* [PATCH net 04/10] net: rmnet: fix suspicious RCU usage
@ 2020-02-26 17:47 Taehee Yoo
2020-02-26 20:33 ` subashab
0 siblings, 1 reply; 3+ messages in thread
From: Taehee Yoo @ 2020-02-26 17:47 UTC (permalink / raw)
To: davem, kuba, subashab, stranche, netdev; +Cc: ap420073
rmnet_get_port() internally calls rcu_dereference_rtnl(),
which checks RTNL.
But rmnet_get_port() could be called by packet path.
The packet path is not protected by RTNL.
So, the suspicious RCU usage problem occurs.
Test commands:
ip netns add nst
ip link add veth0 type veth peer name veth1
ip link set veth1 netns nst
ip link add rmnet0 link veth0 type rmnet mux_id 1
ip netns exec nst ip link add rmnet1 link veth1 type rmnet mux_id 1
ip netns exec nst ip link set veth1 up
ip netns exec nst ip link set rmnet1 up
ip netns exec nst ip a a 192.168.100.2/24 dev rmnet1
ip link set veth0 up
ip link set rmnet0 up
ip a a 192.168.100.1/24 dev rmnet0
ping 192.168.100.2
Splat looks like:
[ 339.775811][ T969] =============================
[ 339.777204][ T969] WARNING: suspicious RCU usage
[ 339.778188][ T969] 5.5.0+ #407 Not tainted
[ 339.779123][ T969] -----------------------------
[ 339.780100][ T969] drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c:389 suspicious rcu_dereference_check() usage!
[ 339.781943][ T969]
[ 339.781943][ T969] other info that might help us debug this:
[ 339.781943][ T969]
[ 339.783475][ T969]
[ 339.783475][ T969] rcu_scheduler_active = 2, debug_locks = 1
[ 339.784656][ T969] 5 locks held by ping/969:
[ 339.785406][ T969] #0: ffff88804cb897f0 (sk_lock-AF_INET){+.+.}, at: raw_sendmsg+0xab8/0x2980
[ 339.786766][ T969] #1: ffffffff92925460 (rcu_read_lock_bh){....}, at: ip_finish_output2+0x243/0x2150
[ 339.788308][ T969] #2: ffffffff92925460 (rcu_read_lock_bh){....}, at: __dev_queue_xmit+0x213/0x2e10
[ 339.790662][ T969] #3: ffff88805a924158 (&dev->qdisc_running_key#3){+...}, at: ip_finish_output2+0x714/0x2150
[ 339.792072][ T969] #4: ffff88805b4fdc98 (&dev->qdisc_xmit_lock_key#3){+.-.}, at: sch_direct_xmit+0x1e2/0x1020
[ 339.793445][ T969]
[ 339.793445][ T969] stack backtrace:
[ 339.794691][ T969] CPU: 3 PID: 969 Comm: ping Not tainted 5.5.0+ #407
[ 339.795946][ T969] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[ 339.797621][ T969] Call Trace:
[ 339.798249][ T969] dump_stack+0x96/0xdb
[ 339.798847][ T969] rmnet_get_port.part.9+0x76/0x80 [rmnet]
[ 339.799583][ T969] rmnet_egress_handler+0x107/0x420 [rmnet]
[ 339.800350][ T969] ? sch_direct_xmit+0x1e2/0x1020
[ 339.801027][ T969] rmnet_vnd_start_xmit+0x3d/0xa0 [rmnet]
[ 339.801784][ T969] dev_hard_start_xmit+0x160/0x740
[ 339.802667][ T969] sch_direct_xmit+0x265/0x1020
[ ... ]
Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial implementation")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
---
drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c | 13 ++++++-------
drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h | 2 +-
.../net/ethernet/qualcomm/rmnet/rmnet_handlers.c | 4 ++--
drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c | 2 ++
4 files changed, 11 insertions(+), 10 deletions(-)
diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
index 7a7d0f521352..93642cdd3305 100644
--- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
+++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
@@ -382,11 +382,10 @@ struct rtnl_link_ops rmnet_link_ops __read_mostly = {
.fill_info = rmnet_fill_info,
};
-/* Needs either rcu_read_lock() or rtnl lock */
-struct rmnet_port *rmnet_get_port(struct net_device *real_dev)
+struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev)
{
