Bug 684643

Summary:	WARNING: at drivers/net/bonding/bond_main.c:5511 bonding_exit()
Product:	Red Hat Enterprise Linux 5	Reporter:	Mark Wu <dwu>
Component:	kernel	Assignee:	Andy Gospodarek <agospoda>
Status:	CLOSED DUPLICATE	QA Contact:	Red Hat Kernel QE team <kernel-qe>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	5.6	CC:	jarod, moshiro, peterm
Target Milestone:	rc
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2011-07-13 16:18:57 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Mark Wu 2011-03-14 05:02:56 UTC

Description of problem:

When we remove the driver using rmmod, the stack trace is given like this.
Mar 11 11:18:12 build kernel: WARNING: at drivers/net/bonding/bond_main.c:5511 bonding_exit()
Mar 11 11:18:12 build kernel: Call Trace:
Mar 11 11:18:12 build avahi-daemon[4651]: Withdrawing address record for
192.168.1.2 on bond5.
Mar 11 11:18:12 build kernel:  [<ffffffff800a8617>]
sys_delete_module+0x196/0x1c5
Mar 11 11:18:12 build kernel:  [<ffffffff8005d28d>] tracesys+0xd5/0xe0 Mar 11 11:18:12 build kernel:

This problem should be caused by only unblock operation is added in functions bonding_store_active_slave and bonding_store_primary. So, if module is removed after these operations , the work counter stays minus and the inconsistency appears.


Version-Release number of selected component (if applicable):
2.6.18-238.6.1.el5 

How reproducible:

Steps to Reproduce:
rmmod bonding
  
Actual results:
Get warning on: WARN_ON(atomic_read(&netpoll_block_tx));

Expected results:
No warning when removed the driver

Additional info:

Comment 2 Mark Wu 2011-03-14 05:07:19 UTC

It looks it need add block_netpoll_tx in functions bonding_store_active_slave and bonding_store_primary too

Comment 4 Mark Wu 2011-03-15 07:48:52 UTC

Oshiro-san,
The test kernel is available at:
https://brewweb.devel.redhat.com/taskinfo?taskID=3178244

The test patch is:

diff --git a/drivers/net/bonding/bond_sysfs.c b/drivers/net/bonding/bond_sysfs.c
index 6ffcdc6..37868bc 100644
--- a/drivers/net/bonding/bond_sysfs.c
+++ b/drivers/net/bonding/bond_sysfs.c
@@ -1097,6 +1097,7 @@ static ssize_t bonding_store_primary(struct class_device *cd, const char *buf, s
        struct bonding *bond = to_bond(cd);

        rtnl_lock();
+       block_netpoll_tx();
        read_lock(&bond->lock);
        write_lock_bh(&bond->curr_slave_lock);

@@ -1258,6 +1259,7 @@ static ssize_t bonding_store_active_slave(struct class_device *cd, const char *b
        struct bonding *bond = to_bond(cd);

        rtnl_lock();
+       block_netpoll_tx();
        read_lock(&bond->lock);
        write_lock_bh(&bond->curr_slave_lock);

Comment 6 Andy Gospodarek 2011-07-13 16:18:57 UTC


*** This bug has been marked as a duplicate of bug 704426 ***

Comment 7 Andy Gospodarek 2011-07-13 16:19:51 UTC

This is fixed in 2.6.18-266.el5.