[PATCH 00/18] net subsystem various refcounter conversions

[PATCH 00/18] net subsystem various refcounter conversions

[Elena Reshetova]

This series, for various network subsystem components, replaces atomic_t reference
counters with the new refcount_t type and API (see include/linux/refcount.h).
By doing this we prevent intentional or accidental
underflows or overflows that can led to use-after-free vulnerabilities.

The patches are fully independent and can be cherry-picked separately.
Since we convert all kernel subsystems in the same fashion, resulting
in about 300 patches, we have to group them for sending at least in some
fashion to be manageable. Please excuse the long cc list.

If there are no objections to the patches, please merge them via respective trees.

Elena Reshetova (18):
  net, llc: convert llc_sap.refcnt from atomic_t to refcount_t
  net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t
  net, l2tp: convert l2tp_session.ref_count from atomic_t to refcount_t
  net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t
  net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t
  net, decnet: convert dn_fib_info.fib_clntref from atomic_t to
    refcount_t
  net, atm: convert atm_dev.refcnt from atomic_t to refcount_t
  net, atm: convert lec_arp_table.usage from atomic_t to refcount_t
  net, atm: convert in_cache_entry.use from atomic_t to refcount_t
  net, atm: convert eg_cache_entry.use from atomic_t to refcount_t
  net, bridge: convert net_bridge_vlan.refcnt from atomic_t to
    refcount_t
  net, calipso: convert calipso_doi.refcount from atomic_t to refcount_t
  net, sched: convert Qdisc.refcnt from atomic_t to refcount_t
  net, lapb: convert lapb_cb.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_interface.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_route.refcnt from atomic_t to refcount_t
  net, netrom: convert nr_neigh.refcount from atomic_t to refcount_t
  net, netrom: convert nr_node.refcount from atomic_t to refcount_t

 drivers/net/vxlan.c            | 10 +++++-----
 include/linux/atmdev.h         |  7 ++++---
 include/net/bluetooth/rfcomm.h |  8 +++++---
 include/net/calipso.h          |  4 ++--
 include/net/dn_fib.h           |  5 +++--
 include/net/ipx.h              | 13 +++++++------
 include/net/lapb.h             |  3 ++-
 include/net/llc.h              |  6 +++---
 include/net/netrom.h           | 13 +++++++------
 include/net/sch_generic.h      |  3 ++-
 include/net/vxlan.h            |  2 +-
 net/atm/lec.c                  |  6 +++---
 net/atm/lec_arpc.h             |  2 +-
 net/atm/mpoa_caches.c          | 26 +++++++++++++-------------
 net/atm/mpoa_caches.h          |  5 +++--
 net/atm/proc.c                 |  2 +-
 net/atm/resources.c            |  2 +-
 net/bluetooth/rfcomm/core.c    |  4 ++--
 net/bridge/br_private.h        |  3 ++-
 net/bridge/br_vlan.c           |  8 ++++----
 net/decnet/dn_fib.c            |  6 +++---
 net/ipv6/calipso.c             | 12 ++++++------
 net/ipx/af_ipx.c               |  6 +++---
 net/ipx/ipx_proc.c             |  2 +-
 net/ipx/ipx_route.c            |  2 +-
 net/l2tp/l2tp_core.c           | 16 ++++++++--------
 net/l2tp/l2tp_core.h           | 13 +++++++------
 net/l2tp/l2tp_debugfs.c        |  5 ++---
 net/l2tp/l2tp_ppp.c            |  2 +-
 net/lapb/lapb_iface.c          |  6 +++---
 net/llc/llc_core.c             |  2 +-
 net/netrom/nr_route.c          |  6 +++---
 net/sched/sch_api.c            |  8 ++++----
 net/sched/sch_generic.c        |  8 ++++----
 34 files changed, 118 insertions(+), 108 deletions(-)

-- 
2.7.4

[PATCH 00/18] net subsystem various refcounter conversions

[Elena Reshetova]

This series, for various network subsystem components, replaces atomic_t reference
counters with the new refcount_t type and API (see include/linux/refcount.h).
By doing this we prevent intentional or accidental
underflows or overflows that can led to use-after-free vulnerabilities.

The patches are fully independent and can be cherry-picked separately.
Since we convert all kernel subsystems in the same fashion, resulting
in about 300 patches, we have to group them for sending at least in some
fashion to be manageable. Please excuse the long cc list.

If there are no objections to the patches, please merge them via respective trees.

Elena Reshetova (18):
  net, llc: convert llc_sap.refcnt from atomic_t to refcount_t
  net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t
  net, l2tp: convert l2tp_session.ref_count from atomic_t to refcount_t
  net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t
  net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t
  net, decnet: convert dn_fib_info.fib_clntref from atomic_t to
    refcount_t
  net, atm: convert atm_dev.refcnt from atomic_t to refcount_t
  net, atm: convert lec_arp_table.usage from atomic_t to refcount_t
  net, atm: convert in_cache_entry.use from atomic_t to refcount_t
  net, atm: convert eg_cache_entry.use from atomic_t to refcount_t
  net, bridge: convert net_bridge_vlan.refcnt from atomic_t to
    refcount_t
  net, calipso: convert calipso_doi.refcount from atomic_t to refcount_t
  net, sched: convert Qdisc.refcnt from atomic_t to refcount_t
  net, lapb: convert lapb_cb.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_interface.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_route.refcnt from atomic_t to refcount_t
  net, netrom: convert nr_neigh.refcount from atomic_t to refcount_t
  net, netrom: convert nr_node.refcount from atomic_t to refcount_t

 drivers/net/vxlan.c            | 10 +++++-----
 include/linux/atmdev.h         |  7 ++++---
 include/net/bluetooth/rfcomm.h |  8 +++++---
 include/net/calipso.h          |  4 ++--
 include/net/dn_fib.h           |  5 +++--
 include/net/ipx.h              | 13 +++++++------
 include/net/lapb.h             |  3 ++-
 include/net/llc.h              |  6 +++---
 include/net/netrom.h           | 13 +++++++------
 include/net/sch_generic.h      |  3 ++-
 include/net/vxlan.h            |  2 +-
 net/atm/lec.c                  |  6 +++---
 net/atm/lec_arpc.h             |  2 +-
 net/atm/mpoa_caches.c          | 26 +++++++++++++-------------
 net/atm/mpoa_caches.h          |  5 +++--
 net/atm/proc.c                 |  2 +-
 net/atm/resources.c            |  2 +-
 net/bluetooth/rfcomm/core.c    |  4 ++--
 net/bridge/br_private.h        |  3 ++-
 net/bridge/br_vlan.c           |  8 ++++----
 net/decnet/dn_fib.c            |  6 +++---
 net/ipv6/calipso.c             | 12 ++++++------
 net/ipx/af_ipx.c               |  6 +++---
 net/ipx/ipx_proc.c             |  2 +-
 net/ipx/ipx_route.c            |  2 +-
 net/l2tp/l2tp_core.c           | 16 ++++++++--------
 net/l2tp/l2tp_core.h           | 13 +++++++------
 net/l2tp/l2tp_debugfs.c        |  5 ++---
 net/l2tp/l2tp_ppp.c            |  2 +-
 net/lapb/lapb_iface.c          |  6 +++---
 net/llc/llc_core.c             |  2 +-
 net/netrom/nr_route.c          |  6 +++---
 net/sched/sch_api.c            |  8 ++++----
 net/sched/sch_generic.c        |  8 ++++----
 34 files changed, 118 insertions(+), 108 deletions(-)

-- 
2.7.4

[Bridge] [PATCH 00/18] net subsystem various refcounter conversions

[Elena Reshetova]

This series, for various network subsystem components, replaces atomic_t reference
counters with the new refcount_t type and API (see include/linux/refcount.h).
By doing this we prevent intentional or accidental
underflows or overflows that can led to use-after-free vulnerabilities.

The patches are fully independent and can be cherry-picked separately.
Since we convert all kernel subsystems in the same fashion, resulting
in about 300 patches, we have to group them for sending at least in some
fashion to be manageable. Please excuse the long cc list.

If there are no objections to the patches, please merge them via respective trees.

Elena Reshetova (18):
  net, llc: convert llc_sap.refcnt from atomic_t to refcount_t
  net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t
  net, l2tp: convert l2tp_session.ref_count from atomic_t to refcount_t
  net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t
  net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t
  net, decnet: convert dn_fib_info.fib_clntref from atomic_t to
    refcount_t
  net, atm: convert atm_dev.refcnt from atomic_t to refcount_t
  net, atm: convert lec_arp_table.usage from atomic_t to refcount_t
  net, atm: convert in_cache_entry.use from atomic_t to refcount_t
  net, atm: convert eg_cache_entry.use from atomic_t to refcount_t
  net, bridge: convert net_bridge_vlan.refcnt from atomic_t to
    refcount_t
  net, calipso: convert calipso_doi.refcount from atomic_t to refcount_t
  net, sched: convert Qdisc.refcnt from atomic_t to refcount_t
  net, lapb: convert lapb_cb.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_interface.refcnt from atomic_t to refcount_t
  net, ipx: convert ipx_route.refcnt from atomic_t to refcount_t
  net, netrom: convert nr_neigh.refcount from atomic_t to refcount_t
  net, netrom: convert nr_node.refcount from atomic_t to refcount_t

 drivers/net/vxlan.c            | 10 +++++-----
 include/linux/atmdev.h         |  7 ++++---
 include/net/bluetooth/rfcomm.h |  8 +++++---
 include/net/calipso.h          |  4 ++--
 include/net/dn_fib.h           |  5 +++--
 include/net/ipx.h              | 13 +++++++------
 include/net/lapb.h             |  3 ++-
 include/net/llc.h              |  6 +++---
 include/net/netrom.h           | 13 +++++++------
 include/net/sch_generic.h      |  3 ++-
 include/net/vxlan.h            |  2 +-
 net/atm/lec.c                  |  6 +++---
 net/atm/lec_arpc.h             |  2 +-
 net/atm/mpoa_caches.c          | 26 +++++++++++++-------------
 net/atm/mpoa_caches.h          |  5 +++--
 net/atm/proc.c                 |  2 +-
 net/atm/resources.c            |  2 +-
 net/bluetooth/rfcomm/core.c    |  4 ++--
 net/bridge/br_private.h        |  3 ++-
 net/bridge/br_vlan.c           |  8 ++++----
 net/decnet/dn_fib.c            |  6 +++---
 net/ipv6/calipso.c             | 12 ++++++------
 net/ipx/af_ipx.c               |  6 +++---
 net/ipx/ipx_proc.c             |  2 +-
 net/ipx/ipx_route.c            |  2 +-
 net/l2tp/l2tp_core.c           | 16 ++++++++--------
 net/l2tp/l2tp_core.h           | 13 +++++++------
 net/l2tp/l2tp_debugfs.c        |  5 ++---
 net/l2tp/l2tp_ppp.c            |  2 +-
 net/lapb/lapb_iface.c          |  6 +++---
 net/llc/llc_core.c             |  2 +-
 net/netrom/nr_route.c          |  6 +++---
 net/sched/sch_api.c            |  8 ++++----
 net/sched/sch_generic.c        |  8 ++++----
 34 files changed, 118 insertions(+), 108 deletions(-)

-- 
2.7.4

[PATCH 01/18] net, llc: convert llc_sap.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/llc.h  | 6 +++---
 net/llc/llc_core.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/llc.h b/include/net/llc.h
index e8e61d4..dc35f25 100644
--- a/include/net/llc.h
+++ b/include/net/llc.h
@@ -55,7 +55,7 @@ struct llc_sap {
 	unsigned char	 state;
 	unsigned char	 p_bit;
 	unsigned char	 f_bit;
-	atomic_t         refcnt;
+	refcount_t		 refcnt;
 	int		 (*rcv_func)(struct sk_buff *skb,
 				     struct net_device *dev,
 				     struct packet_type *pt,
@@ -113,14 +113,14 @@ struct llc_sap *llc_sap_open(unsigned char lsap,
 					struct net_device *orig_dev));
 static inline void llc_sap_hold(struct llc_sap *sap)
 {
-	atomic_inc(&sap->refcnt);
+	refcount_inc(&sap->refcnt);
 }
 
 void llc_sap_close(struct llc_sap *sap);
 
 static inline void llc_sap_put(struct llc_sap *sap)
 {
-	if (atomic_dec_and_test(&sap->refcnt))
+	if (refcount_dec_and_test(&sap->refcnt))
 		llc_sap_close(sap);
 }
 
diff --git a/net/llc/llc_core.c b/net/llc/llc_core.c
index 842851c..8904126 100644
--- a/net/llc/llc_core.c
+++ b/net/llc/llc_core.c
@@ -41,7 +41,7 @@ static struct llc_sap *llc_sap_alloc(void)
 		spin_lock_init(&sap->sk_lock);
 		for (i = 0; i < LLC_SK_LADDR_HASH_ENTRIES; i++)
 			INIT_HLIST_NULLS_HEAD(&sap->sk_laddr_hash[i], i);
-		atomic_set(&sap->refcnt, 1);
+		refcount_set(&sap->refcnt, 1);
 	}
 	return sap;
 }
-- 
2.7.4

[Bridge] [PATCH 01/18] net, llc: convert llc_sap.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/llc.h  | 6 +++---
 net/llc/llc_core.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/llc.h b/include/net/llc.h
index e8e61d4..dc35f25 100644
--- a/include/net/llc.h
+++ b/include/net/llc.h
@@ -55,7 +55,7 @@ struct llc_sap {
 	unsigned char	 state;
 	unsigned char	 p_bit;
 	unsigned char	 f_bit;
-	atomic_t         refcnt;
+	refcount_t		 refcnt;
 	int		 (*rcv_func)(struct sk_buff *skb,
 				     struct net_device *dev,
 				     struct packet_type *pt,
@@ -113,14 +113,14 @@ struct llc_sap *llc_sap_open(unsigned char lsap,
 					struct net_device *orig_dev));
 static inline void llc_sap_hold(struct llc_sap *sap)
 {
-	atomic_inc(&sap->refcnt);
+	refcount_inc(&sap->refcnt);
 }
 
 void llc_sap_close(struct llc_sap *sap);
 
 static inline void llc_sap_put(struct llc_sap *sap)
 {
-	if (atomic_dec_and_test(&sap->refcnt))
+	if (refcount_dec_and_test(&sap->refcnt))
 		llc_sap_close(sap);
 }
 
diff --git a/net/llc/llc_core.c b/net/llc/llc_core.c
index 842851c..8904126 100644
--- a/net/llc/llc_core.c
+++ b/net/llc/llc_core.c
@@ -41,7 +41,7 @@ static struct llc_sap *llc_sap_alloc(void)
 		spin_lock_init(&sap->sk_lock);
 		for (i = 0; i < LLC_SK_LADDR_HASH_ENTRIES; i++)
 			INIT_HLIST_NULLS_HEAD(&sap->sk_laddr_hash[i], i);
-		atomic_set(&sap->refcnt, 1);
+		refcount_set(&sap->refcnt, 1);
 	}
 	return sap;
 }
-- 
2.7.4

