[PATCH net-next 0/2] af_iucv updates

Hi Dave,
please apply two af_iucv patches for net-next.

Thanks,
Julian


Hans Wippel (1):
  net/iucv: improve endianness handling

Mateusz Jurczyk (1):
  af_iucv: Move sockaddr length checks to before accessing sa_family in
    bind and connect handlers

 net/iucv/af_iucv.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

-- 
2.11.2

From: Hans Wippel <redacted>

Use proper endianness conversion for an skb protocol assignment. Given
that IUCV is only available on big endian systems (s390), this simply
avoids an endianness warning reported by sparse.

Signed-off-by: Hans Wippel <redacted>
Reviewed-by: Julian Wiedmann <redacted>
Reviewed-by: Ursula Braun <redacted>
Signed-off-by: Julian Wiedmann <redacted>
---
 net/iucv/af_iucv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/iucv/af_iucv.c b/net/iucv/af_iucv.c
index 2cf9d59f1b72..05112094d76b 100644
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c

@@ -362,7 +362,7 @@ static int afiucv_hs_send(struct iucv_message *imsg, struct sock *sock,
 		else
 			skb_trim(skb, skb->dev->mtu);
 	}
-	skb->protocol = ETH_P_AF_IUCV;
+	skb->protocol = cpu_to_be16(ETH_P_AF_IUCV);
 	nskb = skb_clone(skb, GFP_ATOMIC);
 	if (!nskb)
 		return -ENOMEM;

-- 
2.11.2

From: Mateusz Jurczyk <redacted>

Verify that the caller-provided sockaddr structure is large enough to
contain the sa_family field, before accessing it in bind() and connect()
handlers of the AF_IUCV socket. Since neither syscall enforces a minimum
size of the corresponding memory region, very short sockaddrs (zero or
one byte long) result in operating on uninitialized memory while
referencing .sa_family.

Fixes: 52a82e23b9f2 ("af_iucv: Validate socket address length in iucv_sock_bind()")
Signed-off-by: Mateusz Jurczyk <redacted>
[jwi: removed unneeded null-check for addr]
Signed-off-by: Julian Wiedmann <redacted>
---
 net/iucv/af_iucv.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)

diff --git a/net/iucv/af_iucv.c b/net/iucv/af_iucv.c
index 05112094d76b..ac033e413bc5 100644
--- a/net/iucv/af_iucv.c
+++ b/net/iucv/af_iucv.c

@@ -715,10 +715,8 @@ static int iucv_sock_bind(struct socket *sock, struct sockaddr *addr,
 	char uid[9];
 
 	/* Verify the input sockaddr */
-	if (!addr || addr->sa_family != AF_IUCV)
-		return -EINVAL;
-
-	if (addr_len < sizeof(struct sockaddr_iucv))
+	if (addr_len < sizeof(struct sockaddr_iucv) ||
+	    addr->sa_family != AF_IUCV)
 		return -EINVAL;
 
 	lock_sock(sk);

@@ -862,7 +860,7 @@ static int iucv_sock_connect(struct socket *sock, struct sockaddr *addr,
 	struct iucv_sock *iucv = iucv_sk(sk);
 	int err;
 
-	if (addr->sa_family != AF_IUCV || alen < sizeof(struct sockaddr_iucv))
+	if (alen < sizeof(struct sockaddr_iucv) || addr->sa_family != AF_IUCV)
 		return -EINVAL;
 
 	if (sk->sk_state != IUCV_OPEN && sk->sk_state != IUCV_BOUND)

-- 
2.11.2

On 06/23/2017 07:32 PM, Julian Wiedmann wrote:

From: Mateusz Jurczyk <redacted>

Verify that the caller-provided sockaddr structure is large enough to
contain the sa_family field, before accessing it in bind() and connect()
handlers of the AF_IUCV socket. Since neither syscall enforces a minimum
size of the corresponding memory region, very short sockaddrs (zero or
one byte long) result in operating on uninitialized memory while
referencing .sa_family.

Won't it make sense to generally check the minimum length for .sa_family at a
single point before fixing all called sites?

Regards,
Oliver

From: Oliver Hartkopp <socketcan@hartkopp.net>
Date: Fri, 23 Jun 2017 19:36:12 +0200

On 06/23/2017 07:32 PM, Julian Wiedmann wrote:

quoted

From: Mateusz Jurczyk <redacted>

Verify that the caller-provided sockaddr structure is large enough to
contain the sa_family field, before accessing it in bind() and connect()
handlers of the AF_IUCV socket. Since neither syscall enforces a minimum
size of the corresponding memory region, very short sockaddrs (zero or
one byte long) result in operating on uninitialized memory while
referencing .sa_family.

Won't it make sense to generally check the minimum length for .sa_family at a
single point before fixing all called sites?

We had this discussion last week and we decided that putting it into
the handlers is the way to go for now.

From: Julian Wiedmann <redacted>
Date: Fri, 23 Jun 2017 19:32:27 +0200

From: Hans Wippel <redacted>

Use proper endianness conversion for an skb protocol assignment. Given
that IUCV is only available on big endian systems (s390), this simply
avoids an endianness warning reported by sparse.

Signed-off-by: Hans Wippel <redacted>
Reviewed-by: Julian Wiedmann <redacted>
Reviewed-by: Ursula Braun <redacted>
Signed-off-by: Julian Wiedmann <redacted>

Applied.

From: Julian Wiedmann <redacted>
Date: Fri, 23 Jun 2017 19:32:28 +0200

From: Mateusz Jurczyk <redacted>

Verify that the caller-provided sockaddr structure is large enough to
contain the sa_family field, before accessing it in bind() and connect()
handlers of the AF_IUCV socket. Since neither syscall enforces a minimum
size of the corresponding memory region, very short sockaddrs (zero or
one byte long) result in operating on uninitialized memory while
referencing .sa_family.

Fixes: 52a82e23b9f2 ("af_iucv: Validate socket address length in iucv_sock_bind()")
Signed-off-by: Mateusz Jurczyk <redacted>
[jwi: removed unneeded null-check for addr]
Signed-off-by: Julian Wiedmann <redacted>

Applied.