Patchwork [3.8-rc] tuntap: refuse to re-attach to different tun_struct

login
register
mail settings
Submitter Michael S. Tsirkin
Date Jan. 10, 2013, 10:23 a.m.
Message ID <20130110102319.GB13451@redhat.com>
Download mbox | patch
Permalink /patch/210970/
State RFC
Delegated to: David Miller
Headers show

Comments

Michael S. Tsirkin - Jan. 10, 2013, 10:23 a.m.
On Thu, Jan 10, 2013 at 08:59:48AM +0100, Stefan Hajnoczi wrote:
> Multiqueue tun devices support detaching a tun_file from its tun_struct
> and re-attaching at a later point in time.  This allows users to disable
> a specific queue temporarily.
> 
> ioctl(TUNSETIFF) allows the user to specify the network interface to
> attach by name.  This means the user can attempt to attach to interface
> "B" after detaching from interface "A".
> 
> The driver is not designed to support this so check we are re-attaching
> to the right tun_struct.  Failure to do so may lead to oops.
> 
> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> ---
> This fix is for 3.8-rc.
> 
>  drivers/net/tun.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index fbd106e..cf6da6e 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -491,6 +491,8 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
>  	err = -EINVAL;
>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
>  		goto out;
> +	if (tfile->detached && tun != tfile->detached)
> +		goto out;
>  
>  	err = -EBUSY;
>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
> -- 
> 1.8.0.2


I agree this is a bug but even with this patch, we still allow:

SETIFF
SETQUEUE (DISABLED)
SETIFF

Originally the rule always was that repeated setiff calls fail with
EINVAL. We broke that when we set tun to NULL.  It's probably worth
preserving that, even if queue is disabled.  Applying something like the below
instead will address this concern, won't it?

Note: works with regular userspace but I didn't test
multiqueue userspace. What do you think.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

---

--
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
Jason Wang - Jan. 10, 2013, 10:43 a.m.
On 01/10/2013 06:23 PM, Michael S. Tsirkin wrote:
> On Thu, Jan 10, 2013 at 08:59:48AM +0100, Stefan Hajnoczi wrote:
>> Multiqueue tun devices support detaching a tun_file from its tun_struct
>> and re-attaching at a later point in time.  This allows users to disable
>> a specific queue temporarily.
>>
>> ioctl(TUNSETIFF) allows the user to specify the network interface to
>> attach by name.  This means the user can attempt to attach to interface
>> "B" after detaching from interface "A".
>>
>> The driver is not designed to support this so check we are re-attaching
>> to the right tun_struct.  Failure to do so may lead to oops.
>>
>> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
>> ---
>> This fix is for 3.8-rc.
>>
>>  drivers/net/tun.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
>> index fbd106e..cf6da6e 100644
>> --- a/drivers/net/tun.c
>> +++ b/drivers/net/tun.c
>> @@ -491,6 +491,8 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
>>  	err = -EINVAL;
>>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
>>  		goto out;
>> +	if (tfile->detached && tun != tfile->detached)
>> +		goto out;
>>  
>>  	err = -EBUSY;
>>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
>> -- 
>> 1.8.0.2
>
> I agree this is a bug but even with this patch, we still allow:
>
> SETIFF
> SETQUEUE (DISABLED)
> SETIFF
>
> Originally the rule always was that repeated setiff calls fail with
> EINVAL. We broke that when we set tun to NULL.  It's probably worth
> preserving that, even if queue is disabled.  Applying something like the below
> instead will address this concern, won't it?
>
> Note: works with regular userspace but I didn't test
> multiqueue userspace. What do you think.

