NAK: [F/OEM-5.6][SRU][PATCH 0/1] l2tp: remove skb_dst_set() from l2tp_xmit_skb()

[Po-Hsu Lin]

Hi Tim,
For the bug nomination I think it's correct. OEM-5.6 kernel for Focal.
However I found that the buglink is incorrect, v2 submitted with the right one.
Thanks
Sam

On Sat, Mar 20, 2021 at 1:43 AM Tim Gardner <tim.gardner at canonical.com> wrote:
>
> I'd applied and pushed to the repo before I went to flip the state on
> the bug which is now marked invalid. Please remember to NAK your patches
> in that case.
>
> rtg
>
> On 3/16/21 5:06 AM, Po-Hsu Lin wrote:
> > [Impact]
> > While debugging bug 1880631, I found that this l2tp.sh test from
> > kselftest/net will cause the dmesg output on Focal OEM 5.6 kernel
> > flooded with:
> > "unregister_netdevice: waiting for eth0 to become free. Usage count = 1"
> > for about every 10 seconds.
> >
> > Test log:
> > $ sudo ./l2tp.sh
> > TEST: IPv4 basic L2TP tunnel                                        [ OK ]
> > TEST: IPv4 route through L2TP tunnel                                [ OK ]
> > TEST: IPv6 basic L2TP tunnel                                        [ OK ]
> > TEST: IPv6 route through L2TP tunnel                                [ OK ]
> > TEST: IPv4 basic L2TP tunnel - with IPsec                           [ OK ]
> > TEST: IPv4 route through L2TP tunnel - with IPsec                   [ OK ]
> > TEST: IPv6 basic L2TP tunnel - with IPsec                           [ OK ]
> > TEST: IPv6 route through L2TP tunnel - with IPsec                   [ OK ]
> > TEST: IPv4 basic L2TP tunnel                                        [ OK ]
> > TEST: IPv4 route through L2TP tunnel                                [ OK ]
> > TEST: IPv6 basic L2TP tunnel - with IPsec                           [ OK ]
> > TEST: IPv6 route through L2TP tunnel - with IPsec                   [ OK ]
> > TEST: IPv4 basic L2TP tunnel - after IPsec teardown                 [ OK ]
> > TEST: IPv4 route through L2TP tunnel - after IPsec teardown         [ OK ]
> > TEST: IPv6 basic L2TP tunnel - after IPsec teardown                 [ OK ]
> > TEST: IPv6 route through L2TP tunnel - after IPsec teardown         [ OK ]
> >
> > Tests passed:  16
> > Tests failed:   0
> >
> > dmesg output:
> > $ dmesg
> > [  708.560589] eth1: renamed from tmp
> > [  708.587396] IPv6: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready
> > [  708.587472] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
> > [  708.616950] eth2: renamed from tmp
> > [  708.643341] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
> > [  708.687057] l2tp_core: L2TP core driver, V2.0
> > [  708.689380] l2tp_netlink: L2TP netlink interface
> > [  708.695479] l2tp_ip: L2TP IP encapsulation support (L2TPv3)
> > [  708.700061] l2tp_eth: L2TP ethernet pseudowire support (L2TPv3)
> > [  708.726295] l2tp_ip6: L2TP IP encapsulation support for IPv6 (L2TPv3)
> > [  708.788973] Initializing XFRM netlink socket
> > [  719.448364] unregister_netdevice: waiting for eth0 to become free. Usage count = 1
> > [  729.528359] unregister_netdevice: waiting for eth0 to become free. Usage count = 1
> > [  739.608359] unregister_netdevice: waiting for eth0 to become free. Usage count = 1
> >
> > [Fix]
> > * 27d53323664c54 ("l2tp: remove skb_dst_set() from l2tp_xmit_skb()")
> >
> > We have this patch in all of our generic kernels and OEM-5.10, thus
> > for now I can see just OEM-5.6 is affected. For which this patch can
> > be cherry-picked.
> >
> > [Test Case]
> > This can tested with the l2tp.sh script in the kernel tree:
> > # sh ./tools/testing/selftests/net/l2tp.sh
> >
> > With the patched kernel, it won't get flooded with that
> > unregister_netdevice message anymore.
> >
> > Also, I have it tested with ubuntu_stress_smoke_tests as well, to
> > ensure the sctp stressor won't trigger any issue like in bug 1854968.
> >
> > A test kernel can be found here:
> > https://people.canonical.com/~phlin/kernel/lp-1919277-l2tp/
> >
> > [Where problems could occur]
> > This change is testable. A quick search didn't show any know issue
> > regarding this patch. And with the same fix running in different
> > generic kernel for a while I think the regression potential can be
> > considered as low.
> >
> > Xin Long (1):
> >    l2tp: remove skb_dst_set() from l2tp_xmit_skb()
> >
> >   net/l2tp/l2tp_core.c | 5 +----
> >   1 file changed, 1 insertion(+), 4 deletions(-)
> >
>
> --
> -----------
> Tim Gardner
> Canonical, Inc