oops when bringing up loopback device

oops when bringing up loopback device

[Stephen Hemminger]

I am seeing this on one of my boxes with current net-2.6 tree.

Unable to handle kernel NULL pointer dereference at virtual address 00000000
 printing eip:
00000000
*pde = 00000000
Oops: 0000 [#1]
SMP
Modules linked in: evdev e100 skge mii
CPU:    1
EIP:    0060:[<00000000>]    Not tainted VLI
EFLAGS: 00010286   (2.6.12-rc1-tcp)
EIP is at 0x0
eax: c0413b6c   ebx: 00000001   ecx: 0000000f   edx: f6aa9000
esi: 000007d0   edi: fffc727d   ebp: f6aa9d14   esp: f6aa9d00
ds: 007b   es: 007b   ss: 0068
Process ip (pid: 1597, threadinfo=f6aa9000 task=f64d3020)
Stack: c02fb6ce 00000001 f60022d0 f5fdd740 f6fc1300 f6aa9d70 c0331765 f6aa9d5c
       c0000000 fffffff0 00000000 000000d0 f6aa9d44 00000246 0100007f 00aa9d88
       00000002 00000020 f7f7da80 00000000 f6fc1300 f6aa9dd8 f7f7d780 ffffff97
Call Trace:
 [<c0103ec6>] show_stack+0xa6/0xb0
 [<c0104042>] show_registers+0x152/0x1c0
 [<c010424d>] die+0xed/0x170
 [<c011593d>] do_page_fault+0x4fd/0x685
 [<c0103b33>] error_code+0x2b/0x30
 [<c0331765>] fn_hash_insert+0x205/0x420
 [<c032e9e6>] fib_magic+0x106/0x120
 [<c032ea66>] fib_add_ifaddr+0x66/0x160
 [<c032ed7c>] fib_inetaddr_event+0x1c/0x60
 [<c0129fa8>] notifier_call_chain+0x28/0x50
 [<c0326eb9>] inet_insert_ifa+0xd9/0x170
 [<c03280ec>] inetdev_event+0x1dc/0x1f0
 [<c0129fa8>] notifier_call_chain+0x28/0x50
 [<c02e7ecf>] dev_open+0x5f/0x90
 [<c02e94d2>] dev_change_flags+0x52/0x120
 [<c03279d0>] devinet_ioctl+0x520/0x5d0
 [<c0329968>] inet_ioctl+0x98/0xe0
 [<c02df470>] sock_ioctl+0x1a0/0x250
 [<c016dc36>] do_ioctl+0x76/0x90
 [<c016ddcd>] vfs_ioctl+0x5d/0x1e0
 [<c016df89>] sys_ioctl+0x39/0x60
 [<c0103033>] sysenter_past_esp+0x54/0x75

Re: oops when bringing up loopback device

[David S. Miller]

On Mon, 28 Mar 2005 16:02:48 -0800
Stephen Hemminger <shemminger@osdl.org> wrote:

> I am seeing this on one of my boxes with current net-2.6 tree.

You're missing the fix to net/ipv4/ip_mp_alg.h for some reason,
it is upstream for two days now and I added it to my tree the
weekend you left for spring break.

# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
#   2005/03/19 16:45:16-08:00 davem@sunset.davemloft.net 
#   [IPV4]: Check multipath ops func pointers against NULL.
#   
#   Multipath algorithms are not required to implement
#   all of the methods.  Just the ones they need.
#   
#   Signed-off-by: David S. Miller <davem@davemloft.net>
# 
# include/net/ip_mp_alg.h
#   2005/03/19 16:44:15-08:00 davem@sunset.davemloft.net +6 -4
#   [IPV4]: Check multipath ops func pointers against NULL.
# 
diff -Nru a/include/net/ip_mp_alg.h b/include/net/ip_mp_alg.h
--- a/include/net/ip_mp_alg.h	2005-03-28 16:21:56 -08:00
+++ b/include/net/ip_mp_alg.h	2005-03-28 16:21:56 -08:00
@@ -36,7 +36,8 @@
 #ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
 	struct ip_mp_alg_ops *ops = ip_mp_alg_table[rth->rt_multipath_alg];
 
-	if (ops && (rth->u.dst.flags & DST_BALANCED)) {
+	if (ops && ops->mp_alg_select_route &&
+	    (rth->u.dst.flags & DST_BALANCED)) {
 		ops->mp_alg_select_route(flp, rth, rp);
 		return 1;
 	}
@@ -52,7 +53,7 @@
 	for (i = IP_MP_ALG_NONE; i <= IP_MP_ALG_MAX; i++) {
 		struct ip_mp_alg_ops *ops = ip_mp_alg_table[i];
 
-		if (ops)
+		if (ops && ops->mp_alg_flush)
 			ops->mp_alg_flush();
 	}
 #endif
@@ -66,7 +67,7 @@
 #ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
 	struct ip_mp_alg_ops *ops = ip_mp_alg_table[rth->rt_multipath_alg];
 
-	if (ops)
+	if (ops && ops->mp_alg_set_nhinfo)
 		ops->mp_alg_set_nhinfo(network, netmask, prefixlen, nh);
 #endif
 }
@@ -76,7 +77,8 @@
 #ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
 	struct ip_mp_alg_ops *ops = ip_mp_alg_table[rth->rt_multipath_alg];
 
-	if (ops && (rth->u.dst.flags & DST_BALANCED))
+	if (ops && ops->mp_alg_remove &&
+	    (rth->u.dst.flags & DST_BALANCED))
 		ops->mp_alg_remove(rth);
 #endif
 }