Re: [PATCH] cpuset: Rework sched domains and CPU hotplug handling (take 4)

[Max Krasnyansky <maxk@qualcomm.com>]

Ingo Molnar wrote:
> * Ingo Molnar <mingo@elte.hu> wrote:
> 
>> * Paul Jackson <pj@sgi.com> wrote:
>>
>>>   Acked-by: Paul Jackson <pj@sgi.com>
>>>
>>> ... based on code reading and comparing with the
>>> previous version - looks good.  Nice work, Max.
>>> Thanks.
>> applied to tip/sched/cpuset, thanks. (will show up in tip/sched/urgent 
>> as well soon, for v2.6.27 merging.)
> 
> FYI, this new lockdep warning showed up in -tip testing, after i added 
> this patch.

Hmm, unless I'm missing something this one is unrelated. There are no cpu
hotplug, sched or cpusets paths in the trace besides cpu_maps_update_begin().
But that one is taken in the regular destroy_workqueue() path.

The issues is polldev_mutex and cpu_add_remove_lock nesting. I bet you can
trigger that without cpusets. CC'ing Dmitry Torokhov.

btw It seems to be triggered when X closes polled input device.
There aren't that many of them
   $ git grep input_register_polled_device | wc -l
   16
What do you have hooked up to the test box ?

