 messages Jan 15, 2021 – Jan 19, 2021

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover ipv6: fixes for the multicast routes

From: Matteo Croce <[email protected]>

Fix two wrong flags in the IPv6 multicast routes created by the autoconf code.

Matteo Croce (2):
ipv6: create multicast route with RTPROT_KERNEL
ipv6: set multicast flag on the multicast route

 net/ipv6/addrconf.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
Signature · 1 line
2.29.2
#1 lore raw

PATCH 1/2 ipv6: create multicast route with RTPROT_KERNEL

From: Matteo Croce <[email protected]>

The ff00::/8 multicast route is created without specifying the fc_protocol field, so the default RTPROT_BOOT value is used:

  $ ip -6 -d route
  unicast ::1 dev lo proto kernel scope global metric 256 pref medium
  unicast fe80::/64 dev eth0 proto kernel scope global metric 256 pref medium
  unicast ff00::/8 dev eth0 proto boot scope global metric 256 pref medium

As the documentation says, this value identifies routes installed during boot, but the route is created when interface is set up. Change the value to RTPROT_KERNEL which is a better value.

---
 net/ipv6/addrconf.c | 1 +
 1 file changed, 1 insertion(+)
net/ipv6/addrconf.c +1 -0
index eff2cacd5209..19bf6822911c 100644
@@ -2469,6 +2469,7 @@ static void addrconf_add_mroute(struct net_device *dev)
 		.fc_flags = RTF_UP,
 		.fc_type = RTN_UNICAST,
 		.fc_nlinfo.nl_net = dev_net(dev),
+		.fc_protocol = RTPROT_KERNEL,
 	};
 
 	ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);
Signature · 1 line
2.29.2
#4 lore raw

PATCH 2/2 ipv6: set multicast flag on the multicast route

In reply to Matteo Croce

From: Matteo Croce <[email protected]>

The multicast route ff00::/8 is created with type RTN_UNICAST:

  $ ip -6 -d route
  unicast ::1 dev lo proto kernel scope global metric 256 pref medium
  unicast fe80::/64 dev eth0 proto kernel scope global metric 256 pref medium
  unicast ff00::/8 dev eth0 proto kernel scope global metric 256 pref medium

Set the type to RTN_MULTICAST which is more appropriate.

---
 net/ipv6/addrconf.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv6/addrconf.c +1 -1
index 19bf6822911c..9edc5bb2d531 100644
@@ -2467,7 +2467,7 @@ static void addrconf_add_mroute(struct net_device *dev)
 		.fc_ifindex = dev->ifindex,
 		.fc_dst_len = 8,
 		.fc_flags = RTF_UP,
-		.fc_type = RTN_UNICAST,
+		.fc_type = RTN_MULTICAST,
 		.fc_nlinfo.nl_net = dev_net(dev),
 		.fc_protocol = RTPROT_KERNEL,
 	};
Signature · 1 line
2.29.2
#6 lore raw
Quoted 8 lines · Jakub Kicinski
Quoted 4 lines · On Fri, 15 Jan 2021 19:42:07 +0100 Matteo Croce wrote:

From: Matteo Croce <[email protected]>

Fix two wrong flags in the IPv6 multicast routes created by the autoconf code.

Any chance for Fixes tags here?

Right.
For 1/2 I don't know exactly, that code was touched last time in 86872cb57925 ("[IPv6] route: FIB6 configuration using struct fib6_config"), but it was only refactored. Before 86872cb57925, it was pushed in the git import commit by Linus: 1da177e4c3f4 ("Linux-2.6.12-rc2").
BTW, according the history repo, it entered the tree in the 2.4.0 import, so I'd say it's here since the beginning.

While for 2/2 I'd say:

Fixes e8478e80e5a7 ("net/ipv6: Save route type in rt6_info")
Signature · 1 line
per aspera ad upstream
#2 lore raw

Re: [PATCH net 1/2] ipv6: create multicast route with RTPROT_KERNEL

In reply to Matteo Croce
Quoted 25 lines · Matteo Croce

From: Matteo Croce <[email protected]>

The ff00::/8 multicast route is created without specifying the fc_protocol field, so the default RTPROT_BOOT value is used:

  $ ip -6 -d route
  unicast ::1 dev lo proto kernel scope global metric 256 pref medium
  unicast fe80::/64 dev eth0 proto kernel scope global metric 256 pref medium
  unicast ff00::/8 dev eth0 proto boot scope global metric 256 pref medium

As the documentation says, this value identifies routes installed during boot, but the route is created when interface is set up. Change the value to RTPROT_KERNEL which is a better value.

---
 net/ipv6/addrconf.c | 1 +
 1 file changed, 1 insertion(+)
net/ipv6/addrconf.c +0 -0
index eff2cacd5209..19bf6822911c 100644

net/ipv6/addrconf.c static void addrconf_add_mroute(struct net_device *dev) · line 2475

Quoted 7 lines
net/ipv6/addrconf.c +1 -0
@@ -2469,6 +2469,7 @@ static void addrconf_add_mroute(struct net_device *dev)
 		.fc_flags = RTF_UP,
 		.fc_type = RTN_UNICAST,
 		.fc_nlinfo.nl_net = dev_net(dev),
+		.fc_protocol = RTPROT_KERNEL,
 	};
 
 	ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);

What's the motivation for changing this? ie., what s/w cares that it is kernel vs boot?

#7 lore raw
In reply to Matteo Croce
Quoted 19 lines · Matteo Croce
Quoted 8 lines · Jakub Kicinski
Quoted 4 lines · On Fri, 15 Jan 2021 19:42:07 +0100 Matteo Croce wrote:

From: Matteo Croce <[email protected]>

Fix two wrong flags in the IPv6 multicast routes created by the autoconf code.

