When the link layer connection is broken, the rose->neighbour is
set to null. But rose->neighbour could be used by rose_connection()
and rose_release() later, because there is no synchronization among
them. As a result, the null-ptr-deref bugs will happen.
One of the null-ptr-deref bugs is shown below:
(thread 1) | (thread 2)
| rose_connect
rose_kill_by_neigh | lock_sock(sk)
spin_lock_bh(&rose_list_lock) | if (!rose->neighbour)
rose->neighbour = NULL;//(1) |
| rose->neighbour->use++;//(2)
The rose->neighbour is set to null in position (1) and dereferenced
in position (2).
The KASAN report triggered by POC is shown below:
KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
...
RIP: 0010:rose_connect+0x6c2/0xf30
RSP: 0018:ffff88800ab47d60 EFLAGS: 00000206
RAX: 0000000000000005 RBX: 000000000000002a RCX: 0000000000000000
RDX: ffff88800ab38000 RSI: ffff88800ab47e48 RDI: ffff88800ab38309
RBP: dffffc0000000000 R08: 0000000000000000 R09: ffffed1001567062
R10: dfffe91001567063 R11: 1ffff11001567061 R12: 1ffff11000d17cd0
R13: ffff8880068be680 R14: 0000000000000002 R15: 1ffff11000d17cd0
...
Call Trace:
<TASK>
? __local_bh_enable_ip+0x54/0x80
? selinux_netlbl_socket_connect+0x26/0x30
? rose_bind+0x5b0/0x5b0
__sys_connect+0x216/0x280
__x64_sys_connect+0x71/0x80
do_syscall_64+0x43/0x90
entry_SYSCALL_64_after_hwframe+0x46/0xb0
This patch adds lock_sock() in rose_kill_by_neigh() in order to
synchronize with rose_connect() and rose_release().
Meanwhile, this patch adds sock_hold() protected by rose_list_lock
that could synchronize with rose_remove_socket() in order to mitigate
UAF bug caused by lock_sock() we add.
What's more, there is no need using rose_neigh_list_lock to protect
rose_kill_by_neigh(). Because we have already used rose_neigh_list_lock
to protect the state change of rose_neigh in rose_link_failed(), which
is well synchronized.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <[email protected]>
---
Changes since v2:
- v2: Fix refcount leak of sock.
net/rose/af_rose.c | 6 ++++++
net/rose/rose_route.c | 2 ++
2 files changed, 8 insertions(+)
diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
index bf2d986a6bc..5caa222c490 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -169,9 +169,15 @@ void rose_kill_by_neigh(struct rose_neigh *neigh)
struct rose_sock *rose = rose_sk(s);
if (rose->neighbour == neigh) {
+ sock_hold(s);
rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
rose->neighbour->use--;
+ spin_unlock_bh(&rose_list_lock);
+ lock_sock(s);
rose->neighbour = NULL;
+ release_sock(s);
+ spin_lock_bh(&rose_list_lock);
+ sock_put(s);
}
}
spin_unlock_bh(&rose_list_lock);
diff --git a/net/rose/rose_route.c b/net/rose/rose_route.c
index fee6409c2bb..b116828b422 100644
--- a/net/rose/rose_route.c
+++ b/net/rose/rose_route.c
@@ -827,7 +827,9 @@ void rose_link_failed(ax25_cb *ax25, int reason)
ax25_cb_put(ax25);
rose_del_route_by_neigh(rose_neigh);
+ spin_unlock_bh(&rose_neigh_list_lock);
rose_kill_by_neigh(rose_neigh);
+ return;
}
spin_unlock_bh(&rose_neigh_list_lock);
}
--
2.17.1
On Fri, 2022-06-24 at 09:05 +0800, Duoming Zhou wrote:
> When the link layer connection is broken, the rose->neighbour is
> set to null. But rose->neighbour could be used by rose_connection()
> and rose_release() later, because there is no synchronization among
> them. As a result, the null-ptr-deref bugs will happen.
>
> One of the null-ptr-deref bugs is shown below:
>
> (thread 1) | (thread 2)
> | rose_connect
> rose_kill_by_neigh | lock_sock(sk)
> spin_lock_bh(&rose_list_lock) | if (!rose->neighbour)
> rose->neighbour = NULL;//(1) |
> | rose->neighbour->use++;//(2)
>
> The rose->neighbour is set to null in position (1) and dereferenced
> in position (2).