Max


> [   59.750200] =======================================================
> [   59.750200] [ INFO: possible circular locking dependency detected ]
> [   59.750200] 2.6.27-rc3-tip-00076-g75f9a29-dirty #1
> [   59.750200] -------------------------------------------------------
> [   59.750200] Xorg/6623 is trying to acquire lock:
> [   59.750200]  (cpu_add_remove_lock){--..}, at: [<c0147dae>] cpu_maps_update_begin+0x14/0x16
> [   59.750200] 
> [   59.750200] but task is already holding lock:
> [   59.750200]  (polldev_mutex){--..}, at: [<c07eb66f>] input_close_polled_device+0x22/0x47
> [   59.750200] 
> [   59.750200] which lock already depends on the new lock.
> [   59.750200] 
> [   59.750200] 
> [   59.750200] the existing dependency chain (in reverse order) is:
> [   59.750200] 
> [   59.750200] -> #5 (polldev_mutex){--..}:
> [   59.750200]        [<c01632e3>] __lock_acquire+0x848/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c09eaf41>] __mutex_lock_common+0x8a/0x278
> [   59.750200]        [<c09eb15d>] mutex_lock_interruptible_nested+0x2e/0x35
> [   59.750200]        [<c07eb745>] input_open_polled_device+0x1c/0xa3
> [   59.750200]        [<c07e8f84>] input_open_device+0x5a/0x86
> [   59.750200]        [<c07eded0>] evdev_open+0x103/0x14e
> [   59.750200]        [<c07ea620>] input_open_file+0x44/0x60
> [   59.750200]        [<c019ff1c>] chrdev_open+0x106/0x11d
> [   59.750200]        [<c019c269>] __dentry_open+0x119/0x1f0
> [   59.750200]        [<c019c364>] nameidata_to_filp+0x24/0x38
> [   59.750200]        [<c01a6952>] do_filp_open+0x309/0x5b2
> [   59.750200]        [<c019c04e>] do_sys_open+0x47/0xc1
> [   59.750200]        [<c019c114>] sys_open+0x23/0x2b
> [   59.750200]        [<c011bb2f>] sysenter_do_call+0x12/0x43
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] -> #4 (&dev->mutex){--..}:
> [   59.750200]        [<c01632e3>] __lock_acquire+0x848/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c09eaf41>] __mutex_lock_common+0x8a/0x278
> [   59.750200]        [<c09eb15d>] mutex_lock_interruptible_nested+0x2e/0x35
> [   59.750200]        [<c07ea402>] input_register_handle+0x26/0x80
> [   59.750200]        [<c0424336>] kbd_connect+0x6c/0x95
> [   59.750200]        [<c07e8a46>] input_attach_handler+0x38/0x6b
> [   59.750200]        [<c07ea4d7>] input_register_handler+0x7b/0xaf
> [   59.750200]        [<c0f63917>] kbd_init+0x6b/0x87
> [   59.750200]        [<c0f63a40>] vty_init+0xd3/0xdc
> [   59.750200]        [<c0f63405>] tty_init+0x198/0x19c
> [   59.750200]        [<c0101139>] do_one_initcall+0x42/0x133
> [   59.750200]        [<c0f3f610>] kernel_init+0x17b/0x1e2
> [   59.750200]        [<c011c85f>] kernel_thread_helper+0x7/0x10
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] -> #3 (input_mutex){--..}:
> [   59.750200]        [<c01632e3>] __lock_acquire+0x848/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c09eaf41>] __mutex_lock_common+0x8a/0x278
> [   59.750200]        [<c09eb15d>] mutex_lock_interruptible_nested+0x2e/0x35
> [   59.750200]        [<c07ea2da>] input_register_device+0xff/0x171
> [   59.750200]        [<c04080cd>] acpi_button_add+0x320/0x421
> [   59.750200]        [<c040634f>] acpi_device_probe+0x3f/0x8d
> [   59.750200]        [<c048e8fa>] driver_probe_device+0xc3/0x156
> [   59.750200]        [<c048e9cf>] __driver_attach+0x42/0x64
> [   59.750200]        [<c048e22f>] bus_for_each_dev+0x43/0x65
> [   59.750200]        [<c048e713>] driver_attach+0x19/0x1b
> [   59.750200]        [<c048dc2d>] bus_add_driver+0xaf/0x1b5
> [   59.750200]        [<c048eb72>] driver_register+0x76/0xd2
> [   59.750200]        [<c040665e>] acpi_bus_register_driver+0x3f/0x41
> [   59.750200]        [<c0f61672>] acpi_button_init+0x37/0x56
> [   59.750200]        [<c0101139>] do_one_initcall+0x42/0x133
> [   59.750200]        [<c0f3f230>] do_async_initcalls+0x1f/0x2f
> [   59.750200]        [<c015475f>] run_workqueue+0xb7/0x189
> [   59.750200]        [<c01551ab>] worker_thread+0xbb/0xc7
> [   59.750200]        [<c015764d>] kthread+0x40/0x67
> [   59.750200]        [<c011c85f>] kernel_thread_helper+0x7/0x10
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] -> #2 (async_work){--..}:
> [   59.750200]        [<c01632e3>] __lock_acquire+0x848/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c015475a>] run_workqueue+0xb2/0x189
> [   59.750200]        [<c01551ab>] worker_thread+0xbb/0xc7
> [   59.750200]        [<c015764d>] kthread+0x40/0x67
> [   59.750200]        [<c011c85f>] kernel_thread_helper+0x7/0x10
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] -> #1 (kasyncinit){--..}:
> [   59.750200]        [<c01632e3>] __lock_acquire+0x848/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c015499a>] cleanup_workqueue_thread+0x2b/0x5e
> [   59.750200]        [<c0154a40>] destroy_workqueue+0x61/0x89
> [   59.750200]        [<c0f3f634>] kernel_init+0x19f/0x1e2
> [   59.750200]        [<c011c85f>] kernel_thread_helper+0x7/0x10
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] -> #0 (cpu_add_remove_lock){--..}:
> [   59.750200]        [<c01631b4>] __lock_acquire+0x719/0x9ab
> [   59.750200]        [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]        [<c09eaf41>] __mutex_lock_common+0x8a/0x278
> [   59.750200]        [<c09eb1c7>] mutex_lock_nested+0x2e/0x36
> [   59.750200]        [<c0147dae>] cpu_maps_update_begin+0x14/0x16
> [   59.750200]        [<c0154a05>] destroy_workqueue+0x26/0x89
> [   59.750200]        [<c07eb688>] input_close_polled_device+0x3b/0x47
> [   59.750200]        [<c07e9056>] input_close_device+0x45/0x61
> [   59.750200]        [<c07edd49>] evdev_release+0x7f/0x99
> [   59.750200]        [<c019e7d9>] __fput+0xb3/0x135
> [   59.750200]        [<c019eb5f>] fput+0x1c/0x21
> [   59.750200]        [<c019bffd>] filp_close+0x4c/0x56
> [   59.750200]        [<c019d1d6>] sys_close+0x6d/0xa6
> [   59.750200]        [<c011bb2f>] sysenter_do_call+0x12/0x43
> [   59.750200]        [<ffffffff>] 0xffffffff
> [   59.750200] 
> [   59.750200] other info that might help us debug this:
> [   59.750200] 
> [   59.750200] 3 locks held by Xorg/6623:
> [   59.750200]  #0:  (&evdev->mutex){--..}, at: [<c07edd31>] evdev_release+0x67/0x99
> [   59.750200]  #1:  (&dev->mutex){--..}, at: [<c07e9030>] input_close_device+0x1f/0x61
> [   59.750200]  #2:  (polldev_mutex){--..}, at: [<c07eb66f>] input_close_polled_device+0x22/0x47
> [   59.750200] 
> [   59.750200] stack backtrace:
> [   59.750200] Pid: 6623, comm: Xorg Not tainted 2.6.27-rc3-tip-00076-g75f9a29-dirty #1
> [   59.750200]  [<c016189b>] print_circular_bug_tail+0x5d/0x68
> [   59.750200]  [<c01631b4>] __lock_acquire+0x719/0x9ab
> [   59.750200]  [<c01634b6>] lock_acquire+0x70/0x97
> [   59.750200]  [<c0147dae>] ? cpu_maps_update_begin+0x14/0x16
> [   59.750200]  [<c09eaf41>] __mutex_lock_common+0x8a/0x278
> [   59.750200]  [<c0147dae>] ? cpu_maps_update_begin+0x14/0x16
> [   59.750200]  [<c016261e>] ? trace_hardirqs_on_caller+0x94/0xcd
> [   59.750200]  [<c09eb1c7>] mutex_lock_nested+0x2e/0x36
> [   59.750200]  [<c0147dae>] ? cpu_maps_update_begin+0x14/0x16
> [   59.750200]  [<c0147dae>] cpu_maps_update_begin+0x14/0x16
> [   59.750200]  [<c0154a05>] destroy_workqueue+0x26/0x89
> [   59.750200]  [<c07eb688>] input_close_polled_device+0x3b/0x47
> [   59.750200]  [<c07e9056>] input_close_device+0x45/0x61
> [   59.750200]  [<c07edd49>] evdev_release+0x7f/0x99
> [   59.750200]  [<c019e7d9>] __fput+0xb3/0x135
> [   59.750200]  [<c019eb5f>] fput+0x1c/0x21
> [   59.750200]  [<c019bffd>] filp_close+0x4c/0x56
> [   59.750200]  [<c019d1d6>] sys_close+0x6d/0xa6
> [   59.750200]  [<c011bb2f>] sysenter_do_call+0x12/0x43
> [   59.750200]  [<c0110000>] ? x86_decode_insn+0x46e/0x942
> [   59.750200]  =======================
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>