Any chance for Fixes tags here?

Right.
For 1/2 I don't know exactly, that code was touched last time in 86872cb57925 ("[IPv6] route: FIB6 configuration using struct fib6_config"), but it was only refactored. Before 86872cb57925, it was pushed in the git import commit by Linus: 1da177e4c3f4 ("Linux-2.6.12-rc2").
BTW, according the history repo, it entered the tree in the 2.4.0 import, so I'd say it's here since the beginning.

Quoted 4 lines

While for 2/2 I'd say:

Fixes e8478e80e5a7 ("net/ipv6: Save route type in rt6_info")

As I recall (memory jogging from commit description) my patch only moved the setting from ip6_route_info_create default to here.

The change is correct, just thinking it goes back beyond 4.16. If someone has a system running a 4.14 or earlier kernel it should be easy to know if this was the default prior.

#3 lore raw

Re: [PATCH net 1/2] ipv6: create multicast route with RTPROT_KERNEL

In reply to David Ahern
Quoted 32 lines · David Ahern
Quoted 30 lines · Matteo Croce

From: Matteo Croce <[email protected]>

The ff00::/8 multicast route is created without specifying the fc_protocol field, so the default RTPROT_BOOT value is used:

  $ ip -6 -d route
  unicast ::1 dev lo proto kernel scope global metric 256 pref medium
  unicast fe80::/64 dev eth0 proto kernel scope global metric 256 pref medium
  unicast ff00::/8 dev eth0 proto boot scope global metric 256 pref medium

As the documentation says, this value identifies routes installed during boot, but the route is created when interface is set up. Change the value to RTPROT_KERNEL which is a better value.

---
 net/ipv6/addrconf.c | 1 +
 1 file changed, 1 insertion(+)
net/ipv6/addrconf.c +1 -0
index eff2cacd5209..19bf6822911c 100644
@@ -2469,6 +2469,7 @@ static void addrconf_add_mroute(struct net_device *dev)
              .fc_flags = RTF_UP,
              .fc_type = RTN_UNICAST,
              .fc_nlinfo.nl_net = dev_net(dev),
+             .fc_protocol = RTPROT_KERNEL,
      };
 
Quoted 6 lines

net/ipv6/addrconf.c static void addrconf_add_mroute(struct net_device *dev) · line 2475

Quoted 2 lines
net/ipv6/addrconf.c +0 -0
      ipv6_addr_set(&cfg.fc_dst, htonl(0xFF000000), 0, 0, 0);

What's the motivation for changing this? ie., what s/w cares that it is kernel vs boot?

A practical example: systemd-networkd explicitly ignores routes with kernel flag[1]. Having a different flag leads the daemon to remove the route sooner or later.

[1] https://github.com/systemd/systemd/blob/0b81225e5791f660506f7db0ab88078cf296b771/src/network/networkd-routing-policy-rule.c#L675-L677

Signature · 1 line
per aspera ad upstream
#8 lore raw
In reply to David Ahern
Quoted 26 lines · David Ahern
Quoted 19 lines · Matteo Croce
Quoted 8 lines · Jakub Kicinski
Quoted 4 lines · On Fri, 15 Jan 2021 19:42:07 +0100 Matteo Croce wrote:

From: Matteo Croce <[email protected]>

Fix two wrong flags in the IPv6 multicast routes created by the autoconf code.

Any chance for Fixes tags here?

Right.
For 1/2 I don't know exactly, that code was touched last time in 86872cb57925 ("[IPv6] route: FIB6 configuration using struct fib6_config"), but it was only refactored. Before 86872cb57925, it was pushed in the git import commit by Linus: 1da177e4c3f4 ("Linux-2.6.12-rc2").
BTW, according the history repo, it entered the tree in the 2.4.0 import, so I'd say it's here since the beginning.

Quoted 4 lines

While for 2/2 I'd say:

Fixes e8478e80e5a7 ("net/ipv6: Save route type in rt6_info")
Quoted 6 lines

As I recall (memory jogging from commit description) my patch only moved the setting from ip6_route_info_create default to here.

The change is correct, just thinking it goes back beyond 4.16. If someone has a system running a 4.14 or earlier kernel it should be easy to know if this was the default prior.

Indeed, it was the same long before 4.14:

# uname -a
Linux ubuntu 4.4.0-142-generic #168~14.04.1-Ubuntu SMP Sat Jan 19
11:26:28 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux
# ip -6 -d route show table local dev eth0
unicast ff00::/8 proto boot scope global metric 256 pref medium

Regards,

Signature · 1 line
per aspera ad upstream
#9 lore raw
In reply to Matteo Croce
This series was applied to netdev/net.git (refs/heads/master):

Hello:

This series was applied to netdev/net.git (refs/heads/master):

Quoted 10 lines · On Fri, 15 Jan 2021 19:42:07 +0100 you wrote:

From: Matteo Croce <[email protected]>

Fix two wrong flags in the IPv6 multicast routes created by the autoconf code.

Matteo Croce (2):
ipv6: create multicast route with RTPROT_KERNEL
ipv6: set multicast flag on the multicast route

[...]

Here is the summary with links:
- [net,1/2] ipv6: create multicast route with RTPROT_KERNEL

    https://git.kernel.org/netdev/net/c/a826b04303a4
  - [net,2/2] ipv6: set multicast flag on the multicast route
    https://git.kernel.org/netdev/net/c/ceed9038b278

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html