[PATCH 02/18] net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/l2tp/l2tp_core.c    | 14 +++++++-------
 net/l2tp/l2tp_core.h    |  3 ++-
 net/l2tp/l2tp_debugfs.c |  5 ++---
 net/l2tp/l2tp_ppp.c     |  2 +-
 4 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 8adab63..37e9cf3 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -132,12 +132,12 @@ static inline struct l2tp_net *l2tp_pernet(struct net *net)
  */
 static inline void l2tp_tunnel_inc_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	atomic_inc(&tunnel->ref_count);
+	refcount_inc(&tunnel->ref_count);
 }
 
 static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	if (atomic_dec_and_test(&tunnel->ref_count))
+	if (refcount_dec_and_test(&tunnel->ref_count))
 		l2tp_tunnel_free(tunnel);
 }
 #ifdef L2TP_REFCNT_DEBUG
@@ -145,14 +145,14 @@ static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 do {									\
 	pr_debug("l2tp_tunnel_inc_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_inc_refcount_1(_t);					\
 } while (0)
 #define l2tp_tunnel_dec_refcount(_t)					\
 do {									\
 	pr_debug("l2tp_tunnel_dec_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_dec_refcount_1(_t);					\
 } while (0)
 #else
@@ -1303,7 +1303,7 @@ static void l2tp_udp_encap_destroy(struct sock *sk)
  */
 static void l2tp_tunnel_free(struct l2tp_tunnel *tunnel)
 {
-	BUG_ON(atomic_read(&tunnel->ref_count) != 0);
+	BUG_ON(refcount_read(&tunnel->ref_count) != 0);
 	BUG_ON(tunnel->sock != NULL);
 	l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: free...\n", tunnel->name);
 	kfree_rcu(tunnel, rcu);
@@ -1617,7 +1617,7 @@ int l2tp_tunnel_create(struct net *net, int fd, int version, u32 tunnel_id, u32
 	/* Bump the reference count. The tunnel context is deleted
 	 * only when this drops to zero. Must be done before list insertion
 	 */
-	l2tp_tunnel_inc_refcount(tunnel);
+	refcount_set(&tunnel->ref_count, 1);
 	spin_lock_bh(&pn->l2tp_tunnel_list_lock);
 	list_add_rcu(&tunnel->list, &pn->l2tp_tunnel_list);
 	spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
@@ -1656,7 +1656,7 @@ void l2tp_session_free(struct l2tp_session *session)
 {
 	struct l2tp_tunnel *tunnel = session->tunnel;
 
-	BUG_ON(atomic_read(&session->ref_count) != 0);
+	BUG_ON(refcount_read(&session->ref_count) != 0);
 
 	if (tunnel) {
 		BUG_ON(tunnel->magic != L2TP_TUNNEL_MAGIC);
diff --git a/net/l2tp/l2tp_core.h b/net/l2tp/l2tp_core.h
index aebf281..b942854 100644
--- a/net/l2tp/l2tp_core.h
+++ b/net/l2tp/l2tp_core.h
@@ -7,6 +7,7 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  */
+#include <linux/refcount.h>
 
 #ifndef _L2TP_CORE_H_
 #define _L2TP_CORE_H_
@@ -177,7 +178,7 @@ struct l2tp_tunnel {
 	struct list_head	list;		/* Keep a list of all tunnels */
 	struct net		*l2tp_net;	/* the net we belong to */
 
-	atomic_t		ref_count;
+	refcount_t		ref_count;
 #ifdef CONFIG_DEBUG_FS
 	void (*show)(struct seq_file *m, void *arg);
 #endif
diff --git a/net/l2tp/l2tp_debugfs.c b/net/l2tp/l2tp_debugfs.c
index 2d6760a..189e738 100644
--- a/net/l2tp/l2tp_debugfs.c
+++ b/net/l2tp/l2tp_debugfs.c
@@ -145,8 +145,7 @@ static void l2tp_dfs_seq_tunnel_show(struct seq_file *m, void *v)
 		   "");
 	seq_printf(m, " %d sessions, refcnt %d/%d\n", session_count,
 		   tunnel->sock ? atomic_read(&tunnel->sock->sk_refcnt) : 0,
-		   atomic_read(&tunnel->ref_count));
-
+		   refcount_read(&tunnel->ref_count));
 	seq_printf(m, " %08x rx %ld/%ld/%ld rx %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
@@ -171,7 +170,7 @@ static void l2tp_dfs_seq_session_show(struct seq_file *m, void *v)
 		   "");
 	if (session->send_seq || session->recv_seq)
 		seq_printf(m, "   nr %hu, ns %hu\n", session->nr, session->ns);
-	seq_printf(m, "   refcnt %d\n", atomic_read(&session->ref_count));
+	seq_printf(m, "   refcnt %d\n", refcount_read(&session->ref_count));
 	seq_printf(m, "   config %d/%d/%c/%c/%s/%s %08x %u\n",
 		   session->mtu, session->mru,
 		   session->recv_seq ? 'R' : '-',
diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c
index 36cc56f..eb1a85a 100644
--- a/net/l2tp/l2tp_ppp.c
+++ b/net/l2tp/l2tp_ppp.c
@@ -1607,7 +1607,7 @@ static void pppol2tp_seq_tunnel_show(struct seq_file *m, void *v)
 	seq_printf(m, "\nTUNNEL '%s', %c %d\n",
 		   tunnel->name,
 		   (tunnel == tunnel->sock->sk_user_data) ? 'Y' : 'N',
-		   atomic_read(&tunnel->ref_count) - 1);
+		   refcount_read(&tunnel->ref_count) - 1);
 	seq_printf(m, " %08x %ld/%ld/%ld %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
-- 
2.7.4

[PATCH 02/18] net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/l2tp/l2tp_core.c    | 14 +++++++-------
 net/l2tp/l2tp_core.h    |  3 ++-
 net/l2tp/l2tp_debugfs.c |  5 ++---
 net/l2tp/l2tp_ppp.c     |  2 +-
 4 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 8adab63..37e9cf3 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -132,12 +132,12 @@ static inline struct l2tp_net *l2tp_pernet(struct net *net)
  */
 static inline void l2tp_tunnel_inc_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	atomic_inc(&tunnel->ref_count);
+	refcount_inc(&tunnel->ref_count);
 }
 
 static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	if (atomic_dec_and_test(&tunnel->ref_count))
+	if (refcount_dec_and_test(&tunnel->ref_count))
 		l2tp_tunnel_free(tunnel);
 }
 #ifdef L2TP_REFCNT_DEBUG
@@ -145,14 +145,14 @@ static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 do {									\
 	pr_debug("l2tp_tunnel_inc_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_inc_refcount_1(_t);					\
 } while (0)
 #define l2tp_tunnel_dec_refcount(_t)					\
 do {									\
 	pr_debug("l2tp_tunnel_dec_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_dec_refcount_1(_t);					\
 } while (0)
 #else
@@ -1303,7 +1303,7 @@ static void l2tp_udp_encap_destroy(struct sock *sk)
  */
 static void l2tp_tunnel_free(struct l2tp_tunnel *tunnel)
 {
-	BUG_ON(atomic_read(&tunnel->ref_count) != 0);
+	BUG_ON(refcount_read(&tunnel->ref_count) != 0);
 	BUG_ON(tunnel->sock != NULL);
 	l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: free...\n", tunnel->name);
 	kfree_rcu(tunnel, rcu);
@@ -1617,7 +1617,7 @@ int l2tp_tunnel_create(struct net *net, int fd, int version, u32 tunnel_id, u32
 	/* Bump the reference count. The tunnel context is deleted
 	 * only when this drops to zero. Must be done before list insertion
 	 */
-	l2tp_tunnel_inc_refcount(tunnel);
+	refcount_set(&tunnel->ref_count, 1);
 	spin_lock_bh(&pn->l2tp_tunnel_list_lock);
 	list_add_rcu(&tunnel->list, &pn->l2tp_tunnel_list);
 	spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
@@ -1656,7 +1656,7 @@ void l2tp_session_free(struct l2tp_session *session)
 {
 	struct l2tp_tunnel *tunnel = session->tunnel;
 
-	BUG_ON(atomic_read(&session->ref_count) != 0);
+	BUG_ON(refcount_read(&session->ref_count) != 0);
 
 	if (tunnel) {
 		BUG_ON(tunnel->magic != L2TP_TUNNEL_MAGIC);
diff --git a/net/l2tp/l2tp_core.h b/net/l2tp/l2tp_core.h
index aebf281..b942854 100644
--- a/net/l2tp/l2tp_core.h
+++ b/net/l2tp/l2tp_core.h
@@ -7,6 +7,7 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  */
+#include <linux/refcount.h>
 
 #ifndef _L2TP_CORE_H_
 #define _L2TP_CORE_H_
@@ -177,7 +178,7 @@ struct l2tp_tunnel {
 	struct list_head	list;		/* Keep a list of all tunnels */
 	struct net		*l2tp_net;	/* the net we belong to */
 
-	atomic_t		ref_count;
+	refcount_t		ref_count;
 #ifdef CONFIG_DEBUG_FS
 	void (*show)(struct seq_file *m, void *arg);
 #endif
diff --git a/net/l2tp/l2tp_debugfs.c b/net/l2tp/l2tp_debugfs.c
index 2d6760a..189e738 100644
--- a/net/l2tp/l2tp_debugfs.c
+++ b/net/l2tp/l2tp_debugfs.c
@@ -145,8 +145,7 @@ static void l2tp_dfs_seq_tunnel_show(struct seq_file *m, void *v)
 		   "");
 	seq_printf(m, " %d sessions, refcnt %d/%d\n", session_count,
 		   tunnel->sock ? atomic_read(&tunnel->sock->sk_refcnt) : 0,
-		   atomic_read(&tunnel->ref_count));
-
+		   refcount_read(&tunnel->ref_count));
 	seq_printf(m, " %08x rx %ld/%ld/%ld rx %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
@@ -171,7 +170,7 @@ static void l2tp_dfs_seq_session_show(struct seq_file *m, void *v)
 		   "");
 	if (session->send_seq || session->recv_seq)
 		seq_printf(m, "   nr %hu, ns %hu\n", session->nr, session->ns);
-	seq_printf(m, "   refcnt %d\n", atomic_read(&session->ref_count));
+	seq_printf(m, "   refcnt %d\n", refcount_read(&session->ref_count));
 	seq_printf(m, "   config %d/%d/%c/%c/%s/%s %08x %u\n",
 		   session->mtu, session->mru,
 		   session->recv_seq ? 'R' : '-',
diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c
index 36cc56f..eb1a85a 100644
--- a/net/l2tp/l2tp_ppp.c
+++ b/net/l2tp/l2tp_ppp.c
@@ -1607,7 +1607,7 @@ static void pppol2tp_seq_tunnel_show(struct seq_file *m, void *v)
 	seq_printf(m, "\nTUNNEL '%s', %c %d\n",
 		   tunnel->name,
 		   (tunnel == tunnel->sock->sk_user_data) ? 'Y' : 'N',
-		   atomic_read(&tunnel->ref_count) - 1);
+		   refcount_read(&tunnel->ref_count) - 1);
 	seq_printf(m, " %08x %ld/%ld/%ld %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
-- 
2.7.4

[Bridge] [PATCH 02/18] net, l2tp: convert l2tp_tunnel.ref_count from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/l2tp/l2tp_core.c    | 14 +++++++-------
 net/l2tp/l2tp_core.h    |  3 ++-
 net/l2tp/l2tp_debugfs.c |  5 ++---
 net/l2tp/l2tp_ppp.c     |  2 +-
 4 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 8adab63..37e9cf3 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -132,12 +132,12 @@ static inline struct l2tp_net *l2tp_pernet(struct net *net)
  */
 static inline void l2tp_tunnel_inc_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	atomic_inc(&tunnel->ref_count);
+	refcount_inc(&tunnel->ref_count);
 }
 
 static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 {
-	if (atomic_dec_and_test(&tunnel->ref_count))
+	if (refcount_dec_and_test(&tunnel->ref_count))
 		l2tp_tunnel_free(tunnel);
 }
 #ifdef L2TP_REFCNT_DEBUG
@@ -145,14 +145,14 @@ static inline void l2tp_tunnel_dec_refcount_1(struct l2tp_tunnel *tunnel)
 do {									\
 	pr_debug("l2tp_tunnel_inc_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_inc_refcount_1(_t);					\
 } while (0)
 #define l2tp_tunnel_dec_refcount(_t)					\
 do {									\
 	pr_debug("l2tp_tunnel_dec_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_t)->name,			\
-		 atomic_read(&_t->ref_count));				\
+		 refcount_read(&_t->ref_count));			\
 	l2tp_tunnel_dec_refcount_1(_t);					\
 } while (0)
 #else
@@ -1303,7 +1303,7 @@ static void l2tp_udp_encap_destroy(struct sock *sk)
  */
 static void l2tp_tunnel_free(struct l2tp_tunnel *tunnel)
 {
-	BUG_ON(atomic_read(&tunnel->ref_count) != 0);
+	BUG_ON(refcount_read(&tunnel->ref_count) != 0);
 	BUG_ON(tunnel->sock != NULL);
 	l2tp_info(tunnel, L2TP_MSG_CONTROL, "%s: free...\n", tunnel->name);
 	kfree_rcu(tunnel, rcu);
@@ -1617,7 +1617,7 @@ int l2tp_tunnel_create(struct net *net, int fd, int version, u32 tunnel_id, u32
 	/* Bump the reference count. The tunnel context is deleted
 	 * only when this drops to zero. Must be done before list insertion
 	 */
-	l2tp_tunnel_inc_refcount(tunnel);
+	refcount_set(&tunnel->ref_count, 1);
 	spin_lock_bh(&pn->l2tp_tunnel_list_lock);
 	list_add_rcu(&tunnel->list, &pn->l2tp_tunnel_list);
 	spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
@@ -1656,7 +1656,7 @@ void l2tp_session_free(struct l2tp_session *session)
 {
 	struct l2tp_tunnel *tunnel = session->tunnel;
 
-	BUG_ON(atomic_read(&session->ref_count) != 0);
+	BUG_ON(refcount_read(&session->ref_count) != 0);
 
 	if (tunnel) {
 		BUG_ON(tunnel->magic != L2TP_TUNNEL_MAGIC);
diff --git a/net/l2tp/l2tp_core.h b/net/l2tp/l2tp_core.h
index aebf281..b942854 100644
--- a/net/l2tp/l2tp_core.h
+++ b/net/l2tp/l2tp_core.h
@@ -7,6 +7,7 @@
  * it under the terms of the GNU General Public License version 2 as
  * published by the Free Software Foundation.
  */
+#include <linux/refcount.h>
 
 #ifndef _L2TP_CORE_H_
 #define _L2TP_CORE_H_
@@ -177,7 +178,7 @@ struct l2tp_tunnel {
 	struct list_head	list;		/* Keep a list of all tunnels */
 	struct net		*l2tp_net;	/* the net we belong to */
 
-	atomic_t		ref_count;
+	refcount_t		ref_count;
 #ifdef CONFIG_DEBUG_FS
 	void (*show)(struct seq_file *m, void *arg);
 #endif
diff --git a/net/l2tp/l2tp_debugfs.c b/net/l2tp/l2tp_debugfs.c
index 2d6760a..189e738 100644
--- a/net/l2tp/l2tp_debugfs.c
+++ b/net/l2tp/l2tp_debugfs.c
@@ -145,8 +145,7 @@ static void l2tp_dfs_seq_tunnel_show(struct seq_file *m, void *v)
 		   "");
 	seq_printf(m, " %d sessions, refcnt %d/%d\n", session_count,
 		   tunnel->sock ? atomic_read(&tunnel->sock->sk_refcnt) : 0,
-		   atomic_read(&tunnel->ref_count));
-
+		   refcount_read(&tunnel->ref_count));
 	seq_printf(m, " %08x rx %ld/%ld/%ld rx %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
@@ -171,7 +170,7 @@ static void l2tp_dfs_seq_session_show(struct seq_file *m, void *v)
 		   "");
 	if (session->send_seq || session->recv_seq)
 		seq_printf(m, "   nr %hu, ns %hu\n", session->nr, session->ns);
-	seq_printf(m, "   refcnt %d\n", atomic_read(&session->ref_count));
+	seq_printf(m, "   refcnt %d\n", refcount_read(&session->ref_count));
 	seq_printf(m, "   config %d/%d/%c/%c/%s/%s %08x %u\n",
 		   session->mtu, session->mru,
 		   session->recv_seq ? 'R' : '-',
diff --git a/net/l2tp/l2tp_ppp.c b/net/l2tp/l2tp_ppp.c
index 36cc56f..eb1a85a 100644
--- a/net/l2tp/l2tp_ppp.c
+++ b/net/l2tp/l2tp_ppp.c
@@ -1607,7 +1607,7 @@ static void pppol2tp_seq_tunnel_show(struct seq_file *m, void *v)
 	seq_printf(m, "\nTUNNEL '%s', %c %d\n",
 		   tunnel->name,
 		   (tunnel == tunnel->sock->sk_user_data) ? 'Y' : 'N',
-		   atomic_read(&tunnel->ref_count) - 1);
+		   refcount_read(&tunnel->ref_count) - 1);
 	seq_printf(m, " %08x %ld/%ld/%ld %ld/%ld/%ld\n",
 		   tunnel->debug,
 		   atomic_long_read(&tunnel->stats.tx_packets),
-- 
2.7.4