Or just fail when tun->detached is not NULL at the beginning of
tun_set_iff()?
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
>
> ---
>
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index fbd106e..5ec8b08 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -483,7 +483,7 @@ static void tun_detach_all(struct net_device *dev)
>  	BUG_ON(tun->numdisabled != 0);
>  }
>  
> -static int tun_attach(struct tun_struct *tun, struct file *file)
> +static int tun_attach(struct tun_struct *tun, struct file *file, bool setiff)
>  {
>  	struct tun_file *tfile = file->private_data;
>  	int err;
> @@ -492,6 +492,9 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
>  		goto out;
>  
> +	if (setiff && tfile->detached)
> +		goto out;
> +
>  	err = -EBUSY;
>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
>  		goto out;
> @@ -1561,7 +1564,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
>  		if (err < 0)
>  			return err;
>  
> -		err = tun_attach(tun, file);
> +		err = tun_attach(tun, file, true);
>  		if (err < 0)
>  			return err;
>  
> @@ -1627,7 +1630,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
>  		dev->features = dev->hw_features;
>  
>  		INIT_LIST_HEAD(&tun->disabled);
> -		err = tun_attach(tun, file);
> +		err = tun_attach(tun, file, true);
>  		if (err < 0)
>  			goto err_free_dev;
>  
> @@ -1792,7 +1795,7 @@ static int tun_set_queue(struct file *file, struct ifreq *ifr)
>  		else if (tun_not_capable(tun))
>  			ret = -EPERM;
>  		else
> -			ret = tun_attach(tun, file);
> +			ret = tun_attach(tun, file, false);
>  	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
>  		tun = rcu_dereference_protected(tfile->tun,
>  						lockdep_rtnl_is_held());
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

--
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
Michael S. Tsirkin - Jan. 10, 2013, 11:07 a.m.
On Thu, Jan 10, 2013 at 06:43:20PM +0800, Jason Wang wrote:
> On 01/10/2013 06:23 PM, Michael S. Tsirkin wrote:
> > On Thu, Jan 10, 2013 at 08:59:48AM +0100, Stefan Hajnoczi wrote:
> >> Multiqueue tun devices support detaching a tun_file from its tun_struct
> >> and re-attaching at a later point in time.  This allows users to disable
> >> a specific queue temporarily.
> >>
> >> ioctl(TUNSETIFF) allows the user to specify the network interface to
> >> attach by name.  This means the user can attempt to attach to interface
> >> "B" after detaching from interface "A".
> >>
> >> The driver is not designed to support this so check we are re-attaching
> >> to the right tun_struct.  Failure to do so may lead to oops.
> >>
> >> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> >> ---
> >> This fix is for 3.8-rc.
> >>
> >>  drivers/net/tun.c | 2 ++
> >>  1 file changed, 2 insertions(+)
> >>
> >> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> >> index fbd106e..cf6da6e 100644
> >> --- a/drivers/net/tun.c
> >> +++ b/drivers/net/tun.c
> >> @@ -491,6 +491,8 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
> >>  	err = -EINVAL;
> >>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
> >>  		goto out;
> >> +	if (tfile->detached && tun != tfile->detached)
> >> +		goto out;
> >>  
> >>  	err = -EBUSY;
> >>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
> >> -- 
> >> 1.8.0.2
> >
> > I agree this is a bug but even with this patch, we still allow:
> >
> > SETIFF
> > SETQUEUE (DISABLED)
> > SETIFF
> >
> > Originally the rule always was that repeated setiff calls fail with
> > EINVAL. We broke that when we set tun to NULL.  It's probably worth
> > preserving that, even if queue is disabled.  Applying something like the below
> > instead will address this concern, won't it?
> >
> > Note: works with regular userspace but I didn't test
> > multiqueue userspace. What do you think.
> 
> Or just fail when tun->detached is not NULL at the beginning of
> tun_set_iff()?

Yes that'll also work I think. If you prefer this pls send patch.

> > Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> >
> > ---
> >
> > diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> > index fbd106e..5ec8b08 100644
> > --- a/drivers/net/tun.c
> > +++ b/drivers/net/tun.c
> > @@ -483,7 +483,7 @@ static void tun_detach_all(struct net_device *dev)
> >  	BUG_ON(tun->numdisabled != 0);
> >  }
> >  
> > -static int tun_attach(struct tun_struct *tun, struct file *file)
> > +static int tun_attach(struct tun_struct *tun, struct file *file, bool setiff)
> >  {
> >  	struct tun_file *tfile = file->private_data;
> >  	int err;
> > @@ -492,6 +492,9 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
> >  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
> >  		goto out;
> >  
> > +	if (setiff && tfile->detached)
> > +		goto out;
> > +
> >  	err = -EBUSY;
> >  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
> >  		goto out;
> > @@ -1561,7 +1564,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
> >  		if (err < 0)
> >  			return err;
> >  
> > -		err = tun_attach(tun, file);
> > +		err = tun_attach(tun, file, true);
> >  		if (err < 0)
> >  			return err;
> >  
> > @@ -1627,7 +1630,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
> >  		dev->features = dev->hw_features;
> >  
> >  		INIT_LIST_HEAD(&tun->disabled);
> > -		err = tun_attach(tun, file);
> > +		err = tun_attach(tun, file, true);
> >  		if (err < 0)
> >  			goto err_free_dev;
> >  
> > @@ -1792,7 +1795,7 @@ static int tun_set_queue(struct file *file, struct ifreq *ifr)
> >  		else if (tun_not_capable(tun))
> >  			ret = -EPERM;
> >  		else
> > -			ret = tun_attach(tun, file);
> > +			ret = tun_attach(tun, file, false);
> >  	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
> >  		tun = rcu_dereference_protected(tfile->tun,
> >  						lockdep_rtnl_is_held());
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
--
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
Stefan Hajnoczi - Jan. 10, 2013, 11:53 a.m.
On Thu, Jan 10, 2013 at 12:23:19PM +0200, Michael S. Tsirkin wrote:
> On Thu, Jan 10, 2013 at 08:59:48AM +0100, Stefan Hajnoczi wrote:
> > Multiqueue tun devices support detaching a tun_file from its tun_struct
> > and re-attaching at a later point in time.  This allows users to disable
> > a specific queue temporarily.
> > 
> > ioctl(TUNSETIFF) allows the user to specify the network interface to
> > attach by name.  This means the user can attempt to attach to interface
> > "B" after detaching from interface "A".
> > 
> > The driver is not designed to support this so check we are re-attaching
> > to the right tun_struct.  Failure to do so may lead to oops.
> > 
> > Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
> > ---
> > This fix is for 3.8-rc.
> > 
> >  drivers/net/tun.c | 2 ++
> >  1 file changed, 2 insertions(+)
> > 
> > diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> > index fbd106e..cf6da6e 100644
> > --- a/drivers/net/tun.c
> > +++ b/drivers/net/tun.c
> > @@ -491,6 +491,8 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
> >  	err = -EINVAL;
> >  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
> >  		goto out;
> > +	if (tfile->detached && tun != tfile->detached)
> > +		goto out;
> >  
> >  	err = -EBUSY;
> >  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
> > -- 
> > 1.8.0.2
> 
> 
> I agree this is a bug but even with this patch, we still allow:
> 
> SETIFF
> SETQUEUE (DISABLED)
> SETIFF
> 
> Originally the rule always was that repeated setiff calls fail with
> EINVAL. We broke that when we set tun to NULL.  It's probably worth
> preserving that, even if queue is disabled.  Applying something like the below
> instead will address this concern, won't it?

Sounds good.

Stefan
--
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
Jason Wang - Jan. 10, 2013, 1:53 p.m.
On 01/10/2013 07:07 PM, Michael S. Tsirkin wrote:
> On Thu, Jan 10, 2013 at 06:43:20PM +0800, Jason Wang wrote:
>> On 01/10/2013 06:23 PM, Michael S. Tsirkin wrote:
>>> On Thu, Jan 10, 2013 at 08:59:48AM +0100, Stefan Hajnoczi wrote:
>>>> Multiqueue tun devices support detaching a tun_file from its tun_struct
>>>> and re-attaching at a later point in time.  This allows users to disable
>>>> a specific queue temporarily.
>>>>
>>>> ioctl(TUNSETIFF) allows the user to specify the network interface to
>>>> attach by name.  This means the user can attempt to attach to interface
>>>> "B" after detaching from interface "A".
>>>>
>>>> The driver is not designed to support this so check we are re-attaching
>>>> to the right tun_struct.  Failure to do so may lead to oops.
>>>>
>>>> Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
>>>> ---
>>>> This fix is for 3.8-rc.
>>>>
>>>>  drivers/net/tun.c | 2 ++
>>>>  1 file changed, 2 insertions(+)
>>>>
>>>> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
>>>> index fbd106e..cf6da6e 100644
>>>> --- a/drivers/net/tun.c
>>>> +++ b/drivers/net/tun.c
>>>> @@ -491,6 +491,8 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
>>>>  	err = -EINVAL;
>>>>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
>>>>  		goto out;
>>>> +	if (tfile->detached && tun != tfile->detached)
>>>> +		goto out;
>>>>  
>>>>  	err = -EBUSY;
>>>>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
>>>> -- 
>>>> 1.8.0.2
>>> I agree this is a bug but even with this patch, we still allow:
>>>
>>> SETIFF
>>> SETQUEUE (DISABLED)
>>> SETIFF
>>>
>>> Originally the rule always was that repeated setiff calls fail with
>>> EINVAL. We broke that when we set tun to NULL.  It's probably worth
>>> preserving that, even if queue is disabled.  Applying something like the below
>>> instead will address this concern, won't it?
>>>
>>> Note: works with regular userspace but I didn't test
>>> multiqueue userspace. What do you think.
>> Or just fail when tun->detached is not NULL at the beginning of
>> tun_set_iff()?
> Yes that'll also work I think. If you prefer this pls send patch.

