Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1313773

[PATCH 4.3 38/55] net: fix IP early demux races

From Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Newsgroups linux.kernel
Subject [PATCH 4.3 38/55] net: fix IP early demux races
Date 2016-01-21 02:00 +0100
Message-ID <qTbLu-5Sa-41@gated-at.bofh.it> (permalink)
References <qTbBL-5Oq-3@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw

4.3-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Eric Dumazet <edumazet@google.com>

[ Upstream commit 5037e9ef9454917b047f9f3a19b4dd179fbf7cd4 ]

David Wilder reported crashes caused by dst reuse.

<quote David>
  I am seeing a crash on a distro V4.2.3 kernel caused by a double
  release of a dst_entry.  In ipv4_dst_destroy() the call to
  list_empty() finds a poisoned next pointer, indicating the dst_entry
  has already been removed from the list and freed. The crash occurs
  18 to 24 hours into a run of a network stress exerciser.
</quote>

Thanks to his detailed report and analysis, we were able to understand
the core issue.

IP early demux can associate a dst to skb, after a lookup in TCP/UDP
sockets.

When socket cache is not properly set, we want to store into
sk->sk_dst_cache the dst for future IP early demux lookups,
by acquiring a stable refcount on the dst.

Problem is this acquisition is simply using an atomic_inc(),
which works well, unless the dst was queued for destruction from
dst_release() noticing dst refcount went to zero, if DST_NOCACHE
was set on dst.

We need to make sure current refcount is not zero before incrementing
it, or risk double free as David reported.

This patch, being a stable candidate, adds two new helpers, and use
them only from IP early demux problematic paths.

It might be possible to merge in net-next skb_dst_force() and
skb_dst_force_safe(), but I prefer having the smallest patch for stable
kernels : Maybe some skb_dst_force() callers do not expect skb->dst
can suddenly be cleared.

Can probably be backported back to linux-3.6 kernels

Reported-by: David J. Wilder <dwilder@us.ibm.com>
Tested-by: David J. Wilder <dwilder@us.ibm.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 include/net/dst.h   |   33 +++++++++++++++++++++++++++++++++
 include/net/sock.h  |    2 +-
 net/ipv4/tcp_ipv4.c |    5 ++---
 net/ipv6/tcp_ipv6.c |    3 +--
 4 files changed, 37 insertions(+), 6 deletions(-)

--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -322,6 +322,39 @@ static inline void skb_dst_force(struct
 	}
 }
 
+/**
+ * dst_hold_safe - Take a reference on a dst if possible
+ * @dst: pointer to dst entry
+ *
+ * This helper returns false if it could not safely
+ * take a reference on a dst.
+ */
+static inline bool dst_hold_safe(struct dst_entry *dst)
+{
+	if (dst->flags & DST_NOCACHE)
+		return atomic_inc_not_zero(&dst->__refcnt);
+	dst_hold(dst);
+	return true;
+}
+
+/**
+ * skb_dst_force_safe - makes sure skb dst is refcounted
+ * @skb: buffer
+ *
+ * If dst is not yet refcounted and not destroyed, grab a ref on it.
+ */
+static inline void skb_dst_force_safe(struct sk_buff *skb)
+{
+	if (skb_dst_is_noref(skb)) {
+		struct dst_entry *dst = skb_dst(skb);
+
+		if (!dst_hold_safe(dst))
+			dst = NULL;
+
+		skb->_skb_refdst = (unsigned long)dst;
+	}
+}
+
 
 /**
  *	__skb_tunnel_rx - prepare skb for rx reinsert
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -801,7 +801,7 @@ void sk_stream_write_space(struct sock *
 static inline void __sk_add_backlog(struct sock *sk, struct sk_buff *skb)
 {
 	/* dont let skb dst not refcounted, we are going to leave rcu lock */
-	skb_dst_force(skb);
+	skb_dst_force_safe(skb);
 
 	if (!sk->sk_backlog.tail)
 		sk->sk_backlog.head = skb;
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1509,7 +1509,7 @@ bool tcp_prequeue(struct sock *sk, struc
 	if (likely(sk->sk_rx_dst))
 		skb_dst_drop(skb);
 	else
-		skb_dst_force(skb);
+		skb_dst_force_safe(skb);
 
 	__skb_queue_tail(&tp->ucopy.prequeue, skb);
 	tp->ucopy.memory += skb->truesize;
@@ -1710,8 +1710,7 @@ void inet_sk_rx_dst_set(struct sock *sk,
 {
 	struct dst_entry *dst = skb_dst(skb);
 
-	if (dst) {
-		dst_hold(dst);
+	if (dst && dst_hold_safe(dst)) {
 		sk->sk_rx_dst = dst;
 		inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
 	}
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -93,10 +93,9 @@ static void inet6_sk_rx_dst_set(struct s
 {
 	struct dst_entry *dst = skb_dst(skb);
 
