netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] bonding: bond_slave_info_query() fix
@ 2009-04-23 11:24 Eric Dumazet
  2009-04-23 12:11 ` Jiri Pirko
  2009-05-01 22:16 ` David Miller
  0 siblings, 2 replies; 6+ messages in thread
From: Eric Dumazet @ 2009-04-23 11:24 UTC (permalink / raw)
  To: David S. Miller; +Cc: Linux Netdev List, Jay Vosburgh

bond_slave_info_query() should keep a read lock while accessing slave info,
or risk accessing stale data and corruption.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 63369b6..66697db 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -2213,33 +2213,27 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
 {
 	struct bonding *bond = netdev_priv(bond_dev);
 	struct slave *slave;
-	int i, found = 0;
+	int i, ret = -ENODEV;
 
-	if (info->slave_id < 0) {
+	if (info->slave_id < 0)
 		return -ENODEV;
-	}
 
 	read_lock(&bond->lock);
 
 	bond_for_each_slave(bond, slave, i) {
 		if (i == (int)info->slave_id) {
-			found = 1;
+			ret = 0;
+			strcpy(info->slave_name, slave->dev->name);
+			info->link = slave->link;
+			info->state = slave->state;
+			info->link_failure_count = slave->link_failure_count;
 			break;
 		}
 	}
 
 	read_unlock(&bond->lock);
 
-	if (found) {
-		strcpy(info->slave_name, slave->dev->name);
-		info->link = slave->link;
-		info->state = slave->state;
-		info->link_failure_count = slave->link_failure_count;
-	} else {
-		return -ENODEV;
-	}
-
-	return 0;
+	return ret;
 }
 
 /*-------------------------------- Monitoring -------------------------------*/

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH] bonding: bond_slave_info_query() fix
  2009-04-23 11:24 [PATCH] bonding: bond_slave_info_query() fix Eric Dumazet
@ 2009-04-23 12:11 ` Jiri Pirko
  2009-04-23 13:39   ` Eric Dumazet
  2009-05-01 22:16 ` David Miller
  1 sibling, 1 reply; 6+ messages in thread
From: Jiri Pirko @ 2009-04-23 12:11 UTC (permalink / raw)
  To: Eric Dumazet; +Cc: David S. Miller, Linux Netdev List, Jay Vosburgh

Thu, Apr 23, 2009 at 01:24:07PM CEST, dada1@cosmosbay.com wrote:
>bond_slave_info_query() should keep a read lock while accessing slave info,
>or risk accessing stale data and corruption.
>
>Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
>
>diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>index 63369b6..66697db 100644
>--- a/drivers/net/bonding/bond_main.c
>+++ b/drivers/net/bonding/bond_main.c
>@@ -2213,33 +2213,27 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
> {
> 	struct bonding *bond = netdev_priv(bond_dev);
> 	struct slave *slave;
>-	int i, found = 0;
>+	int i, ret = -ENODEV;
> 
>-	if (info->slave_id < 0) {
>+	if (info->slave_id < 0)
> 		return -ENODEV;

You can return ret; here since -ENODEV is there, but I don't know if it isn't
against policy. Patch looks good anyway.
>-	}
> 
> 	read_lock(&bond->lock);
> 
> 	bond_for_each_slave(bond, slave, i) {
> 		if (i == (int)info->slave_id) {
>-			found = 1;
>+			ret = 0;
>+			strcpy(info->slave_name, slave->dev->name);
>+			info->link = slave->link;
>+			info->state = slave->state;
>+			info->link_failure_count = slave->link_failure_count;
> 			break;
> 		}
> 	}
> 
> 	read_unlock(&bond->lock);
> 
>-	if (found) {
>-		strcpy(info->slave_name, slave->dev->name);
>-		info->link = slave->link;
>-		info->state = slave->state;
>-		info->link_failure_count = slave->link_failure_count;
>-	} else {
>-		return -ENODEV;
>-	}
>-
>-	return 0;
>+	return ret;
> }
> 
> /*-------------------------------- Monitoring -------------------------------*/
>--
>To unsubscribe from this list: send the line "unsubscribe netdev" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] bonding: bond_slave_info_query() fix
  2009-04-23 12:11 ` Jiri Pirko
@ 2009-04-23 13:39   ` Eric Dumazet
  2009-04-23 14:51     ` Jay Vosburgh
  0 siblings, 1 reply; 6+ messages in thread
From: Eric Dumazet @ 2009-04-23 13:39 UTC (permalink / raw)
  To: Jiri Pirko; +Cc: David S. Miller, Linux Netdev List, Jay Vosburgh

