From: Jay Vosburgh <fubar@us.ibm.com>
To: netdev@vger.kernel.org
Cc: Jeff Garzik <jgarzik@pobox.com>,
David Miller <davem@davemloft.net>,
Andy Gospodarek <andy@greyhouse.net>,
Krzysztof Oledzki <olel@ans.pl>, Jay Vosburgh <fubar@us.ibm.com>
Subject: [PATCH 3/3] bonding: fix locking during alb failover and slave removal
Date: Mon, 7 Jan 2008 17:57:00 -0800 [thread overview]
Message-ID: <119975742458-git-send-email-fubar@us.ibm.com> (raw)
In-Reply-To: <1199757423867-git-send-email-fubar@us.ibm.com>
alb_fasten_mac_swap (actually rlb_teach_disabled_mac_on_primary)
requries RTNL and no other locks. This could cause dev_set_promiscuity
and/or dev_set_mac_address to be called with improper locking.
Changed callers to hold only RTNL during calls to alb_fasten_mac_swap
or functions calling it. Updated header comments in affected functions to
reflect proper reality of locking requirements.
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
---
drivers/net/bonding/bond_alb.c | 18 ++++++++++++------
drivers/net/bonding/bond_main.c | 14 ++++++++------
2 files changed, 20 insertions(+), 12 deletions(-)
diff --git a/drivers/net/bonding/bond_alb.c b/drivers/net/bonding/bond_alb.c
index 9b55a12..b57bc94 100644
--- a/drivers/net/bonding/bond_alb.c
+++ b/drivers/net/bonding/bond_alb.c
@@ -979,7 +979,7 @@ static void alb_swap_mac_addr(struct bonding *bond, struct slave *slave1, struct
/*
* Send learning packets after MAC address swap.
*
- * Called with RTNL and bond->lock held for read.
+ * Called with RTNL and no other locks
*/
static void alb_fasten_mac_swap(struct bonding *bond, struct slave *slave1,
struct slave *slave2)
@@ -987,6 +987,8 @@ static void alb_fasten_mac_swap(struct bonding *bond, struct slave *slave1,
int slaves_state_differ = (SLAVE_IS_OK(slave1) != SLAVE_IS_OK(slave2));
struct slave *disabled_slave = NULL;
+ ASSERT_RTNL();
+
/* fasten the change in the switch */
if (SLAVE_IS_OK(slave1)) {
alb_send_learning_packets(slave1, slave1->dev->dev_addr);
@@ -1031,7 +1033,7 @@ static void alb_fasten_mac_swap(struct bonding *bond, struct slave *slave1,
* a slave that has @slave's permanet address as its current address.
* We'll make sure that that slave no longer uses @slave's permanent address.
*
- * Caller must hold bond lock
+ * Caller must hold RTNL and no other locks
*/
static void alb_change_hw_addr_on_detach(struct bonding *bond, struct slave *slave)
{
@@ -1542,7 +1544,12 @@ int bond_alb_init_slave(struct bonding *bond, struct slave *slave)
return 0;
}
-/* Caller must hold bond lock for write */
+/*
+ * Remove slave from tlb and rlb hash tables, and fix up MAC addresses
+ * if necessary.
+ *
+ * Caller must hold RTNL and no other locks
+ */
void bond_alb_deinit_slave(struct bonding *bond, struct slave *slave)
{
if (bond->slave_cnt > 1) {
@@ -1658,12 +1665,11 @@ void bond_alb_handle_active_change(struct bonding *bond, struct slave *new_slave
bond->alb_info.rlb_enabled);
}
- read_lock(&bond->lock);
-
if (swap_slave) {
alb_fasten_mac_swap(bond, swap_slave, new_slave);
+ read_lock(&bond->lock);
} else {
- /* fasten bond mac on new current slave */
+ read_lock(&bond->lock);
alb_send_learning_packets(new_slave, bond->dev->dev_addr);
}
diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index b0b2603..77d004d 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -1746,7 +1746,9 @@ int bond_release(struct net_device *bond_dev, struct net_device *slave_dev)
* has been cleared (if our_slave == old_current),
* but before a new active slave is selected.
*/
+ write_unlock_bh(&bond->lock);
bond_alb_deinit_slave(bond, slave);
+ write_lock_bh(&bond->lock);
}
if (oldcurrent == slave) {
@@ -1905,6 +1907,12 @@ static int bond_release_all(struct net_device *bond_dev)
slave_dev = slave->dev;
bond_detach_slave(bond, slave);
+ /* now that the slave is detached, unlock and perform
+ * all the undo steps that should not be called from
+ * within a lock.
+ */
+ write_unlock_bh(&bond->lock);
+
if ((bond->params.mode == BOND_MODE_TLB) ||
(bond->params.mode == BOND_MODE_ALB)) {
/* must be called only after the slave
@@ -1915,12 +1923,6 @@ static int bond_release_all(struct net_device *bond_dev)
bond_compute_features(bond);
- /* now that the slave is detached, unlock and perform
- * all the undo steps that should not be called from
- * within a lock.
- */
- write_unlock_bh(&bond->lock);
-
bond_destroy_slave_symlinks(bond_dev, slave_dev);
bond_del_vlans_from_slave(bond, slave_dev);
--
1.5.3.4.206.g58ba4-dirty
next prev parent reply other threads:[~2008-01-08 1:57 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-01-08 1:56 [PATCH 0/3] bonding: 3 fixes for 2.6.24 Jay Vosburgh
2008-01-08 1:56 ` [PATCH 1/3] bonding: fix locking in sysfs primary/active selection Jay Vosburgh
2008-01-08 1:56 ` [PATCH 2/3] bonding: fix ASSERT_RTNL that produces spurious warnings Jay Vosburgh
2008-01-08 1:57 ` Jay Vosburgh [this message]
2008-01-08 18:50 ` [PATCH 0/3] bonding: 3 fixes for 2.6.24 Krzysztof Oledzki
2008-01-08 19:17 ` Andy Gospodarek
2008-01-08 20:28 ` Jay Vosburgh
2008-01-09 6:08 ` Herbert Xu
2008-01-08 19:30 ` Jay Vosburgh
2008-01-09 6:35 ` Krzysztof Oledzki
2008-01-09 7:58 ` Jay Vosburgh
2008-01-09 9:36 ` Krzysztof Oledzki
2008-01-09 15:27 ` Andy Gospodarek
2008-01-09 17:54 ` Jay Vosburgh
2008-01-09 20:17 ` Andy Gospodarek
2008-01-09 22:05 ` Herbert Xu
2008-01-09 23:19 ` Jay Vosburgh
2008-01-10 0:58 ` Herbert Xu
2008-01-10 14:51 ` Andy Gospodarek
2008-01-10 20:36 ` Herbert Xu
2008-01-10 20:50 ` Jay Vosburgh
2008-01-10 21:03 ` Andy Gospodarek
2008-01-10 21:05 ` Herbert Xu
2008-01-11 1:06 ` Jay Vosburgh
2008-01-11 4:55 ` Herbert Xu
2008-01-10 20:45 ` Jay Vosburgh
2008-01-12 10:53 ` Krzysztof Oledzki
2008-01-12 17:56 ` Jay Vosburgh
2008-01-13 0:19 ` Herbert Xu
2008-01-14 22:15 ` Krzysztof Oledzki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=119975742458-git-send-email-fubar@us.ibm.com \
--to=fubar@us.ibm.com \
--cc=andy@greyhouse.net \
--cc=davem@davemloft.net \
--cc=jgarzik@pobox.com \
--cc=netdev@vger.kernel.org \
--cc=olel@ans.pl \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).