-	if (dst) {
+	if (dst && dst_hold_safe(dst)) {
 		const struct rt6_info *rt = (const struct rt6_info *)dst;
 
-		dst_hold(dst);
 		sk->sk_rx_dst = dst;
 		inet_sk(sk)->rx_dst_ifindex = skb->skb_iif;
 		inet6_sk(sk)->rx_dst_cookie = rt6_get_cookie(rt);

Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

[PATCH 4.3 00/55] 4.3.4-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 41/55] skbuff: Fix offset error in skb_reorder_vlan_header Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 37/55] sh_eth: fix kernel oops in skb_put() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 34/55] openvswitch: Respect conntrack zone even if invalid Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 39/55] pptp: verify sockaddr_len in pptp_bind() and pptp_connect() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 33/55] openvswitch: Fix helper reference leak Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 22/55] atl1c: Improve driver not to do order 4 GFP_ATOMIC allocation Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 29/55] net: qca_spi: fix transmit queue timeout handling Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 30/55] r8152: fix lockup when runtime PM is enabled Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 10/55] USB: serial: Another Infineon flash loader USB ID Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 47/55] inet: tcp: fix inetpeer_set_addr_v4() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 19/55] usb: Use the USB_SS_MULT() macro to decode burst multiplier for log message Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 05/55] ACPI: Using correct irq when waiting for events Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 11/55] usb-storage: Fix scsi-sd failure "Invalid field in cdb" for USB adapter JMicron Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 55/55] KEYS: Fix keyring ref leak in join_session_keyring() Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 01:50 +0100
  [PATCH 4.3 26/55] sctp: update the netstamp_needed counter when copying sockets Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 07/55] tpm, tpm_tis: fix tpm_tis ACPI detection issue with TPM 2.0 Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 16/55] USB: host: ohci-at91: fix a crash in ohci_hcd_at91_overcurrent_irq Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 09/55] USB: cdc_acm: Ignore Infineon Flash Loader utility Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 14/55] USB: whci-hcd: add check for dma mapping error Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 43/55] net_sched: make qdisc_tree_decrease_qlen() work for non mq Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 24/55] vxlan: fix incorrect RCO bit in VXLAN header Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 42/55] net: check both type and procotol for tcp sockets Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 17/55] usb: musb: USB_TI_CPPI41_DMA requires dmaengine support Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 15/55] usb: gadget: pxa27x: fix suspend callback Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 45/55] net: fix uninitialized variable issue Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 49/55] gianfar: Dont enable RX Filer if not supported Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 44/55] bluetooth: Validate socket address length in sco_sock_bind(). Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 32/55] phy: micrel: Fix finding PHY properties in MAC node. Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 38/55] net: fix IP early demux races Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 21/55] gre6: allow to update all parameters via rtnl Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 35/55] uapi: export ila.h Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 48/55] rhashtable: Enforce minimum size on initial hash table Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 52/55] tcp: restore fastopen with no data in SYN packet Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 06/55] ACPI / PM: Fix incorrect wakeup IRQ setting during suspend-to-idle Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 31/55] ipv6: sctp: clone options to avoid use after free Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 18/55] usb: core : hub: Fix BOS NULL pointer kernel panic Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 25/55] sctp: use the same clock as if sock source timestamps were on Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 20/55] pppoe: fix memory corruption in padt work structure Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 27/55] sctp: also copy sk_tsflags when copying the socket Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 03/55] tipc: Fix kfree_skb() of uninitialised pointer Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 50/55] fou: clean up socket with kfree_rcu Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:00 +0100
  [PATCH 4.3 01/55] Revert "vrf: fix double free and memory corruption on register_netdevice failure" Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-21 02:10 +0100
  Re: [PATCH 4.3 00/55] 4.3.4-stable review Shuah Khan <shuahkh@osg.samsung.com> - 2016-01-21 02:50 +0100
    Re: [PATCH 4.3 00/55] 4.3.4-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-22 09:00 +0100
  Re: [PATCH 4.3 00/55] 4.3.4-stable review Mel Gorman <mgorman@techsingularity.net> - 2016-01-21 10:50 +0100
    Re: [PATCH 4.3 00/55] 4.3.4-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-22 09:00 +0100
      Re: [PATCH 4.3 00/55] 4.3.4-stable review Mel Gorman <mgorman@techsingularity.net> - 2016-01-22 09:20 +0100
  Re: [PATCH 4.3 00/55] 4.3.4-stable review Guenter Roeck <linux@roeck-us.net> - 2016-01-21 13:30 +0100
    Re: [PATCH 4.3 00/55] 4.3.4-stable review Greg Kroah-Hartman <gregkh@linuxfoundation.org> - 2016-01-22 09:00 +0100

csiph-web