netdev - [PATCH net 4/4] iavf: Fix deadlock in initialization

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 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]

Date:   Fri, 12 Aug 2022 10:23:09 -0700
From:   Tony Nguyen <anthony.l.nguyen@...el.com>
To:     davem@...emloft.net, kuba@...nel.org, pabeni@...hat.com,
        edumazet@...gle.com
Cc:     Ivan Vecera <ivecera@...hat.com>, netdev@...r.kernel.org,
        anthony.l.nguyen@...el.com, sassmann@...nic.de,
        Konrad Jankowski <konrad0.jankowski@...el.com>
Subject: [PATCH net 4/4] iavf: Fix deadlock in initialization

From: Ivan Vecera <ivecera@...hat.com>

Fix deadlock that occurs when iavf interface is a part of failover
configuration.

1. Mutex crit_lock is taken at the beginning of iavf_watchdog_task()
2. Function iavf_init_config_adapter() is called when adapter
   state is __IAVF_INIT_CONFIG_ADAPTER
3. iavf_init_config_adapter() calls register_netdevice() that emits
   NETDEV_REGISTER event
4. Notifier function failover_event() then calls
   net_failover_slave_register() that calls dev_open()
5. dev_open() calls iavf_open() that tries to take crit_lock in
   end-less loop

Stack trace:
...
[  790.251876]  usleep_range_state+0x5b/0x80
[  790.252547]  iavf_open+0x37/0x1d0 [iavf]
[  790.253139]  __dev_open+0xcd/0x160
[  790.253699]  dev_open+0x47/0x90
[  790.254323]  net_failover_slave_register+0x122/0x220 [net_failover]
[  790.255213]  failover_slave_register.part.7+0xd2/0x180 [failover]
[  790.256050]  failover_event+0x122/0x1ab [failover]
[  790.256821]  notifier_call_chain+0x47/0x70
[  790.257510]  register_netdevice+0x20f/0x550
[  790.258263]  iavf_watchdog_task+0x7c8/0xea0 [iavf]
[  790.259009]  process_one_work+0x1a7/0x360
[  790.259705]  worker_thread+0x30/0x390

To fix the situation we should check the current adapter state after
first unsuccessful mutex_trylock() and return with -EBUSY if it is
__IAVF_INIT_CONFIG_ADAPTER.

Fixes: 226d528512cf ("iavf: fix locking of critical sections")
Signed-off-by: Ivan Vecera <ivecera@...hat.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@...el.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@...el.com>
---
 drivers/net/ethernet/intel/iavf/iavf_main.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/iavf/iavf_main.c b/drivers/net/ethernet/intel/iavf/iavf_main.c
index 95d4348e7579..f39440ad5c50 100644
--- a/drivers/net/ethernet/intel/iavf/iavf_main.c
+++ b/drivers/net/ethernet/intel/iavf/iavf_main.c
@@ -4088,8 +4088,17 @@ static int iavf_open(struct net_device *netdev)
 		return -EIO;
 	}
 
-	while (!mutex_trylock(&adapter->crit_lock))
+	while (!mutex_trylock(&adapter->crit_lock)) {
+		/* If we are in __IAVF_INIT_CONFIG_ADAPTER state the crit_lock
+		 * is already taken and iavf_open is called from an upper
+		 * device's notifier reacting on NETDEV_REGISTER event.
+		 * We have to leave here to avoid dead lock.
+		 */
+		if (adapter->state == __IAVF_INIT_CONFIG_ADAPTER)
+			return -EBUSY;
+
 		usleep_range(500, 1000);
+	}
 
 	if (adapter->state != __IAVF_DOWN) {
 		err = -EBUSY;
-- 
2.35.1