[PATCH 03/18] net, l2tp: convert l2tp_session.ref_count from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/l2tp/l2tp_core.c |  2 +-
 net/l2tp/l2tp_core.h | 10 +++++-----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 37e9cf3..20ee848 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -1796,7 +1796,7 @@ struct l2tp_session *l2tp_session_create(int priv_size, struct l2tp_tunnel *tunn
 		/* Bump the reference count. The session context is deleted
 		 * only when this drops to zero.
 		 */
-		l2tp_session_inc_refcount(session);
+		refcount_set(&session->ref_count, 1);
 		l2tp_tunnel_inc_refcount(tunnel);
 
 		/* Ensure tunnel socket isn't deleted */
diff --git a/net/l2tp/l2tp_core.h b/net/l2tp/l2tp_core.h
index b942854..568b721 100644
--- a/net/l2tp/l2tp_core.h
+++ b/net/l2tp/l2tp_core.h
@@ -99,7 +99,7 @@ struct l2tp_session {
 	int			nr_oos_count;	/* For OOS recovery */
 	int			nr_oos_count_max;
 	struct hlist_node	hlist;		/* Hash list node */
-	atomic_t		ref_count;
+	refcount_t		ref_count;
 
 	char			name[32];	/* for logging */
 	char			ifname[IFNAMSIZ];
@@ -271,12 +271,12 @@ int l2tp_ioctl(struct sock *sk, int cmd, unsigned long arg);
  */
 static inline void l2tp_session_inc_refcount_1(struct l2tp_session *session)
 {
-	atomic_inc(&session->ref_count);
+	refcount_inc(&session->ref_count);
 }
 
 static inline void l2tp_session_dec_refcount_1(struct l2tp_session *session)
 {
-	if (atomic_dec_and_test(&session->ref_count))
+	if (refcount_dec_and_test(&session->ref_count))
 		l2tp_session_free(session);
 }
 
@@ -285,14 +285,14 @@ static inline void l2tp_session_dec_refcount_1(struct l2tp_session *session)
 do {									\
 	pr_debug("l2tp_session_inc_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_s)->name,			\
-		 atomic_read(&_s->ref_count));				\
+		 refcount_read(&_s->ref_count));			\
 	l2tp_session_inc_refcount_1(_s);				\
 } while (0)
 #define l2tp_session_dec_refcount(_s)					\
 do {									\
 	pr_debug("l2tp_session_dec_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_s)->name,			\
-		 atomic_read(&_s->ref_count));				\
+		 refcount_read(&_s->ref_count));			\
 	l2tp_session_dec_refcount_1(_s);				\
 } while (0)
 #else
-- 
2.7.4

[Bridge] [PATCH 03/18] net, l2tp: convert l2tp_session.ref_count from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/l2tp/l2tp_core.c |  2 +-
 net/l2tp/l2tp_core.h | 10 +++++-----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 37e9cf3..20ee848 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -1796,7 +1796,7 @@ struct l2tp_session *l2tp_session_create(int priv_size, struct l2tp_tunnel *tunn
 		/* Bump the reference count. The session context is deleted
 		 * only when this drops to zero.
 		 */
-		l2tp_session_inc_refcount(session);
+		refcount_set(&session->ref_count, 1);
 		l2tp_tunnel_inc_refcount(tunnel);
 
 		/* Ensure tunnel socket isn't deleted */
diff --git a/net/l2tp/l2tp_core.h b/net/l2tp/l2tp_core.h
index b942854..568b721 100644
--- a/net/l2tp/l2tp_core.h
+++ b/net/l2tp/l2tp_core.h
@@ -99,7 +99,7 @@ struct l2tp_session {
 	int			nr_oos_count;	/* For OOS recovery */
 	int			nr_oos_count_max;
 	struct hlist_node	hlist;		/* Hash list node */
-	atomic_t		ref_count;
+	refcount_t		ref_count;
 
 	char			name[32];	/* for logging */
 	char			ifname[IFNAMSIZ];
@@ -271,12 +271,12 @@ int l2tp_ioctl(struct sock *sk, int cmd, unsigned long arg);
  */
 static inline void l2tp_session_inc_refcount_1(struct l2tp_session *session)
 {
-	atomic_inc(&session->ref_count);
+	refcount_inc(&session->ref_count);
 }
 
 static inline void l2tp_session_dec_refcount_1(struct l2tp_session *session)
 {
-	if (atomic_dec_and_test(&session->ref_count))
+	if (refcount_dec_and_test(&session->ref_count))
 		l2tp_session_free(session);
 }
 
@@ -285,14 +285,14 @@ static inline void l2tp_session_dec_refcount_1(struct l2tp_session *session)
 do {									\
 	pr_debug("l2tp_session_inc_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_s)->name,			\
-		 atomic_read(&_s->ref_count));				\
+		 refcount_read(&_s->ref_count));			\
 	l2tp_session_inc_refcount_1(_s);				\
 } while (0)
 #define l2tp_session_dec_refcount(_s)					\
 do {									\
 	pr_debug("l2tp_session_dec_refcount: %s:%d %s: cnt=%d\n",	\
 		 __func__, __LINE__, (_s)->name,			\
-		 atomic_read(&_s->ref_count));				\
+		 refcount_read(&_s->ref_count));			\
 	l2tp_session_dec_refcount_1(_s);				\
 } while (0)
 #else
-- 
2.7.4

[PATCH 04/18] net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 drivers/net/vxlan.c | 10 +++++-----
 include/net/vxlan.h |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index b791199..21a9251 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -999,11 +999,11 @@ static bool vxlan_group_used(struct vxlan_net *vn, struct vxlan_dev *dev)
 	/* The vxlan_sock is only used by dev, leaving group has
 	 * no effect on other vxlan devices.
 	 */
-	if (family == AF_INET && sock4 && atomic_read(&sock4->refcnt) == 1)
+	if (family == AF_INET && sock4 && refcount_read(&sock4->refcnt) == 1)
 		return false;
 #if IS_ENABLED(CONFIG_IPV6)
 	sock6 = rtnl_dereference(dev->vn6_sock);
-	if (family == AF_INET6 && sock6 && atomic_read(&sock6->refcnt) == 1)
+	if (family == AF_INET6 && sock6 && refcount_read(&sock6->refcnt) == 1)
 		return false;
 #endif
 
@@ -1040,7 +1040,7 @@ static bool __vxlan_sock_release_prep(struct vxlan_sock *vs)
 
 	if (!vs)
 		return false;
-	if (!atomic_dec_and_test(&vs->refcnt))
+	if (!refcount_dec_and_test(&vs->refcnt))
 		return false;
 
 	vn = net_generic(sock_net(vs->sock->sk), vxlan_net_id);
@@ -2757,7 +2757,7 @@ static struct vxlan_sock *vxlan_socket_create(struct net *net, bool ipv6,
 	}
 
 	vs->sock = sock;
-	atomic_set(&vs->refcnt, 1);
+	refcount_set(&vs->refcnt, 1);
 	vs->flags = (flags & VXLAN_F_RCV_FLAGS);
 
 	spin_lock(&vn->sock_lock);
@@ -2791,7 +2791,7 @@ static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
 		spin_lock(&vn->sock_lock);
 		vs = vxlan_find_sock(vxlan->net, ipv6 ? AF_INET6 : AF_INET,
 				     vxlan->cfg.dst_port, vxlan->flags);
-		if (vs && !atomic_add_unless(&vs->refcnt, 1, 0)) {
+		if (vs && !refcount_inc_not_zero(&vs->refcnt)) {
 			spin_unlock(&vn->sock_lock);
 			return -EBUSY;
 		}
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index 49a5920..5c117b4 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -183,7 +183,7 @@ struct vxlan_sock {
 	struct hlist_node hlist;
 	struct socket	 *sock;
 	struct hlist_head vni_list[VNI_HASH_SIZE];
-	atomic_t	  refcnt;
+	refcount_t	  refcnt;
 	u32		  flags;
 };
 
-- 
2.7.4

[PATCH 04/18] net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 drivers/net/vxlan.c | 10 +++++-----
 include/net/vxlan.h |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index b791199..21a9251 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -999,11 +999,11 @@ static bool vxlan_group_used(struct vxlan_net *vn, struct vxlan_dev *dev)
 	/* The vxlan_sock is only used by dev, leaving group has
 	 * no effect on other vxlan devices.
 	 */
-	if (family == AF_INET && sock4 && atomic_read(&sock4->refcnt) == 1)
+	if (family == AF_INET && sock4 && refcount_read(&sock4->refcnt) == 1)
 		return false;
 #if IS_ENABLED(CONFIG_IPV6)
 	sock6 = rtnl_dereference(dev->vn6_sock);
-	if (family == AF_INET6 && sock6 && atomic_read(&sock6->refcnt) == 1)
+	if (family == AF_INET6 && sock6 && refcount_read(&sock6->refcnt) == 1)
 		return false;
 #endif
 
@@ -1040,7 +1040,7 @@ static bool __vxlan_sock_release_prep(struct vxlan_sock *vs)
 
 	if (!vs)
 		return false;
-	if (!atomic_dec_and_test(&vs->refcnt))
+	if (!refcount_dec_and_test(&vs->refcnt))
 		return false;
 
 	vn = net_generic(sock_net(vs->sock->sk), vxlan_net_id);
@@ -2757,7 +2757,7 @@ static struct vxlan_sock *vxlan_socket_create(struct net *net, bool ipv6,
 	}
 
 	vs->sock = sock;
-	atomic_set(&vs->refcnt, 1);
+	refcount_set(&vs->refcnt, 1);
 	vs->flags = (flags & VXLAN_F_RCV_FLAGS);
 
 	spin_lock(&vn->sock_lock);
@@ -2791,7 +2791,7 @@ static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
 		spin_lock(&vn->sock_lock);
 		vs = vxlan_find_sock(vxlan->net, ipv6 ? AF_INET6 : AF_INET,
 				     vxlan->cfg.dst_port, vxlan->flags);
-		if (vs && !atomic_add_unless(&vs->refcnt, 1, 0)) {
+		if (vs && !refcount_inc_not_zero(&vs->refcnt)) {
 			spin_unlock(&vn->sock_lock);
 			return -EBUSY;
 		}
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index 49a5920..5c117b4 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -183,7 +183,7 @@ struct vxlan_sock {
 	struct hlist_node hlist;
 	struct socket	 *sock;
 	struct hlist_head vni_list[VNI_HASH_SIZE];
-	atomic_t	  refcnt;
+	refcount_t	  refcnt;
 	u32		  flags;
 };
 
-- 
2.7.4

[Bridge] [PATCH 04/18] net, vxlan: convert vxlan_sock.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 drivers/net/vxlan.c | 10 +++++-----
 include/net/vxlan.h |  2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index b791199..21a9251 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -999,11 +999,11 @@ static bool vxlan_group_used(struct vxlan_net *vn, struct vxlan_dev *dev)
 	/* The vxlan_sock is only used by dev, leaving group has
 	 * no effect on other vxlan devices.
 	 */
-	if (family == AF_INET && sock4 && atomic_read(&sock4->refcnt) == 1)
+	if (family == AF_INET && sock4 && refcount_read(&sock4->refcnt) == 1)
 		return false;
 #if IS_ENABLED(CONFIG_IPV6)
 	sock6 = rtnl_dereference(dev->vn6_sock);
-	if (family == AF_INET6 && sock6 && atomic_read(&sock6->refcnt) == 1)
+	if (family == AF_INET6 && sock6 && refcount_read(&sock6->refcnt) == 1)
 		return false;
 #endif
 
@@ -1040,7 +1040,7 @@ static bool __vxlan_sock_release_prep(struct vxlan_sock *vs)
 
 	if (!vs)
 		return false;
-	if (!atomic_dec_and_test(&vs->refcnt))
+	if (!refcount_dec_and_test(&vs->refcnt))
 		return false;
 
 	vn = net_generic(sock_net(vs->sock->sk), vxlan_net_id);
@@ -2757,7 +2757,7 @@ static struct vxlan_sock *vxlan_socket_create(struct net *net, bool ipv6,
 	}
 
 	vs->sock = sock;
-	atomic_set(&vs->refcnt, 1);
+	refcount_set(&vs->refcnt, 1);
 	vs->flags = (flags & VXLAN_F_RCV_FLAGS);
 
 	spin_lock(&vn->sock_lock);
@@ -2791,7 +2791,7 @@ static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
 		spin_lock(&vn->sock_lock);
 		vs = vxlan_find_sock(vxlan->net, ipv6 ? AF_INET6 : AF_INET,
 				     vxlan->cfg.dst_port, vxlan->flags);
-		if (vs && !atomic_add_unless(&vs->refcnt, 1, 0)) {
+		if (vs && !refcount_inc_not_zero(&vs->refcnt)) {
 			spin_unlock(&vn->sock_lock);
 			return -EBUSY;
 		}
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index 49a5920..5c117b4 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -183,7 +183,7 @@ struct vxlan_sock {
 	struct hlist_node hlist;
 	struct socket	 *sock;
 	struct hlist_head vni_list[VNI_HASH_SIZE];
-	atomic_t	  refcnt;
+	refcount_t	  refcnt;
 	u32		  flags;
 };
 
-- 
2.7.4

[PATCH 05/18] net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/bluetooth/rfcomm.h | 8 +++++---
 net/bluetooth/rfcomm/core.c    | 4 ++--
 2 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/include/net/bluetooth/rfcomm.h b/include/net/bluetooth/rfcomm.h
index 4190af5..da4acef 100644
--- a/include/net/bluetooth/rfcomm.h
+++ b/include/net/bluetooth/rfcomm.h
@@ -21,6 +21,8 @@
    SOFTWARE IS DISCLAIMED.
 */
 
+#include <linux/refcount.h>
+
 #ifndef __RFCOMM_H
 #define __RFCOMM_H
 
@@ -174,7 +176,7 @@ struct rfcomm_dlc {
 	struct mutex  lock;
 	unsigned long state;
 	unsigned long flags;
-	atomic_t      refcnt;
+	refcount_t    refcnt;
 	u8            dlci;
 	u8            addr;
 	u8            priority;
@@ -247,12 +249,12 @@ struct rfcomm_dlc *rfcomm_dlc_exists(bdaddr_t *src, bdaddr_t *dst, u8 channel);
 
 static inline void rfcomm_dlc_hold(struct rfcomm_dlc *d)
 {
-	atomic_inc(&d->refcnt);
+	refcount_inc(&d->refcnt);
 }
 
 static inline void rfcomm_dlc_put(struct rfcomm_dlc *d)
 {
-	if (atomic_dec_and_test(&d->refcnt))
+	if (refcount_dec_and_test(&d->refcnt))
 		rfcomm_dlc_free(d);
 }
 
diff --git a/net/bluetooth/rfcomm/core.c b/net/bluetooth/rfcomm/core.c
index f7eb02f..8ebca90 100644
--- a/net/bluetooth/rfcomm/core.c
+++ b/net/bluetooth/rfcomm/core.c
@@ -311,7 +311,7 @@ struct rfcomm_dlc *rfcomm_dlc_alloc(gfp_t prio)
 
 	skb_queue_head_init(&d->tx_queue);
 	mutex_init(&d->lock);
-	atomic_set(&d->refcnt, 1);
+	refcount_set(&d->refcnt, 1);
 
 	rfcomm_dlc_clear_state(d);
 