>
> The KASAN report triggered by POC is shown below:
>
> KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
> ...
> RIP: 0010:rose_connect+0x6c2/0xf30
> RSP: 0018:ffff88800ab47d60 EFLAGS: 00000206
> RAX: 0000000000000005 RBX: 000000000000002a RCX: 0000000000000000
> RDX: ffff88800ab38000 RSI: ffff88800ab47e48 RDI: ffff88800ab38309
> RBP: dffffc0000000000 R08: 0000000000000000 R09: ffffed1001567062
> R10: dfffe91001567063 R11: 1ffff11001567061 R12: 1ffff11000d17cd0
> R13: ffff8880068be680 R14: 0000000000000002 R15: 1ffff11000d17cd0
> ...
> Call Trace:
> <TASK>
> ? __local_bh_enable_ip+0x54/0x80
> ? selinux_netlbl_socket_connect+0x26/0x30
> ? rose_bind+0x5b0/0x5b0
> __sys_connect+0x216/0x280
> __x64_sys_connect+0x71/0x80
> do_syscall_64+0x43/0x90
> entry_SYSCALL_64_after_hwframe+0x46/0xb0
>
> This patch adds lock_sock() in rose_kill_by_neigh() in order to
> synchronize with rose_connect() and rose_release().
>
> Meanwhile, this patch adds sock_hold() protected by rose_list_lock
> that could synchronize with rose_remove_socket() in order to mitigate
> UAF bug caused by lock_sock() we add.
>
> What's more, there is no need using rose_neigh_list_lock to protect
> rose_kill_by_neigh(). Because we have already used rose_neigh_list_lock
> to protect the state change of rose_neigh in rose_link_failed(), which
> is well synchronized.
>
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Signed-off-by: Duoming Zhou <[email protected]>
> ---
> Changes since v2:
> - v2: Fix refcount leak of sock.
>
> net/rose/af_rose.c | 6 ++++++
> net/rose/rose_route.c | 2 ++
> 2 files changed, 8 insertions(+)
>
> diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
> index bf2d986a6bc..5caa222c490 100644
> --- a/net/rose/af_rose.c
> +++ b/net/rose/af_rose.c
> @@ -169,9 +169,15 @@ void rose_kill_by_neigh(struct rose_neigh *neigh)
> struct rose_sock *rose = rose_sk(s);
>
> if (rose->neighbour == neigh) {
> + sock_hold(s);
> rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
> rose->neighbour->use--;
> + spin_unlock_bh(&rose_list_lock);
> + lock_sock(s);
> rose->neighbour = NULL;
> + release_sock(s);
> + spin_lock_bh(&rose_list_lock);
I'm sorry, I likely was not clear enough in my previous reply. This is
broken. If a list is [spin_]lock protected, you can't release the lock,
reacquire it and continue traversing the list from the [now invalid]
same iterator.
e.g. if s is removed from the list, even if the sock is not de-
allocated due to the addtional refcount, the traversing will errnously
stop after this sock, instead of continuing processing the remaining
socks in the list.
A possible alternative, not even build-tested would be:
---
diff --git a/include/net/rose.h b/include/net/rose.h
index 0f0a4ce0fee7..090db11d528f 100644
--- a/include/net/rose.h
+++ b/include/net/rose.h
@@ -145,6 +145,7 @@ struct rose_sock {
struct rose_facilities_struct facilities;
struct timer_list timer;
struct timer_list idletimer;
+ struct rose_sock *dl_next;
};
#define rose_sk(sk) ((struct rose_sock *)(sk))
diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
index 5caa222c490e..01f3c50f0921 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -162,25 +162,32 @@ static void rose_remove_socket(struct sock *sk)
*/
void rose_kill_by_neigh(struct rose_neigh *neigh)
{
- struct sock *s;
+ struct rose_sock *del_list = NULL;
+ struct sock *s, *tmp;
spin_lock_bh(&rose_list_lock);
- sk_for_each(s, &rose_list) {
+ sk_for_each_safe(s, tmp, &rose_list) {
struct rose_sock *rose = rose_sk(s);
if (rose->neighbour == neigh) {
- sock_hold(s);
- rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
- rose->neighbour->use--;
- spin_unlock_bh(&rose_list_lock);
- lock_sock(s);
- rose->neighbour = NULL;
- release_sock(s);
- spin_lock_bh(&rose_list_lock);
- sock_put(s);
+ __sk_del_node(s);
+ s->dl_next = del_list;
+ del_list = s;
}
}
spin_unlock_bh(&rose_list_lock);
+
+ while (del_list) {
+ s = del_list;
+ del_list = s->dl_next;
+
+ lock_sock(s);
+ rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
+ rose->neighbour->use--;
+ rose->neighbour = NULL;
+ release_sock(s);
+ sock_put(s);
+ }
}
/*
---
Paolo
Hello,
On Tue, 28 Jun 2022 13:12:40 +0200 Paolo Abeni wrote:
> > When the link layer connection is broken, the rose->neighbour is
> > set to null. But rose->neighbour could be used by rose_connection()
> > and rose_release() later, because there is no synchronization among
> > them. As a result, the null-ptr-deref bugs will happen.
