netdev - Re: [PATCH net 1/3] net: bonding: Unsync device addresses on ndo

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-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <195900.1661916353@nyx>
Date:   Tue, 30 Aug 2022 20:25:53 -0700
From:   Jay Vosburgh <jay.vosburgh@...onical.com>
To:     Benjamin Poirier <bpoirier@...dia.com>
cc:     netdev@...r.kernel.org, Veaceslav Falico <vfalico@...il.com>,
        Andy Gospodarek <andy@...yhouse.net>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>, Jiri Pirko <jiri@...nulli.us>,
        Shuah Khan <shuah@...nel.org>,
        Jonathan Toppins <jtoppins@...hat.com>,
        linux-kselftest@...r.kernel.org
Subject: Re: [PATCH net 1/3] net: bonding: Unsync device addresses on ndo_stop

Benjamin Poirier <bpoirier@...dia.com> wrote:

>Netdev drivers are expected to call dev_{uc,mc}_sync() in their
>ndo_set_rx_mode method and dev_{uc,mc}_unsync() in their ndo_stop method.
>This is mentioned in the kerneldoc for those dev_* functions.
>
>The bonding driver calls dev_{uc,mc}_unsync() during ndo_uninit instead of
>ndo_stop. This is ineffective because address lists (dev->{uc,mc}) have
>already been emptied in unregister_netdevice_many() before ndo_uninit is
>called. This mistake can result in addresses being leftover on former bond
>slaves after a bond has been deleted; see test_LAG_cleanup() in the last
>patch in this series.
>
>Add unsync calls, via bond_hw_addr_flush(), at their expected location,
>bond_close().
>Add dev_mc_add() call to bond_open() to match the above change.
>The existing call __bond_release_one->bond_hw_addr_flush is left in place
>because there are other call chains that lead to __bond_release_one(), not
>just ndo_uninit.
>
>Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")

	I'm just going from memory here, so I'm probably wrong, but
didn't the sync/unsync stuff for HW addresses happen several years after
the git transition?

>Signed-off-by: Benjamin Poirier <bpoirier@...dia.com>
>---
> drivers/net/bonding/bond_main.c | 31 +++++++++++++++++++++----------
> 1 file changed, 21 insertions(+), 10 deletions(-)
>
>diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>index 2f4da2c13c0a..5784fbe03552 100644
>--- a/drivers/net/bonding/bond_main.c
>+++ b/drivers/net/bonding/bond_main.c
>@@ -254,6 +254,8 @@ static const struct flow_dissector_key flow_keys_bonding_keys[] = {
> 
> static struct flow_dissector flow_keys_bonding __read_mostly;
> 
>+static const u8 lacpdu_multicast[] = MULTICAST_LACPDU_ADDR;
>+
> /*-------------------------- Forward declarations ---------------------------*/
> 
> static int bond_init(struct net_device *bond_dev);
>@@ -865,12 +867,8 @@ static void bond_hw_addr_flush(struct net_device *bond_dev,
> 	dev_uc_unsync(slave_dev, bond_dev);
> 	dev_mc_unsync(slave_dev, bond_dev);
> 
>-	if (BOND_MODE(bond) == BOND_MODE_8023AD) {
>-		/* del lacpdu mc addr from mc list */
>-		u8 lacpdu_multicast[ETH_ALEN] = MULTICAST_LACPDU_ADDR;
>-
>+	if (BOND_MODE(bond) == BOND_MODE_8023AD)
> 		dev_mc_del(slave_dev, lacpdu_multicast);
>-	}
> }
> 
> /*--------------------------- Active slave change ---------------------------*/
>@@ -2171,12 +2169,8 @@ int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev,
> 		dev_uc_sync_multiple(slave_dev, bond_dev);
> 		netif_addr_unlock_bh(bond_dev);
> 
>-		if (BOND_MODE(bond) == BOND_MODE_8023AD) {
>-			/* add lacpdu mc addr to mc list */
>-			u8 lacpdu_multicast[ETH_ALEN] = MULTICAST_LACPDU_ADDR;
>-
>+		if (BOND_MODE(bond) == BOND_MODE_8023AD)
> 			dev_mc_add(slave_dev, lacpdu_multicast);
>-		}
> 	}

	Just to make sure I'm clear, the above changes regarding
lacpdu_multicast have no functional impact, correct?  They appear to
move lacpdu_multicast to global scope for use in the change just below.

> 
> 	bond->slave_cnt++;
>@@ -4211,6 +4205,9 @@ static int bond_open(struct net_device *bond_dev)
> 		/* register to receive LACPDUs */
> 		bond->recv_probe = bond_3ad_lacpdu_recv;
> 		bond_3ad_initiate_agg_selection(bond, 1);
>+
>+		bond_for_each_slave(bond, slave, iter)
>+			dev_mc_add(slave->dev, lacpdu_multicast);
> 	}

	After this patch, am I understanding correctly that both
bond_enslave and bond_open will call dev_mc_add for lacpdu_multicast?
Since __dev_mc_add calls __hw_addr_add_ex with sync=false and
exclusive=false, doesn't that allow us to end up with two references?

	-J

> 	if (bond_mode_can_use_xmit_hash(bond))
>@@ -4222,6 +4219,7 @@ static int bond_open(struct net_device *bond_dev)
> static int bond_close(struct net_device *bond_dev)
> {
> 	struct bonding *bond = netdev_priv(bond_dev);
>+	struct slave *slave;
> 
> 	bond_work_cancel_all(bond);
> 	bond->send_peer_notif = 0;
>@@ -4229,6 +4227,19 @@ static int bond_close(struct net_device *bond_dev)
> 		bond_alb_deinitialize(bond);
> 	bond->recv_probe = NULL;
> 
>+	if (bond_uses_primary(bond)) {
>+		rcu_read_lock();
>+		slave = rcu_dereference(bond->curr_active_slave);
>+		if (slave)
>+			bond_hw_addr_flush(bond_dev, slave->dev);
>+		rcu_read_unlock();
>+	} else {
>+		struct list_head *iter;
>+
>+		bond_for_each_slave(bond, slave, iter)
>+			bond_hw_addr_flush(bond_dev, slave->dev);
>+	}
>+
> 	return 0;
> }
> 
>-- 
>2.36.1
>

---
	-Jay Vosburgh, jay.vosburgh@...onical.com