power: remove possible deadlock when unregistering power_supply
authorBenjamin Tissoires <benjamin.tissoires@redhat.com>
Mon, 25 Jun 2018 07:51:48 +0000 (09:51 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 4 Oct 2018 00:00:47 +0000 (17:00 -0700)
commite70f938a605a82c96760ee2afb6f9eac6d0e7088
tree8fbd08595c486914d45a8659366222dd816be174
parent1117e411a46c4df5230b469dc4641dcc46274191
power: remove possible deadlock when unregistering power_supply

[ Upstream commit 3ffa6583e24e1ad1abab836d24bfc9d2308074e5 ]

If a device gets removed right after having registered a power_supply node,
we might enter in a deadlock between the remove call (that has a lock on
the parent device) and the deferred register work.

Allow the deferred register work to exit without taking the lock when
we are in the remove state.

Stack trace on a Ubuntu 16.04:

[16072.109121] INFO: task kworker/u16:2:1180 blocked for more than 120 seconds.
[16072.109127]       Not tainted 4.13.0-41-generic #46~16.04.1-Ubuntu
[16072.109129] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[16072.109132] kworker/u16:2   D    0  1180      2 0x80000000
[16072.109142] Workqueue: events_power_efficient power_supply_deferred_register_work
[16072.109144] Call Trace:
[16072.109152]  __schedule+0x3d6/0x8b0
[16072.109155]  schedule+0x36/0x80
[16072.109158]  schedule_preempt_disabled+0xe/0x10
[16072.109161]  __mutex_lock.isra.2+0x2ab/0x4e0
[16072.109166]  __mutex_lock_slowpath+0x13/0x20
[16072.109168]  ? __mutex_lock_slowpath+0x13/0x20
[16072.109171]  mutex_lock+0x2f/0x40
[16072.109174]  power_supply_deferred_register_work+0x2b/0x50
[16072.109179]  process_one_work+0x15b/0x410
[16072.109182]  worker_thread+0x4b/0x460
[16072.109186]  kthread+0x10c/0x140
[16072.109189]  ? process_one_work+0x410/0x410
[16072.109191]  ? kthread_create_on_node+0x70/0x70
[16072.109194]  ret_from_fork+0x35/0x40
[16072.109199] INFO: task test:2257 blocked for more than 120 seconds.
[16072.109202]       Not tainted 4.13.0-41-generic #46~16.04.1-Ubuntu
[16072.109204] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[16072.109206] test            D    0  2257   2256 0x00000004
[16072.109208] Call Trace:
[16072.109211]  __schedule+0x3d6/0x8b0
[16072.109215]  schedule+0x36/0x80
[16072.109218]  schedule_timeout+0x1f3/0x360
[16072.109221]  ? check_preempt_curr+0x5a/0xa0
[16072.109224]  ? ttwu_do_wakeup+0x1e/0x150
[16072.109227]  wait_for_completion+0xb4/0x140
[16072.109230]  ? wait_for_completion+0xb4/0x140
[16072.109233]  ? wake_up_q+0x70/0x70
[16072.109236]  flush_work+0x129/0x1e0
[16072.109240]  ? worker_detach_from_pool+0xb0/0xb0
[16072.109243]  __cancel_work_timer+0x10f/0x190
[16072.109247]  ? device_del+0x264/0x310
[16072.109250]  ? __wake_up+0x44/0x50
[16072.109253]  cancel_delayed_work_sync+0x13/0x20
[16072.109257]  power_supply_unregister+0x37/0xb0
[16072.109260]  devm_power_supply_release+0x11/0x20
[16072.109263]  release_nodes+0x110/0x200
[16072.109266]  devres_release_group+0x7c/0xb0
[16072.109274]  wacom_remove+0xc2/0x110 [wacom]
[16072.109279]  hid_device_remove+0x6e/0xd0 [hid]
[16072.109284]  device_release_driver_internal+0x158/0x210
[16072.109288]  device_release_driver+0x12/0x20
[16072.109291]  bus_remove_device+0xec/0x160
[16072.109293]  device_del+0x1de/0x310
[16072.109298]  hid_destroy_device+0x27/0x60 [hid]
[16072.109303]  usbhid_disconnect+0x51/0x70 [usbhid]
[16072.109308]  usb_unbind_interface+0x77/0x270
[16072.109311]  device_release_driver_internal+0x158/0x210
[16072.109315]  device_release_driver+0x12/0x20
[16072.109318]  usb_driver_release_interface+0x77/0x80
[16072.109321]  proc_ioctl+0x20f/0x250
[16072.109325]  usbdev_do_ioctl+0x57f/0x1140
[16072.109327]  ? __wake_up+0x44/0x50
[16072.109331]  usbdev_ioctl+0xe/0x20
[16072.109336]  do_vfs_ioctl+0xa4/0x600
[16072.109339]  ? vfs_write+0x15a/0x1b0
[16072.109343]  SyS_ioctl+0x79/0x90
[16072.109347]  entry_SYSCALL_64_fastpath+0x24/0xab
[16072.109349] RIP: 0033:0x7f20da807f47
[16072.109351] RSP: 002b:00007ffc422ae398 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[16072.109353] RAX: ffffffffffffffda RBX: 00000000010b8560 RCX: 00007f20da807f47
[16072.109355] RDX: 00007ffc422ae3a0 RSI: 00000000c0105512 RDI: 0000000000000009
[16072.109356] RBP: 0000000000000000 R08: 00007ffc422ae3e0 R09: 0000000000000010
[16072.109357] R10: 00000000000000a6 R11: 0000000000000246 R12: 0000000000000000
[16072.109359] R13: 00000000010b8560 R14: 00007ffc422ae2e0 R15: 0000000000000000

Reported-and-tested-by: Richard Hughes <rhughes@redhat.com>
Tested-by: Aaron Skomra <Aaron.Skomra@wacom.com>
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Fixes: 7f1a57fdd6cb ("power_supply: Fix possible NULL pointer dereference on early uevent")
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.co.uk>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/power/supply/power_supply_core.c
include/linux/power_supply.h