> >
> > One of the null-ptr-deref bugs is shown below:
> >
> > (thread 1) | (thread 2)
> > | rose_connect
> > rose_kill_by_neigh | lock_sock(sk)
> > spin_lock_bh(&rose_list_lock) | if (!rose->neighbour)
> > rose->neighbour = NULL;//(1) |
> > | rose->neighbour->use++;//(2)
> >
> > The rose->neighbour is set to null in position (1) and dereferenced
> > in position (2).
> >
> > The KASAN report triggered by POC is shown below:
> >
> > KASAN: null-ptr-deref in range [0x0000000000000028-0x000000000000002f]
> > ...
> > RIP: 0010:rose_connect+0x6c2/0xf30
> > RSP: 0018:ffff88800ab47d60 EFLAGS: 00000206
> > RAX: 0000000000000005 RBX: 000000000000002a RCX: 0000000000000000
> > RDX: ffff88800ab38000 RSI: ffff88800ab47e48 RDI: ffff88800ab38309
> > RBP: dffffc0000000000 R08: 0000000000000000 R09: ffffed1001567062
> > R10: dfffe91001567063 R11: 1ffff11001567061 R12: 1ffff11000d17cd0
> > R13: ffff8880068be680 R14: 0000000000000002 R15: 1ffff11000d17cd0
> > ...
> > Call Trace:
> > <TASK>
> > ? __local_bh_enable_ip+0x54/0x80
> > ? selinux_netlbl_socket_connect+0x26/0x30
> > ? rose_bind+0x5b0/0x5b0
> > __sys_connect+0x216/0x280
> > __x64_sys_connect+0x71/0x80
> > do_syscall_64+0x43/0x90
> > entry_SYSCALL_64_after_hwframe+0x46/0xb0
> >
> > This patch adds lock_sock() in rose_kill_by_neigh() in order to
> > synchronize with rose_connect() and rose_release().
> >
> > Meanwhile, this patch adds sock_hold() protected by rose_list_lock
> > that could synchronize with rose_remove_socket() in order to mitigate
> > UAF bug caused by lock_sock() we add.
> >
> > What's more, there is no need using rose_neigh_list_lock to protect
> > rose_kill_by_neigh(). Because we have already used rose_neigh_list_lock
> > to protect the state change of rose_neigh in rose_link_failed(), which
> > is well synchronized.
> >
> > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > Signed-off-by: Duoming Zhou <[email protected]>
> > ---
> > Changes since v2:
> > - v2: Fix refcount leak of sock.
> >
> > net/rose/af_rose.c | 6 ++++++
> > net/rose/rose_route.c | 2 ++
> > 2 files changed, 8 insertions(+)
> >
> > diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
> > index bf2d986a6bc..5caa222c490 100644
> > --- a/net/rose/af_rose.c
> > +++ b/net/rose/af_rose.c
> > @@ -169,9 +169,15 @@ void rose_kill_by_neigh(struct rose_neigh *neigh)
> > struct rose_sock *rose = rose_sk(s);
> >
> > if (rose->neighbour == neigh) {
> > + sock_hold(s);
> > rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
> > rose->neighbour->use--;
> > + spin_unlock_bh(&rose_list_lock);
> > + lock_sock(s);
> > rose->neighbour = NULL;
> > + release_sock(s);
> > + spin_lock_bh(&rose_list_lock);
>
> I'm sorry, I likely was not clear enough in my previous reply. This is
> broken. If a list is [spin_]lock protected, you can't release the lock,
> reacquire it and continue traversing the list from the [now invalid]
> same iterator.
