Re: 2.6.23-rc2-mm1: sleeping function called from invalid context at kernel/mutex.c:86

[Josh Triplett <josh@freedesktop.org>]

Andrew Morton wrote:
> On Fri, 10 Aug 2007 01:23:07 +0200
> Mariusz Kozlowski <m.kozlowski@tuxland.pl> wrote:
> 
>> Hello,
>>
>> 	This probably doesn't have great impact ;) but ...
>> To reproduce: run torture tests for RCU and then sysrq+q.
>>
>> SysRq : Show Pending Timers
>> Timer List Version: v0.3
>> HRTIMER_MAX_CLOCK_BASES: 2
>> now at 1764338760370 nsecs
>>
>> cpu: 0
>>  clock 0:
>>   .index:      0
>>   .resolution: 1 nsecs
>>   .get_time:   ktime_get_real
>>   .offset:     1186699025823815427 nsecs
>> active timers:
>>  clock 1:
>>   .index:      1
>>   .resolution: 1 nsecs
>>   .get_time:   ktime_get
>>   .offset:     0 nsecs
>> active timers:
>>  #0: <3>BUG: sleeping function called from invalid context at kernel/mutex.c:86
>> in_atomic():1, irqs_disabled():1
>> INFO: lockdep is turned off.
>> irq event stamp: 0
>> hardirqs last  enabled at (0): [<00000000>] 0x0
>> hardirqs last disabled at (0): [<c0117def>] copy_process+0x4a8/0x144c
>> softirqs last  enabled at (0): [<c0117e0d>] copy_process+0x4c6/0x144c
>> softirqs last disabled at (0): [<00000000>] 0x0
>>  [<c0104869>] show_trace_log_lvl+0x1a/0x30
>>  [<c01053ad>] show_trace+0x12/0x14
>>  [<c0105515>] dump_stack+0x15/0x17
>>  [<c0114da7>] __might_sleep+0xb7/0xc9
>>  [<c0451771>] mutex_lock+0x15/0x1f
>>  [<c0141b75>] lookup_module_symbol_name+0x17/0xc0
>>  [<c014272a>] lookup_symbol_name+0x3f/0x43
>>  [<c013287e>] print_name_offset+0x1f/0x96
>>  [<c01330f7>] timer_list_show+0x802/0xcbd
>>  [<c01335be>] sysrq_timer_list_show+0xc/0xe
>>  [<c02cc4a1>] sysrq_handle_show_timers+0x8/0xa
>>  [<c02cc3ac>] __handle_sysrq+0x7b/0x115
>>  [<c02cc466>] handle_sysrq+0x20/0x24
>>  [<c02c69c1>] kbd_event+0x3a8/0x5c7
>>  [<c0362f8f>] input_pass_event+0x8f/0x91
>>  [<c0363e77>] input_handle_event+0x98/0x38d
>>  [<c0364e6d>] input_event+0x54/0x67
>>  [<c03682c2>] atkbd_interrupt+0x200/0x59e
>>  [<c0360cd0>] serio_interrupt+0x7c/0x80
>>  [<c0361965>] i8042_interrupt+0x17a/0x289
>>  [<c0147a5d>] handle_IRQ_event+0x28/0x59
>>  [<c01493ca>] handle_level_irq+0xad/0x10b
>>  [<c0105a13>] do_IRQ+0x93/0xd0
>>  [<c010441e>] common_interrupt+0x2e/0x34
>>  [<df39d7e3>] rcu_read_delay+0x8/0x36 [rcutorture]
>>  [<df39d99a>] rcu_torture_reader+0x6e/0x169 [rcutorture]
>>  [<c012c11e>] kthread+0x36/0x58
>>  [<c010451b>] kernel_thread_helper+0x7/0x1c
>>  =======================
> 
> We seem to have made a mess in there.  timer_list_show() ends up calling
> lookup_module_symbol_name(), which takes a mutex.  However print_symbol()
> (which is called at oops time, interrupt time, etc) calls
> module_address_lookup(), which is basically the same, only it doesn't take
> the mutex.
> 
> I guess a quicky fix would be to switch
> kernel/time/timer_list.c:print_name_offset() from
> lookup_module_symbol_name() to module_address_lookup().  But we'd still
> have a mess in there.
> 
> (adds ccs, runs away)

I don't think rcutorture matters for this bug.  As far as I can tell, Andrew's
description of this problem will always apply to this particular sysrq: the
keyboard interrupt leads to handle_sysrq, which leads to timer_list_show,
which leads to lookup_module_symbol_name, which acquires a mutex.

- Josh Triplett