@@ -342,7 +342,7 @@ static void rfcomm_dlc_unlink(struct rfcomm_dlc *d)
 {
 	struct rfcomm_session *s = d->session;
 
-	BT_DBG("dlc %p refcnt %d session %p", d, atomic_read(&d->refcnt), s);
+	BT_DBG("dlc %p refcnt %d session %p", d, refcount_read(&d->refcnt), s);
 
 	list_del(&d->list);
 	d->session = NULL;
-- 
2.7.4

[Bridge] [PATCH 05/18] net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/bluetooth/rfcomm.h | 8 +++++---
 net/bluetooth/rfcomm/core.c    | 4 ++--
 2 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/include/net/bluetooth/rfcomm.h b/include/net/bluetooth/rfcomm.h
index 4190af5..da4acef 100644
--- a/include/net/bluetooth/rfcomm.h
+++ b/include/net/bluetooth/rfcomm.h
@@ -21,6 +21,8 @@
    SOFTWARE IS DISCLAIMED.
 */
 
+#include <linux/refcount.h>
+
 #ifndef __RFCOMM_H
 #define __RFCOMM_H
 
@@ -174,7 +176,7 @@ struct rfcomm_dlc {
 	struct mutex  lock;
 	unsigned long state;
 	unsigned long flags;
-	atomic_t      refcnt;
+	refcount_t    refcnt;
 	u8            dlci;
 	u8            addr;
 	u8            priority;
@@ -247,12 +249,12 @@ struct rfcomm_dlc *rfcomm_dlc_exists(bdaddr_t *src, bdaddr_t *dst, u8 channel);
 
 static inline void rfcomm_dlc_hold(struct rfcomm_dlc *d)
 {
-	atomic_inc(&d->refcnt);
+	refcount_inc(&d->refcnt);
 }
 
 static inline void rfcomm_dlc_put(struct rfcomm_dlc *d)
 {
-	if (atomic_dec_and_test(&d->refcnt))
+	if (refcount_dec_and_test(&d->refcnt))
 		rfcomm_dlc_free(d);
 }
 
diff --git a/net/bluetooth/rfcomm/core.c b/net/bluetooth/rfcomm/core.c
index f7eb02f..8ebca90 100644
--- a/net/bluetooth/rfcomm/core.c
+++ b/net/bluetooth/rfcomm/core.c
@@ -311,7 +311,7 @@ struct rfcomm_dlc *rfcomm_dlc_alloc(gfp_t prio)
 
 	skb_queue_head_init(&d->tx_queue);
 	mutex_init(&d->lock);
-	atomic_set(&d->refcnt, 1);
+	refcount_set(&d->refcnt, 1);
 
 	rfcomm_dlc_clear_state(d);
 
@@ -342,7 +342,7 @@ static void rfcomm_dlc_unlink(struct rfcomm_dlc *d)
 {
 	struct rfcomm_session *s = d->session;
 
-	BT_DBG("dlc %p refcnt %d session %p", d, atomic_read(&d->refcnt), s);
+	BT_DBG("dlc %p refcnt %d session %p", d, refcount_read(&d->refcnt), s);
 
 	list_del(&d->list);
 	d->session = NULL;
-- 
2.7.4

[PATCH 06/18] net, decnet: convert dn_fib_info.fib_clntref from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/dn_fib.h | 5 +++--
 net/decnet/dn_fib.c  | 6 +++---
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/net/dn_fib.h b/include/net/dn_fib.h
index f2ca135..81210a8 100644
--- a/include/net/dn_fib.h
+++ b/include/net/dn_fib.h
@@ -2,6 +2,7 @@
 #define _NET_DN_FIB_H
 
 #include <linux/netlink.h>
+#include <linux/refcount.h>
 
 extern const struct nla_policy rtm_dn_policy[];
 
@@ -28,7 +29,7 @@ struct dn_fib_info {
 	struct dn_fib_info	*fib_next;
 	struct dn_fib_info	*fib_prev;
 	int 			fib_treeref;
-	atomic_t		fib_clntref;
+	refcount_t		fib_clntref;
 	int			fib_dead;
 	unsigned int		fib_flags;
 	int			fib_protocol;
@@ -130,7 +131,7 @@ void dn_fib_free_info(struct dn_fib_info *fi);
 
 static inline void dn_fib_info_put(struct dn_fib_info *fi)
 {
-	if (atomic_dec_and_test(&fi->fib_clntref))
+	if (refcount_dec_and_test(&fi->fib_clntref))
 		dn_fib_free_info(fi);
 }
 
diff --git a/net/decnet/dn_fib.c b/net/decnet/dn_fib.c
index 7af0ba61..85058c3 100644
--- a/net/decnet/dn_fib.c
+++ b/net/decnet/dn_fib.c
@@ -389,7 +389,7 @@ struct dn_fib_info *dn_fib_create_info(const struct rtmsg *r, struct nlattr *att
 	}
 
 	fi->fib_treeref++;
-	atomic_inc(&fi->fib_clntref);
+	refcount_set(&fi->fib_clntref, 1);
 	spin_lock(&dn_fib_info_lock);
 	fi->fib_next = dn_fib_info_list;
 	fi->fib_prev = NULL;
@@ -425,7 +425,7 @@ int dn_fib_semantic_match(int type, struct dn_fib_info *fi, const struct flowidn
 		switch (type) {
 		case RTN_NAT:
 			DN_FIB_RES_RESET(*res);
-			atomic_inc(&fi->fib_clntref);
+			refcount_inc(&fi->fib_clntref);
 			return 0;
 		case RTN_UNICAST:
 		case RTN_LOCAL:
@@ -438,7 +438,7 @@ int dn_fib_semantic_match(int type, struct dn_fib_info *fi, const struct flowidn
 			}
 			if (nhsel < fi->fib_nhs) {
 				res->nh_sel = nhsel;
-				atomic_inc(&fi->fib_clntref);
+				refcount_inc(&fi->fib_clntref);
 				return 0;
 			}
 			endfor_nexthops(fi);
-- 
2.7.4

[Bridge] [PATCH 06/18] net, decnet: convert dn_fib_info.fib_clntref from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/dn_fib.h | 5 +++--
 net/decnet/dn_fib.c  | 6 +++---
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/net/dn_fib.h b/include/net/dn_fib.h
index f2ca135..81210a8 100644
--- a/include/net/dn_fib.h
+++ b/include/net/dn_fib.h
@@ -2,6 +2,7 @@
 #define _NET_DN_FIB_H
 
 #include <linux/netlink.h>
+#include <linux/refcount.h>
 
 extern const struct nla_policy rtm_dn_policy[];
 
@@ -28,7 +29,7 @@ struct dn_fib_info {
 	struct dn_fib_info	*fib_next;
 	struct dn_fib_info	*fib_prev;
 	int 			fib_treeref;
-	atomic_t		fib_clntref;
+	refcount_t		fib_clntref;
 	int			fib_dead;
 	unsigned int		fib_flags;
 	int			fib_protocol;
@@ -130,7 +131,7 @@ void dn_fib_free_info(struct dn_fib_info *fi);
 
 static inline void dn_fib_info_put(struct dn_fib_info *fi)
 {
-	if (atomic_dec_and_test(&fi->fib_clntref))
+	if (refcount_dec_and_test(&fi->fib_clntref))
 		dn_fib_free_info(fi);
 }
 
diff --git a/net/decnet/dn_fib.c b/net/decnet/dn_fib.c
index 7af0ba61..85058c3 100644
--- a/net/decnet/dn_fib.c
+++ b/net/decnet/dn_fib.c
@@ -389,7 +389,7 @@ struct dn_fib_info *dn_fib_create_info(const struct rtmsg *r, struct nlattr *att
 	}
 
 	fi->fib_treeref++;
-	atomic_inc(&fi->fib_clntref);
+	refcount_set(&fi->fib_clntref, 1);
 	spin_lock(&dn_fib_info_lock);
 	fi->fib_next = dn_fib_info_list;
 	fi->fib_prev = NULL;
@@ -425,7 +425,7 @@ int dn_fib_semantic_match(int type, struct dn_fib_info *fi, const struct flowidn
 		switch (type) {
 		case RTN_NAT:
 			DN_FIB_RES_RESET(*res);
-			atomic_inc(&fi->fib_clntref);
+			refcount_inc(&fi->fib_clntref);
 			return 0;
 		case RTN_UNICAST:
 		case RTN_LOCAL:
@@ -438,7 +438,7 @@ int dn_fib_semantic_match(int type, struct dn_fib_info *fi, const struct flowidn
 			}
 			if (nhsel < fi->fib_nhs) {
 				res->nh_sel = nhsel;
-				atomic_inc(&fi->fib_clntref);
+				refcount_inc(&fi->fib_clntref);
 				return 0;
 			}
 			endfor_nexthops(fi);
-- 
2.7.4

[PATCH 07/18] net, atm: convert atm_dev.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/linux/atmdev.h | 7 ++++---
 net/atm/proc.c         | 2 +-
 net/atm/resources.c    | 2 +-
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/linux/atmdev.h b/include/linux/atmdev.h
index c1da539..2a87aea 100644
--- a/include/linux/atmdev.h
+++ b/include/linux/atmdev.h
@@ -11,6 +11,7 @@
 #include <linux/uio.h>
 #include <net/sock.h>
 #include <linux/atomic.h>
+#include <linux/refcount.h>
 #include <uapi/linux/atmdev.h>
 
 #ifdef CONFIG_PROC_FS
@@ -158,7 +159,7 @@ struct atm_dev {
 	struct k_atm_dev_stats stats;	/* statistics */
 	char		signal;		/* signal status (ATM_PHY_SIG_*) */
 	int		link_rate;	/* link rate (default: OC3) */
-	atomic_t	refcnt;		/* reference count */
+	refcount_t	refcnt;		/* reference count */
 	spinlock_t	lock;		/* protect internal members */
 #ifdef CONFIG_PROC_FS
 	struct proc_dir_entry *proc_entry; /* proc entry */
@@ -261,13 +262,13 @@ static inline int atm_may_send(struct atm_vcc *vcc,unsigned int size)
 
 static inline void atm_dev_hold(struct atm_dev *dev)
 {
-	atomic_inc(&dev->refcnt);
+	refcount_inc(&dev->refcnt);
 }
 
 
 static inline void atm_dev_put(struct atm_dev *dev)
 {
-	if (atomic_dec_and_test(&dev->refcnt)) {
+	if (refcount_dec_and_test(&dev->refcnt)) {
 		BUG_ON(!test_bit(ATM_DF_REMOVED, &dev->flags));
 		if (dev->ops->dev_close)
 			dev->ops->dev_close(dev);
diff --git a/net/atm/proc.c b/net/atm/proc.c
index bbb6461..bcb8ee5 100644
--- a/net/atm/proc.c
+++ b/net/atm/proc.c
@@ -61,7 +61,7 @@ static void atm_dev_info(struct seq_file *seq, const struct atm_dev *dev)
 	add_stats(seq, "0", &dev->stats.aal0);
 	seq_puts(seq, "  ");
 	add_stats(seq, "5", &dev->stats.aal5);
-	seq_printf(seq, "\t[%d]", atomic_read(&dev->refcnt));
+	seq_printf(seq, "\t[%d]", refcount_read(&dev->refcnt));
 	seq_putc(seq, '\n');
 }
 
diff --git a/net/atm/resources.c b/net/atm/resources.c
index 0447d5d..9182447 100644
--- a/net/atm/resources.c
+++ b/net/atm/resources.c
@@ -109,7 +109,7 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,
 	else
 		memset(&dev->flags, 0, sizeof(dev->flags));
 	memset(&dev->stats, 0, sizeof(dev->stats));
-	atomic_set(&dev->refcnt, 1);
+	refcount_set(&dev->refcnt, 1);
 
 	if (atm_proc_dev_register(dev) < 0) {
 		pr_err("atm_proc_dev_register failed for dev %s\n", type);
-- 
2.7.4

[Bridge] [PATCH 07/18] net, atm: convert atm_dev.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/linux/atmdev.h | 7 ++++---
 net/atm/proc.c         | 2 +-
 net/atm/resources.c    | 2 +-
 3 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/linux/atmdev.h b/include/linux/atmdev.h
index c1da539..2a87aea 100644
--- a/include/linux/atmdev.h
+++ b/include/linux/atmdev.h
@@ -11,6 +11,7 @@
 #include <linux/uio.h>
 #include <net/sock.h>
 #include <linux/atomic.h>
+#include <linux/refcount.h>
 #include <uapi/linux/atmdev.h>
 
 #ifdef CONFIG_PROC_FS
@@ -158,7 +159,7 @@ struct atm_dev {
 	struct k_atm_dev_stats stats;	/* statistics */
 	char		signal;		/* signal status (ATM_PHY_SIG_*) */
 	int		link_rate;	/* link rate (default: OC3) */
-	atomic_t	refcnt;		/* reference count */
+	refcount_t	refcnt;		/* reference count */
 	spinlock_t	lock;		/* protect internal members */
 #ifdef CONFIG_PROC_FS
 	struct proc_dir_entry *proc_entry; /* proc entry */
@@ -261,13 +262,13 @@ static inline int atm_may_send(struct atm_vcc *vcc,unsigned int size)
 
 static inline void atm_dev_hold(struct atm_dev *dev)
 {
-	atomic_inc(&dev->refcnt);
+	refcount_inc(&dev->refcnt);
 }
 
 
 static inline void atm_dev_put(struct atm_dev *dev)
 {
-	if (atomic_dec_and_test(&dev->refcnt)) {
+	if (refcount_dec_and_test(&dev->refcnt)) {
 		BUG_ON(!test_bit(ATM_DF_REMOVED, &dev->flags));
 		if (dev->ops->dev_close)
 			dev->ops->dev_close(dev);
diff --git a/net/atm/proc.c b/net/atm/proc.c
index bbb6461..bcb8ee5 100644
--- a/net/atm/proc.c
+++ b/net/atm/proc.c
@@ -61,7 +61,7 @@ static void atm_dev_info(struct seq_file *seq, const struct atm_dev *dev)
 	add_stats(seq, "0", &dev->stats.aal0);
 	seq_puts(seq, "  ");
 	add_stats(seq, "5", &dev->stats.aal5);
-	seq_printf(seq, "\t[%d]", atomic_read(&dev->refcnt));
+	seq_printf(seq, "\t[%d]", refcount_read(&dev->refcnt));
 	seq_putc(seq, '\n');
 }
 
diff --git a/net/atm/resources.c b/net/atm/resources.c
index 0447d5d..9182447 100644
--- a/net/atm/resources.c
+++ b/net/atm/resources.c
@@ -109,7 +109,7 @@ struct atm_dev *atm_dev_register(const char *type, struct device *parent,
 	else
 		memset(&dev->flags, 0, sizeof(dev->flags));
 	memset(&dev->stats, 0, sizeof(dev->stats));
-	atomic_set(&dev->refcnt, 1);
+	refcount_set(&dev->refcnt, 1);
 
 	if (atm_proc_dev_register(dev) < 0) {
 		pr_err("atm_proc_dev_register failed for dev %s\n", type);
-- 
2.7.4