if (rmnet_is_real_dev_registered(real_dev))
- return rcu_dereference_rtnl(real_dev->rx_handler_data);
+ return rcu_dereference(real_dev->rx_handler_data);
else
return NULL;
}
@@ -412,7 +411,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
struct rmnet_port *port, *slave_port;
int err;
- port = rmnet_get_port(real_dev);
+ port = rmnet_get_port_rtnl(real_dev);
/* If there is more than one rmnet dev attached, its probably being
* used for muxing. Skip the briding in that case
@@ -427,7 +426,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
if (err)
return -EBUSY;
- slave_port = rmnet_get_port(slave_dev);
+ slave_port = rmnet_get_port_rtnl(slave_dev);
slave_port->rmnet_mode = RMNET_EPMODE_BRIDGE;
slave_port->bridge_ep = real_dev;
@@ -445,11 +444,11 @@ int rmnet_del_bridge(struct net_device *rmnet_dev,
struct net_device *real_dev = priv->real_dev;
struct rmnet_port *port, *slave_port;
- port = rmnet_get_port(real_dev);
+ port = rmnet_get_port_rtnl(real_dev);
port->rmnet_mode = RMNET_EPMODE_VND;
port->bridge_ep = NULL;
- slave_port = rmnet_get_port(slave_dev);
+ slave_port = rmnet_get_port_rtnl(slave_dev);
rmnet_unregister_real_device(slave_dev, slave_port);
netdev_dbg(slave_dev, "removed from rmnet as slave\n");
diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
index cd0a6bcbe74a..0d568dcfd65a 100644
--- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
+++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
@@ -65,7 +65,7 @@ struct rmnet_priv {
struct rmnet_priv_stats stats;
};
-struct rmnet_port *rmnet_get_port(struct net_device *real_dev);
+struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev);
struct rmnet_endpoint *rmnet_get_endpoint(struct rmnet_port *port, u8 mux_id);
int rmnet_add_bridge(struct net_device *rmnet_dev,
struct net_device *slave_dev,
diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
index 1b74bc160402..074a8b326c30 100644
--- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
+++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
@@ -184,7 +184,7 @@ rx_handler_result_t rmnet_rx_handler(struct sk_buff **pskb)
return RX_HANDLER_PASS;
dev = skb->dev;
- port = rmnet_get_port(dev);
+ port = rmnet_get_port_rcu(dev);
switch (port->rmnet_mode) {
case RMNET_EPMODE_VND:
@@ -217,7 +217,7 @@ void rmnet_egress_handler(struct sk_buff *skb)
skb->dev = priv->real_dev;
mux_id = priv->mux_id;
- port = rmnet_get_port(skb->dev);
+ port = rmnet_get_port_rcu(skb->dev);
if (!port)
goto drop;
diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
index 509dfc895a33..a26e76e9d382 100644
--- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
+++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
@@ -50,7 +50,9 @@ static netdev_tx_t rmnet_vnd_start_xmit(struct sk_buff *skb,
priv = netdev_priv(dev);
if (priv->real_dev) {
+ rcu_read_lock();
rmnet_egress_handler(skb);
+ rcu_read_unlock();
} else {
this_cpu_inc(priv->pcpu_stats->stats.tx_drops);
kfree_skb(skb);
--
2.17.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH net 04/10] net: rmnet: fix suspicious RCU usage
2020-02-26 17:47 [PATCH net 04/10] net: rmnet: fix suspicious RCU usage Taehee Yoo
@ 2020-02-26 20:33 ` subashab
2020-02-27 7:33 ` Taehee Yoo
0 siblings, 1 reply; 3+ messages in thread
From: subashab @ 2020-02-26 20:33 UTC (permalink / raw)
To: Taehee Yoo; +Cc: davem, kuba, stranche, netdev
On 2020-02-26 10:47, Taehee Yoo wrote:
> rmnet_get_port() internally calls rcu_dereference_rtnl(),
> which checks RTNL.
> But rmnet_get_port() could be called by packet path.
> The packet path is not protected by RTNL.
> So, the suspicious RCU usage problem occurs.
