Re: v2.6.26-rc7: BUG: unable to handle kernel NULL pointer dereference

["Vegard Nossum" <vegard.nossum@gmail.com>]

On Sun, Jun 22, 2008 at 4:47 PM, Vegard Nossum <vegard.nossum@gmail.com> wrote:
> On Sun, Jun 22, 2008 at 2:56 PM, Vegard Nossum <vegard.nossum@gmail.com> wrote:
>> Initializing CPU#1
>> APIC error on CPU0: 00(40)
>> Stuck ??
>> Inquiring remote APIC #1...
>> ... APIC #1 ID: failed
>> ... APIC #1 VERSION: failed
>> ... APIC #1 SPIV: failed
>
> Arch-specific __cpu_up() failed...
>
>> BUG: unable to handle kernel NULL pointer dereference at 00000024
>> IP: [<c01e0f06>] sysfs_remove_group+0x16/0xc0
>> *pde = 00000000
>> Oops: 0000 [#1] PREEMPT SMP DEBUG_PAGEALLOC
>>
>> Pid: 3994, comm: bash Not tainted (2.6.26-rc7-00002-g8b2e474-dirty #29)
>> EIP: 0060:[<c01e0f06>] EFLAGS: 00010282 CPU: 0
>> EIP is at sysfs_remove_group+0x16/0xc0
>> EAX: 00000008 EBX: 00000001 ECX: 00000004 EDX: c0694c14
>> ESI: 00000004 EDI: c07015fc EBP: f3cbfe9c ESP: f3cbfe80
>>  DS: 007b ES: 007b FS: 00d8 GS: 0033 SS: 0068
>> Process bash (pid: 3994, ti=f3cbe000 task=f3cb4f60 task.ti=f3cbe000)
>> Stack: c0302262 f4bcda40 f3cbfe98 00000008 00000001 00000004 00000000 f3cbfea8
>>       c05880b2 c07576c4 f3cbfec8 c014f567 00000001 00000004 c0753c90 0000001f
>>       00000001 fffffffb f3cbfedc c014f5d9 0000001f 00000000 00000004 f3cbff00
>> Call Trace:
>>  [<c0302262>] ? device_unregister+0x12/0x20
>
> This is line 131 of fs/sysfs/group.c:
>
>        struct sysfs_dirent *dir_sd = kobj->sd;
>
>>  [<c05880b2>] ? topology_cpu_callback+0x32/0x70
>>  [<c014f567>] ? notifier_call_chain+0x37/0x70
>>  [<c014f5d9>] ? __raw_notifier_call_chain+0x19/0x20
>>  [<c0586f6e>] ? _cpu_up+0xee/0x100
>
> This is line 314 of _cpu_up():
>
> out_notify:
>        if (ret != 0)
>                __raw_notifier_call_chain(&cpu_chain,
>                                CPU_UP_CANCELED | mod, hcpu, nr_calls, NULL);
>
>>  [<c0586fc9>] ? cpu_up+0x49/0x70
>>  [<c05678d8>] ? store_online+0x58/0x80
>>  [<c0567880>] ? store_online+0x0/0x80
>>  [<c0302a6b>] ? sysdev_store+0x2b/0x40
>>  [<c01df0f2>] ? sysfs_write_file+0xa2/0x100
>>  [<c01a0d06>] ? vfs_write+0x96/0x130
>>  [<c01df050>] ? sysfs_write_file+0x0/0x100
>>  [<c01a13cd>] ? sys_write+0x3d/0x70
>>  [<c010831b>] ? sysenter_past_esp+0x78/0xd1
>>  =======================
>> Code: 8b 43 04 83 c3 04 85 c0 75 ed 5b 5e 5d c3 8d b4 26 00 00 00 00 55 89 e5 83 ec 1c 89 7d fc 89 d7 89 5d f4 89 75 f8 89 45 f0 8b 12 <8b> 70 1c 85 d2 74 53 89 f0 e8 bc f5 ff ff 85 c0 89 c3 74 59 8b
>> EIP: [<c01e0f06>] sysfs_remove_group+0x16/0xc0 SS:ESP 0068:f3cbfe80
>>
>>
>
> ...adding a few related Ccs.

Sorry, forgot one. This is probably the root cause of the crash:

commit e37d05dad7ff9744efd8ea95a70d389e9a65a6fc
Author: Mike Travis <travis@sgi.com>
Date:   Thu May 1 04:35:16 2008 -0700

    cpu: change cpu_sys_devices from array to per_cpu variable

    Change cpu_sys_devices from array to per_cpu variable in drivers/base/cpu.c.

which had this hunk:

 struct sys_device *get_cpu_sysdev(unsigned cpu)
 {
-       if (cpu < NR_CPUS)
-               return cpu_sys_devices[cpu];
+       if (cpu < nr_cpu_ids && cpu_possible(cpu))
+               return per_cpu(cpu_sys_devices, cpu);
        else
                return NULL;
 }

...so now we're trying to unregister a NULL device (which of course
has no ->kobj).


Vegard

-- 
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036