netdev - bonding reports interface up with 0 Mbps

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: Windows password security audit tool. GUI, reports in PDF.

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

Date:	Wed, 3 Feb 2016 23:10:32 +0000
From:	"Tantilov, Emil S" <emil.s.tantilov@...el.com>
To:	"netdev@...r.kernel.org" <netdev@...r.kernel.org>
CC:	Jay Vosburgh <jay.vosburgh@...onical.com>,
	"gospo@...ulusnetworks.com" <gospo@...ulusnetworks.com>,
	zhuyj <zyjzyj2000@...il.com>,
	"jiri@...lanox.com" <jiri@...lanox.com>
Subject: bonding reports interface up with 0 Mbps

We are seeing an occasional issue where the bonding driver may report interface up with 0 Mbps:
bond0: link status definitely up for interface eth0, 0 Mbps full duplex

So far in all the failed traces I have collected this happens on NETDEV_CHANGELOWERSTATE event:

<...>-20533 [000] .... 81811.041241: ixgbe_service_task: eth1: NIC Link is Up 10 Gbps, Flow Control: RX/TX
<...>-20533 [000] .... 81811.041257: ixgbe_check_vf_rate_limit <-ixgbe_service_task
<...>-20533 [000] .... 81811.041272: ixgbe_ping_all_vfs <-ixgbe_service_task
kworker/u48:0-7503  [010] .... 81811.041345: ixgbe_get_stats64 <-dev_get_stats
kworker/u48:0-7503  [010] .... 81811.041393: bond_netdev_event: eth1: event: 1b
kworker/u48:0-7503  [010] .... 81811.041394: bond_netdev_event: eth1: IFF_SLAVE
kworker/u48:0-7503  [010] .... 81811.041395: bond_netdev_event: eth1: slave->speed = ffffffff
<...>-20533 [000] .... 81811.041407: ixgbe_ptp_overflow_check <-ixgbe_service_task
kworker/u48:0-7503  [010] .... 81811.041407: bond_mii_monitor: bond0: link status definitely up for interface eth1, 0 Mbps full duplex

As a proof of concept I added NETDEV_CHANGELOWERSTATE in bond_slave_netdev_event() along with NETDEV_UP/CHANGE:

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 56b5605..a9dac4c 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -3014,6 +3014,7 @@ static int bond_slave_netdev_event(unsigned long event,
 		break;
 	case NETDEV_UP:
 	case NETDEV_CHANGE:
+	case NETDEV_CHANGELOWERSTATE:
 		bond_update_speed_duplex(slave);
 		if (BOND_MODE(bond) == BOND_MODE_8023AD)
 			bond_3ad_adapter_speed_duplex_changed(slave);

With this change I have not seen 0 Mbps reported by the bonding driver (around 12 hour test up to this point
vs. 2-3 hours otherwise). Although I suppose it could also be some sort of race/timing issue with bond_mii_monitor().

This test is with current bonding driver from net-next (top commit 03d84a5f83).

The bond is configured as such:

mode = 802.3ad
lacp_rate = fast
miimon = 100
xmit_hash_policy = layer3+4

I should note that the speed is reported correctly in /proc/net/bonding/bond0 once the bond0 interface is up,
so this seems to be just an issue with the initial detection of the speed. At least from what I have seen so far.

Thanks,
Emil