[PATCH 08/18] net, atm: convert lec_arp_table.usage from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/lec.c      | 6 +++---
 net/atm/lec_arpc.h | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/atm/lec.c b/net/atm/lec.c
index 09cfe87..431f32c 100644
--- a/net/atm/lec.c
+++ b/net/atm/lec.c
@@ -101,12 +101,12 @@ static void lec_vcc_close(struct lec_priv *priv, struct atm_vcc *vcc);
 /* must be done under lec_arp_lock */
 static inline void lec_arp_hold(struct lec_arp_table *entry)
 {
-	atomic_inc(&entry->usage);
+	refcount_inc(&entry->usage);
 }
 
 static inline void lec_arp_put(struct lec_arp_table *entry)
 {
-	if (atomic_dec_and_test(&entry->usage))
+	if (refcount_dec_and_test(&entry->usage))
 		kfree(entry);
 }
 
@@ -1564,7 +1564,7 @@ static struct lec_arp_table *make_entry(struct lec_priv *priv,
 	to_return->last_used = jiffies;
 	to_return->priv = priv;
 	skb_queue_head_init(&to_return->tx_wait);
-	atomic_set(&to_return->usage, 1);
+	refcount_set(&to_return->usage, 1);
 	return to_return;
 }
 
diff --git a/net/atm/lec_arpc.h b/net/atm/lec_arpc.h
index ec67435..d923f53 100644
--- a/net/atm/lec_arpc.h
+++ b/net/atm/lec_arpc.h
@@ -47,7 +47,7 @@ struct lec_arp_table {
 					 * the length of the tlvs array
 					 */
 	struct sk_buff_head tx_wait;	/* wait queue for outgoing packets */
-	atomic_t usage;			/* usage count */
+	refcount_t usage;		/* usage count */
 };
 
 /*
-- 
2.7.4

[Bridge] [PATCH 08/18] net, atm: convert lec_arp_table.usage from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/lec.c      | 6 +++---
 net/atm/lec_arpc.h | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/atm/lec.c b/net/atm/lec.c
index 09cfe87..431f32c 100644
--- a/net/atm/lec.c
+++ b/net/atm/lec.c
@@ -101,12 +101,12 @@ static void lec_vcc_close(struct lec_priv *priv, struct atm_vcc *vcc);
 /* must be done under lec_arp_lock */
 static inline void lec_arp_hold(struct lec_arp_table *entry)
 {
-	atomic_inc(&entry->usage);
+	refcount_inc(&entry->usage);
 }
 
 static inline void lec_arp_put(struct lec_arp_table *entry)
 {
-	if (atomic_dec_and_test(&entry->usage))
+	if (refcount_dec_and_test(&entry->usage))
 		kfree(entry);
 }
 
@@ -1564,7 +1564,7 @@ static struct lec_arp_table *make_entry(struct lec_priv *priv,
 	to_return->last_used = jiffies;
 	to_return->priv = priv;
 	skb_queue_head_init(&to_return->tx_wait);
-	atomic_set(&to_return->usage, 1);
+	refcount_set(&to_return->usage, 1);
 	return to_return;
 }
 
diff --git a/net/atm/lec_arpc.h b/net/atm/lec_arpc.h
index ec67435..d923f53 100644
--- a/net/atm/lec_arpc.h
+++ b/net/atm/lec_arpc.h
@@ -47,7 +47,7 @@ struct lec_arp_table {
 					 * the length of the tlvs array
 					 */
 	struct sk_buff_head tx_wait;	/* wait queue for outgoing packets */
-	atomic_t usage;			/* usage count */
+	refcount_t usage;		/* usage count */
 };
 
 /*
-- 
2.7.4

[PATCH 09/18] net, atm: convert in_cache_entry.use from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/mpoa_caches.c | 12 ++++++------
 net/atm/mpoa_caches.h |  3 ++-
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/net/atm/mpoa_caches.c b/net/atm/mpoa_caches.c
index a89fdeb..05e89e9 100644
--- a/net/atm/mpoa_caches.c
+++ b/net/atm/mpoa_caches.c
@@ -40,7 +40,7 @@ static in_cache_entry *in_cache_get(__be32 dst_ip,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.in_dst_ip == dst_ip) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -61,7 +61,7 @@ static in_cache_entry *in_cache_get_with_mask(__be32 dst_ip,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if ((entry->ctrl_info.in_dst_ip & mask) == (dst_ip & mask)) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -82,7 +82,7 @@ static in_cache_entry *in_cache_get_by_vcc(struct atm_vcc *vcc,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if (entry->shortcut == vcc) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -105,7 +105,7 @@ static in_cache_entry *in_cache_add_entry(__be32 dst_ip,
 
 	dprintk("adding an ingress entry, ip = %pI4\n", &dst_ip);
 
-	atomic_set(&entry->use, 1);
+	refcount_set(&entry->use, 1);
 	dprintk("new_in_cache_entry: about to lock\n");
 	write_lock_bh(&client->ingress_lock);
 	entry->next = client->in_cache;
@@ -121,7 +121,7 @@ static in_cache_entry *in_cache_add_entry(__be32 dst_ip,
 	entry->count = 1;
 	entry->entry_state = INGRESS_INVALID;
 	entry->ctrl_info.holding_time = HOLDING_TIME_DEFAULT;
-	atomic_inc(&entry->use);
+	refcount_inc(&entry->use);
 
 	write_unlock_bh(&client->ingress_lock);
 	dprintk("new_in_cache_entry: unlocked\n");
@@ -178,7 +178,7 @@ static int cache_hit(in_cache_entry *entry, struct mpoa_client *mpc)
 
 static void in_cache_put(in_cache_entry *entry)
 {
-	if (atomic_dec_and_test(&entry->use)) {
+	if (refcount_dec_and_test(&entry->use)) {
 		memset(entry, 0, sizeof(in_cache_entry));
 		kfree(entry);
 	}
diff --git a/net/atm/mpoa_caches.h b/net/atm/mpoa_caches.h
index 8e5f78c..38a4e7e 100644
--- a/net/atm/mpoa_caches.h
+++ b/net/atm/mpoa_caches.h
@@ -6,6 +6,7 @@
 #include <linux/atm.h>
 #include <linux/atmdev.h>
 #include <linux/atmmpc.h>
+#include <linux/refcount.h>
 
 struct mpoa_client;
 
@@ -25,7 +26,7 @@ typedef struct in_cache_entry {
 	struct   atm_vcc *shortcut;
 	uint8_t  MPS_ctrl_ATM_addr[ATM_ESA_LEN];
 	struct   in_ctrl_info ctrl_info;
-	atomic_t use;
+	refcount_t use;
 } in_cache_entry;
 
 struct in_cache_ops{
-- 
2.7.4

[Bridge] [PATCH 09/18] net, atm: convert in_cache_entry.use from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/mpoa_caches.c | 12 ++++++------
 net/atm/mpoa_caches.h |  3 ++-
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/net/atm/mpoa_caches.c b/net/atm/mpoa_caches.c
index a89fdeb..05e89e9 100644
--- a/net/atm/mpoa_caches.c
+++ b/net/atm/mpoa_caches.c
@@ -40,7 +40,7 @@ static in_cache_entry *in_cache_get(__be32 dst_ip,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.in_dst_ip == dst_ip) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -61,7 +61,7 @@ static in_cache_entry *in_cache_get_with_mask(__be32 dst_ip,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if ((entry->ctrl_info.in_dst_ip & mask) == (dst_ip & mask)) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -82,7 +82,7 @@ static in_cache_entry *in_cache_get_by_vcc(struct atm_vcc *vcc,
 	entry = client->in_cache;
 	while (entry != NULL) {
 		if (entry->shortcut == vcc) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_bh(&client->ingress_lock);
 			return entry;
 		}
@@ -105,7 +105,7 @@ static in_cache_entry *in_cache_add_entry(__be32 dst_ip,
 
 	dprintk("adding an ingress entry, ip = %pI4\n", &dst_ip);
 
-	atomic_set(&entry->use, 1);
+	refcount_set(&entry->use, 1);
 	dprintk("new_in_cache_entry: about to lock\n");
 	write_lock_bh(&client->ingress_lock);
 	entry->next = client->in_cache;
@@ -121,7 +121,7 @@ static in_cache_entry *in_cache_add_entry(__be32 dst_ip,
 	entry->count = 1;
 	entry->entry_state = INGRESS_INVALID;
 	entry->ctrl_info.holding_time = HOLDING_TIME_DEFAULT;
-	atomic_inc(&entry->use);
+	refcount_inc(&entry->use);
 
 	write_unlock_bh(&client->ingress_lock);
 	dprintk("new_in_cache_entry: unlocked\n");
@@ -178,7 +178,7 @@ static int cache_hit(in_cache_entry *entry, struct mpoa_client *mpc)
 
 static void in_cache_put(in_cache_entry *entry)
 {
-	if (atomic_dec_and_test(&entry->use)) {
+	if (refcount_dec_and_test(&entry->use)) {
 		memset(entry, 0, sizeof(in_cache_entry));
 		kfree(entry);
 	}
diff --git a/net/atm/mpoa_caches.h b/net/atm/mpoa_caches.h
index 8e5f78c..38a4e7e 100644
--- a/net/atm/mpoa_caches.h
+++ b/net/atm/mpoa_caches.h
@@ -6,6 +6,7 @@
 #include <linux/atm.h>
 #include <linux/atmdev.h>
 #include <linux/atmmpc.h>
+#include <linux/refcount.h>
 
 struct mpoa_client;
 
@@ -25,7 +26,7 @@ typedef struct in_cache_entry {
 	struct   atm_vcc *shortcut;
 	uint8_t  MPS_ctrl_ATM_addr[ATM_ESA_LEN];
 	struct   in_ctrl_info ctrl_info;
-	atomic_t use;
+	refcount_t use;
 } in_cache_entry;
 
 struct in_cache_ops{
-- 
2.7.4

[PATCH 10/18] net, atm: convert eg_cache_entry.use from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/mpoa_caches.c | 14 +++++++-------
 net/atm/mpoa_caches.h |  2 +-
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/net/atm/mpoa_caches.c b/net/atm/mpoa_caches.c
index 05e89e9..4ccaa16 100644
--- a/net/atm/mpoa_caches.c
+++ b/net/atm/mpoa_caches.c
@@ -339,7 +339,7 @@ static eg_cache_entry *eg_cache_get_by_cache_id(__be32 cache_id,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.cache_id == cache_id) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irq(&mpc->egress_lock);
 			return entry;
 		}
@@ -360,7 +360,7 @@ static eg_cache_entry *eg_cache_get_by_tag(__be32 tag, struct mpoa_client *mpc)
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.tag == tag) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irqrestore(&mpc->egress_lock, flags);
 			return entry;
 		}
@@ -382,7 +382,7 @@ static eg_cache_entry *eg_cache_get_by_vcc(struct atm_vcc *vcc,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->shortcut == vcc) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irqrestore(&mpc->egress_lock, flags);
 			return entry;
 		}
@@ -402,7 +402,7 @@ static eg_cache_entry *eg_cache_get_by_src_ip(__be32 ipaddr,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->latest_ip_addr == ipaddr) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irq(&mpc->egress_lock);
 			return entry;
 		}
@@ -415,7 +415,7 @@ static eg_cache_entry *eg_cache_get_by_src_ip(__be32 ipaddr,
 
 static void eg_cache_put(eg_cache_entry *entry)
 {
-	if (atomic_dec_and_test(&entry->use)) {
+	if (refcount_dec_and_test(&entry->use)) {
 		memset(entry, 0, sizeof(eg_cache_entry));
 		kfree(entry);
 	}
@@ -468,7 +468,7 @@ static eg_cache_entry *eg_cache_add_entry(struct k_message *msg,
 	dprintk("adding an egress entry, ip = %pI4, this should be our IP\n",
 		&msg->content.eg_info.eg_dst_ip);
 
-	atomic_set(&entry->use, 1);
+	refcount_set(&entry->use, 1);
 	dprintk("new_eg_cache_entry: about to lock\n");
 	write_lock_irq(&client->egress_lock);
 	entry->next = client->eg_cache;
@@ -484,7 +484,7 @@ static eg_cache_entry *eg_cache_add_entry(struct k_message *msg,
 	dprintk("new_eg_cache_entry cache_id %u\n",
 		ntohl(entry->ctrl_info.cache_id));
 	dprintk("mps_ip = %pI4\n", &entry->ctrl_info.mps_ip);
-	atomic_inc(&entry->use);
+	refcount_inc(&entry->use);
 
 	write_unlock_irq(&client->egress_lock);
 	dprintk("new_eg_cache_entry: unlocked\n");
diff --git a/net/atm/mpoa_caches.h b/net/atm/mpoa_caches.h
index 38a4e7e..30fe348 100644
--- a/net/atm/mpoa_caches.h
+++ b/net/atm/mpoa_caches.h
@@ -59,7 +59,7 @@ typedef struct eg_cache_entry{
 	uint16_t             entry_state;
 	__be32             latest_ip_addr;    /* The src IP address of the last packet */
 	struct eg_ctrl_info  ctrl_info;
-	atomic_t             use;
+	refcount_t             use;
 } eg_cache_entry;
 
 struct eg_cache_ops{
-- 
2.7.4

[Bridge] [PATCH 10/18] net, atm: convert eg_cache_entry.use from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/atm/mpoa_caches.c | 14 +++++++-------
 net/atm/mpoa_caches.h |  2 +-
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/net/atm/mpoa_caches.c b/net/atm/mpoa_caches.c
index 05e89e9..4ccaa16 100644
--- a/net/atm/mpoa_caches.c
+++ b/net/atm/mpoa_caches.c
@@ -339,7 +339,7 @@ static eg_cache_entry *eg_cache_get_by_cache_id(__be32 cache_id,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.cache_id == cache_id) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irq(&mpc->egress_lock);
 			return entry;
 		}
@@ -360,7 +360,7 @@ static eg_cache_entry *eg_cache_get_by_tag(__be32 tag, struct mpoa_client *mpc)
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->ctrl_info.tag == tag) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irqrestore(&mpc->egress_lock, flags);
 			return entry;
 		}
@@ -382,7 +382,7 @@ static eg_cache_entry *eg_cache_get_by_vcc(struct atm_vcc *vcc,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->shortcut == vcc) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irqrestore(&mpc->egress_lock, flags);
 			return entry;
 		}
@@ -402,7 +402,7 @@ static eg_cache_entry *eg_cache_get_by_src_ip(__be32 ipaddr,
 	entry = mpc->eg_cache;
 	while (entry != NULL) {
 		if (entry->latest_ip_addr == ipaddr) {
-			atomic_inc(&entry->use);
+			refcount_inc(&entry->use);
 			read_unlock_irq(&mpc->egress_lock);
 			return entry;
 		}
@@ -415,7 +415,7 @@ static eg_cache_entry *eg_cache_get_by_src_ip(__be32 ipaddr,
 
 static void eg_cache_put(eg_cache_entry *entry)
 {
-	if (atomic_dec_and_test(&entry->use)) {
+	if (refcount_dec_and_test(&entry->use)) {
 		memset(entry, 0, sizeof(eg_cache_entry));
 		kfree(entry);
 	}
@@ -468,7 +468,7 @@ static eg_cache_entry *eg_cache_add_entry(struct k_message *msg,
 	dprintk("adding an egress entry, ip = %pI4, this should be our IP\n",
 		&msg->content.eg_info.eg_dst_ip);
 