>
> e.g. if s is removed from the list, even if the sock is not de-
> allocated due to the addtional refcount, the traversing will errnously
> stop after this sock, instead of continuing processing the remaining
> socks in the list.
I understand. The following is a new solution:
diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
index bf2d986a6bc..24dcbde88fb 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -165,13 +165,21 @@ void rose_kill_by_neigh(struct rose_neigh *neigh)
struct sock *s;
spin_lock_bh(&rose_list_lock);
+again:
sk_for_each(s, &rose_list) {
struct rose_sock *rose = rose_sk(s);
if (rose->neighbour == neigh) {
+ sock_hold(s);
+ spin_unlock_bh(&rose_list_lock);
+ lock_sock(s);
rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
rose->neighbour->use--;
rose->neighbour = NULL;
+ release_sock(s);
+ spin_lock_bh(&rose_list_lock);
+ sock_put(s);
+ goto again;
}
}
spin_unlock_bh(&rose_list_lock);
diff --git a/net/rose/rose_route.c b/net/rose/rose_route.c
index fee6409c2bb..b116828b422 100644
--- a/net/rose/rose_route.c
+++ b/net/rose/rose_route.c
@@ -827,7 +827,9 @@ void rose_link_failed(ax25_cb *ax25, int reason)
ax25_cb_put(ax25);
rose_del_route_by_neigh(rose_neigh);
+ spin_unlock_bh(&rose_neigh_list_lock);
rose_kill_by_neigh(rose_neigh);
+ return;
}
spin_unlock_bh(&rose_neigh_list_lock);
}
If s is removed from the list, the traversing will not stop erroneously.
Best regards,
Duoming Zhou
On Tue, Jun 28, 2022 at 11:59 PM <[email protected]> wrote:
> Hello,
>
> On Tue, 28 Jun 2022 13:12:40 +0200 Paolo Abeni wrote:
> > [snip]
> > I'm sorry, I likely was not clear enough in my previous reply. This is
> > broken. If a list is [spin_]lock protected, you can't release the lock,
> > reacquire it and continue traversing the list from the [now invalid]
> > same iterator.
> >
> > e.g. if s is removed from the list, even if the sock is not de-
> > allocated due to the addtional refcount, the traversing will errnously
> > stop after this sock, instead of continuing processing the remaining
> > socks in the list.
>
> I understand. The following is a new solution:
>
> diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
> index bf2d986a6bc..24dcbde88fb 100644
> --- a/net/rose/af_rose.c
> +++ b/net/rose/af_rose.c
> @@ -165,13 +165,21 @@ void rose_kill_by_neigh(struct rose_neigh *neigh)
> struct sock *s;
>
> spin_lock_bh(&rose_list_lock);
> +again:
> sk_for_each(s, &rose_list) {
> struct rose_sock *rose = rose_sk(s);
>
> if (rose->neighbour == neigh) {
> + sock_hold(s);
> + spin_unlock_bh(&rose_list_lock);
> + lock_sock(s);
> rose_disconnect(s, ENETUNREACH, ROSE_OUT_OF_ORDER, 0);
> rose->neighbour->use--;
> rose->neighbour = NULL;
> + release_sock(s);
> + spin_lock_bh(&rose_list_lock);
> + sock_put(s);
> + goto again;
It may be worthwhile noting that this changes the time complexity
of the algorithm to be O(n^2) in the number of entries in `rose_list`,
instead of linear. But as that number is extremely unlikely to ever
be large, it probably makes no practical difference.
- Dan C.
> }
> }
> spin_unlock_bh(&rose_list_lock);
> diff --git a/net/rose/rose_route.c b/net/rose/rose_route.c
> index fee6409c2bb..b116828b422 100644
> --- a/net/rose/rose_route.c
> +++ b/net/rose/rose_route.c
> @@ -827,7 +827,9 @@ void rose_link_failed(ax25_cb *ax25, int reason)
> ax25_cb_put(ax25);
>
> rose_del_route_by_neigh(rose_neigh);
> + spin_unlock_bh(&rose_neigh_list_lock);
> rose_kill_by_neigh(rose_neigh);
> + return;
> }
> spin_unlock_bh(&rose_neigh_list_lock);
> }
>
> If s is removed from the list, the traversing will not stop erroneously.
>
> Best regards,
> Duoming Zhou