netdev - Re: [patch for 2.6.24? 1/1] bonding: locking fix

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: <27063.1200620733@death>
Date:	Thu, 17 Jan 2008 17:45:33 -0800
From:	Jay Vosburgh <fubar@...ibm.com>
To:	Krzysztof Oledzki <olel@....pl>
cc:	Andrew Morton <akpm@...ux-foundation.org>, davem@...emloft.net,
	jeff@...zik.org, shemminger@...ux-foundation.org,
	netdev@...r.kernel.org, Andy Gospodarek <andy@...yhouse.net>
Subject: Re: [patch for 2.6.24? 1/1] bonding: locking fix 

Krzysztof Oledzki <olel@....pl> wrote:

>> Andrew Morton <akpm@...ux-foundation.org> wrote:
>> [...]
>>> Can we get this bug fixed please?  Today?  It has been known about for more
>>> than two months.
>>
>> 	I just reposted the complete fix; it's #1 of the series of 7.
>
>Bad news. :( 2.6.24-rc7 + patch #1 (bonding: fix locking in sysfs
>primary/active selection):
[...]
>=========================================================
>[ INFO: possible irq lock inversion dependency detected ]
>2.6.24-rc7 #1
>---------------------------------------------------------
>events/0/9 just changed the state of lock:
> (&mc->mca_lock){-+..}, at: [<c041255a>] mld_ifc_timer_expire+0x130/0x1fb
>but this lock took another, soft-read-irq-unsafe lock in the past:
> (&bond->lock){-.--}

	None of the seven patches I posted just a bit ago will fix this
lockdep warning (which is a different thing that the bug Andrew inquired
about); I'm still working on that one.

	For that one, I had posted this work in progress patch:

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 77d004d..6906dbc 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -3937,8 +3937,6 @@ static void bond_set_multicast_list(struct net_device *bond_dev)
 	struct bonding *bond = bond_dev->priv;
 	struct dev_mc_list *dmi;
 
-	write_lock_bh(&bond->lock);
-
 	/*
 	 * Do promisc before checking multicast_mode
 	 */
@@ -3959,6 +3957,8 @@ static void bond_set_multicast_list(struct net_device *bond_dev)
 		bond_set_allmulti(bond, -1);
 	}
 
+	read_lock(&bond->lock);
+
 	bond->flags = bond_dev->flags;
 
 	/* looking for addresses to add to slaves' mc list */
@@ -3979,7 +3979,7 @@ static void bond_set_multicast_list(struct net_device *bond_dev)
 	bond_mc_list_destroy(bond);
 	bond_mc_list_copy(bond_dev->mc_list, bond, GFP_ATOMIC);
 
-	write_unlock_bh(&bond->lock);
+	read_unlock(&bond->lock);
 }
 
 /*

	which makes the warning go away, but Herbert Xu pointed out that
there is a potential problem with bond_enslave accessing the mc_lists
without sufficient locking.  It's not the only offender, either, and the
bond->mc_list references really need to be protected by the bond_lock,
and the whole thing probably ought to use dev_mc_sync/unsync instead of
what it does now.

	Since the bond_enslave, et al, business isn't a new problem, and
I've never heard of it being hit, I'm thinking now to just leave the
bond_enslave part for 2.6.25, and fix the lockdep warning for 2.6.24.

	But I wanted to make sure dropping the write lock to a read lock
wasn't going to make something worse.

	-J

---
	-Jay Vosburgh, IBM Linux Technology Center, fubar@...ibm.com
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html