-	atomic_set(&entry->use, 1);
+	refcount_set(&entry->use, 1);
 	dprintk("new_eg_cache_entry: about to lock\n");
 	write_lock_irq(&client->egress_lock);
 	entry->next = client->eg_cache;
@@ -484,7 +484,7 @@ static eg_cache_entry *eg_cache_add_entry(struct k_message *msg,
 	dprintk("new_eg_cache_entry cache_id %u\n",
 		ntohl(entry->ctrl_info.cache_id));
 	dprintk("mps_ip = %pI4\n", &entry->ctrl_info.mps_ip);
-	atomic_inc(&entry->use);
+	refcount_inc(&entry->use);
 
 	write_unlock_irq(&client->egress_lock);
 	dprintk("new_eg_cache_entry: unlocked\n");
diff --git a/net/atm/mpoa_caches.h b/net/atm/mpoa_caches.h
index 38a4e7e..30fe348 100644
--- a/net/atm/mpoa_caches.h
+++ b/net/atm/mpoa_caches.h
@@ -59,7 +59,7 @@ typedef struct eg_cache_entry{
 	uint16_t             entry_state;
 	__be32             latest_ip_addr;    /* The src IP address of the last packet */
 	struct eg_ctrl_info  ctrl_info;
-	atomic_t             use;
+	refcount_t             use;
 } eg_cache_entry;
 
 struct eg_cache_ops{
-- 
2.7.4

[PATCH 11/18] net, bridge: convert net_bridge_vlan.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/bridge/br_private.h | 3 ++-
 net/bridge/br_vlan.c    | 8 ++++----
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 2288fca..16cb470 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -21,6 +21,7 @@
 #include <net/ip6_fib.h>
 #include <linux/if_vlan.h>
 #include <linux/rhashtable.h>
+#include <linux/refcount.h>
 
 #define BR_HASH_BITS 8
 #define BR_HASH_SIZE (1 << BR_HASH_BITS)
@@ -127,7 +128,7 @@ struct net_bridge_vlan {
 		struct net_bridge_port	*port;
 	};
 	union {
-		atomic_t		refcnt;
+		refcount_t		refcnt;
 		struct net_bridge_vlan	*brvlan;
 	};
 
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index 62e68c0..e4e9e1a 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -158,7 +158,7 @@ static struct net_bridge_vlan *br_vlan_get_master(struct net_bridge *br, u16 vid
 		if (WARN_ON(!masterv))
 			return NULL;
 	}
-	atomic_inc(&masterv->refcnt);
+	refcount_inc(&masterv->refcnt);
 
 	return masterv;
 }
@@ -182,7 +182,7 @@ static void br_vlan_put_master(struct net_bridge_vlan *masterv)
 		return;
 
 	vg = br_vlan_group(masterv->br);
-	if (atomic_dec_and_test(&masterv->refcnt)) {
+	if (refcount_dec_and_test(&masterv->refcnt)) {
 		rhashtable_remove_fast(&vg->vlan_hash,
 				       &masterv->vnode, br_vlan_rht_params);
 		__vlan_del_list(masterv);
@@ -573,7 +573,7 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 				br_err(br, "failed insert local address into bridge forwarding table\n");
 				return ret;
 			}
-			atomic_inc(&vlan->refcnt);
+			refcount_inc(&vlan->refcnt);
 			vlan->flags |= BRIDGE_VLAN_INFO_BRENTRY;
 			vg->num_vlans++;
 		}
@@ -595,7 +595,7 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 	vlan->flags &= ~BRIDGE_VLAN_INFO_PVID;
 	vlan->br = br;
 	if (flags & BRIDGE_VLAN_INFO_BRENTRY)
-		atomic_set(&vlan->refcnt, 1);
+		refcount_set(&vlan->refcnt, 1);
 	ret = __vlan_add(vlan, flags);
 	if (ret) {
 		free_percpu(vlan->stats);
-- 
2.7.4

[Bridge] [PATCH 11/18] net, bridge: convert net_bridge_vlan.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 net/bridge/br_private.h | 3 ++-
 net/bridge/br_vlan.c    | 8 ++++----
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
index 2288fca..16cb470 100644
--- a/net/bridge/br_private.h
+++ b/net/bridge/br_private.h
@@ -21,6 +21,7 @@
 #include <net/ip6_fib.h>
 #include <linux/if_vlan.h>
 #include <linux/rhashtable.h>
+#include <linux/refcount.h>
 
 #define BR_HASH_BITS 8
 #define BR_HASH_SIZE (1 << BR_HASH_BITS)
@@ -127,7 +128,7 @@ struct net_bridge_vlan {
 		struct net_bridge_port	*port;
 	};
 	union {
-		atomic_t		refcnt;
+		refcount_t		refcnt;
 		struct net_bridge_vlan	*brvlan;
 	};
 
diff --git a/net/bridge/br_vlan.c b/net/bridge/br_vlan.c
index 62e68c0..e4e9e1a 100644
--- a/net/bridge/br_vlan.c
+++ b/net/bridge/br_vlan.c
@@ -158,7 +158,7 @@ static struct net_bridge_vlan *br_vlan_get_master(struct net_bridge *br, u16 vid
 		if (WARN_ON(!masterv))
 			return NULL;
 	}
-	atomic_inc(&masterv->refcnt);
+	refcount_inc(&masterv->refcnt);
 
 	return masterv;
 }
@@ -182,7 +182,7 @@ static void br_vlan_put_master(struct net_bridge_vlan *masterv)
 		return;
 
 	vg = br_vlan_group(masterv->br);
-	if (atomic_dec_and_test(&masterv->refcnt)) {
+	if (refcount_dec_and_test(&masterv->refcnt)) {
 		rhashtable_remove_fast(&vg->vlan_hash,
 				       &masterv->vnode, br_vlan_rht_params);
 		__vlan_del_list(masterv);
@@ -573,7 +573,7 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 				br_err(br, "failed insert local address into bridge forwarding table\n");
 				return ret;
 			}
-			atomic_inc(&vlan->refcnt);
+			refcount_inc(&vlan->refcnt);
 			vlan->flags |= BRIDGE_VLAN_INFO_BRENTRY;
 			vg->num_vlans++;
 		}
@@ -595,7 +595,7 @@ int br_vlan_add(struct net_bridge *br, u16 vid, u16 flags)
 	vlan->flags &= ~BRIDGE_VLAN_INFO_PVID;
 	vlan->br = br;
 	if (flags & BRIDGE_VLAN_INFO_BRENTRY)
-		atomic_set(&vlan->refcnt, 1);
+		refcount_set(&vlan->refcnt, 1);
 	ret = __vlan_add(vlan, flags);
 	if (ret) {
 		free_percpu(vlan->stats);
-- 
2.7.4

[PATCH 12/18] net, calipso: convert calipso_doi.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/calipso.h |  4 ++--
 net/ipv6/calipso.c    | 12 ++++++------
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/include/net/calipso.h b/include/net/calipso.h
index b1b30cd..5f95b11 100644
--- a/include/net/calipso.h
+++ b/include/net/calipso.h
@@ -38,7 +38,7 @@
 #include <linux/skbuff.h>
 #include <net/netlabel.h>
 #include <net/request_sock.h>
-#include <linux/atomic.h>
+#include <linux/refcount.h>
 #include <asm/unaligned.h>
 
 /* known doi values */
@@ -57,7 +57,7 @@ struct calipso_doi {
 	u32 doi;
 	u32 type;
 
-	atomic_t refcount;
+	refcount_t refcount;
 	struct list_head list;
 	struct rcu_head rcu;
 };
diff --git a/net/ipv6/calipso.c b/net/ipv6/calipso.c
index 37ac9de..997611f 100644
--- a/net/ipv6/calipso.c
+++ b/net/ipv6/calipso.c
@@ -338,7 +338,7 @@ static struct calipso_doi *calipso_doi_search(u32 doi)
 	struct calipso_doi *iter;
 
 	list_for_each_entry_rcu(iter, &calipso_doi_list, list)
-		if (iter->doi == doi && atomic_read(&iter->refcount))
+		if (iter->doi == doi && refcount_read(&iter->refcount))
 			return iter;
 	return NULL;
 }
@@ -370,7 +370,7 @@ static int calipso_doi_add(struct calipso_doi *doi_def,
 	if (doi_def->doi == CALIPSO_DOI_UNKNOWN)
 		goto doi_add_return;
 
-	atomic_set(&doi_def->refcount, 1);
+	refcount_set(&doi_def->refcount, 1);
 
 	spin_lock(&calipso_doi_list_lock);
 	if (calipso_doi_search(doi_def->doi)) {
@@ -458,7 +458,7 @@ static int calipso_doi_remove(u32 doi, struct netlbl_audit *audit_info)
 		ret_val = -ENOENT;
 		goto doi_remove_return;
 	}
-	if (!atomic_dec_and_test(&doi_def->refcount)) {
+	if (!refcount_dec_and_test(&doi_def->refcount)) {
 		spin_unlock(&calipso_doi_list_lock);
 		ret_val = -EBUSY;
 		goto doi_remove_return;
@@ -499,7 +499,7 @@ static struct calipso_doi *calipso_doi_getdef(u32 doi)
 	doi_def = calipso_doi_search(doi);
 	if (!doi_def)
 		goto doi_getdef_return;
-	if (!atomic_inc_not_zero(&doi_def->refcount))
+	if (!refcount_inc_not_zero(&doi_def->refcount))
 		doi_def = NULL;
 
 doi_getdef_return:
@@ -520,7 +520,7 @@ static void calipso_doi_putdef(struct calipso_doi *doi_def)
 	if (!doi_def)
 		return;
 
-	if (!atomic_dec_and_test(&doi_def->refcount))
+	if (!refcount_dec_and_test(&doi_def->refcount))
 		return;
 	spin_lock(&calipso_doi_list_lock);
 	list_del_rcu(&doi_def->list);
@@ -553,7 +553,7 @@ static int calipso_doi_walk(u32 *skip_cnt,
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(iter_doi, &calipso_doi_list, list)
-		if (atomic_read(&iter_doi->refcount) > 0) {
+		if (refcount_read(&iter_doi->refcount) > 0) {
 			if (doi_cnt++ < *skip_cnt)
 				continue;
 			ret_val = callback(iter_doi, cb_arg);
-- 
2.7.4

[Bridge] [PATCH 12/18] net, calipso: convert calipso_doi.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/calipso.h |  4 ++--
 net/ipv6/calipso.c    | 12 ++++++------
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/include/net/calipso.h b/include/net/calipso.h
index b1b30cd..5f95b11 100644
--- a/include/net/calipso.h
+++ b/include/net/calipso.h
@@ -38,7 +38,7 @@
 #include <linux/skbuff.h>
 #include <net/netlabel.h>
 #include <net/request_sock.h>
-#include <linux/atomic.h>
+#include <linux/refcount.h>
 #include <asm/unaligned.h>
 
 /* known doi values */
@@ -57,7 +57,7 @@ struct calipso_doi {
 	u32 doi;
 	u32 type;
 
-	atomic_t refcount;
+	refcount_t refcount;
 	struct list_head list;
 	struct rcu_head rcu;
 };
diff --git a/net/ipv6/calipso.c b/net/ipv6/calipso.c
index 37ac9de..997611f 100644
--- a/net/ipv6/calipso.c
+++ b/net/ipv6/calipso.c
@@ -338,7 +338,7 @@ static struct calipso_doi *calipso_doi_search(u32 doi)
 	struct calipso_doi *iter;
 
 	list_for_each_entry_rcu(iter, &calipso_doi_list, list)
-		if (iter->doi == doi && atomic_read(&iter->refcount))
+		if (iter->doi == doi && refcount_read(&iter->refcount))
 			return iter;
 	return NULL;
 }
@@ -370,7 +370,7 @@ static int calipso_doi_add(struct calipso_doi *doi_def,
 	if (doi_def->doi == CALIPSO_DOI_UNKNOWN)
 		goto doi_add_return;
 
-	atomic_set(&doi_def->refcount, 1);
+	refcount_set(&doi_def->refcount, 1);
 
 	spin_lock(&calipso_doi_list_lock);
 	if (calipso_doi_search(doi_def->doi)) {
@@ -458,7 +458,7 @@ static int calipso_doi_remove(u32 doi, struct netlbl_audit *audit_info)
 		ret_val = -ENOENT;
 		goto doi_remove_return;
 	}
-	if (!atomic_dec_and_test(&doi_def->refcount)) {
+	if (!refcount_dec_and_test(&doi_def->refcount)) {
 		spin_unlock(&calipso_doi_list_lock);
 		ret_val = -EBUSY;
 		goto doi_remove_return;
@@ -499,7 +499,7 @@ static struct calipso_doi *calipso_doi_getdef(u32 doi)
 	doi_def = calipso_doi_search(doi);
 	if (!doi_def)
 		goto doi_getdef_return;
-	if (!atomic_inc_not_zero(&doi_def->refcount))
+	if (!refcount_inc_not_zero(&doi_def->refcount))
 		doi_def = NULL;
 
 doi_getdef_return:
@@ -520,7 +520,7 @@ static void calipso_doi_putdef(struct calipso_doi *doi_def)
 	if (!doi_def)
 		return;
 
-	if (!atomic_dec_and_test(&doi_def->refcount))
+	if (!refcount_dec_and_test(&doi_def->refcount))
 		return;
 	spin_lock(&calipso_doi_list_lock);
 	list_del_rcu(&doi_def->list);
@@ -553,7 +553,7 @@ static int calipso_doi_walk(u32 *skip_cnt,
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(iter_doi, &calipso_doi_list, list)
-		if (atomic_read(&iter_doi->refcount) > 0) {
+		if (refcount_read(&iter_doi->refcount) > 0) {
 			if (doi_cnt++ < *skip_cnt)
 				continue;
 			ret_val = callback(iter_doi, cb_arg);
-- 
2.7.4

[PATCH 13/18] net, sched: convert Qdisc.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/sch_generic.h | 3 ++-
 net/sched/sch_api.c       | 8 ++++----
 net/sched/sch_generic.c   | 8 ++++----
 3 files changed, 10 insertions(+), 9 deletions(-)

diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index aeec408..cd7463f 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -8,6 +8,7 @@
 #include <linux/pkt_cls.h>
 #include <linux/percpu.h>
 #include <linux/dynamic_queue_limits.h>
+#include <linux/refcount.h>
 #include <net/gen_stats.h>
 #include <net/rtnetlink.h>
 
@@ -93,7 +94,7 @@ struct Qdisc {
 	struct sk_buff		*skb_bad_txq;
 	struct rcu_head		rcu_head;
 	int			padded;
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 
 	spinlock_t		busylock ____cacheline_aligned_in_smp;
 };
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index bcf49cd..806f7b5 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -828,7 +828,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
 
 			old = dev_graft_qdisc(dev_queue, new);
 			if (new && i > 0)
-				atomic_inc(&new->refcnt);
+				refcount_inc(&new->refcnt);
 
 			if (!ingress)
 				qdisc_destroy(old);
@@ -839,7 +839,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
 			notify_and_destroy(net, skb, n, classid,
 					   dev->qdisc, new);
 			if (new && !new->ops->attach)
-				atomic_inc(&new->refcnt);
+				refcount_inc(&new->refcnt);
 			dev->qdisc = new ? : &noop_qdisc;
 
 			if (new && new->ops->attach)
@@ -1245,7 +1245,7 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n)
 				if (q == p ||
 				    (p && check_loop(q, p, 0)))
 					return -ELOOP;
-				atomic_inc(&q->refcnt);
+				refcount_inc(&q->refcnt);
 				goto graft;
 			} else {
 				if (!q)
@@ -1360,7 +1360,7 @@ static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
 	tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
 	tcm->tcm_parent = clid;
 	tcm->tcm_handle = q->handle;
-	tcm->tcm_info = atomic_read(&q->refcnt);
+	tcm->tcm_info = refcount_read(&q->refcnt);
 	if (nla_put_string(skb, TCA_KIND, q->ops->id))
 		goto nla_put_failure;
 	if (q->ops->dump && q->ops->dump(q, skb) < 0)
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index b052b27..6e0c695 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -633,7 +633,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	sch->dequeue = ops->dequeue;
 	sch->dev_queue = dev_queue;
 	dev_hold(dev);
-	atomic_set(&sch->refcnt, 1);
+	refcount_set(&sch->refcnt, 1);
 
 	return sch;
 errout:
@@ -701,7 +701,7 @@ void qdisc_destroy(struct Qdisc *qdisc)
 	const struct Qdisc_ops  *ops = qdisc->ops;
 
 	if (qdisc->flags & TCQ_F_BUILTIN ||
-	    !atomic_dec_and_test(&qdisc->refcnt))
+	    !refcount_dec_and_test(&qdisc->refcnt))
 		return;
 
