[BUG] net/ppp: A use after free in ppp_unregister_channe

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Each channel in a bridged pair holds a reference on the other.

Hence on return from ppp_unbridge_channels, the channel should not have
been freed (in this code path) because the ppp_register_net_channel
reference has not yet been dropped.

Maybe there is an issue with the reference counting or a race of some
sort?

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

-----原始邮件-----
发件人: "Tom Parkin" [off-list ref]
发送时间: 2021-03-12 18:12:58 (星期五)
收件人: lyl2019@mail.ustc.edu.cn
抄送: paulus@samba.org, davem@davemloft.net, linux-ppp@vger.kernel.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
主题: Re: [BUG] net/ppp: A use after free in ppp_unregister_channe

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Each channel in a bridged pair holds a reference on the other.

Hence on return from ppp_unbridge_channels, the channel should not have
been freed (in this code path) because the ppp_register_net_channel
reference has not yet been dropped.

Maybe there is an issue with the reference counting or a race of some
sort?

quoted

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

This issue was reported by a path-sensitive static analyzer developed by our Lab,
thus i have not a crash or bug log.

As the return type of ppp_unbridge_channels() is a int, can we return a value to
inform caller that the channel is freed?

On  Fri, Mar 12, 2021 at 22:47:53 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

-----原始邮件-----
发件人: "Tom Parkin" [off-list ref]
发送时间: 2021-03-12 18:12:58 (星期五)
收件人: lyl2019@mail.ustc.edu.cn
抄送: paulus@samba.org, davem@davemloft.net, linux-ppp@vger.kernel.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
主题: Re: [BUG] net/ppp: A use after free in ppp_unregister_channe

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Each channel in a bridged pair holds a reference on the other.

Hence on return from ppp_unbridge_channels, the channel should not have
been freed (in this code path) because the ppp_register_net_channel
reference has not yet been dropped.

Maybe there is an issue with the reference counting or a race of some
sort?

quoted

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

This issue was reported by a path-sensitive static analyzer developed by our Lab,
thus i have not a crash or bug log.

Does the analyzer report the sequence of events that lead to a
possible use-after-free?   Is it starting from the assumption that
ppp_unbridge_channels is called with pch->file.refcnt == 1?

As the return type of ppp_unbridge_channels() is a int, can we return a value to
inform caller that the channel is freed?

If ppp_unbridge_channels frees the channel in the
ppp_unregister_channel call path, that implies that
ppp_unregister_channel is called when only the bridge is keeping the
channel alive.  So the caller is attempting to drop a reference it
doesn't in fact have.

It might be preferable to defensively code against that possibility of
course.

On Thu, Mar 11, 2021 at 08:34:44PM +0800, lyl2019@mail.ustc.edu.cn wrote:

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

No. It's freed only if if the refcount drops to 0. And the caller of
ppp_unregister_channel() must hold its own refcount. So
ppp_unbridge_channels() is not going to drop the last refcount in this
case.

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

Next time, please also Cc. the author of the patch.

I have no idea about how to generate a suitable patch, sorry.

There's no patch to send as far as I can see.

On Fri, Mar 12, 2021 at 10:12:58AM +0000, Tom Parkin wrote:

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Agreed: ppp_unregister_channel() is going to drop a refcount from pch
in any case. So holding a refcount on pch is a hard requirement for any
caller of ppp_unregister_channel(), regardless of the channel bridging
code.

To lyl2019: Note that this refcount is (unsurprisingly) held by calling
ppp_register_net_channel().

On Fri, Mar 12, 2021 at 10:47:53PM +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

-----原始邮件-----
发件人: "Tom Parkin" [off-list ref]
发送时间: 2021-03-12 18:12:58 (星期五)
收件人: lyl2019@mail.ustc.edu.cn
抄送: paulus@samba.org, davem@davemloft.net, linux-ppp@vger.kernel.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
主题: Re: [BUG] net/ppp: A use after free in ppp_unregister_channe

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Each channel in a bridged pair holds a reference on the other.

Hence on return from ppp_unbridge_channels, the channel should not have
been freed (in this code path) because the ppp_register_net_channel
reference has not yet been dropped.

Maybe there is an issue with the reference counting or a race of some
sort?

quoted

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

This issue was reported by a path-sensitive static analyzer developed by our Lab,
thus i have not a crash or bug log.

As the return type of ppp_unbridge_channels() is a int, can we return a value to
inform caller that the channel is freed?

I don't think this is going to improve anything, as
ppp_unregister_channel() couldn't take any corrective action anyway.

On  Mon, Mar 15, 2021 at 13:18:24 +0100, Guillaume Nault wrote:

On Fri, Mar 12, 2021 at 10:47:53PM +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

quoted

-----原始邮件-----
发件人: "Tom Parkin" [off-list ref]
发送时间: 2021-03-12 18:12:58 (星期五)
收件人: lyl2019@mail.ustc.edu.cn
抄送: paulus@samba.org, davem@davemloft.net, linux-ppp@vger.kernel.org, netdev@vger.kernel.org, linux-kernel@vger.kernel.org
主题: Re: [BUG] net/ppp: A use after free in ppp_unregister_channe

Thanks for the report!

On  Thu, Mar 11, 2021 at 20:34:44 +0800, lyl2019@mail.ustc.edu.cn wrote:

quoted

File: drivers/net/ppp/ppp_generic.c

In ppp_unregister_channel, pch could be freed in ppp_unbridge_channels()
but after that pch is still in use. Inside the function ppp_unbridge_channels,
if "pchbb == pch" is true and then pch will be freed.

Do you have a way to reproduce a use-after-free scenario?

From static analysis I'm not sure how pch would be freed in
ppp_unbridge_channels when called via. ppp_unregister_channel.

In theory (at least!) the caller of ppp_register_net_channel holds 
a reference on struct channel which ppp_unregister_channel drops.

Each channel in a bridged pair holds a reference on the other.

Hence on return from ppp_unbridge_channels, the channel should not have
been freed (in this code path) because the ppp_register_net_channel
reference has not yet been dropped.

Maybe there is an issue with the reference counting or a race of some
sort?

quoted

I checked the commit history and found that this problem is introduced from
4cf476ced45d7 ("ppp: add PPPIOCBRIDGECHAN and PPPIOCUNBRIDGECHAN ioctls").

I have no idea about how to generate a suitable patch, sorry.

This issue was reported by a path-sensitive static analyzer developed by our Lab,
thus i have not a crash or bug log.

As the return type of ppp_unbridge_channels() is a int, can we return a value to
inform caller that the channel is freed?

I don't think this is going to improve anything, as
ppp_unregister_channel() couldn't take any corrective action anyway.

I agree with you to be honest.

I think the best ppp_unregister_channel could to is to not access pch
again.