Bug#499386: [etch] NULL pointer dereference in sys_exit_group

[Jonathan Nieder <jrnieder@gmail.com>]

Hi Shai,

Four years ago, you wrote[1]:

> After several cases of kernel panic which we blamed on just using old 
> hardware, we caught this one in the log. The system is Etch, it serves as a 
> firewall & dhcp server on one card, and the other is connected to an ADSL 
> modem/router for uplink.
[...]
> tarante kernel: BUG: unable to handle kernel NULL pointer dereference at virtual address 00000045
> tarante kernel:  printing eip:
> tarante kernel: c0751ce0
> tarante kernel: *pde = 00000000
> tarante kernel: Oops: 0000 [#1]
> tarante kernel: Modules linked in: ipt_LOG xt_tcpudp xt_multiport ipt_MASQUERADE iptable_nat xt_state ip6table_filter ip6_tables iptable_filter ip_tables x_tables ip_nat_tftp ip_nat_snmp_basic ip_nat_sip ip_nat_pptp ip_nat_irc ip_nat_h323 ip_nat_ftp ip_nat_amanda ip_conntrack_tftp ip_conntrack_sip ip_conntrack_proto_sctp ip_conntrack_pptp ip_conntrack_netlink ip_nat ip_conntrack_netbios_ns ip_conntrack_irc ip_conntrack_h323 ip_conntrack_ftp ts_kmp ip_conntrack_amanda ip_conntrack nfnetlink ipv6 dm_snapshot dm_mirror dm_mod parport_pc parport psmouse floppy rtc serio_raw pcspkr evdev ext3 jbd ide_generic ide_cd cdrom ide_disk piix generic ide_core 8139cp 8139too mii processor
> tarante kernel: CPU:    0
> tarante kernel: EIP:    0060:[<c0751ce0>]    Not tainted VLI
> tarante kernel: EFLAGS: 00010083   (2.6.18-5-486 #1) 
> tarante kernel: EIP is at 0xc0751ce0
> tarante kernel: eax: c18bde2c   ebx: c18bde2c   ecx: 00000000   edx: 00000003
> tarante kernel: esi: c100de04   edi: 00000001   ebp: c10a5e80   esp: c10a5e60
> tarante kernel: ds: 007b   es: 007b   ss: 0068
> tarante kernel: Process named (pid: 1812, ti=c10a4000 task=c1b5e030 task.ti=c10a4000)
> tarante kernel: Stack: c01114ed 00000000 00000003 c18bdeb8 00c000ff 00000246 c1be93b4 c1b5e030 
> tarante kernel:        c10a5e94 c0112414 00000000 00000000 c1be9380 c10a5ec8 c0115a84 0000000b 
> tarante kernel:        c1b5e030 00000000 c0116c3c 0000000b bfc5e6e4 c10a5f14 c1b5e030 0000000b 
> tarante kernel: Call Trace:
> tarante kernel:  [<c01114ed>] __wake_up_common+0x2e/0x4d
> tarante kernel:  [<c0112414>] complete+0x1f/0x28
> tarante kernel:  [<c0115a84>] exit_mm+0x57/0xa3
> tarante kernel:  [<c0116c3c>] do_exit+0x18e/0x65f
> tarante kernel:  [<c0117169>] sys_exit_group+0x0/0xd
> tarante kernel:  [<c011d91c>] get_signal_to_deliver+0x310/0x320
[...]
> tarante kernel: Code: 81 00 00 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 63 07 00 00 00 00 00 00 da 8b d1 48 00 00 00 00 81 47 ec 45 00 00 00 00 <3e> d3 b1 45 00 00 00 00 0c 00 00 00 00 10 00 00 01 00 00 00 08 
> tarante kernel: EIP: [<c0751ce0>] 0xc0751ce0 SS:ESP 0068:c10a5e60
> tarante kernel:  <1>Fixing recursive fault but reboot is needed!

Alas, we missed this and then lost track of it.

Now I am curious:

 - do you still have access to this machine?
 - did you ever figure out what was causing the kernel panics?  was
   there a workaround or fix?
 - was the backtrace always the same when the bug was encountered?

If you can still reproduce this, I would be especially interested to
learn whether squeeze or wheezy is affected, so we can make sure it
gets fixed.

Sorry for the trouble, and thanks for reporting it.

Sincerely,
Jonathan