 #ifdef CONFIG_NET_SCHED
@@ -739,7 +739,7 @@ struct Qdisc *dev_graft_qdisc(struct netdev_queue *dev_queue,
 	spin_lock_bh(root_lock);
 
 	/* Prune old scheduler */
-	if (oqdisc && atomic_read(&oqdisc->refcnt) <= 1)
+	if (oqdisc && refcount_read(&oqdisc->refcnt) <= 1)
 		qdisc_reset(oqdisc);
 
 	/* ... and graft new one */
@@ -785,7 +785,7 @@ static void attach_default_qdiscs(struct net_device *dev)
 	    dev->priv_flags & IFF_NO_QUEUE) {
 		netdev_for_each_tx_queue(dev, attach_one_default_qdisc, NULL);
 		dev->qdisc = txq->qdisc_sleeping;
-		atomic_inc(&dev->qdisc->refcnt);
+		refcount_inc(&dev->qdisc->refcnt);
 	} else {
 		qdisc = qdisc_create_dflt(txq, &mq_qdisc_ops, TC_H_ROOT);
 		if (qdisc) {
-- 
2.7.4

[Bridge] [PATCH 13/18] net, sched: convert Qdisc.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/sch_generic.h | 3 ++-
 net/sched/sch_api.c       | 8 ++++----
 net/sched/sch_generic.c   | 8 ++++----
 3 files changed, 10 insertions(+), 9 deletions(-)

diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index aeec408..cd7463f 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -8,6 +8,7 @@
 #include <linux/pkt_cls.h>
 #include <linux/percpu.h>
 #include <linux/dynamic_queue_limits.h>
+#include <linux/refcount.h>
 #include <net/gen_stats.h>
 #include <net/rtnetlink.h>
 
@@ -93,7 +94,7 @@ struct Qdisc {
 	struct sk_buff		*skb_bad_txq;
 	struct rcu_head		rcu_head;
 	int			padded;
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 
 	spinlock_t		busylock ____cacheline_aligned_in_smp;
 };
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index bcf49cd..806f7b5 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -828,7 +828,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
 
 			old = dev_graft_qdisc(dev_queue, new);
 			if (new && i > 0)
-				atomic_inc(&new->refcnt);
+				refcount_inc(&new->refcnt);
 
 			if (!ingress)
 				qdisc_destroy(old);
@@ -839,7 +839,7 @@ static int qdisc_graft(struct net_device *dev, struct Qdisc *parent,
 			notify_and_destroy(net, skb, n, classid,
 					   dev->qdisc, new);
 			if (new && !new->ops->attach)
-				atomic_inc(&new->refcnt);
+				refcount_inc(&new->refcnt);
 			dev->qdisc = new ? : &noop_qdisc;
 
 			if (new && new->ops->attach)
@@ -1245,7 +1245,7 @@ static int tc_modify_qdisc(struct sk_buff *skb, struct nlmsghdr *n)
 				if (q == p ||
 				    (p && check_loop(q, p, 0)))
 					return -ELOOP;
-				atomic_inc(&q->refcnt);
+				refcount_inc(&q->refcnt);
 				goto graft;
 			} else {
 				if (!q)
@@ -1360,7 +1360,7 @@ static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
 	tcm->tcm_ifindex = qdisc_dev(q)->ifindex;
 	tcm->tcm_parent = clid;
 	tcm->tcm_handle = q->handle;
-	tcm->tcm_info = atomic_read(&q->refcnt);
+	tcm->tcm_info = refcount_read(&q->refcnt);
 	if (nla_put_string(skb, TCA_KIND, q->ops->id))
 		goto nla_put_failure;
 	if (q->ops->dump && q->ops->dump(q, skb) < 0)
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index b052b27..6e0c695 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -633,7 +633,7 @@ struct Qdisc *qdisc_alloc(struct netdev_queue *dev_queue,
 	sch->dequeue = ops->dequeue;
 	sch->dev_queue = dev_queue;
 	dev_hold(dev);
-	atomic_set(&sch->refcnt, 1);
+	refcount_set(&sch->refcnt, 1);
 
 	return sch;
 errout:
@@ -701,7 +701,7 @@ void qdisc_destroy(struct Qdisc *qdisc)
 	const struct Qdisc_ops  *ops = qdisc->ops;
 
 	if (qdisc->flags & TCQ_F_BUILTIN ||
-	    !atomic_dec_and_test(&qdisc->refcnt))
+	    !refcount_dec_and_test(&qdisc->refcnt))
 		return;
 
 #ifdef CONFIG_NET_SCHED
@@ -739,7 +739,7 @@ struct Qdisc *dev_graft_qdisc(struct netdev_queue *dev_queue,
 	spin_lock_bh(root_lock);
 
 	/* Prune old scheduler */
-	if (oqdisc && atomic_read(&oqdisc->refcnt) <= 1)
+	if (oqdisc && refcount_read(&oqdisc->refcnt) <= 1)
 		qdisc_reset(oqdisc);
 
 	/* ... and graft new one */
@@ -785,7 +785,7 @@ static void attach_default_qdiscs(struct net_device *dev)
 	    dev->priv_flags & IFF_NO_QUEUE) {
 		netdev_for_each_tx_queue(dev, attach_one_default_qdisc, NULL);
 		dev->qdisc = txq->qdisc_sleeping;
-		atomic_inc(&dev->qdisc->refcnt);
+		refcount_inc(&dev->qdisc->refcnt);
 	} else {
 		qdisc = qdisc_create_dflt(txq, &mq_qdisc_ops, TC_H_ROOT);
 		if (qdisc) {
-- 
2.7.4

[PATCH 14/18] net, lapb: convert lapb_cb.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/lapb.h    | 3 ++-
 net/lapb/lapb_iface.c | 6 +++---
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/include/net/lapb.h b/include/net/lapb.h
index 9510f87..85e7737 100644
--- a/include/net/lapb.h
+++ b/include/net/lapb.h
@@ -1,6 +1,7 @@
 #ifndef _LAPB_H
 #define _LAPB_H 
 #include <linux/lapb.h>
+#include <linux/refcount.h>
 
 #define	LAPB_HEADER_LEN	20		/* LAPB over Ethernet + a bit more */
 
@@ -101,7 +102,7 @@ struct lapb_cb {
 	struct lapb_frame	frmr_data;
 	unsigned char		frmr_type;
 
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 };
 
 /* lapb_iface.c */
diff --git a/net/lapb/lapb_iface.c b/net/lapb/lapb_iface.c
index b50b64a..e15314e 100644
--- a/net/lapb/lapb_iface.c
+++ b/net/lapb/lapb_iface.c
@@ -54,12 +54,12 @@ static void lapb_free_cb(struct lapb_cb *lapb)
 
 static __inline__ void lapb_hold(struct lapb_cb *lapb)
 {
-	atomic_inc(&lapb->refcnt);
+	refcount_inc(&lapb->refcnt);
 }
 
 static __inline__ void lapb_put(struct lapb_cb *lapb)
 {
-	if (atomic_dec_and_test(&lapb->refcnt))
+	if (refcount_dec_and_test(&lapb->refcnt))
 		lapb_free_cb(lapb);
 }
 
@@ -136,7 +136,7 @@ static struct lapb_cb *lapb_create_cb(void)
 	lapb->mode    = LAPB_DEFAULT_MODE;
 	lapb->window  = LAPB_DEFAULT_WINDOW;
 	lapb->state   = LAPB_STATE_0;
-	atomic_set(&lapb->refcnt, 1);
+	refcount_set(&lapb->refcnt, 1);
 out:
 	return lapb;
 }
-- 
2.7.4

[Bridge] [PATCH 14/18] net, lapb: convert lapb_cb.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/lapb.h    | 3 ++-
 net/lapb/lapb_iface.c | 6 +++---
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/include/net/lapb.h b/include/net/lapb.h
index 9510f87..85e7737 100644
--- a/include/net/lapb.h
+++ b/include/net/lapb.h
@@ -1,6 +1,7 @@
 #ifndef _LAPB_H
 #define _LAPB_H 
 #include <linux/lapb.h>
+#include <linux/refcount.h>
 
 #define	LAPB_HEADER_LEN	20		/* LAPB over Ethernet + a bit more */
 
@@ -101,7 +102,7 @@ struct lapb_cb {
 	struct lapb_frame	frmr_data;
 	unsigned char		frmr_type;
 
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 };
 
 /* lapb_iface.c */
diff --git a/net/lapb/lapb_iface.c b/net/lapb/lapb_iface.c
index b50b64a..e15314e 100644
--- a/net/lapb/lapb_iface.c
+++ b/net/lapb/lapb_iface.c
@@ -54,12 +54,12 @@ static void lapb_free_cb(struct lapb_cb *lapb)
 
 static __inline__ void lapb_hold(struct lapb_cb *lapb)
 {
-	atomic_inc(&lapb->refcnt);
+	refcount_inc(&lapb->refcnt);
 }
 
 static __inline__ void lapb_put(struct lapb_cb *lapb)
 {
-	if (atomic_dec_and_test(&lapb->refcnt))
+	if (refcount_dec_and_test(&lapb->refcnt))
 		lapb_free_cb(lapb);
 }
 
@@ -136,7 +136,7 @@ static struct lapb_cb *lapb_create_cb(void)
 	lapb->mode    = LAPB_DEFAULT_MODE;
 	lapb->window  = LAPB_DEFAULT_WINDOW;
 	lapb->state   = LAPB_STATE_0;
-	atomic_set(&lapb->refcnt, 1);
+	refcount_set(&lapb->refcnt, 1);
 out:
 	return lapb;
 }
-- 
2.7.4

[PATCH 15/18] net, ipx: convert ipx_interface.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/ipx.h  | 7 ++++---
 net/ipx/af_ipx.c   | 6 +++---
 net/ipx/ipx_proc.c | 2 +-
 3 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/include/net/ipx.h b/include/net/ipx.h
index e5cff68..2de1281 100644
--- a/include/net/ipx.h
+++ b/include/net/ipx.h
@@ -14,6 +14,7 @@
 #include <linux/ipx.h>
 #include <linux/list.h>
 #include <linux/slab.h>
+#include <linux/refcount.h>
 
 struct ipx_address {
 	__be32  net;
@@ -54,7 +55,7 @@ struct ipx_interface {
 	/* IPX address */
 	__be32			if_netnum;
 	unsigned char		if_node[IPX_NODE_LEN];
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 
 	/* physical device info */
 	struct net_device	*if_dev;
@@ -139,7 +140,7 @@ const char *ipx_device_name(struct ipx_interface *intrfc);
 
 static __inline__ void ipxitf_hold(struct ipx_interface *intrfc)
 {
-	atomic_inc(&intrfc->refcnt);
+	refcount_inc(&intrfc->refcnt);
 }
 
 void ipxitf_down(struct ipx_interface *intrfc);
@@ -157,7 +158,7 @@ int ipxrtr_ioctl(unsigned int cmd, void __user *arg);
 
 static __inline__ void ipxitf_put(struct ipx_interface *intrfc)
 {
-	if (atomic_dec_and_test(&intrfc->refcnt))
+	if (refcount_dec_and_test(&intrfc->refcnt))
 		ipxitf_down(intrfc);
 }
 
diff --git a/net/ipx/af_ipx.c b/net/ipx/af_ipx.c
index 8a9219f..0675b4a 100644
--- a/net/ipx/af_ipx.c
+++ b/net/ipx/af_ipx.c
@@ -308,7 +308,7 @@ void ipxitf_down(struct ipx_interface *intrfc)
 
 static void __ipxitf_put(struct ipx_interface *intrfc)
 {
-	if (atomic_dec_and_test(&intrfc->refcnt))
+	if (refcount_dec_and_test(&intrfc->refcnt))
 		__ipxitf_down(intrfc);
 }
 
@@ -876,7 +876,7 @@ static struct ipx_interface *ipxitf_alloc(struct net_device *dev, __be32 netnum,
 		intrfc->if_ipx_offset 	= ipx_offset;
 		intrfc->if_sknum 	= IPX_MIN_EPHEMERAL_SOCKET;
 		INIT_HLIST_HEAD(&intrfc->if_sklist);
-		atomic_set(&intrfc->refcnt, 1);
+		refcount_set(&intrfc->refcnt, 1);
 		spin_lock_init(&intrfc->if_sklist_lock);
 	}
 
@@ -1105,7 +1105,7 @@ static struct ipx_interface *ipxitf_auto_create(struct net_device *dev,
 		memcpy((char *)&(intrfc->if_node[IPX_NODE_LEN-dev->addr_len]),
 			dev->dev_addr, dev->addr_len);
 		spin_lock_init(&intrfc->if_sklist_lock);
-		atomic_set(&intrfc->refcnt, 1);
+		refcount_set(&intrfc->refcnt, 1);
 		ipxitf_insert(intrfc);
 		dev_hold(dev);
 	}
diff --git a/net/ipx/ipx_proc.c b/net/ipx/ipx_proc.c
index c1d247e..7d75e4c 100644
--- a/net/ipx/ipx_proc.c
+++ b/net/ipx/ipx_proc.c
@@ -53,7 +53,7 @@ static int ipx_seq_interface_show(struct seq_file *seq, void *v)
 	seq_printf(seq, "%-11s", ipx_device_name(i));
 	seq_printf(seq, "%-9s", ipx_frame_name(i->if_dlink_type));
 #ifdef IPX_REFCNT_DEBUG
-	seq_printf(seq, "%6d", atomic_read(&i->refcnt));
+	seq_printf(seq, "%6d", refcount_read(&i->refcnt));
 #endif
 	seq_puts(seq, "\n");
 out:
-- 
2.7.4

[Bridge] [PATCH 15/18] net, ipx: convert ipx_interface.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/ipx.h  | 7 ++++---
 net/ipx/af_ipx.c   | 6 +++---
 net/ipx/ipx_proc.c | 2 +-
 3 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/include/net/ipx.h b/include/net/ipx.h
index e5cff68..2de1281 100644
--- a/include/net/ipx.h
+++ b/include/net/ipx.h
@@ -14,6 +14,7 @@
 #include <linux/ipx.h>
 #include <linux/list.h>
 #include <linux/slab.h>
+#include <linux/refcount.h>
 
 struct ipx_address {
 	__be32  net;
@@ -54,7 +55,7 @@ struct ipx_interface {
 	/* IPX address */
 	__be32			if_netnum;
 	unsigned char		if_node[IPX_NODE_LEN];
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 
 	/* physical device info */
 	struct net_device	*if_dev;
@@ -139,7 +140,7 @@ const char *ipx_device_name(struct ipx_interface *intrfc);
 
 static __inline__ void ipxitf_hold(struct ipx_interface *intrfc)
 {
-	atomic_inc(&intrfc->refcnt);
+	refcount_inc(&intrfc->refcnt);
 }
 
 void ipxitf_down(struct ipx_interface *intrfc);
@@ -157,7 +158,7 @@ int ipxrtr_ioctl(unsigned int cmd, void __user *arg);
 
 static __inline__ void ipxitf_put(struct ipx_interface *intrfc)
 {
-	if (atomic_dec_and_test(&intrfc->refcnt))
+	if (refcount_dec_and_test(&intrfc->refcnt))
 		ipxitf_down(intrfc);
 }
 
diff --git a/net/ipx/af_ipx.c b/net/ipx/af_ipx.c
index 8a9219f..0675b4a 100644
--- a/net/ipx/af_ipx.c
+++ b/net/ipx/af_ipx.c
@@ -308,7 +308,7 @@ void ipxitf_down(struct ipx_interface *intrfc)
 
 static void __ipxitf_put(struct ipx_interface *intrfc)
 {
-	if (atomic_dec_and_test(&intrfc->refcnt))
+	if (refcount_dec_and_test(&intrfc->refcnt))
 		__ipxitf_down(intrfc);
 }
 
