diff mbox series

[net-next,v2] net: fix dev_ifsioc_locked() race condition

Message ID 20210131022755.106005-1-xiyou.wangcong@gmail.com (mailing list archive)
State Superseded
Delegated to: Netdev Maintainers
Headers show
Series [net-next,v2] net: fix dev_ifsioc_locked() race condition | expand

Checks

Context Check Description
netdev/cover_letter success Link
netdev/fixes_present success Link
netdev/patch_count success Link
netdev/tree_selection success Clearly marked for net-next
netdev/subject_prefix success Link
netdev/cc_maintainers warning 17 maintainers not CCed: gustavoars@kernel.org colin.king@canonical.com andriin@fb.com f.fainelli@gmail.com toke@redhat.com nikolay@nvidia.com zhudi21@huawei.com roopa@cumulusnetworks.com daniel@iogearbox.net bjorn@kernel.org jwi@linux.ibm.com edumazet@google.com ast@kernel.org ap420073@gmail.com mkubecek@suse.cz davem@davemloft.net laniel_francis@privacyrequired.com
netdev/source_inline success Was 0 now: 0
netdev/verify_signedoff success Link
netdev/module_param success Was 0 now: 0
netdev/build_32bit success Errors and warnings before: 7141 this patch: 7141
netdev/kdoc success Errors and warnings before: 0 this patch: 0
netdev/verify_fixes success Link
netdev/checkpatch success total: 0 errors, 0 warnings, 0 checks, 107 lines checked
netdev/build_allmodconfig_warn success Errors and warnings before: 7540 this patch: 7540
netdev/header_inline success Link
netdev/stable success Stable not CCed

Commit Message

Cong Wang Jan. 31, 2021, 2:27 a.m. UTC
From: Cong Wang <cong.wang@bytedance.com>

dev_ifsioc_locked() is called with only RCU read lock, so when
there is a parallel writer changing the mac address, it could
get a partially updated mac address, as shown below:

Thread 1			Thread 2
// eth_commit_mac_addr_change()
memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);
				// dev_ifsioc_locked()
				memcpy(ifr->ifr_hwaddr.sa_data,
					dev->dev_addr,...);

Close this race condition by guarding them with a RW semaphore,
like netdev_get_name(). The writers take RTNL anyway, so this
will not affect the slow path. To avoid bothering existing
dev_set_mac_address() callers in drivers, introduce a new wrapper
just for user-facing callers in ioctl and rtnetlink.

Fixes: 3710becf8a58 ("net: RCU locking for simple ioctl()")
Reported-by: "Gong, Sishuai" <sishuai@purdue.edu>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
---
 include/linux/netdevice.h |  3 +++
 net/core/dev.c            | 40 +++++++++++++++++++++++++++++++++++++++
 net/core/dev_ioctl.c      | 20 +++++++-------------
 net/core/rtnetlink.c      |  2 +-
 4 files changed, 51 insertions(+), 14 deletions(-)

Comments

Jakub Kicinski Feb. 2, 2021, 3:40 a.m. UTC | #1
On Sat, 30 Jan 2021 18:27:55 -0800 Cong Wang wrote:
> From: Cong Wang <cong.wang@bytedance.com>
> 
> dev_ifsioc_locked() is called with only RCU read lock, so when
> there is a parallel writer changing the mac address, it could
> get a partially updated mac address, as shown below:
> 
> Thread 1			Thread 2
> // eth_commit_mac_addr_change()
> memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);
> 				// dev_ifsioc_locked()
> 				memcpy(ifr->ifr_hwaddr.sa_data,
> 					dev->dev_addr,...);
> 
> Close this race condition by guarding them with a RW semaphore,
> like netdev_get_name(). The writers take RTNL anyway, so this
> will not affect the slow path. To avoid bothering existing
> dev_set_mac_address() callers in drivers, introduce a new wrapper
> just for user-facing callers in ioctl and rtnetlink.

Some of the drivers need to be update, tho, right? At a quick look at
least bond and tun seem to be making calls to dev_set_mac_address() 
on IOCTL paths.

> Fixes: 3710becf8a58 ("net: RCU locking for simple ioctl()")
> Reported-by: "Gong, Sishuai" <sishuai@purdue.edu>
> Cc: Eric Dumazet <eric.dumazet@gmail.com>
> Cc: Jakub Kicinski <kuba@kernel.org>
> Signed-off-by: Cong Wang <cong.wang@bytedance.com>
> ---
>  include/linux/netdevice.h |  3 +++
>  net/core/dev.c            | 40 +++++++++++++++++++++++++++++++++++++++
>  net/core/dev_ioctl.c      | 20 +++++++-------------
>  net/core/rtnetlink.c      |  2 +-
>  4 files changed, 51 insertions(+), 14 deletions(-)
Cong Wang Feb. 2, 2021, 5:27 a.m. UTC | #2
On Mon, Feb 1, 2021 at 7:40 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Sat, 30 Jan 2021 18:27:55 -0800 Cong Wang wrote:
> > From: Cong Wang <cong.wang@bytedance.com>
> >
> > dev_ifsioc_locked() is called with only RCU read lock, so when
> > there is a parallel writer changing the mac address, it could
> > get a partially updated mac address, as shown below:
> >
> > Thread 1                      Thread 2
> > // eth_commit_mac_addr_change()
> > memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);
> >                               // dev_ifsioc_locked()
> >                               memcpy(ifr->ifr_hwaddr.sa_data,
> >                                       dev->dev_addr,...);
> >
> > Close this race condition by guarding them with a RW semaphore,
> > like netdev_get_name(). The writers take RTNL anyway, so this
> > will not affect the slow path. To avoid bothering existing
> > dev_set_mac_address() callers in drivers, introduce a new wrapper
> > just for user-facing callers in ioctl and rtnetlink.
>
> Some of the drivers need to be update, tho, right? At a quick look at
> least bond and tun seem to be making calls to dev_set_mac_address()
> on IOCTL paths.