Will post a patch, thanks.
>>> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
>>>
>>> ---
>>>
>>> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
>>> index fbd106e..5ec8b08 100644
>>> --- a/drivers/net/tun.c
>>> +++ b/drivers/net/tun.c
>>> @@ -483,7 +483,7 @@ static void tun_detach_all(struct net_device *dev)
>>>  	BUG_ON(tun->numdisabled != 0);
>>>  }
>>>  
>>> -static int tun_attach(struct tun_struct *tun, struct file *file)
>>> +static int tun_attach(struct tun_struct *tun, struct file *file, bool setiff)
>>>  {
>>>  	struct tun_file *tfile = file->private_data;
>>>  	int err;
>>> @@ -492,6 +492,9 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
>>>  	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
>>>  		goto out;
>>>  
>>> +	if (setiff && tfile->detached)
>>> +		goto out;
>>> +
>>>  	err = -EBUSY;
>>>  	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
>>>  		goto out;
>>> @@ -1561,7 +1564,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
>>>  		if (err < 0)
>>>  			return err;
>>>  
>>> -		err = tun_attach(tun, file);
>>> +		err = tun_attach(tun, file, true);
>>>  		if (err < 0)
>>>  			return err;
>>>  
>>> @@ -1627,7 +1630,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
>>>  		dev->features = dev->hw_features;
>>>  
>>>  		INIT_LIST_HEAD(&tun->disabled);
>>> -		err = tun_attach(tun, file);
>>> +		err = tun_attach(tun, file, true);
>>>  		if (err < 0)
>>>  			goto err_free_dev;
>>>  
>>> @@ -1792,7 +1795,7 @@ static int tun_set_queue(struct file *file, struct ifreq *ifr)
>>>  		else if (tun_not_capable(tun))
>>>  			ret = -EPERM;
>>>  		else
>>> -			ret = tun_attach(tun, file);
>>> +			ret = tun_attach(tun, file, false);
>>>  	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
>>>  		tun = rcu_dereference_protected(tfile->tun,
>>>  						lockdep_rtnl_is_held());
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>> Please read the FAQ at  http://www.tux.org/lkml/
> --
> 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

--
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

Patch

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index fbd106e..5ec8b08 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c
@@ -483,7 +483,7 @@  static void tun_detach_all(struct net_device *dev)
 	BUG_ON(tun->numdisabled != 0);
 }
 
-static int tun_attach(struct tun_struct *tun, struct file *file)
+static int tun_attach(struct tun_struct *tun, struct file *file, bool setiff)
 {
 	struct tun_file *tfile = file->private_data;
 	int err;
@@ -492,6 +492,9 @@  static int tun_attach(struct tun_struct *tun, struct file *file)
 	if (rcu_dereference_protected(tfile->tun, lockdep_rtnl_is_held()))
 		goto out;
 
+	if (setiff && tfile->detached)
+		goto out;
+
 	err = -EBUSY;
 	if (!(tun->flags & TUN_TAP_MQ) && tun->numqueues == 1)
 		goto out;
@@ -1561,7 +1564,7 @@  static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
 		if (err < 0)
 			return err;
 
-		err = tun_attach(tun, file);
+		err = tun_attach(tun, file, true);
 		if (err < 0)
 			return err;
 
@@ -1627,7 +1630,7 @@  static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
 		dev->features = dev->hw_features;
 
 		INIT_LIST_HEAD(&tun->disabled);
-		err = tun_attach(tun, file);
+		err = tun_attach(tun, file, true);
 		if (err < 0)
 			goto err_free_dev;
 
@@ -1792,7 +1795,7 @@  static int tun_set_queue(struct file *file, struct ifreq *ifr)
 		else if (tun_not_capable(tun))
 			ret = -EPERM;
 		else
-			ret = tun_attach(tun, file);
+			ret = tun_attach(tun, file, false);
 	} else if (ifr->ifr_flags & IFF_DETACH_QUEUE) {
 		tun = rcu_dereference_protected(tfile->tun,
 						lockdep_rtnl_is_held());