@@ -876,7 +876,7 @@ static struct ipx_interface *ipxitf_alloc(struct net_device *dev, __be32 netnum,
 		intrfc->if_ipx_offset 	= ipx_offset;
 		intrfc->if_sknum 	= IPX_MIN_EPHEMERAL_SOCKET;
 		INIT_HLIST_HEAD(&intrfc->if_sklist);
-		atomic_set(&intrfc->refcnt, 1);
+		refcount_set(&intrfc->refcnt, 1);
 		spin_lock_init(&intrfc->if_sklist_lock);
 	}
 
@@ -1105,7 +1105,7 @@ static struct ipx_interface *ipxitf_auto_create(struct net_device *dev,
 		memcpy((char *)&(intrfc->if_node[IPX_NODE_LEN-dev->addr_len]),
 			dev->dev_addr, dev->addr_len);
 		spin_lock_init(&intrfc->if_sklist_lock);
-		atomic_set(&intrfc->refcnt, 1);
+		refcount_set(&intrfc->refcnt, 1);
 		ipxitf_insert(intrfc);
 		dev_hold(dev);
 	}
diff --git a/net/ipx/ipx_proc.c b/net/ipx/ipx_proc.c
index c1d247e..7d75e4c 100644
--- a/net/ipx/ipx_proc.c
+++ b/net/ipx/ipx_proc.c
@@ -53,7 +53,7 @@ static int ipx_seq_interface_show(struct seq_file *seq, void *v)
 	seq_printf(seq, "%-11s", ipx_device_name(i));
 	seq_printf(seq, "%-9s", ipx_frame_name(i->if_dlink_type));
 #ifdef IPX_REFCNT_DEBUG
-	seq_printf(seq, "%6d", atomic_read(&i->refcnt));
+	seq_printf(seq, "%6d", refcount_read(&i->refcnt));
 #endif
 	seq_puts(seq, "\n");
 out:
-- 
2.7.4

[PATCH 16/18] net, ipx: convert ipx_route.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/ipx.h   | 6 +++---
 net/ipx/ipx_route.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/ipx.h b/include/net/ipx.h
index 2de1281..af32b97 100644
--- a/include/net/ipx.h
+++ b/include/net/ipx.h
@@ -81,7 +81,7 @@ struct ipx_route {
 	unsigned char		ir_routed;
 	unsigned char		ir_router_node[IPX_NODE_LEN];
 	struct list_head	node; /* node in ipx_routes list */
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 };
 
 struct ipx_cb {
@@ -164,12 +164,12 @@ static __inline__ void ipxitf_put(struct ipx_interface *intrfc)
 
 static __inline__ void ipxrtr_hold(struct ipx_route *rt)
 {
-	        atomic_inc(&rt->refcnt);
+	        refcount_inc(&rt->refcnt);
 }
 
 static __inline__ void ipxrtr_put(struct ipx_route *rt)
 {
-	        if (atomic_dec_and_test(&rt->refcnt))
+	        if (refcount_dec_and_test(&rt->refcnt))
 			                kfree(rt);
 }
 #endif /* _NET_INET_IPX_H_ */
diff --git a/net/ipx/ipx_route.c b/net/ipx/ipx_route.c
index 3e2a32a..b5d9144 100644
--- a/net/ipx/ipx_route.c
+++ b/net/ipx/ipx_route.c
@@ -59,7 +59,7 @@ int ipxrtr_add_route(__be32 network, struct ipx_interface *intrfc,
 		if (!rt)
 			goto out;
 
-		atomic_set(&rt->refcnt, 1);
+		refcount_set(&rt->refcnt, 1);
 		ipxrtr_hold(rt);
 		write_lock_bh(&ipx_routes_lock);
 		list_add(&rt->node, &ipx_routes);
-- 
2.7.4

[Bridge] [PATCH 16/18] net, ipx: convert ipx_route.refcnt from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/ipx.h   | 6 +++---
 net/ipx/ipx_route.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/ipx.h b/include/net/ipx.h
index 2de1281..af32b97 100644
--- a/include/net/ipx.h
+++ b/include/net/ipx.h
@@ -81,7 +81,7 @@ struct ipx_route {
 	unsigned char		ir_routed;
 	unsigned char		ir_router_node[IPX_NODE_LEN];
 	struct list_head	node; /* node in ipx_routes list */
-	atomic_t		refcnt;
+	refcount_t		refcnt;
 };
 
 struct ipx_cb {
@@ -164,12 +164,12 @@ static __inline__ void ipxitf_put(struct ipx_interface *intrfc)
 
 static __inline__ void ipxrtr_hold(struct ipx_route *rt)
 {
-	        atomic_inc(&rt->refcnt);
+	        refcount_inc(&rt->refcnt);
 }
 
 static __inline__ void ipxrtr_put(struct ipx_route *rt)
 {
-	        if (atomic_dec_and_test(&rt->refcnt))
+	        if (refcount_dec_and_test(&rt->refcnt))
 			                kfree(rt);
 }
 #endif /* _NET_INET_IPX_H_ */
diff --git a/net/ipx/ipx_route.c b/net/ipx/ipx_route.c
index 3e2a32a..b5d9144 100644
--- a/net/ipx/ipx_route.c
+++ b/net/ipx/ipx_route.c
@@ -59,7 +59,7 @@ int ipxrtr_add_route(__be32 network, struct ipx_interface *intrfc,
 		if (!rt)
 			goto out;
 
-		atomic_set(&rt->refcnt, 1);
+		refcount_set(&rt->refcnt, 1);
 		ipxrtr_hold(rt);
 		write_lock_bh(&ipx_routes_lock);
 		list_add(&rt->node, &ipx_routes);
-- 
2.7.4

[PATCH 17/18] net, netrom: convert nr_neigh.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/netrom.h  | 7 ++++---
 net/netrom/nr_route.c | 4 ++--
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/net/netrom.h b/include/net/netrom.h
index 110350a..cecb4fd 100644
--- a/include/net/netrom.h
+++ b/include/net/netrom.h
@@ -11,6 +11,7 @@
 #include <linux/list.h>
 #include <linux/slab.h>
 #include <net/sock.h>
+#include <linux/refcount.h>
 
 #define	NR_NETWORK_LEN			15
 #define	NR_TRANSPORT_LEN		5
@@ -93,7 +94,7 @@ struct nr_neigh {
 	unsigned short		count;
 	unsigned int		number;
 	unsigned char		failed;
-	atomic_t		refcount;
+	refcount_t		refcount;
 };
 
 struct nr_route {
@@ -128,11 +129,11 @@ static __inline__ void nr_node_put(struct nr_node *nr_node)
 }
 
 #define nr_neigh_hold(__nr_neigh) \
-	atomic_inc(&((__nr_neigh)->refcount))
+	refcount_inc(&((__nr_neigh)->refcount))
 
 static __inline__ void nr_neigh_put(struct nr_neigh *nr_neigh)
 {
-	if (atomic_dec_and_test(&nr_neigh->refcount)) {
+	if (refcount_dec_and_test(&nr_neigh->refcount)) {
 		if (nr_neigh->ax25)
 			ax25_cb_put(nr_neigh->ax25);
 		kfree(nr_neigh->digipeat);
diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index d72a4f1..6b72970 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -149,7 +149,7 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
 		nr_neigh->count    = 0;
 		nr_neigh->number   = nr_neigh_no++;
 		nr_neigh->failed   = 0;
-		atomic_set(&nr_neigh->refcount, 1);
+		refcount_set(&nr_neigh->refcount, 1);
 
 		if (ax25_digi != NULL && ax25_digi->ndigi > 0) {
 			nr_neigh->digipeat = kmemdup(ax25_digi,
@@ -431,7 +431,7 @@ static int __must_check nr_add_neigh(ax25_address *callsign,
 	nr_neigh->count    = 0;
 	nr_neigh->number   = nr_neigh_no++;
 	nr_neigh->failed   = 0;
-	atomic_set(&nr_neigh->refcount, 1);
+	refcount_set(&nr_neigh->refcount, 1);
 
 	if (ax25_digi != NULL && ax25_digi->ndigi > 0) {
 		nr_neigh->digipeat = kmemdup(ax25_digi, sizeof(*ax25_digi),
-- 
2.7.4

[Bridge] [PATCH 17/18] net, netrom: convert nr_neigh.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/netrom.h  | 7 ++++---
 net/netrom/nr_route.c | 4 ++--
 2 files changed, 6 insertions(+), 5 deletions(-)

diff --git a/include/net/netrom.h b/include/net/netrom.h
index 110350a..cecb4fd 100644
--- a/include/net/netrom.h
+++ b/include/net/netrom.h
@@ -11,6 +11,7 @@
 #include <linux/list.h>
 #include <linux/slab.h>
 #include <net/sock.h>
+#include <linux/refcount.h>
 
 #define	NR_NETWORK_LEN			15
 #define	NR_TRANSPORT_LEN		5
@@ -93,7 +94,7 @@ struct nr_neigh {
 	unsigned short		count;
 	unsigned int		number;
 	unsigned char		failed;
-	atomic_t		refcount;
+	refcount_t		refcount;
 };
 
 struct nr_route {
@@ -128,11 +129,11 @@ static __inline__ void nr_node_put(struct nr_node *nr_node)
 }
 
 #define nr_neigh_hold(__nr_neigh) \
-	atomic_inc(&((__nr_neigh)->refcount))
+	refcount_inc(&((__nr_neigh)->refcount))
 
 static __inline__ void nr_neigh_put(struct nr_neigh *nr_neigh)
 {
-	if (atomic_dec_and_test(&nr_neigh->refcount)) {
+	if (refcount_dec_and_test(&nr_neigh->refcount)) {
 		if (nr_neigh->ax25)
 			ax25_cb_put(nr_neigh->ax25);
 		kfree(nr_neigh->digipeat);
diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index d72a4f1..6b72970 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -149,7 +149,7 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
 		nr_neigh->count    = 0;
 		nr_neigh->number   = nr_neigh_no++;
 		nr_neigh->failed   = 0;
-		atomic_set(&nr_neigh->refcount, 1);
+		refcount_set(&nr_neigh->refcount, 1);
 
 		if (ax25_digi != NULL && ax25_digi->ndigi > 0) {
 			nr_neigh->digipeat = kmemdup(ax25_digi,
@@ -431,7 +431,7 @@ static int __must_check nr_add_neigh(ax25_address *callsign,
 	nr_neigh->count    = 0;
 	nr_neigh->number   = nr_neigh_no++;
 	nr_neigh->failed   = 0;
-	atomic_set(&nr_neigh->refcount, 1);
+	refcount_set(&nr_neigh->refcount, 1);
 
 	if (ax25_digi != NULL && ax25_digi->ndigi > 0) {
 		nr_neigh->digipeat = kmemdup(ax25_digi, sizeof(*ax25_digi),
-- 
2.7.4

[PATCH 18/18] net, netrom: convert nr_node.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/netrom.h  | 6 +++---
 net/netrom/nr_route.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/netrom.h b/include/net/netrom.h
index cecb4fd..443a4ff 100644
--- a/include/net/netrom.h
+++ b/include/net/netrom.h
@@ -110,7 +110,7 @@ struct nr_node {
 	unsigned char		which;
 	unsigned char		count;
 	struct nr_route		routes[3];
-	atomic_t		refcount;
+	refcount_t		refcount;
 	spinlock_t		node_lock;
 };
 
@@ -119,11 +119,11 @@ struct nr_node {
  *********************************************************************/
 
 #define nr_node_hold(__nr_node) \
-	atomic_inc(&((__nr_node)->refcount))
+	refcount_inc(&((__nr_node)->refcount))
 
 static __inline__ void nr_node_put(struct nr_node *nr_node)
 {
-	if (atomic_dec_and_test(&nr_node->refcount)) {
+	if (refcount_dec_and_test(&nr_node->refcount)) {
 		kfree(nr_node);
 	}
 }
diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index 6b72970..0c59354 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -184,7 +184,7 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
 
 		nr_node->which = 0;
 		nr_node->count = 1;
-		atomic_set(&nr_node->refcount, 1);
+		refcount_set(&nr_node->refcount, 1);
 		spin_lock_init(&nr_node->node_lock);
 
 		nr_node->routes[0].quality   = quality;
-- 
2.7.4

[Bridge] [PATCH 18/18] net, netrom: convert nr_node.refcount from atomic_t to refcount_t

[Elena Reshetova]

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.

Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: David Windsor <dwindsor@gmail.com>
---
 include/net/netrom.h  | 6 +++---
 net/netrom/nr_route.c | 2 +-
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/include/net/netrom.h b/include/net/netrom.h
index cecb4fd..443a4ff 100644
--- a/include/net/netrom.h
+++ b/include/net/netrom.h
@@ -110,7 +110,7 @@ struct nr_node {
 	unsigned char		which;
 	unsigned char		count;
 	struct nr_route		routes[3];
-	atomic_t		refcount;
+	refcount_t		refcount;
 	spinlock_t		node_lock;
 };
 
@@ -119,11 +119,11 @@ struct nr_node {
  *********************************************************************/
 
 #define nr_node_hold(__nr_node) \
-	atomic_inc(&((__nr_node)->refcount))
+	refcount_inc(&((__nr_node)->refcount))
 
 static __inline__ void nr_node_put(struct nr_node *nr_node)
 {
-	if (atomic_dec_and_test(&nr_node->refcount)) {
+	if (refcount_dec_and_test(&nr_node->refcount)) {
 		kfree(nr_node);
 	}
 }
diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index 6b72970..0c59354 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -184,7 +184,7 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
 
 		nr_node->which = 0;
 		nr_node->count = 1;
-		atomic_set(&nr_node->refcount, 1);
+		refcount_set(&nr_node->refcount, 1);
 		spin_lock_init(&nr_node->node_lock);
 
 		nr_node->routes[0].quality   = quality;
-- 
2.7.4

Re: [PATCH 05/18] net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t

[Marcel Holtmann]

Hi Elena,

> refcount_t type and corresponding API should be
> used instead of atomic_t when the variable is used as
> a reference counter. This allows to avoid accidental
> refcounter overflows that might lead to use-after-free
> situations.
> 
> Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
> Signed-off-by: Kees Cook <keescook@chromium.org>
> Signed-off-by: David Windsor <dwindsor@gmail.com>
> ---
> include/net/bluetooth/rfcomm.h | 8 +++++---
> net/bluetooth/rfcomm/core.c    | 4 ++--
> 2 files changed, 7 insertions(+), 5 deletions(-)

patch has been applied to bluetooth-next tree.

Regards

Marcel

Re: [Bridge] [PATCH 05/18] net, bluetooth: convert rfcomm_dlc.refcnt from atomic_t to refcount_t

[Marcel Holtmann]

Hi Elena,

> refcount_t type and corresponding API should be
> used instead of atomic_t when the variable is used as
> a reference counter. This allows to avoid accidental
> refcounter overflows that might lead to use-after-free
> situations.
> 
> Signed-off-by: Elena Reshetova <elena.reshetova@intel.com>
> Signed-off-by: Hans Liljestrand <ishkamiel@gmail.com>
> Signed-off-by: Kees Cook <keescook@chromium.org>
> Signed-off-by: David Windsor <dwindsor@gmail.com>
> ---
> include/net/bluetooth/rfcomm.h | 8 +++++---
> net/bluetooth/rfcomm/core.c    | 4 ++--
> 2 files changed, 7 insertions(+), 5 deletions(-)

patch has been applied to bluetooth-next tree.

Regards

Marcel