>
> Test commands:
> ip netns add nst
> ip link add veth0 type veth peer name veth1
> ip link set veth1 netns nst
> ip link add rmnet0 link veth0 type rmnet mux_id 1
> ip netns exec nst ip link add rmnet1 link veth1 type rmnet mux_id 1
> ip netns exec nst ip link set veth1 up
> ip netns exec nst ip link set rmnet1 up
> ip netns exec nst ip a a 192.168.100.2/24 dev rmnet1
> ip link set veth0 up
> ip link set rmnet0 up
> ip a a 192.168.100.1/24 dev rmnet0
> ping 192.168.100.2
>
> Splat looks like:
> [ 339.775811][ T969] =============================
> [ 339.777204][ T969] WARNING: suspicious RCU usage
> [ 339.778188][ T969] 5.5.0+ #407 Not tainted
> [ 339.779123][ T969] -----------------------------
> [ 339.780100][ T969]
> drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c:389 suspicious
> rcu_dereference_check() usage!
> [ 339.781943][ T969]
> [ 339.781943][ T969] other info that might help us debug this:
> [ 339.781943][ T969]
> [ 339.783475][ T969]
> [ 339.783475][ T969] rcu_scheduler_active = 2, debug_locks = 1
> [ 339.784656][ T969] 5 locks held by ping/969:
> [ 339.785406][ T969] #0: ffff88804cb897f0 (sk_lock-AF_INET){+.+.},
> at: raw_sendmsg+0xab8/0x2980
> [ 339.786766][ T969] #1: ffffffff92925460 (rcu_read_lock_bh){....},
> at: ip_finish_output2+0x243/0x2150
> [ 339.788308][ T969] #2: ffffffff92925460 (rcu_read_lock_bh){....},
> at: __dev_queue_xmit+0x213/0x2e10
> [ 339.790662][ T969] #3: ffff88805a924158
> (&dev->qdisc_running_key#3){+...}, at: ip_finish_output2+0x714/0x2150
> [ 339.792072][ T969] #4: ffff88805b4fdc98
> (&dev->qdisc_xmit_lock_key#3){+.-.}, at: sch_direct_xmit+0x1e2/0x1020
> [ 339.793445][ T969]
> [ 339.793445][ T969] stack backtrace:
> [ 339.794691][ T969] CPU: 3 PID: 969 Comm: ping Not tainted 5.5.0+
> #407
> [ 339.795946][ T969] Hardware name: innotek GmbH
> VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
> [ 339.797621][ T969] Call Trace:
> [ 339.798249][ T969] dump_stack+0x96/0xdb
> [ 339.798847][ T969] rmnet_get_port.part.9+0x76/0x80 [rmnet]
> [ 339.799583][ T969] rmnet_egress_handler+0x107/0x420 [rmnet]
> [ 339.800350][ T969] ? sch_direct_xmit+0x1e2/0x1020
> [ 339.801027][ T969] rmnet_vnd_start_xmit+0x3d/0xa0 [rmnet]
> [ 339.801784][ T969] dev_hard_start_xmit+0x160/0x740
> [ 339.802667][ T969] sch_direct_xmit+0x265/0x1020
> [ ... ]
>
> Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial
> implementation")
> Signed-off-by: Taehee Yoo <ap420073@gmail.com>
> ---
> drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c | 13 ++++++-------
> drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h | 2 +-
> .../net/ethernet/qualcomm/rmnet/rmnet_handlers.c | 4 ++--
> drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c | 2 ++
> 4 files changed, 11 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> index 7a7d0f521352..93642cdd3305 100644
> --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> @@ -382,11 +382,10 @@ struct rtnl_link_ops rmnet_link_ops __read_mostly
> = {
> .fill_info = rmnet_fill_info,
> };
>
> -/* Needs either rcu_read_lock() or rtnl lock */
> -struct rmnet_port *rmnet_get_port(struct net_device *real_dev)
> +struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev)
> {
> if (rmnet_is_real_dev_registered(real_dev))
> - return rcu_dereference_rtnl(real_dev->rx_handler_data);
> + return rcu_dereference(real_dev->rx_handler_data);
> else
> return NULL;
> }
> @@ -412,7 +411,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
