vxlan: test dev->flags & IFF_UP before calling netif_rx()
Subsystems (2)
PATCH 1/1 vxlan: test dev->flags & IFF_UP before calling netif_rx()
netif_rx() must be called under a strict contract.
At device dismantle phase, core networking clears IFF_UP and flush_all_backlogs() is called after rcu grace period to make sure no incoming packet might be in a cpu backlog and still referencing the device.
Most drivers call netif_rx() from their interrupt handler, and since the interrupts are disabled at device dismantle, netif_rx() does not have to check dev->flags & IFF_UP
Virtual drivers do not have this guarantee, and must therefore make the check themselves.
Otherwise we risk use-after-free and/or crashes.
Note this patch also fixes a small issue that came with commit ce6502a8f957 ("vxlan: fix a use after free in vxlan_encap_bypass"), since the dev->stats.rx_dropped change was done on the wrong device.
--- drivers/net/vxlan.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
drivers/net/vxlan.c +11 -3
@@ -2219,7 +2219,7 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,struct pcpu_sw_netstats *tx_stats, *rx_stats;union vxlan_addr loopback;union vxlan_addr *remote_ip = &dst_vxlan->default_dst.remote_ip; -struct net_device *dev = skb->dev; +struct net_device *dev;int len = skb->len; tx_stats =this_cpu_ptr (src_vxlan->dev->tstats);
@@ -2239,9 +2239,15 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan,#endif } +rcu_read_lock (); + dev = skb->dev; +if (unlikely (!(dev->flags &IFF_UP ))) { +kfree_skb (skb); + goto drop; + } +if (dst_vxlan->cfg.flags &VXLAN_F_LEARN ) -vxlan_snoop (skb->dev, &loopback,eth_hdr (skb)->h_source,0 , - vni); +vxlan_snoop (dev, &loopback,eth_hdr (skb)->h_source,0 , vni);u64_stats_update_begin (&tx_stats->syncp); tx_stats->tx_packets++;
@@ -2254,8 +2260,10 @@ static void vxlan_encap_bypass(struct sk_buff *skb, struct vxlan_dev *src_vxlan, rx_stats->rx_bytes += len;u64_stats_update_end (&rx_stats->syncp); }else { +drop: dev->stats.rx_dropped++; } +rcu_read_unlock (); } static int encap_bypass_if_local(struct sk_buff *skb, struct net_device *dev,
Signature · 1 line
2.20.1.791.gb4d0f1c61a-goog
Quoted 8 lines · Eric Dumazet
netif_rx()must be called under a strict contract.At device dismantle phase, core networking clears
IFF_UPandflush_all_backlogs()is called after rcu grace period to make sure no incoming packet might be in a cpu backlog and still referencing the device.
Quoted 6 lines
Most drivers call
netif_rx()from their interrupt handler, and since the interrupts are disabled at device dismantle,netif_rx()does not have to checkdev->flags&IFF_UPVirtual drivers do not have this guarantee, and must therefore make the check themselves.
Since other netif_rx() callers seem to have the same issue, I wonder if we simply could add a generic check.
Most high performance drivers no longer call netif_rx() so an extra check in it wont be a problem.
David Miller
maintainerFrom: Eric Dumazet <[email protected]>
Date: Thu, 7 Feb 2019 12:27:38 -0800
Quoted 16 lines
netif_rx()must be called under a strict contract.At device dismantle phase, core networking clears
IFF_UPandflush_all_backlogs()is called after rcu grace period to make sure no incoming packet might be in a cpu backlog and still referencing the device.Most drivers call
netif_rx()from their interrupt handler, and since the interrupts are disabled at device dismantle,netif_rx()does not have to checkdev->flags&IFF_UPVirtual drivers do not have this guarantee, and must therefore make the check themselves.
Otherwise we risk use-after-free and/or crashes.
Quoted 8 lines
Note this patch also fixes a small issue that came with commit
ce6502a8f957("vxlan: fix a use after free invxlan_encap_bypass"), since thedev->stats.rx_droppedchange was done on the wrong device.Fixes d342894c5d2f ("vxlan: virtual extensible lan")Fixes ce6502a8f957 ("vxlan: fix a use after free in vxlan_encap_bypass")Signed-off-by Eric Dumazet <[email protected]>
Applied and queued up for -stable.