[SRU][Xenial][Bionic][Cosmic][PATCH] Fix for LP:#1801878

Gavin Guo gavin.guo at canonical.com
Tue Nov 6 14:46:33 UTC 2018


BugLink: https://bugs.launchpad.net/bugs/1801878

[Impact]

NULL pointer access happens when trying to access dst_orig->ops.

The function xfrm_lookup() calls xfrm_lookup_with_ifid() and there is 
a line inside trying to access dst_orig->ops and it's exactly where the 
panicing happens: 

u16 family = dst_orig->ops->family; 

As you can see that the symbol offset of ops is about 32(0x20) which 
definitely is the error message shows in the kern.log: 

[267265.140511] BUG: unable to handle kernel NULL pointer dereference 
at 0000000000000020 

struct dst_entry { 
struct callback_head callback_head; /* 0 16 */ 
struct dst_entry * child; /* 16 8 */ 
struct net_device * dev; /* 24 8 */ 
struct dst_ops * ops; <-- /* 32 8 */ 

The oops:
BUG: unable to handle kernel NULL pointer dereference at 0000000000000020 
IP: xfrm_lookup+0x31/0x870 
PGD 0 P4D 0 
Oops: 0000 [#1] SMP PTI 
CPU: 5 PID: 0 Comm: swapper/5 Not tainted 4.15.0-36-generic #39~16.04.1-Ubuntu 
Hardware name: Xen HVM domU, BIOS 4.2.amazon 08/24/2006 
RIP: 0010:xfrm_lookup+0x31/0x870 
RSP: 0018:ffff98b542343a48 EFLAGS: 00010246 
RAX: 0000000000000000 RBX: ffff98b542343ac8 RCX: 0000000000000000 
RDX: ffff98b542343ac8 RSI: 0000000000000000 RDI: ffffffffb39e4380 
RBP: ffff98b542343ab8 R08: 0000000000000002 R09: 0000000000000000 
R10: 0000000000000020 R11: 000000007fb56465 R12: 0000000000000000 
R13: ffffffffb39e4380 R14: 0000000000000002 R15: ffffffffb39e4380 
FS: 0000000000000000(0000) GS:ffff98b542340000(0000) knlGS:0000000000000000 
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 
CR2: 0000000000000020 CR3: 00000004ed40a001 CR4: 00000000001606e0 
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 
Call Trace: 
<IRQ> 
? __xfrm_policy_check+0x41d/0x630 
__xfrm_route_forward+0xa3/0x110 
ip_forward+0x38c/0x470 
? ip_route_input_noref+0x28/0x40 
ip_rcv_finish+0x124/0x410 
ip_rcv+0x28e/0x3b0 
? inet_del_offload+0x40/0x40 
__netif_receive_skb_core+0x879/0xba0 
? __skb_checksum+0x188/0x2c0 
__netif_receive_skb+0x18/0x60 
? __netif_receive_skb+0x18/0x60 
netif_receive_skb_internal+0x37/0xe0 
? tcp4_gro_complete+0x86/0x90 
napi_gro_complete+0x73/0x90 
dev_gro_receive+0x2ee/0x5c0 
napi_gro_frags+0xa3/0x230 
ena_clean_rx_irq+0x486/0x7c0 [ena] 
ena_io_poll+0x41d/0x770 [ena] 
net_rx_action+0x265/0x3b0 
__do_softirq+0xf5/0x28f 
irq_exit+0xb8/0xc0 
xen_evtchn_do_upcall+0x30/0x40 
xen_hvm_callback_vector+0x84/0x90 

[Fix]
The patch tries to avoid the NULL pointer access before the line
mentioned "dst_orig->ops->family" in function __xfrm_route_forward.
And the function calling sequence is: 

__xfrm_route_forward -> xfrm_lookup -> xfrm_lookup_with_ifid 

It definitely avoids the NULL pointer access in the 
xfrm_lookup_with_ifid.

commit c5e39ef174ad34cd4d26af3a83bdbccddd2ad9d6
Author: Steffen Klassert <steffen.klassert at secunet.com>
Date:   Tue Sep 11 10:31:15 2018 +0200

    xfrm: Fix NULL pointer dereference when skb_dst_force clears the dst_entry.
    
    Since commit 222d7dbd258d ("net: prevent dst uses after free")
    skb_dst_force() might clear the dst_entry attached to the skb.
    The xfrm code don't expect this to happen, so we crash with
    a NULL pointer dereference in this case. Fix it by checking
    skb_dst(skb) for NULL after skb_dst_force() and drop the packet
    in cast the dst_entry was cleared.
    
[Test]
The fix has been tested in the production system with the IPSec enabled.

Steffen Klassert (1):
  xfrm: Fix NULL pointer dereference when skb_dst_force clears the
    dst_entry.

 net/xfrm/xfrm_output.c | 4 ++++
 net/xfrm/xfrm_policy.c | 4 ++++
 2 files changed, 8 insertions(+)

-- 
2.7.4




More information about the kernel-team mailing list