> struct rmnet_port *port, *slave_port;
> int err;
>
> - port = rmnet_get_port(real_dev);
> + port = rmnet_get_port_rtnl(real_dev);
>
> /* If there is more than one rmnet dev attached, its probably being
> * used for muxing. Skip the briding in that case
> @@ -427,7 +426,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
> if (err)
> return -EBUSY;
>
> - slave_port = rmnet_get_port(slave_dev);
> + slave_port = rmnet_get_port_rtnl(slave_dev);
> slave_port->rmnet_mode = RMNET_EPMODE_BRIDGE;
> slave_port->bridge_ep = real_dev;
>
> @@ -445,11 +444,11 @@ int rmnet_del_bridge(struct net_device
> *rmnet_dev,
> struct net_device *real_dev = priv->real_dev;
> struct rmnet_port *port, *slave_port;
>
> - port = rmnet_get_port(real_dev);
> + port = rmnet_get_port_rtnl(real_dev);
> port->rmnet_mode = RMNET_EPMODE_VND;
> port->bridge_ep = NULL;
>
> - slave_port = rmnet_get_port(slave_dev);
> + slave_port = rmnet_get_port_rtnl(slave_dev);
> rmnet_unregister_real_device(slave_dev, slave_port);
>
> netdev_dbg(slave_dev, "removed from rmnet as slave\n");
> diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> index cd0a6bcbe74a..0d568dcfd65a 100644
> --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> @@ -65,7 +65,7 @@ struct rmnet_priv {
> struct rmnet_priv_stats stats;
> };
>
> -struct rmnet_port *rmnet_get_port(struct net_device *real_dev);
> +struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev);
> struct rmnet_endpoint *rmnet_get_endpoint(struct rmnet_port *port, u8
> mux_id);
> int rmnet_add_bridge(struct net_device *rmnet_dev,
> struct net_device *slave_dev,
> diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> index 1b74bc160402..074a8b326c30 100644
> --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> @@ -184,7 +184,7 @@ rx_handler_result_t rmnet_rx_handler(struct sk_buff
> **pskb)
> return RX_HANDLER_PASS;
>
> dev = skb->dev;
> - port = rmnet_get_port(dev);
> + port = rmnet_get_port_rcu(dev);
>
> switch (port->rmnet_mode) {
> case RMNET_EPMODE_VND:
> @@ -217,7 +217,7 @@ void rmnet_egress_handler(struct sk_buff *skb)
> skb->dev = priv->real_dev;
> mux_id = priv->mux_id;
>
> - port = rmnet_get_port(skb->dev);
> + port = rmnet_get_port_rcu(skb->dev);
> if (!port)
> goto drop;
>
> diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> index 509dfc895a33..a26e76e9d382 100644
> --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> @@ -50,7 +50,9 @@ static netdev_tx_t rmnet_vnd_start_xmit(struct
> sk_buff *skb,
>
> priv = netdev_priv(dev);
> if (priv->real_dev) {
This rcu lock shouldn't be needed as it is acquired already in
__dev_queue_xmit().
> + rcu_read_lock();
> rmnet_egress_handler(skb);
> + rcu_read_unlock();
> } else {
> this_cpu_inc(priv->pcpu_stats->stats.tx_drops);
> kfree_skb(skb);
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH net 04/10] net: rmnet: fix suspicious RCU usage
2020-02-26 20:33 ` subashab
@ 2020-02-27 7:33 ` Taehee Yoo
0 siblings, 0 replies; 3+ messages in thread
From: Taehee Yoo @ 2020-02-27 7:33 UTC (permalink / raw)
To: subashab; +Cc: David Miller, Jakub Kicinski, stranche, Netdev
On Thu, 27 Feb 2020 at 05:33, <subashab@codeaurora.org> wrote:
>
Hi,
Thank you for the review!
> On 2020-02-26 10:47, Taehee Yoo wrote:
> > rmnet_get_port() internally calls rcu_dereference_rtnl(),
> > which checks RTNL.
> > But rmnet_get_port() could be called by packet path.
> > The packet path is not protected by RTNL.
> > So, the suspicious RCU usage problem occurs.