Ah, good catch! Clearly I missed those special IOCTL's.

Thanks,
diff mbox series

Patch

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index e9e7ada07ea1..8f13d3177130 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -3907,6 +3907,9 @@  int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
 			      struct netlink_ext_ack *extack);
 int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
 			struct netlink_ext_ack *extack);
+int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
+			     struct netlink_ext_ack *extack);
+int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
 int dev_change_carrier(struct net_device *, bool new_carrier);
 int dev_get_phys_port_id(struct net_device *dev,
 			 struct netdev_phys_item_id *ppid);
diff --git a/net/core/dev.c b/net/core/dev.c
index aae116d059da..ecc7914784d2 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -8773,6 +8773,46 @@  int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
 }
 EXPORT_SYMBOL(dev_set_mac_address);
 
+static DECLARE_RWSEM(dev_addr_sem);
+
+int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
+			     struct netlink_ext_ack *extack)
+{
+	int ret;
+
+	down_write(&dev_addr_sem);
+	ret = dev_set_mac_address(dev, sa, extack);
+	up_write(&dev_addr_sem);
+	return ret;
+}
+
+int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name)
+{
+	size_t size = sizeof(sa->sa_data);
+	struct net_device *dev;
+	int ret = 0;
+
+	down_read(&dev_addr_sem);
+	rcu_read_lock();
+
+	dev = dev_get_by_name_rcu(net, dev_name);
+	if (!dev) {
+		ret = -ENODEV;
+		goto unlock;
+	}
+	if (!dev->addr_len)
+		memset(sa->sa_data, 0, size);
+	else
+		memcpy(sa->sa_data, dev->dev_addr,
+		       min_t(size_t, size, dev->addr_len));
+	sa->sa_family = dev->type;
+
+unlock:
+	rcu_read_unlock();
+	up_read(&dev_addr_sem);
+	return ret;
+}
+
 /**
  *	dev_change_carrier - Change device carrier
  *	@dev: device
diff --git a/net/core/dev_ioctl.c b/net/core/dev_ioctl.c
index db8a0ff86f36..478d032f34ac 100644
--- a/net/core/dev_ioctl.c
+++ b/net/core/dev_ioctl.c
@@ -123,17 +123,6 @@  static int dev_ifsioc_locked(struct net *net, struct ifreq *ifr, unsigned int cm
 		ifr->ifr_mtu = dev->mtu;
 		return 0;
 
-	case SIOCGIFHWADDR:
-		if (!dev->addr_len)
-			memset(ifr->ifr_hwaddr.sa_data, 0,
-			       sizeof(ifr->ifr_hwaddr.sa_data));
-		else
-			memcpy(ifr->ifr_hwaddr.sa_data, dev->dev_addr,
-			       min(sizeof(ifr->ifr_hwaddr.sa_data),
-				   (size_t)dev->addr_len));
-		ifr->ifr_hwaddr.sa_family = dev->type;
-		return 0;
-
 	case SIOCGIFSLAVE:
 		err = -EINVAL;
 		break;
@@ -274,7 +263,7 @@  static int dev_ifsioc(struct net *net, struct ifreq *ifr, unsigned int cmd)
 	case SIOCSIFHWADDR:
 		if (dev->addr_len > sizeof(struct sockaddr))
 			return -EINVAL;
-		return dev_set_mac_address(dev, &ifr->ifr_hwaddr, NULL);
+		return dev_set_mac_address_user(dev, &ifr->ifr_hwaddr, NULL);
 
 	case SIOCSIFHWBROADCAST:
 		if (ifr->ifr_hwaddr.sa_family != dev->type)
@@ -418,6 +407,12 @@  int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr, bool *need_c
 	 */
 
 	switch (cmd) {
+	case SIOCGIFHWADDR:
+		dev_load(net, ifr->ifr_name);
+		ret = dev_get_mac_address(&ifr->ifr_hwaddr, net, ifr->ifr_name);
+		if (colon)
+			*colon = ':';
+		return ret;
 	/*
 	 *	These ioctl calls:
 	 *	- can be done by all.
@@ -427,7 +422,6 @@  int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr, bool *need_c
 	case SIOCGIFFLAGS:
 	case SIOCGIFMETRIC:
 	case SIOCGIFMTU:
-	case SIOCGIFHWADDR:
 	case SIOCGIFSLAVE:
 	case SIOCGIFMAP:
 	case SIOCGIFINDEX:
diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
index c313aaf2bce1..0edc0b2baaa4 100644
--- a/net/core/rtnetlink.c
+++ b/net/core/rtnetlink.c
@@ -2660,7 +2660,7 @@  static int do_setlink(const struct sk_buff *skb,
 		sa->sa_family = dev->type;
 		memcpy(sa->sa_data, nla_data(tb[IFLA_ADDRESS]),
 		       dev->addr_len);
-		err = dev_set_mac_address(dev, sa, extack);
+		err = dev_set_mac_address_user(dev, sa, extack);
 		kfree(sa);
 		if (err)
 			goto errout;