netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Michael S. Tsirkin" <mst@redhat.com>
To: Jiri Pirko <jiri@resnulli.us>
Cc: jtluka@redhat.com, netdev@vger.kernel.org,
	bridge@lists.linux-foundation.org, stephen@networkplumber.org,
	edumazet@google.com, laine@redhat.com, zhiguohong@tencent.com,
	paulmck@linux.vnet.ibm.com, davem@davemloft.net
Subject: Re: [patch net/stable v2] br: fix use of ->rx_handler_data in code executed on non-rx_handler path
Date: Mon, 9 Dec 2013 13:58:35 +0200	[thread overview]
Message-ID: <20131209115835.GA15564@redhat.com> (raw)
In-Reply-To: <1386257257-25258-1-git-send-email-jiri@resnulli.us>

On Thu, Dec 05, 2013 at 04:27:37PM +0100, Jiri Pirko wrote:
> br_stp_rcv() is reached by non-rx_handler path. That means there is no
> guarantee that dev is bridge port and therefore simple NULL check of
> ->rx_handler_data is not enough. There is need to check if dev is really
> bridge port and since only rcu read lock is held here, do it by checking
> ->rx_handler pointer.
> 
> Note that synchronize_net() in netdev_rx_handler_unregister() ensures
> this approach as valid.
> 
> Introduced originally by:
> commit f350a0a87374418635689471606454abc7beaa3a
>   "bridge: use rx_handler_data pointer to store net_bridge_port pointer"
> 
> Fixed but not in the best way by:
> commit b5ed54e94d324f17c97852296d61a143f01b227a
>   "bridge: fix RCU races with bridge port"
> 
> Reintroduced by:
> commit 716ec052d2280d511e10e90ad54a86f5b5d4dcc2
>   "bridge: fix NULL pointer deref of br_port_get_rcu"
> 
> Please apply to stable trees as well. Thanks.
> 
> RH bugzilla reference: https://bugzilla.redhat.com/show_bug.cgi?id=1025770
> 
> Reported-by: Laine Stump <laine@redhat.com>
> Debugged-by: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> Signed-off-by: Jiri Pirko <jiri@resnulli.us>
> ---
> v1->v2: moved br_port_get_check_rcu definition below br_handle_frame definition
> 
>  net/bridge/br_private.h  | 10 ++++++++++
>  net/bridge/br_stp_bpdu.c |  2 +-
>  2 files changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/net/bridge/br_private.h b/net/bridge/br_private.h
> index 229d820..045d56e 100644
> --- a/net/bridge/br_private.h
> +++ b/net/bridge/br_private.h
> @@ -426,6 +426,16 @@ netdev_features_t br_features_recompute(struct net_bridge *br,
>  int br_handle_frame_finish(struct sk_buff *skb);
>  rx_handler_result_t br_handle_frame(struct sk_buff **pskb);
>  
> +static inline bool br_rx_handler_check_rcu(const struct net_device *dev)
> +{
> +	return rcu_dereference(dev->rx_handler) == br_handle_frame;

Actually this started to bother me.
rcu_dereference is for when we dereference, isn't it?
I think we should use rcu_access_pointer here.


> +}


Given all the confusion, how about we create an API to
access rx handler data outside rx handler itself in a
safe, documented way?

If everyone agrees, we can then re-implement
br_port_get_check_rcu on top of this API.

What do others think?

---

netdevice: allow access to rx_handler_data outside rx handler

rx_handler_data is easy to use correctly within
rx handler itself. Outside of that context, one must
validate the handler first.

Add an API to do this in a uniform way.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

-->

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 7f0ed42..7a353b1 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -1320,6 +1320,9 @@ struct net_device {
 #endif
 
 	rx_handler_func_t __rcu	*rx_handler;
+	/* rx_handler itself can use rx_handler_data directly.
+	 * Others must use netdev_rx_handler_data_rcu_dereference.
+	 */
 	void __rcu		*rx_handler_data;
 
 	struct netdev_queue __rcu *ingress_queue;
@@ -2399,6 +2402,31 @@ int netdev_rx_handler_register(struct net_device *dev,
 			       void *rx_handler_data);
 void netdev_rx_handler_unregister(struct net_device *dev);
 
+/**
+ *	netdev_rx_handler_data_rcu_dereference - access receive handler data
+ *	@dev: device to get handler data for
+ *	@rx_handler: receive handler used to register this data
+ *
+ *	Check that the receive handler is valid for the device.
+ *	Return handler data if it is, NULL otherwise.
+ *
+ *	Use this function if you want to access rx handler data
+ *	outside rx handler itself.
+ *
+ *	The caller must invoke this function under RCU read lock.
+ *
+ *	For a general description of rx_handler, see enum rx_handler_result.
+ */
+static inline
+void *netdev_rx_handler_data_rcu_dereference(struct net_device *dev,
+					     rx_handler_func_t *rx_handler)
+{
+	if (rcu_access_pointer(dev->rx_handler) != rx_handler)
+		return NULL;
+
+	return rcu_dereference(dev->rx_handler_data);
+}
+
 bool dev_valid_name(const char *name);
 int dev_ioctl(struct net *net, unsigned int cmd, void __user *);
 int dev_ethtool(struct net *net, struct ifreq *);

  parent reply	other threads:[~2013-12-09 11:58 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-12-05 15:27 [patch net/stable v2] br: fix use of ->rx_handler_data in code executed on non-rx_handler path Jiri Pirko
2013-12-05 15:35 ` Michael S. Tsirkin
2013-12-05 15:37 ` Eric Dumazet
2013-12-05 16:55 ` Stephen Hemminger
2013-12-06  2:26   ` Gao feng
2013-12-07  1:44     ` Vlad Yasevich
2013-12-06 20:43 ` David Miller
2013-12-06 21:10   ` Stephen Hemminger
2013-12-06 21:16     ` David Miller
2013-12-07  8:51     ` Jiri Pirko
2013-12-07 17:42       ` Stephen Hemminger
2013-12-07 18:18         ` Jiri Pirko
2013-12-07 19:10       ` Vlad Yasevich
2013-12-07 20:07         ` Jiri Pirko
2013-12-09  2:07           ` Vlad Yasevich
2013-12-09  9:36             ` Jiri Pirko
2013-12-09 11:58 ` Michael S. Tsirkin [this message]
2013-12-09 12:13   ` Jiri Pirko
2013-12-09 19:31   ` Vlad Yasevich
2013-12-09 21:52     ` Jiri Pirko

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20131209115835.GA15564@redhat.com \
    --to=mst@redhat.com \
    --cc=bridge@lists.linux-foundation.org \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=jiri@resnulli.us \
    --cc=jtluka@redhat.com \
    --cc=laine@redhat.com \
    --cc=netdev@vger.kernel.org \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=stephen@networkplumber.org \
    --cc=zhiguohong@tencent.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
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).