> >
> > Test commands:
> > ip netns add nst
> > ip link add veth0 type veth peer name veth1
> > ip link set veth1 netns nst
> > ip link add rmnet0 link veth0 type rmnet mux_id 1
> > ip netns exec nst ip link add rmnet1 link veth1 type rmnet mux_id 1
> > ip netns exec nst ip link set veth1 up
> > ip netns exec nst ip link set rmnet1 up
> > ip netns exec nst ip a a 192.168.100.2/24 dev rmnet1
> > ip link set veth0 up
> > ip link set rmnet0 up
> > ip a a 192.168.100.1/24 dev rmnet0
> > ping 192.168.100.2
> >
> > Splat looks like:
> > [ 339.775811][ T969] =============================
> > [ 339.777204][ T969] WARNING: suspicious RCU usage
> > [ 339.778188][ T969] 5.5.0+ #407 Not tainted
> > [ 339.779123][ T969] -----------------------------
> > [ 339.780100][ T969]
> > drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c:389 suspicious
> > rcu_dereference_check() usage!
> > [ 339.781943][ T969]
> > [ 339.781943][ T969] other info that might help us debug this:
> > [ 339.781943][ T969]
> > [ 339.783475][ T969]
> > [ 339.783475][ T969] rcu_scheduler_active = 2, debug_locks = 1
> > [ 339.784656][ T969] 5 locks held by ping/969:
> > [ 339.785406][ T969] #0: ffff88804cb897f0 (sk_lock-AF_INET){+.+.},
> > at: raw_sendmsg+0xab8/0x2980
> > [ 339.786766][ T969] #1: ffffffff92925460 (rcu_read_lock_bh){....},
> > at: ip_finish_output2+0x243/0x2150
> > [ 339.788308][ T969] #2: ffffffff92925460 (rcu_read_lock_bh){....},
> > at: __dev_queue_xmit+0x213/0x2e10
> > [ 339.790662][ T969] #3: ffff88805a924158
> > (&dev->qdisc_running_key#3){+...}, at: ip_finish_output2+0x714/0x2150
> > [ 339.792072][ T969] #4: ffff88805b4fdc98
> > (&dev->qdisc_xmit_lock_key#3){+.-.}, at: sch_direct_xmit+0x1e2/0x1020
> > [ 339.793445][ T969]
> > [ 339.793445][ T969] stack backtrace:
> > [ 339.794691][ T969] CPU: 3 PID: 969 Comm: ping Not tainted 5.5.0+
> > #407
> > [ 339.795946][ T969] Hardware name: innotek GmbH
> > VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
> > [ 339.797621][ T969] Call Trace:
> > [ 339.798249][ T969] dump_stack+0x96/0xdb
> > [ 339.798847][ T969] rmnet_get_port.part.9+0x76/0x80 [rmnet]
> > [ 339.799583][ T969] rmnet_egress_handler+0x107/0x420 [rmnet]
> > [ 339.800350][ T969] ? sch_direct_xmit+0x1e2/0x1020
> > [ 339.801027][ T969] rmnet_vnd_start_xmit+0x3d/0xa0 [rmnet]
> > [ 339.801784][ T969] dev_hard_start_xmit+0x160/0x740
> > [ 339.802667][ T969] sch_direct_xmit+0x265/0x1020
> > [ ... ]
> >
> > Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial
> > implementation")
> > Signed-off-by: Taehee Yoo <ap420073@gmail.com>
> > ---
> > drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c | 13 ++++++-------
> > drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h | 2 +-
> > .../net/ethernet/qualcomm/rmnet/rmnet_handlers.c | 4 ++--
> > drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c | 2 ++
> > 4 files changed, 11 insertions(+), 10 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> > b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> > index 7a7d0f521352..93642cdd3305 100644
> > --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> > +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.c
> > @@ -382,11 +382,10 @@ struct rtnl_link_ops rmnet_link_ops __read_mostly
> > = {
> > .fill_info = rmnet_fill_info,
> > };
> >
> > -/* Needs either rcu_read_lock() or rtnl lock */
> > -struct rmnet_port *rmnet_get_port(struct net_device *real_dev)
> > +struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev)
> > {
> > if (rmnet_is_real_dev_registered(real_dev))
> > - return rcu_dereference_rtnl(real_dev->rx_handler_data);
> > + return rcu_dereference(real_dev->rx_handler_data);
> > else
> > return NULL;
> > }
> > @@ -412,7 +411,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
> > struct rmnet_port *port, *slave_port;
> > int err;
> >
> > - port = rmnet_get_port(real_dev);