Jiri Pirko a écrit :
> Thu, Apr 23, 2009 at 01:24:07PM CEST, dada1@cosmosbay.com wrote:
>> bond_slave_info_query() should keep a read lock while accessing slave info,
>> or risk accessing stale data and corruption.
>>
>> Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
>>
>> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>> index 63369b6..66697db 100644
>> --- a/drivers/net/bonding/bond_main.c
>> +++ b/drivers/net/bonding/bond_main.c
>> @@ -2213,33 +2213,27 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
>> {
>> 	struct bonding *bond = netdev_priv(bond_dev);
>> 	struct slave *slave;
>> -	int i, found = 0;
>> +	int i, ret = -ENODEV;
>>
>> -	if (info->slave_id < 0) {
>> +	if (info->slave_id < 0)
>> 		return -ENODEV;
> 
> You can return ret; here since -ENODEV is there, but I don't know if it isn't
> against policy. Patch looks good anyway.
>

Yes, we can just delete this extra test, since requested slave_id wont
be found anyway if negative.

Thanks

[PATCH] bonding: bond_slave_info_query() fix

bond_slave_info_query() should keep a read lock while accessing slave info,
or risk accessing stale data and corruption.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 63369b6..67515b7 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -2213,33 +2213,24 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
 {
 	struct bonding *bond = netdev_priv(bond_dev);
 	struct slave *slave;
-	int i, found = 0;
-
-	if (info->slave_id < 0) {
-		return -ENODEV;
-	}
+	int i, res = -ENODEV;
 
 	read_lock(&bond->lock);
 
 	bond_for_each_slave(bond, slave, i) {
 		if (i == (int)info->slave_id) {
-			found = 1;
+			res = 0;
+			strcpy(info->slave_name, slave->dev->name);
+			info->link = slave->link;
+			info->state = slave->state;
+			info->link_failure_count = slave->link_failure_count;
 			break;
 		}
 	}
 
 	read_unlock(&bond->lock);
 
-	if (found) {
-		strcpy(info->slave_name, slave->dev->name);
-		info->link = slave->link;
-		info->state = slave->state;
-		info->link_failure_count = slave->link_failure_count;
-	} else {
-		return -ENODEV;
-	}
-
-	return 0;
+	return res;
 }
 
 /*-------------------------------- Monitoring -------------------------------*/


^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH] bonding: bond_slave_info_query() fix
  2009-04-23 13:39   ` Eric Dumazet
@ 2009-04-23 14:51     ` Jay Vosburgh
  2009-05-01 22:17       ` David Miller
  0 siblings, 1 reply; 6+ messages in thread
From: Jay Vosburgh @ 2009-04-23 14:51 UTC (permalink / raw)
  To: Eric Dumazet; +Cc: Jiri Pirko, David S. Miller, Linux Netdev List

Eric Dumazet <dada1@cosmosbay.com> wrote:

>Jiri Pirko a écrit :
>> Thu, Apr 23, 2009 at 01:24:07PM CEST, dada1@cosmosbay.com wrote:
>>> bond_slave_info_query() should keep a read lock while accessing slave info,
>>> or risk accessing stale data and corruption.
>>>
>>> Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
>>>
>>> diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>>> index 63369b6..66697db 100644
>>> --- a/drivers/net/bonding/bond_main.c
>>> +++ b/drivers/net/bonding/bond_main.c
>>> @@ -2213,33 +2213,27 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
>>> {
>>> 	struct bonding *bond = netdev_priv(bond_dev);
>>> 	struct slave *slave;
>>> -	int i, found = 0;
>>> +	int i, ret = -ENODEV;
>>>
>>> -	if (info->slave_id < 0) {
>>> +	if (info->slave_id < 0)
>>> 		return -ENODEV;
>> 
>> You can return ret; here since -ENODEV is there, but I don't know if it isn't
>> against policy. Patch looks good anyway.
>>
>
>Yes, we can just delete this extra test, since requested slave_id wont
>be found anyway if negative.
>
>Thanks
>
>[PATCH] bonding: bond_slave_info_query() fix
>
>bond_slave_info_query() should keep a read lock while accessing slave info,
>or risk accessing stale data and corruption.

