On 09/29/2014 07:11 AM, Peter Zijlstra wrote:
> On Sun, Sep 28, 2014 at 12:09:09AM -0400, Sasha Levin wrote:
>
>> > [ 690.801720] 2 locks held by trinity-c95/17888:
>> > [ 690.801738] #0: (cpu_hotplug.lock){++++++}, at: get_online_cpus
>> > (kernel/cpu.c:92)
>> > [ 690.801754] #1: (&ctx->lock){-.-...}, at: perf_lock_task_context
>> > (kernel/events/core.c:988)
>> > [ 690.801758]
>> > [ 690.801758] stack backtrace:
>> > [ 690.801766] CPU: 21 PID: 17888 Comm: trinity-c95 Not tainted
>> > 3.17.0-rc6-next-20140926-sasha-00051-g9253dff-dirty #1242
>> > [ 690.801779] ffffffff92b7f320 0000000000000000 ffffffff92afbee0
>> > ffff8804078179c8
>> > [ 690.801798] ffffffff8ef0070f 0000000000000011 ffffffff92ab6aa0
>> > ffff880407817a18
>> > [ 690.801813] ffffffff8a24ec2c ffff880407817aa8 ffff880409c00000
>> > ffff880407817a18
>> > [ 690.801818] Call Trace:
>> > [ 690.801836] dump_stack (lib/dump_stack.c:52)
>> > [ 690.801845] print_circular_bug (kernel/locking/lockdep.c:1217)
>> > [ 690.801856] __lock_acquire (kernel/locking/lockdep.c:1842
>> > kernel/locking/lockdep.c:1947 kernel/locking/lockdep.c:2133
>> > kernel/locking/lockdep.c:3184)
>> > [ 690.801872] lock_acquire (kernel/locking/lockdep.c:3610)
>> > [ 690.801892] _raw_spin_lock (include/linux/spinlock_api_smp.h:143
>> > kernel/locking/spinlock.c:151)
>> > [ 690.801921] __queue_work (kernel/workqueue.c:1325)
>> > [ 690.801943] queue_work_on (kernel/workqueue.c:1403)
>> > [ 690.801956] free_object (lib/debugobjects.c:209)
>> > [ 690.801967] __debug_check_no_obj_freed (lib/debugobjects.c:718)
>> > [ 690.801983] debug_check_no_obj_freed (lib/debugobjects.c:727)
>> > [ 690.801995] kmem_cache_free (mm/slub.c:2687 mm/slub.c:2715)
>> > [ 690.802016] free_task (kernel/fork.c:221)
>> > [ 690.802026] __put_task_struct (kernel/fork.c:251)
>> > [ 690.802037] put_ctx (include/linux/sched.h:1864
>> > kernel/events/core.c:904)
>> > [ 690.802049] find_get_context (kernel/events/core.c:913
>> > kernel/events/core.c:3222)
>> > [ 690.802078] SYSC_perf_event_open (kernel/events/core.c:7347)
>> > [ 690.802111] SyS_perf_event_open (kernel/events/core.c:7210)
>> > [ 690.802120] tracesys_phase2 (arch/x86/kernel/entry_64.S:529)
> This doesn't make sense; perf_lock_task_context() isn't supposed to
> return with ctx->lock held and therefore it should not still be held in
> find_get_context() when calling put_ctx().
>
> Now, the only put_ctx() call in find_get_context() is in the !ctx path
> of the perf_lock_task_context() call, furthermore there is a
> mutex_lock() - which implies a might_sleep() - before that, so we can't
> still be holding a spinlock().
I think you missed the put_ctx() call in the other branch in find_get_context(),
which is the call described by the trace above:
find_get_context()
unclone_ctx()
put_ctx()
Thanks,
Sasha
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