> > + port = rmnet_get_port_rtnl(real_dev);
> >
> > /* If there is more than one rmnet dev attached, its probably being
> > * used for muxing. Skip the briding in that case
> > @@ -427,7 +426,7 @@ int rmnet_add_bridge(struct net_device *rmnet_dev,
> > if (err)
> > return -EBUSY;
> >
> > - slave_port = rmnet_get_port(slave_dev);
> > + slave_port = rmnet_get_port_rtnl(slave_dev);
> > slave_port->rmnet_mode = RMNET_EPMODE_BRIDGE;
> > slave_port->bridge_ep = real_dev;
> >
> > @@ -445,11 +444,11 @@ int rmnet_del_bridge(struct net_device
> > *rmnet_dev,
> > struct net_device *real_dev = priv->real_dev;
> > struct rmnet_port *port, *slave_port;
> >
> > - port = rmnet_get_port(real_dev);
> > + port = rmnet_get_port_rtnl(real_dev);
> > port->rmnet_mode = RMNET_EPMODE_VND;
> > port->bridge_ep = NULL;
> >
> > - slave_port = rmnet_get_port(slave_dev);
> > + slave_port = rmnet_get_port_rtnl(slave_dev);
> > rmnet_unregister_real_device(slave_dev, slave_port);
> >
> > netdev_dbg(slave_dev, "removed from rmnet as slave\n");
> > diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> > b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> > index cd0a6bcbe74a..0d568dcfd65a 100644
> > --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> > +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_config.h
> > @@ -65,7 +65,7 @@ struct rmnet_priv {
> > struct rmnet_priv_stats stats;
> > };
> >
> > -struct rmnet_port *rmnet_get_port(struct net_device *real_dev);
> > +struct rmnet_port *rmnet_get_port_rcu(struct net_device *real_dev);
> > struct rmnet_endpoint *rmnet_get_endpoint(struct rmnet_port *port, u8
> > mux_id);
> > int rmnet_add_bridge(struct net_device *rmnet_dev,
> > struct net_device *slave_dev,
> > diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> > b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> > index 1b74bc160402..074a8b326c30 100644
> > --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> > +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_handlers.c
> > @@ -184,7 +184,7 @@ rx_handler_result_t rmnet_rx_handler(struct sk_buff
> > **pskb)
> > return RX_HANDLER_PASS;
> >
> > dev = skb->dev;
> > - port = rmnet_get_port(dev);
> > + port = rmnet_get_port_rcu(dev);
> >
> > switch (port->rmnet_mode) {
> > case RMNET_EPMODE_VND:
> > @@ -217,7 +217,7 @@ void rmnet_egress_handler(struct sk_buff *skb)
> > skb->dev = priv->real_dev;
> > mux_id = priv->mux_id;
> >
> > - port = rmnet_get_port(skb->dev);
> > + port = rmnet_get_port_rcu(skb->dev);
> > if (!port)
> > goto drop;
> >
> > diff --git a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> > b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> > index 509dfc895a33..a26e76e9d382 100644
> > --- a/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> > +++ b/drivers/net/ethernet/qualcomm/rmnet/rmnet_vnd.c
> > @@ -50,7 +50,9 @@ static netdev_tx_t rmnet_vnd_start_xmit(struct
> > sk_buff *skb,
> >
> > priv = netdev_priv(dev);
> > if (priv->real_dev) {
>
> This rcu lock shouldn't be needed as it is acquired already in
> __dev_queue_xmit().
>
You're right, __dev_queue_xmit() already acquires rcu_read_lock_bh().
So, rcu_read_lock() is unnecessary in rmnet_vnd_start_xmit().
And I think rcu_dereference_bh() should be used instead of
rcu_dereference() in the rmnet_get_port_rcu().
I will send a v2 patch, which changes these two things.
Thank you!
Taehee Yoo
> > + rcu_read_lock();
> > rmnet_egress_handler(skb);
> > + rcu_read_unlock();
> > } else {
> > this_cpu_inc(priv->pcpu_stats->stats.tx_drops);
> > kfree_skb(skb);
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-02-27 7:34 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-26 17:47 [PATCH net 04/10] net: rmnet: fix suspicious RCU usage Taehee Yoo
2020-02-26 20:33 ` subashab
2020-02-27 7:33 ` Taehee Yoo
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).