	I don't think that's actually true, since changes to the data
being referenced are (should be) protected by RTNL, and
bond_slave_info_query runs via ioctl, and is holding RTNL.

	Nevertheless, the patch does make the function cleaner, so I
have no objections.

	-J

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>

>Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
>
>diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
>index 63369b6..67515b7 100644
>--- a/drivers/net/bonding/bond_main.c
>+++ b/drivers/net/bonding/bond_main.c
>@@ -2213,33 +2213,24 @@ static int bond_slave_info_query(struct net_device *bond_dev, struct ifslave *in
> {
> 	struct bonding *bond = netdev_priv(bond_dev);
> 	struct slave *slave;
>-	int i, found = 0;
>-
>-	if (info->slave_id < 0) {
>-		return -ENODEV;
>-	}
>+	int i, res = -ENODEV;
>
> 	read_lock(&bond->lock);
>
> 	bond_for_each_slave(bond, slave, i) {
> 		if (i == (int)info->slave_id) {
>-			found = 1;
>+			res = 0;
>+			strcpy(info->slave_name, slave->dev->name);
>+			info->link = slave->link;
>+			info->state = slave->state;
>+			info->link_failure_count = slave->link_failure_count;
> 			break;
> 		}
> 	}
>
> 	read_unlock(&bond->lock);
>
>-	if (found) {
>-		strcpy(info->slave_name, slave->dev->name);
>-		info->link = slave->link;
>-		info->state = slave->state;
>-		info->link_failure_count = slave->link_failure_count;
>-	} else {
>-		return -ENODEV;
>-	}
>-
>-	return 0;
>+	return res;
> }
>
> /*-------------------------------- Monitoring -------------------------------*/
>
>--
>To unsubscribe from this list: send the line "unsubscribe netdev" in
>the body of a message to majordomo@vger.kernel.org
>More majordomo info at  http://vger.kernel.org/majordomo-info.html

---
	-Jay Vosburgh, IBM Linux Technology Center, fubar@us.ibm.com

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] bonding: bond_slave_info_query() fix
  2009-04-23 11:24 [PATCH] bonding: bond_slave_info_query() fix Eric Dumazet
  2009-04-23 12:11 ` Jiri Pirko
@ 2009-05-01 22:16 ` David Miller
  1 sibling, 0 replies; 6+ messages in thread
From: David Miller @ 2009-05-01 22:16 UTC (permalink / raw)
  To: dada1; +Cc: netdev, fubar

From: Eric Dumazet <dada1@cosmosbay.com>
Date: Thu, 23 Apr 2009 13:24:07 +0200

> bond_slave_info_query() should keep a read lock while accessing slave info,
> or risk accessing stale data and corruption.
> 
> Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>

Applied, thanks Eric.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] bonding: bond_slave_info_query() fix
  2009-04-23 14:51     ` Jay Vosburgh
@ 2009-05-01 22:17       ` David Miller
  0 siblings, 0 replies; 6+ messages in thread
From: David Miller @ 2009-05-01 22:17 UTC (permalink / raw)
  To: fubar; +Cc: dada1, jpirko, netdev

From: Jay Vosburgh <fubar@us.ibm.com>
Date: Thu, 23 Apr 2009 07:51:36 -0700

> Eric Dumazet <dada1@cosmosbay.com> wrote:
> 
> Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
> 
>>Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>

Aha, I'll apply this updated version instead, ignore my
other email.

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2009-05-01 22:17 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-04-23 11:24 [PATCH] bonding: bond_slave_info_query() fix Eric Dumazet
2009-04-23 12:11 ` Jiri Pirko
2009-04-23 13:39   ` Eric Dumazet
2009-04-23 14:51     ` Jay Vosburgh
2009-05-01 22:17       ` David Miller
2009-05-01 22:16 ` David Miller

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).